Revitalizing the Ganges Coastal Zone: Turning Science into Policy and Practices CONFERENCE PROCEEDINGS E. Humphreys, T.P. Tuong, M.C. Buisson, I. Pukinskis, M. Phillips Revitalizing the Ganges Coastal Zone: Turning Science into Policy and Practices CONFERENCE PROCEEDINGS E. Humphreys, T.P. Tuong, M.C. Buisson, I. Pukinskis, M. Phillips May 2015 Copyright ©2015, CGIAR Challenge Program on Water and Food Unless otherwise noted, you are free to copy, duplicate or reproduce, and distribute, display, or transmit any part of this paper or por ons thereof without permission, and to make transla ons, adapta ons or other deriva ve works under the following condi ons: ATTRIBUTION. The work must be a ributed but not in any way that suggests endorsement by CPWF or the author(s). NON-COMMERCIAL. This work may not be used for commercial purposes. SHARE ALIKE. If this work is altered, transformed, or built upon, the resul ng work must be distributed only under the same or similar license to this one. Any views expressed in this publica on are those of the authors. They do not necessarily represent the views of CPWF, the authors’ ins tu ons or the financial sponsors of this publica on. ISBN 978-984-33-9151-3 Cita on Humphreys, E., T.P. Tuong, M.C. Buisson, I. Pukinskis and M. Phillips. 2015. Revitalizing the Ganges Coastal Zone: Turning Science into Policy and Prac ces Conference Proceedings. Colombo, Sri Lanka: CGIAR Challenge Program on Water and Food (CPWF). 600pp. The Revitalizing the Ganges Coastal Zone conference was organized by the CGIAR Challenge Program on Water and Food’s Ganges Basin Development Challenge. The CGIAR Challenge Program on Water and Food (CPWF) was launched in 2002 and concluded in 2013. CPWF aimed to increase the resilience of social and ecological systems through be er water management for food produc on (crops, fisheries and livestock). It did so through an innova ve research and development approach that brought together a broad range of scien sts, development specialists, policy makers and communi es, in six river basins, to address the challenges of food security, poverty and water scarcity. Learn more at waterandfood.org CPWF was a partner of the CGIAR Research Program on Water, Land and Ecosystems, which combines the resources of 11 CGIAR Centers and numerous interna onal, regional and na onal partners to provide an integrated approach to natural resource management research. This program is led by the Interna onal Water Management Ins tute (IWMI). Learn more at wle.cgiar.org. Financial support for the Revitalizing the Ganges Coastal Zone conference was provided by the CGIAR Research Program on Water, Land and Ecosystems and the CGIAR Global Rice Science Partnership. Front cover photo: A high yielding rice variety in polder 30 on 2 November 2012 that will be ready for harvest by mid-November, in me for mely sowing of a high yielding/high value rabi crop. Photo credit: Liz Humphreys Back cover photo: A tradi onal aman variety is planted in Patuakhali on 21 August 2013. Late plan ng of old (tall) seedlings is due to the high water depth (a result of poor water management, lack of drainage). The crop will not be ready for harvest un l late December. Photo credit: Liz Humphreys Table of Contents Foreword.............................................................................................................................................................. 6 Conference Summary........................................................................................................................................... 7 Reviewers ........................................................................................................................................................... 10 Conference Commi ees..................................................................................................................................... 11 Sec on 1: Coastal Water Resources................................................................................................................... 13 Present surface water resources of the Ganges coastal zone in Bangladesh Z.H. Khan, F.A. Kamal, N.A.A. Khan, S.H. Khan and M.S.A. Khan ....................................................................... 14 External drivers of change, scenarios and future projec ons of the surface water resources in the Ganges coastal zone of Bangladesh Z.H. Khan, F.A. Kamal, N.A.A. Khan, S.H. Khan, M.M. Rahman, M.S.A. Khan, A.K.M.S. Islam and B.R. Sharma. 27 Climate change impacts on crop produc on and water requirement in southern Barisal, Bangladesh M. Maniruzzaman, J.C. Biswas, M.A.I. Khan, G.W. Sarker, S.S. Haque, J.K. Biswas, M.H. Sarker, M.A. Rashid, N.U. Sekhar, A. Nemes, S. Xenarios and J. Deelstra ........................................................................................... 39 Groundwater salinity zoning for development plans: A case study of four sub-districts in the southwestern coastal region of Bangladesh M.R. Hasan, M. Shamsuddin, M.S. Masud and A.F.M.A. Hossain ...................................................................... 53 Effect of groundwater use on groundwater salinity, piezometric level and boro rice yield in the Sundarbans of West Bengal D. Burman, K.K. Mahanta, S.K. Sarangi, S. Mandal, B. Maji, U. K. Mandal, B.K. Bandyopadhyay, E. Humphreys and D. K. Sharma ............................................................................................................................................... 61 Reducing irriga on water requirement of dry season rice (boro) in coastal areas using mely seeding and short dura on varie es S.K. Sarangi, D. Burman, S. Mandal, B. Maji, T.P. Tuong, E. Humphreys, B. K. Bandyopadhyay and D. K. Sharma ....................................................................................................................................................... 68 Sec on 2: Coastal Environment ......................................................................................................................... 81 Bacteriological assessment of managed aquifer recharge (MAR) water in southwest coastal areas of Bangladesh M. P. Kabir, M. A. Islam and M. A. Akber ........................................................................................................... 82 Effects of controlling saline water intrusion in an empoldered area of Bangladesh M. C. Rahman, T. H. Miah and M. H. Rashid ...................................................................................................... 89 Sec on 3: Water Governance ............................................................................................................................ 97 Indo-Bangladesh Ganges water interac ons: From water sharing to collec ve water management P. Saikia and B. Sharma ...................................................................................................................................... 98 Community water management and cropping system synchroniza on: The keys to unlocking the produc on poten al of the polder ecosystems in Bangladesh M.K. Mondal, E. Humphreys, T.P. Tuong, M.N. Rahman and M.K. Islam ........................................................ 119 How successful are community-led organiza ons for water management? Evidence from an assessment of Water Management Organiza ons in Coastal Bangladesh N. Kenia and M.-C. Buisson .............................................................................................................................. 131 Mul ple actors, conflic ng roles and perverse incen ves: The case of poor opera on and maintenance of coastal polders in Bangladesh F.Naz and M-C Buisson ..................................................................................................................................... 147 The imposi on of par cipa on? The case of par cipatory water management in coastal Bangladesh C. Dewan, M.-C. Buisson and A. Mukherji........................................................................................................ 162 Predic ng success in community-driven water infrastructure maintenance: Evidence from public goods games in coastal Bangladesh A. Das, M.-C. Buisson and A. Mukherji ............................................................................................................. 183 Determinants of contract choice in groundwater irriga on markets in Bangaldesh M. Saidur Rahman, M. A. Sa ar Mandal and Humnath Bhandari ................................................................... 197 Sec on 4: Homestead Produc on Systems...................................................................................................... 215 Do homestead food produc on systems hold promise for household food security? Empirical evidence from the southwest coastal zone of Bangladesh M. Karim, M.H. Ullah, K.A. Kabir and M. Phillips.............................................................................................. 216 Homestead farming: A biodiverse system to enhance resilience to climate vulnerability J.K. Sundaray, A. Bha acharya, A.G. Ponniah, T.K. Ghoshal, A.D. Deo, J.P. Sharma and M. Phillips ............... 231 Homestead produc on systems in Sundarbans region of West Bengal, India – Current status and opportuni es S. Mandal, D. Burman, S. K. Sarangi, B. K. Bandyopadhyay and B. Maji ........................................................... 241 Homestead farming system: compara ve characteriza on and role in resource poor farmers’ livelihood in Bangladesh and West Bengal K.A. Kabir, J.K. Sundaray, S. Mandal, D.A. Deo, D. Burman, S.K. Sarangi, A. Bha acharya, M. Karim, M.B. Shahrier, S. Cas ne and M. Phillips.................................................................................................................. 251 Producing fish in small shaded homestead ponds: finding solu ons with rural women K. A. Kabir, G. Faruque, R. Sarwar, B. Barman, A. Choudhury, M. Hossain, E. Hossain, N. A. Aleem, M. Karim, K. Kamp and M. Phillips........................................................................................................................................ 265 Sec on 5: Aquaculture ..................................................................................................................................... 279 Produc vity, diversifica on and resilience of saline aquaculture systems in coastal southern Bangladesh K.A. Kabir , S.B. Saha, M. Karim, C.A. Meisner and M. Phillips ......................................................................... 280 Community-based fisheries management: Improving fish biodiversity in inland fisheries of Bangladesh M. G. Mustafa .................................................................................................................................................. 290 Sec on 6: Rice-Based System Intensifica on and Diversifica on .................................................................... 303 Promising rice genotypes for the wet and dry seasons in coastal West Bengal S.K. Sarangi, D. Burman, S. Mandal, B. Maji, E. Humphreys, T.P. Tuong, B. K. Bandyopadhyay and D. K. Sharma ..................................................................................................................................................... 304 Performance of improved aman rice varie es in the coastal zone of Bangladesh M.R.A. Sarker, M.A. Rahman, N. Sharma, M.R. Islam, M.K. Mondal, G.B. Gregorio, E. Humphreys and T.P. Tuong ............................................................................................................................................................... 320 Challenges and opportuni es for aman rice cul va on in ghers used for brackish water shrimp produc on M. A. Rahman, M. R. A. Sarker, N. Sharma, M. K. Mondal, M. R. Islam, G. B. Gregorio, E. Humphreys and T. P. Tuong........................................................................................................................................................ 333 Op mum sowing window for boro cul va on in the coastal zone of Bangladesh M.K. Mondal, N.K. Saha, S.P. Ritu, P.L.C. Paul, A.K.M. Sharifullah, E. Humphreys, T.P. Tuong, and M.A. Rashid 342 An aus-aman system for increasing the produc vity of a moderately saline region of the coastal zone of Bangladesh S.P. Ritu, M.K. Mondal, T.P. Tuong, S.U. Talukdar and E. Humphreys ............................................................. 361 Rice-sunflower: An alterna ve cropping system for sustained livelihoods in the coastal zone of Bangladesh M. Afsar and T.H. Miah .................................................................................................................................... 389 Oilseed crops in rice-based cropping systems in southern Bangladesh M.H. Rashid, F. Hossain, D.K. Nath, P.C. Sarker, A.K.M. Ferdous and T. Russell .............................................. 405 Triple rice in a year: Is it a feasible op on for the low salinity areas of the coastal zone of Bangladesh? N. K. Saha, M. K. Mondal, E. Humphreys, J. Bha acharya, M. H. Rashid, P. C. Paul and S. P. Ritu .................. 421 Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh J. Bha acharya, M. K. Mondal, E. Humphreys, N. K. Saha, M. H. Rashid, P. C. Paul and S. P. Ritu .................. 436 Opportuni es for cropping system intensifica on in the coastal zone of Bangladesh M.K. Mondal, P.L.C. Paul, E. Humphreys, T.P. Tuong, S.P. Ritu and M.A. Rashid............................................. 449 Op mizing use of fresh and saline water for irriga on of boro rice in salt affected areas of Bangladesh using the crop model ORYZA v3 A.M. Radanielson, O. Angeles, T. Li, A.K. Rahman and D. Gaydon .................................................................. 477 Response of wheat, mustard and watermelon to irriga on in saline soils A.R. Akanda, S.K. Biswas, K.K. Sarker, M.S. Mondal, A.F. Saleh, M.M. Rahman and A.Z.M. Mosleuddin ........ 492 Rabi crop establishment methods for increasing land produc vity in the coastal zone of Bangladesh M.N. Rahman, M.G.M. Amin, M.K. Mondal and E. Humphreys ....................................................................... 504 Screening of watermelon varie es for the coastal area of Khulna M.M. Hossain, S.M. Zaman, P.K. Sardar and M.M. Howlader .......................................................................... 516 Sec on 7: Outscaling........................................................................................................................................ 521 Targe ng improved cropping systems in the coastal zone of Bangladesh: A decision tree approach for mapping recommenda on domains P.K. Chandna, A. Nelson, M.Z.H. Khan, M.M. Hossain, M.S. Rana, M. Mondal, S. Mohanty, E. Humphreys, F. Rashid and T.P. Tuong .................................................................................................................................. 522 Decentralized surface water irriga on as a pathway for sustainable intensifica on in southern Bangladesh: On how much land can the drop be brought to the crop? U. Schulthess, T.J. Krupnik, Z.U. Ahmed, and A.J. McDonald ......................................................................... 542 Poten al for expansion of surface water irriga on through axial flow pumps to increase cropping intensifica on in southern Bangladesh A.S. Qureshi, S. Yasmin, N.C. Howlader, K. Hossain and T.J. Krupnik ............................................................... 553 Increasing agricultural and aquacultural produc vity in the coastal zone of Bangladesh M. Sirajul Islam, S.K. Biswas, D. Gain, M.A. Kabir and T.A. Quarashi ............................................................... 566 Agricultural machinery ownership and intensifica on in South Asia: What can we learn from Bangladesh? K.A. Mo aleb and T.J. Krupnik ......................................................................................................................... 576 Conference A endees...................................................................................................................................... 593 CPWF Ganges Partner Organiza ons ............................................................................................................... 597 Foreword The Ganges coastal zone of Bangladesh and West Bengal, India, is characterized by extremes in terms of both challenges and opportuni es. Despite the huge investment in the coastal zone over the past 50 years, the poverty of farming families in the region remains extreme. The CGIAR Challenge Program on Water and Food’s conference ‘Revitalizing the Ganges Coastal Zone: Turning Science into Policy and Prac ces’ brought together researchers, extensionists, development partners and policy makers to share plans, progress and ideas for unlocking the produc on poten al of the coastal zone. The 41 papers in these proceedings provide up-to-date informa on on current and likely future water resource availability, and opportuni es for improving water management, increasing produc vity, and improving livelihoods. Specific topics include: • the current status of coastal zone surface water resources and likely impacts of key drivers of change, such as climate change, sea level rise and construc on of the Ganges Barrage • opportuni es for increasing the produc vity of available surface and groundwater resources • current water governance policy and prac ce at local and Indo-Bangladesh scales, and proposals for improving water management through improved ins tu onal arrangements • the importance of homestead produc on systems, their current (low) produc vity, and opportuni es for improving produc vity and livelihoods and empowering women • opportuni es for increasing the produc vity of brackish water aquaculture • opportuni es for increasing the produc vity, diversity and resilience of rice-based produc on systems • outscaling of improved produc on technologies and extrapola on domains – which improved cropping systems are most suited to which parts of the coastal zone landscape The informa on presented in these papers reveals tremendous opportuni es to improve food security and livelihoods in the coastal zone through making much be er use of exis ng land and water resources, through use of improved germplasm, management and cropping systems. The entry point to unlocking the poten al is improved drainage management, which requires a community approach. We trust that you will find these proceedings a valuable resource for donors and policy makers involved in the planning and implementa on of development projects for the coastal zones of Bangladesh and West Bengal, as well as for researchers, extensionists, teachers and students. Manoranjan Mondal IRRI Chair, Organizing Commi ee 6 Pamela George WorldFish Secretary, Organizing Commi ee Elizabeth Humphreys IRRI Chair, Science Program Conference Summary The ‘Revitalizing the Ganges Coastal Zone: Turning Science into Policy and Prac ces’ conference was held in Dhaka, Bangladesh from 21-23 October 2014. Hosted by the CGIAR Challenge Program on Water and Food (CPWF), the event brought together researchers, extensionists, developers and policy makers in a forum to share plans, progress and ideas for unlocking the poten al of the coastal zone. In total, 224 people a ended the event, with most par cipa ng for two to three days. The conference was sponsored by the CGIAR Research Program on Water, Land and Ecosystems and the CGIAR Research Program on Rice, known as the Global Rice Science Partnership. Day 1 The first day of the conference provided an overview of the challenges and opportuni es for the coastal zone before delving into a more focused discussion on the policies, plans and development projects that have poten al to shape the region. Md. Shahidur Rahman, Director General of the Bangladesh Water Development Board (BWDB), chaired the inaugural session. Dr. Craig Meisner, co-leader of the CPWF Ganges Coordina on and Change Project set the scene with an overview of the region. He touched on the many biophysical constraints to produc on in the coastal zone, which will be exacerbated by climate change, before emphasizing CPWF’s mo va on for working in the area: the belief that there is an opportunity to make a ‘quantum leap’ in food produc on in the coming decades. Dr. Andrew Noble, director of the CGIAR Research Program on Water, Land and Ecosystem (WLE), followed with a presenta on on WLE’s plans to build upon the work of CPWF beyond 2014, when the CPWF program concluded. Photo 1. Inaugural session of the ‘Revitalizing the Ganges Coastal Zone’ conference. From le to right: Dr. Andrew Noble, Director, WLE; Dr. Shelina Afroza, Secretary, Ministry of Fisheries and Livestock; Md. Shahidur Rahman, Director General, Bangladesh Water Development Board; Barrister Anisul Islam Mahmud, MP and Honorable Minister, Ministry of Water Resources; Dr. Zafar Ahmed Khan, Secretary, Ministry of Water Resources; Dr. Craig Meisner, co-leader of CPWF Ganges Basin Development Challenge. 7 Following these introductory remarks, special guest Dr. Shelina Afroza, Secretary of the Ministry of Fisheries and Livestock, spoke of the Government of Bangladesh’s aspira ons to become a middle income country by the year 2021. She underscored the need for collabora on between the research, private and government sectors in order to achieve this goal, as well as effec ve integra on across ministries. Conference a endees then heard from special guest Dr. Zafar Ahmed Khan, Secretary of the Ministry of Water Resources who discussed the role of policy in taking science and knowledge and turning it into programs and prac ce. He also noted the diversity of ministries in a endance and the value of such an integrated approach to addressing development challenges. Chief guest Barrister Anisul Islam Mahmud, MP and Honorable Minister of the Ministry of Water Resources devoted his presenta on to discussing the gap that exists between the knowledge and understanding of policy makers and the reality of what is occurring on the ground. This disconnect, he stated, can result in situa ons where researchers, policy makers and development prac oners are not working in a concerted effort. He noted the value of mee ngs such as this for providing a space for professionals from different backgrounds to converge and provide useful input to policy makers. The session concluded with a speech from the Chair, Md. Shahidur Rahman, Director General of BWDB, who underscored the importance of integrated water management in the polders. The next session was tled ‘Coastal Zone Development Program: Towards ‘Water Smart Communi es’’. It featured presenta ons from Government of Bangladesh officials who each provided an overview of their agency’s work in the coastal zone. A endees heard presenta ons from: Saiful Alam, Water Resources Planning Organiza on, on the Bangladesh Water Act and Coastal Zone Strategy; Sarafat Hossain Khan, BWDB, on the Coastal Embankment Improvement Project of the World Bank, and Tahmina Begum, Deputy Director of the Department of Agriculture Extension, on the Blue Gold project. Photo 2. Members of coastal zone communi es par cipate in Day 1 of the conference. Farmer Altaf Boya from Amtali Upazila of Barguna District is speaking. 8 Dr. T.P. Tuong, formerly of IRRI, then provided an overview of the main messages that emerged from the CPWF’s decade of work in the Ganges coastal zone. The primary message, he stated, is “Water resources in the coastal zone have largely been misconceived and underu lized. In reality, they are a rich and valuable resource to support agricultural and aquacultural produc on and livelihood improvement of farming families and communi es.” Representa ves from different development programs and donors were then invited to discuss their work and reflect on whether the CPWF’s messages reflected their understanding of the region. Presenta ons were given by Dirk Smits of Blue Gold, Nicholas Syed of IFAD, Mike Robson of FAO, Md. Khaleduzzaman of the Embassy of the Netherlands, and Jaap de Heer of the Bangladesh Delta Plan. The presenters iden fied many overlaps. The final session of the first day was tled ‘Revitalizing the Ganges Coastal Zone: Influencing Policies and Implementa on Strategies’. The session began with a joint presenta on on water smart communi es from Liz Humphreys (Interna onal Rice Research Ins tute) and Marie-Charlo e Buisson (Interna onal Water Management Ins tute), who each lead one of the five CPWF Ganges projects. They highlighted the changes needed in water management to realize the huge produc on poten al in the polders of the coastal zone. Furthermore, integrated policy implementa on will require greater coordina on between ins tu ons responsible for water management, food produc on and dissemina on. The twenty-four representa ves of coastal zone communi es, including one member of the Parliament were then invited to comment on what they had heard during the day, and their percep ons of the remaining challenges for intensifying produc on in the coastal zone. The representa ve agreed that it is possible to grow three rice crops or two rice crops and one rabi crop in parts of the coastal zone, and integrated rice-fish in highly saline areas, but stated that opera on of the sluice gates and coordina on of community members remained a large challenge. The session concluded with a panel discussion on policies and priori es for future investments to raise the produc on poten al of the coastal zone. Discussions centered around the need to address waterlogging due to silta on in order to improve the lives of those in the coastal zone. Days 2 and 3 The second and third days of the conference were devoted to oral and poster presenta ons on coastal zone findings from a large range of research projects conducted by many Bangladesh, Indian and interna onal organiza ons. Topics covered included: current and future status of coastal zone water resources and biodiversity; improving water governance; homestead produc on systems—current status, future opportuni es; increasing the produc vity of brackish water aquaculture; increasing the produc vity and diversity of rice-based produc on systems, and; outscaling produc on technologies—what will work best where. Presenters were invited to develop their abstracts into full papers, many of which are presented in the following pages. 9 Reviewers All papers published in these proceedings were reviewed by at least one independent reviewer, and most papers had two reviewers. We are grateful to the following people for their contribu ons to the review process. 10 M.A.R. Akanda Mac Kirby M.R. Akhlas Tim Krupnik Olivyn Angeles Ruben Lampayan Frederic Aubery Minhaj Mahmud Randy Barker M. Mainuddin Ram Basko M. Maniruzzaman Ja sh Biswas Archisman Mitra Marie-Charlo e Buisson C.V. Mohan Romy Cabangon Manoranjan Mondal Sarah Cas ne Venkatesh Moorthy Camelia Dewan Andy Nelson Nepal Dey Phong Ngo Willie Erskine Mike Phillips Animesh Gain Ando Radanielson Don Gaydon M.A. Rahman Glenn Gregorio Abudal Rashid Andrew Hartly M.A. Rashid Chu Thai Hoanh M.H. Rashid M.M. Hossain Sanjida Ritu Elizabeth Humphreys Sukanta Sarangi M.R. Islam T.P. Tiwari Abdel Ismail T.P. Tuong Fazlul Karim Sudhir Yadav Nandish Kenia Murat Yakubov Conference Commi ees Organizing Commi ee Manoranjan Mondal (chair) Pamela George (secretary) Mohammad Alamgir Benoy Barman Elizabeth Humphreys Zahirul Haque Khan Craig Meisner Science and Editorial Commi ee Elizabeth Humphreys (chair) To Phuc Tuong Marie-Charlo e Buisson Manoranjan Mondal Michael Phillips Logis cs Commi ee Nazneen Ahmed Bijoy Bhusan Debnath Hamida Lipi Akter Minam Huq Communica ons Commi ee Nitasha Nair Ilse Pukinskis Mohammad Mahabubur Rahman Michael Victor Budget Commi ee Pamela George 11 12 Sec on 1 Coastal Water Resources 13 Present surface water resources of the Ganges coastal zone in Bangladesh Z.H. Khan1 , F.A. Kamal1, N.A.A. Khan1 , S.H. Khan2 and M.S.A. Khan3 1 Ins tute of Water Modelling, Bangladesh, zhk@iwmbd.org, fal@iwmbd.org, nkh@iwmbd.org 2 Bangladesh Water Development Board, Bangladesh, sarafat.khan@gmail.com 3 Bangladesh University of Engineering and Technology, Bangladesh, msalamkhan@iwfm.buet.ac.bd Abstract The water resources in the coastal zone of the Ganges delta are vital for crop produc on, fisheries, ecosystem sustenance and livelihoods. Beyond water ’s func ons in the hydrological cycle, it also has social, economic and environmental values and is essen al for enhancing food produc on and sustainable development. The main objec ve of the research presented here was to inves gate the temporal and spa al varia on of salinity and to map freshwater availability in the Ganges coastal zone of Bangladesh. The present condi ons of salinity intrusion and availability of fresh water were analyzed based on historical data, detailed monitoring of salinity, water flow and water level, and on numerical modeling. Field measurements show that salinity remains below 2 ppt in the Kazibacha River and Lower Sholmari River at loca ons adjacent to Polder 30 un l February each year. Storage capacity of the canal systems of Polder 30 is sufficient to store adequate fresh water to enable produc on of (irrigated) boro (dry season rice) on about 25% of the cul vable area. Scarcity of fresh water is acute in Satkhira District. Here, salinity remains above 2 ppt (~3 dS/m) throughout the year and peak salinity is about 20 ppt. Thus the river water is never suitable for irriga on of dry season agricultural crops, however, it is a valuable resource for shrimp cul va on. In contrast to Satkhira, there is abundant fresh water for irriga on in most of Barisal Division throughout the en re year. This area receives a huge volume of fresh water from the Lower Meghna River through the Tentulia, Bishkhali and Burishawar (Payra) rivers. The monthly minimum average flow in the Payra River is about 5400 m3/s during the dry season. The river salinity remains below 1 ppt during the whole year. This implies that there is plenty of water for irriga on for the rabi, kharif-1 and kharif-2 crop seasons. Furthermore, gravity irriga on is feasible during the kharif-1 and kharif-2 seasons as the river level is above the average land level for significant periods of me due to diurnal dal water level fluctua ons of 2 to 3 m. In addi on, drainage is feasible as the river level is also below the average land level for significant periods of me. The coastal zone is rich in water resources that could be used to greatly increase the land and water produc vity of agriculture and aquaculture in the Ganges coastal zone of Bangladesh. Key message: There is abundant fresh water for irriga on in much of the Barisal Division of Bangladesh throughout the dry season. Cropping system intensifica on and land produc vity can be increased u lizing the available water. Keywords: de, salinity, water availability, salinity modeling, coastal polder, agriculture, aquaculture 1. Introduc on With increasing pressure from popula on growth and industrializa on, it is more important than ever to assess the availability and produc vity of water for agriculture and aquaculture systems in Bangladesh. This is especially so in the Ganges coastal zone of Bangladesh since this area is characterized by mul ple problems and opportuni es. Coastal water resources not only support agriculture, aquaculture and industrial ac vi es but also provide naviga on routes and other ecosystems. 14 The Ganges coastal zone is characterized by a vast network of river systems, an ever-dynamic estuary, and is the drainage path of a huge river basin covering parts of India, Bangladesh, China and Nepal. In the late 1960s and early 1970s, 139 polders were built in the coastal zone of Bangladesh for intensifica on of agriculture and protec on of lives and livelihoods of the coastal communi es. A polder is a low-lying land area protected from dal inunda on and salinity intrusion by a peripheral earthen embankment. About 54 polders are located in the Ganges coastal zone. The salinity front moves seasonally as well as spa ally in the coastal area. The seasonal movements of the salinity front are predominantly governed by the varia ons in freshwater discharge from upstream, des, coastal currents and mixing processes. Maps presen ng the zones of availability of fresh and brackish water could be used to inform the planning and design of agriculture and aquaculture systems. Research on temporal and spa al varia on of salinity, and mapping and quan fica on of the availability of freshwater in the Ganges coastal zone has been limited. There are huge knowledge gaps on the poten al for irriga on during the dry season for cropping intensifica on in the greater Barisal and Khulna Divisions. Therefore the objec ves of the present research were to assess the temporal and spa al varia on of salinity and to map freshwater availability in the Ganges coastal zone of Bangladesh. 2. Study area The study area comprises the coastal zone of the Ganges delta in southwest and south-central Bangladesh (Fig. 1). The area is characterized by a vast network of dal rivers and creeks. There are 56 polders in the region. Three polders were selected for detailed analysis based primarily on salinity, availability of fresh water and cropping systems. These polders are situated in high (polder 3), medium (polder 30) and low (polder 43/2F) salinity zones. Tides in the Bay of Bengal are semi-diurnal in nature, exhibi ng two high waters and two low waters per day. The amplitudes of the two cycles differ slightly. Over a longer term, a fortnightly varia on in amplitude between spring and neap des is also evident, with spring de amplitude approximately 2.5 to 3 mes higher than that of the neap de. The dal range is about 1.8 to 3.3 m during spring de in the dry season in the Pussure Estuary and the dal effect penetrates more than 150 km upstream during the dry season. The salinity front moves seasonally as well as spa ally in the coastal area. These movements are predominantly governed by varia ons in freshwater discharge, des, coastal currents and mixing processes. The availability of freshwater for agriculture and domes c use in the coastal area is mainly dependent on upstream river water flows. The main source of freshwater in the Khulna region is the Ganges River and freshwater flows through the Gorai River to this region. Salinity starts to increase from December with the decrease in upstream fresh water flow in the medium saline zone i.e. in the Rupsha, Kazibacha and Pussur river systems. The main source of freshwater in the Barisal Region is the Lower-Meghna River and salinity in most of Barisal Region remains low throughout the year. 15 Study Area Legend Di stri ct Hq Wa te rbod y Interna ti ona l Boun dary Sun drba n Are a 0 Ri ver Sel ected Pol der for detai led Investig ation Projection : Bangladesh Transvers Mercator Pol der 5 10 20 30 Kil om et ers Path: D:\P00 20\51152_Economics_Clim ateChange\Mxdfile\SW_Region _Up to_Ganges_Selected_Polder_map .mxd [*L pb*]Date Saved: 16 /03/2015 11:3 4:54 AM Fig. 1. Map showing the study area. Pink areas are (le to right) polder 3, 30 and 43/2F. 16 Tidal water level fluctua ons in the rivers in the study area are semi-diurnal (Fig. 2). The amplitudes of the two daily dal cycles differ slightly. Over a longer term, a fortnightly varia on in amplitude between spring and neap des is also evident, with spring de amplitude approximately 2.5 to 3 mes higher than the neap de amplitude. 2.5 Tidal range = 3.4m 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 00:00 2011-03-14 00:00 03-16 00:00 03-18 00:00 03-20 00:00 03-22 00:00 03-24 00:00 03-26 Fig. 2. Varia on of de level during spring and neap de at Hiron Point in the Pussur River. Field visits, community consulta on and data analysis show that the problems in the polders are diverse and that there are opportuni es for intensifica on of cropping systems if proper water management is put in place. In Polder 3, located in Satkhira District, salinity is a major problem for agriculture as well as for drinking water. This polder is surrounded by the Ichhamo River, Kansiali River and Habra Khal. Salinity in these rivers rises above 20 ppt (~31 dS/m) in the dry season. Salinity in the Ichhamo and Kansiali rivers starts decreasing in mid-June and drops below 5 ppt in mid-July but remains above 2 ppt throughout the year. However, salinity in Habra Khal remains high because there is lack of freshwater flow from upstream. Farmers mainly prac ce shrimp farming in the southern half of this polder but also transplant aman rice in rota on with shrimp in some parts. Shrimp farmers have installed many pipes and cuts in the polder embankment to flush the shrimp ghers using saline river water during every spring de (i.e. during the new and full moon) of the shrimp season. However, these cuts and pipes weaken the embankment and thus jeopardize the safety of the polder, and generate huge social problems and conflicts between the Bangladesh Water Development Board (BWDB) and shrimp farmers. The present scenario is very different from that in the 1970s when BWDB constructed the polders and drainage system. Revisi ng the problems and improving the water infrastructure of the polder could ensure proper management and improvement of the drainage and flushing systems. This should include construc on of planned flushing sluices for shrimp produc on and excava on of the drainage channels. Polder 30 is surrounded by the Lower-Sholmari, Kazibacha and Jhapjhapia-Monga rivers and is located in Khulna District. The river receives freshwater through the Gorai River. There is plenty of water flow throughout the year in the peripheral river systems. Salinity varies seasonally and remains below 2 ppt from mid-June to February. Polder 43/2F is surrounded by the Payra and Gulishakhali rivers, which receive plenty of flow from the Lower-Meghna River. Consequently, the river salinity remains very low, even in the dry season. 17 3. Methodology Water level (depth), flow rate and salinity were monitored in the river systems of the study area for over three years. Detailed water level and flow measurements were carried in the peripheral rivers of the three selected polders. The measurements were made at half hour intervals over a full dal cycle (12 hours and 25 minutes) during neap and spring des in the dry and monsoon seasons. To assess the varia on of salinity over me, measurements were carried out on alternate days at high and low water at 36 sta ons across the Ganges coastal zone. In addi on, secondary data from BWDB were used, especially water discharge in the Ganges River at Hardinge Bridge and in the Gorai River at Railway Bridge. The current condi ons of salinity intrusion and availability of fresh water were analyzed based on historical data, the detailed field measurements over 3+ years as outlined above, and by applying numerical modeling techniques. Numerical models are effec ve tools to characterize and predict the quan ty and quality of water. These models provide a means of genera ng a con nuous me series of water level, flow and salinity at a large number of loca ons in the study area. The river model developed by IWM for the southwest region (based on the MIKE modeling system) was used to establish hydrologic condi ons and varia on of salinity over space and me across the en re study area (IWM 2005). The river systems and boundary condi ons of the salinity model are shown in Figure 3. Fig. 3. The river systems and boundary condi ons of the southwest region salinity model. The model was calibrated using measured data for 2012 and validated against measured data for 2011, for water flow, water level and salinity in the different rivers. The agreement between model generated results and measured data was good for both water flow (R2 from 0.87 to 0.95 in different rivers) and salinity (R 2 from 0.73 to 0.97). Figure 4 shows an example of the good performance of the model in predic ng salinity of the Pussur River at Mongla. 18 14 12 Salinity (PPT ) 10 8 6 4 2 0 February 2012 March 2012 April 2012 Time May 2012 June 2012 Fig. 4. Calibra on results of salinity model in the Pussur River at Mongla (model calibra on data set). 4. Results and discussion The main source of freshwater in Khulna Division is the Ganges River. The flow during November to April is vital in pushing the salinity front downstream and enhancing the freshwater area in the river for irriga on of boro rice and for other domes c and industrial uses. Bangladesh Water Development Board measures the daily water flow in the Ganges River at the Haridinge Bridge. Average daily flow over the period from November to April in 1998 to 2000 was in excess of 3000 m3/s, but since then it has generally been below 2500 m3 /s, with a record low of 2013 m3/s (Table 1). 19 Table 1. Flow sta s cs at Hardinge Bridge (November-April) 1 Year Maximum flow1 (m3 /s) Minimum flow1 (m3 /s) Average flow (m3/s) 1998 11061 1038 3247 1999 20965 558 3452 2000 16396 768 3139 2001 6794 452 2388 2002 9297 728 2479 2003 7007 909 2608 2004 13629 1039 3462 2005 7854 775 2230 2006 9328 567 2465 2007 6949 951 2144 2008 8902 765 2456 2009 8789 1058 3108 2010 NA NA NA 2011 7916 766 2309 2012 10757 657 2555 2013 5715 826 2013 Instantaneous flow The dry season flow in the Gorai River depends on the available flow in the Ganges River and on the connec vity of the Gorai with the Ganges at the upstream end. The Gorai River is usually disconnected from the Ganges in the dry season because of the huge amount of sedimenta on. Restora on of the Gorai River was done in 2011 to 2013 and as a result there was flow through the Gorai from the Ganges. The average daily flow rate in the Gorai River over the period from November 2011 to April 2012 was 261 m3/s, while the maximum instantaneous flow rate was about 693 m3 /s. The average daily flow rate is computed as the average of all the daily river flow determina ons (m3/s) for the period 1 November to 30 April. The flow dura on curve for that period shows that 90% dependable flow is less than 1000 m3/s in the Ganges River at Hardinge Bridge during the dry season (Fig. 5). 25000 Flow (m³/s) 20000 15000 10000 5000 0 0 10 20 30 40 50 60 70 80 Probability of exceedance (%) 90 100 Fig. 5. Flow dura on curve for the Ganges River based on average flow for the period November to April from 1998 to 2008. 20 Salinity in the Lower Meghna River varies from 0 to 2 ppt during the dry season (Fig. 5). The flow in the Lower Meghna River during the dry season is enormous since it carries the combined flow of the Ganges, Brahmaputra and Upper Meghna Rivers. The average daily flow from February to March is about 24000m3 /s in the Lower Meghna River. The Barisal area is connected with the Lower Meghna River through the Sawrupka Kocha-Baleswar, Buriswar and Bishkhali rivers. The area also receives flow from the Padma River through Arial Khan River during the dry season. Field observa ons showed that river water salinity remained below 0.6 ppt throughout the year in much of Barisal Division (e.g. Fig. 6), and the same was true of simulated salinity (data not presented). Simulated monthly mean daily flow in the Payra River varied from 5400 to 6300m3 /s. It is thus evident that much of the Barisal Division has abundant fresh water available for irriga on, aquaculture and other industrial and domes c uses throughout the whole year. The salinity front of 5 ppt remains along the coastline in Barisal during the dry season (Akhter et al. 2012) 5.0 Aus Aman Rabi (Boro) 6200 6000 3.0 5800 2.0 5600 5400 Available flow (cumec) Salinity (ppt) 4.0 6400 1.0 5200 0.0 Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar 2012 2012 2012 2012 2012 2012 2012 2012 2012 2013 2013 2013 5000 Fig. 6. Varia on of salinity (measured, red markers) in the Payra River at a loca on adjacent to Polder 43/2F and monthly mean daily flow (model-generated data, blue line). Simula on and monitoring also showed a limited area with available fresh water in March in Khulna Division. Figure 7 presents the map of water salinity in the Ganges coastal zone in March, a cri cal month for dry season crops, as this is when soil salinity starts to increase rapidly. The map shows that the river water is fresh throughout most of Barisal Division in March, thus the poten al for irriga on. However, the present agricultural prac ces in much of Barisal Division are mainly a rainy season rice crop (aman) using low yielding local varie es, which is some mes preceded by an aus crop and/or followed by a low yielding (non-irrigated) rabi crop such as grass pea or mungbean. The water resources are underu lized in this area although there is huge poten al for enhancing land and water produc vity through the use of modern high yielding varie es of rice and rabi crops and cropping system intensifica on (Bha acharya et al. 2015; Mondal et al. 2015b; Saha et al. 2015). 21 Salinity (ppt) Salinity Zoning Map Base Condition: March, 2012 South West Region LEGEND 0-1 20.01 - 25 1.01 - 2 25.01 - 35 2.01 - 4 4.01 - 5 5.01 - 10 10.01 - 15 15.01 - 20 District Boundary Upazila Boundary Water Bodies N 0 5 10 20 30 40 Km Fig. 7. Model-generated (simulated) river water salinity in the Ganges coastal zone in March 2012 (the blue area, 0-1 ppt, shows that the river water is fresh in large parts of the coastal zone). 22 Gravity irriga on is feasible in the coastal polders since the high de level exceeds the average ground level for quite a long me during at least two cropping seasons (Fig. 8). For example, comparison of the dal water level in the Payra River adjacent to Polder 43/2F and the land eleva on inside the polder shows that the river level is above the median land level for 43% of the me during the kharif-2 season, and for 23% of the me during the kharif-1. In the rabi season the poten al for gravity irriga on is very low. However, the canals can s ll be filled by gravity by opening the regulators at high de, and water can be pumped for irriga on using low li or axial flow pumps. Drainage is also feasible as the river level is below the average ground level for much of the me. This creates the possibility of growing high yield varie es of rice that have shorter stature than tradi onal varie es, bringing the advantage of earlier maturity atnd the opportunity to intensify cropping systems (Bha acharya et al. 2015; Mondal et al. 2015a,b; Saha et al. 2015). Land Level mPWD Legend Below0 .2 0.21 - 0.40 0.41 - 0.60 0.61 - 0.80 0.81 - 1.00 1.01 - 1.20 1.21 - 1.40 1.41 - 1.60 1.61 - 1.80 1.81 - 2.00 2.01 - 2.20 2.21 - 2.40 2.41 - 2.60 2.61 - 2.80 2.81 - 3.00 Over 3. 00 N 0 0.5 1 2 Km Percent of Percent of Percent of Level Area Time Water Time Water Time Water (mPW below Level Exceed Level Exceed Level D) (%) During During Kharif- Exceed Kharif-1 2 During Rabi 1.0 1.2 1.4 1.8 2.0 9 23 52 92 98 46 34 23 5 2 65 53 41 16 7 37 25 14 1 0 Fig. 8. The poten al for gravity irriga on in Polder 43/2F as determined by comparison of land and dal water levels. 23 Polder 30 and adjacent areas in Khulna District are characterized as medium salinity areas. Salinity in the peripheral rivers of this polder starts to increase from early December and reaches a peak in April (Fig. 9). Salinity decreases rapidly to less than 2 ppt in June and remains below 2 ppt un l late January to mid-February depending on upstream flow and rainfall. The simulated monthly mean daily water flow rate in the Kazibacha-Pussur River ranges from 3700 to 5300 m3 /s throughout the year. The river water can thus be used for irriga on from June to early February. The internal drainage canal network is vast and can be filled by gravity at the end of January. The stored water can then be used for irriga on of boro rice and rabi crops in March and April. Analysis of the canal network for the whole polder shows that about 25 to 30% of the cul vable area of Polder 30 could be brought under rice in the dry season using internal storage to finish off the crop. 5500 Aus Aman Rabi (Boro) 12 5000 Salinity (ppt) 10 4500 8 6 4000 2 PPT 4 Available flow (cumec) 14 3500 2 0 Apr May 2012 2012 Jun Jul 2012 2012 Aug Sep Oct Nov Dec 2012 2012 2012 2012 2012 Jan Feb Mar 2013 2013 2013 3000 Fig. 9. Varia on of salinity (measured, red markers) in the Kazibacha River at a loca on adjacent to Polder 30 and monthly mean daily water flow (simulated, blue line) in the medium salinity zone. Salinity is higher in Satkhira District than in the low and medium salinity zones. In this western coastal zone, salinity remains above 2 ppt throughout the year (Fig. 10). Salinity starts to increase gradually in October as rainfall decreases, and increases rapidly from February to May. The high river salinity is due to the absence of freshwater flow from the upstream since the Kobadak and Betna rivers have been cut-off from the Ganges for several decades by sedimenta on. 24 Aus 20 Aman Rabi (Boro) Salinity (ppt) 15 10 5 2 PPT 0 Apr 2012 May 2012 Jun 2012 Jul 2012 Aug Sep 2012 2012 Oct Nov 2012 2012 Dec 2012 Jan Feb Mar 2013 2013 2013 Fig. 10. Varia on of salinity in the Ichamo River adjacent to Polder 3. In the southwestern part of the study region salinity is more than 4 ppt during the dry season, which has enabled wide-scale prac ce of brackish water shrimp farming in Satkhira District. To do this, farmers have constructed many informal structures to bring in the saline river water. Thus saline water can be considered an important livelihood resource. It is important to revisit polder design considering the autonomous change of land use by farmers. 5. Conclusions Although the Ganges Coastal Zone is besieged with mul ple problems and constraints it has tremendous poten al to create innumerable opportuni es for agricultural and aquacultural produc on through improved use of available water resources. River salinity in the Tentulia, Bishkhali, Buriswar and upstream stretch of Baleswar rivers (i.e. most of Barisal Division) was found to be very low throughout the year. The availability of water (high river flows) is high; simula on results show that the minimum flow in the Payra River (the peripheral river of polder 43/2F) during the dry season is 5400 m3/s. Simula on results also show that both gravity irriga on and drainage are feasible during the kharif-1 and kharif-2 seasons. Irriga on can also be prac ced during the dry season by filling the canals at high de using gravity, storing the water in the canal systems and pumping from the canal systems using low li pumps. Fresh water is available in the peripheral rivers of Polder 30 and adjacent polders from June to early February, which can be used for gravity irriga on if needed. The internal drainage canal network is vast and water can be stored during March and April for irriga on of boro rice once the salinity of the river is too high for use for irriga on. About 25 to 30% of the cul vable area of Polder 30 could be brought under rice cul va on in the dry season using water directly from the river un l early February, and water stored in the internal canal network to finish the crop off. In the southwestern part of the study region salinity is higher than 4 ppt during the dry season and many farmers cul vate brackish water shrimp during the dry season using the saline river water. To do this, the farmers constructed many informal structures to bring saline water into the polder. This implies that saline water should be considered an important resource and not a curse. The autonomous change of land use by 25 farmers and opera on of regulators for both drainage and flushing also imply that there is a need to revisit the polder design. The present study reveals that the water resources of the coastal zone are enormous and that the polder systems offer huge poten al for enhancing land and water produc vity in the Ganges coastal zone through making be er use of the available water resources and the dal ecosystem. Acknowledgements This paper presents findings from ‘G4 Assessment of the impact of an cipated external drivers of change on water resources of the coastal zone’, a project of the CGIAR Challenge Program on Water and Food. References Akhter, S., Hasan, M. And Khan, Z.H. 2012. Impact of Climate Change on Saltwater Intrusion in the Coastal Area of Bangladesh; Paper presented in the Eighth Interna onal Conference on Coastal and Port Engineering in Developing Countries, IIT Madras Chennai, 20-24 February, 2012. Bha acharya, J., Mondal, M.K., Humphreys, E., Saha, N.K., Rashid, M.H., Paul, P.C. and Ritu, S.P. 2015. Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh. These proceedings. Brammer, H. 1996. Geography of soils of Bangladesh, University Press Ltd. Dhaka, Bangladesh. Dasgupta S. and C. Meisner. 2009b. Climate Change and Sea-level Rise: A Review of the Scien fic Evidence. Environment Department Paper # 118.The World Bank. Dhaka Tribune. 2013. Salinity in coastal aquifers alarming. h p://www.dhakatribune.com/bangladesh/2013/ oct/20/salinity-coastal-aquifers-alarming Government of Bangladesh: Ministry of Water Resources and Water Resources Planning Organiza on. 2006. Integrated Coastal Zone Management Program, Bangladesh: State of the Coast. Hasan, M.R., M. Shamsuddin, and A.F.M.A. Hossain. 2013. Salinity status in groundwater: A study of selected upazilas of southwestern regions in Bangladesh. J. Global Science and Technology (1): 112-122. IWM. 2014. Closure Report, Ganges Basin Development Challenge. IWM. 2005. Impact Assessment of Climate Changes on the Coastal Zone of Bangladesh, Final Report. Kabir, K. A., M. Karim, M. B. Shahrier, S. B. Saha, S. Chakraborty, P. Jharendu and M. J. Phillips 2014. Homestead farming systems in southwest Bangladesh: A survey. CPWF Working Paper Series (submi ed). Karim M.F., and Mimura N. 2008. Impacts of climate change and sea-level rise on cyclonic storm surge floods in Bangladesh, Global Environmental Change 18 (3), 490-500. Mondal, M.K., Humphreys, E., Tuong, T.P., Rahman, M.N. and Islam. M.K. 2015a. Community water management and cropping system synchroniza on: The keys to unlocking the produc on poten al of the polder ecosystems in Bangladesh. These proceedings. Mondal, M.K., Paul, P.L.C., Humphreys, E., Tuong, T.P., Ritu, S.P. and Rashid, M.A. 2015b. Opportuni es for cropping system intensifica on in the coastal zone of Bangladesh. These proceedings. Saha, N.K., Mondal, M.K., Humphreys, E., Bha acharya, J., Rashid, M.H., Paul, P.L.C. and Ritu, S.P. 2015. Triple rice in a year: Is it a feasible op on for the low salinity areas of the coastal zone of Bangladesh? These proceedings. Tingsanchali, T. and Karim, M. F. 2005. Flood hazard and risk analysis in the southwest region of Bangladesh. Hydrol. Process., 19: 2055–2069. doi: 10.1002/hyp.5666 26 External drivers of change, scenarios and future projec ons of the surface water resources in the Ganges coastal zone of Bangladesh Z.H. Khan1 , F.A. Kamal1, N.A.A. Khan1 , S.H. Khan2 , M.M. Rahman2, M.S.A. Khan3 A.K.M.S. Islam3 and B.R. Sharma4 1 Ins tute of Water Modelling, Bangladesh, zhk@iwmbd.org, fal@iwmbd.org, nkh@iwmbd.org Bangladesh Water Development Board, Bangladesh, sarafat.khan@gmail.com, mmahfuz82@yahoo.com 3 Bangladesh University of Engineering and Technology, Bangladesh, msalamkhan@iwfm.buet.ac.bd, akmsaifulislam@iwfm.buet.ac.bd 4 Interna onal Water Management Ins tute, b.sharma@cgiar.org 2 Abstract The surface water resource of the coastal zone of the Ganges basin is vital for crop produc on, ecosystem sustenance and livelihoods. In the future this resource is likely to be shaped by various drivers of change. Through a series of par cipatory processes the most important external drivers for the Ganges delta were iden fied, then future scenarios that are likely to affect water availability and produc vity were devised and evaluated using numerical modeling. The main drivers of change were iden fied as trans-boundary river flow, popula on growth, and changes in land use, climate, water governance and water infrastructure development. At present, there is an enormous amount of freshwater suitable for irriga on of agricultural crops throughout the year in much of Barisal Division. The quan ty of fresh water in the coastal zone is likely to decrease due to the combined effects of external drivers and inadequate transboundary flows that will increase river salinity during the dry season. By 2030, the area suitable for irriga on (less than 2 ppt river water salinity) is likely to decrease by about 11% under a moderate climate change scenario (A1B). However under this scenario in 2030 salinity of the rivers in Barguna, Patuakhali and Jhalokathi Districts will not exceed 2 ppt, meaning con nued high availability of river water for irriga on in these regions, even with sea level rise of 22 cm. However, this fresh water pocket in the south-central zone is likely to become more saline (2-4 ppt) with climate change and 52 cm sea level rise in 2050, damaging fish habitat and sources of irriga on water. If the proposed Ganges barrage is built in the Ganges River to divert flow through the Gorai River, an addi onal 2660 km2 area would be exposed to river water salinity of less than 2 ppt compared to the base condi on, i.e. a larger land area would have access to irriga on. With sea level rise of 22 cm, the Ganges barrage would s ll result in an addi onal 886 km2 of land with river water suitable for irriga on. Key message: River water salinity is predicted to remains less than 2 ppt (suitable for irriga on) in 2030 with moderate (22 cm) sea level rise in much of Barisal Division. However, salinity in that area will increase to 4 ppt and above with 52 cm sea level rise in 2050. Keywords: climate change, salinity, sea level rise, transboundary flow, Ganges barrage 1. Introduc on The problems related to surface water resource management in the coastal zone of the Ganges delta are numerous, complicated and challenging. Effec ve resolu on of these problems requires a clear understanding of future demand for water and water availability. Hydrological condi ons vary seasonally, annually and during extreme weather events. Changes in ecosystem services are almost always caused by mul ple interac ng drivers. The present and future agriculture, aquaculture and ecosystem services in the coastal zone of the Ganges delta depend to a large extent on the availability of fresh water. Iden fica on of the key drivers of change and assessment of their effects on water resources are important for future planning for the mul ple uses of water. Of par cular importance is knowledge of the effects of drivers of change on future salinity intrusion, availability of fresh water and drainage. 27 A driver of change is any natural or human-induced factor that directly or indirectly causes a change in an ecosystem (Carpenter et al. 2006). The categories of global driving forces are demographic, economic, socio-poli cal, cultural and religious, scien fic and technological, and physical and biological. Drivers in all categories other than physical and biological are considered indirect. Important direct drivers include changes in climate, land use and water use. Climate change, high popula on growth rate, rapid urbaniza on, expansion of infrastructure, increasing pollu on and changes in land and water use all translate into changes in water flows and the availability of water. Previous research has been limited to the effects of sea level rise and found that with a 60 cm sea level rise along the coast of the south-central zone the 5 ppt salinity front would move 55 km inland (Akhter 2012). Detailed research was carried out by the World Bank to determine the impacts of sea level rise on salinity intrusion. “Slight saline (<1dS/m)” river area is likely to decrease from 22 percent at the baseline to 13 percent in the worst-case scenarios in the southwest region with 67 cm sea level rise (Dasgupta et al. 2014). The present research was undertaken to iden fy the key drivers of change in surface water resources in the Ganges delta coastal zone of Bangladesh, to priori ze likely future scenarios and to assess the likely impact of these scenarios on salinity intrusion and fresh surface water availability. 2. Study Area The study area is the southwest coastal region of Bangladesh, which is bound by the Ganges and the Padma rivers to the north, the Lower Meghna and Tentulia rivers in the east, the Ichamo River and the Bangladesh-India interna onal boundary to the west and the Bay of Bengal to the south (Fig. 1). A vast network of river systems that are influenced by dal ac on characterizes the area. Tides are semi-diurnal, i.e. two high and two low des per day, the dura on of one dal cycle being 12 hours and 25 minutes. The topography of the region is very flat and strong dal effects are propagated up to 200 km upstream of the coast at mes. The huge freshwater ou low from the Ganges, Brahmaputra and Meghna rivers induces a large zone of brackish water (salinity from 1 to 30 ppt) in the coastal region of Bangladesh through mixing with seawater (32 to 35 ppt). Seasonal movements of the front between seawater and brackish water govern salinity in the northern-most part of the Bay of Bengal. The Ganges River is the main trans-boundary river in the southwest region. The Bhairab, Kobadak, Nabaganga and Chitra rivers used to receive freshwater flow from the Ganges during the dry season. However, these rivers have been disconnected from the Ganges for several decades due to silta on. There is no freshwater flow through these rivers to the study area from the Ganges, but during the monsoon there are freshwater flows from the local catchment due to rainfall. The water from the Ganges River, which flows through its distributary, the Gorai River, is the only major source of fresh water in the southwest zone during the dry season. However, the offtake of the Gorai remains dry during the dry season (December to May) if it is not dredged. Salinity in the Bay of Bengal is high during the dry season, and saline water intrudes inland through the major rivers, namely the Baleswar, Jamuna, lower Meghna, Malancha, Pussur, Sibsa and Tnetulia, through dal effects. There is very inadequate upstream flow coming down the Gorai during the dry season. Consequently, the region is severely affected by salinity intrusion. The Bangladesh Water Development Board built about 56 polders (lands surrounded by large earthen embankments) in the southwest coastal region in the late 1960s and early 1970s. The embankments prevented dal flooding and salinity intrusion, enabling the produc on of a tradi onal rainy season crop and protec ng lives. However, silta on of the rivers (a result of reduced flows from India and the construc on of the polders), sea level rise and the sinking of the land mean that the restora on of the peripheral rivers and planned sedimenta on of the polders is required. The average annual rainfall of the southwest coastal region is about 1,780 mm and the 80% dependable rainfall is 1,480 mm (Mainuddin et al. 2014). Rainfall is lower towards the Indian border and higher in the east. Average annual poten al evapotranspira on (ETo) is 1370 mm. There is a clear surplus of rainfall during the monsoon period from June to October, a balance between rainfall and ETo in the pre- and post-Monsoon months of May and October, and a clear deficit in the boro season months from November to April (Fig. 2). 28 Study Area Legend Di stri ct Hq Wa te rbod y Interna ti ona l Boun dary Sun drba n Are a 0 Ri ver Sel ected Pol der for detai led Investig ation Projection : Bangladesh Transvers Mercator Pol der 5 10 20 30 Kil om et ers Path: D:\P00 20\51152_Economics_Clim ateChange\Mxdfile\SW_Region _Up to_Ganges_Selected_Polder_map .mxd [*L pb*]Date Saved: 16 /03/2015 11:3 4:54 AM Fig. 1. The study area in the southwest coastal region of Bangladesh. 29 600 SW-SC region Rain Rain, ETo, mm 500 ETo 400 300 200 100 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Fig. 2. Mean monthly rainfall and evapotranspira on (ETo) in the southwest coastal region of Bangladesh, during 1985-2010 (Source: Mainuddin et al. 2014). The availability of fresh surface water resources during the dry season is variable, more so towards the southwestern corner. In the far southwest (Satkhira District) salinity in the river systems rises above 20 ppt during the dry season and starts to decline in June, but remains above 2 ppt even during the monsoon (Khan and Kamal 2015). However, fresh surface water resources in Khulna District (to the east of Satkhira) are plen ful during, and for a couple of months a er, the monsoon. Here, river salinity falls to less than 2 ppt in the middle of June. A er the end of the monsoon, fresh water (<2 ppt) is available in the Rupsa, Kazibach–Pussur river system from November to mid-February. In the south-central area (Barisal Division) freshwater is available in much of the region throughout the year since the area receives huge freshwater flows from the Lower Meghna River, even in the dry season. Consequently, salinity remains below 0.6 ppt. 3. External drivers of change – methodology and iden fica on The process for iden fica on and priori za on of external drivers of change for scenario analysis involved a wide range of stakeholders including farmers (agricultural crops and fish), fishers, researchers, academics, and representa ves of planning and implemen ng agencies working in Bangladesh (Fig. 3). 30 Literature review, field informa on and interac on with partners Experts of different disciplines Ques onnair e Survey Triangula on Workshop Researchers of GBDC and other projects in Bangladesh Farmers and Fishers Focus Group Discussions Key external drivers and ranking Workshop on scenario genera on involving stakeholders and local community Scenarios Fig. 3. Flow chart of the process used for iden fica on and selec on of key drivers of change for surface water resources in the southwest coastal region of Bangladesh. GBDC stands for the CGIAR Challenge Program on Water and Food’s Ganges Basin Development Challenge (h p://waterandfood.org/river-basins/ganges/). Two approaches were used to iden fy key drivers of change. Firstly, a ques onnaire survey was conducted with experts in a range of relevant disciplines, researchers of the CGIAR Challenge Program on Water and Food’s Ganges Basin Development Challenge (GBDC), water resources planning professionals and implemen ng/ opera ng organiza ons. The purpose was to compile a list of drivers of change and their ranking. The ques onnaire was developed based on the findings of a literature review, field informa on and consulta on with partner organiza ons. Secondly, focus group discussions were conducted in polders 3 (high salinity), 30 (medium salinity) and 43/2F (low salinity) with farmers, fishers, local traders and women to iden fy important issues on water, agriculture and fisheries, problems and their causes. The purpose of the discussions was to iden fy the local communi es’ an cipated external drivers of change and their likely impacts. Triangula on and integra on of the findings of the two approaches was done through a facilitated workshop involving all stakeholders. The result was consensus on the key drivers of change and their rela ve importance (Table 1). Table 1. List and rank of external drivers of change External Drivers Change in trans-boundary flow Popula on growth Change in water management prac ces Land use change Climate change (including precipita on, temperature) and sea level rise Change in water governance and ins tu ons (including policy change) Water use change Water infrastructure development Urbaniza on Rank 1 2 3 4 5 6 7 8 9 31 4. Development of scenarios Several scenarios were devised (Table 2) based on the priori zed drivers of change using a par cipatory approach in another workshop with all stakeholders including local community representa ves. This was a pioneering effort in Bangladesh to generate scenarios of mul ple drivers for integrated water resources management through a par cipatory approach involving a wide range of stakeholders. A moderate climate change scenario (A1B), with predicted sea level rise of 22 cm in 2030 and 52 cm in 2050, was used in the simula ons. Table 2. Scenarios for assessing the effects of external drivers on water resources in the coastal zone of Bangladesh No. B I II III IV 1 Scenario 1 Outputs Base case Water flow and salinity for the year 2012 is considered as base case/base condi on Minimum trans-boundary flow + Popula on growth + Land use change+ Climate change (including pptn, temp & SLR) • Salinity zoning and water availability map Average trans-boundary flow + Popula on growth + Land use change + Climate change (including pptn, temp & SLR) Maximum trans-boundary flow + Popula on growth + Land use change + Climate change (including pptn, temp & SLR) Minimum trans-boundary flow + Popula on growth + Land use change + Climate change (including pptn, temp & SLR) + Ganges Barrage • as above • Salinity zoning and water availability map for the coastal zone of Ganges delta in Bangladesh • Water level and flow sta s cs (max, min, mean) • Projec on of water for food produc on • Projec on of domes c water demand • Projec on of impact of climate change, SLR • as above • as above pptn = precipita on, temp = temperature, SLR = sea level rise 5. Scenario analysis - water modeling Numerical modeling was used to examine the likely impacts of the scenarios on surface water resources. The Ins tute of Water Modelling (IWM) maintains hydrological, hydrodynamic models for river systems and the Bay of Bengal for simula ng water flow, drainage and salinity. These models have been tested and verified for the river systems of Bangladesh. We used four IWM models in this study—Rainfall-Runoff, Hydrodynamic, Bay of Bengal, and Salinity—to determine the present and future dal varia on, water flows, spa al varia on of salinity, fresh water availability and drainage condi ons of the polders. River salinity in the southwest region depends on the freshwater flow from upstream, rainfall runoff from the adjacent catchment area and the dal dynamics of the river systems. Tidal waves from the Indian Ocean travel through the deeper part of the Bay of Bengal and approach the coast of Bangladesh from the south. The process of mixing freshwater from the upstream river system and saline water from the Bay of Bengal in the coastal water occurs as turbulence, which is generated by wind and dal currents. The water flow and salinity modeling processes are illustrated in the following sec ons. 32 5.1 Rainfall-runoff modeling The Rainfall-Runoff module of the MIKE11 modeling system was applied to es mate runoff from rainfall in various catchments of the southwest coastal region. The Rainfall-Runoff model takes into account catchment characteris cs, rainfall, soil moisture, irriga on and water extrac on from surface or groundwater sources in the catchments, evapora on, percola on and other losses. The model generates runoff, which is the input data for the water flow/hydrodynamic modeling. 5.2 Hydrodynamic modelling To generate me series de and water flow/discharge in the river systems of the study region, the hydrodynamic module of the MIKE 11 modelling system was used. Inputs for the Southwest Regional Model include the rainfall runoff es mates from the Rainfall-Runoff model, water flow data (flow gauge data) at the upstream end of the study area (upstream boundary), water level data/ dal varia on at the downstream ends of the rivers (downstream boundary), and the cross-sec on river data at loca ons along the en re stretch of the rivers. The model simulates con nuity and momentum equa ons and provides water levels at each cross-sec on loca on and water flow me series between successive river cross-sec ons. The model was calibrated by adjus ng the model parameters to achieve good agreement between simulated and observed discharge and water level data (e.g. Fig. 4). 30000 25000 20000 15000 10000 5000 0 -5000 -10000 -15000 -20000 -25000 -30000 -35000 00:00 2012-03-04 00:00 03-06 00:00 03-08 00:00 03-10 00:00 03-12 00:00 03-14 Fig. 4. Simulated (lines) and observed (dots) flow for the calibra on of the water flow model at Char Doani in the Bhaleswar River from 4 to 14 March 2012 (-ve sign is for flood de flow, +ve is ebb de flow). 33 5.3 Salinity modeling Salinity of the Bay of Bengal was simulated using the Bay of Bengal (BoB) model based on the MIKE 21FM modeling system. The BoB model domain extends from Chandpur on the lower Meghna River (23° N) to 16 °N in the Bay of Bengal. The BoB model used the water flow es mates at the upstream boundary. The me-series water level data at the downstream boundary were generated from the DHI Global Tide model. Measured concentra ons of salinity along the upstream boundary and constant 32 ppt salinity at the downstream boundary were used as salinity inputs. A salinity model (the advec on-dispersion module of MIKE 11) was applied to assess river water salinity in the southwest coastal region. Measured concentra ons of salinity along the upstream boundary and generated salinity by the BoB model for the downstream boundary were used as inputs for boundary condi ons. The flow chart of the salinity model is presented in Figure 5. Catchment Informa on Rainfall, Evapora on & water extrac on Rainfall-Runoff Model Global Tide Model Water level at d/s boundary Catchment Runoff Primary and Secondary Data Sources Water Flow at u/s boundary Water Level at u/s boundary Hydrodynamic Model Water flow at u/s boundary Hydrodynamic Model Cross-sec on Water Flow Water Flow Salinity boundary at u/s Salinity Model Salinity boundary at d/s Regional 1-D models Salinity Model 2-D Bay of Bengal Model Fig. 5. Flow chart of the salinity model for the southwest coastal zone of Bangladesh. 6. Salinity and fresh water availability under different scenarios Salinity levels in the southwest coastal zone show a dis nct seasonal varia on due to varia on of upstream flow from monsoon to dry season. Average salinity concentra ons are higher in the dry season than in the monsoon season due to lack of freshwater flow from upstream. Salinity generally builds up during October to early/late May. During May, salinity levels drop sharply due to increased upstream flow and rainfall. Salinity levels are highest during the la er part of the dry season, usually from March to May. 34 The effect of transboundary flow on salinity was assessed in combina on with other drivers of change under different scenarios. Three levels of transboundary flow were used: the minimum (Fig. 6), average and maximum daily flows during the dry season in the Ganges River at Hardinge over a period from 1996 to 2013 (i.e. since the implementa on of the Ganges Water Treaty). Decreased flow in the Ganges River with the construc on of the Farraka Barrage (completed in 1975) resulted in large increases in river salinity downstream. For example, prior to 1975 salinity was below 1 ppt at Khulna and the river water was used for drinking, agriculture and industrial purposes. At present, river water salinity at Khulna increases to more than 15 ppt during the dry season and is unsuitable for most purposes. Fig. 6. Average minimum daily flow (average of daily minimum flow for the period January to May in the Ganges River at Hardinge Bridge from 1960 to 2009). To examine the change in salinity and its spa al varia on under each scenario, the calibrated and validated salinity model was used to prepare maps of surface water salinity for baseline condi ons (2012). The map for March 2012 (Fig. 7) shows that there is fresh water in the river systems of Narial, Jessore and Gopalganj Districts in the northwest, and throughout most of Barisal Division. 35 Salinity Zoning Map Base Co ndition March, 2012 South West Regio n Salinity (ppt) LEGEND 0 -1 20. 01 - 25 1. 01 - 2 25. 01 - 35 2. 01 - 4 4. 01 - 5 5. 01 - 10 10. 01 - 15 15. 01 - 20 District Boundary Water Bodies N 0 5 10 20 30 40 Km Document Path: D:\P002551408_Salinity_Information\MXDFile\Base_march2012.mxd[*mtn*22.03.2015] Fig. 7. Map of surface water salinity distribu on for the baseline condi on of March 2012. Simula ons of scenarios I to III were carried out to examine the combined effects of change in transboundary flow and sea level rise along with popula on growth and land use change. In the model popula on growth is considered as extrac on of domes c water and urban water for major ci es. Land use change is taken as addi onal water extrac on for irriga on for future cropping systems. Under all scenarios except scenario IV the area exposed to freshwater and irriga on water decreases because of sea level rise and decreased transboundary flow in the Ganges. Under scenario-I, the freshwater area (0-1 ppt) is predicted to decrease by 19.3% and the irrigated area by 13.5% (Table 3). In the case of maximum transboundary flow with the same sea level rise, the area suitable for irriga on is 9% less than baseline. However, in January to March a larger area would be exposed to salinity than in April as salinity intrusion would affect a larger area. However under these scenarios salinity in the rivers in Barguna, Patuakhali and Jhalokathi Districts does not exceed 2 ppt, which implies con nued availability of river water suitable for irriga on in these regions. Thus, considerable water will be available for irriga on in much of Barisal Division in 2030 under the A1B climate change scenario with 22 cm sea level rise (Fig. 8). However, part of Narial, Jessore and Gopalganj are likely to become saline, despite the fact that under the base condi on river water in these areas is fresh (Fig. 7). 36 Salinity Intrusion in Coastal Zone of Bangladesh due to Climate Change Salinit y Zoning Map Minimum Transboundary Flow+ Sea L evel Rise (SLR) of 22cm March, 2023 Sout h Weast Region Salinity (ppt) LEGEND 0 -1 20. 01 - 25 1. 01 - 2 25. 01 - 35 2. 01 - 4 4. 01 - 5 5. 01 - 10 10. 01 - 15 15. 01 - 20 District Boundary Water Bodies N 0 510 20 30 40 Km Document Path: D:\P0025\51408_Salinity_Information\MXDFile\MinTBF+SLR_22cm_march2030.mxd[*mtn*22.03.2015] Fig. 8. Spa al distribu on of salinity in the southwest coastal region with 22 cm sea level rise and minimum trans-boundary flow through the Ganges River. Table 3. Effect of climate change (including precipita on, temperature change and sea level rise) and transboundary flow on freshwater availability (flow, m3/s) in the southwest region of Bangladesh during April Salinity level (ppt) Base Condi on (2012) Scenario, 2030 Scenrio-I: MinTBF + CC+popula on+land use change Scenario-II. Average TBF + CC+popula on+land use change Scenario-III:MaxTBF + CC+popula on+land use change Area (sq. km) Area (sq. km) Area (sq. km) Change of Area over base condi on Change of Area over reference condi on Change of Area over reference condi on 0-1 (Potable water) 19,575 15,792 -19.3% 16,738 -14.4% 17,255 -12% 0-2 (Suitable for irriga on) 21,261 18,377 -13.5% 18,909 -11% 19,387 -9% 0-5 (suitable for specific fish species) 26,066 23,517 -9.7% 23,892 -8% 24,722 -5% >5 (Suitable for shrimp) 12,919 15,403 +19% 15,029 +16% 14,199 +10% Note: -ve sign refers to a decrease in area and +ve sign refers to an increase in area over the 2012 baseline condi on 37 Simula on results show that the freshwater zone (0-1 ppt) in Barguna, Patuakhali and Jhalokathi Districts is likely to be lost due to sea level rise of 52 cm in the year 2050. This increase in river water salinity is likely to impact freshwater aquaculture, drinking water supply and agriculture. Installa on of the Ganges barrage to divert flow through the Gorai River and other distributaries of the Ganges River is predicted to result in substan al seaward movement of the dry season salinity fronts. This would result in an addi onal 2660 km2 of land being exposed to less than 2 ppt river water salinity in comparison with the base case, i.e. a larger area would have access to irriga on compared to the present condi on. However, with sea level rise of 22 cm, this addi onal area would decrease to 886 km2. Thus, the Ganges barrage would bring considerable benefits, even under 22 cm SLR. 7. Conclusions An innova ve methodology was established for selec ng and ranking of external drivers of change involving farmers, fishers, researchers, academics and representa ves of planning and implemen ng agencies. The key drivers were considered to be transboundary flow in the Ganges, popula on growth, land use change and climate change. The likely impacts of these changes were considered to be increased salinity, loss of fresh water fish habitat and rice produc on during the dry season, and damage to ecosystems. Reduc on in transboundary flow reduces the area where river water is suitable for irriga on (0-2 ppt salinity) and is most felt in the Gorai dependent area. Currently, there is abundant fresh water for irriga on in much of Barisal Division throughout the dry season. Under moderate climate change (scenario A1B) and with 22 cm sea level rise in 2030, salinity does not exceed 2 ppt in Barguna, Patuakhali and Jhalokathi Districts, ensuring con nued availability of river water suitable for irriga on throughout the year. However, this region is likely to become more saline (2-4 ppt) with 52 cm sea level rise in 2050, damaging fish habitat and sources of irriga on water. Regional coopera on is crucial for transboundary flow sharing, which is in turn necessary for salinity control and sustenance of river system ecosystem services in the face of a changing climate. Adapta on measures like the proposed Ganges and Brahmaputra barrages are important for fresh water security under climate change. New salt tolerant cropping technology will also be needed. Acknowledgements This paper presents findings from G4 ‘Assessment of the impact of an cipated external drivers of change on water resources of the coastal zone’, a project of the CGIAR Challenge Program on Water and Food. References Akhter, S. 2012. Impact of Climate Change on Saltwater Intrusion in the Coastal Area of Bangladesh; Paper presented in the Eighth Interna onal Conference on Coastal and Port Engineering in Developing Countries, IIT Madras Chennai, 20-24 February, 2012. Carpenter, S. R., Benne , E.M. and Peterson, G.S. 2006. Editorial: Special Feature on Scenarios for Ecosystem Services. Ecology and Society 11(2), 32. Available at h p://www.ecologyandsociety.org/vol11/iss2/art32/ (last accessed 15 March 2015). Dasgupta, S., Kamal, F.A., Khan, Z.H., Choudhury, S. and Nishat, A. 2014, River Salinity and Climate Change: Evidence from Coastal Bangladesh. Policy and Research Working Paper no. WPS 6817. Volume 1. The World Bank. Mainuddin, M., Kirby, M., Chowdhury, R.A.R., Sanjida, L., Sarker, M.H. and Shah-Newaz, S.M. 2014. Bangladesh integrated water resources assessment supplementary report: land use, crop produc on and irriga on demand. CSIRO: Water for a Healthy Country Flagship, Australia. Khan, Z.H., Kamal, F.A., Khan, N.A.A, Khan, S.H. and Khan, M.S.A. 2015. Present surface water resources of the Ganges coastal one in Bangladesh. These proceedings. 38 Climate change impacts on crop produc on and water requirement in southern Barisal, Bangladesh M. Maniruzzaman1, J.C. Biswas1, M.A.I. Khan1, G.W. Sarker1, S.S. Haque1, J.K. Biswas1, M.H. Sarker2, M.A. Rashid2, N.U. Sekhar3, A. Nemes3, S. Xenarios3 and J. Deelstra3 1 Bangladesh Rice Research Ins tute, Bangladesh, mzamaniwm@yahoo.com, ja shb@yahoo.com, ashikjp@gmail.com, gwsarker@yahoo.com, shamiulbrri@gmail.com, biswas.jiban@gmail.com 2 Center for Environmental and Geographical Informa on Services, Bangladesh, mhsarker@cegisbd.com, arashid@cegisbd.com 3 The Norwegian Ins tute for Agricultural and Environmental Research, Norway, Nagothu.UdaySekhar@bioforsk.no, A la.Nemes@bioforsk.no, Stefanos.Xenarios@bioforsk.no, Johannes.Deelstra@bioforsk.no Abstract Crop produc on in Bangladesh’s southern Barisal Division is vulnerable because of increasing erra c rainfall, cyclones, dal surge and salinity intrusion. Climate change and resultant sea level rise will exacerbate the situa on. The effects of global climate change on climate in three upazilas of Patuakhali and Barguna Districts in southern Barisal were determined by downscaling, and used to assess the impacts of climate change on surface water resource availability, rainfed crop yields and net irriga on requirement. Mean annual temperature was predicted to increase by 1 to 30 C in the 21st century under IPCC (Intergovernmental Panel on Climate Change) scenario A1B, and by 0.6 to 2.5°C under scenario A2, compared to the baseline of 1971 to 2010. Annual rainfall was usually predicted to decrease in both scenarios throughout most of the 21st century, by 2 to 11%. The excep ons were 4% increases from 2011 to 2040 using scenario A1B in two of the three upazilas, and by 1 to 6% in all three upazilas from 2071 to 2100. A 12 to 30% reduc on in river flows was predicted during the pre-monsoon period using the SWAT (Soil and Water Assessment Tool) model, while monsoon season and total annual water flows were predicted to increase during the 21st century. The DRAS (drought assessment) model predicted a 3 to 7% yield increase of rainfed T. Aus throughout most of the 21st century. On the other hand, rainfed T. Aman yield was predicted to decrease by 2 to 19% due to decreased rainfall and increased temperature. Irriga on water requirement to meet evapora ve demand was predicted to decrease for T. Aus and to increase for T. Aman under the A1B and A2 scenarios. Concerted efforts by regional research and extension organiza ons are needed for the development and promo on of strategies for adapta on of cropping systems to climate change in southern Barisal. Key message: Climate change is predicted to increase grain yield of pre-monsoon rainfed aus crops by 3 to 7% throughout most of the 21st century, and to decrease yield of monsoon rainfed aman crops by 2 to 19%, in southern Barisal, Bangladesh, in the absence of interven ons. Keywords: Climate change, SWAT model, DRAS model, fresh water 1. Introduc on The coastal areas of Bangladesh can broadly be divided into three dis nct regions—eastern, central and western—based on morphological features. The PDO–ICZMP (2003a) classified the coastal areas of Bangladesh under two broad categories: interior coast and exterior coast. Out of 19 coastal districts (147 upazilas) a total of 48 upazilas in 12 districts (23,900 km2) are exposed to the sea and/or lower estuaries and defined as the exposed coast, and the remaining 99 upazilas are termed interior coast with a total area of 47,200 km2 (PDO-ICZMP 2003b) (Figure 1). 39 COASTAL ZONE BANGLADESH BAY OF BEN GAL Dis_line.shp Exposed Coast Interior Coast Bay & River Inc luding Exc lus ive Economic Zone Fig. 1. Coastal regions in Bangladesh (source: BOBLME 2011). The coastal zone of Bangladesh has many rivers and distributaries and complex ecology, and is strongly affected by cyclones, flooding, dal surges and salinity. The coastline is 734 km long and the coastal zone has a popula on of about 50 million people, nearly one-third of the total popula on of the country (Miyan 2009). The majority of the people in the coastal zone are dependent on primary produc on (agriculture and aquaculture), and in par cular on rice produc on, for their food security and livelihoods. The suitability of crops and cropping systems for any region depends on climate, soils and water availability. The climate of the coastal zone is subtropical monsoon, with average annual rainfall of about 2,000 mm, most of which falls from June to October. The rainy season is followed by a cool and dry winter from December to February, and a hot summer in April to June during which there are occasional pre-monsoon rains. During winter and summer, salinity of the river water and soil increases, and availability of fresh water for irriga on is limited in many regions. Therefore, the main crop is aman rice grown during the rainy season over most of the coastal zone. Very limited areas of irrigated rice (boro) are grown during the dry season/early summer in the more northern parts of the coastal zone, and limited areas of aus rice are grown during summer/early rainy season (aus) in the south-central coastal zone. Both the aman and aus crops are predominantly rainfed. Global warming is increasing the risk of drought and sea level rise in Bangladesh. The IPCC Report 5 (2013) indicated that temperature is likely to increase by 1.8 to 3.4°C at the end of the 21st century along with sea level rise of 0.26 to 0.82 m. Sea level rise will shi the coastal periphery inland and could reduce the land area by 17% (Dewan and Nizamuddin 1998; SRDI 1997). A sea level rise of 32 cm by 2050 is also predicted to increase the salt-affected area to 10,600 km2, and to 14,500 km2 with 88 cm sea level rise at the end of this century (Figure 2, CEGIS 2006). The frequency of cyclones, severe flooding and drought is also expected to increase in coming years. 40 0 cm SLR (Area : 9239 sq km) 32 cm SLR (Area : 10612 sq km) 88 cm SLR (Area : 14468 sq km) Fig. 2. Change in the salt-affected area due to sea level rise of 32 cm in 2050 and of 88 cm at the end of the 21st century (source: Center for Environmental and Geographic Informa on Services (CEGIS), Dhaka, Bangladesh). Any varia ons in seasonal monsoon rainfall, temperature and cyclonic storms could significantly influence rice produc on and food security. To meet such challenges, analysis of the impacts of change on crop produc on and iden fica on of poten al adapta on strategies are needed. Therefore, a study was undertaken to es mate the impacts of global warming on the climate at three loca ons in the south-central coastal zone, and on the availability of surface water resources (river flows). The results were then used to determine the impacts of global warming on rice yields and crop water requirements. 2. Methodology 2.1 Site details The study sites were three upazilas in southern Barisal Division - Amtali and Patharghata Upazilas of Barguna District and Kalapara Upazila of Patuakhali District (Figure 3). All upazilas are within polders. Soil proper es were determined based on 30 samples per upazila collected a er harvest of the aman crop. Soil proper es varied widely. Most of the soils are acidic to neutral with generally sa sfactory organic ma er content, but with widespread deficiencies of nitrogen and phosphorus (Table 2). Soil salinity in the upazilas ranged from low to high. Soil salinity (electrical conduc vity of satura on 1:5 soil:water extract, EC1:5) ranged from 1.5 -11.5 dS/m. High soil salinity impairs crop performance at some loca ons, especially in the dry season. 41 88°0'0''E 89°0'0''E 90°0'0''E 91°0'0''E 92°0'0''E 90°0'0''E 90°10'0''E 90°20'0''E 90°10'0''E 90°20'0''E N 0 55 110 K ilomet ers 1:3,80 0,000 Leg en d Dist rict Boundary N St udy Upazila Boundary Non saline wit h some very slight ly saline 0 Very slightly saline with some sli ghtl y saline 3 6 K ilomet ers Sli ghtl y saline wit h some m oderat aly saline 1:44 0,000 Moderately saline with some st rongly saline St rongly saline wi th some very strongly sali ne 90°0'0''E Sundarban River 88°0'0''E 89°0'0''E 90°0'0''E 91°0'0''E 92°0'0''E Fig. 3. Loca on and salinity of the study upazilas (purple boundaries) in southern Barisal, Bangladesh. Table 2. Soil chemical proper es of the study sites (range, median value in parentheses) Upazila (n=30 per upazila) Amtali Kalapara Patharghata Cri cal values 42 EC1:5 (dS/m) pH 1.446.85 (3.50) 1.665.86 (3.21) 1.9111.51 (6.93) 4.37.1 (5.18) 4.75.4 (4.97) 4.57.2 (6.17) - - OM Total N Mg K Meq/100 g soil 0.932.63 (1.73) 0.762.16 (1.55) 0.834.17 (2.25) 0.050.14 (0.09) 0.040.12 (0.09) 0.050.24 (0.13) 5.098.08 (6.62) 4.977.99 (5.83) 3.326.56 (5.20) - 0.12 0.50 % P B µ/g Zn 0.210.72 (0.30) 0.200.34 (0.31) 0.200.34 (0.31) 0.425.76 (2.67) 0.858.76 (3.97) 1.2711.78 (6.55) 0.221.10 (0.80) 0.500.95 (0.71) 0.330.98 (0.71) 0.541.60 (1.11) 0.691.28 (0.99) 0.522.77 (1.23) 0.12 8.00 0.20 0.60 2.2 Climate change scenarios Future climate and water resource scenarios were developed by downscaling the predic ons of global climate change models to the upazila level. The downscaled predic ons were used as inputs for predic on of water resource availability using hydrological modeling, and for predic on of crop irriga on requirement and yield using a crop model. 2.2.1 Climate change The process of assessing future likely climate change for the region involved several steps including: i) review of exis ng model-based climate scenarios, ii) evalua on of climate model outputs and scenarios for Bangladesh, iii) assessment of the limita ons of climate models and downscaling in the context of Bangladesh, iv) iden fica on of currently adopted models, v) selec on of climate models for use in this study, vi) downscaling of climate model results to upazila level, and vii) genera on of climate scenarios. Two IPCC (2013) future climate scenarios were selected for the study: A1B (balanced emphasis on all energy sources) and A2 (self-reliant na ons, con nuously increasing popula on and regionally-oriented economic development). A dynamic downscaling process using the PRECIS (Providing Regional Climates for Impact Studies) regional climate model (Jones et al. 2004) by the Ins tute of Water and Flood Management from 1981 to 2100 was used to generate future climate data for the upazilas. The predicted rainfall pa ern by 2050 was determined based on Rahman et al. (2012). 2.2.2 Hydrological modeling Fresh surface water availability was es mated based on inflows (from rainfall and rivers) to the study area generated using the SWAT hydrological model (Neitsch et al. 2009; Arnold et al. 1998; Arnold et al. 2009a, b). The SWAT model is a physically based, con nuous simula on model developed for watershed assessment of short and long-term hydrology and water quality. It is a widely used catchment-scale model that can predict the impacts of land management prac ces (human ac vi es) and climate change over me on water, sediment, nutrient and pes cide yields with reasonable accuracy in large, ungauged river basins (Santhi et al. 2001). The model was calibrated and validated on daily and monthly scales for Amtali, Kalapara and Patharghata Upazilas (Figures 4a, b) using observed climate data (daily precipita on, maximum and minimum air temperature, wind speed, rela ve humidity and solar radia on) and hydrological data (watershed boundaries, river network, land use, flow rates). This part of the study area is situated in the coastal region and there are no observed discharge data to calibrate the model. For that reason, the model was calibrated against the monthly observed groundwater level at Patharghata sta on (Id BAG 005) for the period of 1986 to 1990 and validated for the period of 1981 to 1985. The model was also validated against the observed groundwater level at Amtali sta on (Id BAG002) for the period of 1981 to 1985. The Nash-Sutcliffe Efficiency (NSE) value for the calibra on period was 0.54, which is in the range of “adequate” (Rossi et al. 2008). The values of mean rela ve bias (PBIAS), RSR (ra o of the root mean square error to the devia on) and coefficient of determina on (R2) were 14.15, 0.67 and 0.73, respec vely, indica ng sa sfactory predic ve capability of the calibrated model against the calibra on data set. The NSE values for the valida on period were 0.27 and 0.36 for Patharghata and Amtali sta on, respec vely, and the correla on coefficients were 0.56 and 0.67, respec vely. These results are acceptable, considering the simplicity of the SWAT soil water model with respect to other models (CEGIS, 2013). The calibrated and validated SWAT model was run for baseline condi ons to simulate the temporal and spa al distribu on of surface water yields in the study areas. The model was then run using future climate predic ons to es mate the impacts of climate change on fresh water availability. 43 Ba se Map (Saline Prone Are a) Upazila : Patha rghata , Amtali, Kalapara Legend Elevation mPWD Bay of Beng al Hi gh : 3 m Legend Lo w: 1 m Bay of Bengal Fig. 4a. Base map of the study area. Gro wth Cente r Gro und Wate r Well Dischar ge stat io n Water level statio n Th ana b ound ary Amtali Kalapar a Pathar ghat a Water bodies Nation al roa d Region al roa d Fe eder roa d-A Fe eder roa d-B Rail line Emban kment River N km 0 1.5 3 6 Scale : 1:184, 302 Fig. 4b. Land eleva on of the study area. 2.2.3 Irriga on requirement and crop yield Net irriga on requirement and crop yield reduc on as a result of water deficit were es mated using the Drought Assessment (DRAS) model (Figure 5) (Hossain et al. 2008). This was done separately for pre-monsoon (aus), monsoon (aman) and winter rice (boro) crops in the study areas under baseline and A1B and A2 climate change scenarios. The DRAS model uses three types of data: i) climate data, ii) soil data, and (iii) crop and management data. Thirty years (1981-2010) of climate data from the Bangladesh Meteorological Department for Khepupara in Patuakhali District were used. Poten al evapora on (ETo) was calculated using ETo Calculator Version 3.2 (Raes 2012). Decadal ETo was used to determine maximum ET of a crop and to calculate actual ET under an exis ng soil moisture regime. 44 Fig. 5. Structure of the irriga on water demand and yield reduc on model. 1 Reference evapo-transpira on (ETo) is calculated separately by ETo Calculator Version 3.2 (Raes 2012) using climate data 2 Parameters are applicable for transplanted rice crops 3. Results and discussion 3.1 Climate analysis Mean annual temperature at Khepupara during 1971 to 2010 was around 26.2° C. Monthly mean minimum temperature was least in January (13.7° C) while monthly mean maximum temperature was highest in April and May (32.7° C). The highest recorded temperature was 38.1° C and the lowest was 8.5° C (CEGIS 2013). Under both the A1B and A2 climate change scenarios, annual mean temperature of the study region is predicted to be higher during each of the 30-year periods from 2011 to 2040, 2041 to 2070 and 2071 to 2100 than during the base period (1971 to 2010). Increases of 1 to 3° C are predicted under A1B, and of 0.6 to 2.5° C under A2 (Figures 6-8). However, the difference will vary depending on the me period, season and loca on. Daily average maximum temperature from April to June during the baseline period was more than 32° C (Figure 9), above the cri cal values for flowering of wheat, pea, mustard and tomato (Table 3). With climate change, maximum temperature will be more than 35° C from March to May by 2050 (IPCC 2013), indica ng unsuitability for cul va on of those crops unless heat-tolerant varie es are developed and disseminated, or the cropping window is changed so that flowering occurs at lower temperatures. 45 Mean temperature – Range of temperature 40 35 30 25 20 15 10 Time (years) Fig. 6. Seasonal temperature varia on in Amtali Upazila under baseline and future climate scenarios. Mean temperature – Range of temperature 40 35 30 25 20 15 10 1971-2010 2011-2040 2041-2070 2071-2100 2011-2040 2041-2070 20712100 Base Period A1B Scenario A2 Scenario Time (years) Fig. 7. Seasonal temperature varia on in Kalapara Upazila under baseline and future climate scenarios. Mean temperature – Range of temperature 40 35 30 25 20 15 10 1971-2010 2011-2040 2041-2070 2071-2100 2011-2040 2041-2070 2071-2100 Base Period A1B Scenario A2 Scenario Time (years) Fig. 8. Seasonal temperature varia on in Parthaghata Upazila under baseline and future climate scenarios. 46 Table 3. Cri cal temperature for sterility at flowering stage in selected field crops Name of crop Cri cal temperature for sterility (°C) Reference Rice (Oryza sa va L.) 35 BRRI 2011 Wheat (Tri chum aes vum L.) 30 Saini and Aspinal 1982 Chickpea (Cicer arie num L.) 35 Devasirvatham et al. 2012 Pea (Pisum sa vum) 30 McDonald and Paulsen 1997 Groundnut (Arachis hypogaea L.) 35 Prasad et al. 1999 Mustard (Brassica juncea L.) 27 Chauhan et al. 2013 Mung bean (Vigna radiata L.) 40 Tickoo et al. 19996 Tomato (Lycopersicon esculentum L.) 32 Peet et al. 1998 Tmax Tmin Tmax+1.8 Tmin+1.8 Temperature, °C and rainfall, mm Rainfall Tmax+3.4 Fig. 9. Mean daily average maximum and minimum temperatures and rainfall (1983-2012 at Khepupara), and the effects of temperature increases of 1.8 and 3.4°C and 3% precipita on increase. The predicted effects of climate change on rainfall are variable over me and across scenarios and upazilas. In Amtali, annual mean precipita on is expected to be 3% higher in 2011 to 2040 than the baseline under A1B scenario, but to be 3% lower from 2041 to 2100. In the A2 scenario, expected mean annual precipita on will be 9 to 11% lower from 2041 to 2070 than during the base period (Figure 10). In Kalapara Upazila the overall trend will be similar to that in Amtali Upazila. However, annual mean precipita on in Patharghata Upazila is predicted to decline in all three future periods in both scenarios except for A1B from 2071 to 2100. The predicted decrease from the base period is highest (about 11%) from 2041 to 2070 under A1B. Predic ons of the effects of climate change on rainfall are inconsistent. Hussain (2011) reported increased rainfall in various regions of Bangladesh, including Barisal, during March through November in both 2050 and 2070 based on the Geophysical Fluid Dynamics Laboratory Transient (GFDL-TR) Global circula on model, but decreased rainfall in the same period based on the Hadley Center (HadCM2) Global circula on model. Rahman et al. (2012) reported increased rainfall by 107% in the dry season (December to February) using the Regional Climate Model version 3 (RegCM3), but Hussain (2011) predicted variable results in the dry season. Increased winter rainfall could hamper establishment of non-rice winter crops in southern Bangladesh due to excessive soil moisture and delayed recession of flood water. However, such a scenario would reduce irriga on cost for boro rice cul va on. 47 A1B 2011-2040 A2 2041-2070 A2 2011-2040 A1B 2071-2100 A1B 2041-2070 A2 2071-2100 Fig. 10. Percentage change in annual mean precipita on in Amtali, Kalapara and Patharghata Upazilas considering A1B and A2 scenarios. 3.2 Predicted water availability There are many natural canals (khals) crisscrossing Barisal region, which give the opportunity to irrigate using surface water. Tidal river water flows into the canals twice a day, with peak des during the full and dark moons. As a consequence, surface water availability for crop produc on is sufficient in Barisal region if properly managed. However, a er the middle of February, river water salinity increases due to increasing salinity (Miah 2010). The increase in salinity varies with loca on and is higher towards the coast – in much of Barisal the river water is of low salinity throughout the year (Khan et al., these proceedings), but this is not the case in the southern-most regions of Barisal. The problem of waterlogging during the rainy season is also increasing in some loca ons due to silta on of the khals, which also serve as drainage outlets. The SWAT model was used to assess surface water availability under the A1B and A2 scenarios for the 2011 to 2040, 2041 to 2070 and 2071 to 2100 periods. Water availability trends are similar for all three study upazilas, so only the results of Patharghata are presented here. Annual water availability is predicted to decrease for both scenarios throughout the 21st century, except for A1B in 2011 to 2040 and A2 in 2071 to 2100 (Figure 11), however, the impact on availability within seasons varies. The reduc on in water availability will be highest (12 to 30%) during the pre-monsoon period. Water availability will be reduced by 2 to 20% and 3 to 15% during the post-monsoon period and dry season, respec vely. During the monsoon (Jun-Sep), predicted percentage changes in water availability are generally small. Predicted changes in total annual freshwater availability are also small for scenario A1B, while availability is predicted to be 10% lower in scenario A2 in 2040 to 2070, but 7% higher in 2070 to 2100. Fig. 11. Change in seasonal and annual surface water availability for A1B and A2 climate change scenarios of Patharghata Upazila, Barguna District. Values in the upper part of the figure are surface water availability during the base period (1981-2010). Note: 2020s = 2010-2040; 2050s = 2040-2070; 2080s = 2070-2100 48 3.3 Predicted grain yield and water requirement Grain yield of rainfed T. Aus is predicted to be 3 to 7% higher in all scenarios for most of the 21st century, apart from 2070 to 2100 in scenario A1B (Table 3). The yield is associated with increased rainfall by up to 15% under A1B and 12% under A2 scenarios throughout the 21st century in the T. Aus growing season. In contrast, yield of rainfed T. Aman rice yield is predicted to be 2 to 19% lower under the two scenarios throughout the 21st century, and this is associated with a decrease in rainfall of up to 2% under A1B and 10% under A2 scenarios together with an increase in evapora on rate of up to 12% under A1B and 8% under A2 scenarios during the T. Aman growing period. Table 3. Yield of rainfed rice under different climate change scenarios in the study region Crop name % yield increase (+) or decrease (-) from base scenario 2010-2040 2040-2070 2070-2100 A1B A2 A1B A2 A1B A2 T. Aus +(6 to 7) +(6 to 7) +(3 to 4) +(6 to 7) +(0 to 1) +(6 to 7) T. Aman -(8 to 11) -(5 to 7) -(3 to 10) -(13 to 19) -(4 to 12) -(2 to 8) The irriga on water demand of T. Aus is predicted to decrease by 8 to 111 mm compared to the base period under both the A1B and A2 scenarios, with the biggest difference for the 2011 to 2040 period (Table 4) due to increased rainfall in the T. Aus growing season. The effect of climate change on irriga on water demand of T. Aman is smaller (increases of 7 to 49 mm) due to decreased rainfall and increased evapora on rate. Throughout the 21st century there is a large surplus of surface water available for both the T. Aus and T. Aman crops. This implies that improving infrastructure for irriga on could mi gate seasonal droughts. In contrast to the situa on for T. Aus and T. Aman, there is always a large surface water deficit for boro rice cul va on in the study sites. The effect of climate change on irriga on demand for boro is variable – small increases with scenario A1B, and slightly larger decreases with scenario A2 – and effects on surface water availability are very small. Table 4. Comparison of irriga on water demand and surface water availability (SWA, mm) at Amtali Upazila, Barguna District Scenarios Year A1B A2 T. Aus T. Aman Boro Surplus/Deficit Demand SWA Demand SWA Demand SWA T. Aus T. Aman Boro Base 117 833 97 473 881 58 716 376 -823 2020s 39 950 118 486 898 55 911 368 -843 2050s 74 802 146 469 890 55 728 323 -835 2080s 109 823 116 475 919 51 714 359 -868 2020s 6 772 107 388 824 65 766 281 -759 2050s 24 762 120 402 820 51 738 282 -769 2080s 31 903 104 508 859 48 872 404 -811 Note: Base year = 1981-2010; SWA = Surface Water Availability; 2020s = 2010-2040; 2050s = 2040-2070; 2080s = 2070-2100 49 Es mated net irriga on water requirement (NIR) for T. Aus, T. Aman and boro crops is 100 to 120 mm, 80 to 125 mm and 830 to 880 mm, respec vely, in the base period (Table 5). Irriga on water requirement is predicted to decrease under both scenarios for T. Aus, with reduc ons usually in the range of 60 to 100 mm. In contrast, Rahman et al. (2012) reported increased irriga on water requirements for T. Aus cul va on. There were generally small increases in NIR for T. Aman under both the scenarios. However, NIR would s ll be low because of high rainfall during the aman crop (Rahman et al. 2012). The predicted effect of climate change on NIR for boro is generally small but highly variable across scenarios and me periods during the 21st century. Table 5. Net irriga on requirement (NIR) and effect of climate change on NIR in the study region Crop Name Base year Change of NIR (mm) (1981-2010) 2010-2040 2040-2070 A1B A2 A1B 2070-2100 A2 A1B A2 -(80 to 90) -(3 to 8) -(70 to 90) T. Aus 100-120 -(60 to 80) -(90 to 110) -(25 to 40) T. Aman 80-125 +(20 to 30) +(10 to 30) +(50 to 70) +(20 to 40) +(20 to 25) +(10 to 30) Boro 830-880 +(10 to 20) -(40 to 60) +(10 to 30) -(40 to 60) +(35 to 40) -(10 to 30) Note: (+) increased and (-) decreased in comparison with the base year 4. Conclusions and recommenda ons In southern Barisal Division, Bangladesh, temperatures are predicted to increase by 1 to 3° C under IPCC (2013) climate change scenario A1B, and by 0.6 to 2.5° C under scenario A2, during most of the 21st century, in comparison with the baseline of 1981 to 2010. In the future, daily maximum temperature of 35° C or more will prevail from March to May, with poten al adverse effects on wheat and mustard during the flowering and grain filling periods. Annual mean precipita on is generally predicted to decrease, but predic ons are variable across scenarios and me periods, with changes ranging from +3% to -11%. A small increase in the yield of rainfed T. Aus is predicted. However, grain yield of rainfed T. Aman is predicted to decrease. While availability of surface water is predicted to decrease during the pre-monsoon period, water availability would s ll be more than adequate for irriga on of both T. Aus and T. Aman rice. But there will con nue to be insufficient surface water available for irriga on of boro. Net irriga on requirement is predicted to be lower for T. Aus and slightly higher for T. Aman under both the A1B and A2 scenarios from 2011 to 2040 and 2041 to 2070 compared with the baseline. The effects of climate change on NIR for boro are variable and rela vely small (only a small frac on of NIR). Further studies are needed on the effects of climate change on fresh water availability taking salinity into account, on rainfed crop response taking into account soil salinity as well as soil moisture, and on poten al adapta on strategies to increase crop produc on in the future climate. Acknowledgements The authors acknowledge the “Climate Change Impacts Vulnerability and Adapta on: Sustaining Rice Produc on in Bangladesh” project funded by the Ministry of Foreign Affairs, Norway through the Royal Norwegian Embassy, Dhaka for funding the research in collabora on with BRRI, CEGIS and Bioforsk. References Arnold J. G., R. Srinivasan, R. S. Mu ah and J. R. Williams. 1998. Large area hydrologic modeling and assessment: Part I. Model development. J. American Water Resources Assoc. 34(1): 73-89. Arnold J. G., S. L. Neitsch, J. R. Kiniry and J. R. Williams. 2009a. Soil and Water Assessment Tool – Theore cal Documenta on, Version 2009, Texas, USA. 50 Arnold J. G., J. R. Kiniry, R. Srinivasan, J. R. Williams, E. B. Haney and S. L. Neitsch. 2009b. Soil and Water Assessment Tool– Input/Output File Documenta on, Version 2009, Texas, USA. BOBLME (Bay of Bengal Large Marine Ecosystem Project). 2011. Country report on pollu on in the BOBLME – Bangladesh. BOBLME-2011-Ecology-01 BRRI (Bangladesh Rice Research Ins tute). 2011. Modern Rice Cul va on, 16th Edi on. Pub No 5. 72pp. CEGIS 2006. Impacts of Sea Level Rise on Land-use Suitability and Adapta on Op ons. Final Report, Submi ed to the Ministry of Environment and Forest, Government of Bangladesh and United Na ons Development Programme (UNDP), Dhaka, Bangladesh CEGIS 2013. Descrip on of physical and environmental parameters. CEGIS, Dhaka. CEGIS, 2013. Report on Water Availability Assessment using SWAT Model. CEGIS, Dhaka. Chauhan, J. S., M. L. Meena, M. Saini, D.R. Meena, M. Singh, S. S. Meena, and K. H. Singh. 2013. Heat stress effects on morpho-physiological characters of Indian mustard (Brassica juncea L.). Directorate of Rapeseed-Mustard Research, Sewar, Bharatpur 321 303, Rajasthan, India. E-mail: js_chau @yahoo.com Devasirvatham V., P. M. Gaur, N. Mallikarjuna, R. N. Tokachichu, R. M. Trethowan, and D. K. Y. Tan. 2012. Func onal Plant Biol., 39(12): 1009-1018 Dewan A. M. and K. Nizamuddin. 1998. A geographical Analysis of Salinity Intrusion in the South-West Region of Bangladesh. In Geographical paper No. 17. Hossain, M. A., M. A. Haque, M. A. Ibrahim, E. H. Chaoudhury and M. A. Rashid. 2008. Bangladesh J of Agril. Engg. 19(1&2): 19-26. Hussain, S. G. 2011. Assessing impact of climate change on cereal produc on and food security in Bangladesh. Pages 459-476. In R. Lal et al (ed) Climate Change and Food Security in South Asia. Springer Science+Business Media B. V. 2011. IPCC. 2013. Climate Change 2013: The Physical Science Basis. Summary for Policymakers. Contribu on of the Working Group I to the Fi h Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. Jones, R. G., M. Noguer, D. C. Hassell, D. Hudson, S. S. Wilson, G. J. Jenkins and J. F. B. Mitchell. 2004. Genera ng high resolu on climate change scenarios using PRECIS, Met Office Hadley Centre, Exeter, UK, pp. 40. McDonald, G. K. and G. M. Paulsen. 1997. High temperature effects on photosynthesis and water rela ons of grain legumes. Plant Soil, 196, 47–58. Miah M. M. U. 2010. Final report of UNEP/APFED project. Field Tes ng of Innova ve Farming Prac ces Related to Climate Change in the Vulnerable Areas of Bangladesh. BCAS and UNEP, Dhaka, Bangladesh. 140p. Miyan. M. A. 2009. Monsoon in Bangladesh; Changes and Adapta on: Asian Monsoon Years AMY-6, Kunming China November 30-December 01, 2009 Mondal, M. K., S. P. Ritu, M. H. K. Choudhury, A. M. Chasi, P. K. Majumdar, M. M. Islam and S. K. Adhikari. 2004. Performance of high yielding varie es of transplanted Aman rie in southwest coastal region of Bangladesh. 19-27. Proc. worksho: coastal water management and uptake technologies. IRRI and PETRRA. Neitsch S. L., J. G. Arnold, J. R. Kiniry, J. R. Williams. 2009. Soil and Water Assessment Tool Theore cal documenta on, Version 2009, Texas, USA. PDO-ICZMP. 2003a. Delinea on of the Coastal Zone. Project Development Office for Integrated Coastal Zone Management Plan. December 2003. 51 PDO-ICZMP (Program Development Office for Integrated Coastal Zone Management Plan). 2003b. Coastal Zone Management– an Analysis of Different Policy Documents; WP009, Dhaka Peet M. M., S. Sato, and R. G. Gardner. 1998. Comparing heat stress effects on male-fer le and male-sterile tomatoes. Plant, Cell & Environ., 21: 225–231. Prasad, P. V., P. Q. Craufurd, and R. J. Summerfield. 1999. Fruit Number in Rela on to Pollen Produc on and Viability in Groundnut Exposed to Short Episodes of Heat Stress. Annals Bot., 84: 381-386. Raes, D. (2012) The ETo Calculator. Reference Manual Version 3.2. FAO, Rome, Italy Rahman, M.M., M.N. Islam, A.U. Ahmed, and F. Georgi, 2012. Rainfall and temperature scenarios for Bangladesh for the middle of 21st century using RegCM. J. Earth Syst. Sci. 121(2): 287-295. Rossi, C. G., Dybala, T. J., Moriasi, D. N., Arnold, J. G., Amone , C. and Marek, T. 2008. Hydrologic calibra on and valida on of the soil and water assessment tool for the Leon River watershed. Soil Water Conserv. 6, 533–541. Saini, H.S. and D. Aspinal. 1982. Abnormal sporogenesis in wheat (Tri cum aes vum L.) induced by short periods of high temperature. Ann. Bot., 49: 835-846. Santhi C., J. G. Arnold, J. R. Williams, W. A. Dugas, R. Srinivasan and L. M. Hauck. 2001. Valida on of the SWAT model on a large river basin with point and nonpoint sources. J. American Water Resources Assoc. 37(5): 1169-1188. SRDI. 1997.Agricultural ecology and salinity in coastal areas of Bangladesh. Soil Resources development Ins tute, Dhaka. Tickoo, J. L., Gajraj, R. Matho, and C. Manji. 1996. Plant type in mungbean (Vigna radiata (L.) Wilczek). In: Proceedings of Recent Advances in Mungbean Research, A.N. Asthana and D.H. Kim (eds.), Indian Society of Pulses Research and Development, Indian Ins tute of Pulses Research, Kanpur, India, pp: 197-213. 52 Groundwater salinity zoning for development plans: A case study of four sub-districts in the southwestern coastal region of Bangladesh M.R. Hasan, M. Shamsuddin, M.S. Masud and A.F.M.A. Hossain Ins tute of Water Modelling, Bangladesh, mrh@iwmbd.org, smd@iwmbd.org, msm@iwmbd.org, a @iwmbd.org Abstract The southwestern region of Bangladesh is part of the Ganges floodplain, with low relief and numerous rivers and channels linked with the Bay of Bengal. Surface water salinity is increasing due to reduced upstream water flow. As a result, dependence on groundwater has become vital. The risk of saliniza on of groundwater is greater in regions closer to the sea and the situa on can be further aggravated by anthropogenic influences (global warming, changes in trans-boundary river flows). For op mal use of groundwater resources in the coastal area a salinity zoning map would be a useful tool to help iden fy how to meet drinking water requirements as well as boost agricultural produc on. Therefore, a pilot study was undertaken to iden fy groundwater salinity zones in four upazilas of Khulna and Satkhira Districts in the southwestern coastal region of Bangladesh. Groundwater samples were collected in April 2013 from 245 randomly selected tubewells including 130 shallow hand tubewells (SHTW, screening depths from 35 to 120 m) and 115 deep hand tubewells (DHTW, 150 to 340 m). Chloride (Cl-) concentra on of the groundwater was used to develop groundwater salinity maps. The maps were fine-tuned using the findings of qualita ve surveys (focus group discussions and key informant interviews) with a range of stakeholders. Only 20% of the SHTW and 54% of the DHTW had fresh water suitable for drinking (Cl-< 600 ppm), while 86% of the shallow tubewells and 74% of the deep tubewells had water suitable for irriga on of rice (Cl-< 2,000 ppm). In Paikgacha, there was no fresh water (Cl-< 600 ppm) in DHTWs and only 4% of SHTWs had fresh water. Key message: Studies of groundwater quality across the coastal zone of Bangladesh are urgently needed. In a 92,000 ha pilot area in Khulna Division, only 20% of the study area had fresh water in the upper aquifer suitable for drinking, while the percentage suitable for drinking in the deep aquifer was 54%. Keywords: chloride, shallow aquifer, deep aquifer, Khulna, Satkhira 1. Introduc on Bangladesh has 710 km of coastline and the lands of the coastal zone cover about 32% of the country (MoWR 2005). Islam (2001) classified the coastal lands of Bangladesh into three dis nct regions, namely southeast, central and southwest. The southwest coastal region is part of the Ganges dal floodplain with low relief and is crisscrossed by rivers, dal marshes and swamps. Although groundwater is abundant in the region saline water intrusion into the upper aquifer system is increasing due to reduc on of upstream freshwater flows, shrimp farming and over-abstrac on of groundwater. The region has also been iden fied as one of the most vulnerable parts of the world due to increasing temperature and salinity and sea level rise caused by climate change (FAO 2008; Roy 2009). In the southwest region, the main focus of research to date has been on irriga on, drainage, flood control and diversion of river flows for irriga on during the dry season. Very few studies have been undertaken on groundwater salinity. Groundwater can be abstracted from aquifers at various depths. During the last four decades, thousands of hand tubewells were installed all over the country including the coastal region. The tubewells in the coastal area were generally installed at depths between 150 m and 350 m. In the Khulna region, Haskoning and Iwaco (1981) found that the upper aquifer contains brackish to saline water and that the fresh-saline groundwater interface lies at a depth between 200 and 300 m below ground level. Saline 53 pockets also occur in both shallow and deep aquifers due to the presence of paleo-brackish water that was entrapped in small areas during rapid regressive events that occurred between 12,000 and 10,000 years a er a transgression period, which occurred between 18,000 and 12,000 years ago (Acharyya 1999). In Bangladesh, the recommended chloride concentra on (Cl-) in drinking water is 150 to 600 ppm (ECR 1997). However, higher values (up to 1000 ppm) are considered acceptable in problem areas including the coastal belt (DPHE 2006). For rice produc on, a chloride concentra on of less than 2000 ppm is considered suitable (BARC 2013). The Department of Public Health Engineering (DPHE) engaged the Ins tute of Water Modelling (IWM) to conduct a study on salt water intrusion into groundwater in a pilot area of the southwestern coastal region. This included the delinea on of groundwater salinity zones through two approaches: hydro-chemical inves ga ons and par cipatory rapid appraisal (PRA). While hydro-chemical inves ga on indicates actual salinity of the groundwater, sociological indicators are also useful in understanding the impacts of salinity. The main objec ve of the study was to iden fy groundwater salinity zones for mee ng drinking water requirements and for other domes c uses. The study also sought to delineate poten al areas for agriculture and brackish water shrimp farming. 2. Methodology 2.1 Study area The study area is in Khulna Division and the boundary of the study area mainly follows the Kazibacha (local name of the Rupsha) and Kobadak rivers on the southern, eastern and western sides. The study area covers four upazilas (sub-districts): Dumuria, Ba aghata and Paikgacha Upazilas of Khulna District and Tala Upazila of Satkhira District, with a total area of about 918 km2 (91,800 ha) (Fig. 1). The northern part of the study area is congruent with the northern boundaries of the upazilas. Land use in the study area is dominated by agriculture. Rice (T. Aman) is the main crop grown in the area. Most of the lands remain fallow in the dry season as no crop can be grown due to salinity. According to WARPO (2000), about 0.37 million ha of arable land have been badly affected by salinity in the greater Khulna region. Bagda (shrimp, P. Monodon, grown under saline condi ons) farming is prac ced in parts of the study area. About 21,000, 7,500, 4,400 and 3,700 ha were under Bagda in Paikgacha, Dumuria, Ba aghata, and Tala Upazilas, respec vely, in 2005 (CEGIS 2006). 89 °1 0'0''E 89 °2 0'0''E 89 °3 0'0''E N 7 3.5 0 89 °4 0'0''E S tudy Area 7 K ilomet ers 89 °1 0'0''E Fig. 1. Loca on of the study area. 54 89 °2 0'0''E 89 °3 0'0''E 89 °4 0'0''E 2.2 Study design The study involved hydro-chemical inves ga ons and qualita ve surveys, which were conducted in April 2013. 2.2.1 Hydro-chemical inves ga ons Groundwater samples were collected from 130 shallow hand tube wells (SHTW) and from 115 deep hand tube wells (DHTW) (Figure 2, Table 1). The SHTWs had screen depths varying from 35 to 120 m below ground level (bgl), and the DHTW had screens at 150 to 340 m bgl. Groundwater pH, temperature, EC, total dissolved salts (TDS) and chloride concentra on were determined at each tube well based on rela onships with EC, determined using a field test kit (Hach Salinometer model EC5DL, a conduc vity meter with a minimum detec on level of 300 ppm). Water samples of eight wells were analyzed for EC and Cl- at laboratories of Bangladesh University of Engineering and Technology (BUET), Bangladesh Council of Scien fic and Industrial Research, and DPHE. BUET laboratory results are considered to be the standard benchmark in Bangladesh. The results of the DPHE laboratory analyses and BUET were consistent, so the DPHE laboratory was selected for carrying out laboratory analyses of 12% of the samples for the development of rela onships between EC and Cl- from EC, determined using the test kit. 89 °16 '0'' E 89 °23 '0'' E 89 °30 '0'' E 89 °16 '0'' E 89 °23 '0'' E 89 °30 '0'' E 89 °16 '0'' E 89 °16 '0'' E 89 °23 '0'' E 89 °23 '0'' E 89 °30 '0'' E 89 °30 '0'' E Fig. 2. Loca ons of ground water sampling sites: shallow aquifer (le ) and deep aquifer (right). Table 1. Distribu on of the sampled tube wells by upazila District Upazila No. of SHTW No. of DHTW Khulna Dumuria 46 41 Ba aghata 13 45 Paikgacha 47 5 Tala 24 24 130 115 Satkhira Total 55 2.2.2 Qualita ve survey A par cipatory rapid appraisal (PRA) including focus group discussions (FGD) and key informant interviews (KII) was carried out to collect qualita ve informa on and users’ percep ons and understanding about the tube wells and water quality. The FGDs were conducted in six unions with local level public representa ves and representa ves of a range of professional groups including farmers, businessmen, service holders, shopkeepers, teachers and housewives. Eight to twelve people par cipated in each of the FGD sessions. The KIIs were mainly held with staff of DPHE based in the study area. To verify the results of the qualita ve survey in rela on to groundwater salinity, especially in areas reported to be less saline, field kit tests were carried out at most of the PRA loca ons. Samples were taken from at least three tube wells at each loca on. Before taking the water samples, sufficient water was pumped out to ensure that the sample was representa ve of the surrounding groundwater. The informa on from the qualita ve social survey thus verified by field test kit was used to further inform the groundwater map delinea on. 2.3 Mapping groundwater salinity zones Both quan ta ve (i.e., hydro-chemical inves ga on) and qualita ve (i.e., FGD, KII) data were used for delinea ng groundwater zones according to salinity. Salinity zone maps were generated based on Cl- concentra on using ArcGIS 10.1 (inverse distance weight (IDW) method). Four groundwater salinity zones were delineated based on the suitability of the water for various purposes (Table 2). The qualita ve informa on was superimposed on the salinity distribu on map to improve the detail in delinea ng the groundwater salinity zones. Table 2. Classes of groundwater salinity and their suitability for various uses Classes (Cl-, ppm) Usage and limita ons < 600 Very good quality water; suitable for drinking, domes c, irriga on 600 - 1000 Good waterl suitable for domes c uses and irriga on 1000 - 2000 Suitable for irriga on > 2000 Saline-affected water 3. Results and discussion 3.1 Calibra on of the field test kit The rela onship between EC measured using the field test kit and in the BUET laboratory was strong and there was a near perfect polynomial rela onship (y = -0.00001* x2+1.061*x – 54.33, R2=0.999, where y is EC measured in the laboratory at BUET and x is EC measured using the field test kit) (Fig. 3a). There were also strong rela onships between EC measured using the field test kit and chloride concentra on in the water determined in both the BUET (R2=0.98) and DPHE (R2=0.96) laboratories. For EC > 4000 µS/cm the BUET equa on was used, while for lower EC values the DPHE equa on was used. a) b) 4500 4000 3500 3000 2500 2000 1500 1000 500 0 2,500 y = -1E-05x² + 1.061x - 54.33 R² = 0.999 y = 4E-05x ²+ 0.4 10x -214.8 R² = 0.957 2,000 1,500 y = 2E-05x² + 0.417x -2 06.5 R² = 0.9 82 1,000 500 - - 1,000 2,000 3,000 E C a t F IE L D (µS /c m ) 4,000 5,000 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 Electrical Conductivity, EC (µS/cm) BUET DPHE Fig. 3. Calibra on of the field test kit against laboratory values from BUET and DPHE: a) field kit EC vs BUET- EC, and; b) field kit EC vs chloride concentra on from BUET and DPHE. 56 3.2 Groundwater quality Around 50% of the SHTWs had chloride concentra ons exceeding 1,000 ppm (Table 3). However, in Tala Upazila, Cl- concentra on of the SHTWs was below 600 ppm and thus suitable for drinking at all loca ons. In the shallow aquifer of Paikgacha Upazila, there were some areas with low Cl- concentra on and other areas with very high values, indica ng that in some parts of this upazila the shallow aquifer is suitable for drinking water while in others it is not. This variability may be due to the fact that, historically, fresh river water used to flow down the Kobadak River, recharging the shallow aquifers of adjacent areas and reducing their salinity. Throughout Paikgacha Upazila Cl- concentra on of the deep aquifer exceeded 1,700 ppm and was generally unsuitable for irriga on. Table 3. Range (median) of groundwater parameters in shallow and deep aquifers pH EC (dS/m) Cl- (g/l) Upazila SHTW DHTW SHTW DHTW SHTW DHTW 7.2-8.2 7.5-8.6 0.6-6.1 1.8-4.3 0.3-2.1 0.3-5.8 Dumuria (7.6) (8.1) (2.5) (3.1) (0.95) (0.3) 7.2-8.3 7.5-8.7 1.1-8.5 0.7-6.0 0.3-3.4 0.3-1.3 Ba aghata (7.6) (8.3) (2.7) (1.1) (0.98) (0.3) 7.1-7.8 7.4-8.3 0.6-12.9 0.6-8.6 0.3-12.9 1.7-3.5 Paikgacha (7.5) (7.5) (3.6) (4.0) (1.5) (2.2) 7.2-7.7 7.3-8.7 0.5-1.6 0.6-7.6 0.3-0.5 0.3-2.9 Tala (7.5) (8.4) (0.9) (1.0) (0.3) (0.3) 1 TDS1 (g/l) SHTW 0.3-3.0 (1.3) 0.5-4.6 (1.4) 0.3-17.0 (1.9) 0.3-0.8 (0.5) DHTW 0.3-6.7 (0.4) 0.4-3.2 (0.6) 2.1-4.7 (2.1) 0.3-4.1 (0.5) TDS = total dissolved salts; 1 g/l = 1,000 ppm 3.3 Findings of the qualita ve surveys Changes in groundwater salinity over me were not perceived by the people consulted during FGDs, and tube well drinking water quality was considered sa sfactory by most (Table 4). Yet, at the same me, there was a strong percep on that salinity was causing dysentery and diseases, and that salinity was increasing and affec ng agriculture. In contrast with the opinions of the FGD par cipants, key informants stated that groundwater salinity has increased over the past few decades (Table 5). In some places in the south, suitable water from a deep aquifer was not available. In these places people used SHTWs for domes c purposes. In the dry season most of the agricultural land could not be brought under cul va on due to high salinity of the available water sources. However, there was limited boro cul va on in areas where suitable groundwater was available. Shrimp farming was prac ced throughout the area. In some fields, shrimp was raised in the dry months while paddy was grown in the wet season. Table 4. Percep ons of people interviewed during FGDs on groundwater quality and related issues Sl no. Descrip on Increment of salinity in groundwater Use of water for drinking Source of irriga on water Ownership of tube wells Impact on health People’s percep ons (% of people interviewed) Almost 100% did not perceive any increase in salinity. About 93% were comfortable drinking their tube well water. Groundwater-24.7%; pond-36.8%; river/canal-38.5% Government-49%; NGO-1%; private-50% Almost 100% stated that they suffered from dysentery and skin diseases due to salinity. Impact of climate change (sea level About 91% stated that agriculture was affected by increasing salinity rise, temperature and natural caused by sea level rise. disasters) More than 80% said that temperatures had been rising. About 26% stated that the frequency of cyclones had increased. 57 Table 5. Informa on on salinity of shallow and deep aquifers from the key informant interviews Upazila Descrip on Dumuria Major por on does not have good drinking water except western areas (Maguraghona, Atlia, Sobhana unions), some por ons of central and northern parts. Ba aghata Only the central por on of Ba aghata union has a freshwater aquifer. Paikgacha Paikgacha does not have fresh water for drinking, except for some sca ered pockets in the northern, northwestern and southwestern parts of the upazila. Tala The northeastern and southeastern parts of the study area do not have a good aquifer. The rest of the study area has fresh water Dumuria Most parts have fresh water except the southern part (Magurkhali and Sarafpur unions). Also some sca ered areas of Maguraghona, Rudaghara and A lia unions have saline aquifers. Ba aghata Most of Ba aghata has fresh water with sca ered pockets of salinity exis ng in the northern (Jalma union), southeastern (Gangarampur) and western parts (western half of Surkhali). Paikgacha There are hardly any freshwater aquifers. Tala Most of the northern part has a fresh water aquifer; there are sca ered freshwater pockets in the southern part. 3.4 Groundwater salinity zoning maps For the shallow aquifers, Cl- concentra on exceeded 1,000 ppm in two-thirds of the tube wells, covering the eastern side of the study area from north to south (Table 6, Fig. 4). For deep aquifers, 54% of the tubewells had Cl-<600 ppm and salinity was highest (Cl->1,000 ppm) in the southern part. Twenty-six percent of the DHTWs were highly saline (Cl- >2000 ppm). Overall, the deep aquifer was less saline than the shallow aquifer. But in Paikgacha there was no fresh water (Cl- <600 ppm) in DHTWs, and only 4% of SHTWs had fresh water. Table 6. Percent of SHTWs and DHTWs in different classes of salinity (expressed in Cl- concentra on) in different upazilas. Upazila 600 ppm 600-1000 ppm 1000-2000 ppm >2000 ppm Total SHTW DHTW SHTW DHTW SHTW DHTW SHTW Dumuria 6 37 11 3 28 1 5 9 50 Ba aghata 0 12 1 1 16 4 1 1 18 Paikgacha 4 0 0 1 7 5 8 13 19 Tala 10 5 0 4 3 1 0 3 13 Total 20 54 12 9 54 11 14 26 100 58 DHTW SHTW + DHTW 89°10'0''E 89°20'0''E Cl (mg/ l) Map Area April , 201 3 Upazila Boun dary 600 - 1, 000 Union Bou ndary N Proje ction : BTM , Evere st 18 30 4 1, 000-2,000 2 89°30'0''E 0 Cl (mg/ l) Map Area April , 201 3 Lege nd River < 600 89°20'0''E Groundwater Salinity Zoning (Shallow Aquifer) Groundwater Salinity Zoning (Deep Aquifer) Lege nd 89°10'0''E 89°30'0''E River < 600 Upazila Boun dary 600 - 1, 000 Union Bou ndary N Proje ction : BTM , Evere st 18 30 4 2 0 4 km 1, 000-2,000 > 2,000 > 2,000 [GIS7 6] D:\ P00 24 \51 21 2 JAR\Mxd Fi le \So ci al _Su rvey _Ma p_ Upa zi la s\4Up az il as _Soc ia l_ Surve y_ A4 _P_ De ep _r2 _v 2.mxd [*eu z*21 /09 /2 01 4] [GIS7 6] D:\ P00 24 \51 21 2 JAR\Mxd Fi l e\Soci a l Surve y Ma p Upa zi la s\4Up az il as Soc ia l Surve y A4 P Sha l ow r2 v2 .mxd [*eu z*2 1/09 /201 4] Fig. 4. Groundwater salinity map based on chloride concentra on - shallow aquifer (le ) and deep aquifer (right). Different colors designate different classes of chloride concentra on: green<600 ppm; yellow 600-1000 ppm, orange 1000-2000 ppm and red>2000 ppm. 4. Conclusions and recommenda ons This study revealed that the salinity of deep aquifers was less than that of shallow aquifers in most of the study area, except in Paikgacha Upazila. In general, Dumuria Upazila had be er groundwater quality than the other upazilas. Only 20% of the study area had fresh water in the upper aquifer suitable for drinking, while the percentage suitable for drinking in the deep aquifer was 54%. About 14% of SHTWs and 26% of DHTWs in the study areas did not have water suitable for rice cul va on. The study covered a limited area and was a short dura on pilot. Similar studies are needed throughout the coastal zone to determine groundwater quality. A long-term groundwater monitoring strategy is also needed to iden fy changes in water quality over me. These studied are needed to inform planning and development ac vi es. Acknowledgements The authors are highly grateful to Department of Public Health Engineering (DPHE) to engage the Ins tute of Water Modelling (IWM) to carry out the study. The authors would like to thank all the field staff of IWM at the Khulna Project office who provided data required for carrying out the study. 59 References Acharyya, S.K., Chakraborty, P., Lahiri, S., Raymahashay, B.C., Guha, S., Bhowmik, A. 1999. Arsenic poisoning in the Ganges delta. Nature 401:545. BARC. 2013. Suitable Agricultural Technology for Southwest Bangladesh. Bangladesh Agricultural Research Council, Ministry of Agriculture, Government of the People’s Republic of Bangladesh. (In Bengali) BGS and DPHE. 2001. Arsenic Contamina on of Groundwater in Bangladesh. Kinniburgh, D.G. and Smedley, P. L. (Editors), Vol 2, Final Report, BGS Technical Report WC/00/19, Bri sh Geological Survey, Keyworth, United Kingdom. CEGIS 2006. Impact of Sea Level Rise on Landuse Stability and Adapta on Op ons, BGD/96/007-Sustainable Environment Management Programme, Component 1.4.3. Ministry of Environment and Forest, Government of the People’s Republic of Bangladesh. DPHE 2006. Development of Deep Aquifer Database and Preliminary Deep Aquifer Map, Final Report. Ministry of LGRD and Coopera ves, Government of the People’s Republic of Bangladesh. ECR 1997. Environmental Conserva on Rules, Bangladesh Gaze e. Ministry of Environment and Forest, Government of the People’s Republic of Bangladesh. FAO 2008. Community Based Adapta on in Ac on: A case study from Bangladesh. Food and Agriculture Organiza on of UN, Rome. Haskoning & Iwaco B.V.. 1981. Khulna Water Supply Project - Feasibility Study, Final Report, Vol 3. Department of Public Health Engineering, Dhaka, Bangladesh. Islam, M.S. 2001. Sea-level Changes in Bangladesh: The Last Ten Thousand Years. Asia c Society of Bangladesh. MoWR 2005. Coastal Zone Policy (CZPo). Ministry of Water Resources (MoWR), Government of the People’s Republic of Bangladesh. Roy, K. 2009. Future Climate Change and Moisture Stress: Impact on Crop Agriculture in South-Western Bangladesh. vol: 1 issue:1 Unnayan Onnesion Khulna. WARPO 2000. Na onal Water Management Plan (NWMP), Dra Development Strategy, Volume-4, Annex-C, Land and Water Resources. Water Resources Planning Organiza on (WARPO), MoWR, Government of the People’s Republic of Bangladesh. 60 Effect of groundwater use on groundwater salinity, piezometric level and boro rice yield in the Sundarbans of West Bengal D. Burman1, K.K. Mahanta1, S.K. Sarangi1, S. Mandal1, B. Maji1, U. K. Mandal1, B.K. Bandyopadhyay1, E. Humphreys2 and D. K. Sharma1 1 ICAR-Central Soil Salinity Research Ins tute, India, burman.d@gmail.com, mahantakk@gmail.com, sksarangicanning@gmail.com, subhasis2006@gmail.com, b.maji57@gmail.com, u am_icar@yahoo.com, bimalbkb@gmail.com, dksharma@cssri.ernet.in 2 Interna onal Rice Research Ins tute, Philippines, e.humphreys@irri.org Abstract Agriculture in the Sundarbans of India is mainly mono-cropped with kharif rice (aman) grown in the monsoon season. Scarcity of fresh surface water for irriga on during the post-monsoon period is the main constraint on growing boro (post monsoon/dry season) rice and other crops. Therefore, there is increasing exploita on of groundwater through shallow tube wells (STWs). At present, STWs are used to irrigate about 32% of the total irrigated area of approximately 55,300 ha in the Sundarbans. However, there is concern about the sustainability of current rates of groundwater use and its con nued expansion. Therefore a series of surveys was conducted on the status of STWs in South 24 Parganas and North 24 Parganas districts from 2000 to 2014. The depth of STWs varied from 67 to 128 m below ground level (bgl). A study throughout 2003 showed that the piezometric pressure level was highest and salinity was least a er the monsoon (November-December) and that piezometric level was least and salinity highest before the onset of the monsoon (May-June). The piezometric level and water quality recovered during the 2003 monsoon. Detailed study of STWs used for irriga ng boro rice in 2003, 2009 and 2014 showed that the piezometric level decreased significantly from an average of 2.1 m bgl in January to 3.9 m bgl in May while the discharge rate decreased significantly from 10.8 to 7.2 ls-1. Over the same period, groundwater salinity increased significantly from an average of 2.0 to 3.7 dS m-1. Salinity (ECe) of the topsoil (0-15cm) in boro rice fields irrigated from STWs increased significantly from an average of 3.0 dS m-1 at sowing to 5.8 dS m-1 at harvest. The grain yield of boro rice increased significantly with increasing installa on depth of STWs. STW depth accounted for 70% varia on in grain yield and this was at least partly due to a significant decrease in water salinity with increasing STW depth. While the piezometric level and water quality recovered during the 2003 monsoon, whether this would be the case in more intensively irrigated areas is not known. The sustainable level of groundwater use needs to be determined across the region. Measures for more judicious use of groundwater are also needed to increase produc vity of this precious resource. Key message: There is an urgent need to be er characterize the aquifers of the Indian Sundarbans and to develop a plan for the sustainable use of groundwater for irriga on. Keywords: shallow tube well, soil salinity, coastal area, 24 Parganas, India 1. Introduc on The coastal region in the delta of the river Ganges in India and Bangladesh is known as the ‘Sundarbans’. The Indian Sundarbans lies between 21032’and 21040’ N and between 88005’and 89000’ E. It comprises 102 islands, of which 54 are inhabited and spread across 19 blocks of the two southernmost districts of West Bengal, namely North 24 Parganas and South 24 Parganas. Out of 19 blocks, 13 are in South 24 Parganas and six are in North 24 Parganas. This region is one of the most disadvantaged and poverty stricken areas in India. Farmers are mostly very resource poor marginal and smallholder farmers with highly fragmented land holdings (Mandal et al. 2011). The produc vity of the land is very low due to soil and water salinity. 61 Agriculture in the Indian Sundarbans is mainly mono-cropped with kharif (aman) rice grown in the monsoon season. Although the region receives high rainfall (about 1800 mm per annum), there is an acute shortage of irriga on water during the dry season as more than 80% of the rainfall occurs from July to September. Most of the croplands remain fallow during the post-monsoon period due to scarcity of fresh surface water for irriga on (Burman et al. 2013). Therefore, farmers in the Sundarbans resort to pumping groundwater using shallow tube wells (STWs) for growing crops (mostly boro rice) during the post-monsoon period. Boro is the preferred crop as its yield (3 t ha-1) is higher than that of rice grown during the monsoon season (2 t ha-1) or during the pre-monsoon rains (aus, 2.5 t ha-1). In the coastal region of West Bengal, alluvium of Recent to Pleistocene ages and Ter ary sediments form aquifers that are in the general depth range of 70-360 m bgl and are mostly confined (CGWB 2014). Different aquifer systems (shallow to deep) are separated from each other by thick layers of clay. Each aquifer system comprises one or several interconnected aquifers. Both fresh and brackish/saline water-bearing aquifers occur, with fresh groundwater mostly found in the deeper aquifers (Ray and Shekhar 2009; Gayan2009; Misra and Nag 2009; CGWB 2014). Farmers in the Sundarbans pump ground water from shallow aquifers through STWs to grow boro rice, as extrac ng water from deeper aquifers is not economic. The number of STWs installed for irriga ng boro rice has been increasing over me (Mahanta and Burman 2004). But some farmers report loss of boro rice yield due to the high salinity of irriga on water. Informa on on the status and use of groundwater through STWs in the Sundarbans is meager. Informa on on the dynamics, in me and space, of the quality and quan ty of groundwater from STWs is needed to help understand the sustainability of the use of STWs for irriga on, and in par cular for boro rice because of its high irriga on water requirement. The present paper deals with the expansion of STWs and the effects of groundwater pumping on water quality, piezometric level and produc vity of boro rice in the Sundarbans of West Bengal. 2. Methods Surveys were conducted on the status of STWs in 2000-01, 2002-03, 2008-09 and 2013-14 in 16 villages in Canning-I block and one village in Basan block in South 24 Parganas District, and in one village in Sandeshkhali-I block and two villages in Sandeshkhali-II block in North 24 Parganas District. About 150 farmers were interviewed to collect informa on on the depth of installa on of STWs, crops cul vated and yields, and on their percep ons of groundwater salinity. The survey team measured salinity of the groundwater and piezometric levels. 2.1 Study 1 During 2003, five STWs were selected in Dumki village for monthly determina on of the piezometric level and salinity of the groundwater over a period of 12 months. 2.2 Study 2 In a separate study, the discharge rate, piezometric level and water salinity of 44 STWs in Dumki, Nikarighata, Chandkhali, Kathalberia, Simulha , Daspara villages were measured at the beginning (January) and end (May) of the boro seasons in 2003, 2009 and 2014. The salinity of the topsoil (0-15cm) (EC of the satura on extract, ECe) in the farmers’ fields irrigated from those STWs was also monitored during the boro season. Different STWs were monitored in different years. 2.3 Study 3 During 2014, 30 farmers’ fields were selected in Dumki, Nikarighata, Chandkhali, Kathalberia, Simulha , and Daspara villages where similar boro rice varie es (Lalminikit, WGL20471) were grown. The grain yield of boro rice and salinity of the surface soil (ECe, 0-15cm) at harvest were determined. 62 2.4 Monitoring Water and soil salini es were measured using a conduc vity meter and piezometric pressure level (depth below the soil surface) was determined using an electronic water level meter. The discharge rate of the STWs was measured by the coordinate method (Murty 1985). 3. Results and discussion 3.1 Status of STWs in the Sundarbans The number of STWs in the Sundarbans roughly tripled from about 2,200 in 1994 to 7,656 in 2002 and has con nued to increase, but at a slower rate, since then (Fig. 1). The area irrigated by this source roughly doubled from about 9,000 to 18,000 ha between 1994 and 2012. The slower rate of increase in STW installa on in recent years was mainly due to increases in the costs of diesel and other inputs (Mahanta and Burman 2004), making the cul va on of boro rice less profitable. At present, about 32% of the total irrigated area of approximately 55,300 ha in the Sundarbans is irrigated using STWs (BAES 2012). On average, about 2.34 ha (17.4 bigha) are irrigated by each tube well. 20000 15000 10000 5000 0 1994 1998 2002 Year No. 2007 2012 Area Fig.1. Number of STWs and area under STW irriga on during the last two decades in the Sundarbans of West Bengal (source: Bureau of Applied Economics and Sta s cs (BAES), Govt. of West Bengal, 1994, 1998, 2002 & 2012). In the study area, the depth of the STWs varied from 67 to 128 m below ground level (bgl). The tube wells were installed in low, medium and high lands according to the land owned by the farmers and ability to irrigate the adjoining lands. All STWs were constructed using PVC pipe of diameter 7.6 cm, a strainer and a 5 horsepower diesel pump. Most strainers were made of PVC, some were bamboo, and a few were made from aluminum. The length of the strainers varied from 9.1 to 18.3 m with an average of 14.2 m. 3.2 Piezometric level, discharge rate, water and soil salinity 3.2.1 Study 1 The depth of the piezometric level and water salinity increased from January to May, just before the onset of the monsoon, which occurred in mid-June in 2003 (Fig. 2). The depth decreased steadily over the next few months to its lowest values in December. At the same me, salinity decreased sharply between May and June, and then more slowly to its lowest values in November and December (Fig. 3). The piezometric level and salinity in all STWs recovered to their original values at the start of the boro season (December). Total monthly rainfall in 2003 was similar to the long-term averages, except for unusually high rainfall in October (Fig. 4), which occurred a er the piezometric levels and water salinity were measured in October. 63 600 4.5 4 500 3.5 3 400 2.5 300 2 1.5 200 1 100 0.5 0 0 Rainfall STW1 STW2 Fig. 2. Monthly rainfall in 2003 and monthly depth (below ground level) of the piezometric level in selected STWs in Dumki village. 4 600 3.5 500 3 400 2.5 300 2 200 1.5 100 1 0 Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep Oct Nov. Dec. Month Rainfall STW1 STW2 STW3 STW4 STW5 Fig. 3. Monthly rainfall in 2003 and monthly water salinity in selected STWs in Dumki Village. 600 500 400 300 200 1966-2013 2003 100 0 Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec. Fig. 4. Monthly rainfall in 2003 and average monthly rainfall from 1966-2013 at Dumki village. 64 3.2.2 Study 2 The 44 monitored STWs were installed at depths ranging from 73 to 116 m. The piezometric level and discharge rate decreased significantly (P<0.01) from January to May while salinity of the groundwater increased significantly during that period (Table 1). There was significant rela onship (P<0.05) between piezometric level and discharge rate in both January and May, with piezometric level accoun ng for 36% of the varia on in discharge rate in both months. The decrease in piezometric level and the decrease in discharge rate from January to May were also related significantly, however the decrease in piezometric level only accounted for 15% of the decrease in discharge rate. The increase in salinity of the groundwater suggests contamina on of the aquifer from other sources (e.g. naviga on canals, estuaries, saline aquifers) as a result of the decrease in pressure level. The rela onships between STW depth and ini al (January) or final (May) water salinity were significant. STW depth accounted for 19% and 25% of the varia on in ini al and final water salinity, respec vely, with higher salinity in shallower STWs. There was no rela onship between STW depth and ini al soil salinity, nor between ini al and final soil salinity. However, there was a strong rela onship (p<0.01) between ini al water salinity and final soil salinity, with ini al water salinity accoun ng for 41% of the varia on in final soil salinity (Fig. 5). STW depth accounted for 42% of the varia on in final soil salinity (p<0.01), reflec ng the significant rela onship between STW depth and ini al water salinity, which was in turn related to final water salinity, reflec ng the fact that irriga on with higher salinity water results in greater accumula on of salt in the topsoil due to evapora on. Similar observa ons were also reported by Burman et al. (2009) in the Sundarbans. Table 1. Changes in the piezometric level, discharge rate, water salinity of STWs and soil salinity of rice fields during the boro season (data are for STWs monitored in 2003, 2009 and 2014; different STWs were monitored each year, total number of observa ons 44) Depth of STWs (m) Depth of piezometric level (m bgl) Jan May Discharge rate (ls-1) Water Salinity (dSm-1) Jan Jan Range 73-116 1.3-4.1 2.1-6.5 Mean 94 2.1 3.9 10.8 2.18 0.1 0.2 0.3 SEm ± Paired t test (P<0.01) Significant May 7.8-14.2 3.0-9.5 Soil Salinity (ECe, dSm-1) May Jan May 1.3-3.9 2.5-5.4 2.0-3.9 2.6-11.3 7.2 2.0 3.7 3.0 5.8 0.3 0.1 0.1 0.1 0.1 Significant Significant Significant 14 12 10 8 6 y = 3.708 x -1.780 R² = 0.407 4 2 1 2 3 4 5 Water salinity (dSm-1) Fig. 5. Rela onship between ini al water salinty of STW and final soil salinity. 65 3.3 Effect of irriga on water on boro rice produc vity There was a significant (p<0.01) rela onship between STW depth and grain yield of boro rice, with STW depth accoun ng for 70% of the varia on in grain yield (Fig. 6). This was due to the rela onships between STW depth and ini al water salinity, between ini al water salinity and final soil salinity, and between yield and final soil salinity (Fig. 7). Final soil salinity accounted for 74% of the varia on in grain yield (p<0.01). 4.5 4.0 3.5 R² = 0.697 3.0 2.5 2.0 70 80 90 100 110 120 Depth of SWT (m) Fig. 6. Rela onship between STW depth and grain yield of boro rice. 4.5 R² = 0.736, P<0.01 4.0 3.5 3.0 2.5 2.0 2 4 6 8 Soil salinity (dSm-1) 10 12 Fig. 7. Rela onships between soil salinity at harvest and boro grain yield in farmers’ fields irrigated through STWs. 4. Conclusions and recommenda ons STWs are being used for irriga on of crops, mainly boro rice, in the salt affected Sundarbans in the coastal region of West Bengal, India. Decreases in piezometric level and discharge rate and increasing salinity of the irriga on water between the me of establishment and harvest of boro rice were observed. In 2003, the piezometric level and salinity of the STWs recovered during monsoon season in the study area. However, whether this would be the case in intensively irrigated areas with STWs and in all years is not known. Long-term studies of the status of the aquifers used for irriga on of agricultural crops, and of the deeper fresh aquifers, are needed to help determine the sustainable level of extrac on. 66 Soil salinity increased during the boro season and rice yield was strongly and posi vely related to final soil salinity and strongly and nega vely related to STW depth. Final soil salinity was inversely related to STW installa on depth, at least partly due to the higher water salinity of the shallower STW. There is an urgent need to be er characterize the aquifers of the region and to develop a plan for the sustainable use of groundwater for irriga on. Strategies for increasing the produc vity of ground water are also needed, including conjunc ve use of ground water and surface water. Strategies for increasing the availability of fresh surface water are also needed, such as rainwater harves ng through surface storage structures. Harvested fresh water could be used in various ways, such as at cri cal crops stages (when sensi vity to salinity is greatest), and later in the season when STW salinity is highest. Crop prac ces to reduce irriga on requirement during the dry season are also needed, such as early establishment of boro rice and replacement of rice with crops with lower water requirement such as sunflower. The development of boro varie es with improved tolerance to salinity could also increase the produc vity of groundwater in the region. Acknowledgements The authors acknowledge the financial and technical support to this research work by the Indian Council of Agricultural Research (ICAR) and the CGIAR Challenge Program on Water and Food (CPWF). Along with findings from research projects of ICAR-Central Soil Salinity Research Ins tute, this paper presents findings from ‘G2 Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of CPWF. References BAES 2012.District Sta s cal Handbook, SOUTH/ NORTH 24-PARGANAS, Bureau of Applied Economics & Sta s cs, Department of Sta s cs & Programme Implementa on, Government of West Bengal. Burman, D., Bandyopadhyay, B.K., Mandal, Subhasis, Mandal, U.K., Mahanta, K.K., Sarangi, S.K., Maji, B., Rout, S., Bal, A.R., Gupta, S.K. and Sharma, D.K. 2013. Land shaping – A unique technology for improving produc vity of coastal land. Technical Bulle n, CSSRI/Canning Town/Bulle n/ 2013/02, Central Soil Salinity Research Ins tute, Canning Town, South 24 Parganas, West Bengal.p.38. Burman, D., Sarangi, S.K., Mandal, Subhasis and Bandyopadhyay, B.K. 2009. Water quality of tube-wells used for irriga on during rabi abd summer seasons in the coastal areas of Sundarbans, West Bengal. Journal of Indian Society of Coastal Agricultural Research 27(2): 76-79. CGWB 2014. Report on status of ground water quality in coastal aquifers of India. Central Ground Water Board, Ministry of Water Resources, Government of India, Bhujal Bhawan, Faridabad, Haryana, India. p. 121. Gayen, Anadi.2009. Sustainable ground water management op ons in the delta mouth area of Sundarban islands, West Bengal. Bhu-Jal News 24(1):79-84. Mahanta, K.K. and Burman, D. 2004. Studies on the development and u lity status of the shallow tubewells in the Sundarbans area. Journal of Indian Society of Coastal Agricultural Research 22: 124-25. Mandal, Subhasis, Bandyopadhyay, B.K., Burman,D., Sarangi, S.K. and Mahanta, K.K. 2011. Baseline Report of the NAIP project on ‘Strategies for Sustainable Management of Degraded Coastal Land and Water for Enhancing Livelihood Security of Farming Communi es’. Central Soil Salinity Research Sta on, Regional Research Sta on, West Bengal, India. P.74. Misra, A. K. and Nag, S. K. 2009. Aquifer characteris cs in south 24-Parganas and Kolkata Municipal Corpora on (KMC) area, West Bengal. Bhu-Jal News 24(1):28-36. Murty, V.V.N. 1985. Land and water management engineering. New Delhi: Kalyani Publishers. Ray, Abhijit and Shekhar, S. 2009. Ground water issues and development strategies in West Bengal. Bhu-Jal News 24(1): 2-17. 67 Reducing irriga on water requirement of dry season rice (boro) in coastal areas using mely seeding and short dura on varie es S.K. Sarangi 1, D. Burman1, S. Mandal1, B. Maji1, T.P. Tuong 2, E. Humphreys2, B. K. Bandyopadhyay1 and D. K. Sharma1 1 ICAR-Central Soil Salinity Research Ins tute, India, sksarangicanning@gmail.com, burman.d@gmail.com, subhasis2006@gmail.com, b.maji57@gmail.com, bimalbkb@gmail.com, dineshksharma@rediffmail.com 2 Interna onal Rice Research Ins tute, Philippines, e.humphreys@irri.org, t.tuong@irri.org Abstract Irrigated dry season rice (boro) can help to bridge the gap between produc on and consump on in eastern India. Due to scarcity of fresh surface water during the dry season, farmers resort to pumping ground water for irriga on. However this leads to the lowering of groundwater piezometric levels, increased pumping costs and saline water intrusion. Conserva ve use of irriga on water (IW) is essen al to enable boro cul va on while sustaining the produc vity of this fragile ecosystem. Therefore, research was conducted to test the hypotheses that early seeding and the use of short dura on varie es would reduce IW requirements and increase the IW produc vity (IWp) of boro rice. The experiment was conducted in the 2012–2013 and 2013–2014 boro seasons at Canning Town in West Bengal, India. Eight rice varie es including four from Bangladesh were sown during the first week (early seeding) and the last week (late seeding) of November. Each season, yields of early sown Binadhan-8, BRRI dhan47 and CSR 22 were similar (about 6 t ha-1) and significantly higher than yields of the other varie es. With late seeding, Binadhan-8 produced significantly higher grain yield (5.9 t ha -1) than all other varie es. Late seeding reduced grain yield of most varie es, more so in the case of longer dura on varie es - by up to 24%. Irriga on water input increased with variety growth dura on and was lowest with BRRI dhan55, followed by IR 10206-29-2-1-1 and BRRI dhan47. Average IW input with early seeding was 17% less than with late seeding, mainly due to lower irriga on requirements for land prepara on. Irriga on water produc vity with early seeding (41-45 kg grain ha-1cm-1) was 30% higher than with late seeding (31-35 kg grain ha-1cm-1) each year. With early seeding BRRI dhan47 had the highest IWp, while with late seeding Binadhan-8 had the highest. Two of the three varie es with highest yield and highest IWp came from Bangladesh, signifying the importance of cross-country germplasm exchange. There was an indica on that early seeding resulted in higher yields and IWp, but this needs further verifica on. Key message: Irriga on water input to boro crops in the coastal zone of west Bengal can be reduced while achieving yields in excess of 5 t ha-1 through the use of high yielding, short dura on varie es, and possibly through earlier seeding. Keywords: cropping intensifica on, eastern India, water management, water produc vity 1. Introduc on The importance of rice as a staple food crop is known worldwide and Asian countries depend on rice as the major source of daily dietary requirements. It is the staple food of half the world’s popula on and is grown by more than half the world’s farmers (Fairhurst and Dobermann 2002). Rice is the most widely grown crop in India, which makes the country the second highest rice producer (close to 100 Mt per year) in the world. India needs to produce 120 Mt per year by 2030 to feed what will be by then its popula on of over one and a half billion (Adhya 2011). To meet the requirement, innova ons are needed for higher produc on and sustained produc vity. One of the important strategies for increasing produc on is cropping intensifica on. In India, rice is mostly grown during three seasons, locally known as aus or autumn or pre-kharif (pre-monsoon), aman or kharif 68 (monsoon), and boro or rabi or summer (post-monsoon or dry season). Among these three growing seasons, rice yield is highest during the boro season due to ample sunshine, controlled water management (irriga on) and higher efficiency of inputs like fer lizer and crop protec on chemicals (Sarangi et al. 2014). But the area planted during boro is the least—around 4 Mha, in comparison to 38 Mha during kharif—mainly due to limited availability of irriga on water and suitable varie es. These constraints are par cularly true for eastern India where surface water is affected by salinity during the dry season (Kukal et al. 2010; Bouman and Tuong 2001). The source of water for dry season rice irriga on in eastern India is mainly pumping of groundwater through tube wells. Indiscriminate pumping of groundwater lowers the groundwater table during the dry season by an alarming magnitude, which has caused many tube well systems to go dry or deliver pulsa ng discharges during summer months (Goswami 2006). This situa on results in increased pumping costs, salinity intrusion to the aquifers below rice growing land and build up of soil salinity in the top soil. In the future, groundwater u liza on is likely to increase with expansion of irrigated agriculture and efforts to achieve na onal food produc on targets. Climate change is also likely to affect groundwater availability in terms of both quan ty and quality (in par cular coastal aquifers) due to changes in precipita on and evapo-transpira on (Pathak et al. 2014). Therefore, judicious use of irriga on water is essen al to sustain boro rice produc on and to op mize the boro rice cul vated area. To reduce the irriga on water requirement of boro rice effec ve use of residual soil moisture a er the kharif crop, op mum me of seeding and transplan ng, and salt tolerant rice varie es are needed. Shorter dura on boro varie es are also needed as rice requirements for irriga on water decrease with dura on. There is great scope for increasing rice produc on in the eastern coastal plains of India through the development and adop on of suitable varie es with site-specific crop and natural resource management (Saha et al. 2008). Plan ng date plays a crucial role in irriga on requirement and yield. Usually, farmers sow the seeds of the boro crop in the nursery from late November to the first week of December, about three weeks a er the harvest of wet season rice. This plan ng date fails to tap the residual soil moisture in the topsoil of the paddy fields, which dries rapidly due to evapora on and recession of the groundwater table. Furthermore, the late sown rice crop is exposed to higher temperature and evapora on demand during the la er part of the crop growing period, from March to April (Fig. 1). The flowering and grain filling periods are exposed to warmer weather (heat stress), which increases spikelet sterility and shortens the grain filling period leading to lower yield (Cas llo et al. 2006). Modelling studies for two boro rice varie es (BR3 and BR14) in Bangladesh indicated significant reduc on (23 to 41%) in rice yield for delayed plan ng (Basak et al. 2010). Sterility percentage increased from 24 to 45% due to delayed plan ng (Mannan et al. 2012). Soil salinity also increases as the dry season progresses. Salt tolerant varie es can be used to maintain economic yield under moderate soil salinity and may need less irriga on water for leaching of salt than suscep ble varie es. Keeping these facts in view, this study was conducted to test the hypotheses that early seeding and the use of short dura on varie es would reduce irriga on water (IW) requirement and increase the IW produc vity (IWp) of boro rice. 2. Materials and methods 2.1 The study site The experiment was conducted during the boro seasons of 2012-13 and 2013-14 at the ICAR-CSSRI Regional Research Sta on, Canning Town (La tude: 22°15’N, Longitude: 88°40’ E; Al tude 3.0 m above MSL), West Bengal. The climate is tropical monsoon with average annual rainfall of 1802 mm, of which 89% occurs during the monsoon season (June-October). Rainfall during the boro season is not enough to meet the crop water requirement. Both maximum and minimum air temperatures, as well as poten al evapora on, increase from January to April (Fig. 1) and the sky is usually cloudless with sufficient bright sunshine (6–7 h d-1) for photosynthesis. 69 a) Long term rainfall Rainfall (2013-14) BSH (2012-13) Rainfall (2012-13) Longterm BSH BSH (2013-14) 70 10 9 8 7 6 5 4 3 2 1 0 60 50 40 30 20 10 0 Nov b) Dec Jan Feb Mar Apr 40 35 30 25 Min 20 Long term 15 2012-13 10 2013-14 5 0 Nov Dec Jan Feb Mar Apr Fig. 1. Weather parameters during the crop growing months of the dry season: (a) rainfall, bright sunshine hours (BSH) and pan evapora on; and (b) maximum and minimum temperatures at ICAR-CSSRI RRS Canning Town. Long term data are for the period 1984-85 to 2013-14. The experiment was conducted in two adjacent fields with very similar soil proper es and groundwater table dynamics (Table 1, Fig. 2). The soil was a silty clay with a neutral pH, low organic carbon and available nitrogen, medium available phosphorus, and high available potassium (Table 1). Both fields were under paddy cul va on (one rainy season crop per year) for the three years prior to conduc ng this experiment. Table 1. Ini al physical and chemical proper es of the topsoil (0-15 cm) at the experimental site. Area 1 and Area 2 refer to two fields (see sec on 2.2, experimental design). Loca on Clay Sand % BD N P g cm-3 K Zn kg ha-1 pH OC ECe ppm % dS m-1 Area 1 40 10 1.49 210 12.1 419 7.6 6.8 0.42 3.6 Area 2 44 10 1.52 227 11.1 378 7.8 7.1 0.46 4.2 70 The depth to the groundwater table (as observed in piezometers installed to a depth of 3 m) was similar in both experimental areas (Fig. 2). The water table was about 0.4 to 0.6 m below the soil surface in November each year, and the depth increased to about 1.2 m in April as the boro season progressed. Days a er sowing or transplan ng (month/year) 0 20 40 Area 1 60 Area 2 80 100 120 140 Fig. 2. Depth of groundwater in the study areas during the boro seasons of 2012-13 and 2013-14. Area 1 and Area 2 refer to two fields of the experiment (see sec on 2.2, experimental design). (For simplicity, data for May to October are not presented.) 2.2 Experimental design The experiment was conducted in two adjacent fields with two seeding dates (early and late November), one seeding date per field. Eight varie es or lines were evaluated in each field. Seeding dates: Area 1: early November (6 Nov. 2012, 4 Nov 2013) Area 2: late November (28 Nov. 2012, 24 Nov. 2013) Varie es: BRRI dhan47 BRRI dhan53 BRRI dhan55 Binadhan-8 CSR 34, CSR 22 IR 10206-29-2-1-1 CSRC (S) 50-2-1-1-4-B All varie es tested were semi-dwarf high yielding salt tolerant varie es. Four varie es from Bangladesh were included, three of which are short dura on (BRRI dhan47, BRRI dhan55 and Binadhan-8), with the other (BRRI dhan53) being medium short dura on (www.knowledgebank-brri.org, www.bina.gov.bd). Three varie es/lines (CSR 22, CSR 34 and CSRC(S) 50-2-1-1-4-B) from India and one line from IRRI (IR 10206-29-2-1-1) were also included. These varie es were selected on the basis of previous evalua ons done at CSSRI, RRS, Canning Town (Annual Report, CSSRI, 2011-12). 71 There were three replicates of each variety in a randomized block design within each field. Each variety plot measured 5 m x 19 m (gross area, including bunds). The plots were surrounded by 20 cm high x 50 cm wide bunds. The bunds were first built with dry soil, irriga on was applied for puddling, and the bunds were then plastered with mud to minimize lateral seepage. Irriga on for land soaking and puddling was applied 4 d before transplan ng. 2.3 Management Thirty day-old seedlings were transplanted at a spacing of 20 cm x 10 cm with one to two seedlings per hill. A fer lizer dose of 120:20:0 kg N:P2O5:K2O ha-1 was applied to each plot as urea, single super phosphate and muriate of potash, respec vely. All of the P and K and 25% of the N were applied prior to leveling. Half of the N was broadcast 21 d a er transplan ng (DAT) and the remaining 25% was broadcast at 60 DAT. Hand weeding was done twice at 20 and 40 DAT to remove all weeds. Chloropyriphos @ 2 ml l-1 water and tricyclazole @ 0.6 g l-1 water was used to control insects and diseases, respec vely, as recommended. The plots were kept flooded (2.5-7.5 cm) throughout the season un l about 20 d before harvest maturity. 2.4 Parameters monitored 2.4.1 Soil Ini al physico-chemical characteris cs of the soil of the two fields were determined from top soil samples (0–15 cm) which were sun dried and sieved through a 2 mm sieve. Samples were then analyzed for texture, salinity, pH, organic carbon (Walkley and Black 1934), available nitrogen (Subbiah and Asija 1956), available phosphorus (Olsen et al. 1954) and available potassium (Hanway and Heidel 1952). Bulk density at 5-10 cm soil depth was determined on samples collected using a core sampler. 2.4.2 Water The volume of irriga on water applied to each plot for puddling was measured from the discharge rate of the pump and the me of pumping. The depth of applied water was calculated by dividing volume by the plot area; the calcula on assumes that lateral seepage was negligible as all plots were irrigated and puddled at the same me. Water depth s cks with cm scales were installed in the middle and four corners of each plot a er puddling. Irriga on was applied when water depth (as indicated by 3 out of the 5 water s cks) in the plot fell below 2.5 cm, to bring the water depth to 7.5 cm. 2.4.3 Crop dura on Crop dura on was determined as the number of days from seeding to harvest. Harves ng was done when 80% of the grains turned straw colour. 2.4.4 Yield and yield components and irriga on water produc vity At harvest, grain yield was detemined on an area of 5 m x2 m in the middle of each plot. The grain was sun dried, weighed, moisture content was determined using a moisture meter, and grain yield was adjusted to a moisture content of 14%. Yield components (panicle density (no. m-2), number of spikelets panicle-1, spikelet fer lity (%), 1000-grain weight) were also determined. The number of panicles was counted in three randomly selected areas of 1 m2 in each plot, and the average of three was used for sta s cal analysis. Ten panicles were randomly selected from each plot to count spikelets. These panicles were hand-threshed, filled (grains) and empty spikelets (chaff) were separated by submerging in water (floa ng spikelets considered empty), then the number of grains and empty spikelets was counted. Grain samples were adjusted to 14% moisture content to determine 1000-grain weight. 72 Irriga on water produc vity (IWP) of each treatment was calculated by using the following formula: IWP (kg ha-1cm-1) = grain yield (kg ha -1)/volume of irriga on water applied (ha.cm) All irriga on water, from land soaking and pudlling to the end of irriga on period, was included in the determina on of IWP. 2.5 Sta s cal analysis The results for each seeding date were analyzed separately. The effects of variety and year were determined by analysis of variance using the Sta s cal Tool for Agricultural Research (STAR) so ware developed by the Interna onal Rice Research Ins tute (h p://bbi.irri.org). Treatment means were compared using the least significant difference (LSD) test and compared at P=0.05 level of significance (Gomez and Gomez 1984). When the interac on between variety and year was not significant, means were averaged across the two years. Lack of randomiza on of seeding date did not allow sta s cal analysis to be carried on the effects of seeding date. 3. Results and discussion 3.1. Crop dura on, yield and yield components There were no significant interac ons between year and variety for either seeding date on dura on, all yield components, and yield (Table 2). However, there were significant effects of variety and year on all these parameters for both seeding dates. Table 2. Level of significance of the effects of year and variety and their interac on on crop dura on and yield parameters for early and late sown boro Dura on Panicle m-2 Spikelet panicle-1 % fer lity 1000-grain weight Grain yield (t ha -1) P>F P>F P>F P>F P>F P>F Variety (V) <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 Year (Y) <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 V x Y interac on 0.712 0.199 0.094 0.499 0.162 0.665 Variety (V) <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 Year (Y) <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 V x Y interac on 0.999 0.081 0.099 0.739 0.759 0.351 Effects Early seeding (D1) Late seeding (D2) The dura on of the varie es varied from 125 to 143 d with early seeding, and from 120 to 142 d with late seeding. The dura on of each variety with late seeding was consistently 3 to 5 d shorter than that of the same variety with early seeding. In both sowing dates, BRRI dhan55, IR 10206-29-2-1-1, and Binadhan-8 had the shortest dura on, while CSR 34 and CSR 22 had the longest dura on. Yields of CSR 22 and BRRI dhan47 did not differ significantly from that of Binadhan-8 in the first seeding, but were significantly lower than that of Binadhan-8 in the second seeding. On average, grain yield was 11% more with early seeding (mean 5.5 t ha-1) than late seeding (4.9 t ha-1). There was a general trend for delayed seeding to reduce grain yield of the longer dura on varie es by more than the shorter dura on varie es. Yield 73 of BRRI dhan55 (dura on 120-125 d) was not affected by seeding date, while the effect of seeding date on yield of Binadhan-8 (130-134 d) was small. Variety dura on accounted for 56% of the effect of seeding date on yield (data not presented). The reduc on in grain yield with late seeding was associated with reduced spikelet fer lity, and to a very small degree with reduced panicle density (by 1%, but consistent across varie es) (Table 3). On average, spikelet fer lity decreased from 84% in D1 to 77% in D2. This was probably due to higher temperature at flowering with late seeding each season. Flowering of the early seeding occurred in February while flowering of the late seeding occurred in March. The greater effect of late seeding on yield reduc on with longer dura on varie es is also probably due to flowering at higher temperatures. The higher yield of Binadhan-8, CSR 22 and IR 10206-29-2-1-1 was associated with higher panicle density than most or all other varie es and more spikelets per panicle than some varie es. Bindhan-8 also had much higher grain weight than all other varie es except BRRI dhan47. BRRI dhan47 also had rela vely high panicle density and significantly more spikelets per panicle than all other varie es. The higher yield in 2013-14 was associated with unusually high rainfall in February and March 2014 (total about 120 mm, Fig. 1a) which probably reduced the salinity experienced by both the early and late sown crops. 3.2 Irriga on water (IW) The interac on between variety and year was not significant for all IW parameters except IWP at late seeding. Therefore, mean data of the two years are presented for all parameters except the la er. There was a significant effect of variety on all parameters except the amount of irriga on water for land soaking and puddling (IWL) in both years, and a significant effect of year on all parameters except irriga on number. The depth of irriga on water required for land prepara on for the early seeding (22 cm) was much less than that for late seeding (38 cm) (Table 4). This is a ributed to the higher water table depth (Fig. 2) and higher water content of the soil profile at the me of seeding. The amount of irriga on water for land prepara on in 2012 was about 20 mm less than in 2013 for both crops. In 2012 there were 34 and 36 mm of rain between the mes of harvest and transplan ng of the early and late seeded crops, respec vely, compared with no rain during this period in 2013, hence the higher irriga on input for land prepara on in 2012. The irriga on input during the cropped period was 30 to 40 mm lower in 2013-14 than in 2012-13, and this was associated with the unusually high rainfall in February and March 2014 (Fig. 1a). 74 Table 3. Crop dura on, yield and yield components of early and late sown boro varie es. Means are the average of three replicates over two years Dura on Panicle m-2 Spikelet panicle -1 % fer lity 1000-grain weight Grain yield (t ha -1) Early seeding (D1) Variety Binadhan-8 CSR 22 BRRI dhan47 IR 10206-29-2-1-1 BRRI dhan53 CSR 34 BRRI dhan55 CSRC(S) 50-2-1-1-4-B LSD (P=0.05) 134b 143a 135b 134b 141a 143a 125c 135b 1.49 490a 485a 455b 483a 387c 325e 357d 286f 5.3 121bc 118c 139a 126b 112d 97f 107e 97f 3.1 89.1a 84.7c 83.9c 81.4d 80.6de 85.0bc 86.6b 79.6e 1.08 26.7b 22.9f 27.5a 23.9d 22.3g 23.1e 25.5c 22.3g 0.07 6.04a 5.95a 5.89a 5.57b 5.51b 5.25c 4.97d 4.56e 0.12 Year 2012-13 2013-14 LSD (P=0.05) 138a 134 b 0.73 415 a 401b 2.65 115 a 114 b 1.53 84.8 a 83.0 b 0.54 24.2 b 24.3 a 0.04 5.75 a 5.19 b 0.06 Mean 136 408 115 83.9 24.24 5.47 Late seeding (D2) Variety Binadhan-8 CSR 22 BRRI dhan47 IR 10206-29-2-1-1 BRRI dhan53 CSR 34 BRRI dhan55 CSRC(S) 50-2-1-1-4-B LSD (P=0.05) 130d 141a 132c 131cd 137b 142a 120e 132c 1.73 485a 479ab 451c 477b 383d 320f 352e 282g 7.1 122c 119c 140a 126b 113d 99f 107e 98f 3.8 81.6a 73.9cd 79.0ab 76.3bc 73.7cd 75.3c 81.7a 71.5d 3.15 26.5b 22.7f 27.3a 23.8d 22.1g 22.9e 25.3c 22.1g 0.13 5.86a 5.21c 5.34b 5.46b 4.57e 4.24f 4.92d 3.79g 0.18 Year 2012-13 2013-14 LSD (P=0.05) 134 a 133 b 0.62 410 a 396 b 2.51 116 a 115 b 1.35 77.31 a 75.96 b 1.10 24.0b 24.2 a 0.05 5.22 a 4.63 b 0.06 Mean 134 403 116 76.6 24.1 4.92 Note: In the same seeding date, means with the same le er in a column are not significantly different at the 5% level by LSD. The interac on between year and variety was not significant for any parameters. 75 Table 4. Level of significance of the effects of year and variety and their interac on on number of irriga ons (In), depth of irriga on for land soaking and puddling (IWL), depth of irriga on during crop period (IWC), total irriga on water (Total IW) and irriga on water produc vity (IWP), for early and late sown boro In IWL (cm) Total IW (cm) IWP (kg ha-1cm-1) P>F P>F P>F P>F P>F Variety (V) <0.05 0.788 <0.05 <0.05 <0.05 Year (Y) 1.000 <0.05 <0.05 <0.05 <0.05 V and Y interac on 1.000 0.999 1.000 1.000 0.878 Variety (V) <0.05 0.708 <0.05 <0.05 <0.05 Year (Y) 1.000 <0.05 <0.05 <0.05 <0.05 V and Y interac on 1.939 0.944 0.941 0.998 <0.05 Effects IWC (cm) Early seeding (D1) Late seeding (D2) The number of irriga ons increased with crop dura on. About 17 irriga ons were required for early sown CSR 22 and 34 (dura on 141-143 d), whereas early sown IR 10206-29-2-1-1 and BRRI dhan47 (dura on 131-135 d) needed only 13 irriga ons. The number of irriga ons under late seeding increased to 23 in the case of CSR 22 and CSR 34, and to 18 in the case of IR 10206-29-2-1-1and BRRI dhan47. The amount of irriga on water applied between transplan ng and harvest of the early seeding (107 cm) was approximately five mes that needed for land prepara on. The amount of irriga on water applied between transplan ng and maturity (IWC) was similar for early (107 cm) and late (114 cm) seedings. There were small but significant differences in the amount of water applied during the cropped period to each variety , with differences of up to about 20 cm between varie es within both early and late seedings. Both IWC and total IW increased with crop dura on, and were highest for CSR 22 (early seeding total 138 cm), followed by BRRI dhan53. Irriga on requirement was lowest in BRRI dhan55 followed by BRRI dhan47. BRRI dhan55 needed about 20 and 13 cm less irriga on water than CSR 22 under early and late seeding, respec vely. On average, early seeding needed 17% less irriga on water than late seeding. 3.3 Irriga on water produc vity (kg ha-1cm-1) There was a consistent trend across years and seeding dates for highest IWP in Binadhan-8, BRRI dhan47 and IR 10206-29-2-1-1. This was due to both higher grain yield and lower irriga on water requirement during the crop growing period. The low IWP of BRRI dhan55 was due to lower grain yield which more than offset its lower irriga on water requirement. Within variety, there was a consistent trend for higher IWP with early seeding (41-45 kg ha-1cm-1) than late seeding (31-35 kg ha-1cm-1) each year. This was mainly due to the larger amount of irriga on water required for land soaking and puddling and lower yield of the late seeding. Sarangi and Lenka (2000) observed that IWP in summer rice (variety: Lalat) in coastal areas of Odisha (India) could be increased from 38 kg grain ha-1 cm-1 with flooding throughout the growing period to 50 kg grainha-1cm-1of irriga on water by maintaining satura on from one week a er transplan ng to maturity in comparison with a ponded water depth of 3-5 cm. 76 Table 5. Number of irriga ons from TP to maturity (In), depth of irriga on for land soaking and puddling (IWL), depth of irriga on during crop period (IWC)and total irriga on water (Total IW) used for boro rice crop at Canning Town, West Bengal, India (mean of 2012-13 and 2013-14) Treatment In IWL (cm) IWC (cm) Total IW (cm) Early seeding (D1) Variety Binadhan-8 CSR 22 BRRI dhan47 IR 10206-29-2-1-1 BRRI dhan53 CSR 34 BRRI dhan55 CSRC(S) 50-2-1-1-4-B LSD (P=0.05) 14b 17a 13b 13b 16a 17a 13b 16a 1 21.1 21.9 21.4 21.5 21.7 22.3 20.9 22.1 NS1 111ab 116a 102c 101c 113ab 112ab 97c 109b 4 132bc 138a 123c 123d 135ab 134bc 118d 131c 3 Year 2012-13 2013-14 LSD (P=0.05) 15 15 NS 20.7 b 22.5 a 0.9 109 a 106 b 2 130 a 128 b 1 Mean 15 21.6 107 129 Late seeding (D2) Variety Binadhan-8 CSR 22 BRRI dhan47 IR 10206-29-2-1-1 BRRI dhan53 CSR 34 BRRI dhan55 CSRC(S) 50-2-1-1-4-B LSD (P=0.05) 19c 23a 18c 18c 21b 23a 18c 21b 1 37.7 38.1 37.4 37.5 37.8 38.0 36.7 38.0 NS 109d 120b 107d 109d 117c 124a 107d 120b 2 147d 158b 144e 146de 155c 162a 144e 158b 2 Year 2012-13 2013-14 LSD (P=0.05) 20 20 NS 36.4 b 38.9 a 0.8 116 a 112 b 1 153 a 151 b 0.8 Mean 20 37.7 114 152 1NS= not significant Note: In the same seeding date, means with the same le er in a column are not significantly different at the 5% level by LSD. 77 Table 6. Water produc vity of rice varie es for early and late seeding (the interac on between year and variety was not significant for early seeding) Early seeding (D1) Mean Variety Binadhan-8 45.8b BRRI dhan47 47.6a BRRI dhan53 40.8d BRRI dhan55 42.1c CSR 22 43.1c CSR 34 39.2e CSRC(S) 50-2-1-1-4-B 34.9f IR 10206-29-2-1-1 45.3b Mean 42.4 LSD (P=0.05) 1.1 LSD (P=0.05) for variety and year interac on 1.2 Late seeding (D2) 2012-13 a 41.7 38.9b 33.1d 35.9c 38.0b 31.1d 26.0e 39.2b 35.5 2013-14 37.9a 35.1b 27.8d 32.3c 31.2c 24.4e 22.9f 35.3b 30.9 Note: Means with the same le er in a column are not significantly different at the 5% level by LSD. 4. Conclusions and recommenda ons The present study iden fied promising boro varie es for the coastal salt affected areas of West Bengal. The best varie es were Binadhan-8, BRRI dhan47 and CSR 22, which yielded > 5 t ha-1. Because of their short dura on, Binadhan-8 and BRRI dhan47 used less irriga on water and had higher irriga on water produc vity than CSR 22. Thus, two of the three varie es with highest yield and highest irriga on water produc vity came from Bangladesh, signifying the importance of cross-country germplasm exchange. The two season, non-replicated comparison between early seeding (during the first week of November) and late seeding (during the last week of November, farmers’ prac ce) suggested that early seeding produced higher grain yield, required less irriga on water and thus had higher irriga on water produc vity. The effects of seeding date, however, need to be confirmed by further research with adequate replica on of date of seeding. As availability of irriga on water is a constraint during the boro season, combining varie es with high yield and high irriga on water produc vity with op mum seeding me can contribute to increasing rice produc on in coastal salt-affected areas. Acknowledgments This paper presents findings from G2 project on “Produc ve, profitable, and resilient agriculture and aquaculture systems”, a project of the CGIAR Challenge Program on Water and Food. We also greatly acknowledge the support provided by the Interna onal Rice Research Ins tute (IRRI) and the Director, Central Soil Salinity Research Ins tute (CSSRI), Karnal in terms of providing necessary facili es to conduct the experiments at CSSRI, Regional Research Sta on, Canning Town. References Adhya, T. K. (2011). Vision 2030, Central Rice Research Ins tute, Cu ack, Odisha, India. www.crri.nic.in. Annual Report (2011-2012). Central Soil Salinity Research Ins tute, Karnal, Haryana, India, p.115. 78 Basak, J.K., Ali, M..A., Islam, M.N. and Rashid, M.A. (2010). Assessment of the effect of climate change on boro rice produc on in Bangladesh using DSSAT model. Journal of Civil Engineering (IEB) 38(2):95-108. Bouman, B.A.M. and Tuong, T.P. (2001). Field water management to save water and increase its produc vity in irrigated lowland rice. Agricultural Water Management 49: 11-30. Cas llo, E.G., Tuong, T.P., Singh, U., Inubushi, K., Padilla, J., 2006. Drought response of dry-seeded rice to water stress ming and N-fer lizer rates and sources. Soil Science and Plant Nutri on 52: 496-508. Fairhurst, T. H. Abd Dobermann, A. (2002). Rice in the global food supply. Be er crops interna onal, vol. 16, special supplement. www.ipni.net/publica on. Gomez, K.A., and A.A. Gomez. 1984. Sta s cal procedures for agricultural research. John Wiley and Sons, New York. Goswami, S.B. (2006). Water balance and water produc vity of summer rice as influenced by land situa on in tube well commands. Indian Journal of Agricultural Sciences 76(1): 11-15. Hanway, J.J., and H. Heidel. 1952. Soil analyses methods as used in Iowa State College Soil Tes ng Laboratory. Iowa Agric. 57:1-31. h p://bbi.irri.org. Interna onal Rice Research Ins tute, Manila, Philippines. Kukal, S. S., Humphreys, E., Yadav, S. and Singh, Y. 2010. Irriga on water produc vity of rice grown with resource conserva on technologies. 19th World Congress of Soil Science, Soil Solu ons for a Changing World 1 – 6 August 2010, Brisbane, Australia. Mannan, M.A., Bhuiya, M.S.U., Akhand, M.I.U. and Rana, M.M. (2012). Influence of date of plan ng on the growth and yield of locally popular tradi onal aroma c rice varie es in boro season. Journal of Science Founda on 10(1): 20-28. DOI: h p://dx.doi.org/10.3329/jsf.v10i1.16275. Olsen, S.R., C.V. Cole, F.S. Watanabe,and L.A. Dean. 1954. Es ma on of available phosphorus in soils by extrac on with sodium bicarbonate. USDA Circular 939:1-19, Gov. Prin ngoffice Washington D.C. Pathak, H., Pramanik, P., Khanna, M. And Kumar, A. (2014). Climate change and water availability in Indian Agriculture: Impacts and adapta on. The IndianJournal of Agricultural Sciences 84(6): 671-679. Saha, S., Singh, D.P., Sinhababu, D.P., Mahata, K.R., Behera, K.S., Pandy, M.P., 2008. Improved rice based produc on systems for higher and sustainable yield in eastern coastal plain in India. Journal of the Indian Society of Coastal Agricultural Research 26(2): 74-79. Sarangi, S. K. and Lenka, D. (2000). Increasing water-use efficiency in summer rice (Oryza sa va) at Bhubaneswar. TheIndian Journal of Agricultural Sciences 70(11): 779-780. Sarangi, S.K., Maji, B., Singh, S., Sharma, D.K., Burman, D., Mandal, S., Ismail, A.M., Haefele, S.M., 2014. Crop establishment and nutrient management for dry season (boro) rice in coastal areas. AgronomyJournal106 (6), 2013-2023. doi:10.2134/agronj14.0182. Subbiah, B.V., and G.L. Asija. 1956. A rapid procedure for the determina on of available nitrogen in soils. Current Science 25:259-60. Walkley, A.J., and I.A. Black. 1934. Es ma on of soil organic carbon by the chromic acid tra on method. Soil Science 37:29-38.doi: 10.1097/00010694-193401000-00003. 79 80 Sec on 2 Coastal Environment 81 Bacteriological assessment of managed aquifer recharge (MAR) water in southwest coastal areas of Bangladesh M. P. Kabir, M. A. Islam and M. A. Akber Environmental Science Discipline, Khulna University, Bangladesh, mdpervezkabir@gmail.com, a kku_es@yahoo.com, shamim.05es@gmail.com Abstract Drinking water is scarce in the coastal areas of Bangladesh due to the unavailability of freshwater aquifers at suitable depths and high salinity of surface water. Households mainly depend on rainwater harves ng, pond sand filters and rain-fed ponds for drinking water. Thus, people in these areas o en suffer from waterborne diseases. Recently the government and non-government organiza ons started promo ng Managed Aquifer Recharge (MAR) in the coastal areas of Bangladesh to provide safe drinking water. This study was conducted to assess the bacteriological quality of MAR water. Five MARs were selected for the study and samples were taken from source pond, abstrac on well (treated water) and household storage water. Samples were collected in two terms consecu vely on June and August 2014 and were evaluated for indicator bacteria (E. coli and enterococci) and physico-chemical parameters (pH, electrical conduc vity, turbidity and salinity). Although the MAR treatment reduces bacterial contamina on, the treated water showed indicator bacterial contamina on and an increase in salinity. For the source pond, abstrac on well and household storage water the concentra on of E. coli were 2570, 20 and 36cfu/100ml, respec vely; concentra ons of enterococci were 2400, 134 and 209cfu/100ml; and salinity concentra ons were 0.86, 1.21 and 1.27 ppt, respec vely. The study findings suggest that consump on of MAR water may pose a substan al risk to public health. Key message: Applica on of Managed Aquifer Recharge (MAR) for rain-feed pond water treatment may reduce indicator bacterial concentra on significantly, however point of use (in house) water treatment is required to ensure potability. Keywords: drinking water, coastal area, MAR, indicator bacteria, salinity 1. Introduc on About 884 million people in the world, most of whom are from developing countries, do not get their drinking water from improved sources (WHO and UNICEF, 2010). Waterborne diseases are responsible for approximately 7 million deaths every year. Every eight seconds, a child dies from a waterborne disease, about 4 million in a year (Lalzad 2007). Access to safe drinking water is a fundamental human right and, like other developing countries, it is an important na onal goal of Bangladesh (Khan et al. 2008). In addi on to risks from microbial contamina on, the safety of drinking water in Bangladesh is also threatened by chemical contamina on. The widespread fecal indicator bacteria and viral pathogens found in tube wells are known to be predic ve of diarrheal disease in diverse popula ons around the world (Peter et al. 2012). Bangladesh is one of the most densely populated countries of the world; about 145 million people live in 145,000 km2 area. Among the waterborne diseases diarrhea and cholera are s ll the major killers of infant and child mortality in this populous na on (Khan et al. 2008). The Bangladesh Bureau of Sta s cs and UNICEF es mated that children under five years of age suffer from three to five episode of diarrhea each year, each of which lasts for two to three days and some mes more than two weeks (Arsenic Policy Support Unit 2006). Scarcity of drinking water is acute in the coastal areas of Bangladesh as freshwater aquifers are not available at suitable depths and surface water is highly saline. WHO (2004) reported that in the southwest coastal area of Bangladesh (Khulna, Satkhira and Bagerhat Districts) the groundwater is unsuitable for human consump on due to high salinity (Harun and Kabir 2012). At present, the coastal popula on is mainly dependent on natural 82 sources for drinking water, such as rainwater and pond water (Farhana 2011). Several studies (Kamruzzaman and Ahamed 2006; Karim 2010; Islam et al. 2011) showed that natural sources used as alterna ve sources of drinking water have microbial contamina on. Rain-fed pond waters are heavily polluted due to unhygienic sanita on and eventually act as excellent carriers of water-borne pathogens (Karim 2010). Several studies (APSU 2005; Howard et al. 2006) in Bangladesh showed that roo op harvested rainwater is of consistently high quality, free from arsenic and sa sfies the physical and chemical water quality standard of Bangladesh. However, microbiological contamina on was found to occur to a great extent, which may cause significant health hazards for rural people (Karim 2010). As a result, several water treatment technologies emerged in this region such as solar desaliniza on, reverse osmosis, pipe line water supply, and Managed Aquifer Recharge (MAR). MAR can restore groundwater levels and provide a barrier against seawater intrusion, and aquifers can be used as a reservoir facility for both seasonal and long term storage (Rahman et al. 2012). Now government and non-government organiza ons are promo ng MAR to provide safe drinking water for the coastal people of Bangladesh. However, no detailed study on rou ne bacteriological assessment of MAR has been conducted for this area. It is therefore necessary to evaluate the effec veness of MAR systems in producing safe drinking water by analyzing bacteriological contamina on and salinity of the water from source to household storage. 2. Methodology In the southwestern coastal region of Bangladesh, Dacope Upazila and Ba aghata Upazila in Khulna District and Mongla Upazila in Bagerhat District were selected as study areas. In these areas the problem of safe drinking water is likely to be severely acute, and many MARs have already been established as a supplementary drinking water source. A reconnaissance survey was conducted before site selec on to understand the overall scenario of drinking water sources in this area. Five MAR sites were selected as sampling sta ons, out of which two sites were selected from Mongla, two from Dacop and one from Ba aghata. Water sampling and laboratory analysis was conducted in two terms between April and July 2014 to cover both dry and wet seasons. A total of 20 samples were collected during each sampling term where the number of samples from source pond, abstrac on well (treated water) and household storage were 5, 5 and 10, respec vely. For microbiological analysis, 500 ml water samples were asep cally collected in sterile Nalgene plas c bo les. All samples were placed in an insulated box filled with ice packs and transported to the Environmental Microbiology Laboratory of Khulna University for bacteriological analysis within 6 hours of sampling. We analyzed for two indicator organisms (E. coli and enterococci) and four physico-chemical parameters (pH, salinity, EC and turbidity). Proper quality assurance and control were adopted in water sampling, preserva on and laboratory analysis according to the method of the American Public Health Associa on (APHA 1998). The indicator organisms were tested by membrane filtra on technique. For E. coli enumera on the first petri dish was prepared with m-TEC Ager media. A 100 ml water sample was filtered through a 0.45µm membrane filter and then placed in the center of the petri dish. Then the dishes were closed, inverted, and incubated in 35±0.2⁰C for 2 hours. A er 2 ours the plates were sealed with zip-lock bags and placed on a rack in a 44.5±0.2⁰C water bath for 24 hours. A er 24 hours the plates were removed from the water bath. Then the absorbent pad was placed in a new petri dish, and the pad was saturated using urea substrate medium. Pale yellow, yellow-brown, yellow-green colonies are considered to be E. coli. Enumera on of enterococci is nearly similar to the enumera on of E. coli. For enterococci, the first petri dish was prepared by enterococcus agar media. A 100 ml water sample was filtered through a 0.45µm membrane filter and placed in the center of the petri dish and incubated at 35±0.2⁰C for 48 hours. Dark red color colonies of enterococci were formed. The pH was measured by Hanna instrument (Hi 8424), salinity and electrical conduc vity (EC) were measured by Hanna instrument (Hi 9635, microprocessor, Conduc vity and TDS meter) and turbidity was measured by turbidity meter (HACH DR/820 portable spectrophotometer). 83 3. Results 3.1 Indicator bacterial contamina on Microbiological quality of the collected water sample is summarized in Table 1. E. coli and enterococci was present in 100% of the source pond water samples. However, E. coli and enterococci contamina on was concentrated at the abstrac on well and quan ty of E. coli and enterococci slightly increased again at household storage. Table 1. E. coli and enterococci concentra on at source pond, abstrac on well (treated water) and household storage water Parameter Mean Median Range 20 8 0-80 Av. % reduc on Household storage (source pond to abstrac on well) Mean Median Range 36 40 0-130 99.65 134 209 Source pond Abstrac on well (Treated water) Mean Median Range E. coli 2570 2300 15005400 Enterococci 2400 2900 3004200 90 7-460 120 1-530 96.89 Maximum concentra ons of E. coli and enterococci were found in the source pond water. About 37% of abstrac on wellwater showed E. coli concentra ons of 0 cfu/100ml while enterococci were found in all abstrac on well water. At the abstrac on well E. coli and enterococci concentra ons ranged from 0-80 cfu/100ml and 7-460 cfu/100ml, respec vely, and concentra ons tended to increase from abstrac on well to household storage water. E. coli and enterococci removal by MAR were 99.65% (p < 0.01) and 96.89% (p < 0.05), respec vely (Table 1). 3500 3000 2500 2000 1500 1000 500 0 Source pond Abstraction well (Treated water) Household storage Dry season 3165 10 30 Rainy season 2125 30 40 Fig. 1. Seasonal varia on of E. coli concentra on in source pond, abstrac on well (treated water) and household storage water. 84 E. coli concentra ons of source pond, abstrac on well and household storage water in both dry and rainy season are shown in Figure 1. E. coli concentra on of source pond water was higher during the dry season than the rainy season but at the abstrac on well the result was considerably lower. During dry season E. coli removal by MAR was 99.69% (p < 0.01) whereas in rainy season it was 98.58% (p < 0.05). Both dry and rainy season E. coli concentra ons increased from abstrac on well to household storage water. 3000 2500 2000 1500 1000 500 0 Source pond Abstraction well (Treated water) Household storage 1680 2775 44 220 87 475 Dry season Rainy season Fig. 2. Seasonal varia on of enterococci concentra on in source pond, abstrac on well (treated water) and household storage water. Figure 2 shows that the enterococci concentra ons for both source pond and abstrac on well water were higher during the rainy season, unlike E. coli concentra ons. During the dry season enterococci decrease from source to abstrac on well was 97.38% (p < 0.05), and in the rainy season it was 92.07% (p < 0.05). Like E. coli, enterococci concentra ons increased from abstrac on well to household storage water in both dry and rainy seasons. 3.2 Results of physico-chemical analysis The results of the physico-chemical analyses of the water samples collected from source pond, abstrac on well and household storage in both dry and rainy season are summarized in Table 2. Mean pH, EC and turbidity values for all op ons were within the guideline values of WHO (WHO 2004) but salinity is the prime concern, especially at the abstrac on well. Table 2. Concentra ons of physico-chemical parameters during dry and rainy season at source pond, abstrac on well (treated water) and household storage water Source pond Parameter Dry season Wet season Mean Range Mean Range pH 7.35 7.17.9 7.657.88 8.34 EC 2737 820- 923.7 375(µS/cm) 5520 1497 Salinity 1.35 0.40- 0.37 .09(ppt) 2.74 .64 Turbidity 59.37 48.7- 75.86 55.7(NTU) 66.8 95.5 Abstrac on well (Treated water) Dry season Wet season Mean Range Mean Range 7.32 6.87.33 7.087.84 7.87 3526 1817- 1572 5004940 2040 1.73 0.90- 0.67 0.152.45 0.90 34.85 2.41- 40.8 7.6390.5 86.9 Household storage Dry season Wet season Mean Range Mean Range 7.54 7.08.5 8.317.67 8.7 330 1765- 1263 6884575 1838 1.63 0.88- 0.55 0.252.27 0.85 23.9 5.46- 33.45 17.268 70.3 85 As shown in Table 2, mean pH values for all op ons are within the guideline values for drinking water of 6.5 to 8.5 (WHO 2004). During the dry season the mean ECs for all source pond and abstrac on well water were above 1500 µS/cm, whereas during the rainy season mean EC values ranged from 923.7 to 1572 µS/cm. During the dry season, the mean salinity concentra on for all source pond water was 1.35 ppt and for abstrac on well water was 1.73 ppt; results indicate a 28% salinity increase from source pond to abstrac on well. Overall salinity concentra on decreased in the rainy season. In the rainy season, the mean salinity concentra on for all source ponds was 0.37 ppt with salinity of 0.67 ppt at the abstrac on well; results indicate an 81% increase in salinity from source pond to abstrac on well. Source pond water showed the highest turbidity in the rainy season, whereas during the dry season the highest turbidity was observed at the abstrac on well. In the dry season, the mean turbidity of source ponds and abstrac on wells was 57.37 NTU and 34.85 NTU, respec vely. Overall turbidity tends to increase in the rainy season. However, the mean turbidity of all source ponds and abstrac on wells in the rainy season was 75.86 NTU and 40.8 NTU, respec vely (Table 2). 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Source pond Abstraction well (Treated water) Household storage Dry season 1.35 1.73 1.63 Rainy season 0.37 0.67 0.55 Fig. 3. Seasonal varia on of salinity concentra on in source pond, abstrac on well (treated water) and household storage water. Salinity levels of source pond, abstrac on well and household storage water in both dry and rainy seasons are shown in Figure 3. The salinity of the source pond water was lower than the abstrac on well. Salinity levels at the abstrac on well were higher in both dry and rainy seasons, but tended to decline with household storage. 4. Discussion The study suggests that source pond water experiences higher incidents of microbial contamina on than abstrac on well (treated water) and household storage water. Highest median E. coli and enterococci were found in source pond water, which may be a ributed to contamina on from surface runoff including agricultural, household and domes c runoff. It was observed that presence of embankments did not protect any of the ponds from contamina on. Almost 100% of the source pond water samples showed E. coli and enterococci concentra ons above WHO recommended guidelines (Table 1). At the abstrac on well microbial contamina on was rela vely low. Mean E. coli and enterococci concentra ons found at abstrac on wells were 20 cfu/100ml and 134 cfu/100ml, respec vely (Table 1); from source pond to abstrac on well water about 99.65% (p < 0.01) of E. coli and 96.89% (p < 0.05) of enterococci were reduced. However, microbial quality of collected water samples deteriorated from abstrac on well to household storage water. The increase in E. coli and enterococci may be a ributed to poor hygienic and sanitary prac ces. Lack of hygienic prac ces, such as improper cleaning of storage pots before water collec on, water collected by dipping of pot with unwashed hands and water stored in open containers are responsible for microbial contamina on (Blum et al. 1990; 86 Tu le et. al. 1995). During water collec on from various sites we observed that water is collected in unwashed pots and stored in open containers. While 99.65% of E. coli and 96.89% of enterococci was removed by MAR (Table 2), salinity concentra ons increased greatly. From source pond to abstrac on well water, salinity increased by 28% during the dry season and 81% during the rainy season (Table 2). Salinity intake with water may be responsible for le ventricular hypertrophy, increasing the number of strokes and incidences of cardiovascular disease among coastal communi es (Strazzullo et al. 2009). The aim of MAR technology is to remove microbial contamina on as well as decrease the salinity level of groundwater. This study found that MAR technology significantly reduces microbial contamina on, but at the abstrac on well E. coli and enterococci concentra ons were above WHO drinking water guideline values. At the household consump on point different water treatment op ons like chlorina on, alum treatment and halazone tablet treatment may be helpful for reducing microbial contamina on. This study also found that salinity levels increased with MAR treatment. References American Public Health Associa on (APHA) 1998. Standard Methods for the Examina on of Water and Waste water. 20th Edi on, United Book Press, Inc., Bal more, Maryland. APSU 2005. Progress with Provision of Arsenic Mi ga on Op ons to the End of December 2004. Dhaka: Arsenic Policy Support Unit. APSU 2006. Experiences from pilot projects to implement water safety plans in Bangladesh. Dhaka: Arsenic Policy Support Unit. Blum, D., Emeh, R.N., Hu ly, S.R., Dosunmu-Ogunbi, O., Okeke, N., Ajala, M., Okoro, J.I., Akujobi, C., Kirkwood, B.R. and Feachem, R.G. 1990. The Imo State (Nigeria) Drinking Water Supply and Sanita on Project, 1. Descrip on of the project, evalua on methods, and impact on intervening variables. Tropical Medicine and Hygiene, 84: 309-315. Farhana, S. 2011. Suitability of pond sand filters as safe drinking water solu on in storm surge prone areas of Bangladesh: a case study of post-Aila situa on in Shyamnagar, Satkhira district, Khulna. MSc. Diss., BRAC University. Harun, M. A. Y. A. and Kabir, G. M. M. 2012. Evalua ng pond sand filter as sustainable drinking water supplier in the Southwest coastal region of Bangladesh. Journal of the Applied Water Science, 3:161–166. Howard, G., Ahmed, M. F., Shamsuddin, A. F., Mahmud, S. G. and Deere, D. 2006. Risk assessment of arsenic mi ga on op ons in Bangladesh. Journal of Health, Popula on, and Nutri on, 24(3), 346–355. Islam, A.M.; Sakakibara, H.; Karim, M. .; Sekine, M. and Mahmud, Z.H. 2011. Bacteriological assessment of drinking water supply op ons in coastal areas of Bangladesh. Journal of Water and Health (09.2), 415 – 428. Kamruzzaman, A. K. M. and Ahmed, F. 2006. Study of performance of exis ng pond sand filters in different parts of Bangladesh. Proceedings of 32nd WEDC Interna onal Conference s, 377–380. Karim, M. R. 2010. Microbial contamina on and associated health burden of rainwater harves ng in Bangladesh. Journal of Water Science and Technology, 61(8) 2129− 2135. Khan,M. M. H., Kraemer,A. and Mori, M. 2008. Drinking water crisis due to arsenic contamina on in Bangladesh: public health consequences, mi ga on strategies and sustainability. University of Bielefeld, Germany and Sapporo Medical University, Japan. Lalzad, A. 2007. An Overview of the Global Water Problems and Solu ons. h p://www.go aman.com/daten/ en/ar cles/An%20Overview%20of%20the%20Global%20Water%20Problems%20and%20Solu ons.pdf 87 Peter S. K., Larry D. M., Alice L., Daniel E., Williams, Md. J., Alam, B. J., Mailloux, A. S., Ferguson, P. J., Culligan, M. L., Serre, M. E., Ahmed, K. M., Sayler, G. S. and Alexander, V. G. 2012. Unsealed tubewells lead to increased fecal contamina on of drinking water. Journal of Water Science, 10(4): 565−577. Rahman, M. A., Oberdorfer, P., Jin, Y., Pervin, M., and Holzbecher, E. 2012. Impact assessment of hydrologic and opera onal factors on the efficiency of managed aquifer recharge scheme. Excerpt from the Proceedings of the 2012 COMSOL Conference. Strazzullo P., D'Elia L., Kandala, N.B. and Cappuccio, F.P. 2009. "Salt intake, stroke, and cardiovascular disease: meta-analysis of prospec ve studies". BMJ 339: b4567. Tu le, J., Rise, A.A., Chimba, R.M., Perera, C.U., Bean, N.H., Griffin, P.M. 1995. An microbial-resistant epidemic Shigella dysenteriae type 1 in Zambia: modes of transmission. Journal of Infec ous Diseases, 171(2): 371-375. WHO 2004. Occurrence of cyanobacterial toxins (microcys ns) in surface waters of rural Bangladesh-Pilot Study. Geneva: World Health Organiza on. h p://www.who.int/water_sanita on_health/emerging/wsh0406.pdf WHO and UNICEF 2000. Global Water Supply and Sanita on Assessment Report. USA: World Health Organiza on. h p://www.unicef.org/wash/files/gafull.pdf 88 Effects of controlling saline water intrusion in an empoldered area of Bangladesh M. C. Rahman1, T. H. Miah2 and M. H. Rashid2 1 Bangladesh Rice Research Ins tute, Bangladesh, siddiquer07@gmail.com Bangladesh Agricultural University, Bangladesh, tofazzal_miah@yahoo.com, mharunar@yahoo.com 2 Abstract The study explores the effects of controlling saline water intrusion into coastal polder areas of Bangladesh. The study was conducted at the Dacope Upazila of Khulna District (Polder 31) using a predesigned and pretested interview schedule involving 200 farmers. Profitability and undiscounted benefit-cost ra os of rice and shrimp were calculated. Saline water has a significant impact on crop produc on in the study area and the two unions of Pankhali and Tildanga have a contras ng choice of crops and cropping pa erns. Considering the impact of saline water on environment and rice produc on, the farmers of the Pankhali union decided to control saline water intrusion and follow a rice-cum-golda (freshwater prawn) cropping pa ern. By contrast, the farmers of Tildanga prac ced a rice-cum-bagda (brackishwater shrimp) pa ern. Water salinity differed significantly between the two unions. Average monthly salini es throughout the year in Pankhali and Tildanga Unions were 7.85 and 12.34 with standard devia ons of 7.39 and 8.85, respec vely. Lands remained fallow in the boro (dry) season. The benefit-cost ra os of golda and bagda culture were 2.12 and 3.60, respec vely. Farmers in Pankhali Union were able to compensate for lower benefits from golda farming with higher returns from T. Aman rice (BCR 1.76) and other benefits such as grazing land, drinking water, vegetables cul va on, boro farming, leasing arrangements and the higher price of golda. The people of Pankhali Union were ac vely controlling saline water inflow to the polder, whereas those of Tildanga Union preferred saline water because of be er financial benefits from bagda farming. Poor farmers of Tildanga were thinking differently; they wanted to grow more rice and vegetables to escape from Hurry (a leasing system) and improve access to their land. According to farmers, the salinity of Pankhali Union is gradually declining due to changed water management and cropping prac ces. Farmers’ are interested in cul va ng rice in the boro season and introducing salt tolerant rice varie es to improve produc vity and food security. Separate water management policies are needed to facilitate increased produc vity of farming at the two different areas. Key messages: Controlling saline water intrusion into the empoldered area of coastal Bangladesh has an impact on rice produc vity, salinity level, society and environment. Keywords: water salinity, golda, bagda, T. aman rice 1. Introduc on Bangladesh has a large coastal area of 47,211 km2 (32 percent of the country’s geographical area), where 35 million people (28 percent of the country’s total popula on) live in 6.85 million households. Of the country’s 64 districts, 19 are considered to be coastal districts (Rahman et al. 2013). The coastal area has been inhabited for centuries but is now suffering from various development pressures; salinity intrusion is a key problem par cularly affec ng rice cul va on. Bagda (brackishwater shrimp) farming creates an addi onal pressure on coastal areas of Bangladesh and has been associated with various nega ve social and ecological externali es. The situa on has precipitated significant debate and controversies in recent years, resul ng in a number of government ini a ves to manage nega ve impacts including enac ng laws governing land leasing, release of guidelines for site selec on and opera on of bagda farms, requiring consent of local farmers before se ng up bagda farms, and the forma on of Bagda Culture Steering Bodies at na onal, regional and local levels. These government regula ons have been cri cized on the grounds of their inadequacy, weak enforcement and non-sensi vity to environmental concerns (WorldFish Center 2009). 89 The effects on rice produc on are considered significant in coastal Bangladesh; the produc vity of rice in the region is lower (3.0 ton/ha) than the poten al yield (5.5 ton/ha) of modern varie es. Salinity has become a problem for livestock and forestry also, reducing op ons for diversifica on par cularly by poorer farmers. Cul va on of rice-cum-golda (freshwater prawn) rather than rice-cum-bagda has become a suitable alterna ve in some areas, avoiding use of saline water and allowing rice culture. Widespread adop on of this method appears to be limited at present as larger farmers prefer shrimp culture with saline water, a prac ce that constrains op ons for less influen al farmers. Rice takes the monopoly posi on as a staple food of the country and salinity has significant impact on produc on of this staple. Recognizing this, people of the Pankhali Union of polder 31 in Khulna District banned the intrusion of saline water and brought more land under the cul va on of T. Aman rice, following a rice-cum-golda cropping pa ern instead of rice-cum-bagda (Rahman et al. 2013). This study was designed to evaluate the effect of this ini a ve to control saline water intrusion into polders. 2. Methodology Dacope Upazila of Khulna District was selected as the study area, with Pankhali and Tildanga Unions selected for in-depth study because they reflected contras ng cropping and salinity condi ons. A stra fied random sampling was used to select 200 (100 from each Union) farmers with two different cropping pa erns (rice-cum-golda and rice-cum-bagda). Primary data were collected using a structured interview schedule. Both descrip ve sta s cs and ac vity budge ng were used for data analysis. The produc vity and profitability of rice and golda/bagda cropping was assessed using the following algebraic equa on:  TR  TC n   Q y . Py   Qb . Pb   ( X i . Pxi )  TFC i 1 Where,  Net returns from produc on (Tk/ha); Qy Total quan ty of product (kg/ha); Py Per unit prices of the product (Tk/kg); Qb Total quan ty of the concerned byproduct (kg/ha); P Per unit prices of the relevant byproduct (Tk/kg); Xi Quan ty of the concerned ith inputs; Pxi Per unit price of the relevant ith inputs; TFC Total fixed cost involved in produc on; i 1, 2, 3,…, n (number of inputs). 3. Result and discussion 3.1 Impacts of bagda farming in Tildunga Union This sec on provides a qualita ve descrip on of the impacts of bagda farming in the Tildunga Union. A comparison of environment and farming systems is provided in Sec on 3.2. 3.1.1 Effect on forestry A significant impact of salinity on presence and growth of trees was observed in Tildunga Union, with few indigenous fruit trees present. Older people reported a con nuous reduc on of indigenous trees in Tildunga, 90 in contrast to the considerable development of indigenous trees in Pankhali Union. Reduc on of indigenous trees and dominance of mangrove trees provides an indica on of the impact of salinity on forestry. Mangrove trees (namely ‘Gewa’- Excoecaria agallocha and ‘Golpata’- Nypa fru cans) were notably more abundant in the Bagda gher areas. Fig. 1. Development of mangrove trees in bagda ghers. 3.1.2 Unfavorable environment for crops and vegetables Farmers wishing to grow rice and vegetables in Tildunga Union faced constraints associated with salinity in the water and soil. The effect of salinity on crops depends on the degree of salinity at cri cal stages of growth, but overall declines in produc vity with exposure to salinity were reported. Only a small number of farmers grew some vegetables. Several aqua c plants and weeds were reported to be absent from now saline areas including Durba, Baju, Chehur, Tankuni, Halencha, Malanshak, Kalmisak, Ambalisak, Kachuri Pana and Shapla. Older people in par cular reported such historical changes. 3.1.3 Scarcity of grazing land for livestock produc on Saline water brought into gher areas for bagda farming in Tildunga Union has also resulted in a reduc on in land for livestock grazing, limi ng opportuni es for livestock produc on among farmers in the area. The shrinkage of paddy area associated with expansion of bagda farming has also led to an acute shortage of rice straw, which resulted in nega ve impacts on the livelihoods of poorer families dependent on ca le rearing. Fig. 2 No feed for livestock; the cow is grazing on algae in the gher. 91 3.1.4 Reduc on in rice producing areas Saline intrusion associated with bagda culture has le areas unsuitable for rice cul va on in Tildunga Union. Standing saline water in the polder damages rice crops and has an adverse effect on rice yields. Poor dyke construc on by gher owners exacerbates the problem, leading to saline water leaking out or flooding adjoining rice fields. Poor rice farmers have less influence than richer and influen al gher owners and as a consequence rice growing opportuni es have become more restricted within the Union. 3.1.5 Shortage of drinking water People of Tildunga Union area generally drink pond water that is also used for other homestead ac vi es. Bagda cul va on has led to flooding of the area with saline water and contamina on of pond water with salt. Local farmers have tried to resist construc on of bagda ponds near freshwater ponds, leading to social tensions and clashes in Tildunga Union. Poorer people in par cular faced acute problems with access to drinking water, with rainwater becoming the only source of water for drinking and other domes c uses. Deple on and saliniza on of freshwater supplies means women and children have to walk increasingly long distances to collect clean water. Fig. 3. Women and children have been the greatest vic ms of bagda culture developments. 3.1.6 Impact of bagda on society Bagda farming has significant impact on society. Increased income for bagda farmers increases the ability to save and further invest in other employment opportuni es, leading to overall be er socioeconomic condi ons among bagda farmers. Small farmers within Tildunga Union suffer from improper leasing arrangements (known locally as ‘Hurry’) that led to social conflict. In the Hurry system, large farmers lease the land of small farmers. Small farmers have no choice but to lease out the land for whatever money the rich farmers offer for it, with limited bargaining power. Large farmers use their control of canal water to their benefit, as there is no alterna ve irriga on system. 92 The overall impact of bagda farming is shown in the flow diagram below. Be er housing and sanita on system Improved socioeconomic condi on Improved standard of living Expansion of bagda culture Investment on farm and non-rarm business Increased produc on Increased technical knowledge Savings Created employment opportunity for men and women Transporta on facility improved Market and communica on developed More people educated Infrastructural development Income form bagda farm Educa onal ins tute developed Advantages + Impact of Bagda Farming Disadvantages Keeping saline water in farm Less grazing land Acidic soil Decreased livestock produc on Decreased crops and vegetables Destroyed trees of all kinds Administra ve corrup on Improper leasing arrangement Less organic manure Decreased soil fer lity Decreased organic ma er of soil Destroyed bio diversity Affected soil health Affected ecological reserve Land and water use conflict (Between large and small farmers) Increased social conflict and tension Fig. 4. Impact flow diagram of bagda farming. 93 3.2 Effect of controlling saline water intrusion in Pankhali Union 3.2.1 Water salinity of Pankhali and Tildanga Unions Following the restric on of saline water intrusion in Pankhali Union, its water salinity levels changed significantly from those of Tildanga Union. Salinity varies from month to month, with lower salinity during the months of August to December and higher during the months March to June (Figure 5). The average salinity of Pankhali and Tildunga was 7.85 dS/m and 12.34 dS/m, respec vely. The lowest salinity (0.50 dS/m) of Pankhali Union occurred during September and that of Tildanga (1.95) was during November. The highest water salinity occurred in Panlhali and Tildanga during the months of April (19.75 dS/m) and May (26.50 dS/m), respec vely (Table 1). Table 1. Compara ve salinity of Pankhali and Tildanga unions Area (union) Pankhali Tildunga Monthly water salinity (dS/m) Minimum Maximum Mean Std. Devia on 0.50 1.95 7.85 12.34 7.39 8.85 19.75 26.50 Source: Soil Resource Development Ins tute (2012), author’s calcula on 30 26.5 25 23.4 21.5 20 10 8.55 0 19.75 15.65 15 11.6 5 20.3 19.6 15 6.95 Water salinity (dS/m) Tildunga Water salinity (dS/m) Pankhali 11.95 11.15 8.3 4.3 0.5 4.3 0.75 2.63 1.95 0.9 1.3 2.8 2.05 Fig. 5. Compara ve salinity of Pankhali and Tildanga Unions (source: Soil Resource Development Ins tute (2012). 3.2.2 Produc vity of T. aman rice and prawn/shrimp in Tildunga and Pankhali Unions The produc vity of rice also differed between the different salinity areas. The produc vity of rice is marginally higher in the lower salinity areas than that of higher salinity areas. On average the produc vity of rice in Pankhali Union was 3,098 kg/ha and 3,005 kg/ha in Tildunga Union (Table 2). This indicates that water salinity has some impact on the produc vity of rice in the study area. However, the produc vity of golda at Pankhali was higher than that of bagda at Tildanga. 94 Table 2. Produc vity of rice and shrimp in Tildunga and Pankhali unions Area Mean produc vity (kg/ha) Pankhali Union Rice 3098.20 Shrimp (golda/bagda) 254.94 Tildunga Union 3004.72 233.18 3.2.3 Compara ve gross margion, net return and BCR of T. Aman rice and shrimp Although the gross return from golda was higher in Pankhali Union than that of bagda in the Tildunga Union, the benefit cost ra o (BCR) of bagda (3.60) was higher than that of golda (2.12). The reason for this is the low investment cost of bagda (shrimp). The gross return and BCR of rice produc on, though low, was higher in Pankhali Union than Tildunga Union (Table 3), as salinity was lower in Pankhali Union. Table 3. Compara ve per hectare costs and return of T. Aman rice and shrimp farming Items Shrimp farming Pankhali Union (mostly golda) Gross returns (Tk/ha) Variable costs (Tk/ha) Gross margin (Tk/ha) Fixed costs (Tk/ha) Total costs (Tk/ha) Net returns (Tk/ha) BCR 446,025.00 131,898.00 314,127.00 78,547.00 210,445.00 235,580.00 2.12 T. Aman rice farming Tildunga Union (mostly bagda) Pankhali Union Tildunga Union 221205.00 54458.00 58,237.00 25,736.00 47958.00 31999.00 41592.00 32,501.00 15959.00 7612.00 62070.00 7,418.00 33,154.00 8033.00 40032.00 159135.00 3.60 25,083.00 1.76 7926.00 1.20 Source: Field survey (2012). 4. Conclusion Realizing the environmental, social and rice yield impacts of bagda cul va on with saline water, the people of Pankhali Union became interested in cul va ng golda instead of bagda. The community has restricted the intrusion of saline water into the polder area. People of that area were happy enough as the system was beneficial for rice farming and resulted in be er quality water for drinking. Small farmers reported that they had removed the curse of Hurry. Although poor and small farmers benefited by restric ng the entry of saline water, larger gher farmers were reported to be unhappy as it hampered bagda farming. Rice takes the monopoly posi on as the staple food item throughout the country and saline water intrusion for bagda culture hampered rice cul va on. The preference for bagda is clearly driven by significantly higher financial benefit (BCR was 3.60). Farmers in Tildunga Union were also thinking of introducing-salt tolerant rice varie es to solve the exis ng problem of culturing rice alongside bagda. The farmers of both areas want to be sa sfied with the exis ng farming situa on of rice-golda and rice-bagda. However, different water management policies should be adopted within the respec ve areas to facilitate be er management of water for rice-golda and rice-bagda farming. 95 Acknowledgements The author is grateful to the Interna onal Water Management Ins tute (IWMI), which led the ‘G3: Water Governance and Community Based Management’ project of the CGIAR Challenge Program on Water and Food’s Ganges Basin Development Challenge, for gran ng a fellowship to undertake this research. References Rahman, M.C., M.A.B. Siddique, M.A. Salam, M.A. Islam, and M.S. Al-faisal. 2013. Assessment of technical efficiency of rice farmers in a selected empoldered area of Bangladesh. European Journal of Agricultural Sciences 102, no. 10. h p://bellpress.org/Journals/index.php/EJAS/ar cle/viewFile/959/199. Rahman, M.C., M.A.B. Siddique, M.A. Salam, M.A. Islam, M.F. Kabir, and M.S. Mahamud. 2013. Farm level evalua on of T. aman rice cul va on in selected saline and non-saline areas of Bangladesh. IOSR Journal of Agriculture and Veterinary Science (IOSR-JAVS) 91, no. 4. h p://www.iosrjournals.org/iosr-javs/papers/vol4issue4/O0449198.pdf?id=6498. SRDI. 2012. Survey report, 'Soil Resource Development Ins tute', Brackish water sta on, Shatkhira, Bangladesh. WorldFish Center and N.M. Ahsan. 2009. Community-related social issues in shrimp and prawn farming in Bangladesh. WorldFish Center Bangladesh, Dhaka. 96 Sec on 3 Water Governance 97 Indo-Bangladesh Ganges water interac ons: From water sharing to collec ve water management P. Saikia and B. Sharma Interna onal Water Management Ins tute, India, p.saikia@cgiar.org, b.sharma@cgiar.org Abstract Sharing of Ganges water resources has been a long, contested issue between India and Bangladesh. It dates back to the pre-independence era when Bangladesh was part of Pakistan. The crux of the dispute is the water diversion and dry season alloca on at Farakka Barrage in India. The problem is exacerbated by an historical background of mistrust and antagonism between the two countries along with bureaucra c inefficiency within each country. The signing of the Indo-Bangladesh Ganges Water Treaty (GWT) in 1996, one of the landmark trea es in South Asia, gave a new dimension to rela ons between India and Bangladesh with the promise of ending the longstanding dispute over the opera on of the Farakka Barrage. But to what degree the dispute has been se led remains a ma er of conten on. Despite the working agreement, controversies con nue unabated on the inefficient implementa on of the treaty, water diversion at the barrage and the low freshwater flow of the Ganges downstream of the barrage in both Bangladesh and West Bengal, India. This raises major ques ons on the current management prac ces, ins tu onal capacity and policy ini a ves of the Indo-Bangladesh Ganges water sharing arrangements. This paper describes the current challenges faced by downstream areas due to inadequate freshwater flow in the dry season, including increased sedimenta on, salinity intrusion, dying river streams, silta on of naviga onal channels, and river bank erosion in the lower Gange c plains. The paper also reveals many inefficiencies in the ins tu onal arrangements between India and Bangladesh over Ganges water sharing, and presents poten al avenues for improving coopera on between the two countries to address the problem of inadequate dry season freshwater flow. This requires the development of a strong ins tu onal framework to achieve an equitable win-win situa on for both the upper and lower riparian popula ons. The paper concludes with suggested policy op ons and recommenda ons for considera on by the na onal policy makers of India and Bangladesh. Key message: The Indo-Bangladesh transboundary complexi es and challenges related to Ganges water management can be resolved through an effec ve, strong, sustainable and collabora ve ins tu onal arrangement. Keywords: Ganges Water Treaty, Farakka Barrage, Joint River Commission, Gorai River, Hooghly-Bhagirathi River, Kolkata Port 1. Introduc on The Ganges basin covers an area of about 109 Mha and is shared by Nepal, India, Bangladesh and China (Fig. 1). The sharing and management of Ganges water resources between India and Bangladesh has not been an easy process. Despite a joint ins tu onal arrangement and a working agreement through the Ganges Water Treaty (GWT) signed in 1996, the two countries have not been very successful in addressing the challenges of the downstream Ganges Dependent Areas (GDA), par cularly in rela on to mee ng water needs during the dry season (Salehin et al. 2011). 98 GANGES RIVER BASIN RIVER COUNTR Y B OUNDARY CAPITAL C ITY MAJOR CITY BANGLADESH Fig. 1. Ganges River Basin (shaded green), showing the Ganges Dependent Area in India, Nepal and Bangladesh (source: WLE, CGIAR 2012). The crux of the water sharing dispute is the construc on of the Farraka Barrage (Fig. 2) and water diversion into the Hooghly-Bhagirathi River channel, and the resultant reduc on in dry season flow to Bangladesh. Construc on of the barrage dates back to the pre-independence era when Bangladesh was the eastern province of Pakistan. Concern over the navigability of Kolkata port in the state of West Bengal in India pushed Indian officials to take a unilateral decision to construct a barrage at Farakka in West Bengal (Fig. 3) to feed the naviga onal channels of the Bhagirathi River (Government of India 2004). The Ganges splits into two channels about 40 km downstream of the Farakka barrage in Murshidabad District of West Bengal. The right arm is the Hooghly-Bhagirathi River system and the le arm enters Bangladesh through Chapi Nawabganj District of Bangladesh and flows into the Ganges/Padma River, the distributary of the Ganges in Bangladesh. Both the Hooghly-Bhagirathi and Ganges/Padma Rivers flow into the Bay of Bengal. The Hooghly-Bhagirathi River channel, with a catchment area of 55,600 km2, is one of the major distributaries of the Ganges in West Bengal. It is the only perennial river in the Gange c plains of that state. Kolkata port is 299 km downstream of Farakka Barrage, and the naviga onal channel of the port is fed by the Hooghly-Bhagirathi River. The 2,225 m Farakka Barrage was constructed about 300 km north of Kolkata and about 16 km from the border with Bangladesh. A 38.38 km long feeder canal was constructed with a capacity of 1133 m3 s-1 (40,000 3 s-1) to divert water from the Ganges into the Hooghly-Bhagirathi River. The feeder canal offtake is located in the right bank of the river above the barrage, and ou alls into the Hooghly-Bhagirathi River about 40 km from Farakka Barrage. This water diversion interven on only benefi ed India and triggered conflict between India and Bangladesh. Experts in Bangladesh claimed that it reduced the flow of fresh water from the Ganges into Bangladesh and that this caused considerable hydrological, socio-economic and environmental changes in the GDA of Bangladesh, comprising roughly one-third of the country (Mirza and Hossain 2004; Khan 1996). 99 Fig. 2. Map showing the loca on of Farakka Barrage and the Gorai River (source: Mirza 1997). There is also widespread concern in the region that na onal priori es, poli cal dispari es in rela on to increasing human interven on in the river basin, and the impacts of climate change will further exacerbate the transboundary complexi es between the two countries (Crow et al. 1995; Verghese 1997). A study under the Na onal Communica on (NATCOM) project of the Government of India concluded that climate change has already reduced flows in the dry season and increased flows during the wet season in river systems such as the Ganges, Krishna, Cauvery and Narmada (Gosain et al. 2006). Reduced dry season flows have resulted in severe droughts in some parts of the country, and increased flows during the wet season have enhanced intensity of floods in other parts (AK Gosain, Indian Ins tute of Technology, India, personal communica on, 2013). The projected general rise in temperature, temperature extremes and changes in precipita on are also likely to affect river and groundwater flows in the Ganges basin (Bhara et al. 2009). Together with climate change and increased climate variability, hydrological and morphological changes in the Ganges system will have major impacts on the water flow reaching Farakka Barrage, further challenging the transboundary arrangements between India and Bangladesh. Future flows reaching Farakka Barrage will also be modified by increased abstrac on of river water for irriga on, industrializa on and urbaniza on, and by the construc on of hydropower schemes, flood control embankments and revetments in the upper reaches. Excessive exploita on of groundwater will also affect the ow dynamics of the river system (Bhara et al. 2011). 100 The Gange threatens to outllank Farakka Barrage along this course. Kolkata port 299 kilometers downstream of Far akka Barrage Fig. 3. Map showing Farakka Barrage, Farakka feeder canal and Bhagirathi River (source: Rudra 2004). In addi on to future loss of surface water and lowering of the groundwater table, the Ganges River basin faces problems of riverbed and coastal erosion, destruc on of naviga onal routes, deteriora ng water quality and saline water intrusion (Adel 2013). With increasing demand for surface water resources coupled with excessive groundwater withdrawal and contamina on of aquifers, conflict may soon extend to rights on the shared groundwater resources. Resolving the conflicts in water resources sharing between India and Bangladesh would contribute greatly to the development of the region, and help to conserve Ganges water resources and ecology. Among the first steps towards conflict resolu on are understanding the complexi es and the ins tu onal setup of the Indo-Bangladesh transboundary water resources sharing treaty and how these impact on water resources of the coastal zone in Bangladesh and India. The literature on transboundary flow has mostly focused on technical aspects and on the impacts of the Farakka Barrage. Research on ins tu onal arrangements has been scarce. It is possible that that the extensive degrada on of the Ganges and the serious conflicts in water sharing are not just the results of imprudent physical interven ons but, more importantly, due to negligence in river basin management by all its riparian countries, inadequate informa on sharing, and lack of joint management and ins tu onal capacity. This paper aims to increase understanding of the causes of the water sharing conflicts and to iden fy viable approaches for Indo-Bangladesh transboundary coopera on on Ganges water sharing. The objec ves of this paper are: To outline the historical background of the Indo-Bangladesh ins tu onal arrangements over sharing of Ganges water, focusing on the Farakka Barrage dispute, implementa on of the 1996 GWT, and the opera on of the Indo-Bangladesh Joint River Commission (JRC) To iden fy the transboundary complexi es between India and Bangladesh over the sharing of Ganges waters To iden fy the drivers of these transboundary complexi es and how these transboundary issues impact on changes in water resources in coastal Bangladesh To explore and recommend avenues for coopera on over the Ganges River, not just for water sharing but towards a basin-wide resource management approach 101 2. Methods The research methods included review of qualita ve and quan ta ve data in the literature, secondary data analysis and interviews with key personnel involved in decision making about the transboundary flows. Discussions were also held with scien sts, engineers, hydrologists, water experts and professionals from government agencies and academic ins tutes in both India and Bangladesh to get informa on and addi onal literature, and to get their views on Ganges water sharing and other current bilateral arrangements. The literature reviewed is listed by topic in Table 1. 3. Results and discussion 3.1 Common concerns over the low dry season flow of the Ganges in West Bengal and Bangladesh The major concern for both India and Bangladesh is low freshwater flow during the dry season, which has resulted in increasing sedimenta on, salinity intrusion, dying river streams, silta on of naviga onal channels and river bank erosion in the lower Gange c plains (Salehin et al. 2011). Table 1. Literature reviewed by topic Topic Cita ons Transboundary flow of Ganges Afroz and Rahman 2013; Bahadur 2004; Biswas 2011; Crow and Singh 1999;Crow et al. 1995; Joint River Commission (JRC) 1996; Khan 1996; Government of India 2004; Nishat 1996; Rahaman 2009; Rahaman 2006; Salman and Uprety 2011; Salman 1998; Pant 2002 Indo-Bangladesh Ganges water sharing dispute over Farakka Barrage Government of India 2004; Mirza and Hossain 2004; Khan 1996; Crow et al. 1995; Verghese 1997; Rahaman 2006; Bahadur 2004; Iyer 2008 Indo-Bangladesh Ganges water sharing agreements and ins tu onal arrangements Rahaman 2009; Rahaman 2006; Salman and Uprety 2011; Crow and Singh 1999; Rahaman 2006; Upre 1993; Bangladesh Water Development Board 2010; Government of Bangladesh 1977; Government of India 1972 The Indo-Bangladesh Joint River Commission (JRC) Joint River Commission (JRC) 1996 1996 Ganges Water Treaty Crow et al. 1995; Joint River Commission (JRC) 1996; Government of India 2004;Rahaman 2009; Rahaman 2006; Salman and Uprety 2011; Salman 1998) Impact of inadequate freshwater flow in Ganges Dependent Area of Bangladesh and West Bengal Salehin et al. 2011; Ahmad et al. 2001; Nishat 1996; Verghese 1997 Bandyopadhyay and Nandy 2011; Panda and Bandyopadhyay 2011; Adel 2013 Climate change impacts on hydrology of Ganges basin Bhara et al. 2011; Bhara et al. 2009; Gosain et al. 2006 Hydrological and morphological changes of Ganges Bangladesh Water Development Board 2010; Islam and Gnauck 2011; Rudra 2004; Sinha and Ghosh 2012; Panda and Bandyopadhyay 2011 Socio-economic impacts of Farakka Barrage, inadequate freshwater flow of Ganges Bandyopadhyay and Nandy 2011; Banerjee 1999; Basu et al. 2****; Chakravar 2004; Du 2004; Islam and Gnauck 2011; Mai et al. 2014; Mirza and Hossain 2004; Mirza 1997; Panda and Bandyopadhyay 2011; Rudra, 2004; Sinha and Ghosh 2012; Sawin 2009);Government of West Bengal 2007 Ins tu onal arrangements and in transboundary water coopera on Mapendere 2007; Kappameier et al. 2012; Crow et al. 1995; Wolf et al. 2003; Wolf 1998; Wolf 1999; Schmeier 2010 102 3.1.1 Dying and disconnected river channels of the Ganges in Bangladesh The Ganges River feeds the rivers in the western and southwestern regions of Bangladesh. With the rising popula on, the per capita availability of fresh water in Bangladesh will decrease rapidly (Ahmad et al. 2001), and this is also a major concern and challenge in mee ng water demand in Bangladesh. The reduced flows from the Ganges along with the construc on of polders have resulted in sil ng up of stretches of these rivers and increased water salinity in the dry season. The construc on of polders reduced dal flooding of the lands (and associated deposi on of silt), resul ng in accumula on of silt in the river beds (Zahirul Khan, Ins tute of Water Modelling (IWM), Bangladesh, personal communica on, 2013). As a result, many of the tributaries and distributaries of the Ganges/Padma are now becoming disconnected from the main stream due to the rise in sediment bars along the offtakes of these streams (Zahirul Khan, Ins tute of Water Modelling (IWM), Bangladesh, personal communica on, 2013). Many distributaries and tributaries of the Ganges/Padma and the numerous (formerly) perennial water bodies (locally known as pukur, baors, khals, beels, dinghies) are almost dry during the low flow period. Silta on has greatly affected the naviga onal channels in the GDA of Bangladesh (Nishat 1996). The 1989 Bangladesh Inland Water Transporta on Authority study of classified waterways found that the length of the waterways varies from 3865 km in the dry season to 5968 km in the rainy season (Mishra and Hussain 2012). The southwest coastal zone of Bangladesh (Fig. 4) depends on the Ganges for freshwater flow. However, the Gorai River (Fig. 2), the major distributary of the Ganges/Padma in Bangladesh, has been completely dry during the dry season since 1989 (Bangladesh Water Development Board 2010; BG Verghese, Centre for Policy Research, India, personal communica on, 2013). Silta on has created a huge silt dam, about 5.5 m high and 30 km long in the river channel (Verghese, 1997). A vast network of rivers in the southwest region of Bangladesh is dependent on the Gorai (Afroz and Rahman, 2013). The reduced flow has led to increasing sedimenta on and salinity intrusion into the estuaries of these rivers and reduced crop produc on in the region (Nabiul Islam, Senior Research Fellow, Bangladesh Ins tute of Development Studies, BIDS, personal communica on, 2013). To try and be er manage the water scarcity problems and to augment the flow of the Gorai River, the government of Bangladesh has ini ated various projects such as the Ganges Barrage project, the Gorai River Restora on project and water control schemes under the Bangladesh Water Development Board (BWDB), along with increased dredging ac vi es at the inflow points. However, future ac vi es and projects upstream of Farraka Barrage will also have a cri cal influence on downstream dry season flows. In the present scenario of decreasing freshwater flow at Farraka, the next ten years will pose major challenges for the Bangladesh government to manage the food, water and energy needs of its people. Having an effec ve arrangement with its riparian countries will be crucial (Ainun Nishat, personal communica on, 2013). 103 Fig. 4. Map showing the southwest coastal zone of Bangladesh (Khulna, Satkhira, etc.) fed by the Gorai River (source: Bangladesh Water Development Board 2010). 3.1.2 Dying Bhagirathi-Hooghly River system in India Stress as a result of the low dry season flow of the Ganges is also felt in India, mostly in West Bengal. Decreased water flow into the Hooghly-Bhagirathi River has increased the salinity of the river downstream, with reverse flow high de effects as far inland as Tribeni, 205 km from the sea in Hooghly District (Bandyopadhyay and Nandy 2011). Increasing sedimenta on, loose texture of the soil and varying discharge from the Farakka Barrage into the Hooghly-Bhagirathi River have led to meandering of the river channel and the forma on of ox-bow lakes (Kolkata Port Trust, personal communica on, 2013). The catchment areas of the western tributaries to the Hooghly-Bhagirathi River have been dras cally modified during the last two centuries due to these major sediment induced cut-offs, reducing the length of the river (Panda and Bandyopadhyay 2011). The hydraulic regime of this area has also been modified as a result of deple on of forest cover, expansion of agriculture, indiscriminate exploita on of groundwater, expansion of road and railway networks, and the building of dams and barrages across the rivers. All these events combined to contribute to reduced flow and increasing sediment load in the Hooghly-Bhagirathi River. Dams built across some of the western tributaries also reduced the peak discharge of the Hooghly-Bhagirathi River which affected the ability to flush the sediment load into the sea (Chief Hydraulic Engineer, Hydraulic Department, Kolkata Port Trust, personal communica on, 2013). 104 The morphological changes in the Hooghly-Bhagirathi River system (up to Hooghly Point) greatly influence the Hooghly estuary. The increasing sedimenta on in the upper part of the river system is leading to sediment deposits in the estuarine part of the river system. 3.2 Lack of effec ve ins tu onal arrangements 3.2.1 Indo-Bangladesh Ganges water sharing agreements Several bilateral arrangements have been made in the past between the two countries over sharing of the Ganges water and management of downstream dry season flows. In July 1970 an agreement was signed between India and Bangladesh (then East Pakistan) in which it was agreed that the barrage would be the main point of water sharing between the two countries. However, there were no clear provisions on how the water would be shared. On 29 March 1972, a friendship agreement (‘Treaty of Friendship, Coopera on and Peace’) was signed by the two countries. Under this treaty the Indo-Bangladesh Joint River Commission (JRC) was established to facilitate nego a on over sharing of the many common rivers (Government of India 1972). Following this, the two countries have made several a empts to find a suitable and equitable solu on through several legal agreements over sharing of the Ganges water. But due to changing na onal poli cs (changes in governments in both countries) and ups and downs in the rela onship between the governments of India and Bangladesh, the arrangements over water sharing were either ignored or overshadowed by other foreign policy issues. The nego a ons repeatedly failed to make substan ve progress during that phase (Crow and Singh 1999). A breakthrough was achieved in 1975 with the signing of an agreement to temporarily operate the Farraka Barrage for 41 days. Under this agreement, varying discharges at Farakka Barrage were implemented in 10-day periods during April and May 1975 (Table 2). A joint team of experts from the two governments was formed to observe the flow of the Ganges at appropriate places in each country to monitor the effects of the various withdrawals at Farakka. This involved flow observa ons in the discharge canal and the Hooghly-Bhagirathi River, and the flows into Bangladesh. Table 2: Flow in ten-day periods during April and May, 1975 at Farakka barrage (observa ons by the joint team based on which the agreement was signed). Month April,1975 Ten-day period 21st to 30th Withdrawal (m3 s -1) 11,000 May, 1975 1st to 10th 11th to 20th 21st to 31st 12,000 15,000 16,000 Source: Salman and Uprety 2011 The 1975 agreement expired a er two years, however India con nued unilateral diversion of the Ganges flow beyond the s pulated period of the 1975 agreement. Bangladesh then brought the issue to the United Na ons (UN). In 1976, the UN General Assembly adopted a consensus statement and directed both countries to nego ate and se le the issue. Under the direc on of the UN General Assembly both countries signed a five-year agreement in 1977 (Salman and Uprety 2011). Under this agreement a Joint Commi ee was formed to monitor and implement the provisions of this new agreement and to examine the opera on of the barrage. The Joint Commi ee was responsible for observing and recording daily flows at Farakka, in the feeder canal in India, and at Hardinge Bridge in Bangladesh, and to keep a record of the water shared at Farakka between the two countries. One of the provisions, in Ar cle II(i) of the agreement, was that the sharing of the Ganges waters at Farakka from the 1 January to the 31 May every year would be based on 75% availability calculated from the recorded flows of the Ganges at Farakka from 1948 to 1973, with 60% of the flow for Bangladesh and 40% for 105 India (Rahaman 2006) (Table 3). Another provision, the “80 per cent guarantee clause”, guaranteed Bangladesh a minimum of 80% of its share during each 10-day period, however low the flow of the Ganges may be. The provisions of the agreement were highly cri cized by the Kolkata Port Authority as the required amount of water for the port was not met under this arrangement. Thus, with varying demands and individual na onal interest priori es, the two countries failed to come up with an equitable solu on to sustain this agreement. Table 3. Shares of dry season ow at Farakka ( February March April May sec-1) according to the 1977 agreement Flow reaching Farakka1 Period January 3 Withdrawal by India at Farakka2 Minimum release to Bangladesh3 1-10 98,500 40,000 58,500 11-20 89,750 38,500 51,250 21-30 82,500 35,000 47,500 1-10 79,250 33,000 46,250 11-20 74,000 31,500 42,500 21-28/29 70,000 30,750 39,250 1-10 11-20 65,250 63,500 26,750 25,500 38,500 38,000 21-30 61,000 25,000 36,000 1-10 59,000 24,000 35,000 11-20 55,500 20,750 34,750 21-30 55,000 20,500 34,500 1-10 56,500 21,500 35,000 11-20 59,250 24,000 35,250 21-31 65,500 26,750 38,750 1 Based on 75% availability 2 Less if flow reaching Farakka is less than the 75% availability amount 3 At least 80% of this amount if the flow reaching Farakka is less than the 75% availability amount – this “80% guarantee clause” was later removed Note: 1 3 sec-1 = 0.028316847 m3s-1 Source: Upre 1993 As a result of the failure to reach an equitable agreement, a short-term measure was agreed and a Memorandum of Understanding (MoU) was signed on 7 October 1982 on specified water use between January and May for two years (1983 and 1984). There was no arrangement for sharing the dry season flow in 1985. It was only on 22 November 1985 that another MoU was signed for sharing the dry season flow for three years, star ng in 1986. The provisions of these two MoUs were similar to the 1977 agreement except that the 80% guarantee clause was removed. A er the MoU expired no further legal agreements on management of the dry season flow at Farakka were made between the two countries for almost eight years (1989-1996). Finally, with improved bilateral rela ons, the two countries nego ated and signed a 30-year Indo-Bangladesh Ganges Water Treaty (GWT) on 12 December 1996 (JRC 1996). This was one of the landmark water trea es in South Asia, the first me a formally termed ‘treaty’ was signed between India and Bangladesh over Ganges water sharing. The treaty also called for future coopera on between the two countries over its 53 common rivers and emphasized reaching water sharing agreements on all these rivers. 106 The GWT provided a formula for sharing the Ganges water at Farakka barrage during the dry season, opera onal between 1 January 1 and 31 May each year (Table 4, Rahman 2009). The GWT further states that the water sharing arrangement shall be reviewed at five-year intervals, or earlier if needed, by the requirement of either party. Table 4. Formula for Farakka barrage water sharing during January–May, according to the Ganges Water Treaty, 1996 Flow at Farakka ( 3 sec-1 ) India’s share Bangladesh’s share < 70,000 50% 50% 70,000–75,000 Balance of ow 35,000 > 75,000 40,000 Note: 1 3 3 sec-1 3 sec-1 Balance of ow sec -1 = 0.028316847 m3s-1 Source: Joint River Commission (JRC) 1996 A Joint Commi ee was established under the treaty to monitor implementa on of the treaty. The Joint Commi ee is responsible for observing and recording the daily flow at Farakka, along the feeder canal, below the barrage and at Hardinge Bridge, and for submi ng reports and sharing of informa on. The Joint Commi ee is also responsible for reviewing the treaty every five years based on the field observa ons (Rahaman 2009). 3.2.2 Ineffec veness of the ins tu onal arrangements Despite the signing of the GWT, the Indo-Bangladesh water sharing arrangements have not been successful. The two countries have not been able to reach agreement for augmenta on of the dry season flow of the Ganges. A major weakness has been the lack of collec ve ac on and failure in the implementa on of the ins tu onal arrangements. There has been considerable change in the flow of water reaching Farakka since the implementa on of the GWT. The dry season flow at Farakka has seldom reached the historic average flow (during the period 1949-88) of 12,105 m3 s-1 (Fig. 5). It has been over 18 years since the signing of the GWT. During this period, the Joint Commi ee has not conducted any reviews of the implementa on of the treaty. Nor has the commi ee studied the hydrological and morphological changes in the Ganges. The opera on and func oning of the Joint Commi ee remains unclear. Ar cle IV of the GWT states that the Joint Commi ee is responsible for monitoring the Ganges flow. Ar cles V and VI state that the Joint Commi ee shall decide its own procedures and methods of func oning and shall submit an annual report to the two governments (Ganges Water Treaty 1996). 107 14000 12000 10000 8000 6000 4000 2000 0 1996 1998 2000 2002 2004 2006 2008 Flow reaching Farakka Mm3 Feeder Canal of Farakka Mm3 Farakka Mm3 Hardinge Bridge Point Mm3 2010 2012 Fig. 5. Cri cal period of dry season (Jan-May) flow (m 3s-1) from 1996 to 2010 (source: Bangladesh Water Development Board, 2010). The func oning and role of the GWT-JC is debatable. The JC emerged from a bilateral arrangement as a joint ins tu on to liase between the two countries and maximise the benefits from common river systems. But in reality, it is operates as two parallel na onal river commissions with separate offices, JRC Bangladesh in Dhaka and JRC India in the Union Ministry of Water Resource in Delhi. There is no uniformity in the organiza onal structure of JRC India and JRC Bangladesh. JRC Bangladesh comprises technical experts, mostly engineers, and is headed by a director and a member, from Bangladesh, both with technical backgrounds. JRC India includes both policy makers and engineers in its structure, and is headed by a Commissioner of the Department of Ganges River Management within the Ministry of Water Resources. There are no social scien sts, environmentalists or any legal exprts in either JRC. The JRC Bangladesh has created a website where it uploads an annual press release on the 10-day flow sharing data at Farrakka and flows at Hardinge Bridge in Bengali. The website of JRC India has a page with its organiza onal structure on the Union Minister of Water Resources website, but has not made any data available to the public on the Ganges water sharing at Farakka, flow at Farakka and flow to the feeder canal. This limits scien fic research, in-depth analysis and informed discussion on water sharing. Implementa on of the GWT is monitored separately by these two na onal ins tu ons. The different terminologies used by the two ins tu ons make shared informa on ambiguous. Flow data at various points of the Ganges are collected by each ins tu on on its own side (country) and a mee ng is held every year to share the data. These mee ngs are limited to technical data sharing such as dry season flow data and rainy season floods, and issues related to the construc on and repair of embankments. However, there is no clear process for se ling any disagreements between the two par es about the data. There is a provision in the GWT which states that if the Joint Commi ee fails to se le any dispute arising out of the implementa on of the treaty, the dispute should be referred to the Indo-Bangladesh JRC. If the dispute is not resolved, it should be le to the two governments, which would meet urgently at the appropriate level to resolve the issue 108 (Salman 1998). But there is no agreement on what level of government is required to se le disputes, nor any legal mechanism that binds the two par es to resolve disputes within a specified me frame (Rahaman 2009). At the bilateral level, interac ons are limited to just two diplomacy tracks, namely Track-1 and Track-2. Track-1 involves official discussions between high-level poli cal leaders and ministries on issues related to trea es, agreements, cease fires or peace talks. Through Track-1 interac ons, the two governments signed the “Framework Agreement on Coopera on for Development” in 2011, emphasizing the promo on of transboundary coopera on in the management of shared water resources and ecosystems, and joint development and financing of water resource management and hydropower projects (Government of India 2011). This, however, remains just a paper agreement and nothing concrete has been done towards mee ng any common basin management research studies or projects. On Track-2, unofficial interac ons are held involving influen al academics, water experts, NGO leaders and other civil society actors. But this, too, has been limited to closed-door discussions without any concrete results. 3.3 Loss of navigability of the Bhagirathi-Hooghly: How relevant is Farakka Barrage and the 1996 GWT? 3.3.1 Provisions in the 1996 GWT versus reality The 1996 GWT working agreement provided minimum dry season flows to both Kolkata Port in West Bengal and to Bangladesh. The flow to the port was expected to flush out sediments to the deeper part of the estuary, and keep the naviga on channel silt free. However, this has not been very successful in improving the navigability of Kolkata Port (Chief Hydraulic Engineer, Hydraulic Department, Kolkata Port Trust, personal communica on, 2013). The minimum requirement for achieving this, a flow of 40,000 m3s-1 from the feeder canal, is not being met. As a result, sediments are accumula ng in the naviga on channels and big vessels are not able to enter Kolkata Port or Haldia Dock. Some experts in India claim that implementa on of the 1996 GWT has diminished the dry season diversion into the feeder canal, affec ng the maximum possible dra for naviga on (Biswas 2011). 3.3.2 Is the maintenance of Kolkata Port becoming a liability? Kolkata Port was commissioned on 17 Oct 1870 under the Calcu a Port Act. Construc on of Kolkata Port was a poor decision given the hydraulic characteris cs of the river channel. The inadequate dry season flow in the Bhagirathi channel was evident during 1768 and 1777 (Mukherjee 2011). The Hooghly-Bhagirathi River had lost its connec on with the main channel of the Ganges long before construc on of the port. A steady water flow during the dry season is needed to flush the silt from the naviga onal channels of Kolkata Port into the Bay of Bengal (Case Study 2: The Ganges Basin-with Focus on India and Bangladesh 2007). But with inadequate freshwater flow, increasing silta on along with morphological changes in the Hooghly-Bhagirathi River system have limited the use of Kolkata Port by big vessels. The port suffers from a low dra of 7 to 9 m and traffic dropped from 43.4 Mt in 2011-12 to 39.9 Mt in 2012-13 (Sanyal 2013). Even small vessels are not able to navigate the system during the low flow period. The port is suffering heavy economic loss due to reduced cargo volume. Downstream of Kolkata Port, Haldia Port is also losing cargo volume due to sedimenta on in the naviga on channels. There is an urgent need to explore alterna ves to these ports (Jayanta Bandhyopadyay, Indian Ins tute of Management, Kolkata, personal communica on, 2013). Dredging is used to sustain the naviga on channel and to mi gate the heavy silta on in the Hooghly-Bhagirathi River, with enormous subsidies from the government. The increasing dredging cost has become a major burden. The amount of dredging increased from 7 million cubic meters in 1974-75 to 14 million cubic meters in 1995-96 and to 22 million cubic meters more recently (Bahadur 2004). The development of Mongla and Chi agong Ports in Bangladesh is another challenge for Kolkata and Haldia Ports. Kolkata will lose two of its major traders, Nepal and Bhutan, if they shi their dependency to the ports in Bangladesh. Another big challenge for Kolkata Port is implementa on of the proposed Si we Port in Myanmar, under the Kaladan Mul -Modal project to connect Kolkata and Si we port. 109 3.3.3 Farakka Barrage: from solu on to problem The failure of the Farakka Barrage to reduce silta on of the naviga onal channels of Kolkata Port raises major ques ons about the existence and usefulness of Farakka Barrage. First, the original objec ve for which it was constructed has not been met, although the feeder canal does provide water for domes c and agricultural purposes in the adjoining municipali es, and for a few industries and thermal power plants. Second, the changing course of the Ganges poses a major threat to Farakka Barrage, with the possibility of the river ou lanking the barrage and forming a new route through Kalindri and Mahananda channels (Rudra 2004). If the Ganges bypasses Farakka Barrage this will have serious consequences for the 1996 GWT arrangement. Third, instead of genera ng benefits, the construc on of Farraka Barrage has resulted in environmental degrada on and socio-economic problems in the river basin. The Ganges is a meandering river with natural processes of riverbank erosion and accre on. This has been exacerbated by the Farakka Barrage interven on; the problem of silta on was exacerbated, leading to erosion upstream in Malda and increasing silta on downstream of the barrage along the Bhagirathi-Hooghly and Ganges/Padma River systems (Islam and Gnauck 2011). According a report of the Government of West Bengal, as a result of the devasta ng erosion, 236 villages upstream of the barrage lost their land holdings, with an es mated 5043 ha of cul vable land lost between 2000 and 2001 (Government of West Bengal 2007). Downstream of the barrage, in the district of Murshidabad in West Bengal, 79,190 people were displaced from 1988 to 1994 as a result of floods and land erosion (Basu et al. 20**). The increased erosion and silta on have greatly changed the demographics of the area (Mai et al. 2014), with large socio-economic impacts. The large quan es of silt flowing over the plains create vast stretches of land patches/islands in the riverbed (Mukherjee 2011). These riverine landmasses are called chars in Bengali (in West Bengal and Bangladesh) or diaras in the middle Gange c plains of northern Bihar and eastern U ar Pradesh Districts of India. As the chars are temporary, they are usually occupied by unauthorized migrant se lements (Du 2004). The chars are eroded by the monsoon floods every year. The eroded silt is carried by the river, and the silt accumulates and forms new land patches elsewhere. The char inhabitants are highly vulnerable. With the disappearance of the char where they were previously se led, inhabitants are forced to migrate to newly emerged chars. Some of the chars turn into permanent human se lements but their legal status as ‘land’ remains contested (Du 2004). There is a high incidence of migra on of people from the disappearing chars to the newly emerging chars. Due to the fer le soil of the chars, many inhabitants from Malda and Murshidabad Districts of West Bengal also choose to migrate to them (Banerjee 1999). Within India, a dispute arose between West Bengal and Jharkhand over newly formed chars that formed across the border between the states. The erosion and forma on of new chars has also created border disputes between India and Bangladesh. The disappearances of chars on one side of the border and emergence of new chars on other side results in forced migra on. The concept of a border seems to remain vague for the inhabitants, who migrate between the chars in search of be er livelihoods. Between 1992 and 1994, nearly 10,000 hectares of chars emerged that were inaccessible from the Indian side but were readily accessible from Bangladesh. Many inhabitants of chars in India migrated across the border without legal documenta on to reach these chars. Similarly, the char lands in Jalangi in India have been occupied by Bangladesh immigrants. This raises major ques ons regarding responsibility for the management of resources and socio-economic development of the inhabitants of the chars (Du 2004). There are major risks and vulnerabili es in such unpredictable and unstable human migra on and se lement including poverty, environmental hazards, border disputes, illegal migra on, trafficking of women and children, and health problems (Sawin 2009). 110 3.4 Moving forward to effec ve and sustainable collabora ve water management 3.4.1 From water sharing to holis c water resource management According to Ilyer (2008), “It is greed that lies at the heart of water conflict. Agreements, accords and trea es may temporarily bring peace, but conflict will erupt again unless we re-define development challenges and learn to view water as a scarce and precious resource to be conserved, protected and used with extreme economy”. Indo-Bangladesh Ganges coopera on has been limited to just water sharing. It needs to expand to a more collec ve regional water management approach, and to view the Ganges as a precious resource, one that must be jointly managed to maximize benefits. A mul lateral arrangement including all the riparian countries would be more useful. To address the challenge of downstream dry season freshwater flow, the involvement of Nepal in ins tu onal arrangements and interac ons over Ganges water sharing and management is essen al. Nepal is the upper riparian country of the Ganges and nearly 71% of the dry season freshwater flow originates in this Nepal. There have been studies proposing the construc on of storage reservoirs in the basin area of Nepal to augment the dry season flow of the Ganges (Rahaman 2009), which would change the denominator for water sharing. However, different countries view the benefits/detriments of these reservoirs differently (Khan 1996; Jayanta Bandhopadhya, Indian Ins tute of Management, Kolkata, personal communica on, 2013), indica ng that any “water sharing” agreement needs to take into account upstream developments. To develop arrangements for the mutual benefit of all the riparian countries, collec ve efforts and joint studies are needed. These include studies on the hydrological and morphological changes in the Ganges, feasibility studies of the construc on of storage reservoirs upstream and modeling of the hydrology of the basin (Bhara et al. 2011). Water sharing arrangements should be reviewed and revised based on the findings and new agreements could be made. Joint op mum water u liza on, legal water sharing arrangements and ins tu ons for sustainable management of the Ganges water resources would result in socio-economic development and ecological benefits and improve the livelihoods of the lower Ganges popula on in addi on to preven ng poten al future conflicts. A more equitable and sustainable solu on to the contemporary challenges could be achieved by shi ing the focus from just water sharing to the wider development objec ves of u lizing the benefits from integrated water management and development of the river basin. To achieve these objec ves, the riparian countries must approach the basin as a single ecological en ty and the elements of sustainability and equity should be incorporated in water planning and policy goals for the basin. 3.4.2 Enhancing ins tu onal setups The GWT successfully se led the Farakka Barrage dispute, but only temporarily. The exis ng Indo-Bangladesh JRC could act as the driving force to foster mul lateral coopera on and successfully implement its policy ini a ves. But this would require major changes in the role, func on, structure and supervision mechanism of the JRC (Fig. 8) such as: Strengthening the capacity and func on of the JRC. The JRC must be seen as a basin-wide ins tu on supervising the management of the river basin. Policy development and implementa on and dispute se lement over the Ganges basin must be supervised by the JRC. Towards this, frequent interac on between the riparian countries is essen al to bring transparency in informa on sharing. To reach mutual agreement on the organiza onal structure, objec ves and func oning of the JRC the riparian countries must organize interac ons at various levels and include all stakeholders. There is a need to create a pla orm where official and non-official actors meet at a common forum (Track-1.5 diplomacy, Fig. 6). Along with these interac ons, a pla orm for interac on between stakeholders at the grassroots level is important (Track 3 diplomacy, Fig. 6), as proposed by various experts and authors (Mapendere 2007). To understand the real problems and challenges of 111 the basin mee ngs must be conducted involving all stakeholders, including smallholder producers/farmers, processors-traders/private sector, local and provincial government of all member countries. More o en than not, the voices of the grassroots level remain unheard. To disseminate the informa on from these interac ons and bridge the gap between the grassroots and higher policy levels, a pla orm of Track-2.5 diplomacy (Fig. 6) is proposed. Track-2.5 diplomacy would involve the conduc ng of dialogue workshops involving grassroots stakeholders and the classical par cipants of Track-2 level, and is proposed as a useful medium for the development of coopera on in a poli cally tense region (Kappameier et al. 2012). Task forces at Track-1.5 and Track-2.5 levels would be formed to conduct frequent discussions and share reports of the interac ons held at Tracks-1 and -2 and Tracks-2 and -3, respec vely. These interac ons and the forma on of task forces would facilitate the achievement of a joint ins tu onal framework. TRACK-1 diplomacy TRACK-2 diplomacy People to people dialogue at grassroots level; including smallholder producers/ framers, processors-traders/privates sectors, local & provincial government of these riparian countries Interac on between a cademicians, water experts, journalists, business elites, NGO, INGOs of these riparian countries Interac ons between government officials, leaders ministers of a ll riparia n countries TRACK-1.5 diplomacy Task Force Sig n an agreement on the ins tu onal farmework TRACK-3 diplomacy TRACK-2.5 diplomacy Frequent interac on between these task forces Task Force Mutually reach a joint ins tu on framework to strengthen the Joint River Commission Fig. 6. Interac ons between the riparian countries of the Ganges at various levels; proposed levels in blue boxes are Track-3; Track-1.5 and Track-2.5 diplomacy; Track 1.5 means interac on involving representa ves of both Tracks 1 and 2; Track 2.5 means interac on involving representa ves of Tracks 2 and 3. Through these discussions and informa on sharing, the riparian countries could mutually reach and sign an agreement on a joint ins tu onal framework (Fig. 6). Through this framework the func ons and organiza onal structure of the JRC could be strengthened (Fig. 7). Inclusion of all the riparian countries of the Ganges basin in the JRC. Establish the JRC as a joint ins tu on instead of several parallel ins tu ons, with headquarters (JRC Secretariat) in the region headed by one of the riparian countries on a rota ng basis. Under the JRC Secretariat atTechnical team should be established consis ng of technical experts including engineers, hydrologists, environmental scien sts and social scien sts from the riparian countries. This team would coordinate the research/technical studies by establishing a joint study team that would work towards the development of an integrated basin management plan for sustainable development of the basin. An administra ve team under the JRC Secretariat would coordinate administra ve ac vi es such as mee ngs, financial management and communica ons. The monitoring and evalua on mechanism of the JRC must be reviewed and strengthened under a Monitoring and Evalua ng (M&E) team. The M&E team would monitor implementa on of the policy 112 frameworks and development of projects across the river basin. An effec ve informa on technology system with a satellite service to monitor river basin development would make the process more transparent and effec ve. A JRC website must be introduced for sharing and upda ng of data and informa on, and with press releases on JRC mee ngs and projects. Frequent mee ngs of the JRC should be conducted to exchange data and informa on on various aspects and issues of the river basin, discuss development issues along the transboundary river basin, translate the feedback into ac ons, iden fy the tradeoffs, find equitable solu ons to the challenges and facilitate long-term planning and investment. The interac ons and func on of the JRC should be expanded from just water sharing to other water-related issues such as watershed management, water quality, coastal ecosystem management and sustainable development of the river basin (Nishat 1996). Under the JRC, a training ins tute should be set up at its headquarters to provide training workshops for the stakeholders. Na onal commi ees will be needed to provide coordina on between the JRC and the governments of each riparian country, to monitor the work at the na onal level and to report to the Technical and Administra ve sec on at the JRC Secretariat. These na onal commi ees would be independent of the JRC and under the supervision of governments. A Na onal Community Commi ee could be set up under each na onal Commi ee to coordinate work at the grassroots level, working with local NGOs, farmers, etc. The Na onal Community Commi ee could organize pilot projects and training workshops for the stakeholders. Along with interac on at the JRC level, new mechanisms for dialogue and informa on sharing at the na onal level under each Na onal Commi ee would be useful. At this pla orm, interac on among all relevant stakeholders including policy makers, scien sts, engineers, academics, NGOs (na onal and interna onal) and farmers could be conducted. Planning and decision making body Donors and development Partners Consulta ve Group Ministerial Council (Members from respec ve governments of the riparian countries) Joint River Commission for Ganges basin Headquarter (Secretariat) Technical & Administra ve team (Joint research study team, Monitoring & Evalua ng team, Working groups on various sectors, organize frequent interacons and mee ng of the JRC, Training programmes) Na onal Commi es (independent of the JRC and under the supervision of ministry of water resources in respec ve riparian countries) Na onal Community Commi ees (Pilot projects, conduc ng interac ons at the grassroots level, organizing training programme) Fig. 7. Proposed governance structure of the JRC. 113 • Na onal level and sub-basin level studies conducted by the Na onal Commi ees of the respec ve countries and findings reported to the JRC Secretariat Technical Team. This would provide the best available informa on on the available resources and challenges of each country and the region. The exchange of informa on and data among the countries would assist in the development of policies for sustainable management of water resources at both na onal and regional levels. • Under the Technical Team at the Secretariat, and working in close coordina on with the Na onal Commi ees, joint studies could be conducted to iden fy, analyze and communicate the poten al benefits of the shared water resources in the basin area. Informa on on the water resources such as on shared aquifers in the Indo-Bangladesh Ganges basin area would help to maximize the available resources and address water scarcity problems in the GDA during the dry season. Despite the vast amount of freshwater stored as groundwater fed by heavy recharge from Himalayan snowmelt, the countries sharing the basin have not been able to make full use of this resource (Pant 2002). To devise and support technical interven ons and policies towards sustainable management of the groundwater resources and meet the growing water demands in the GDA of West Bengal and Bangladesh, a detailed understanding of the groundwater resources and poten al socio-economic impacts of their further development is needed. • Exchange of informa on across the riparian countries and joint studies will be important to collec vely address water management challenges and to propose joint projects towards the sustainable management of the resources. This could be done through area studies/macro-basin level studies of surface waters and shared aquifer systems, and future climate change impacts on the quan ty and quality of those resources. Joint development ini a ves and demographic studies of the chars could be conducted. At the Na onal Community Commi ee-level, rese lement and rehabilita on programs, disaster management training and monitoring devices and warning systems for the char popula ons could be ini ated. 4. Conclusions The literature on Indo-Bangladesh transboundary water coopera on over the Ganges River generally ignores the poten al role that the Indo-Bangladesh Joint River Commission (JRC) could play in enhancing transboundary coopera on between India and Bangladesh. If expanded, the JRC could provide a pla orm for coopera on among all the Ganges riparian countries. The inefficiency in the present ins tu onal arrangements between the riparian countries, the absence of joint assessment, the lack of transparency in exchange of informa on, and data with poli cal asymmetries between the riparian countries have been major hindrances towards achieving collec ve benefits from the river basin. An ins tu on that manages and governs the transboundary river basin, with all riparian countries as members, is essen al to achieve adap ve integrated water management, to resolve disputes, and to prevent future conflict among the riparian countries. This paper iden fies the importance of strengthening and redesigning the organiza onal and ins tu onal configura on of the JRC. The proposed ins tu onal arrangement would be an effec ve policy ini a ve for building trust and confidence among the riparian countries and would enhance coopera on in informa on and data sharing. This would also provide a pla orm to collec vely manage and govern the river basin, to coordinate and formulate policy, to implement joint projects, and to encourage stakeholder interac ons, from policy makers to the grassroots level. Acknowledgements This paper presents findings from ‘G4 Assessment of the impact of an cipated drivers of change on water resources of the coastal zone’, a project of the CGIAR Challenge Program on Water and Food. The authors would like to acknowledge the coopera on and contribu on of the government departments, academics, water experts and professionals from both India and Bangladesh interviewed for the study. The authors are grateful to B.G Varghese for his guidance at the incep on of the study and his valuable sugges ons on the 114 literature. Sincere thanks also to Jeremy Bird, Director General of IWMI and the Joint River Commission, Bangladesh, for providing construc ve comments and cri cal review that helped improve the quality of the report. References Ahmad, Q.K., Ahmed, A.U., Khan, H.R. and Rasheed, K.B.S. 2001. GBM Regional Water Vision: Bangladesh Perspec ves. In Ganges-Brahmaputra-Meghna Region: A Framework for Sustainable Development. eds. Q.K. Ahmad, A.K. Biswas, R. Rangachari and M.M. Sainju. University Press Limited, Dhaka. pp.31-80. Adel, Miah Muhammad. 2013. Upstream Water Piracy, the Strongest Weapon of Cornering a Downstream Na on. Environment and Ecology Research 1(3). pp 85-128. Afroz, Rounak. and Rahman, Md. Ataur. July 2013. Transboundary River Water for Ganges and Teesta Rivers in Bangladesh: An Assessment Global Science and Technology Journal Vol. 1. No. 1. July 2013 Issue. pp. 100-111. Bahadur, Pun Santa. 2004. Overview: Conflicts over the Ganga? In PSA, Disputes over the Ganga: A Look at Poten al Water-Related Conflicts in South Asia. eds. Subba, Bhim.; Pradhan, Kishor. Panos Ins tute South Asia, Kathmandu, Nepal. pp. 2-20. Bandyopadhyay, Sunando. and Nandy, Sreetapa. 2011. Trend of sea level change in the Hugli estuary, India. Indian Journal of Geo-Marine Sciences. Vol. 40(6). pp. 802-812. Banerjee, Manisha. November 1999. A Report on the Impact of Farakka Barrage on the Human Fabric. South Asia Network on Dams, Rivers and People. New Delhi, India. pp.6-29. Bangladesh Water Development Board (BWDB), September 2010. Interim Report of Consultancy Services for the feasibility study and detailed engineering for Ganges Barrage project. Volume II(Annex A-Hydrology) and Volume III(Annex B-Environmental and Social Baseline Condi on in the Ganges Barrage Projects Study Area; Annex C- Es ma on of Irriga on Water Demand in GDA and Inventory of FCD/FCDI projects of BWDB. Ministry of Water Resources. Government of the People’s Republic of Bangladesh. Basu, Parthiba., Sayani, Mukhopadhyay. and Jayaraman, T. 2***. Climate Change Adapta on in Flood Plain of West Bengal. Climate Change Policy Paper I. World Wildlife Fund (WWF). h p://awsassets.wwfindia.org/ downloads/climate_change_adapta on_in_flood_plains_of_west_bengal.pdf (accessed in August-September 2013). Bhara , L., Lacombe, G., Gurung, P., Jayakody, P., Hoanh, C. T. and Smakh n, V. 2011. The impacts of water infrastructure and climate change on the hydrology of the Upper Ganges River Basin. Colombo, Sri Lanka: Interna onal Water Management Ins tute. IWMI Research Report 142. 36p. Bhara , Luna .,Hosterman,Heather R., McCornick, Peter G., Kis n, Elizabeth J., Pant, Anju ., Sharma, Bharat., August 2009, 'Water, Climate Change, and Adapta on: Focus on the Ganges River Basin' NI-WP 09-03. h p://nicholasins tute.duke.edu/sites/default/files/publica ons/water-climate-change-and-adapta on-1-pap er.pdf (accessed in August 2013) Biswas, A.K. 2011. Coopera on or conflict in transboundary water management: case study of South Asia. Hydrological Sciences Journal 56:4. pp. 662-670. Chakravar , Indranath. 2004. Water Problems for Kolkata Metropolitan Region. h p://rwes.dpri.kyotou.ac.jp/~tanaka/APHW/APHW2004/proceedings/WRD/56-WRD-M108/56-WRD-M108.pdf (accessed in JulyAugust 2013). Case Study 2: The Ganges Basin-with Focus on India and Bangladesh. 10 July 2007. h p://www.caee.utexas.edu/prof/mckinney/ce397/Readings/case-study2_Ganges.pdf(accessed in May 2013). 115 Crow, Ben. and Singh, Nirvikar. August 1999. Impediments and Innova on in Interna onal Rivers: The Waters of South Asia. World Development, 28:1907-25. Crow, Ben. Lindquist, A. and Wilson, D. 1995, Sharing the Ganges: The Poli cs and Technology of River Development. Sage. New Delhi. 272p. Du , Kuntala Lahiri. 2004. Flee ng Land, Flee ng People: Bangladesh Women in a Charland Environment in Lower Bengal, India. Asia and Pacific Migra on Journal, Vol. 13, No. 4, 2004. Government of India, 19 March 1972. Treaty of Peace and Friendship between [1972] Intser 6. Treaty of Peace and Friendship between the Government of India and the Government of the People's Republic of Bangladesh. h p://www.commonlii.org/in/other/trea es/INTSer/1972/6.html (accessed in April 2013). Government of India, 6 September 2011. Framework agreement on coopera on for development between India and Bangladesh. Ministry of External Affairs, Government of India. h p://mea.gov.in/bilateraldocuments.htm?dtl.5218/Framework+Agreement+on)Coopera on+for+Development+between+India+and+B angladesh (accessed in August 2013). Government of India. 2004, Farakka Barrage Project. Chapter 14. Ministry of Water Resources (MWR). h p://wrmin.nic.in/writereaddata/Publica ons/anu141200246187.pdf (accessed in June 2013). Government of West Bengal. April 2007. District Human Development Report: Malda. Development and Planning Department. h p://wbplan.gov.in/htm/HumanDev/Human%20Development%20Malda.pdf(accessed in July 2013). Government of Bangladesh. 5 November 1977. Agreement between the Government of the People’s Republic of Bangladesh and the Government of the Republic of India on sharing of the Ganges waters at Farakka and on augmen ng its flows. Ministry of Foreign Affairs (External Publicity Division) Dacca. Government of Bangladesh. h p://ocid.nacse.org/ dd/ dddocs/394ENG.pdf (accessed in April 2013). Gosain, A.K. and Rao, Sandhya. Basuray, D. 2006. Climate change impact assessment on hydrology of Indian River basins. Current Science 90(3). pp. 346-353. Islam, S.N. and Albrecht, Gnauck. 2011. Water salinity inves ga on in the Sundarbans Rivers in Bangladesh. Int. J. Water, Vol. 6, Nos. 1/2, pp.74–91. Iyer, Ramaswamy. 2008. Trans-boundary Water Conflicts: A Review’. In Water Conflicts in India: A Million Revolts in the Making. eds. Joy, K.J..; Gujja, Biksham.; Paranjpye, Suhas.; Goud, Vinod.; Vispute, Shru . Routledge: New Delhi. pp. 375. Joint River Commission (JRC), 1996. Treaty between the government of the People’s Republic of Bangladesh and the government of the Republic of India on sharing of the Ganga/Ganges waters at Farakka. h p://www.jrcb.gov.bd/a achment/Gganges_Water_Sharing_treaty,1996.pdf (accessed in May 2013). Joint River Commission (JRC), 1996, ‘Treaty between the Government of the People's Republic of Bangladesh and the Government of the Republic of India on Sharing of the Ganga / Ganges Waters at Farakka, Government of People’s Republic of Bangladesh. h p://www.jrcb.gov.bd/a achment/Gganges_Water_Sharing_ treaty,1996.pdf (accessed in May 2013). Kappameier, Mariska. Redlich, Alexander. and Knyazev, Evgeniya. 2012. Building Local Capacity in Eastern Europe. In From Iden ty Based Conflict to Iden ty Based Coopera on-The ARIA Approach in Theory and Prac ce. ed. Rothman, Jay. Springer. Peace Psychology Boom Series. Series Editor: Daniel J. Chris e. pp. 101-106. Khan, Tauhidul Anwar. 1996. Management and Sharing of the Ganges. Natural Resources Journal. Vol.36.pp.455-479. Mai , Somnath. Ghorai,Debrabata. Satpathy, Rajat. and Bandyopadhyay, Ja sankar. October 2014. Monitoring of Land Use Land Cover Change over the years due to River Course Shi ing: A Case Study on Ganga River basin 116 near Malda district, West Bengal using Geo-Informa ces Techniques. Interna onal Journal of Research in Advent Technology, Vol. 2. No. 10. E-ISSN: 2321-9637. h p://www.ijrat.org/downloads/oct-2014/paper%20id210201430.pdf (accessed in December 2014). Mapendere, Jeffrey. 2007. Track One and a Half Diplomacy and the Complementarity of Tracks. COPOJ – Culture of Peace Online Journal, 2(1), 66-81. h p://peacemaker.un.org/sites/peacemaker.un.org/files/ TrackOneandaHalfDiplomacy_Mapendere.pdf (accessed in November 2013). Mirza, Monirul Qader. and Hossain, Md. Altaf. 2004. Adverse effects on Agriculture in the Ganges Basin in Bangladesh. In The Ganges Water Diversion: Environmental Effects and Implica ons. eds. Mirza, M. Monirul Qader. Water Science and Technology Library. USA-Kluwer Academic Publisher. pp. 177-191. Mirza, Monirul Qader, 1997. Hydrological changes in the Ganges system in Bangladesh in the post-Farakka period, Hydrological Sciences Journal, 42:5, 613-631, DOI: 10.1080/02626669709492062. Mishra, Dinesh Kumar. and Hussain, Syed Monowar. 2012. Situa on Analysis on Inland Naviga on. Ecosystems for Life: A Bangladesh-India Ini a ve. Interna onal Union for Conserva on of Nature and Natural Resources (IUCN). h p://cmsdata.iucn.org/downloads/02inlandnavia.pdf (accessed on August 2013). Mukherjee, Jenia. July 2011. No Voice, No Choice: Riverine Changes and Human Vulnerability in the ‘Chars’ of Malda and Murshidabad. Occasional Paper No.28, Ins tute of Development Studies Kolkata (IDSK). Nishat, A.1996. Impact of Ganges Water Dispute on Bangladesh. In Asian Interna onal water: From Ganges-Brahmaputra to Mekong. eds. Biswas, A.K. and Hashimoto, T. Oxford University Press, Bombay. pp. 60-80. Panda, Surajit and Bandyopadhyay, Ja sankar. 2011, ‘Morphodynamic Changes of Bhagirathi River at Murshidabad District Using Geoinforma cs’, Journal of Geographic Informa on System, 2011, 3, 85-97. file:///C:/Users/773611/Downloads/JGIS20110100004_53874435.pdf (accessed in July 2013). Pant, Niranjan. 2002. Groundwater Issues in Eastern and Western Alluvium of Ganga Basin. Interna onal Water Management Ins tute (IWMI). IWMI Tata Water Policy Program Annual Partners Meet 2002 Discussion Paper. 25p. Rahaman, M.M. 2009. Principles of Transboundary Water Resources Management and Ganges Trea es: An Analysis, Water Resource Development, Vol. 25, No. 1, 159-173. Rahaman, M.M. 2006. The Ganges water Conflict-A compara ve analysis of 1977 agreement and 1996 treaty. Interna onal Water Law. 1/2: 195-208. ISSN 1886-5860/ h p://www.interna onalwaterlaw.org/bibliography/ ar cles/general/Rahaman-Ganges-Asteriskos.pdf (accessed in August 2013). Rudra, Kalyan. 2004. The Encroaching Ganga and Social Conflicts: The Case of West Bengal India. West Bengal: Dept. of Geography, Habra H.C. Mahavidyalaya, 2003. pp 9-40. Schmeier, Susanne, August 2010, ‘Governing Interna onal Watercourse: Perspec ve FROM Different Disciplines’, Her e School of Governance-Working Paper, No. 53. url:h p://www.transboundarywaters. orst.edu/publica ons/publica ons/Schmeier%20RBO%20Literature%20Review%2010.pdf (accessed in May 2013). Salman, M.A.S. and Uprety, K. 2011. Legal aspects of sharing and management of transboundary waters in South Asia: preven ng conflicts and promo ng coopera on. Hydrological Sciences Journal, 56:4, 641-661. Salman, M.A.S. 1998. Sharing the Ganges Waters between India and Bangladesh: An Analysis of the 1996 Treaty. In Interna onal Watercourse: Enhancing Coopera on and Managing Conflicts. eds. Salman M.A.S. and De Chazournes, L.B. World Bank Technical Paper No. 414, The World Bank, Washigton, DC. 117 Salehin, M., Khan, M.S.A., Prakash, A. and Goodrich, CG. 2011. Opportuni es for Trans-boundary Water Sharing in The Ganges, The Brahmaputra and The Meghna Basins. In India Infrastructure Report 2011: Water-Policy and Performance for Sustainable Development. eds. Infrastructure Development Finance Company Limited, India, New Delhi: Oxford University press. Chapter 3. pp. 29-43. Sanyal, Shantanu. April 2013. Kolkata Port Trust traffic drops 8% in 2012-13. The Hindu. h p://www.thehindubusinessline.com/industry-and-economy/logis cs/kolkata-port-trust-traffic-drops-8-in-20 1213/ar cle4576743.ece (accessed in October 2013). Sinha, R. and Ghosh, S. 2012.Understanding dynamics of large rivers aided by satellite remote sensing: a case study from Lower Ganga plains, India. Geocarto Interna onal, 27:3. pp. 207-219. Sawin, Ashok. 2009. Environmental Trap: The Ganges River Diversion, Bangladeshi Migra on & Conflicts in India. Report No. 41, Department of Peace and Conflict Research Uppsala University, Sweden. pp. 63-65. Upre , B.C. 1993. Poli cs of Himalayan River Waters. Nirala Series 32. Jaipur, India: Nirala Publica ons. Verghese, BG. 1997.Water Conflicts in South Asia. Centre for Policy Research. Studies in Conflict & Terrorism, 20:185-194, 1997. Taylor & Francis. pp. 185-194. Wolf, Aaron T. Yoffee, S. and Giordano, M. 2003. Interna onal Waters: Iden fying basins at Risk. Water Policy. Water Policy, 5(1): 29-60. Wolf A T, 1998 Conflict and Coopera on along Interna onal Waterways, Water Policy 1, 251-65. Wolf, Aaron T, 1999, Çriteria for Equitable Alloca on: The heart of Interna onal Water Conflict, Natural resources Forum, 23: 3-30. WLE, CGIAR. 2012. Dependent Area Map, h p://wle.cgiar.org/wp-content/uploads/2012/09/Ganges1.jpg (accessed in April 2013). 118 Community water management and cropping system synchroniza on: The keys to unlocking the produc on poten al of the polder ecosystems in Bangladesh M.K. Mondal1, E. Humphreys1, T.P. Tuong1, M.N. Rahman2 and M.K. Islam2 1 Interna onal Rice Research Ins tute, Bangladesh and Philippines, m.mondal@irri.org, e.humphreys@irri.org, t.tuong@irri.org 2 Bangladesh Agricultural University, Bangladesh, nefaur25@gmail.com, khairulislam521@gmail.com Abstract Despite huge investments in the polders of the coastal zone over the past 50 years, crop produc vity remains much lower than in most of Bangladesh. Most farmers in the polders cannot adopt high yielding varie es (HYV) of aman rice due to the risk of submergence a er transplan ng. The local aman varie es mature late, delaying establishment of rabi (non-rice) crops, which, as a consequence, are o en damaged or destroyed by pre-monsoon rains or cyclones prior to harvest. A pilot was undertaken to test the hypothesis that community water management is the key to increasing produc vity in a six ha watershed with 37 farmers in polder 30 in Khulna district from May 2012 to June 2014. The research team worked with the community to define the boundaries of the pilot watershed and to construct small drains/levees separa ng lower and higher lands. A drainage outlet was installed to connect the watershed to the adjacent khal. The farmers were trained in drainage water management and produc on of HYV aman, and in the early establishment and produc on of high yielding rabi crops. In 2012 the farmers were provided with enough HYV rice seed to plant the whole area, but only approximately 50% of the area was planted with HYV while the rest of the land was planted with tradi onal aman varie es. The pilot watershed was inundated twice in 2012 as a result of heavy rainfall, but the community drained the excess water within three to four d through systema c opera on of the sluice gate. This indicated that with community involvement and proper infrastructure management, water stagna on can be avoided. However, at the me of maturity of the HYV rice, the local rice was just flowering so instead of terminal drainage, more water was brought into the watershed to finish the local rice crop. As a result, rabi (mostly sesame) crop establishment was delayed un l late February/early March and the crops were subsequently destroyed by heavy rainfall in mid-May 2013. In the 2013-14 cycles, six farmers in a con guous area of 1.3 ha strictly followed the water management and cropping system plan; they cul vated HYV aman, drained the fields prior to topdressing nitrogen (N) fer lizer (twice), drained at the end of October, and then planted maize, sunflower and wheat on 10 December 2013. Yields of rice, sunflower, maize and wheat were approximately 5.0, 2.8, 5.9 and 1.3 t/ha, respec vely. Yields of maize and wheat were lower than expected due to inability to irrigate, as instead of storing fresh water in the canal for use during the dry season once the river water became too saline, the sluice gate was le slightly open to prevent silta on of the canal intake. Therefore, the water in the khal was too saline for irriga on from the beginning of February un l June. The yield of the local rice crops of the farmers in the rest of the original pilot watershed averaged 2.6 t/ha and yield of the HYV rice averaged 3.4 t/ha (with no or very li le fer lizer). As in the previous year, there was no sesame harvest due to late establishment and early monsoon rains. This pilot demonstrated that due to the prevailing hydrology, individuals alone cannot successfully modify their cropping system schedule and adopt improved agricultural technologies in coastal polders. It requires synchronized cropping pa erns among farmers within community water management units, and community coordina on at a range of scales – within small community water management units, within a sub-polder and at the polder scale. Further engagement with polder communi es is needed to determine how they can take advantage of opportuni es for increasing cropping system produc vity through improving water management and use of improved varie es and management. Key message: Community water management and cropping system synchroniza on are essen al if farmers of the polders of the coastal zone are to benefit from the tremendous opportuni es to increase produc vity through adop on of improved varie es and cropping system technologies. 119 Keywords: drainage, rice, maize, sunflower, Khulna 1. Introduc on Most of the world’s rice produc on takes place in the Asian mega deltas where large areas of rice land are subjected to stagnant flooding—0.3 to 0.5 m water depth for prolonged periods (several weeks to several months)—during the monsoon. Modern high yielding varie es (HYV) of rice adapted to stagnant flooding are not yet available. Therefore, in loca ons such as the coastal zone of Bangladesh most farmers grow a single rice crop (aman) during the rainy season using tall, photoperiod sensi ve, local landraces that can survive stagnant flooding but have low yield (2.0 to 3.5 t/ha) and mature late (growth dura on 155 to 170 d). The aman crop is o en followed by a late sown, low input and low yielding rabi crop (0.5 to 1.0 t/ha, but with severe damage or crop failure in about 40% of years due to the pre-monsoon rains and cyclones in May). Thus much of the coastal zone remains as fallow land for three to seven months every year. There is huge scope for increasing cropping system produc vity in the coastal zone. Trials conducted in farmers’ fields under the CGIAR Challenge Program on Water and Food have shown that the use of HYV aman can double rice yield, provided that water is managed to avoid stagnant flooding (Ruhul et al. 2015). On-farm trials have also shown that the earlier maturity of HYV aman (growth dura on 115 to 145 d) enables double and triple cropping with rice or rice and rabi crops (Bha acharya et al. 2015; Mondal et al. 2015a; Ritu et al. 2015; Saha et al. 2015). The earlier maturity of HYV aman creates the possibility of drainage in early November and earlier rabi crop establishment. This in turn allows for diversifica on to higher yielding and/or higher value rabi crops. However, the key to being able to change to HYV rice and thus to cropping system intensifica on and diversifica on is improved water management – through improved drainage management and separa on of lands of higher and lower eleva on to prevent accumula on of water in the lower lands (Bha acharya et al. 2015; Saha et al. 2015). The challenge is how to achieve wide scale implementa on of improved water management to enable cropping system intensifica on and diversifica on. The rivers of the coastal zone of Bangladesh are dal and this effect extends about 150 km inland. The dal fluctua on is more extreme (up to several meters) during the rainy season, resul ng in dal flooding of adjacent lands. About 1.2 Mha of the agricultural lands of the coastal zone of Bangladesh were therefore poldered by humanitarian projects during the 1960s and 1970s to prevent inunda on and saline water intrusion, crea ng scope for improving produc vity through improved water management. Salinity is commonly perceived to be the main reason for non-adop on of improved agricultural technologies in the coastal zone. But we hypothesized that water stagna on and waterlogging are the primary constraints and that community coordina on is needed to enable wide scale adop on of improved agricultural technologies in polder ecosystems. At low de (twice daily) river water levels are usually lower than the land level within the polders, crea ng the opportunity for drainage of excess water by gravity to a level that would allow good growth and yield of HYV aman. We also hypothesized that drainage shortly prior to aman harvest would allow the soil to dry sufficiently for mely establishment of rabi crops. However, these opportuni es have not yet been recognized. The present study was therefore undertaken to work with a community to demonstrate the benefits of improved management of water in a polder, and how to achieve this. Specific objec ves of the study were: To work with a community to implement improved water management and improved cropping systems within a pilot watershed To evaluate the performance of the pilot project in terms of water management, crop performance and community coordina on 120 2. Methodology 2.1 Site The hypotheses were tested in a six ha pilot watershed (la tude 22.70N, longitude 89.50 E) with 37 farmers in polder 30 (at Kisma ultola village, Ba aghata Upazila, Khulna District) from July 2012 to June 2014. An area adjacent to a khal (canal) and near the sluice gate connec ng the khal to the Kazibacha River, enclosed on two sides by a regional highway and rural roads, was selected in consulta on with the local community (Fig. 1). Earthen dykes (50 cm high x 50 cm wide) were constructed on the eastern and western sides in order to a ain hydrological separa on of the pilot area from the adjacent lands. Fig. 1. Map of the study site showing the pilot watershed area in polder 30, Kismat Fultola, Ba aghata, Khulna, Bangladesh. The canal (khal) is used for drainage to the river at low de during the rainy season and for water intake during the dry season. The Kazibacha is a dal river with water depth fluctua ons of two to three m at the loca on of the sluice gate. The water users’ associa on (WUA) generally operates the gate. Monitoring over several years has shown that the river becomes too saline for irriga on (4 dS/m) in early to mid-February and salinity increases to around 18 dS/m in early June (Mondal et al. 2015b). The salinity then decreases rapidly to <1 dS/m un l January, when it begins to increase again. The predominant soil texture within the mini-watershed is silty clay and the dominant cropping pa ern involves local aman landraces transplanted in August and harvested in late December/January followed by sesame sown in late February and harvested in late May/early June. This cropping system predominates throughout polder 30. 121 2.2 Development of collabora ve arrangements with the community on sluice gate management From January to June 2012 several informal mee ngs were conducted with the pilot watershed farmers, neighboring farmers, officials of the local water management group (WMG) and the Union Parishad (UP – the lowest administra ve unit) to discuss the community water management proposal. Three formal mee ngs involving the above groups were also organized from April to June 2012 to discuss implementa on details. Several mee ngs were also organized with the local WMG officials and pilot watershed farmers to come to an agreement on the cropping systems to be prac ced in the watershed and on the water management prac ces (mainly management of the sluice gate to enable growth of the agreed cropping systems). The (informally) agreed upon water management and cropping systems involved: HYV aman, using non photoperiod sensi ve, medium dura on varie es (135 to 145 d) that would be transplanted (using 25 to 30 d-old seedlings) in the first week of August to enable harvest by the third week of November Rabi crops of maize and sunflower to be sown in early December, and sesame and mungbean to be sown in early February Drainage during the aman crop whenever the water was too deep (taking into account the development stage and height of the crop) by systema cally opening the sluice gate at low de to drain the khal, and draining water from the pilot into the khal un l the desired field water depth was achieved Terminal drainage of the aman crop – drainage of the standing water (if any) in early November, about two weeks before harvest maturity Establishment of maize and sunflower by dibbling into the moist soil; llage for the sesame and mung in early February (once the soil had dried enough for llage using a power ller driven by a two-wheel tractor) followed by sowing in the first week of February Filling of the canal with river water and closure of the sluice gate in early February before the river water became too saline to use for irriga on Irriga on of the rabi crops as needed (two to three irriga ons for the maize and sunflower); one irriga on of the sesame and mungbean if needed Thirty-seven farmers grow crops within the six ha pilot area, 30 farmers on their own land and seven farmers on leased land. The landowner farmers cropped 0.01 to 0.40 ha within the pilot watershed and the tenant farmers 0.04 to 0.45 ha. All the farmers in the pilot watershed were male, 69% of whom were middle-aged (40-60 years), 14% were <40 years and 17% were >60 years old. Almost all the famers had some level of formal educa on; about 38% had completed primary (grade I-V), 35% secondary (VI-X), 13% higher secondary (XI-XII) educa on and about 11% had completed a bachelor’s or master’s degree. 2.3 Construc on of the drainage system A topographic map of the study area was collected from the Bangladesh Water Development Board (BWDB). Using this map and farmer local knowledge, the loca ons of the watershed boundary and internal drains were determined. Internal drains (~20 cm x 20 cm) and bunds (~30 cm x 30 cm) were constructed to separate low, medium and high land, together with a drain (~25 cm x 30 cm) around the perimeter of the watershed. A gated drainage outlet (25 cm diameter) connec ng the watershed to the khal was also installed at the lowest point. The farmers did all the earthworks, but were paid to do so. 2.4 Rainfall and water depth monitoring A standard rain gauge was installed in the vicinity of the pilot watershed and 20 staff gauges were installed on a grid within the watershed. These devices were read each morning between 9 and 11 o’clock and average water depth was determined. 122 2.5 Training on crop cul va on and sluice gate opera on Formal training on the produc on of HYV aman varie es and on drainage management was provided to the farmers prior to the start of the rainy season. Enough seed of HYV was provided to each farmer so that they could grow HYV on all of the land they cul vated in the watershed. Prior to rice harvest training was provided in the cul va on of maize and sunflower and seeds were provided to farmers interested in growing these crops. The farmers were mentored during the rainy and dry seasons, on-farm at mes of key ac vi es such as transplan ng, fer lizer applica on, and sowing of rabi crops. This included guidance on water management, especially drainage in mes of excessive water depth following rainfall, prior to topdressing nitrogen (N) fer lizer, and two weeks prior to rice harvest. 3. Major achievements and lessons learned 3.1 Year 1 (July 2012 to June 2013) 3.1.1 Water management in 2012-13 In 2012, there were two heavy rainfall events in excess of 250 mm each, the first one during 8-14 August, immediately a er transplan ng, and the second during 3-5 September. Both events led to flooding of the en re polder including the pilot watershed area. The farmers successfully drained the excess water within three to four d on both occasions in collabora on with the WMG officials, whereas other parts of the polder were waterlogged for seven to ten d. At the me of the first heavy rainfall the HYV seedlings were only ~15 cm tall and were completely submerged. The mean water depth in the pilot watershed was ~23 cm on 6 September 2012 following the second heavy rainfall (Fig. 2a). At this stage the height of the HYV plants was 20 to 25 cm and thus they were almost completely submerged. The water depth in the pilot watershed was lowered to ~11 cm by 9 September 2012. As a result of rapid drainage on both occasions the HYV rice in the pilot watershed was not damaged. However, the rice crops in some low-lying areas of the polder were damaged and many farmers had to re-plant using older seedlings, increasing their produc on cost and decreasing yield due to the use of old seedlings. Although the farmers were very efficient in handling the drainage of floodwater during the aman 2012 crop they were unable to implement the terminal drainage plan in 2012 because approximately 50% of the farmers planted long-dura on tradi onal rice varie es instead of HYV. The plan was to drain the rice field in the first week of November. But at that me the tradi onal rice was only in the early grain filling stage and instead of draining, more water was brought into the field (Fig. 2a) to finish the tradi onal rice crops that were sca ered across the area. As a result, soil drying was delayed and most farmers could not establish rabi crops un l late February/early March 2013. In the 2012-13 dry season the sluice gate was not closed once salinity of the Kazibacha River started to increase beyond 4 dS/m. This decision was made by the WMG in order to prevent silta on of the intake canal because of a lack of volunteers to de-silt the canal at the end of the dry season. As a result, the salinity of the water in the khal beside the pilot watershed gradually increased to 20 dS/m in mid-May as the salinity of the river increased (Fig. 3). Therefore, the farmers could not irrigate the dry season crops. 123 250 a) Rainfall 2012 200 Water depth 2012 Target drainage me–but irriga on because local variety, late maturing 150 N Topdressing 100 50 0 b) 120 Rainfall 2013 W ater depth 2013 100 80 N Topdressing Terminal Drainage 60 40 20 0 Fig. 2. Rainfall and water depth in the pilot watershed in the aman seasons of 2012 (a) and 2013 (b) showing drainage for N topdressing (blue arrow) and terminal drainage (red arrow). 24 20 Canal River 16 12 8 4 0 1-Dec 1-Jan 1-Feb 1-Mar 1-Apr 1-May 1-Jun 1-Jul Fig. 3. River and canal water salinity (electrical conduc vity, EC) at Kismat Fultola during the 2012-13 rabi season. (EC meter was out of order during the 2013-14 dry season.) 124 3.1.2 Crop management and performance in 2012-13 The farmers transplanted rice (HYV and tradi onal) in mid-August, within the recommended me, using 30 to 35 d and 30 to 40 d old seedlings of HYV and tradi onal varie es, respec vely (recommended HYV seedling age is ~30 d). Those who grew HYV adopted wider row spacing than recommended (20 cm x 20 cm or 15 cm x 25 cm) and applied no or very li le fer lizer. Therefore, yield of HYV rice as reported by the farmers was low (~3 t/ha) and similar to that of the tradi onal varie es. The HYV rice was harvested in the second half of November and the tradi onal rice was harvested in the second half of December. As a result of delayed drainage and harvest of tradi onal rice, rabi crop establishment was delayed un l the second half of February/early March 2013. About 60% of the farmers cul vated tradi onal sesame (broadcast on plowed land) with no fer lizer input or management other than weeding. The rest of the farmers cul vated improved varie es of sesame, maize, sunflower and mungbean. The project provided good quality seeds of the improved rabi crops and the farmers bought and applied fer lizer, but at about half the recommended rate. While most rabi crops were sown late, two farmers were able to establish sunflower on 1 January 2013 by dibbling into the moist soil. But despite moderate to good growth and development of the crops, they did not take proper care of the sunflower crops (no fer lizer, irriga on or weeding). These crops were harvested before the end of April with a yield of 1.4 t/ha (about half the expected yield with proper management). Ini ally, growth of the late sown crops was poor due to lack of rainfall and inability to irrigate, but all the crops (sesame, mungbean, sunflower and maize) recovered strongly a er rainfall in mid-March (~22 mm) and mid-April (~70 mm) of 2013 (Figs 4a; 5a,b). Just when the farmers’ hopes for a good harvest were rising, a cyclone developed in the Bay of Bengal and cyclone Mohasen made landfall in mid-May, bringing huge rainfall (420 mm). As a result, the en re polder including the pilot watershed was flooded and all the rabi crops (except the early sown sunflower in the pilot watershed, which had already been harvested) were damaged or destroyed due to waterlogging (Figs 5c,d). Only two farmers near the drainage outlet were able to drain their late sown maize and sunflower and harvested 5.4 and 1.5 t/ha, respec vely. Yield of both these crops was low as the maize farmer applied only about 50% of the recommended fer lizer and the sunflower farmer did not apply any (although weeding and irriga on were done, with water carried from the nearby hand pump in an earthen pitcher). Some farmers picked mature sesame pods from the waterlogged fields—about 42% harvested an average of ~340 kg/ha of poor quality sesame (less than half of the harvest in a good year, 700-1000 kg/ha), while others abandoned the crops, having invested about Tk. 11,000 per ha (total produc on cost) (Fig. 6). Thus, the poten al crop produc on benefits through improved management of water could not be demonstrated for both the aman and the rabi crops in the first crop cycle (July 2012 to June 2013). 70 140 (a) Rabi 2012-13 60 120 50 100 40 80 30 60 20 40 10 20 0 0 1-Dec 1-Feb 1-Apr 1-Jun 1-Dec (b) Rabi 2013-14 1-Feb 1-Apr 1-Jun Fig. 4. Rainfall in the pilot watershed in the rabi seasons of 2012-13 (a) and 2013-14 (b). 125 a) b) c) d) Fig. 5. Early growth of late sown mungbean and sesame (a) and sunflower (b); following the cyclone, flooded sesame (c) and sunflower (d) fields in the pilot watershed in 2013 (destroyed by rain prior to harvest). 14000 Produc on Cost 12000 Gross Income 10000 8000 6000 4000 n=52 n=6 Watershed Outside 2000 0 Fig. 6. Average produc on cost and gross income (net income is nega ve due to cyclone damage, not shown in the figure) from sesame in 2012-13 cul vated by the farmers inside (HYV and tradi onal) and outside (tradi onal) the pilot watershed. Ver cal and capped bars indicate standard error. (Tk. 2000 @ USD 25) 3.2 Year 2 (July 2013 to June 2014) Following the experience of the first year, a sub-watershed (1.3 ha) was created within the pilot watershed near the drainage outlet. A farm dyke was constructed with an internal drainage canal alongside the dyke. The six farmers agreed to properly implement the water management and cropping plan during the second crop season (July 2013 to June 2014). Both technical and financial support was provided to all six farmers to cul vate HYV rice on the whole area, followed by rabi crops. 126 3.2.1 Water management There were no high rainfall events in 2013 and water depth in the sub-watershed fluctuated between two and ten cm during the 2013 aman season (Fig. 2 b). The sub-watershed farmers drained out ponded water three mes to topdress N fer lizer (Fig. 2b). Terminal drainage was done as planned on 31 October 2013, about two weeks before harvest. Water depth in the rest of the watershed varied from two to sixteen cm, and terminal drainage was done by the farmers from 23-26 November 2013, later than in the sub-watershed because of the cul va on of tradi onal rice by the majority of the farmers in the rest of the watershed. As in the previous year, the sluice gate was le open during the dry season, meaning that the water in the khal was too saline for irriga on by the me that the rabi crops required irriga ng. 3.2.2 Crop management and performance The farmers of the sub-watershed transplanted HYV aman on 7 August 2013, two weeks earlier than their tradi onal prac ce, and implemented recommended fer lizer and crop management prac ces (BRRI 2012). The crop was harvested on 27 November with an average yield of about 5.0 t/ha (Fig. 7). The pilot watershed farmers outside the sub-watershed and those adjacent to the watershed cul vated both HYV and tradi onal rice using no or very li le fer lizer (<1/4th of the recommended dose) and harvested their crops around mid-January, obtaining 2.6-3.4 t/ha. 6000 Recommended Mangment Farmer Management 5000 4000 3000 n=5 2000 n=10 n=39 1000 0 HYV Tradi onal Fig. 7. Yield of HYV rice in the 2013 aman season under recommended management in the sub-watershed, and yield of HYV aman and tradi onal varie es in the rest of the watershed and adjacent areas. Ver cal and capped bars indicate standard error. During the following rabi season (December 2013 to May 2014), maize, sunflower and wheat were cul vated by the research team in the sub-watershed with the help of the farmers as paid laborers. The crops were established by both conven onal llage with a two-wheel tractor and power ller, and by zero/minimum llage. Maize and sunflower were sown by dibbling into moist soil on 10 December 2013 and following conven onal llage on 04 February 2014. Wheat was established by surface seeding on the moist soil surface and by making furrows with a tradi onal hand plough (achra or kota) on 9 December 2013. Sunflower, maize and wheat were grown with and without rice straw mulch applied immediately a er sowing. The farmers in the main watershed area sowed sesame during the second half of February 2014. The 2013-14 rabi season was excep onally dry, with only 36 mm rainfall from 1 December 2013 to 30 April 2014 (Fig. 4b). As a result, establishment, growth and development of the farmers’ sesame crops were very poor. The sesame was also damaged by waterlogging as a result of 245 mm rain in late May and early June 2014 and the sesame yielded only 0.4 t/ha. On the other hand, despite the drought and lack of irriga on, growth of the early-established maize, sunflower and wheat crops on residual soil moisture in the sub-watershed was reasonable. However, these crops suffered from N deficiency as the top dressed urea was not effec vely used due to lack of topsoil moisture and inability to irrigate following urea applica on. Despite the lack of rain or irriga on, sunflower yield was good (2.8 t/ha) (Fig. 8), revealing its greater drought 127 tolerance than that of wheat and maize. Yield of maize and wheat were 5.9 and 1.3 t/ha, respec vely (Rahman et al. 2015; Islam et al. 2015). There was no effect of mulch on yield, probably because all crops ul mately suffered from water deficit stress. Soil moisture at a depth of 37.5 cm in the mulched plots was higher than in the no-mulch plots throughout the growth period of sunflower, and up to the end of February 2014 in the maize and wheat; a er that both mulched and non-mulched maize and wheat plots had similar moisture content (Rahman et al. 2015; Islam et al. 2015). 8 Mulch No Mulch 6 LSD 0.05 4 2 0 Maize Sunflower Wheat Fig. 8. Yield of rabi crops with and without mulch in the 2013-14 dry season 2013-14 (source: adapted from Rahman et al. 2015; Islam et al. 2015). 4. Major challenges to the success of the community watershed pilot 4.1 Non-adop on of HYV rice and recommended cultural prac ces Although HYV rice seed and prac cal training were provided to all the farmers in the pilot watershed only about 50% of them cul vated HYV rice in 2012. The reasons probably include: (1) lower market price of HYV rice than local varie es in 2012 (HYV prices improved in 2013); (2) tenant farmers have to give two-thirds of their harvest to the land owner while having to bear all cul va on costs; (3) higher labor requirement/cost for transplan ng HYV rice due to closer plant spacing than that used for tradi onal rice, and; (4) the cost of fer lizer (although it is subsidized by the government). Some farmers also felt that fer lizer is bad for the soil and were reluctant to apply synthe c fer lizer, although many farmers noted that soil fer lity is declining and similarly rice yield. Another possible reason for the reluctance to apply fer lizer may be the fact that there were several projects in the area giving the farmers cash or in-kind support, and the pilot watershed farmers may have been wai ng to see what support they could extract from our project. 4.2 Non-availability of llage equipment for early crop establishment Land prepara on is highly mechanized in Bangladesh, and farmers depend on rented two-wheel tractors with a power ller to prepare their land for plan ng. In 2013, the pilot watershed farmers contracted local service providers well ahead of me to plw their land in the dry season. But once the soil was dry enough for llage, the small tractor owners doubled their price because they knew that the farmers wanted to prepare the land quickly to establish the crops as early as possible. So the farmers had to wait for about two weeks un l four-wheel tractors reached the area and did the llage at a lower price. The big tractors usually come from Jessore once there is sufficient land ready for llage. Hence, most farmers have to establish their rabi crops late every year, even if their fields are dry enough for earlier llage. 128 4.3 Non-availability of fresh water for irriga on and N topdressing As described above, the pilot watershed farmers could not irrigate their dry season crops because the sluice gate was le open to prevent sil ng of the intake canal and the water in the khal became too saline for irriga on in mid-February. This also hindered effec ve topdressing of N fer lizer. 4.4 Soil drying Despite early drainage from the rice field of the sub-watershed, drying of the soil was slow, probably due to a range of factors: (1) a high perched water table as the water in the khal was high, and adjacent rice fields were s ll flooded; (2) heavy textured soil with slow internal drainage, and; (3) low evapora on in December/January. For these reasons, the soil reached field capacity later than desired, which delayed plowing and sowing of rabi crops (although the soil in the sub-watershed was ready for llage about two weeks earlier than in the adjacent area). While crops can be established much earlier by dibbling into the moist soil rather than wai ng for the soil to dry to moisture suitable for llage, cracking is a serious problem on these soils. 4.5 Soil cracking Rabi crops can be established by zero or minimum llage (dibbling or crea on of a sowing furrow only), but soil cracking in these soils is severe in the absence of llage. This results in root breakage and crop damage, and increases the rate of soil drying due to exposure of the soil surface in the cracks to the air. Straw mulching reduced the cracking considerably and conserved soil moisture for be er growth (Rahman et al. 2015; Islam et al. 2015), but requires more labor. 5. Conclusions and recommenda ons The possibility of reducing drainage conges on for growing HYV aman, and for advancing and diversifying rabi cropping, has not yet been recognized by policy makers, water management and agricultural extension authori es, and the millions of farming families living inside the polders. The pilot proved that with appropriate opera on of the sluice gate, excess water can be drained from inside a polder during low de within a few days to enable the cul va on of HYV aman. Furthermore, mely drainage shortly before rice harvest hastened soil drying and allowed “early” ( mely) establishment of rabi crops. Early sowing meant that the rabi crops were harvested by the end of April, before the onset of the pre-monsoon rains and the May cyclones. This pilot study also demonstrated that individuals alone cannot successfully modify their cropping system schedule and adopt improved agricultural technologies due to the prevailing hydrology in the polder. Adop on of improved technologies requires synchronized cropping pa erns among farmers within community water management units and community coordina on at a range of scales – within small hydrological units, within a sub-polder based on the catchment area of a sluice gate and finally at the polder scale. Un l this is addressed, farming communi es of the polders of the coastal zone will not be able to adapt and benefit from the green revolu on technologies that much of Bangladesh benefits from. To harness the produc on poten al of the polders, the watershed area of each sluice gate needs to be delineated and hydrologically separated (with gates where appropriate) from other catchments of the polder. Within each sluice gate watershed, smaller water management units need to be defined and hydrologically separated, taking advantage of exis ng infrastructure (roads) and separa ng lands of different eleva on with small levees. Crop and water management need to be synchronized based on the water management units and management of the sluice gate. Further, engagement with polder communi es is needed to test these recommenda ons at a wider scale, with par cular emphasis on empowering communi es to enable them to take advantage of the available opportuni es to increase produc vity and improve livelihoods. 129 Acknowledgements This paper presents findings from ‘G2: Produc ve, profitable, and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food. The authors are grateful for high quality technical assistance from Amal Ray, Lincoln Ray, Tanmoy Ray and Mithun Ray. References Bangladesh Rice Research Ins tute (BRRI) 2012. Modern rice cul va on (in Bangla). BRRI, Gazipur, Bangladesh. Bha acharya, J., Mondal, M.K., Humphreys, E., Saha, N.K., Rashid, M.H., Paul, P.C. and Ritu, S.P. 2015. Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh. These proceedings. Islam, M.K., Amin, M.G.A., Mondal, M.K. and Humphreys, E. 2015. Appropriate crop establishment techniques for rabi crops cul va on in the saline coastal zone of Bangladesh. These proceedings. Mondal, M.K., Paul, P.L.C., Humphreys, E., Tuong, T.P., Ritu, S.P. and Rashid, M.A. 2015a. Opportuni es for cropping system intensifica on in the coastal zone of Bangladesh. These proceedings. Mondal, M.K., Saha, N.K., Ritu, S.P., Paul, P.L.C, Sharifullah, A.K.M., Humphreys, E., Tuong, T.P. and Rashid, M.A. 2015b. Op mum sowing window for boro cul va on in the coastal zone of Bangladesh. These proceedings. Rahman, M.N., Amin, M.G.A., Mondal, M.K. and Humphreys, E. 2015. Residual soil moisture u liza on techniques for improving land produc vity in the coastal zone of Bangladesh. These proceedings. Ritu, S.P., Mondal, M.K., Tuong, T.P.,Talukdar, S.U. and Humphreys, E. 2015. An aus-aman system for increasing the produc vity of a moderately saline region of the coastal zone of Bangladesh. These proceedings. Saha, N.K., Mondal, M.K., Humphreys, E., Bha acharya, J., Rashid, M.H., Paul, P.L.C. and Ritu, S.P. 2015. Triple rice in a year: Is it a feasible op on for the low salinity areas of the coastal zone of Bangladesh? These proceedings. 130 How successful are community-led organiza ons for water management? Evidence from an assessment of Water Management Organiza ons in Coastal Bangladesh N. Kenia and M.-C. Buisson 1 Interna onal Water Management Ins tute, India, n.kenia@cgiar.org, m.buisson@cgiar.org Abstract In Bangladesh, water is abundant and o en excessive during the monsoon season and scarce during the dry season. These issues are exacerbated in the coastal zone, which is subject to dal and saline intrusion and highly vulnerable to climate change. Combined, these factors make water management one of the most impera ve issues for Bangladesh. Polders were ini ally created in the 1960’s to prevent flooding and intrusion of saline water. Implementa on of a successful management system to deal with opera on and maintenance of the polders is required for their long term sustainability. In 2001 the Government of Bangladesh (GoB) decided to establish Water Management Organiza ons (WMOs) to facilitate community-based natural resource management, collec ve decision-making and par cipa on in water management. In this ar cle, we discuss and analyze the success of these WMOs in the coastal zone of Bangladesh. On the basis of a literature review and qualita ve insights, we iden fy criteria used to assess the WMOs: par cipa on, transparency, financial accountability, legi macy and ac vi es carried out in terms of water management (opera on and maintenance). Then we analyze these criteria using qualita ve and quan ta ve data from extensive fieldwork conducted in 44 villages across the coastal zone in loca ons with formal and informal water management ins tu ons. The in-depth analysis underlines the lack of legi macy, interest and considera on given to these community-led organiza ons by most households. However, we also establish that most influen al households, larger farmers and WMO members frequently overrate the success of these organiza ons that de facto serve their interests. We conclude by ques oning the sustainability of these organiza ons and their role in promo ng equity in water management. Finally, we suggest to formalize the role of local government ins tu ons (Union Parishads) in water management for a more sustainable and legi mate governance framework and to apply a targeted approach for more vulnerable segments of the communi es in order to promote equity in natural resources management. Keywords: community based natural resources management, water management, coastal Bangladesh 1. Introduc on Bangladesh occupies one of the largest deltas in the world. The whole country is crisscrossed by creeks, canals, and rivers including the Ganges, the Brahmaputra and the Meghna. In this part of the world, water is abundant and o en excessive during the monsoon and scarce during the dry season. Managing water resources in Bangladesh is consequently a major challenge. This is especially true in the coastal zone where the challenges are exacerbated by dal surge-induced floods and salinity intrusion. In response, embankments were built throughout the coastal zone in the 1960s and ‘70s. An addi onal aim of the embankments was to increase agricultural produc on1 and to develop the area, s ll the coastal zone remains locked in a poverty trap. Indeed, of the 8 million inhabitants of the polders2 85% of rural households live under the na onal poverty line (Tuong et al. 2014). 1 Ini ally only a single crop—low yielding paddy—was cul vated in the coastal zone. By controlling water intrusion, polders allowed for increased yields and the cul va on of a second or even a third crop (sesame, boro paddy, pulses or oilseeds) in some loca ons. 2 Polders are low-lying lands that are surrounded by rivers and have been ar ficially isolated through enclosure by embankments. 131 With water being so central in driving economic ac vi es (mostly for agriculture and aquaculture) and in the day-to-day livelihoods of coastal communi es, water management is one of the most impera ve issues for the coastal zone of Bangladesh. Dealing with water management in this complex environment is akin to dealing with conflic ng interests: between agriculture and aquaculture, between lower and higher land, between the landless and landowners or even between produc ve and domes c water uses. Due to the specifici es of the ecology of coastal Bangladesh, the issues around quan ty, quality, ming and targe ng of water access are much more complex than in any part of the world. Policy has consequently played a central role in crea ng a framework around which to organize water management. The Na onal Water Policy (MoWR 1999) was an a empt to link various ins tu ons and mainly considered cross-sectorial coordina on (Dewan 2012). However, the policy suffered from an urban focus and did not specify which ins tu on would be responsible for water issues related to agriculture or aquaculture in the rural areas. Thus, the Na onal Water Policy created areas of ambiguity amongst various ins tu ons. Then in 2001 the Guidelines for Par cipatory Water Management (MoWR 2001) prac cally defined how the policy should be implemented. The ins tu onal framework gives importance and responsibili es to local stakeholders and especially to coastal communi es for organizing themselves and leading decision-making over water infrastructures. Local stakeholders are supposed to par cipate into water management by forming Water Management Organiza ons (WMOs). In polders, WMOs were created based on a three- er system comprising Water Management Groups (WMGs) – the smallest hydrological or social unit at the village level, Water Management Associa ons (WMAs) and Water Management Federa ons (WMFs). These organiza ons are given the responsibility for opera ng the gates that flush water in or out between the polder and the river via canals. In addi on, these organiza ons are responsible for minor maintenance over the infrastructures (canals, gates, and embankments). Implemen ng agencies including the Bangladesh Water Development Board (BWDB) and Local Government Engineering Department (LGED) are supposed to finance major maintenance. Our purpose in this ar cle is to understand to what extent these community-led WMOs have been successful in coastal Bangladesh. The following sec on presents the sources of the data on which the present analysis is anchored. Our objec ve ques ons the criteria used to assess the success or lack of success of these organiza ons. We exploit qualita ve insights and the rich, available literature to iden fy five criteria that will be used to assess the community organiza ons. The selected criteria are presented in sec on 3. Then, based on a quan ta ve survey conducted among 1000 households from 44 villages across the coastal zone, we analyze community organiza ons on the criteria in sec on 4. In the fi h part of the ar cle, we propose an overall ra ng and analyze the differences. Finally the conclusion underlines the main findings, ques ons the sustainability and the equity of the community organiza ons for water management and suggests implementable policy recommenda ons. 2. Methodology and data sources For the purpose of this analysis, we use two complementary sources of data: qualita ve data from one side and quan ta ve data from the other side. Learnings from the qualita ve data together with the literature review helped us to iden fy the criteria considered as essen al for the success of the WMOs. Then quan ta ve data allow us to measure to what extent these criteria are sa sfied. Qualita ve data were collected in 2012 in five polders (polder 3, polder 30, polder 43-2F, polder 31, polder 24G) covering different agro-ecological zones (low, medium and high salinity zones) and four sub-projects of the coastal zone (Jabusa, Jainka , Latabunia, Bagh-Anchra Badurgacha). These loca ons were purposely selected with the aim of covering different degrees of salinity and different ins tu onal set ups. The exact loca ons of the data collec on were selected a er analyzing maps of the area, transect walks and consulta ons with the communi es. Qualita ve data collec on took the form of Focus Groups Discussions (FGDs) and Key Informant Interviews (KIIs). In both cases the discussions followed a structured outline. FGDs were conducted with general groups of community members, WMO members and women. KIIs targeted farmers, women headed households, Labour Contrac ng Socie es’ group members, gatemen, Union Parishad 132 (UP) members, WMO members, and BWDB and LGED officials. In total, transcripts were recorded for 57 FGDs and 92 KIIs. The transcripts were coded using Atlas Ti and the frequency of the codes is used in the following sec on to highlight the predominance of some criteria as perceived by the community members. The assessment of community WMOs is also based on a quan ta ve survey conducted in the same loca ons of coastal Bangladesh in January 2013. A sample of 1000 household from 44 villages in three polders and three sub-projects were surveyed. Villages were randomly selected in the three surveyed polders (polder 3, polder 30 and polder 43-2F). Thanks to this random selec on, some of the villages selected are located close to the canals and the gates whereas others are located in the middle of the polder, with different water access and resources. The three sub-projects (Jabusa, Jainka , Latabunia) selected were purposely chosen to represent different agro-ecological zones. Figure 1 presents a map of the coastal zone and the loca on of the study areas. With this sample we gathered informa on on different community-led WMOs. The par cipatory policy for water management men oned above is implemented by two major implemen ng agencies and by the projects they lead in partnership with the Government of Bangladesh (GoB) and donors. The BWDB is responsible for schemes of up to 5000 hectares and therefore leads the work conducted in the polders. The Integrated Planning for Sustainable Water Management (IPSWAM) project and its successor Blue Gold, started in 2013, are the most important projects implemented by BWDB in the coastal zone for community management of water. These two projects put the policy into prac ce and create Water Management Groups in the loca ons where they operate. In total IPSWAM created 242 WMGs from 2003 to 2011 (EKN and BWDB 2011). Formed in 1995, LGED is in charge of areas of less than 1000 hectares 3 and led the Small Scale Water Resources (ADB 2007) project funded by the Japan Interna onal Coopera on Agency and Asian Development Bank. The project completed 280 sub-projects in its first phase and 300 in its second. This project works outside of the main polders and provides infrastructure work and community organiza on to communi es located outside of the main embankments. The project also follows Bangladeshi policy and takes a par cipatory approach through the crea on of Water Management Coopera ve Associa ons (WMCA). Even if the implemen ng agencies and projects apply the government policy in the loca ons where they work, large spans of the coastal zone are le without implementa on of the policy. For example, IPSWAM worked in only nine of the 123 polders and to date Blue Gold works in 12 polders. As a consequence, in spite of the exis ng policy that creates a framework and pushes for a par cipatory approach to water management, informal ins tu ons s ll govern water management decisions and opera ons in many loca ons4. Therefore, three different types of community ins tu ons for water management can be found in our sample. WMCAs were found in the sub-projects, whereas in former IPSWAM project areas (polder 30 and polder 43/2F) we met with WMGs. Finally, the third type of organiza on, iden fied mostly (but not only) in polder 3, can be considered to be informal organiza ons taking the form of gher commi ees, gate commi ees, beel commi ees or any other informal groups managing water in the area. All three types of groups are community-led WMOs; this assessment will consequently consider the different forms of WMOs and the differences between these three types of ins tu ons. Table 1 summarizes the types of ins tu ons surveyed in each loca on. 3 These small areas are usually called sub-projects whereas the word ‘polder’ refers to larger areas created in the 1960s and ‘70s when the embankments were built. 4 Before the 90s gatemen were appointed and paid by BWDB throughout the coastal zone, the role played by communi es was reintroduced in the 90s a er the system of state employed gates operators disappeared. 133 Table 1. Number and type of ins tu ons surveyed, by loca on Type of WMO Polder 3 Polder 30 Polder 43/2F Latabunia Jainka Jabusa TOTAL 4 4 WMCA WMG 12 WMA 12 24 1 1 Gate commi ee 2 2 Gher commi ee 4 4 Other informal commi ee 8 Total 14 1 12 9 14 4 44 The survey instruments included a household ques onnaire and a WMO ques onnaire. The household survey targeted 20 households per village in polders and 40 households in each sub-project. The households were randomly selected using a walking transect method in the villages. The WMO ques onnaire was answered in each surveyed village by members of the community-led organiza ons responsible for decisions regarding water management in the village. As a result some respondents were members of formal WMOs while others were not. The WMO ques onnaire contained five sec ons and the objec ve was to gain perspec ve about the func oning of the organiza on, ins tu onal structures and financial accoun ng that was taking place in the organiza on. Similarly, ten sec ons were included in the household ques onnaire with one sec on focusing on water management. In the analysis below we use data from the WMO ques onnaire as well as informa on collected individually with the household ques onnaire. This allows us to get a balanced perspec ve that includes the points of view of organiza on members and non-members. The data used in the below analysis also combine responses to behavioural ques ons and to a tudinal ques ons. A tudinal ques ons were used to collect opinions and beliefs of the respondents regarding WMOs. These responses are based on the percep ons of respondents and can therefore be highly subjec ve. In order to limit varia on in the way the a tudinal ques ons were understood, enumerators had to read the defini on of the criteria that respondents were asked to rate. In addi on, a Likert scale was used for the responses and the scale was shown on a piece of paper in order to help the respondent. Stu dy Are a M ap o f Pro je ct G3 Lege nd District Boun dary Division Bound ary Inte rnat io nal Bound a Th ana Bou ndar y Selected Po ld er fo r Study Polder Bo undar y River District HQ Th ana HQ Study Are a N 20 10 0 20 K ilomet ers D:\P0 0 2 5\5 1 4 09 _ Wa te r_ G ov e rn an c e \M XD Fi l e \ Stu d y Are a _G 3 .mx d [*mtn *1 1 03 1 2 ] Fig. 1. Map of the study areas (source: Ins tute of Water Modelling for the CGIAR Challenge Program on Water and Food G3 project). 134 3. Iden fica on of criteria to assess water management organiza ons The criteria for assessing the success of WMOs in coastal Bangladesh were ini ally based on insights from the qualita ve data and are here further jus fied by the way they are considered in the literature on community natural resource management. As men oned by Bardhan (2000), the literature on natural resource management and community coopera on for irriga on or water management remains largely led by social and anthropological empirical studies. We dig into this large body of literature with the addi on of a few exis ng quan ta ve analyses to iden fy the criteria to be used. The success of these community-led organiza ons depends first on the objec ves they were given. Here we refine the defini on of success from the Bangladeshi policy that assigned roles to WMOs. As per the Guidelines for Par cipatory Water Management (MoWR 2001), the main objec ve of those guidelines, both immediate and long term, was to improve and develop the par cipa on of local stakeholders in water resource management and to achieve sustainable par cipatory water management. If par cipa on is clearly the first aim of Bangladesh water management policy, it is also one of the main concerns of the community members. Hence, in the qualita ve transcripts, the first topic discussed was the par cipatory process employed at the forma on of the organiza on 5. The focus on par cipa on emerged again while discussing conflicts in the community and was even the second cause of conflict men oned6. Par cipa on emerged as an essen al indicator of success from the qualita ve data and was also frequently assessed in the literature, with ques ons regarding the true par cipatory nature of par cipatory water management both in Bangladesh (Sultana 2009; Dewan et al. 2014) and in other contexts (Cleaver 2001). While the par cipatory process is considered to be essen al for establishing sustainable water management prac ces(Thompson et al. 2003), s ll too o en par cipa on remains imposed by donors and par cipatory management fails to achieve a high degree of control over the decisions and resources of communi es. With par cipa on also comes the equitable distribu on of benefits. This perspec ve assumes that the views of all community members with conflic ng interests are considered. In an impact evalua on from Peru that employed a spa al regression discon nuity design, Datar (2009) studied the equity impact of irriga on rehabilita on projects with a community-led component and found significant and posi ve impacts on the poorer beneficiaries. But more o en, the equity of programs that transfer water resource management responsibili es to communi es is ques oned (Social Development Department 2008; Dewan et al. 2014). The degree of par cipa on of local stakeholders in these community-led WMOs will therefore be one of our criteria. In the analysis below, we will consider par cipa on at the crea on of the WMO as well as the inclusion of all the stakeholders in the decision making process on a day-to-day basis. Governance of the organiza on is also key for explaining the level of success of WMOs and for understanding their sustainability. Governance refers to the rules that guide the func oning of the organiza on. These rules include the elec on process, decision making, par cipa on in mee ngs, the way water should be allocated, solving disputes (Haider 2009; Gain 2012) and the collec on and alloca on of resources. We will consider two governance-related criteria: transparency and financial accountability. The transparency of an ins tu on refers to the degree of informa on that is shared by the ins tu on. In our case, this concerns the level of publicity for decisions or mee ngs that have taken place. Financial accountability refers to the policy organiza ons adopt to meet their responsibility for ensuring an efficient and fair collec on and alloca on of financial resources. These governance issues were also highlighted during our qualita ve discussions with groups and key informants, appearing under different topics including conflicts, maintenance and ins tu ons7. 5 6 7 Atlas Ti code: ‘PROJECT_Process_forma on_WMG/WMCA’, frequency 84, 1st code related to WMOs Atlas Ti code: ‘CONFLICT_Par cipa on:exclusion_non-elites’, frequency: 135, 2nd code related to conflicts. For example: Atlas Ti code: ‘BWDB_percep on: nega ve: corrup on’, frequency 82, 1st code related to implemen ng agencies. 135 Another body of literature assessing community-led natural resource management considers the linkages between the ins tu ons in charge of water management and other ins tu ons, and par cularly local government elected bodies (Shackelton 2002). Even if the idea that the water sector is part of a broader framework that includes the social, poli cal and economic spheres is widely recognized (Social Development Department 2008), the role of poli cs and of local government ins tu ons is s ll discussed. This also relates to acknowledging that water management is a poli cal process (Mollinga 2008), from the day-to-day poli cs of conflic ng interests regarding gate opera on to the global poli cs of water and environment. Duyne (1998) similarly underlines the existence of poli cs as well as informal ac vi es for suppor ng the water management needs of the coastal communi es of Bangladesh. The role of local government ins tu ons in community-led natural resource management organiza ons has been posi vely reviewed for community-based fisheries management (Thompson et al. 2003; Rab 2009). In those cases the link between community organiza ons and local government ins tu ons has been iden fied as a tool to sustain the organiza ons in the long-term, especially a er project withdrawal. Also, in an impact evalua on led by the Independent Evalua on Group of the World Bank on Andhra Pradesh irriga on projects (Social Development Department 2008), the authors conclude that it is necessary to involve local governments to ensure con nuity in water management beyond the community-led organiza ons. However, local government ins tu ons can also be seen as compe ve actors that can weaken the legi macy of the community-led organiza ons and may bring poli cal interference and elite capture. Local government ins tu ons, and especially the Union Parishad, are a subject that has been largely covered in the qualita ve data. Among the 693 quotes related to Union Parishads in the transcripts, 68% can be considered as posi ve and support an increased role for this ins tu on in terms of coordina on, supervision and funding of water management. We will consequently consider the legi macy of the community-led water management organiza ons that exist in coastal Bangladesh as well as their interac ons with other formal and informal ins tu ons, including local government ins tu ons, as one of the assessment criteria. Finally, the ac vi es of the WMOs must be considered, namely opera on and maintenance. The way in which opera ons are conducted, the existence of conflicts related to these opera ons and the quality of infrastructure maintenance are commonly considered as outcomes of the community-led management of natural resources. De facto, the transfer of responsibili es from central governments to communi es for opera on and especially for maintenance happened at a me where the limited resources allocated by the governments resulted in infrastructure deteriora on ( Shackleton 2002; Heer and Jenkins 2012; Dewan et al. 2013). Hence, the ability of the community-led organiza ons to perform these ac vi es indicates their ability to succeed. For example, in a qualita ve analysis Bardhan (2000) considers the role of community-led organiza ons in coopera on. He measures coopera on by the quality of the maintenance, the absence of conflicts and the respect of water-alloca on rules. These three indicators clearly decipher the quality of the opera on and maintenance ac vi es carried out by WMOs. In another case study from the Philippines, Bandyopadhyay (2007) assesses the impact of irriga on management transfer on revenue collec on for irriga on and iden fies the significant impact of community-led water management on maintenance ac vi es. The importance of opera on and maintenance criteria are clearly confirmed by the FGDs and KIIs; these two topics gathered the most number of quotes8 and are in the top five codes related to WMOs. Consequently, the last criteria for our assessment is the ac vi es in which the WMOs are involved and especially their opera on and maintenance performance. 8 Atlas Ti codes: ‘OPERATION’, frequency 616; ‘MAINTENANCE’, frequency: 1005. 136 4. Assessing community-led water management organiza ons 4.1 Par cipa on One of the vital factors for crea ng sustainable WMOs is par cipa on by all stakeholders. A dynamic and integrated approach should be designed such that each and every member is a part of this group and has the opportunity to par cipate in all decision-making processes. From Figure 2 we can observe that at the me of WMO crea on, the level of par cipa on was high. During this period donors were s ll ac ve and they ini ally involved all the community members in the organiza on. However, we can clearly see that there is barely any par cipa on in the informal set-up. People organized themselves and the larger gher- and landowners took the ini a ve for the opera ons of the gates and maintenance of the embankments. This is not the most viable solu on, however, as in many cases small and landless farm owners suffer due to intrusion of saline water. What is also seen in most of the organiza ons surveyed is that the ini a ve for crea ng the group was taken by external bodies (BWDB, LGED donors, NGOs) rather than by the villagers themselves. This calls into ques on the par cipatory process and the sustainability of the involvement of community members in these organiza ons. The par cipa on of specific groups and par cularly more marginalized groups is also important to consider. In polders 30 and 43/2F, where all the WMOs surveyed are formal and were registered to the coopera ve department, the ra o of male to female members is 3:1 as per the guidelines (MoWR 2001). However in most villages women are primarily only figureheads, as men make most of the water management decisions. Further analysis reveals that female WMO members of are not representa ve of all women in the communi es and do not consider the needs and interests of women in decision making related to natural resources management (Buisson et al. 2014). Fig. 2. Percep on of par cipatory process at the crea on of WMOs (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 44 villages selected in polders and sub-projects). Qualita ve and quan ta ve data also reveals that the WMOs struggle once donors or projects end their support. The expecta on is that the community organiza ons will func on independently. This can be seen in Figure 3: six years a er the crea on of the WMOs, the par cipa on level has declined steeply. At this me, nearly half of the organized WMOs don’t conduct any mee ngs during the year. Around one-third of them claim to have conducted mee ngs once in a quarter but this could be a formal rou ne with a small number of par cipants and without inclusive discussion or any construc ve ac vity. Amongst the informal commi ees no such mee ngs are held. Indeed, since the gher or large farm owners have poli cal influence they maintain their decision-making authority and dominate decisions regarding opera on and maintenance without par cipa on from other stakeholders. In the sub-projects, the WMO ques onnaire indicates that mee ngs are conducted every quarter, but qualita ve insights also suggest a decline in interest in these organiza ons over me. 137 Finally, par cipa on can be considered through the number of members, trends in membership or the elec on processes of WMOs. We note that the number of members in the commi ees ranges from 20 to 25 per village, but in some villages there are more than 100 members. Most of the WMO member respondents men oned that over the past few years the number of members has remained the same, but this has been made possible only because of the lack of payment of annual membership fees. Considering elec ons, vo ng has taken place recently in only a few villages, which reflects the lack of renewal in execu ve commi ees. When elec ons did occur, there was no contest for the posi on of president or secretary; this again points to the inac ve status of most of these organiza ons and the lack of long-term par cipa on. Fig. 3. Number of mes organiza ons held mee ngs in 2012 (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 44 villages selected in polders and sub-projects). 4.2 Transparency We next consider transparency and note that the overall ra ng for transparency remains very low in all polders and sub-projects. It also remains low among the households regardless of the size of their farms, or whether household members are members and non-members of the WMO. According to our findings, there is no formal mode of communica on between various actors such as community members, local government ins tu ons, LGED or BWDB. This means that large amounts of poten ally relevant informa on are unavailable in the public sphere. This could include mes for opening and closing of gates, publicizing decisions taken and the agendas of public mee ngs to name a few. FGDs reveal that many community members could feel excluded from water management and their faith in community-led ins tu ons such as WMOs has eroded over me. One prac cal consequence is that the community’s willingness to contribute (through money or volunteering work) declines. This is in turn has a direct effect on the opera on and maintenance of the embankments, canals and the gates. In Figure 4 we can see that the transparency ra ng from WMOs members is neither good nor bad, sugges ng that members themselves don’t have much faith in the organiza on to which they belong. Fig. 4. Percep on of transparency by WMO respondents (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 44 villages selected in polders and sub-projects). 138 4.3 Financial Accountability The difficulty of collec ng informa on on income and expenses from WMOs gives an idea of the financial accountability situa on. It further denotes the lack of informa on and some mes the lack of ac vity in these organiza ons. There are no WMOs registered in polder 3 and hence there are no accounts and no organized structure. However, most of the WMOs in polders 30 and 43/2F are registered and consequently have savings accounts. In addi on, some villages also have accounts dedicated to maintenance or other ac vi es. But the existence of an account is not a guarantee for solving the issues of these organiza ons or for maintaining ac vi es. Indeed, most of these accounts are non-func onal or have become dormant. Even in polder 30 and polder 43/2F, only a few villages have registered some expenses for the year. This confirms that in villages in which organiza ons are present, community members are too o en unwilling to contribute to the cost of water management opera ons and rather expect support from donors or projects. Fig. 5. Who makes the decision on the expenses in the WMO? (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 44 villages selected in polders and sub-projects). However, in polder 3, in spite of the absence of formal organiza on, it is of interest to note that communi es have indeed spent money on maintenance. In those loca ons, influen al people or large gher owners take charge and take responsibility for maintaining the gates. Although there is no official record of bank accounts, the maintenance ac vi es are s ll carried out and supported by contribu ons given by these gher owners. In terms of finance-related decisions, we note that generally, decisions are made jointly by the execu ve commi ee members in polders 3 and 43/2F, whereas in polder 3 a single person makes the decisions (Fig. 5). In the sub-projects, all three WMCAs are registered as coopera ves but the case of Jabusa is worth highlight. The WMCA of Jabusa is quite financially ac ve9. The community organiza on has a savings account, a maintenance account and an emergency account into which money is deposited. This financial ac vity can be linked to the ability of the organiza on to receive income, mainly from canal leasing but also from interest from microcredit, interest from savings and from a grant from LGED. All the records are well maintained in the WMCA office and are said to be open to all members. In Jainka and Latabunia, the WMCAs have savings accounts into which money is deposited but the amounts remain very limited due to the lack of regular contribu ons and the absence of addi onal sources of income. Finally, in all the surveyed WMCAs, all elected members of the execu ve commi ee take decisions on contribu ons or resource alloca on collec vely. Hence, amongst the project and sub-projects, the la er have been more financially ac ve and their accoun ng systems seem more sustainable. 9 This was at least the case in 2012; since then conflicts have started hindering the ac vi es of this WMCA. 139 4.4 Legi macy The legi macy of an organiza on is a func on of the reliance that people place on it for solving their problems. In this regard, the success of community-led models is largely limited. While analyzing the data collected we realized that the powers held by the communi es are rela vely limited compared to those of local government ins tu ons. Interes ngly, the legi macy given by community members to WMOs—supposedly community-led organiza ons—is lower than the legi macy given to other ins tu onal actors. When household respondents were asked which ins tu on they would go to first to solve a water management issue they were least likely to approach a WMO to resolve their problems and instead rather chose to go to Union Parishads. These results are displayed in Figure 6. Respondents located in areas where informal commi ees are in place for water management ma ers (polder 3) are most inclined to approach Union Parishads and only around 21% of them consider community members capable of solving these issues. But the scenario is not so different in loca ons with formal WMOs. Where WMGs are in place, most of the respondents (37%) would first approach BWDB. Where WMCAs are in place, respondents indicated they would first approach LGED. In no loca ons were community organiza ons considered legi mate actors for solving water management issues; less than 3% of the respondents choose WMOs as the organiza on they would contact first. These results underline the inability of WMOs to establish themselves as legi mate ins tu ons for solving water management issues and leading water governance; the households who should be the beneficiaries of these ins tu ons clearly don’t recognize their usefulness. On contrast, the legi macy of Local Government Ins tu ons, especially of Union Parishads, and the level trust accorded to them by community members is a robust result throughout the different polders and sub-project considered in this analysis. Fig. 6. Which ins tu on would you contact first to solve a water-related issue? (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 1000 households selected in polders and sub-projects) 4.5 Ac vi es There are various ac vi es required to ensure that an organiza on runs efficiently. In our case there are two main ac vi es that take place in the polders and sub-projects: maintenance and opera on. Maintenance applies to the canals, embankments and gates whereas opera on mainly applies to the gates. Below we consider the ac vi es undertaken on these three types of infrastructure. First, annual maintenance ac vi es on the canals are rarely observed. In polder 30, polder 43/2F, Latabunia, Jabusa and Jainkathi, the quality of the canals is especially poor and these water bodies are heavily silted. There is no work or money invested in these canals. Indeed, most of the villages surveyed claim that there has 140 been no re-excava on or de-silta on in the canals for decades. Some others men oned that the last re-excava on occurred in 2009. Thus the state of the canals is worsening year a er year. If one argues that canal leasing may be a way to collect funds for improving the maintenance of the canals, the case of Jabusa does not support this assump on. Despite leasing the canal for several years and receiving substan al money from the leasing contract, the main canal of Jabusa is not well maintained and, on the contrary, leasing has contributed to segmenta on of the canal and flow disrup on. Annual gate opera on in these polders tells a similar story. Opening and closing of the gates remains one of the most controversial and debated issues in the polders. The conflicts occurring between shrimp culture and agriculture, small and large farmers, and low and high land farmers are due to issues coordina ng water quan ty, quality (salinity) or ming. The ming of gate opening and closing is supposed to be decided jointly by all the members of the WMO but more o en than not the decision is made by influen al people to suit their own needs. It is common for the gates to be operated by a gateman who is formally or informally appointed by the decision makers. When formal WMOs exist, the execu ve commi ee of the WMO appoints the gateman. But in the case of polder 3, gher owners select the gateman themselves. The gateman generally lives near the gate and is then compensated in kind or in cash (as is the case in polder 3). Fig. 7. Who makes decisions regarding the opera on of the gates? (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 1000 households selected in polders and sub-projects) Lastly, embankments are the pillars of the polders and need constant maintenance to prevent water intrusion into the polder. But, despite being so essen al, the amount of me and money invested in embankment maintenance is inadequate. Across polders and sub-projects, only half of the surveyed villages have carried out maintenance ac vi es on their embankments. In more than half of these villages, no rehabilita on work has been undertaken in the last four years. This lack of maintenance is again related to the absence of contribu ons from community members. In addi on, the act of a few farmers or gher owners digging and adding pipes through the embankment further puts the infrastructure at risk. Fig. 8. Annual maintenance ac vity to the embankments (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 44 villages selected in polders and sub-projects). 141 5. Ranking the community-led water management organiza ons and discussion One way to synthesize the above-men oned results is to consider rankings from the WMO respondents and households for the five indicators. Rankings were included in the WMO and household ques onnaires. Respondents were asked to rate the different parameters on a Lickert-type scale from one to six. One was the lowest value and meant the community-led organiza on completely failed in achieving this objec ve. Six meant that the organiza on was very successful with respect to that indicator. The six10 topics considered were: (i) Par cipa on, to what extent all members were involved in the organiza on when created and are s ll involved in decision making; (ii) Transparency, to what extent community members have access to informa on related to the ac ons of the WMO, to WMOs mee ngs, decisions or rules; (iii) Financial accountability, to what extent the organiza on is able to provide clarity on budgets and expenditures or to what extent the organiza on is free of corrup on; (iv) Legi macy, to what extent the organiza on plays a fair role in conflict resolu on and is able to uphold the rules and decisions related to water management in the community; (v) Maintenance, to what extent the organiza on is able to maintain the state of the water infrastructure and the quality of the infrastructure in the village; and (vi) Opera on, to what extent the community-led organiza on is able to take joint decisions regarding the opera on of the gates and to what extend the decisions sa sfy user needs. We first examine the ranking from the WMO respondents and then from the households respondents. Figure 9 gives an overview of the average ranking for the different polders and sub-projects by WMO respondents. These graphs demonstrate the self-sa sfac on of WMO respondents on the success of their respec ve organiza ons regarding mul ple parameters. Obviously, since the respondents are members of these organiza ons (most of the me, presidents or secretaries), the responses can be biased and driven by a willingness to show the strength of their ins tu ons. In the case of the community-led organiza ons from polders 3, 30 and 43-2F, the average ra ng was between 4 and 3 deno ng a neither good nor bad situa on. The only clear differences is seen in the ra ng for maintenance in polder 43-2F, which seems rela vely be er than in the two other polders. Also, polder 30’s rankings on several criteria (maintenance, opera on, legi macy and financial accountability) are lower than in other loca ons. Considering that most of the respondents from polder 30 were members of the WMG established by IPSWAM, this is somewhat surprising and may translate a difference in expecta ons. By ini a ng community mobiliza on and strengthening the role of communi es in water management, these projects also create new expecta ons among the communi es. Another explana on is related to the status of these organiza ons in polder 30; many were de facto non-func onal at the me of the survey and the respondents were consequently keen to share their real feelings and disappointment regarding these organiza ons. 10 Maintenance and opera on are considered independently for the ranking, we consequently give six indicators instead of the five previously men oned. 142 Fig. 9. Ra ngs by WMO respondents (source: CGIAR Challenge Program on Water and Food G3 project from 44 villages selected in polders and sub-projects). Overall the ra ngs are higher in sub-projects than they are in polders. All the sub-project ra ngs exceed 3, except for financial accountability in Jabusa. Maintenance was considered to be very successful in Jabusa, where large sluice gates are indeed well maintained even if the embankment is weakened by erosion. Interes ngly, Jainka appears to be the most successful among the three sub-projects with high rankings for par cipa on, legi macy and financial accountability. This situa on is consistent with the qualita ve insights. We then consider the same rankings from the household survey. Rankings from Figure 10 are averages from all respondents in each loca on. Our purpose here is to understand the percep on from the households, i.e. from the water users or beneficiaries of the services provided by the water organiza ons in each village. We are also interested in no ng consistencies or differences in the ra ngs given by households and by the WMO respondents. The first observa on is that the ra ngs dropped down to 2.5 on most of the parameters. On average, household members rated each criterion below 3. This clearly indicates that community members are not sa sfied with the func oning of the WMOs. It may also suggest that the members from the organiza ons overes mate the benefits the ins tu ons have on the stakeholders. Fig. 10. Ra ngs by households respondents (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 1000 households selected in polders and sub-projects). From the household ra ngs, the par cular case of polder 3 comes to light. In this polder, the average household ra ng for the six indicators is close to 1.5. The primary reason for this very low ra ng could be the informal way in which water is controlled and the frequency of conflicts. Indeed in this polder, where no large-scale projects have been implemented, what we call community-led water management is actually 143 informal organiza ons led by influen al gher owners who dominate gate opera on. Small and marginal farmers generally do not have any say in the decision making process related to water management in this polder. In polder 30 ra ngs are higher than in the other polders. The fact that IWPSAM was involved in this polder for a long me, set up the organiza ons through a par cipatory process and empowered the community members may help explain this result. Of the sub-projects, Jabusa received the lowest scores. While WMO respondents rated Jabusa highly, a clear difference can be seen with household respondents, who indicated their dissa sfac on with the organiza on and its ac vi es. Since Jabusa is a compara vely large sub-project, at least in terms of popula on, the ac vi es require more vigorous monitoring and evalua on and coordina on may become more difficult to achieve. Also, this is the only sub-project where the WMO respondents acknowledged a poli cal influence on water management. In Jainkathi, household members seemed rela vely sa sfied with the WMO. Finally, intrigued by the difference of perspec ve between the WMO and household respondents, we analyze the responses for different categories of respondents from the household survey. First, we consider members of the WMOs and non-members and second we consider the ra ngs from different land size ownership groups (Figure 11). The results obtained confirm that the WMOs primarily serve the needs of larger landowners, whereas smaller farmers and the landless are commonly excluded. Fig. 11. House respondent ra ng, by WMO membership and land size group (source: CGIAR Challenge Program on Water and Food G3 project quan ta ve survey from 1000 households selected in polders and sub-projects). The same issue appears when comparing the ra ngs of WMO members and non-members. On average, 11.3% of the households from the sample have at least one member in a WMO. These households ranked the WMOs significantly higher than the non-members did for all six criteria, and especially for par cipa on. 6. Conclusion In 2001, the Government of Bangladesh decided to establish Water Management Organiza ons as the main tool for improved water management and for community involvement in this ma er. Following this policy, donors and implemen ng agencies ini ated the crea on of community-led WMOs in their projects, mostly for the renova on of water infrastructure. Water management is a common requirement across coastal Bangladesh, so in loca ons without projects endogenous and informal organiza ons emerged from the communi es to take care of opera on and maintenance. Our purpose here was to understand to what extent these community-led WMOs have been successful in the context of coastal Bangladesh. On the basis of a literature review and insights from qualita ve data, we iden fied the criteria used to assess WMOs: par cipa on, transparency, financial accountability, legi macy and water management ac vi es carried out (opera on and maintenance). Then, we analyzed the indicators using quan ta ve data from extensive fieldwork in 44 villages across the coastal zone in loca ons with formal and informal water management ins tu ons. 144 The in-depth analysis underlines a lack of par cipa on, which declines over me along with the interest and the legi macy that community stakeholders accord to these organiza ons. Both transparency and financial accountability are very poor, indica ng that most of these organiza ons became dormant a er project and donor withdrawal. Finally, elite capture is a widespread issue across the study areas, and not only in loca ons with informal water management. We indeed establish that most influen al households, larger farmers and WMO members frequently overrate the success of these organiza ons that de facto serve their own interests. These results lead us to ques on the long-term sustainability of these organiza ons and their role in promo ng equity in water management. The analysis presented in this paper also indicates that a series of reforms need to be implemented on mul ple fronts to improve water governance in coastal Bangladesh and to find a be er fit between the water management needs of various stakeholders and the way water is controlled and shared. Through the above assessment the paper a empts to shine light on the following first-level policy recommenda ons to tackle. First, there is an immediate need for improved coordina on among stakeholder ins tu ons through implementa on of a clear water governance framework. While basic ins tu ons on water governance at the local community level are already present, clear understanding of their roles and responsibili es is lacking. This is confirmed by the clear lack of legi macy for community-led organiza ons underlined in this analysis. Currently, according to the guidelines (MoWR 2001), the Union Parishads are supposed to provide support to projects and sub-projects by facilita ng effec ve coordina on among the exis ng ins tu ons. However, the lack of specificity regarding their responsibili es limits the span of control of the Union Parishad and curtails their effec veness in performing a coordina on role. This reality is mismatched with the level of trust that community members give these ins tu ons and with the legi macy they benefit from in terms of water management. Giving Union Parishads a formal role to coordinate the actors involved in water management would reduce conflicts and promote more sustainability in the water governance framework. Second, equity is seen as being at risk. This is indicated by the lack of par cipa on from more vulnerable segments of the communi es and from elite capture. This is especially relevant for women, who in spite of the water needs they may have, are excluded from decision making over the management of the resource. While female par cipa on needs to be encouraged in all roles, it is extremely important to tap their inherent strengths and support their par cipa on in roles that make them feel comfortable and thus ensure their greater involvement. For example, women could provide feedback regarding the ac vi es taking place near sluice gates or act as key informants in case any illegal ac vity takes place. Although it is already mandatory for an organiza on to be inclusive, we suggest adop ng a more targeted approach that would reconcile individual water needs with the requirement for coordina on at the community scale. Acknowledgements This ar cle is based on data collected for the project ‘G3: Water Governance and Community-based Management in Coastal Bangladesh’, part of the Ganges Basin Development Challenge of the CGIAR Challenge Program on Water and Food and the CGIAR Research Program on Water, Land and Ecosystems. The authors thank the NGO Shushilan for the data collec on throughout the southwest coastal zone of Bangladesh and the respondents. Camelia Dewan is also thanked for her work in the qualita ve data and especially coding of the transcripts that are used here. References Asian Development Bank. 2007. Small-Scale Water Resources Development Sector Project. Performance Evalua on Report. Opera ons Evalua on Department. World Bank Group. (2008). An Impact Evalua on of India’s Second and Third Andhra Pradesh Irriga on Projects. World Bank Group . World Bank. (2008). The Poli cal Economy of Policy Reform: Issues and Implica ons for Policy Dialogue and Development Opera ons. Social Development Department . 145 Buisson, M.-C., & Mitra, A. a. (2013). Households and Water Management Organiza ons Quan ta ve Surveys. Dhaka, Bangladesh: Challenge Program on Water and Food. Bandyopadhyay, S. S. (2007). Yield Impact of Irriga on Management Transfer: Story from the Philippines. World Bank Group. Bank, A. D. (2007). SSWRDP. Bangladesh: ADB. Bardhan, P. (July 2000). Irriga on and Coopera on: An Empirical Analysis of 48 Irriga on Communi es in South. Economic Development and Cultural Change, Vol. 48, No. 4 , 847-865. Cleaver, F. (2001). Ins tu on, agency and the limita ons of par cipatory approaches to development . In B. K. Cooke, Par cipa on: The New Tyranny (pp. 36-55). London: Zed Books. Cooke, B. a. (2001). The case for par cipa on as tyranny. Embassy of the Kingdom of the Netherlands Dhaka, B. (2011). IPSWAM. Dhaka. Duyne, J. E. (1998). Local Ini a ves: People's Water Management Prac ces in Rural Bangladesh. Development Policy Review Vol 16 , 265-280. Datar, G. X. (2009). Are Irriga on Rehabilita on Projects Good for Poor Farmers in Peru? World Bank Group. Dewan, C. M. (2013). Evolu on of Par cipatory Water Management in Coastal Bangladesh: A review of project and policies. Dewan, C. M. (2014). The Imposi on of Par cipa on? The Case of Par cipatory Water Management in Coastal Bangladesh. Water Alterna ves 7(2) Volume 7 | Issue 2 , 342-366. Dewan, C. (2012). SWOT analysis of Bangladesh’s key water. Dewan, C. (April 2012). Review of the historical evolu on of policies and ins tu ons of community based management in coastal Bangladesh. G3: Water Governance and Community Based Management. Gain, A. a. (2012). An Assessment of Water Governance trends: the case of bangladesh. Water Policy 14 , 821-840. Humphreys, E. B. (2014). Messages from the Ganges Basin Development Challenge. Research for Development Series. Haider, H. (2009). Community-based Approaches to Peacebuilding in Conflict-affected and Fragile Contexts. Governance and social development . Heer, J. a. (2012). Prac ces of Cross Cultural Collabora on in Sustainable Water. Interna onal Journal of Business Anthropology . MoWR. (2001). Guidelines Par cipatory Water Management, GoB. Ministry of Water Resources. Mollinga, P. P. (2008). Water, Poli cs and Development: Framing a Poli cal Sociology of Water Resources Management. Water Alterna ves 1(1) , 7-23. Pal, S. a. (2010). Challenges and adequacy of water resources management policies in Bangladesh. School of the Built-Environment, Heriot-Wa University, UK . Sultana, F. (2008). Community and par cipa on in water resources management: gendering and naturing development debates from Bangladesh. Transac ons of the Ins tute of Bri sh Geographers . Silva, S. d. (2012). The experiences of Water Management Organiza ons in Bangladesh. Shackleton, S. C. (March 2002). Devolu on and community - based natural resource management: crea ng space for local people to par cipate and benefit. Natural Resources Perspec ves (ODI) 76: 1-6. , Number 76. Rab, A. (2009). River fisheries management in Bangladesh: Drawing lessons from community Based Fisheries Management (CBFM) experiences. Ocean & Coastal Management 52 , 533-538. Thompson P., S. P. (2003). Lessons from community based management of floodplain. Journal of Environmental Management 69 (2003) 307–321, 15. 146 Mul ple actors, conflic ng roles and perverse incen ves: The case of poor opera on and maintenance of coastal polders in Bangladesh F.Naz1 and M-C Buisson2 1 2 World Agroforestry Centre (ICRAF), Vietnam, f.naz@cgiar.org Interna onal Water Management Ins tute, India, m.buisson@cgiar.org Abstract The government of Bangladesh invested in large scale coastal embankment projects in the 1960s and 1970s. The polders that were developed played an important role in protec ng coastal communi es from water-related disasters and in increasing agricultural produc vity. However, over me maintenance of these infrastructures became a major concern leading to the crea on of a na onal policy that requires local communi es to par cipate in their opera on and maintenance. In this paper we are interested in understanding what determines the poor state of affairs of the polders. One way to examine this will be through the lens of opera on and maintenance (O&M) and the prac cal strategies adopted by different actors for O&M. This paper consequently discusses the roles and responsibili es of these mul ples actors in opera on and maintenance of water infrastructure in the coastal zone of Bangladesh. The analysis is based on primary data collected in 2012 and 2013 in nine study sites from the coastal zone. Qualita ve data was collected in these nine sites through focus group discussions and key informa on interviews. An in-depth analysis of how opera on and maintenance ac vi es actually take place reveals that the mul plicity of actors involved in opera on creates overlaps and conflicts, resul ng in the strategic deferral of maintenance by different actors and eventual disrepair and degrada on of the infrastructures. Ul mately, the unclear demarca on of roles and responsibili es for these actors curtails the short and long term sustainability of water management in the polder area. The paper recommends revising the legal water management framework, improving coordina on and giving a formal role to local government ins tu ons. Keywords: water management, community-based natural resources management, decentraliza on, actors, power 1. Introduc on In the coastal areas of Bangladesh inunda on, salinity intrusion and severe flooding are frequent occurrences. To overcome these challenges, the Bangladesh government has invested in coastal zone management through construc on and rehabilita on of polders. A polder is a low-lying tract of land enclosed by embankments that create an independent hydrological en ty. Gates and sluices ensure both irriga on and drainage of the area. In the 1960s the former Government of East Pakistan through its Water and Power Development Board (WAPDA), now Bangladesh Water Development Board (BWDB), constructed polders to protect agricultural crops, land and human se lements. Efforts intensified in 1967 with the Coastal Embankment Project, which was funded with USAID assistance (Islam 2005; Chowdhury and Rasul 2011). In total 123 polders were built along the coastal zone. Much later, in late 1990s and early 2000s, the Local Government Engineering Department (LGED) of Bangladesh also constructed polders, but on a smaller scale, which were therefore called sub-projects. Benefits from the polderiza on of the coastal zone were short lived. The structures delinked the wetlands from the rivers and caused drainage problems and waterlogging. This further added to the natural processes of river erosion and silta on in the very ac ve delta, resul ng in the need for con nuous rounds of rehabilita on to preserve the infrastructures (embankments, gates and canals). In many developing countries, 147 with Bangladesh no excep on, large-scale public irriga on systems are o en characterized by “inefficient, unreliable, and inequitable water service; chronic underinvestment in maintenance; rapid deteriora on of infrastructure; and reduc on in service areas” (Araral 2005: 113). In conjunc on with a shi in donor discourses, the con nuous challenge of opera on and maintenance of water infrastructure induced regular shi s in water management governance in the coastal zone and the introduc on of new actors. What had been a local and indigenous system prior to the 1960s became a top-down, engineering-driven system in the 1960s and 1970s before returning to a decentralized and depoli cized community water management system in the 2000s (Dewan et al. 2015). In this paper we are interested in understanding what determines the poor state of affairs in the polders. One way to examine this issue is through the lens of Opera on and Maintenance (O&M) and the prac cal strategies adopted by different actors toward O&M. The concepts of O&M are o en considered together although they are quite different: whereas opera on has short term benefits and involves daily acts in this context, maintenance is less immediate and its effects are only perceived in the medium to long term. Apart from their meframes, the incen ves for the different actors to undertake O&M and their funding sources also differ. Consequently, we consider separately ac vi es related to O&M and analyze how different stakeholders contribute to these two sets of ac vi es. The ar cle is organized as follows. Sec on 2 discusses the methods used for data collec on and analysis. The third sec on then provides context on water management in Bangladesh and the main actors involved in the sector. Sec on 4 focuses on the results with an emphasis on the roles of actors in opera on and sec on 5 focuses on the actors involved in maintenance. Finally sec on 6 concludes the analysis and provides policy recommenda ons. 2. Methods This analysis is based on primary data collected in 2012 and 2013 in the coastal zone of Bangladesh. Fieldwork was conducted at nine study sites: five large polders built by BWDB and four sub-projects (less than 1000 hectares) under the supervision of LGED. These nine loca ons were purposely selected from three different agro-ecological zones in order to capture differences in environmental constraints (e.g. salinity and waterlogging) and ins tu onal backgrounds (small scale vs. large scale; project coverage). The loca on of the study area is given in Figure 1. Table 1 summarizes the agro-ecological and ins tu onal features of these study sites. Qualita ve data were collected in these nine sites through 57 Focus Group Discussions (FGDs) and 92 Key Informant Interviews (KIIs). Data collected was representa ve of different contexts in terms of distance from the main rivers and sluice gates, level of silta on of the surrounding canals and concentra on of various types of cropping systems. The KIIs were held with different stakeholders: farmers, women-headed households, Labor Contrac ng Society group members, gatemen, Union Parishad members, Water Management Organiza on (WMO) members, and BWDB and LGED officials. Apart from the qualita ve primary data, secondary data such as government and donors reports and sta s cs were also used to support our analysis. 148 N 20 10 0 20 Kilometers Fig. 1. Map of the study area, coastal zone of Bangladesh. Source: Ins tute of Water Modelling Table 1. Main characteris cs of the study sites Study site Area (sq. km) Approximate popula on (2011) Number of sluice gates Salinity levels Agency Project coverage for water management Polder 3 194.3 39,584 32 Very high BWDB None Polder 24G 258.56 61,867 8 Medium to low BWDB KJDRP Polder 31 Polder 30 148.31 72.09 32,576 36,017 67 28 High Medium BWDB BWDB 4th Fisheries IPSWAM Polder 43-2F 56.22 28,485 11 Very low BWDB IPSWAM Latabunia Jabusha 2.0 4.11 446 6195 1 5 Medium Low to medium LGED LGED SSWRDP SSWRDP Jainka 1.0 325 2 Very low LGED SSWRDP Bagarchra 3.5 1299 2 Medium to high LGED SSWRDP 3. Water management of Bangladesh: a mul ple actor framework 3.1 Decentraliza on and formaliza on of water management In parallel with and resul ng from the physical, poli cal and economic changes of the coastal zone, ins tu ons involved in water management have witnessed several evolu ons. The introduc on of new ins tu ons follows two historical trends: decentraliza on and formaliza on. In Bangladesh decentraliza on 149 has largely been a poli cal tool employed by ruling par es (Islam and Fujita 2012). Poli cal decentraliza on can be defined as the transfer of authority to a sub-na onal body. Poli cal decentraliza on aims to give ci zens, or more o en, their elected representa ves, more power in public decision making. In rural Bangladesh there are three local government ers: Zila Parishad at the district level, Upazila Parishad at the sub-district level and Union Parishad for groups of villages. In spite of decentraliza on, local governments in Bangladesh are s ll largely dependent on central governance. For example, the central government can legally dissolve a local authority that is not able to meet its objec ves (due to inefficiency, power abuse, financial bankruptcy, etc.) (Habibullah 1996). Therefore, the predominant sen ment is that Bangladesh’s local government ins tu ons were created to spread the control of the central state to remote loca ons, rather than for reasons of empowerment. In that respect, the lowest er of rural administra on, Union Parishads, are largely dependent on Upazila Parishads and thus play only a limited role in rural development programs (Islam and Fujita 2009). Decentraliza on of Bangladesh’s water management was ini ated in 1999 with the formula on of the Bangladesh Na onal Water Policy (GoB 1999) and then opera onalized in 2001 with the Guidelines for Par cipatory Water Management (GoB 2001). These guidelines clearly state that communi es are the main stakeholders. The Na onal Water Policy of 1999 recognized for the first me the role of water in poverty allevia on and introduced inclusive water management (Quassem 2001). At the same me, these policies argued for a formaliza on of the community ins tu ons involved in water management and de facto weakened exis ng formal and informal organiza ons. The guidelines opened par cipa on in water management to a large range of actors, but also created confusion on the respec ve roles of each actor. It is also worth men oning that despite the Na onal Water Policy’s focus on par cipa on, the different steps defined in the Guidelines for Par cipatory Water Management gave more importance to communi es for consulta on rather than implementa on and did not define effec ve mechanisms for transferring the decision-making responsibili es (Dewan 2012, Dewan et al. 2015). 3.2 Defining the actors: top-down scale, formaliza on and power Actors are defined here as individuals or group of individuals ac vely involved in water management who influence water access and water control based on their degree of power. Actors can be formal or informal and they are classified in this paper on a scale of formal and informal ins tu ons11. The defini on of power used in this paper is drawn from Lukes (2005); power manifests itself by shaping the values, norms and preferences of a group. It is close to the form of power illustrated by Foucault (1975) who argues that power is not the apanage of a central state, but rather is consolidated in the daily enforcement of social and poli cal prac ces. Therefore in this paper we are interested to assessing power in the Foucault sense, where ins tu ons and power are closely interrelated. Bangladesh Water Development Board (BWDB) is the oldest actor in water management in coastal zone, and a key formal one. BWDB can be considered a governmental implemen ng agency in terms of water management. It started its opera ons in 1959 as the water wing of the erstwhile East Pakistan WAPDA. BWDB has held the responsibility of execu ng flood control, drainage and irriga on projects to boost produc vity in agriculture and fisheries through major investments in the water sector supported by the Ministry of Water Resources and interna onal donors. It was and s ll is predominately an engineering, construc on-oriented agency, characterized by a centralized structure that was suited to the type of large-scale investments implemented in the 1960s and 1970s (Chadwick and Da a 2003). As per the Na onal Water Policy, BWDB is responsible for water management in polders larger than 1000 hectares. BWDB implemented the Integrated Planning for Sustainable Water Management (IPSWAM) project from 2003 to 2011 and now leads the Blue Gold project funded by the Dutch Embassy. 11 Ins tu ons in this paper are referred to by the “rules of the game” (North 1990). We recognize that the terms formal (i.e., modern, bureaucra c) and informal (i.e., social and tradi onal) may some mes be misleading; indeed tradi onal ins tu ons can also be formalized though not necessarily in the bureaucra c forms that are considered here (Cleaver 2001). 150 Local Government and Engineering Department (LGED) entered into the water management arena in the 1980s. LGED is also an implemen ng agency in terms of water management but pertains to the Ministry of Local Government Rural Development and Coopera ves. LGED formalized its role in the water sector through the Small-Scale Water Resources Development Sector Project (SSWRDSP), which began in 1995. Through this project, LGED has provided flood control, drainage and irriga on infrastructures to sub-project areas of less than 1000 hectares. Their approach relies heavily on local stakeholder ini a ve to iden fy interven ons and support engineering design (De Silva 2012). As previously men oned, Union Parishads are the lowest er of elected local government ins tu ons. In that respect they are a formal actor and stand at an intermediate level between the government and the communi es. Union Parishads are under the supervision of the Ministry of Local Government. They are comprised of 12 members: nine members from nine wards of the union and three women members, one each from three wards. All members are elected through direct universal adult suffrage (Mujeri and Singh 1997). As defined by the Guidelines for Par cipatory Water Management (GoB 2001) Union Parishads were supposed to be ‘advisors’ to the Water Management Organiza ons. In addi on, in command areas of less than 1000 hectares, they were to gradually receive ownership of the infrastructures. However, facing a lack of legal mechanisms and resources to take on these roles, the Union Parishads are only informally involved in water management. Their limited role is not only contrary to the intent of the policy but also to the wishes of most community members (Dewan et al 2014). As part of the process of decentraliza on in the water sector and following the Guidelines for Par cipatory Water Management (GoB 2001), Water Management Organiza ons (WMOs) were also created (de Silva 2012). These WMOs are intended to func on as the ins tu onal mechanism by which local stakeholders par cipate in water management. In areas of more than 1000 hectares, WMOs should ideally be comprised of Water Management Groups (WMGs) and managed by BWDB; in areas of less than 1000 hectares they should be comprised of Water Management Coopera ve Associa ons (WMCAs) and under LGED management. They are intended to hold decision-making power at all stages of local water resource management and are responsible for planning, implemen ng, opera ng and maintaining local water schemes (GoB 2001). Previous research highlighted how these organiza ons have failed to enhance the par cipa on of the most vulnerable community members (notably women and the landless) and have resulted in elite capture (Dewan et al 2014). Due to the process of decentraliza on of the water sector in Bangladesh, LGED follows a one- er system, whereas BWDB follows a three- er model: associa ons (WMA) welcome representa ves from the different WMGs, with federa ons (WMF) at the upper-most level. In spite of their ins tu onal differences, all these WMOs are registered as coopera ves12 and are therefore formal ins tu ons. Apart from the above-men oned formal actors, there are many informal actors, individuals or groups involved in water management in Bangladesh. Gate commi ees, gher commi ees and beel commi ees are some examples of informal actors. Some of these actors are related to formal ins tu ons whereas some are not. It is typical however to find individuals within these informal groups who have a formal role in other ins tu ons. Figure 2 locates the different actors in a two-dimensional space. The first dimension is related to the level of decentraliza on; the actors are located in a top-down scale. The second dimension is the formal recogni on of the actors, for which a formal-informal scale is applied. As noted in the following sec ons, the power of these actors is context-specific, determined at a small scale and is not consistently aligned with their loca on in this framework. 12 Since February 2014 and as per the Water Management Rules (GoB 2014), Water Management Organiza ons ini ated by BWDB or its projects are not required to register with the coopera ve department but must register with the local Water Management Department. 151 Actors involved in water management in polders (more then 1000 ha), BWDB BWDB Actors involved in water management in sub-projects (less then 1000 ha), LGED Other actors LGED IPSWAM project SSWRSP project WMF WMA Union Parishad WMG WMCA Gate commi ee Gher Commi ee Individual farmer Formal Formal Informal scale Informal Fig. 2. Actors involved in water management in the coastal zone of Bangladesh. 4. Actors in opera ons: from decentraliza on to informal management 4.1 Centralized and formal opera ons: the former khalashi system Ini ally, a er the crea on of polders in the 1960s, BWDB employed government-funded gatemen called khalashis to operate the gates based on local requests. At this me the government did not expect communi es to be involved in the day-to-day management of water infrastructures. The khalashis system worked efficiently, as there was someone responsible for the opera on of the gates and the communi es did not hold financial responsibili es for O&M. “There used to be khalashis in the BWDB sluice gates but not anymore. They were paid by the East Pakistan government and the system was quite efficient.” -Polder 43/2F, KII, 10-04-2012 However, the 1990s saw a push for decentraliza on and people’s par cipa on, which led to a structural adjustment process requiring BWDB to downscale its ac vi es and payroll, and ul mately to change its approach to water management (Dewan et al 2015). As a result, the system of state funded ‘gatekeepers’ was abolished and the responsibility was placed with communi es. 4.2 Formal water management groups, the principal actors in opera on According to the Guidelines for Par cipatory Water Management (GoB 2001), WMOs are responsible for 152 internal water management; in sub-projects WMCAs are singularly responsible, while in BWDB polder the responsibility is shared between the different levels (WMG, WMA, WMF). In reality, this means that the WMOs are responsible for opera on of the gates as well as for preliminary discussions required to reach a consensus on opera on. This situa on is indeed happening in some loca ons. For example, in polder 43/2F a farmer described the WMG’s coordina ng role for reaching consensus on opera on. “We open the sluice gate at the me of preparing seed beds in the month of Chaitro (March-April). Usually the group’s members sit together, discuss among themselves and make decisions on opening the gate. The beneficiaries residing beside the canals a end the mee ngs. Some mes people are divided in their opinions regarding opening of the gate. But with discussion, we are able to reach a consensus.” -Polder 43-2F, FGD, 10-04-2012 Similarly in sub-projects, LGED has given responsibility for opening and closing the gates to the WMCAs and there is no strong monitoring of these ac vi es by local-level LGED officials. Through the inclusion of different stakeholders in the WMCA, this system can be effec ve in balancing power in the decision-making process. “WMCA takes the decision regarding when to open and close the sluice gates. There are influen al people in the WMCA. But nevertheless at least to some extent the WMCA has created a situa on of balance of power to control water management of rivers and canals. Large and influen al farmers cannot totally dominate the decisions of the associa ons.” -Bagchara-Badurghacha Sub-Project, FGD, 25-03-2012 4.3 Responsibility of opera ons delegated to gate commi ees Given that Water Management Groups may supervise water management for several villages, they are o en responsible for opera ng several gates. Consequently, they frequently delegate their gate opera on role to lower-level gate commi ees. This delega on should simplify the decision making process by reducing the stakeholders to the users of this par cular gate. These commi ees can some mes be informal but they can also be formally related to the formal WMG. Even if gate commi ees are supposed to report to the WMG, by holding the decision on opera on, these small groups also hold the power. “Gate commi ees are the most powerful in opera ng gates as they are in charge of everything.” -Polder 43-2F, KII, 10-04-2012 WMCAs also delegate their responsibility to decide on the opera on of gates, including their opening and closing, to gate commi ees. For example, in Jabusha sub-project which has 10 gates, there are sub-commi ees i.e., gate commi ees formed with local farmers and fishermen. Opening and closing the gate is generally done through voluntary work, without a fixed gateman, and the responsibility of opera ng the gates rotates amongst the members. In these cases, no remunera on is given to gate operators. In some other cases, the commi ee assigns the responsibility to the owners of a house located close to the gate. When opening of gate is required, people approach the gateman and he opens the gate, if needed with the permission of the gate commi ee president or with the agreement of the other members, as in the case of Jainka . “The gate commi ee has a gateman but he is not specified. They do not receive any allowance. Any fisherman or farmer can be a gateman.” Jabusha sub-project, FGD, 30-03-2012 4.4 Informal opera ons and elite capture If the inclusion of all the stakeholders in the WMOs is supposed to ensure a balanced decision making process, in prac ce WMOs are not always able to prevent elite capture in opera on of the gates. For example, it was commonly found that gher owners or local elite dominate the decision over the opera on of gates. 153 “The influen al people make the final decision about the closing and opening of the gate. They take the decision as they have the economic power and direct connec ons with the Union Parishad chairman. This badly hampers water management. Moreover it is the large land owners who have unlawfully grabbed the khal13.” Polder 30, KII, 15-04-2012 Due to elite capture, gate opera on o en fails to reach an efficient and fair outcome. In Jainka , the WMCA only controls one of the two gates due to a land dispute involving influen al families. In Jabusha, fac ons within the WMCA have resulted in canals being blocked and used for aquaculture, disrup ng irriga on through the main canal. Officially, WMCAs are supposed to make decisions regarding flushing and draining of water during different farming seasons and within farming seasons. However, the adjoining gher owner effec vely holds effec ve control in Latabunia. In Latabunia where approximately 50% of land belongs to outsiders, elite capture over water control is a clear source of conflict. Conflicts occur between outside leaseholders or fish farmers and local paddy farmers, par cularly regarding the drainage of water from the gher before plan ng aman paddy. In these cases, the theore cal role of conflict resolu on that lies at the WMOs level is unlikely to be seen. The data collected establishes that elite capture is prevalent in most of the polders but this is predominantly in areas where no water management project was implemented and consequently where no WMOs have been formed, such as polder 3. In these polders opera ons are informally managed and capture by influen al elites is the standard. Diverse situa ons have been observed with different degrees of informal management and capture. For example for some gates in polder 3, a BWDB sec on officer gives a decision or intervenes on decisions regarding the opening or closing of par cular gates. Although the BWDB officer insists on keeping the gate closed in the dry season, it can be opened any me if the interested person bribes the staff or gives ps to the proxy gateman. “In order to get the gate open, if one pays a bribe to the proxy gateman he opens the gate. A prominent gher owner pays 1000 Taka and the gateman open the gates.” -Polder 3, FGD, 17-02-2012 Thus in polder 3, larger gher owners are o en the elite and the main decision makers regarding the opening and closing of the gates. In the absence of any formal WMOs informal commi ees have been formed under the leadership of gher owners who require frequent renewal of saline water in their ghers. “We do not have any formal commi ee. Gher owners have an informal commi ee to open and close the gates. It is usually the gher owners who decide when and where water is needed. The one who owns more ghers is the most powerful, leads the decision and calls the final shot.” --Polder 3, FGD, 18-02-2012 Some of these informal commi ees hire a gateman for the opera on of the gate, who is paid through fishing rights or given cash remunera on. In some villages in polder 3 it was also found that gher commi ees collect contribu ons from the gher owners, some mes based on the size of their land under opera on. The same phenomenon occurs under informal management when formal WMOs become inac ve a er project withdrawal. In those cases, informal commi ees tend to take over opera on and benefit from the vacuum of power regarding water control and access. For example, informal beel commi ees decide the opening or closing of gates in polder 30 and in some villages of polder 43-2F; these commi ees are supported by gher owners or large landowners rather than all local stakeholders. Again, the involvement of informal actors means that influen al elites dominate decisions related to the opera ons of the gates. This is apparent from a large number of focus group discussions. 13 Khals are internal canals, mainly found inside the polders. 154 4.5 From collec ve to private opera ons Community water management, even when the decisions are subject to elite capture, induce collec ve decisions over control and access of the resource. Although it has been noted that a large number of decisions related to water access are taken out of the collec ve sphere and can therefore be considered to be private opera ons. Indeed, private actors develop their own strategy in terms of water management to fulfil their water requirements. In polder 31 there are about 24 private gates and seven pipes according to the mapping done for this analysis. Respondents indicated that the private gates are operated by neither the gate commi ees nor by the WMGs; they are privately operated by individual gher owners. Similarly, despite opera ng within a smaller area, sub-projects also face issues of private opera ons. There are number of private gates in LGED polders, which are apparently under the control of the landowners. In Latabunia, the embankment is crossed by tens of underground pipes, some temporarily closed with mud and some with more sophis cated closing systems. Individual gher owners decide the opening and closing of these pipes without any coordina on with their neighbours. This situa on makes the WMCA, who is hardly in control of one gate, powerless in preven ng salinity intrusion and in draining the area for paddy cul va on. Therefore, private opera ons are ed up with vested interests and prevent mul ple community stakeholders from efficiently controlling their access to water. 4.6 Union Parishads in opera on: subs tutes, elite capture and conflict resolu on As previously stated Union Parishad involvement in water management is supposed to be limited to an advisory role. In many places, Union Parishads do not even play this role and are absolutely removed from the opera on of the gates. However, the data collected also suggests that in the absence of any formal and func onal WMO, the Union Parishad can also become an important subs tute for opera ng the gates. Thus a number of local gate commi ees are related to or headed by Union Parishads. The Union Parishad as an organiza on is not involved in opera ng the sluice gates but the Union Parishad chairman and the members of the concerned union could be involved. For example, Union Parishad representa ves can be found in the beel commi ees and some local gate commi ees are headed by a Union Parishad member or by the Union Parishad chairman. However, as noted in the below quota on from polder 30, the involvement of Union Parishads in water opera on does not always result in balanced power. “Most of the commi ees are no longer func oning. The Union Parishad chairman controls the gates and he gives responsibility for opening and closing of the gates to his favorite people.” -Polder 30, FGD, 16-03-2012 Indeed, there is o en a clear overlap between influen al people and Union Parishad members. For example in polder 3, Union Parishad members tend to be gher owners and are also members of the informal gate commi ees leading the opera on of the gates. The same can also be true when there are func onal WMOs. Even if the Union Parishad as an organiza on is not directly involved in gate opera on some elected representa ves could be members of the WMOs and can consequently influence the opening and closing of gates. These situa ons have been iden fied in polder 30 and Jabusha sub-project. Finally, in terms of opera on Union Parishads play a role in conflict resolu on. Conflicts relate to paddy versus shrimp farming, to low land versus high land water access or drainage, and to ming depending on the cropping pa ern chosen by the farmer and on the maturity of the crop. When these conflicts cannot be resolved locally by mutual agreement between farmers then media on from the Union Parishad is required, which de facto involves the local government ins tu on in opera on. 155 4.7 Gaps and overlaps in opera on Ins tu onal arrangements for regarding gate opening and water management decision making vary across and within polders. Indeed, from one gate to another one, the ins tu onal arrangement that leads to a decision and to the physical opera on of the gate is never exactly the same. The na onal policy, and especially the Guidelines for Par cipatory Water Management (GoB 2001), locates the responsibility for opera on at the community level, making it a decentralized decision making process. But in all the loca ons, whether BWDB polders or LGED sub-projects, the informal, elite actors tend to dominate the opera on of the sluice gates. Opera on of the gates is vital to the livelihoods of those in the coastal zone. Taking advantage of this fact as well as the missing or unsustainable formal groups and of the vagueness of the policy, several actors have tried to put forward their own water management strategy that services their own interests. In some places this leads to conflicts between different actors, as in the case of Polder 31, Latabunia or Jabusha. Thus, the mul plicity of actors involved in opera on produces overlaps and conflicts. Due to the structure of power in rural Bangladesh, these overlaps and conflicts contribute to the benefit of private interests. 5. Actors in maintenance: deferred and subs tuted responsibili es 5.1 Breach at the central level: deferred major maintenance Ins tu onally, BWDB owns the water-related infrastructure in the polders. In each district BWDB has a special wing called Opera on and Maintenance for the polders and an O&M office headed by an execu ve engineer. However, findings from this fieldwork indicate that BWDB executes repair work only occasionally, when funds are available. These funds are typically only given some disaster takes place or when minor maintenance becomes major and a racts the a en on of higher authori es or donors. BWDB engages contractors by tender or Labor Contrac ng Socie es to carry out such maintenance. There is also a general view among community members that BWDB staff are outsiders, lacking both local knowledge and ownership. “The employees of BWDB comes on motorcycles and just take rounds and do nothing. They do not represent us.” -Polder 3, FGD, 18/02/2012 As per the LGED’s handover agreement with each WMCA, WMCAs hold responsibility for minor maintenance and repair while major repairs remain a responsibility of LGED. LGED also supports community organiza ons to design their maintenance plans for maintenance expenditure each year. For example in Jainkathi sub-project, to monitor, supervise and plan maintenance work, LGED staff visit the sub-project twice a year. A post-monsoon assessment of damages is established, which informs cost es mates and yearly budgets. The WMCAs typically inform LGED of their repair needs, raise demands for repairs, lobby if needed, and thus get funding allocated by LGED, usually once a year. However the funds allocated are o en less than what is required for maintenance and may take me to reach the community. This is the case, for example, in Latabunia: “LGED does not maintain the gates directly. But if the closing device of the sluice gate is damaged or it is oxidized or the plaster comes off, then LGED gives money to repair. But we have to communicate with them many mes to get the work done.” -Latabunia, FGD, 27-03-2012 5.2 Water management groups: deferred minor maintenance According to the Guidelines for Par cipatory Water Management, WMGs are responsible for planning, implemen ng, opera ng and maintaining local water resource schemes in a sustainable manner. WMGs are tasked with: (i) preven ve maintenance of the medium and minor hydraulic structures, bridges, culverts, etc.; (ii) preven ve maintenance of the main embankment and secondary embankment; (iii) rou ne/annual maintenance (desal ng) of field channels, drains, etc.; (iv) clearing weeds, obstacles from secondary and 156 ter ary channels, canals drains, etc.; (v) regular greasing of gates; and (vi) annual pain ng and minor repair of minor gates and replacement of fall board. Even if these repairs are referred to as minor, they are most of the me beyond the capacity of WMOs. To finance minor and regular maintenance the WMOs relies on its own funds. These resources come from either contribu ons by the community or from some addi onal sources of income (e.g. interest from micro-credit, leasing of canals). For example, several WMOs started to propose saving accounts and offer micro-credit services to their members in order to generate sustainable maintenance funds. Nevertheless, respondents across the study sites stated that savings and loan services were defunct; default of loans was more frequent than repayment. This has occurred in sub-projects as well as polders. “In Latabunia polder, collec on of monthly savings is effec vely discon nued. The membership fee is not being generated as there are no new members and the exis ng members are not buying new shares. The microfinance program was intended to cover the cost of occasional repair and build up the capital of the WMCA. This objec ve has also not been achieved. The loan disbursed to members before Aila14 has also not been recovered.” -Latabunia, KII, 28-3-2012 Similarly, where canal leasing has been implemented it has failed in the long run due to conflicts over the choice of tenant, water flow blockages and legal vagueness over who should be the leaser. Jabusha has experienced challenges with canal leasing. Finally, very few WMOs have been able to maintain regular collec on of contribu on fees from their members. As such, all WMO sources of funding have dwindled over me. The financial failure of the WMOs o en reflects a more general situa on. O&M fund inadequacy has been a ributed to a general disinterest in assuming responsibility for maintenance and lack of competency in the case of WMCA management commi ees (ADB 2008; BIDS 2008). In an external evalua on commissioned by the Asian Development Bank it was found that some O&M sub-commi ees were inac ve and half of the WMCAs had no O&M plan, despite this being a requirement for handover (ADB 2003). Interes ngly, the state of infrastructure in areas where WMOs have managed to obtain some sources of funds is not any be er. In Bagachra-Badurgacha, despite considerable funds from canal leasing, addi onal funds were required from LGED. Similarly in Jainka , where a fairly good system of contribu on collec on (based on land size) is in place, one gate is blocked due to property conflict and the canal from the second gate is silted. These situa ons bring into ques on the incen ves that these WMOs may have for inves ng in minor maintenance. In summary, the role of WMGS in minor maintenance is one of solving only the most urgent infrastructure problems and relying on rudimentary and unsustainable repairs. Excep ons occur during emergencies when WMOs uses their own saving fund, and collect special fees and material contribu ons from the community. They also play an important role in mobilizing people and organizing voluntary work in order to repair the embankments. At a higher level, WMAs do not play any direct role in maintenance, as they don’t have any dedicated funds for doing so. They nevertheless play a role in iden fying needs and repor ng them to BWDB. They can also be involved in supervising maintenance work. For example in polder 43/2F, through the IPSWAM project, once the funds were allocated to specific works the WMA made a list of Labor Contrac ng Society groups and members in the concerned area. The work was then allocated to the Labor Contrac ng Socie es and the WMA monitored progress and reported on quality, which determined BWDB payments. 14 Cyclone Aila struck the coast of Bangladesh in May 2009. Numerous villages were completely submerged. Apart from the short-term impact of lost housing and belongins the cyclone has had a long-term effect on households by increasing the salinity levels of both land and water. 157 5.3 Informal and local level involvement: safety net for essen al maintenance The deferred maintenance that has resulted from central actors as well as community-level formal organiza ons not fulfilling their roles has resulted in a lack of trust from water users and community members. Many community members believe they must rely on themselves to conduct maintenance ac vi es. “The local people work voluntarily to maintain the embankment. LGED does not work as it used to do earlier. Moreover the WMCA is also not working well, and we cannot complain against them, we have to work by ourselves.” -Jabusha, FGD, 30-03-2012 Communi es o en report that they have to protect themselves with low-cost repair work done on their own ini a ve. For example, in Bagachra-Badurgacha sub-project most households own land and it is therefore understood that they have to work voluntarily toward maintenance of the infrastructures in order to protect their land. For example, local farmers build bamboo pilings to prevent damage or to repair damaged parts of the embankments. The role of individual community member becomes all the more crucial in emergency situa ons. Whereas formal and centralized levels need me to mobilize resources, the informal community level has more flexibility. Moreover, while households may have difficul es in valuing their interest in contribu ng to maintenance on a regular basis, emergency situa ons bring clear and short-term incen ves. Then, similar to opera on, the main users and the ones for whom maintenance is essen al take over formal organiza ons to finance maintenance. Thus the gher owners are very o en involved in maintenance or repair of the gates. In the case of BWDB gates, the landowners and gher owners provide financial contribu ons for repairing or re-excava ng the canal when it becomes essen al and risks threatening their interests. These influen al people can also play a role in mobilizing the material and human resources required for maintenance. “Local people manage the small amount of the cost of repair of the gate. We made a wooden shu er with our own ini a ve as the door of the gate was damaged. In this a local elite person […] played a key role in mobilising people.” -Latabunia, FGD, 27-03-2012 5.4 Union Parishads: the constrained and subs tute actor Union Parishads play a suppor ng role in maintaining water infrastructure alongside the informal and local actors. The ini al role of Union Parishads in maintenance was limited to emergency repairs. As the lowest level of public administra on in Bangladesh, they are also the first level of relief. Their role was therefore important a er the Sidr and Aila cyclones. Apart from mobilizing financial resources they also mobilized communi es and organized the voluntary work. This was par cularly important in Latabunia sub-project where the embankment and the village were submerged. “During floods our lives were saved because of the UP Chairman. He did some emergency work at the me of the disaster by organizing the villagers and took ini a ve in the repair work. Moreover we protected ourselves by collec ng money, bricks and sand.” -Polder 3, FGD, 18/02/2012 Facing deferred maintenance in their unions, some Union Parishad members have gradually increased their involvement in maintenance in order to respond to the increasing demands of their voters. But Unions Parishads face a number of problems that limit their role in polder maintenance. First, as per the legisla on, their role is to coordinate and to advise the WMOs; they therefore don’t have any resources dedicated to 158 water infrastructure maintenance. In addi on, embankments are under the ownership of BWDB, so Union Parishads cannot rehabilitate the embankments without BWDB’s consent. They similarly have to coordinate with LGED for maintenance work in sub-projects. Despite these financial, technical and ins tu onal capacity constraints, Union Parishads execute some repair work and re-excava on. Their involvement in maintenance uses at least two tools. The first is mobiliza on. The chairman of the Union Parishad o en mobilizes people to repair embankments, work which is regularly done on a voluntary basis. In addi on people donate bamboo, mber and other materials. “Villagers temporarily repaired the embankment in 2011 using bamboo fencing and UP chairman also mobilized people to do this.” -Polder 43/2F, FGD, 10-04-2012 Another way of being involved is through dedica ng rural employment schemes to water infrastructures. “UP has no fund by which sluice gate or embankments can be repaired but it s ll it conducts repair work to the road and the embankments by the 40 days programme.” -Polder 30, FGD, 16-03-2012 Indeed, some Union Parishads use rural employment schemes such as KABHIKA (Food for Work), KABITA (Cash for work) and 40-days work, funds for which are allocated from the Upazila Parishad to maintain roads, repair embankments, and re-excavate canals. This happened for example in polder 3 and in polder 30. Union Parishads do so either by sub-contrac ng NGOs or LGED, or by directly forming Labor Contrac ng Socie es made up of rural and disadvantaged community members. Despite these examples, financial and ins tu onal constraints mean the role of Union Parishads in water infrastructure maintenance remains limited. They are unable to carry out the regular maintenance of all infrastructures that is required to sustain the livelihoods of the coastal areas. 6. Conclusion and recommenda ons This analysis has shown that water management in the coastal zone of Bangladesh is much more confused than the procedures and roles defined by the policy may suggest. Indeed, a large number of actors anchored in different poli cal, social, economic or administra ve frameworks are involved in decisions and ac ons related to opera on and maintenance. The roles of these actors have been assessed through a top-down scale to point out the level of decentraliza on and through a formal-informal scale. The mul plicity of actors involved in opera on produces overlaps and conflicts; however the structure of power in rural Bangladesh results these overlaps and conflicts benefi ng only private interests. On the maintenance side, gaps and deferred maintenance arise from the mul plicity of actors involved. This leads to disrepair and degrada on of the infrastructures, which steadily weakens the sustainability of coastal zone livelihoods. Thus we find that the policy has created confusion regarding the respec ve roles of each of these actors and does not take into account the social and ins tu onal structure of Bangladesh and exis ng power rela ons. Where formal actors miss-fill their role and responsibili es, many informal actors, individuals or groups (gate commi ees, gher commi ees or beel commi ees) fill the gap. Similarly, while local government ins tu ons have been largely overlooked in the policy, they remain informally involved in water management when required. These results bring forward a number of recommenda ons. First, the water management policy must be revised and clarified. This policy has to take into account the par culari es of the coastal zone and the exis ng power rela ons between central and decentralized actors as well as between formal and informal actors. Rethinking the water management policy may involve redefining the role of each stakeholder in terms of opera ons and maintenance. By devia ng from the common discourse on the inclusion of all the stakeholders, the policy could be able to create a more efficient water management system. 159 Secondly, analysis from the qualita ve data points out a lack of formal coordina on between the different actors. A latent and rudimentary form of coordina on occurs through conflicts but this type of coordina on brings power rela ons into the game and prevents some actors from being taken into account. All actors involved in opera on and maintenance should have access to a democra c pla orm for discussing and coordina ng their ac ons. This coordina on also requires a set of rules and a leader who will ensure that rules are respected. Following the Union Parishad Act of 2009, Union Development Coordina on Commi ees were created at the union level to improve coordina on in terms of development in the union. Water management is not clearly included as one of the mandates of these commi ees, but it should be formally included. Finally, the role of local government ins tu ons in water management should be formally recognized. The Union Parishad, the lowest- er rural local government, closest to the rural people and elected by them, has a realis c possibility of playing a vital role in water management. Union Parishad involvement would ensure long term sustainability of the process and balanced adjudica on. Nevertheless, this would only be possible through increasing their control over local resources and over choices regarding resource alloca on (Ullah and Pongquan 2010). Improved water governance and successful opera on and maintenance in the polders requires the defini on of a new legal framework that is more inclusive of the ins tu onal reali es of Bangladesh, improved coordina on between all stakeholders and the formal recogni on of the essen al role played by local government ins tu ons in water management. Acknowledgements This ar cle is based on data collected for the project G3: Water Governance and Community-based Management in Coastal Bangladesh , part of the Ganges Basin Development Challenge of the CGIAR Challenge Program on Water and Food and the CGIAR Research Program on Water, Land and Ecosystems (WLE). The authors thank the NGO Shushilan for data collec on throughout the southwest coastal zone of Bangladesh and the respondents. Camelia Dewan is also thanked for her work on the qualita ve data and especially coding of the transcripts that are used here. We would like to thank Adi Mukherji for her insigh ul comments on the earlier versions of this paper. References ADB. 2003. External evalua on: Small-scale water resources development sector project-I. Final Report. Dhaka: Asian Development Bank. ADB. 2008. Opera on and maintenance study: Final report. Incremental Support to TA-7041-BAN: Par cipatory Small Scale Water Resource Projects. Dhaka: Asian Development Bank. Araral, E. 2005. Bureaucra c incen ves, path dependence and foreign aid: An empirical ins tu onal analysis of irriga on in the Philippines. Policy Science 38:131-157. BIDS, B.I. 2008. Impact evalua on study: Report of selected 10 sub-projects, SSWRDSP-I, Benefits Monitoring and Evalua on (BME) Study, GoB, LGED, IWRM. Dhaka: Local Government Engineering Development, Government of Bangladesh. Chadwick, M. and A. Da a. 2003. Water Resources Management in Bangladesh: A Policy Review. Livelihood-Policy Rela onships in South Asia. London: Department for Interna onal Development. Chowdhury, A.K.M.J.U. and G. Rasul. 2011. Equity and social jus ce in water resource governance: The case of Bangladesh. South Asian Water Studies 2(2): 44-58. Cleaver, F. 2001. "Ins tu on, agency and the limita ons of par cipatory approaches to development" In Par cipa on: The New Tyranny, by B. Kothari, U. Cooke, 36-55. London: Zed Books. 160 De Silva, S. 2012. Li erature Review: The experiences of Water Management Organiza ons in Bangladesh. Dhaka, Bangladesh: CPWF. h ps://cgspace.cgiar.org/bitstream/handle/10568/35137/Lit%20Reviews_Water%20Management%20Organi za ons.pdf?sequence=1 Dewan, C. 2012. Literature Review: Review of the Historical Evolu on of Policies and Ins tu ons of Community Based Management in Coastal Bangladesh. Dhaka, Bangladesh: CPWF. h ps://cgspace.cgiar.org/bitstream/handle/10568/35139/Lit%20Review_Historical%20Analysis.pdf?sequence =1 Dewan, C., M.-C. Buisson, and A. Mukherji. 2014. The imposi on of par cipa on? The case of par cipatory water management in Bangladesh. Water Alterna ves 7(2): 342-366. Dewan, C., A. Mukherji and M.-C. Buisson. 2015. Historical evolu on of par cipatory water management in coastal Bangladesh: From indigenous temporary earthen embankments before 1960s to depoli cized community based management of Dutch styled polders in 2000s. Water Interna onal, forthcoming. Foucault, M. 1975. Surveiller et Punir. Naissance de la prison. Paris: Gallimard. Habibullah, M. 1996. The Cons tu on of the People’s Republic of Bangladesh, Government Prin ng Press, Tejgeon, Dhaka, 10-65. Islam, R. 2005. Managing diverse land use in coastal Bangladesh: Ins tu onal approaches. Paper Presented at the Interna onal Conference on Environment and Livelihoods in Coastal Zones: Managing Agriculture-Fishery-Aquaculture, March 1-3, Bac Lieu, Vietnam. Islam, M.T., and K. Fujita. 2009. Empowering Rural Administra on at the Lowest-level in Bangladesh: A Comparison with India. Journal of Regional Development Studies 13, United Na ons Center for Regional Development (UNCRD), Japan. Islam, M.T., and K. Fujita. 2012. Dimension of decentraliza on process and rural local Government in India: A comparison with Bangladesh. Kyoto Working Papers on Area Studies 130:1-26. Kyoto University, Japan. Lukes, S. 2005. Power: A radical view (2nd ed.). Basingstoke: Palgrave Macmillan. Ministry of Water Resources GoB. 1999. Na onal Water Policy. Ministry of Water Resources GoB. 2001. Guidelines Par cipatory Water Management. Ministry of Water Resources GoB. 2014. Par cipatory Water Management Rules. Mujeri, M. K., and L.S. Singh. 1997. Case studies on decentraliza on: Bangladesh. Working Paper prepared for SDA Technical Consulta on on Decentraliza on (TCD) FAO HQ ROME, CIRDAP Research Division Centre on Integrated Rural Development for Asia and the Pacific. North, D. C. 1990. Ins tu ons, ins tu onal change and economic performance. Cambridge university press. Quassem, M. A. 2001. Experience with opera on of par cipatory water management: Country paper – Bangladesh. Dhaka. Ullah, M.A., and S. Pongquam. 2010. Financial resources mobiliza on performance of rural local government: A case study of three Union Parishad in Bangladesh. Asian Social Science 6(11): 95-115. 161 The imposi on of par cipa on? The case of par cipatory water management in coastal Bangladesh C. Dewan1, M.-C. Buisson2 and A. Mukherji 3 1 University of London, England, c_dewan@soas.ac.uk Interna onal Water Management Ins tute, India, m.buisson@cgiar.org 3 The Interna onal Centre for Integrated Mountain Development, Nepal, amukherji@icimod.org 2 Abstract Community-based Natural Resources Management (CBNRM) has been promoted as part of the development discourse on sustainable natural resources management since the mid-1980s. It has influenced recent water policy in Bangladesh through the Guidelines for Par cipatory Water Management (GPWM) where community-based organiza ons are to par cipate in the management of water resources. This paper reviews the extent of success of such par cipatory water management. It does so by first discussing the changing discourses of par cipa on in Bangladesh’s water policy from social mobiliza on to decentralized CBNRM. Second, Bangladesh is used as a case study to draw a en on to how the crea on of separate water management organiza ons has been unable to promote inclusive par cipa on. It argues that the current form of decentraliza on through a CBNRM framework has not resulted in its stated aims of equitable, efficient, and sustainable management of natural resources; rather it has duplicated exis ng local government ins tu ons. Finally, it ques ons the current investments into community-based organiza ons and recommends maintenance funds should be increased and made permanent through exis ng funding channels. Keywords: Community-based natural resources management, par cipatory water management, local government ins tu ons, Bangladesh 1. Introduc on Community-based Natural Resources Management (CBNRM) is based on a simple and a rac ve assump on that communi es, defined by their dis nct and integrated social structure and common interests, can manage their natural resources in an efficient, equitable, and sustainable way (Blaikie, 2006). CBNRM has been promoted by most major Interna onal Financial Ins tu ons (IFIs) since the mid-1980s as part of the development discourse on sustainable natural resources management (Blaikie, 2006; Mansuri and Rao, 2003). Decentraliza on is defined in this paper as any poli cal act in which a central government formally cedes powers to actors and ins tu ons at lower levels in a poli cal-administra ve and territorial hierarchy (Ribot et al., 2006). It has been argued that any form of decentraliza on should increase efficiency, equity, and democracy "by linking the costs and benefits of local public services more closely" (World Bank, 1988: 154). CBNRM involves decentraliza on of power to community-based organiza ons where the underlying ra onale is that decentraliza on to communi es may increase local ownership, responsiveness to local needs and accountability to local people (Ingham and Kalam, 1992). Bangladesh has seen significant involvement of major IFIs and donors in promo ng CBNRM and decentralized water management in its water policy reforms. The Na onal Water Policy (MoWR, 1999) and the Guidelines for Par cipatory Water Management (MoWR, 2001), shi ed away responsibili es for water management from state implemen ng agencies to externally ini ated community-based Water Management Organiza ons (WMOs), with limited formal involvement of local government ins tu ons. To date, evalua ons of par cipatory water management projects in Bangladesh’s coastal infrastructure (embankments/polders, sluice regulators, canals) limit themselves to analyzing outcomes in light of stated project aims without ques oning the theore cal framework of CBNRM (cf. MoWR 2001, 2005; ADB 2007a, 2007b; BIDS, 2008; Fujita, 2011); or how WMOs as parallel structures may undermine or compete with the role of local government ins tu ons (Summers, 2001). 162 By analyzing the gaps between par cipa on in policy versus par cipa on in prac ce, this paper seeks to illuminate the weaknesses of decentralisa on of water management through CBNRM and its inability to address coastal water challenges, while highligh ng how it marginalises local government ins tu ons. The paper first discusses the changing discourses of 'par cipa on' in Bangladesh’s water policy. Second, it uses field data from coastal Bangladesh to evaluate CBNRM against its stated aims of efficient, equitable and sustainable water management. Third, it discusses these findings in rela on to the role that democra cally elected local governments play in water management. It will conclude that maintenance funds should be increased and made permanent through exis ng funding channels and that the role of local government in water management must be acknowledged. 2. Methodology This paper draws on large and original qualita ve and quan ta ve data sets. First, informa on on how donors have been important in shaping water policy were gathered through 28 Key Informant Interviews (KIIs) with government officials, donors, academics and project consultants from par cipatory water management projects in Bangladesh. They were asked about their experiences of community par cipa on, the role of the various stakeholders and the degree of 'success' of par cipatory approaches. The interviews were conducted in Dhaka from December 2011 to March 2012. Second, to understand how local popula ons from various socio-economic groups and interests perceive water management and the performance of community-based WMOs, 57 semi-structured Focus Group Discussions (FGDs) and 92 KIIs were conducted in the southwest coastal zone. This qualita ve work was conducted from January 2012 to September 2012 in five Bangladesh Water Development Board polders: P3 (19,430 hectares [ha]), P31 (14,831 ha), P30 (7209 ha), P24G (25,856 ha), P43-2F (5622 ha), and four Local Government Engineering Department sub-projects: Jainkathi (31 ha), Jabusha-Beel (1211 ha), Bagachra-Badurgacha (385 ha), and Latabunia (168 ha). These nine study areas were purposively selected from three different agro-ecological zones in coastal Bangladesh in order to capture differences in terms of environmental constraints (salinity, waterlogging) and their differing ins tu onal backgrounds for water management (small scale vs. large scale; managed by different government implemen ng agencies). In each selected area, FGDs were first conducted with a general group of community members and then separately with the execu ve commi ees of WMOs and with Labor Contrac ng Socie es consis ng of male or female day laborers. KIIs were held with local government officials (male and female, respec vely), project field staff, the execu ve chairs of WMOs, women WMO members, paddy farmers, shrimp farmers, women household heads and the landless (men and women, respec vely). Ul mately, eight FGDs were conducted with female only groups and 12 key women informants were interviewed. The resul ng 2000 pages of transcripts were then coded and entered into the Atlas Ti qualita ve analysis so ware. Queries were generated on percep ons of par cipa on, ability to influence water management and the state of the infrastructure, and were disaggregated based on the type of respondents. Third, descrip ve sta s cs from a quan ta ve survey conducted in a subset of the study areas (P3, P30, P43-2F, Latabunia, Jabusha and Jainkathi) were used in order to illustrate qualita ve findings. The survey drew a sample of 1000 representa ve households from 44 villages randomly selected in the study areas. 3. Water policies in Bangladesh: Par cipa on in theory 3.1 History of water projects: From top-down engineering to small-scale interven ons The current prac ce of CBNRM in the water sector of Bangladesh is closely ed to a long-standing discourse of people’s par cipa on and the perceived top-down mentality of government engineering departments. For example, the BWDB held key responsibility for irriga on, flood control, and drainage in Bangladesh from the 1950s to the late 1990s. It constructed over 100 embankments across the coastal zone in the 1960s to protect coastal communi es from flooding, established irriga on systems and employed local gatemen called 163 khalashis for the opera on of sluice regulators. This ini al construc on was seen as an infrastructural investment in the hands of engineers, without any par cipa on from, or consulta on with, local communi es. Nevertheless, the embankments, known as polders, with their canals and sluice gates became fundamental in the struggle against flooding and salinity intrusion while they simultaneously established themselves as a key source of water for agriculture, aquaculture, and other produc ve ac vi es. In the 1970s and 1980s, donors focused on projects that moved from top-down mega construc ons to small-scale local interven ons. Donors such as the Swedish Interna onal Development Agency (Sida) and the Embassy of the Kingdom of Netherlands (EKN) introduced and financed par cipatory projects to be implemented by the Bangladesh Water Development Board. They were to use social mobiliza on NGOs to organize excluded and marginalized groups to take part in income-genera ng maintenance work through local groups named 'Target Groups' for the poor and 'Landless Contrac ng Socie es'. This trend con nued un l the early 1990s and includes the Early Implementa on Project (1972-1995), the Land Reclama on Project (1978-1991) and the Delta Development Project (1981-1991) (Du a, 1997; Duyne, 1997). It was further emphasized that the landless and the poor were to become ac ve in the decision-making processes of water management. The emphasis on social equity and challenging power inequali es through focusing on the poor reflected a wider movement of par cipa on at the me. The 1970s and 1980s saw a prolifera on of social mobiliza on NGOs that promoted women’s empowerment and the strengthening of the rights of the landless. Notable NGOs were Nijera Kori ('We do it ourselves') and Gono Shahajjo Sangstha (GSS, 'People’s Help/Aid Organiza on') that effec vely encouraged their members to compete in local government elec ons and/or engage in local poli cs (Hashemi, 1996; Thörlind, 2000). These examples illustrate the early interest in par cipa on emerging in the 1980s within the NGO community, strongly influenced by Robert Chamber’s (1983) idea of 'pu ng the last first' to promote a power shi among stakeholders (Williams, 2004). 3.2 Paradigm shi in the 1990s: Par cipa on as maintenance Swedish and Dutch donors were funding both social empowerment NGOs and par cipatory water management projects, where the poli cized par cipa on in the former affected the discourse of par cipa on in the la er. However, as Mollinga (2008) points out, 'par cipa on' is a central theme in water policy discussions and has obtained several different meanings over me. There are thus divergent views on par cipa on, how it is defined, whom it is expected to involve, what it is expected to achieve and how it is to be brought about (Agarwal, 2001). Over the 1990s and 2000s, a depoli cized concept of par cipa on consolidated in the donor community, who came to prefer service delivery to social mobiliza on as the la er became too poli cally conten ous (Wood, 1994; Hashemi, 1996; Edwards and Hulme, 1997; Holloway, 1998; Sogge, 2002; Rahman, S., 2006; Dewan, 2009). By the 1990s, par cipatory discourse rapidly became part of the official aims and objec ves of governments and interna onal development agencies (Williams, 2004). However, the shi away from social mobiliza on changed the meaning of par cipa on to one that increasingly obscured power inequali es. Depoli cized terms such as 'stakeholder consulta on' replaced the use of 'Target Groups' that had explicitly focused on the poor and the use of 'Labor Contrac ng Socie es' removed the focus on the landless from 'Landless Contrac ng Socie es'. The early 1990s saw a growing tension between these compe ng meanings of par cipa on. On the one hand, civil society and NGOs promoted par cipa on as 'an end in itself', reflec ng the legacies of the 1970s and 1980s (Du a, 1997; Duyne, 1997; Hanche , 1997). On the other hand, donor-funded projects began to increasingly advocate par cipa on as a means to an end, the end being involving communi es for maintenance and upkeep of water infrastructures. In the la er, par cipa on was relegated to public consulta on, while it was used as an excuse for transferring responsibili es without delega ng actual decision-making power (Hanche , 1997: 278; Cornwall and Gaventa, 2001). Williams (2004) also suggests that enlis ng and demonstra ng 'popular par cipa on' became a crucial measure of scheme success and a key condi on of donor approval in development projects. Cri cs argue that policy labeled as 'par cipatory' or 'community-driven' provides more effec ve instruments to advance external interests and agendas while further concealing the agency of outsiders, or poli cal manipula ons of local elites (Cook and Kothari, 2001; Mosse, 2001). The problem of water management is inherently poli cal and as such a empts to keep poli cs out of it is fu le and even 164 counterproduc ve. As Mollinga (1998: i) aptly put it: "water management and use are contested at all these levels, that is, that water control needs to be understood as a poli cal process". Considering the different meanings and uses of 'par cipa on', Arnstein’s (1969) ladder of par cipa on (Figure 1) will be used to differen ate between the different levels of poli cized and depoli cized par cipa on in which water management organiza ons operate. 8 Ci zen Control 7 Delegated Power 6 Partnership 5 Placa on 4 Consulta on 3 Informing 2 Therapy 1 Manipula on Ci zen Power Tokenism Nonpar cipa on Fig. 1. Arnstein’s ladder of par cipa on (source: Arnstein 1969). The top of the ladder envisages par cipa on similar to that of the empowerment work of social mobiliza on NGOs of the 1970s and 1980s, where the redistribu on of power enables ci zens presently excluded from the poli cal and economic processes ('have-nots') to be deliberately included in decision-making. This is directly ed to discussions of ci zen power, where Ribot et al. (2006) define ci zenship as the right and ability of people to be poli cally engaged and shape the fate of their polity. As such, high levels of ci zen power may therefore also correlate with democra c par cipa on. The lower half of the ladder reflects a depoli cized approach to par cipa on, or 'tokenism', where par cipa on is limited to informa on, consulta on, and placa on. In Bangladesh, the depoli cized and 'mainstreamed' version of par cipa on came about in a context of a wider decentraliza on agenda, where 'community par cipa on' included the devolu on of responsibility over O&M from the state to communi es, while state agencies such as the BWDB were being simultaneously weakened. For example, in 1992, the World Bank recommended that the Land and Water User Directorate would be closed, ending the unit that provided BWDB with the staff and exper se to interact with local water users and farmers (MoWR, 2005). The BWDB Act of 1998 reduced staff size from 24,000 to 8000, replacing government-employed gate operators with operators who were to be appointed and paid by communi es, while many of the staff that had worked with the empowerment projects of the 1980s re red and were not replaced. The Na onal Water Policy of 1999 formally transferred responsibility over O&M to WMOs (MoWR, 1999). Despite a considerable reduc on in size and the closure of Land and Water User Directorate, BWDB was now required to engage with communi es on ma ers of water management (MoWR, 2000). As a result, both the BWDB and the Local Government Engineering Department (LGED) have relied heavily on donor funding to implement par cipatory projects in order to comply with the Na onal Water Policy and the Guidelines for Par cipatory Water Management (GPWM) (MoWR, 2001). Despite the Na onal Water Policy’s a empt towards decentraliza on, in the GPWM no formal men on is made of local government ins tu ons beyond that they 'raise awareness' of water management issues and suppor ng, facilita ng and coordina ng assistance to the concerned WMOs (MoWR, 2001). It frames a decentraliza on agenda where the central government transfers powers to private actors, in this case WMOs, 165 rather than democra cally elected local ins tu ons (Larson, 2003). As will be discussed later in the ar cle, local governments are nevertheless highly ac ve in water-related issues and coordinate various development projects and social programs in local areas. The crea on of water management organiza ons in order to implement projects may therefore, as Summers (2001) points out, duplicate the func ons of local government in a way that detracts funding and legi macy away from exis ng democra cally elected local ins tu ons. The next sec on discusses how the GPWM with its depoli cized framework and limited conceptualiza on of par cipa on is prone to tokenism. 4. Par cipa on in prac ce: Top-down blue prints Figure 2 illustrates the key par cipatory stages of the GPWM that implemen ng agencies BWDB and LGED must follow when execu ng new water projects. The GPWM aims to ensure community ownership and involvement in water management, both in planning, decision-making, and financial and physical par cipa on. 1. Iden fica on/Pre-feasibility study 2. Feasibility study 3. Detailed planing, design and stakeholders’ ins tu on building 4. Implementa on and trial opera on 5. Opera on and Maintenance 6. Monitoring and Evalua on Fig. 2. Guidelines for par cipatory water management (source: Ministry of Water Resources 2001). 4.1. Lack of efficacy: Top-down and sub-op mal planning of infrastructures The first three stages in the GPWM (Figure 2) aim to ensure that local stakeholders have ample opportuni es to provide feedback and shape water management projects. The GPWM requirement of feasibility studies sought to address the perceived top-down planning associated with BWDB and place the decision-making power in the hands of local stakeholders through the WMO. This component was integrated into the methodologies of LGED’s Small-Scale Water Resources Development Sector Project (SSWRDSP) for sub-projects under 1000 hectares and BWDB’s Integrated Planning for Sustainable Water Management (IPSWAM) for larger projects. IPSWAM is seen as one of BWDB’s most successful examples of par cipatory water management and was implemented in nine out of 123 polders.15 For IPSWAM and other BWDB polder communi es a key constraint arose with the requirement of WMOs; they must create community organiza ons 50 years a er the ini al construc on of polders in order to receive government assistance for maintenance and rehabilita on. BWDB polders struggle with the lack of resources needed to create and sustain the WMOs required for accessing rehabilita on and maintenance funds.16 G3’s fieldwork found that without a project budget to create and support WMOs in the local communi es, BWDB field engineers would rarely consult communi es on periodic maintenance, e.g. where to excavate canals or repair the embankment.17 Instead, BWDB tend to use use external contractors rather than hiring local people, a prac ce seen as removing rural employment opportuni es18. 15 Two of the five BWDB study areas (P30 and P43-2F) were included in IPSWAM. FGDs in P3, P31, P30 and P24G from February to August 2012. 17 KIIs with BWDB Upazila field engineers in Khulna and Satkhira Districts, February to August 2012. 18 FGDs from P3, P31, P30, P43-2F and P24G, February to September 2012. 16 166 The difficulty of incorpora ng local feedback was also evident in LGED ‘s SSWRDSP sub-projects. The data collected revealed several examples of inadequate technical solu ons. These include an unsa sfactory number of regulators, too low or weakly constructed embankments, flawed sluice gate shu ers, and superficial canal re-excava on. The inability to incorporate local needs was further respondents in these sub-projects sta ng that their request for a larger, wider, and more robust embankment had been ignored, resul ng in the embankment now being in poor condi on.19 A striking example of poor design can be found in Badurgacha-Bagachra where the gate is operated provisionally using bamboos and rope, a consequence of LGED s disregard of local residents’ request for a steel shu er rather than a now-broken wooden shu er. Similarly, respondents in the Jainkathi sub-project stated that they had warned LGED about placing regulators on private land. The result was that the landowner took control over the infrastructure; consequently, only one out of two regulators in Jainkathi is ac ve and the second canal in the sub-project has become silted and unproduc ve. The emerging picture from LGED sub-projects is that the final decision-making power over physical construc on also remains largely in the hands of the implemen ng agency rather than those of the WMO or community, a finding supported by an evalua on of the LGED project (ADB, 2003). LGED has been able to ins tu onalize community engagement (local contact at sub-district level, permanent coordina ng unit at headquarters). However, even with such ins tu onalized support, the sub-projects s ll suffer from flawed technical problems. The experiences of IPSWAM and SSWRDSP provide a case to ques on the efficacy of the feasibility studies and forma on of WMOs made mandatory by GPWM. Rather than providing a high degree of ci zen power, the current arrangement seems to fit on the lower end of Arnstein’s ladder, near 'tokenism' through placa on, consulta on, and informing. The degree of ci zen power, i.e. to exert control in decision-making, at this stage seems to be that of par cipa on that is limited to a specific project interven on and then only through the channel of an externally ini ated non-func onal water management organiza on. Though the GPWM a empts to involve and empower local communi es in implemen ng agencies, the current guise of par cipa on is perhaps used as a 'tool' to give a 'human face' to depoli cized and technocra c projects (Palmer-Jones et al., 2010). This may reflect a mismatch of incen ves between project implementers and stated goals of the projects, where more tangible and measurable goals like physical construc on of infrastructure can easily take precedence over longer-term goals like par cipa on and empowerment (Mosse, 2001; Araral, 2005). This type of CBNRM is therefore arguably ineffec ve in involving local stakeholders in decision-making. 4.2 Lack of equity: WMO obscuring power differences within communi es 4.2.1 Elite capture of WMOs The GPWM s pulates broad involvement of local stakeholders from all cross-sec ons of society. In all study sites, WMO members had internalized the rhetoric of par cipa on and broad stakeholder involvement where statements such as, "The water management commi ee is formed with equal emphasis to all classes of people, nobody is excluded from the commi ee" were common.20 However, in prac ce the WMO composi on o en consisted of teachers, local poli cians, and businessmen who in some instances were either not directly involved in using water for produc ve uses, or could benefit greatly from deciding the distribu on of water. These findings are corroborated by other studies that found frequent domina on of rural male elites in WMO execu ve commi ees in Bangladesh preven ng par cipa on of general people (MoP, 2005; Rahman et al., 2007; ADB, 2007a, 2008; BIDS, 2008; Nowreen et al., 2011). 19 20 Code: 'Q3. Embankment_condi on_poor'. Frequency: 48 total in the four small LGED sub-projects. FGDs in P31, P30, P43-2F, Latabunia, Bagachra-Badurgacha, Jainkathi and Jabusha, February to June 2012. 167 90 80 70 60 50 In the population* 40 30 In the WMOs executive committee members 20 10 0 Small Medium Large farmer (less farmer (2.5 farmer than 2.49 to 7.49 (more than acres) acres) 7.5 acres) Fig. 3. Elite frequency in execu ve commi ee composi on (source: CGIAR Challenge Program on Water and Food G3 quan ta ve survey (IWMI, 2013)). Note: * From 1000 representa ve households selected in polders and sub-projects; 1 acre = 0.405 ha. Figure 3 illustrates how the WMO execu ve commi ee does not reflect the composi on of the local community. One of the most common percep ons in the qualita ve survey was that the ability to par cipate is defined by power and economic status,21 where non-elites are excluded from par cipa on in water management.22 In par cular, the majority of poor and women respondents stated that they did not have any informa on on the ac vi es of the WMOs, WMOs were not seen at pla orms where voices of ordinary men and women are heard. This confirms findings from the anthropological literature on CBNRM in general, that though community par cipa on is meant to involve and benefit all sec ons of the community, they can effec vely exclude significant social segments, such as women (Agarwal, 2001; Sultana, 2009), while masking power dynamics and inequality (Kothari, 2001; Mosse, 2001, Bandiaky, 2008). 4.2.2 Quotas and tokenism: Exclusion of women and landless Coastal water infrastructure projects in the coastal zone of Bangladesh have aimed at flood control and at suppor ng the most visible produc ve uses of water, e.g.; irriga on for paddy cul va on and water supply for shrimp farming. This la er focus has o en ignored other uses of water, where women use a variety of water sources, such as ponds, rivers, canals for produc ve (kitchen garden, livestock) and domes c purposes. WMOs are solely in charge of the produc ve uses of water and rarely consider other water uses that are par cularly important for women: drinking water, bathing, sanita on, livestock and homestead garden irriga on (Crow and Sultana, 2002; Faisal and Kabir, 2005; MoWR and EKN 2008; Clement, 2012). For example, Crow and Sultana (2002) in their case study of coastal Bangladesh found that the neglect of the mul ple uses of water in polder management can adversely affect women – shrimp farming affects them through loss of ponds and saliniza on of water. For women, this has increased me to fetch water and find suitable places for bathing, poorer nutri on due to decreased vegetable cul va on and increased reliance on the cash economy for food items such as rice and fish. The expanding use of groundwater for irriga on has caused many hand pumps used for drinking and domes c water to run dry, worsening women’s tasks to fetch safe water especially in arsenic-contaminated areas (ibid). A major ra onale for women’s par cipa on in WMOs is therefore that it can improve the integra on of their needs within water management and therefore improve their livelihoods (Clement, 2012). 21 22 Atlas Ti code: 'Par cipa on:power_economic_status'. Frequency: 109. Top 3rd code. Atlas Ti code: 'Par cipa on:exclusion_general'. Frequency: 89. Top 5th code. 168 Equity is a key aim of CBNRM as communi es are perceived as able to manage resources for the common interest. In order to ensure 'women’s par cipa on', the guidelines (GPWM) s pulate that one-third of the execu ve commi ee members must be women. However, household surveys in the study sites reveal that 80% of execu ve commi ee members are male and less than 20% are women. Previous project evalua ons have argued that women WMO representa ves are o en token members with no real power in WMO decision-making processes (ADB, 2003; MoP, 2005). During the data collec on for the qualita ve survey, it was difficult to locate women WMO members for KIIs in the majority of sites. FGDs revealed that though women are formally included in the WMOs, they are not no fied of, or involved in, water management mee ngs. This appears to be connected to an inherent bias against women’s involvement in water management expressed both explicitly, "Women should not be involved in this work [water management]",23 and implicitly through the forma on of WMOs consis ng of only male elites who used their spouses to complement the quota requirement of the project. A majority of women respondents emphasized the importance of drinking water and water for food, yet would s ll state that formal water management and WMOs belonged to the male domain. "No, I am not involved in any water management organiza on. I am a woman, why will I be involved there"?24 The above findings of women being marginalized in the par cipa on process are corroborated with other studies on gender and CBNRM (Cornwall and Gaventas, 2001; Agarwal, 2001; Sultana 2009). Agarwal (2001) argues that par cipatory ins tu ons can exclude people through 'par cipatory exclusions' that can individually or interac vely constrain a woman’s par cipa on in natural resources management. She iden fies these exclusionary mechanisms as rules of entry (e.g. only one member for each household in WMOs), social norms of women’s behavior and ac ons (mee ngs held in public spaces deemed inappropriate for women), social percep ons of women ’s abili es (unknowledgeable, 'illiterate'), all of which are exclusions expressed by our respondents. Some excep ons were found in IPSWAM’s polders 30 and 22, where gender awareness training of both male and female WMO members was perceived to have increased the confidence in women engaging as ac ve execu ve commi ee members in the WMOs (BARD, 2009; EKN and BWDB, 2011). This was further facilitated by long-term empowerment ac vi es taking place during the 1980s Delta Development project implemented by the social mobiliza on NGO Nijera Kori.25 The women execu ve commi ee members in both these polders were vocal and proac ve in the WMO and its water management decisions. Nevertheless, in both cases these women were married. In Polder 22 respondents stated that "women without husbands have nothing, no food, no clothes",26 indica ng that the gendered issue in water management is further divided along socio-economic lines. As Agarwal (2001) points out, by having quotas for women, the differences between women in a locality may become obscured, and ins tu onalize exclusions and privilege where rich or elite women hold the nominal memberships. Women-headed households revealed a sense of exclusion from most ins tu ons due to their marital status as divorcees or widows. In addi on, women who are landless, poor and/or of religious minori es o en expressed that they lack the social standing to par cipate in decision-making processes. When asked why they were not members of a WMO frequent replies were: "[w]e are women, poor and Hindu, why would they listen to us? Nobody hears us, nobody cares about us (…). We only go to mee ngs to provide our signature".27 Furthermore, adding women to a project does not necessarily address power issues between men and women, and does not capture that many poor and marginalized men are excluded (Agarwal, 2001). The guidelines (GPWM) recognize the par cular vulnerability of those without landholding by requiring at least one landless representa ve in the WMO execu ve commi ees. This recogni on is important because, though they do not own the land on which they work, they are also local stakeholders who are affected in various ways by issues of water management. For example, it may inhibit their right to fish for themselves in public canals or reduce or increase their chance of employment depending on whether aquaculture or agriculture is pursued.28 However, 23 KII, Shrimp farmer, Tildanga Union, P31, 12 March 2012. KII, Woman household head, Kaliganj Union, P3, 16 February 2012. 25 KIIs, Former BWDB consultants to the Delta Development Project, Dhaka, February 2012. 26 FGD, WMO execu ve commi ee in Dumuria Union, P22, 8 December 2011. 27 FGDs, Women day labourers’ earthwork groups (LCS) in P3 and P31, also the most conflicted aquaculture areas. 28 FGDs with day labourers’ earthwork groups (LCS) and KIIs with landless respondents in P3, P31, Bagachra-Badurgacha and Latabunia. 24 169 a majority of the WMOs lacked landless representa ves in both the general and execu ve commi ees and when landless members have been found in these commi ees, they have always been male, thus ignoring the par cular social exclusion of women of the poorest class. A key weakness of the CBNRM’s focus on 'community' is that by viewing the WMO as capable of represen ng the interests of a homogenous 'community', it ignores the various levels of conflict of interest among rural popula ons. The mechanism of quotas as they have been implemented thus far generally fails to empower these target groups in the decision-making processes of WMOs. 4.2.3 Rural inequali es and conflict: Opera on for compe ng water uses The inability of implemen ng agencies and of the GPWM to address social inequali es and compe ng interests is further revealed by the conflicts between different water users in the opera on of sluice gates that regulate the water entering the polders and the usage of canals that distribute the water inside the polders. With the crea on of polders in the 1960s, BWDB employed government-funded gatemen to operate the gates according to set protocols and through local requests. Since 1999, local communi es have to fund and operate sluice gates themselves. The study found that sluice gate commi ees were created regardless of the presence of a WMO, where local operators would be paid through rice or fishing rights in the canal. This was the case even where LGED’s WMOs s ll con nued to formally exist ten years a er they had first been created. Figure 4 below illustrates the various ways in which operators for different gates may be appointed in the different polders. 100 90 80 70 60 50 40 30 20 10 0 Informal arrangements (unpaid) Informal arrangements operator (paid) WMO appointed operator (paid/unpaid) Local operator (unpaid) Local operator (paid) From 1000 representa ve households selected in polders and sub-projects, households answered on the opera on of the most important sluice gate for their produc ve ac vi es. Fig. 4. Distribu on of different modes of opera on of sluice gates across six case study polders. Source: G3 quan ta ve survey (IWMI, 2013). For all SSWRDSP sub-projects, the WMO is responsible for a limited number of gates in a small area (less than 1000 ha). Figure 4 establishes that only a minority of the gates is operated through the WMOs, with a considerable number of informal arrangements. Through our qualita ve study, we found that even in an area as small as Jainkathi, the WMO can only control one of the two gates due to a land dispute. In Latabunia, shrimp farmers have created private gates and pipes that they regulate unilaterally, rendering the WMO powerless. In Jabusha, fac ons in the WMO have led to leasing and blockage of the canals used for aquaculture, which disrupts irriga on through the main canal. The qualita ve survey revealed how water management is affected by a leasing system dominated by influen al elites. Due to their autocra c use of canals and their domina on over opera on of sluice gates, existence of a gate commi ee becomes redundant. Instead, the canals are drying up, o en as a deliberate a empt to increase landholding size.29 For regular farmers, the slow annihila on of the canals is detrimental as their main source of irriga on is removed. As such, 'canal grabbing' impedes effec ve, equitable, and sustainable water management. 29 FGDs and KIIs from BWDB polders P31, P3, P30, 24G and P43-2F and LGED sub-projects Bagachra-Badurgacha, Jabusha, Patuakhali, Jainkathi, February to September 2012. 170 Some canals are possessed by powerful peoples through corrup on. They use these canals as they wish. We cannot excavate, repair canals, or maintain water drainage. These privately possessed canals cannot retain water. So there is no way for us to cul vate our crops.30 The informal pipes, gates and incisions to the embankment by aquaculture interests increase the risk of flooding during disasters and reduce agricultural yields by increasing the salinity level of the land. A group of day laborers in Bagachra voiced their feeling of exclusion from any real 'par cipa on'. If they would listen to our voice, they could stop aquaculture and stop drawing in saltwater. All of the influen al people are prac cing aquaculture using saltwater. Actually, they get much benefit from this, but we are not ge ng anything. We are going from poor to poorer.31 The problems outlined above are arguably related to the depoli cized view of water management in the CBNRM inspired GPWM. Our respondents did not equate par cipa on with a water management organiza on. Instead, the qualita ve findings suggest that the WMO is an external idea that prompts polder popula ons to create commi ees as a condi on to receive project funding. Without a working mechanism to ensure that power differences and the needs of marginalized stakeholders are taken into account, CBNRM cannot be truly considered par cipatory and equitable, rather the use of 'community' in the Bangladeshi context is based on a flawed assump on of shared common interests. 4.3 Lack of sustainability: The panacea of financial cost-sharing With the debate on 'par cipa on' in the 1990s, there was an increasing focus on 'par cipa on' as a means of local stakeholders financially contribu ng to O&M. With the GPWM, communi es were now encouraged to contribute financially; fully for minor maintenance costs and par ally for periodic maintenance defined as canal excava on and embankment repair, while being ac ve in the regular upkeep and maintenance of infrastructure. In addi on to transferring the full responsibility of opera on of the gates from a state-employed gatekeeper to 'communi es', the Na onal Water Policy of 1999, also s pulates that WMOs takes full responsibility for 'minor', or day-to-day, maintenance. Therefore, given the new discourse on par cipa on, its outcomes have to be measured not only in terms of empowerment of communi es or their voices in decision-making, but in terms of state of maintenance of infrastructure. Below, we assess to what extent this decentraliza on to communi es has supported maintenance of the coastal water infrastructure. 70 60 50 40 30 20 10 0 Very bad condition of the gate Very bad condition of the canal Very bad condition of the embankment Fro m 1000 representa ve hou seholds selected in polders and sub-pro jects households were asked to rate the co ndi on of the most important wa ter infrastructures (gate, canal, emb ankment segment) fo r their produc ve ac vi es. Fig. 5. Households’ percep on of the quality of water infrastructures across six study polders (source: CGIAR Challenge Program on Water and Food G3 quan ta ve survey (Buisson et al., 2013)). 30 31 FGD General, Khona village, Pankhali Union, P31, 10 March 2012. FGD, Male earthwork group, East Bagachra, Shobhona Union, Bagachra-Badurgcha sub-project, 24 March 2012. 171 Figure 5 highlights a poor state of maintenance. Arguably, WMOs have not only performed sub-op mally in terms of processes of community mobiliza on through elite capture and marginaliza on of the voices of the poor and the women, but also in their results and outcomes (O&M). The problems of maintenance are also seen in the LGED’s flagship par cipatory water management projects, the SSWRDSP. LGED’s division of labor between units is well defined since major donors have promoted standardiza on through long and comprehensive technical assistance. A permanent unit has been created within LGED to promote coordina on of water management coopera ve associa ons from various phases (there are currently four successive SSWRDSP projects running), while ins tu onalizing coopera on with several different agencies (BWDB, departments of coopera ves, agricultural extension, and fisheries) (Fujita, 2011). According to our key informants, LGED’s success is o en a ributed to its 'water-plus' approach that combines water management with micro-credit and income-sharing ac vi es, a financial incen ve perceived to facilitate the longevity and con nuity of their WMOs. In LGED’s WMOs, local stakeholders contribute a monthly fee to be a member of the WMO, seen as a means to increase their sense of ownership over the water infrastructure, while providing funds for local micro-credit. The accrued interest will then be used for WMO savings and maintenance funds (LGED, 2012). The 'water-plus' approach proved popular among other donors and was at the me of fieldwork being incorporated into the planning of BWDB’s Dutch-funded Blue Gold project.32 WMOs created by all LGED sub-projects that we studied were s ll ac ve in one form or another even up to ten years a er their crea on, while a majority of those created by BWDB in Polders 31 (Fourth Fisheries Project, World Bank) and Polder 24G (Khulna-Jessore Drainage Rehabilita on Project, Asian Development Bank) are conspicuous by their absence. However, even though LGED-created WMOs are s ll ac ve, this does not necessarily mean that they have carried out their mandated tasks of O&M. In fact, the amount spent by WMOs on maintenance is less than micro-credit amounts disbursed (LGED, 2009). Rather, respondents across the study sites reported high levels of default of micro-loans. Despite having ac ve WMO coopera ves, the state of the embankment, canals, and gates tended to be poor in the four LGED sub-projects reviewed. When the WMO-financial incen ve leads to higher default than repayment without contribu ng to effec ve water management, it is unlikely that micro-finance in itself is key to a sustainable WMO and warrants cau on for promo ng micro-finance as a sustainable prac ce for CBNRM. It is equally doub ul that current cost-sharing arrangements are realis c given maintenance demands in the coastal zone. The two most frequently men oned issues from the qualita ve survey was that increased excava on of canals is necessary33 and that the canals are heavily silted in a way that disrupts their func on to retain and distribute water: 34 "The Bhadra River is now the Mora (dead) Bhadra".35 The high frequencies of both responses were prevalent throughout the study sites and were followed by problems rela ng to damaged embankments and inac ve sluice regulators. In Bagachra-Badurgacha, despite funds from leasing out canals, the WMO funding was insufficient and required addi onal amounts from LGED that, in turn, found itself reques ng funds from the current donor. This is an example of how WMO funding from membership collec on was insufficient to pay for half of the periodic maintenance costs. Moreover, GoB itself lacks the funding required for maintenance, as illustrated by Figure 6. 32 KII, Embassy of the Kingdom of Netherlands to Bangladesh. Dhaka, 28 February 2012. 33 Atlas Ti code: 'SUG:maintenance_re-excava on of canal'. Frequency: 218. 34 Atlas Ti code: 'Khal_condi on:silted'. Frequency: 163. 35 FGD, Male earthwork group, Shobhona Union, Bagachra-Badurgacha sub-project, 24 March 2012. 172 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 Es mated Demand for GoB (BWDB) Actual Expenditure by GoB (BWDB) Fig. 6. Funding gap in maintenance (source: Adapted from BWDB data 36). LGED suffers from a similar problem, but on a much lower scale, and o en fails to finance its part of the cost-sharing arrangements (LGED 2009). This funding gap in maintenance has therefore led to an oversubscrip on to LGED’s Emergency Fund and BWDB’s Non-Revenue Development Budget as a financial source for addi onal maintenance costs. The reality of financial decentraliza on is that the concept has remained alien to local stakeholders, who state that it is unrealis c for them to contribute half of the required maintenance costs given the problems of con nuous and extensive silta on that congests both rivers and canals, while river erosion and cyclones regularly damage and weaken embankments. This regular need for major repairs is costly, me-consuming and beyond the capacity of WMOs and local communi es. A significant weakness of the GPWM is therefore that it does not take into account the considerable hydrological challenges in the coastal zone, the consequent funding deficiency or the means by which WMOs can generate such large funds. Instead, donors tend to a ribute blame for deferred maintenance to communi es, who they argue are unwilling to contribute financially (ADB, 2007a). In contrast, others have argued that donors are to blame, as they prefer 'visible projects' (Araral, 2005). Instead of providing con nuous maintenance support, donors have let maintenance lapse a er the project interven on is concluded, effec vely rese ng the en re rehabilita on process.37 Arguably, the cost-sharing approach has not been able to ensure the sustainability of community-based water management organiza ons or their contribu on to water management. In addi on, the 'flagship' examples of successful par cipatory water management projects only take place in a limited number of polders in the coastal zone. These projects also require substan al funding for mobilizing and sustaining WMOs. LGED ’s SSWRDSP project has seen consistent funding from 1994 to 2017; arguably, the LGED project has never really ended. This makes it difficult to compare it with BWDB polders where project funding has terminated. Most BWDB WMOs tend to collapse within two years of the end of a project interven on. In the case of IPSWAM, WMOs were in limbo from 2011 to 2012, awai ng addi onal project funding. Furthermore, though IPSWAM is deemed a 'successful' par cipatory project, most of its polders reviewed are less than 7,000 ha. With difference in size comes difference in size of communi es and number of villages that must be mobilized. Many LGED projects span just one village, making community mobiliza on rela vely easier than in most BWDB polders that are above 5000 ha spread across ten or more villages. Furthermore, since the late 1990s BWDB has undergone ins tu onal reforms that have significantly reduced its manpower, 36 37 Email correspondence. Planning Wing, Bangladesh Water Development Board. Dhaka, 13 April 2012. KIIs with project consultants for BWDB’s IPSWAM and LGED’s SSWRDSP, Dhaka, February 2012 as well as LGED, 2010. 173 community par cipa on exper se and local field presence. To both create sustainable WMOs in a true par cipatory process and to be able to respond to the various requests and needs of these communi es with over 1.21 million ha of land, BWDB requires a large number of mul disciplinary manpower, which they cannot hire without substan al and long-term external funding. To quote an ex-BWDB official: "[w]e are engineers. We cannot promote community ownership. We cannot manage these things, nor can we hire the people that can. They [donors] just forced community par cipa on on us". Lack of sustainability is a key reason to ques on the validity of the current model of WMOs, where examples of successful par cipatory water management projects are represented by costly processes that are difficult to replicate, a finding which matches those of a global review of par cipatory water projects (Mukherji et al., 2009). The use of WMOs is par cularly unsustainable given their record of being unable to address the acute maintenance challenges facing the coastal zone, making local communi es increasingly vulnerable to cyclones and rising sea levels. The CBNRM model in Bangladesh’s coastal water management is unsustainable for two key reasons. One is that this form of decentraliza on does not equip WMOs financially or structurally to deal with an ongoing and accelera ng water management crisis. Second, it is a costly and me-consuming process to create and sustain these WMOs, resources that could have been allocated to state and local government ins tu ons to be er equip them with the acute issues of periodic maintenance. 5. Filling the gap: The role of exis ng ins tu ons Local government ins tu ons (LGIs) are not formally acknowledged in the GPWM, but play an important role in water management. For example, the Union Parishad (UP) arranges for evacua on when alerted to cyclones, while it also organizes immediate repair in the face of embankments breaking during disasters. Similarly, when the WMO system fails to address acute maintenance needs, the Upazila Parishad responds to the requests of its cons tuents by using rural employment schemes such as KABHIKA (Food for Work), KABITA (Cash for Work) and 40-day Work Order allocated from the to maintain roads, embankments, and canals. It does so through either sub-contrac ng it to the Union Parishad, NGOs or LGED, who then organize the forma on of Labor Contrac ng Socie es consis ng of the rural and disadvantaged poor. Such rural employment schemes are popular and a majority of respondents suggested that permanent funding should be made available to these schemes to address silta on and river erosion. However, even in these schemes, there were problems with the canals not being properly excavated and respondents have suggested that there is an independent body that measures the depth of canals before and a er an interven on and that it will have the authority to pay the sub-contrac ng party only a er the work has been successfully completed. In BWDB polders where the WMOs have become inac ve a er the project has ended, such as in P24G in Jessore, P31 in Dacope and P30 in Khulna, and in areas without formal par cipatory projects, e.g. in P3 Satkhira, the local governments in the form of UPs are ac ve members of gate commi ees. They are o en the first point of contact for water management issues, "If we face any problem, we inform the Union Parishad",38 and are part of the decision-making process rela ng to gate opera ons and coordina on of different requests. Since the UP representa ves are democra cally and locally elected, this arrangement is seen as generally favorable in managing disputes of opera on, though incidences of elite capture have also been men oned.39 In contrast to the lack of confidence in externally ini ated WMOs, respondents from various categories voiced that they perceived the UP as locally accessible, accountable and working for the local community. The second most frequently men oned sugges on in the qualita ve survey, a er canal excava on, was that the role of UP in water management should increase.40 The data collected generally depict the UPs as having a unique posi on as grassroots representa ves situated within the local government system. This allows them to further coordinate between various sectors, from drinking water to agriculture, fisheries, infrastructure and health, 38 Atlas Ti code: 'Union Parishad:percep on_First point of contact', Frequency: 82. Top 9th code. A key excep on to this was found in P3, where the UP officials tend to be part of the system of influen al elites/shrimp farmers. This is discussed in-depth by De Silva (2012). Similar cri cims and allega ons of corrup on and collusion were voiced in Polder 31’s Tildanga Union, Latabunia and Jabusha, where financial interests from the aquaculture industry are meshed with local poli cs. 40 Atlas Ti code: 'SUG: increase_role_water_management_UnionParishad'. Frequency: 91. Top 2nd code. 39 174 and thus help avoid the replica on and duplica on that otherwise tend to occur in a 'project' approach. This is in stark contrast to the isolated task of the WMO. Figure 7 below illustrates the gap between par cipa on as seen by the GPWM versus par cipa on as seen by the local Bangladeshi people themselves. In total, more than 70% believed that the Government (BWDB, UP, LGED) should be responsible for water management, with a majority favoring the UP. Figure 7 illustrates that respondents preferred the UP to act in water management ques ons (35%) over temporary project en es such as WMOs (2%). LGED 9% Other 2% Community people 24% BWDB 28% WMO 2% UP 35% Fig. 7. Percep on of the responsibility for water management (source: CGIAR Challenge Program on Water and Food G3 quan ta ve data (IWMI 2013)). From 1000 representa ve households selected in polders and sub-projects, households were asked who should act to solve the water-management problems. Any ins tu on situated in a context of deep inequali es and social strife is prone to elite capture (Bardhan, 2000). Local government ins tu ons such as the UP in Bangladesh and Gram Panchayat in India are just as prone to elite capture as WMOs and are as likely to be corrupt and exclude the voices of the poor and the marginalized (Bardhan, 2000; Khan, 2004; Lewis and Hossain, 2008). Increasing the role of local government ins tu ons in water management is not a panacea. They have their own share of problems ranging from several factors that impede their responsiveness and accountability to the people (As-Saber and Rabbi, 2009). This encompasses role confusion and a lack of authority and accountability between local poli cians (the UP) and local bureaucrats (at Upazilla/district level) (Toufique and Turton, 2002) to being weak and constrained by the central government through regula ons and inadequate local resources (Toufique and Turton, 2002; As-Saber and Rabbi, 2009). However, there are several points that favor inclusion of UPs in water management. First, chairmen and members of UPs are already involved in water management and other local development ac vi es, including local employment genera ng earthwork ac vi es. Second, unlike WMOs, whose funding is restricted through specific donor projects, UPs can access wider arrays of developmental funds from the Upazila level that they can deploy more effec vely for water-management-related work. Finally, Union Parishads are subjected to regular local elec ons. As elected representa ve of the people, the local residents demand these services from them and given the nature of electoral poli cs, UP members feel obliged to meet these demands. Thus, the polls might exert a posi ve pressure on the UP members in favor of their electorates needs and in favor of accountability. The deficient electoral process of the WMOs and the lack of long-term vision prevent this pressure to work in the case of the community organiza ons. 175 While there is considerable change required to strengthen LGIs to become more accountable and effec ve (As-Saber and Rabbi, 2009), each successful local elec on means that local governments, their power and authority are validated by the electorate. Strengthening of grassroots democracy through regular elec ons and inherent compe on for votes in a mul -party democracy is likely to lessen chances of elite capture and exclusion (Lewis and Hossain, 2008). Moving beyond the CBNRM model could involve more focus on democra c decentraliza on of water management through local governments. Though it must be noted that this is not a 'silver bullet', it could, however, contribute towards a democra c decentraliza on that is (a) more efficient as they can coordinate between various departments at the local level while using exis ng channels for maintenance; (b) more equitable as they face re-elec on and are therefore held accountable to their voters to a greater extent than the WMO execu ve commi ees where elec ons are exclusive to those with economic or social power; (c) more sustainable as it would involve a shi away from temporary projects to strengthening exis ng government channels such as Food/Cash for Work schemes, and in the process give focus on making these ins tu ons more responsive and accountable to their ci zens. Yet, despite their merits and performance in water management, local governments are neglected in favor of WMOs. Arguably, one key weakness with the CBNRM model is that it acts as an alterna ve decentraliza on working in parallel with exis ng local government ins tu ons, while simultaneously enabling capture of resources. 6. Conclusion The GPWM were established to ensure that local people from all segments of society could influence water decisions that affected them, with a par cular emphasis on the control of gates and canals. At the same me, it departed from previous discourses of people s par cipa on by focusing on decentralizing responsibility to local stakeholders, rather than mobilizing their degree of decision-making on development outcomes. Furthermore, it imposed par cipa on and CBNRM on the main implemen ng state agency, the BWDB, while having removed its Land and Water User Directorate and reduced staff who had the exper se to engage and consult with local communi es. It was also apoli cal in its nature by limi ng representa on of local stakeholders to externally created community-based water management organiza ons, thus obfusca ng deep inequali es embedded in society. The GPWM model of quotas has resulted in high degrees of tokenism among women and landless representa ves, two groups that are rarely involved in decision-making processes. As externally ini ated commi ees, these WMOs tend to lack both transparency and accountability through their ar ficial elec ons, and instead become resources for elites. This model is unable to address underlying conflicts ed to socio-economic inequali es, evidenced by the prevalence of illegal salinity intrusion and the misappropria on of public canals. It has therefore proven unsuccessful in ensuring equitable water management. In addi on, the model has proved ineffec ve, as engineering design remains top-down. Par cipa on is limited to consulta on while decision-making power remains in the hands of the implemen ng agency. The WMO model is also unsustainable, as its unrealis c cost-sharing requirements do not take into account the periodic maintenance challenges posed by silta on, river erosion, canal grabbing and illegal cuts/pipes in the embankment further contribu ng to deferred maintenance. Rather, millions of dollars are spent on each individual donor-funded project in order to create and sustain WMOs, yet these o en collapse within two years. In addi on, WMOs are disassociated from the local government structure and established channels for maintenance, and instead rely heavily on temporary project funding. In contrast, the UP is perceived as embedded in the local government ins tu onal structure, with access to rural employment schemes from the Upazila office along with NGOs and LGED. If donors and the Government of Bangladesh were to establish a permanent maintenance funds and allocate it through the Upazila Office, this would arguably be a more sustainable system to address the severe hydrological and socio-economic challenges facing the coastal zone of Bangladesh This, in turn, would lead to support for real and democra c decentraliza on, rather than for the limited effec veness of CBNRM in the water sector. Considering the prevalence of inadequate canal excava on independent of sub-contractor, a strengthening of rural employment work schemes should also be accompanied by the establishment of an independent quality assurance body that would approve payment a er sa sfactory comple on of maintenance work. 176 Acknowledgements This paper presents findings from G3 ‘Water Governance and Community-based Management in Coastal Bangladesh ’, a part of the Ganges Basin Development Challenge funded by the CGIAR Challenge Program on Water and Food. The authors wish to thank colleagues from the Interna onal Water Management Ins tute, in par cular Mark Giordano, Sanjiv De Silva, Floriane Clement, and Ravinder Malik for their feedback and support. They also wish to thank all their G3 project partners, including Shushilan for their qualita ve and quan ta ve data collec on throughout the southwest costal zone of Bangladesh. Last, a special thanks to the government officials, donors and project staff who gave access to valuable informa on on water management in Bangladesh. The original version of this paper was published in Water Alterna ves 7(2) and should be cited as: ‘Dewan, C.; Buisson, M.-C. and Mukherji, A. 2014. The imposi on of par cipa on? The case of par cipatory water management in coastal Bangladesh. Water Alterna ves 7(2): 342-366’. References ADB (Asian Development Bank). 2003. External evalua on. Small-scale water resources development sector project. Final Report. Dhaka, Bangladesh: ADB. ADB. 2007a. Project performance evalua on report. Small scale water resources development sector project (Loan 1381-BAN). Dhaka, Bangladesh: ADB (Opera ons and Evalua on Department). ADB. 2007b. Project performance evalua on report, Bangladesh: Khulna Jessore Drainage Rehabilita on Project. Dhaka, Bangladesh: ADB (Opera ons and Evalua on Department). ADB. 2008. Opera on and maintenance study: Final report. Incremental support to TA-7041-BAN: Par cipatory small-scale water resources project. Dhaka, Bangladesh: ADB. Agarwal, B. 2001. Par cipatory exclusions, community forestry, and gender: An analysis for South Asia and a conceptual framework. World Development 29(10): 1623-1648. Agrawal, A. and Gibson, C.C. 1999. Enchantment and disenchantment: The role of community in natural resource conserva on. World Development 27(4): 629-649. Araral, E. 2005. Bureaucra c incen ves, path dependence, and foreign aid: An empirical ins tu onal analysis of irriga on in the Philippines. Policy Sciences 38(2-3): 131-57. Arnstein, S.R. 1969. A ladder of ci zen par cipa on. Journal of the American Ins tute of Planners 35(4): 216-24. Arntzen, J.W.; Molokomme, D.L.; Terry, E.M.; Moleele, N.; Tshosa, O. and Mazambani, D. 2003. Main findings of the review of CBNRM in Botswana. Occasional Paper No. 14. Gaborone: IUCN/Netherlands Development Organisa on (SNV) CBNRM Support Programme. As-Saber, S. and Rabbi, M. 2009. Democra sa on of the Upazila Parishad and its impact on responsiveness and accountability: Myths versus reali es. Journal of Administra on and Governance 4(2): 53-71. Bandiaky, S. 2008. Gender inequality in Malidino biodiversity community-based reserve, Senegal: Poli cal par es and the village approach. Conserva on & Society 6(1): 62-73. BRAC (Bangladesh Rural Advancement Commi ee). 1983. The net: Power structure in ten villages. Dhaka: BRAC. BARD (Bangladesh Academy of Rural Development). 2009. Review of the socioeconomic impact assessment report on IPSWAM project in Southern and South-western zones of Bangladesh. Comilla, Bangladesh: BARD. Bardhan, P. 2000. Irriga on and coopera on: An empirical analysis of 48 irriga on communi es in South India. Economic Development and Cultural Change 48(4): 847-866. 177 BIDS (Bangladesh Ins tute of Development Studies). 2008. Impact evalua on study: Report of selected 10 sub-projects, SSWRDSP-I, Benefits monitoring and evalua on (BME) study, GoB, LGED, IWRM. Dhaka, Bangladesh: Local Government Engineering Development, Government of Bangladesh. Blaikie, P. 2006. Is small really beau ful? Community-based natural resource management in Malawi and Botswana. World Development 34(11): 1942-57. Boyce, J. 1987. Agrarian impasse in Bengal: Ins tu onal constraints to technological change. Oxford: Oxford University Press. Buisson, M.-C.; Mitra, A. and Kenia, N. 2013. Report from Households and Water Management Organiza ons Quan ta ve Surveys. Dhaka, Bangladesh: Challenge Program on Water and Food Chambers, R. 1983. Rural development: Pu ng the last first. London: Longman. Chambers, R. 1997. Whose reality counts? Pu ng the first last. London: Intermediate Technology Publica ons. Clement, F. 2010. Analysing decentralised natural resource governance: Proposi on for a 'poli cised' ins tu onal analysis and development framework. Policy Sciences 43(2): 129-56. Clement, F. 2012. Literature review: Gender and water management organisa ons in Bangladesh. Dhaka, Bangladesh: Challenge Program on Water and Food. Cooke, B. and Kothari, U. 2001. Par cipa on: The new tyranny? London and New York: Zed Books. Cornwall, A. and Gaventa, J. 2001. Bridging the gap: Ci zenship, par cipa on and accountability. PLA Notes 40(February): 32-5. Crook, R.C. and Manor, J. 1998. Democracy and decentraliza on in South Asia and West Africa: Par cipa on, accountability and performance. New York, US: Cambridge University Press. Crow, B. and Sultana, F. 2002. Gender, class and access to water: Three cases in a poor and crowded delta. Society and Natural Resources 15(8): 709-724. Custer, P. 1993. Bangladesh’s flood ac on plan: A cri que. Economic and Poli cal Weekly July (17-24): 1501-03. De Silva, S. 2012. G3 situa on analysis report: Polder 3, Satkhira, Bangladesh. Colombo, Sri Lanka: Interna onal Water Management Ins tute. Mimeo. Devas, N. and Grant, U. 2003. Local government decision-making ci zen par cipa on and local accountability: Some evidence from Kenya and Uganda. Public Administra on and Development 23: 307-16. Dewan, C. 2009. Piper calls the tune query: A case study of power-dynamics between Northern donors and Southern recipients in Bangladesh. MA disserta on, University of Edinburgh, Edinburgh, UK. Du a, A. 1997. Planning and management of water resources – Lessons from two decades of early implementa on projects, Bangladesh. Dhaka, Bangladesh: University Press Limited. Duyne, J. 1997. Local ini a ves and na onal projects in Bangladesh. In Du a, A.K. (Ed), Planning and management of water resources – Lessons from two decades of early implementa on projects, Bangladesh. Dhaka, Bangladesh: University Press Limited. Edwards, M. and Hulme, D. 1997. Beyond the magic bullet: NGO performance and accountability in the post-cold war world. West Har ord, Connec cut, US: Kumarian Press. EKN (Embassy of the Kingdom of Netherlands) and BWDB (Bangladesh Water Development Board). 2011. Integrated planning for sustainable water management (IPSWAM): Evalua on report. Dhaka, Bangladesh: EKN and Government of Bangladesh. 178 Faisal, L.M. and Kabir, M.R. 2005. An analysis of gender-water nexus in rural Bangladesh. Journal of Developing Socie es 21(2): 175-194. FAO (Food and Agriculture Organiza on of the United Na ons). 1985. Report on dal area study.: FISHERIES RESOURCES SURVEY SYSTEM, FAO/UNDP-BGD/79/015. Fariba, A. 2010. Contested waters, conflic ng livelihoods and water regimes in Bangladesh. Hague, The Netherlands: Interna onal Ins tute of Social Studies. Fox, W.F. and Menon, B. 2008. Decentraliza on in Bangladesh: Decentraliza on has been elusive. Interna onal Studies Program Working Paper Series 0829. Atlanta, Georgia, US: Interna onal Center for Public Policy, Andrew Young School of Policy Studies, Georgia State University. Fujita, Y. 2011. What makes the Bangladesh Local Government Engineering Department (LGED) so effec ve? Complementarity between LGED capacity and donor capacity development support. Tokyo: JICA Research Ins tute. Hanche , S. 1997. Par cipa on and policy development: The case of the Bangladesh Flood Ac on Plan. Development Policy Review 15(3): 277-95. Hardin, G. 1968. The tragedy of the commons. Science 162(3859): 1243-1248. Hashemi, S. 1996. NGO accountability in Bangladesh: Beneficiaries, donors and the state. In Hulme, D. and Edwards, M. (Eds), Beyond the magic bullet: NGO performance and accountability in the post-cold war world, pp. 123-31. West Har ord, Connec cut, US: Kumarian Press. Holloway, R. 1998. Suppor ng ci zens ini a ves: Bangladesh s NGOs and society. London, UK: Intermediate Technology Publica on. Hossain, M. A. 2004. Administra ve decentraliza on: A framework for discussion and its prac ces in Bangladesh. Rajshahi, Bangladesh: University of Rajshahi. Hulme, D. and Murphree, M. 2001. African wildlife and livelihoods: The promise and performance of community conserva on. Oxford, UK: James Currey. Ingham, B. and Kalam, A. 1992. Decentraliza on and development: Theory and evidence from Bangladesh. Public Administra on and Development 12(4): 375-85. IWMI (Interna onal Water Management Ins tute). 2013. Results from G3 quan ta ve survey. Project report. Colombo: IWMI (Mimeo). Khan, M. M. 2004. Corrup on: Concepts, compara ve experiences and Bangladesh case. A paper prepared for Transparency Interna onal – Bangladesh Chapter. h p://unpan1.un.org/intradoc/groups/public/documents/APCITY/UNPAN019105.pdf (accessed 13 April 2014) Kothari, U. 2001. Power, knowledge and social control in par cipatory development. In Cooke, B. and Kothari, U. 2001. Par cipa on: The new tyranny?, pp. 148. London and New York: Zed Books. Larson, A. 2003. Decentralisa on and forest management in La n America: Towards a working model. Public Administra on and Development 23(3): 211-226. Larson, A. 2008. Indigenous peoples, representa on and ci zenship in Guatemalan forestry. Conserva on and Society 6(1): 35-48. 179 Larson, A.M. and Ribot, J.C. 2007. The poverty of forestry policy: Double standards on an uneven playing field. Sustainability Science 2(2): 189-204. Leach, M.; Mearns, R. and Scoones, I. 1999. Environmental en tlements: Dynamics and ins tu ons in community-based natural resource management. World Development 27(2): 225-247. Lewis, D. and Hossain, A. 2008. A tale of three villages: Power, difference and locality in rural Bangladesh. Journal of South Asian Development 3(1): 33-51. LGED (Local Government Engineering Department). 2012. Incep on report: Support to Department of Coopera ves under par cipatory Small Scale Water Resources Sector Project. Dhaka, Bangladesh: LGED (Mimeo). LGED (Local Government Engineering Department). 2009. Opera on and maintenance strategy development for small scale water resources sub-projects: Small scale water resources support strategy and O&M manual. Dhaka, Bangladesh: LGED and Asian Development Bank. Mansuri, G. and Rao, V. 2003. Evalua ng community-based and community-driven development: A cri cal review of the evidence. Development Research Group. Washington, DC: The World Bank. Molle, F. 2008. Nirvana concepts, narra ves and policy models: Insights from the water sector. Water Alterna ves 1(1): 131-56. Mollinga, P. M. 1998. Editorial introduc on. In Mollinga, P.M.; Merrey, D.J.; Hvidt, M. and Radwan, L.S. 1998. Water control in Egypt s canal irriga on: A discussion of ins tu onal issues at different levels, interna onal ins tute for land reclama on and improvement. Wageningen, the Netherlands: Wageningen Agricultural University. Mollinga, P. 2008. Water, poli cs and development: Framing a poli cal sociology of water resources. Water Alterna ves 1(1): 7-23. MoP (Ministry of Planning). 2005. Evalua on study on small scale water resources development sector project at Upazila and Union level 1996-2002. Dhaka, Bangladesh: MoP, Government of Bangladesh. Mosse, D. 1999. Colonial and contemporary ideologies of 'community management': The case of tank irriga on development in South India. Modern Asian Studies 33(2): 303-38. Mosse, D. 2001. 'People's knowledge', par cipa on and patronage: Opera ons and representa ons in development. In Cooke, B. and Kothari, U. (Eds), Par cipa on – The new tyranny?, pp. 16-35. London: Zed Books. Mosse, D. 2004. Is good policy unimplementable? Reflec ons on the ethnography of aid policy and prac ce. Development and Change 35 (4): 639-71. Mosse, D. 2005. Cul va ng development: An ethnography of aid policy and prac ce. Anthropology, culture and society. London, UK: Pluto Press. MoWR (Ministry of Water Resources). 1999. Na onal Water Policy. Dhaka, Bangladesh: MoWR, Government of Bangladesh. MoWR. 2000. Bangladesh Water Development Board Act 2000. Dhaka, Bangladesh: Government of Bangladesh. MoWR. 2001. Guidelines for par cipatory water management. Dhaka, Bangladesh: Government of Bangladesh. MoWR. 2005. Technical report no 2: Report of the ins tu onal analysis study. Integrated Planning for Sustainable Water Management (IPSWAM). Dhaka, Bangladesh: Government of Bangladesh. 180 MoWR and EKN (Embassy of the Kingdom of the Netherlands). 2008. Integrated Planning for Sustainable Water Management (IPSWAM). Gender in IPSAWM. Dhaka, Bangladesh: Government of Bangladesh. Mukherji, A.; Fuleki, B.; Shah, T.; Suhardiman, D.; Giordano, M. and Weligamage, P. 2009. Irriga on reforms in Asia: A review of 108 cases of irriga on management transfer. Colombo, Sri Lanka: Interna onal Water Management Ins tute. Nowreen, S.M.; Khan, S.A. and Huq, H. 2011. Development of an opera onal IWRM framework in a selected small scale water resources development sector project in Bangladesh. South Asian Water Studies 2(2): 6-26. Ostrom, E. 1990. Governing the commons: The evolu on of ins tu ons for collec ve ac on. New York, Cambridge University Press. Poteete, A. and Ribot, J. 2011. Repertoires of domina on: Decentraliza on as process in Botswana and Senegal. World Development 39(3): 439-49. Palmer-Jones, R.; Arvidson, M and Mandal, S. 2010. Whatever happened to the water sellers? Dra paper presented at festschri for Geoff Wood, University of Bath, 13-14/9/2010. www.uea.ac.uk/~d280/other/bangladesh/water%20sellers%207%20Sept.pdf (accessed 4 May 2014) Rahman, A. 1981. Rural power structure: A study of local leaders in Bangladesh. Dhaka: Bangladesh Books Interna onal. Rahman, A.; Rahman, T. and Rahman, T. 2007. The role of organiza ons in the growth of the rural non-farm sector in Bangladesh: The case of LGED. University of Manchester Discussion Paper Series No. 11. Manchester, UK: Improving Ins tu ons Pro-Poor Growth (IPPG) Programme Office, Ins tute for Development Policy and Management, University of Manchester. Rahman, S. 2006. Development, democracy and the NGO sector: Theory and evidence from Bangladesh. Journal of Developing Socie es 22(4): 451-73. Rasul, G. and Chowdhury, A. 2010. Equity and social jus ce in water resource management in Bangladesh. Gatekeeper Series Interna onal Ins tute for Environment and Development (IIED) 146(July): 1-20. Ribot, J.; Agrawal, A. and Larson, A. M. 2006. Recentralizing while decentralizing: How na onal governments reappropriate forest resources. World Development 34(11): 1864-86. Ribot, J.; Chhatre, A. and Lankina, T. 2008. Introduc on: Ins tu onal choice and recogni on in the forma on and consolida on of local democracy. Conserva on and Society 6(1): 1-11. Shackleton, S. and Campbell, B. 2001. Devolu on in natural resources management: Ins tu onal arrangements and shi s – A synthesis of case studies for Southern Africa. Report to USAID SADC NRM Project no. 690-0251.12. 79. Mimeo. Shackleton, S.; Campbell, B.; Wollenberg, E. and Edmunds, D. 2002. Devolu on and community based management: Crea ng space for local people to par cipate and benefit? Natural Resources Perspec ves (ODI) 76: 1-6. Sogge, D. 2002. Give and take: What s the ma er with foreign aid? London; New York: Zed Books. Sultana, F. 2009. Community and par cipa on in water resources management: Gendering and naturing development debates from Bangladesh. Transac ons of the Ins tute of Bri sh Geographers 34(3): 346-63. Sultana, P. and Thompson, P. 1997. Effects of flood control and drainage on fisheries in Bangladesh and the design of mi ga ng measures. Regulated Rivers: Research and Management 13(1): 43-55. 181 Summers, L. 2001. Speech at World Bank Country Director s Retreat – 2001, June 2001. Thörlind, R. 2000. Development, decentraliza on and democracy: Exploring social capital and poli ciza on in the Bengal region. Copenhagen, Denmark: Nordic Ins tute of Asian Studies. Toufique, K.A. and Turton, C. 2002. Hands not land: How livelihoods are changing in rural Bangladesh. Dhaka, Bangladesh: Bangladesh Ins tute of Development Studies. Wester, P. and Bron, J. (n.d.). Coping with water: Water management in flood control and drainage systems in Bangladesh. h p://edepot.wur.nl/78392 (accessed 24 April, 2014) Williams, G. 2004. Evalua ng par cipatory development: Tyranny, power and (re)poli cisa on. Third World Quarterly 25(3): 557-578. Wi ayapak, C. and Vandergeest, P. 2009. The poli cs of decentraliza on: Natural resource management in Asia. Chiang Mai, China: Mekong Press. Wood, G.D. 1994. Bangladesh: Whose ideas, whose interests? Dhaka, Bangladesh: University Press Limited. World Bank. 1988. World development report. New York, US: Oxford University Press. 182 Predic ng success in community-driven water infrastructure maintenance: Evidence from public goods games in coastal Bangladesh A. Das1, M.-C. Buisson2 and A. Mukherji 3 1 Center for Public Affairs and Cri cal Theory, Shiv Nadar University, India, arijitdas22@gmail.com 2 Interna onal Water Management Ins tute, India, m.buisson@cgiar.org 3 Interna onal Centre for Integrated Mountain Development, Nepal, amukherji@icimod.org Abstract The policy of the Government of Bangladesh requires local communi es to organize themselves into Water Management Organiza ons (WMOs) and contribute to ‘minor’ maintenance of water infrastructure. For this purpose, WMOs are required to collect contribu on from their members. This paper is an in-depth analysis of the condi ons under which these contribu ons are collected from the communi es to serve the sustainability of the water infrastructure. This paper uses an experimental game, a public good game, replica ng on-the-ground reali es of coastal areas of Bangladesh, which was played in 14 different places across coastal Bangladesh to understand group coopera ve behavior towards infrastructure maintenance. Here we analyze the determinants of the success or failure of the game regarding group contribu on to maintenance and further simulate the results for predic ng success under different scenarios. The results highlight the role of ins tu ons in building coopera ve behavior toward sustainable management of the infrastructure as well as propor onal sharing of the benefits. These results argue for a stronger involvement of local stakeholders at a smaller scale while transferring water management responsibili es to communi es, and for strengthening group incen ves for community contribu ons. Keywords: Experimental games, water management, contribu on, simula on 1. Introduc on The en re coastal belt of Bangladesh is one of the most ac ve deltas of the world. Every day millions of ton of silt floats through the Ganges and Brahmaputra rivers to this newly formed land, con nuously forming and destroying it. Other features including daily dal surges and occurrence of seasonal cyclones and floods make for a harsh and uncertain life in the coastal area. During the 1960s a major geophysical interven on took place in the coastal areas of Bangladesh. The Dutch government along with Bangladesh Water Development Board (BWDB) started building embankments along the riversides. These embankments (called polders) were to serve a dual purpose: first, to protect the coastal communi es from daily dal surges, flooding, and natural calami es, and; second, to increase agricultural produc vity in the coastal regions. Over the years polder infrastructure, with its embankments, canals and sluice gates, became a key source of water for mul ple uses like irriga on, aquaculture, domes c use and drainage. The embankments also became useful for transporta on. For forty years the embankments served their ini al purpose and the regular minor maintenance of the water infrastructures was undertaken by central agencies (BWDB). The second round of interven on started in the year 1990, when the Local Government Engineering Department (LGED) built embankments in smaller areas that were earlier le out from the main polders. Developed by a different agency, the LGED sub-projects have a different governing structure than that of the polders. Under the current water policy, polders of 1000 hectares and more are supervised by BWDB and managed jointly by BWDB and community-led water management organiza ons (WMOs). WMOs are comprised of all water users in the area. Those of less than 1000 hectares (called sub-projects to dis nguish them from polders) are under the jurisdic on of local governments, built by LGED and jointly managed by LGED, WMOs and local governments. Policies regarding opera on and maintenance of polders have changed over me (see 183 Dewan et al. 2014; Du a 1999). The Government of Bangladesh’s current policy framework asks local communi es to organize themselves into WMOs and to contribute towards ‘minor’ maintenance of water infrastructure. This is embodied in the Na onal Water Policy of 1999 (MoWR 1999) and in the Guidelines for Par cipatory Water Management (MoWR 2001). For this purpose, WMOs are required to create a maintenance fund and collect contribu on from their members. Interes ngly, not all WMOs maintain such funds; some WMOs keep some funds that are some mes used for emergencies. Similarly, a series of focus group discussion revealed that some WMOs/Water Management Coopera ve Associa ons are successfully repairing their water infrastructures while others are not. The purpose of this paper is to understand the factors that may explain varia ons in success of ‘minor’ maintenance in water infrastructure in coastal Bangladesh. We use experimental games to understand success in mul ple simulated environments. Our par cipants are drawn from coastal communi es – more specifically from BWDB polders and LGED sub-projects. We consequently try to understand how ins tu onal mechanisms are important for water infrastructure management. This study further tries to simulate the results in real life condi ons and predict success in simulated condi ons. Based on the predic ons, our objec ve is to prescribe policies for ‘minor’ water infrastructure management in coastal areas. To answer these research objec ves an arte-factual public goods game is used as the analy cal framework in this study. Most common-pool resource experiments are inspired by the groundbreaking work of Ostrom, Gardener and Walker (1994). A pure public good is characterized by non-excludability and non-rivalness in consump on. The current maintenance problem in the polder areas of Bangladesh replicates similar common pool problems. This paper consequently uses the typical framework of a public goods game, which essen ally replicates on-the-ground condi ons in coastal areas of Bangladesh. The principal of the WMO maintenance fund is contribu ons, which are dependent on the agent’s choice. No ins tu onal pressure or direc ve is imposed on the agents to contribute to the common fund. This situa on is therefore exactly similar to a public goods game. 2. Literature Neo-classical literature posits that since these goods are non-rival and non-excludable in consump on, any a empts to elicit a user’s willingness to voluntarily pay for such goods will inevitably fail. This is known as a free riding problem. To test this neoclassical free riding problem, experimental work using voluntary contribu on mechanism games has been designed to replicate public goods situa ons. These studies have revealed that a large frac on of experimental cohorts do not play ra onal Pareto dominated equilibrium (i.e. zero contribu on for public goods provisions) (Baland et al. 1996). Indeed, this paradox of ra onality has led to a large number of experimental studies in both economics and psychology that have explored this devia on from neo-classical free rider assump ons (Andreoni et al. 1993). Similar devia on from free-riding principle is also observed in real-life examples (Baviskar 1994; North 1990; Ostrom 1992), which in the process avoids occurrence of the ‘tragedy of the commons’ (Hardin 1968). Numerous experiments have made significant contribu ons to studying the behavioral par culari es of the human psychology that may affect environmental governance. Fehr et al. (2000) show that costly punishment can deter free-riding and maintain coopera on among agents, but its excessive use lowers the total surplus. Studies also indicate that communica on among par cipants improves coopera on (Janssen et al. 2013; Shah 1996). Extensive research has been done in the direc on of economic valua on and discoun ng of future benefits and costs with alterna ve behavioral paradigm (see Kahnemann et al. 1990; Horowitz et al. 2002; and Shogren 2004 for a survey). Another branch of experimental work deals with the study of how incen ves and ins tu ons affect decisions and outcomes. Experimentalists have used experimental designs to study how incen ves can influence water management systems (Cummings et al. 2003; Ward et al. 2006) and network markets for resource alloca on (e.g. energy network markets, Denton et al. 2001). A third group of experimental economists work on the typical problem of externali es or social dilemmas associated with 184 common-pool resources and public goods. Ledyard’s survey (1995) of voluntary contribu on mechanism games revealed the behavioral and ins tu onal founda ons of why would people, despite the clear incen ves to free ride, engage in coopera ve behavior and refrain from over-exploi ng a common-pool resource or contribute voluntarily to the provisioning of a public good. Field experiments in this area include Velez et al. (2008), Cardenas et al. (2000, 2004) and Rodrigues et al. (2008). Numerous experiments have been conducted to understand not only the contribu on to public goods, but also how coopera on can be increased by punishments. Fehr et al. (2000) carried out seminal work in this direc on. The work of Fehr et al. (2002) and others shows that costly punishment can deter free riding and maintain coopera on among agents. Though costly punishments are effec ve, they are seen as socially wasteful, thus their excessive use lowers the total surplus. Communica on and informal punishment are both seen to have an impact on augmen ng contribu on behavior. Pre-play communica on between subjects is one such tool for maintaining coopera on. Issac et al. (1988) found that communica on fosters coopera on. Ostrom et al. (1992) corroborated this finding in a common pool extrac on game. Various types of pre-play communica ons have been explored in the literature (Isaac and Walker 1988; Bochet et al. 2009). Botchet et al. (2006) found that communica on, both face-to-face as well as anonymously through a chat room, allowed subjects to cooperate efficiently. Important theories consider social preference as a root cause for the pro-social behavior in public goods game. They posit that social preference is a func on of social capital. Social capital is seen as a medium to reduce transac on costs of monitoring contracts, and nego a ng and enforcing contracts. Dasgupta (2008) argues that trust is the key to coopera on and what scholars meant by social capital is just one of the ways to channel trust. He further argues that social capital means some kind of social network and social network developed on the basis of caste, religion or race. Bouma et al. (2008) use a trust game (an NPD: N-Player Prisoner’s Dilemma) experiment combined with a household survey in rural India to explore the linkages between social capital, community characteris cs, and the provision of a semi-public good: investment in soil and water conserva on maintenance. They find that coopera on in the trust game is posi vely correlated with community par cipa on in the provision of the public goods and social homogeneity. It is found from the earlier discussion that game design and socio-cultural backgrounds influence the results of the games. Moreover, ins tu onal setup some mes shapes players preference of coopera on and contribu on or division of wealth. Henrich et al. (2004) compiled a large number of field experiments to argue that people’s decisions are not uniform across con nents or socie es. The outcomes of simple Dictator Games, Ul matum Games and Public Goods Games vary widely across socie es. Travers et al. (2011) find ins tu onal set up to be significant for resource extrac on from common property in Cambodia. Mitra and Gupta (2009) also observed the effects of ins tu on on human decision in the Indian context. They ran a non-computerized, single-period voluntary contribu on mechanism game. The subjects are from different social backgrounds. The first group consists of undergraduate students from colleges. The subjects in the second and third groups have a similar socio-economic background and are members of Community Based Organiza ons (CBO). But the players in the second group belong to the same CBO, while players in the third group are pooled from different CBOs. They found that the members in the second group made the highest contribu on to public goods in all the treatments, with an average contribu on rate of above 85% of endowment. They argue that weak bonds between the members of different CBOs could be the reason for lower contribu on in the third group. According to the authors, the primary cause for lower contribu on in the first group could be a ributed to the player’s urban backgrounds and exposure to western culture, which promotes individualis c behavior. In a different example for Thye et al. (2009), it is observed that subjects coming from collec vist socie es behave more pro-socially compared to those who come from individualist socie es. These results clearly indicate how being in a specific ins tu onal structure can change human behavior. Few studies show how informa on about social dona on pa ern affects individual contribu on. Croson and Shang (2008) used a voluntary contribu on mechanism game to explore the effect of social influence on contribu on behavior. Frey and Meier (2004) studied the behavior of students from the University of Zurich 185 with respect to contribu on to two separate funds. They found that students are willing to contribute more, if others contribute more in accordance to condi onal coopera on theory. Heterogeneity in wealth and preference has shown an ambiguous effect on contribu on. Both posi ve and nega ve effect of heterogeneity has been registered in the literature. Ostrom (1992) found the importance of homogeneity of social and individual capital for people to cra successful ins tu ons and Oslon (1993) suggested that homogeneity in preference and income may “increase welfare and reduce conflict”. Other researchers (Marwell et al. 1988; Heckathorn 1993) described scenarios in which the impact of wealth heterogeneity in preferences and incomes may lead to increased collec ve ac on. Some studies found more coopera ve behavior in same-sex groups than in mixed-sex groups (Nowell and Tinkler 1994; Greig et al. 2009). Others obtained the opposite results with women being more coopera ve in mixed-sex than in same-sex groups (Sell 1997; Carpenter et al. 2004). Several other studies however provide no evidence of sex or sex composi on differences in the voluntary provision of public goods (Sell et al. 1993). In a nutshell, game design as well as socio-economic background and par cularly ins tu onal background (Travers et al. 2011) of the subjects can have a strong impact on the outcome of the experiment. Factors such as ethnic group, gender, and the societal norms that the subjects bring with them can significantly alter their decisions. The game and the subsequent analysis then build on the learning from the literature. 3. Game Design An arte-factual field experiment (herea er referred to as field experiment) gave us an opportunity to study decisions in the field, in controlled situa ons, involving subjects who would be difficult to get to a laboratory in an urban se ng (see Binswanger 1981 for an early field experiment; Harrison et al. 2004 and and Chakravarty et al.2011 for comprehensive reviews). Contribu on dilemma in coastal Bangladesh could be replicated through a public goods game environment. A public goods game creates an environment where players choose to contribute to a common pool. A similar voluntary contribu on mechanism is followed in coastal areas of Bangladesh for minor maintenance works. This paper uses a par cular variant of a public goods game and calibrated treatments according to the ground condi ons. 3.1. Selec on of game venues and players Our public good games were played 18 mes in 14 different villages in the coastal zone of Bangladesh. Of these villages, five were located in LGED sub-projects and 9 were located in BWDB polders. In each village, we designated households into different wealth strata: poor, rich and intermediate. Five players were then randomly selected from these three strata. Stra fied selec on ensured a subject pool of diverse economic backgrounds. The players are users and beneficiaries of the same water infrastructure. This also means that they are people who may have to make decisions together in real life. 3.2. Rules of the game The game is a public goods game and the design follows Issac et al. (1984), however, changes have been incorporated to sa sfy local context and research focus. One control and four treatment sessions are played. The treatment rules are designed to capture three separate effects: publicly available informa on, unequal endowment, and differences in redistribu on of benefits. Each set of rules (or sessions) was played for five rounds of replica on. Rules were explained to the players by reading out a wri en text before the start of every new treatment. Players were instructed not to talk to each other during the rounds. The first session was the control session. Each player received 20 tokens and had to decide how much to contribute to the common maintenance fund. Contribu ons were made in secret so that other players did not know who contributed how much, but the total collec on in the common fund is declared a er every round. The threshold of total contribu on to the common fund is fixed at 50 tokens. Total contribu on of less than 186 50 tokens was forfeited, but contribu ons above 50 tokens were rewarded by an addi onal contribu on of 25 tokens or 75 tokens depending on toss of a coin. This amount represents government contribu on and mimics the real life situa on when the government tops up community contribu on for repair and maintenance work. The common fund so created (minimum 50 tokens, plus either 25 or 75 addi onal) was then equally re-distributed among the five players. At the end of the round, each player was informed of individual earnings. The earnings of the players were the sum of tokens kept with them plus the gains received from the sharing of the common fund if any. In these five rounds both the individual contribu on to the common fund and the individual earnings were kept secret from the other players. Thus, a player’s payoff at the end of each round was: The first treatment session was similar to the control session in that every player started with 20 tokens; the only difference was that informa on on individual contribu on to the common fund was made public. The second treatment remained iden cal to first treatment, except that the common fund was redistributed in the same propor on as the contribu ons. In the third treatment, the ini al endowment (or tokens) received by the players became unequal; two players were considered “rich” and received 35 tokens, whereas three “poor” players received 10 tokens each. Poor and rich players were decided by drawing of a lo ery at the beginning of each round. The threshold level of success remained the same (50 tokens) and the collec ve fund was equally divided among the players, irrespec ve of their contribu on to the common fund – akin to Treatment 1. The fourth treatment incorporated inequali es in endowment (similar to Treatment 3) and redistributed common funds in the same propor on as contribu on to the common fund (similar to Treatment 2). Put in real life context, this is somewhat akin to a situa on when farmers with more landholding contribute more to maintenance funds and receive more benefits from upkeep of the water infrastructure. The rounds in which the players were able to reach the threshold of 50 tokens were considered to be successful. However, when players failed to accumulate half of their endowment in the public pool we consider the round a failure. The success or failure of each round will here be considered as our variable of interest. We also collected demographic, socio-economic and percep on related data from each player at the end of the game using a structured ques onnaire. In addi on, we generated discussion among par cipants about their behavior at the end and kept notes of those discussions to inform our results qualita vely. 4. Descrip ve sta s cs Here we describe the average success rate in 25 rounds that are divided into five sessions. We can observe the changes in success due to changes in the rules of the game. Informa on introduced in Treatment 1 had a nega ve effect on success and, moreover, success declined over rounds. A sudden jump in success can be seen in Treatment 2, where benefit progressed with contribu on. This clearly indicates that incen ve has a dominant effect on contribu on. Indeed both Treatment 2 and Treatment 4 show exactly the same pa ern. The commonality between these two rounds is propor onal division of benefits. Irrespec ve of ini al endowment, the success rate rises as the game proceeds during these two rounds, confirming that incen ve is a primary driving force both in homogeneous and heterogeneous setups. 187 Control 0 Treatment 1 5 Treatment 2 10 Rounds Treatment 3 15 Treatment 4 20 25 Fig. 1. Percentage of success across the rounds. 5. Econometric modeling 5.1 Empirical model We derived a simple logis c model to explain the varia on in ‘success’. Based on the exis ng findings and literature review we assume that success in each round depends on specific game design as well as external group characteris cs. The basic model is as follows: Xij is a binary variable with two values 0 and 1 where 0 and 1 indicates failure and success respec vely. i stands for rounds and i = 1, 2, 3, 4……, 25. j stands for group and j = 1, 2, 3, …,18. G is the vector of game variables and varies over the rounds. Game-specific characteris cs like informa on, unequal endowment and propor onal division are included in this category. U stands for the group’s characteris cs. This vector includes group characteris cs like average educa on level of the group, average age of the group, gender and religious heterogeneity of the group and the number of WMO and other local body members within the group. We also tried to capture the learning that occurred during the gaming process. The learning effects are represented by L in the equa on and try to capture how the groups coordinate among themselves through their ini al contribu on pa ern in the common pool. This learning vector also captures how the last two rounds explain the success in the current round. Most importantly we have incorporated a binary variable, I, which represents the specific ins tu onal environment in order to capture the effect of ins tu on on coopera on and coordina on. The descrip ve sta s cs of the variables used in the regression are given in Table 1. There is a comfortable level of varia on in the variables making them significant to explain varia on in success. Coming back to the regression model, the varia on in success is es mated by five dis nct models. The first and second models are es mated through logis c regression with and without clustering, respec vely. The third and fourth models are akin to model one and two; the only difference is that we dropped the learning variables from the model. Model 5 is more interes ng and dis nct from the other models. This model is focused on the treatment-specific effect on success, controlling other covariates whereas the other four models looked into effects of informa on, wealth heterogeneity and unequal sharing on success keeping all other covariates the same. The results are presented in Table 2. 188 Finally, robustness checks were conducted. Results from a test for specifica on of the errors in the model highlight that there are no specifica on errors. We also ran goodness-of-fit tests for all five models and found that model two is the one explaining the highest varia on in success. The ROC curve displayed in Appendix 1 confirms that the second model has the maximum explanatory power. Table 1. Descrip ve sta s cs of the variables Obs. Mean Std. Dev. Min Max 1 if collected sum ≥50, otherwise 0 450 0.68 0.47 0 1 Ins tu on 1 if loca on is at LGED area, otherwise 0 450 0.33 0.47 0 1 Informa on 1 if informa on about contribu on and earning is publicly available 450 0.80 0.40 0 1 1 if ini al endowment is not equal, 0 if ini al endowment is equal (i.e. 20 BDT) of 1 if division of common pool is propor onal to individual contribu on, 0 otherwise 450 0.40 0.49 0 1 450 0.40 0.49 0 1 Variable Defini on Success Unequal endowment Propor onal common pool distribu on Average age of the group Average age of the 5 players 450 39.88 5.59 26 47 Varia on in Age Standard devia on of age of 5 players 450 10 3.48 5 16 Avg. educa on level of the group Average educa on level of the group 450 8 1.97 4 11 Varia on in Educa on Standard devia on of educa on level of 5 players 450 2.72 1.28 1 5 Sex homogeneity 1 if all the players are from same sex, 0 otherwise 450 0.33 0.47 0 1 Religion homogeneity 1 if all the players are from same religion, 0 otherwise No of players engaged is aquaculture or agriculture 450 0.61 0.49 0 1 450 4.06 1.08 2 5 450 0.28 0.45 0 1 450 0.28 0.45 0 1 450 11.45 3.38 2.32 16.16 450 3.95 2.22 1.40 8.75 432 0.68 0.47 0 1 414 0.61 0.49 0 1 Occupa on Majority WMCA/WMO members 1 if more than 3 players are member of WMCA/WMO, 0 otherwise Majority member of other local 1 if more than 3 players are member of bodies other local bodies, 0 otherwise Avg. contribu on in first 5 rounds Average contribu on by the five players in first five rounds S.D. of contribu on in first 5 Standard devia on of five players of the rounds average contribu on in first five rounds Success in previous round 1 if previous round is a successful round i.e. sum of contribu on is ≥50 Success in second last previous 1 if previous two rounds are successful , 0 round otherwise 5.2 Econometric results 5.2.1 Ins tu ons We will start explaining varia on in success with the covariate ‘ins tu on’. This is a characteris c that players bring with them to the game (Dasgupta 2008; Mitra and Gupta 2009). Through this binary variable we capture how exposure to par cular governance setup can influence coordina on as well as success. Ins tu on is significant in all five models. The coefficient of ins tu on is posi ve and significant. Taking all variables at their mean, the probability of success is 40%when the game is played at a LGED sub-project, meaning that LGED polder areas have 40% higher probability of success than BWDB polders. The result shows posi ve and significant influence of ins tu on on success. It is an indica on that certain governance structures can help some communi es to coordinate more effec vely than others. A related policy recommenda on could be to improve water governance by replica ng the LGED governance module at the community scale in larger polders. 189 5.2.2 General rule variables Treatment variables show very consistent results in four models. This study shows that informa on has a nega ve and significant effect on success. The coefficients of informa on dummy are significant and nega ve in the four models. This result confirms that disclosing contribu on amount to other stakeholders of water management could result in backlash. The reason could be mismatch of expected contribu on on the part of other players results into lower contribu on and coordina on in later rounds. This fact is corroborated by declining success in later rounds in the Treatment 1 session in model 5. Heterogeneity in ini al endowment is not significant in all the models. This result indicates that in-built heterogeneity does not have any influence on success (Oliver et al. 1950; Marwell et al. 1988; Heckathorn 1993). In a real life context this means no significant difference is expected in homogeneous villages and heterogeneous villages. Propor onal distribu on of the common pool has a posi ve and significant influence on success rate. The coefficients are posi ve and significant in all the models, which confirms the results obtained by Buisson et al. (2013). On the ground, this result indicates that locals would able to manage ‘minor’ maintenance more successfully if repair work was linked with water use or canal use. 5.2.3 Group characteris cs We assume that the players carry some inherent characteris cs while making decisions. This sec on describes the effects of such characteris cs. Average age of the group as well as varia on in the age among the players is not important for achieving success. Educa on level seems important in achieving success. Average educa on level of the group is significant for model 2, which was the best fi ed model among others, though varia on in educa on level does not have any effect. The coefficient for sex homogeneity is posi ve and significant for all models. Hence, gender-wise homogeneous groups are more likely to achieve success than mixed sex groups. This result aligns with other findings where researchers found lower contribu ons from mixed sex groups. Social capital also has interes ng effects. If social capital is defined as par cipa on in the local community organiza on then we see a very interes ng result. The fact that a majority of players are associated with WMOs does not have any influence on success, but a greater associa on with other local organiza ons significantly increases the success rate. This indicates that socializa on and social capital does have a significant influence on coordina on, even if it is not specifically related to water management. Interes ngly, the occupa on variable comes out as nega ve and significant. As more and more players are associated with agriculture and aquaculture the success rate goes down. This is quite surprising and contrary to the findings of Buisson et al. (2013) who considered individual occupa ons and obtained a posi ve effect on contribu ons for being farmers. In that case, it might be that even if individually the agriculturist or aqua-culturist is contribu ng more to the ‘minor’ maintenance fund, when it comes to a group strategy, players prac cing agriculture and aquaculture play more strategically, preven ng them from reaching the required threshold. Hence, successful water management policy would promote par cipa on of different stakeholders with different interests. 5.2.4 Learning effect The same group played the game for 25 consecu ve rounds. Hence, a lot of learning took place at the player and group levels. This analysis focuses at the group level and we hence concentrate more on group level learning. Controlling such learning is always challenging. In this regression we considered mul ple levels of learning: learning from ini al contribu on behavior, learning from immediate past experiences and learning over me. On the first case we use average contribu on in the first five rounds to determine the coopera ve nature of the group. The average of the first five rounds is considered as it is least affected by game design and leaning bias. The coefficient of average contribu on is 0.898 and significant, which means that higher contribu on in the ini al rounds predicts higher probability of success. Interes ngly, varia on in the contribu on is also 190 posi vely related to success. Success in immediate previous rounds shows significant and posi ve influence on next round’s success. If the group experiences success in one round probability of success increases in the next round. But if the group is experiencing success for two consecu ve rounds then that does not have any significant influence over the third round. Finally round learning variable, designed to capture linear round-wise learning, is insignificant. Due to regular changes in the rules of the game, no specific round-wise linear learning happened in this game. Table 2. Regression results VARIABLES Institution Information Unequal endowment Proportional distribution of common pool (1) Success 1.169 (1.142) -2.388*** (0.873) -1.157 (1.363) 1.516* (0.799) (2) Success 1.169* (0.636) -2.388*** (0.790) -1.157 (1.220) 1.516* (0.874) (3) Success 1.603*** (0.489) -1.602*** (0.473) 0.242 (0.301) 1.700*** (0.311) (4) Success 1.603* (0.867) -1.602*** (0.567) 0.242 (0.299) 1.700*** (0.445) Treatment 1 0.038 (0.129) 0.168* (0.0986) 1.092 (0.811) -0.0622 (0.679) 2.688** (1.299) -1.077 (1.644) -1.442*** (0.548) 1.555 (1.022) 6.517*** (1.523) 0.898*** (0.128) 0.659** (0.331) 1.710*** (0.625) -0.046 (0.645) 0.141 (0.134) -21.48* (12.91) 0.038 (0.099) 0.168 (0.114) 1.092** (0.551) -0.062 (0.644) 2.688*** (0.903) -1.077 (0.890) -1.442*** (0.340) 1.555*** (0.577) 6.517*** (1.251) 0.898*** (0.150) 0.659*** (0.216) 1.710*** (0.430) -0.046 (0.469) 0.141 (0.128) -21.48** (9.090) -0.007 (0.067) -0.0655 (0.0443) 0.528 (0.370) 0.923*** (0.337) 2.250*** (0.621) 1.737*** (0.663) -0.968*** (0.207) 2.513*** (0.492) 5.919*** (0.674) -0.007 (0.133) -0.0655 (0.119) 0.528 (0.559) 0.923* (0.533) 2.250* (1.257) 1.737 (1.394) -0.968*** (0.353) 2.513** (1.015) 5.919*** (1.283) -4.427 (6.328) -4.427 (10.79) -2.510*** (0.871) -0.528 (1.322) -3.375** (1.708) -2.291 (2.496) 0.044 (0.099) 0.166 (0.115) 1.129** (0.551) -0.044 (0.647) 2.771*** (0.919) -1.174 (0.882) -1.460*** (0.353) 0.903*** (0.153) 0.669*** (0.218) 1.593*** (0.594) 6.552*** (1.278) 1.713*** (0.445) -0.087 (0.471) 0.139 (0.127) -22.04** (9.130) 414 414 450 450 414 Treatment 2 Treatment 3 Treatment 4 Average age of the group Variation in Age Avg. education level of the group Variation in Education Gender homogeneity Religion heterogeneity Occupation Majority WMCA/WMO members Majority member of Other local bodies Avg. contribution in first 5 rounds S.D. of contribution in first 5 rounds Success in previous round Success in second last previous round Round learning Constant Observations (5) Success 1.201* (0.654) Note: Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 191 6. Predic ons and simula ons This par cular sec on is very important with respect to the focus of this study as well for policy formula on. This sec on will simulate the best-fi ed model (in this case model 2) and replicate real life condi ons into the model results to predict probability of success. Such an exercise enables us to simulate probable scenarios that could help policy makers get a hands-on policy prescrip on. These simula ons are based on different assump ons from which the different scenarios in terms of failure or success for community coopera on toward contribu ons for water infrastructure maintenance are derived. Specifically, the probability of failure is predicted under different assump ons. The simula ons are repeated 10,000 mes in order to check the consistency of the coefficients and the robustness of the results. 6.1. Observed effect of ins tu ons 0 0 .2 .2 .4 .4 .6 .8 .8 Community water management in coastal Bangladesh is under two types of governance structures: Water Management Groups created and supported by donors and projects in BWDB-led polders and Water Management Coopera ve Associa ons established by LGED programs in sub-projects. Stark differences are visible between these two bodies in their forma on, involvement with locals, fundraising ac vi es and outreach (Dewan 2012). Previous regression analysis also indicates that the differences between these two ins tu onal structures has significant effect on the probability of success. The simula on again confirms these results (Figure 2). The figures indicate that in a large number of events LGED water management bodies would have more prolonged success than BWDB polders. The probability of failure is higher under the BWDB ins tu onal framework and increased much faster while progressing over the number of replica ons than under the LGED assump on. The LGED-type of community level ins tu ons are consequently more likely to support effec ve community contribu ons for maintenance and sustained infrastructure. 0 20 40 100*P 60 80 BWDB 100 0 20 40 100*P 60 80 100 LGED Fig. 2. Simula on with ins tu onal varia on. 6.3 Observe effect of propor onal sharing The second important finding of the regression analysis is the posi ve and significant impact of propor onal distribu on of benefits on the probability of group success. We simulated two scenarios: one with propor onal sharing of benefits and one with equal sharing of benefits (Figure 3). Figure 3 indicates that the probability of failure is minimal for a large number of replica ons under the scenario with propor onal sharing of benefits. On the contrary, when benefits are equally divided among the stakeholders the probability of failure is much higher and increases more rapidly over the replica ons. Efforts made by the community members for the maintenance of water infrastructure are consistent with access to benefits. This should be used as a tool to improve contribu ons from different stakeholders. 192 .8 .8 .6 .6 .2 .6 .4 .2 0 0 0 20 40 100*P 60 80 100 Equal sharing of benefits 0 20 40 100*P 60 80 100 Propor onal sharing of benefits Fig. 3. Simula on with equal or propor onal sharing rules. 7. Conclusion Based on an experimental game played with coastal communi es, this analysis aimed to understand how such communi es can undertake coopera ve behavior to keep their water infrastructures, here a public good, well kept and regularly maintained with contribu ons from community members. Using regression analysis and simula ons we established that in this se ng, success in collec ve ac on varied across ins tu ons and with the rules of the game. The simula on exercise shows that propor onal sharing of benefits is a strong tool for posi vely influencing the coopera ve behavior of a group and for prolonging success. This result is consistent with Buisson et al. (2013), who demonstrated a posi ve and significant effect of propor onal sharing of benefits on individual contribu ons toward maintenance of water infrastructure. The second most important result relates to ins tu ons; the probability of success and coopera on is higher in communi es within LGED projects and where Water Management Coopera ve Associa ons have been created than in BWDB polders where Water Management Groups exist. To understand the scope of this result, the differences between these two ins tu onal se ngs have to be understood. The first difference is the scale. In LGED sub-projects, water management organiza ons are organized at a small scale; with less than 1000 hectares in a sub-project, usually only one village is included. In contrast, in BWDB-administered communi es, the resources of several villages were commonly considered together. The second difference is more subjec ve and highlights the par cipatory processes followed during the crea on of these community-led water management organiza ons (Dewan et al. 2014). From these results, the policy recommenda ons that emerge highlight the need to reconcile the contribu ons from the communi es and the benefits they are granted. Due to the deferred maintenance cycle, too o en infrastructure is so damaged that even if communi es contribute, they will hardly see any benefit. If the Government of Bangladesh and donors were able to cover major maintenance costs as men oned in the policy, then community members would have an incen ve to contribute as they would ul mately be able to enjoy the benefits of the water infrastructure. Secondly, the results in rela on to ins tu ons underline the need to mobilize communi es at a small scale and to strengthen the ownership they may have over water infrastructure through a long term par cipatory process rather than through technical rehabilita on works. Acknowledgements This ar cle is based on data collected for the project ‘G3: Water Governance and Community-based Management in Coastal Bangladesh’, part of the Ganges Basin Development Challenge of the CGIAR Challenge Program on Water and Food and the CGIAR Research Program on Water, Land and Ecosystems (WLE). The authors wish to thank the players involved in these games as well as Dr. Anannya Dasgupta and anonymous referees for their comments on the manuscript. 193 References Andreoni J. and J. H. Miller. 1993. Ra onal coopera on in the finitely repeated prisoner's dilemma: Experimental evidence. The Economic Journal 103: 570-585. Asian Developement Bank. 2008. External Evalua on. Small-Scale Water Resources Development Sector Project – I. Final Report. Asian Development Bank, Dhaka. Baland J. M.and J. P. Pla eau. 1996. Hal ng degrada on of natural resources: is there a role for rural communi es?. Oxford: FAO and Clarendon Press. Bangladesh Ins tute of Development Studies. 2008. Impact Evalua on Study Report, Selected 10 Subprojects, SSWRDSP-I. Baviskar A. 1994. Fate of the forest: Conserva on and tribal rights. Economic and poli cal weekly 29(38): 2493-2501. Binswanger, Hans P. 1981. A tude Toward Risk, Theore cal Implica ons of an Experiment in Rural India. Economic Journal 91: 867-90. Bochet, O., T. Page, and L. Pu erman. 2006. Communica on and punishment in voluntary contribu on experiments. Journal of Economic Behavior & Organiza on, 60(1): 11-26. Bochet O. and Pu erman, L. 2009. Not just babble: Opening the black box of communica on in a voluntary contribu on experiment. European Economic Review 53:309-326. Bouma J., E. Bulte, and D. van Soest, 2008. Trust and coopera on: Social capital and community resource management. Journal of Environmental Economics and Management 56: 155-166. Buisson, M-C., Das, A., and Mukherji, A. 2013. What determines contribu on to a common fund for upkeep of water infrastructures? Evidence from experimental game in Coastal Bangladesh. Poster. Colombo, Sri Lanka: CPWF. Cardenas, J. C., and Ostrom, E. 2004. What do people bring into the game? Experiments in the field about coopera on in the commons. Agricultural System 82(3): 307-326. Cardenas, J. C., Stranlund, J., and Willis, C. 2000. Local environmental control and ins tu onal crowding-out. World Development 28(10): 1719-1733. Carpenter J. P., A. G. Daniere, and L. M. Takahashi, 2004. Social capital and trust in South-east Asian ci es. Urban Studies 41:853-874. Chakravarty, S., D. Friedman, G. Gupta, N. Hatekar, S. Mitra, and S. Sunder. 2011. Experimental economics: a survey. Economic and Poli cal Weekly 46(35): 39-78. Croson, R., and J. Y. Shang. 2008. The impact of downward social informa on on contribu on decisions. Experimental Economics 11(3): 221-233. Cummings, R. G., C. A. Holt, and S. K. Laury. 2004. Using laboratory experiments for policymaking: An example from the Georgia irriga on reduc on auc on. Journal of Policy Analysis and Management 23(2): 341-363. Dasgupta P. 2005. Economics of Social Capital. Economic Record 81:S2-S21. Da a A. K. 1999. Planning and management of water resources: lessons from two decades of early implementa on projects, Bangladesh. Dhaka: University Press Ltd. Denton, M. J., S. J. Rassen , and V. L. Smith. 2001. Spot market mechanism design and compe electric power. Journal of Economic Behavior & Organiza on 44(4):435-453. 194 vity issues in Dewan, C., M.-C. Buisson, and A. Mukherji. 2014. The imposi on of par cipa on? the case of par cipatory water management in coastal bangladesh. Water Alterna ve 7(2): 342-366 Dewan, C. 2012. Literature Review: Review of the Historical Evolu on of Policies and Ins tu ons of Community Based Management in Coastal Bangladesh. Dhaka, Bangladesh: CPWF. Fehr, E. and S. Gächter. 2000. Coopera on and punishment in public goods experiments. The American Economic Review 90 :980-994. Fehr, E., and S. Gächter. 2002. Altruis c punishment in humans. Nature 415(6868): 137-140. Frey B. S. and S. Meier, 2004. Social comparisons and pro-social behavior: Tes ng "condi onal coopera on" in a field experiment. The American Economic Review 94:1717-1722. Greig, F., and I. Bohnet. 2009. Exploring gendered behavior in the field with experiments: Why public goods are provided by women in a Nairobi slum. Journal of Economic Behavior & Organiza on 70(1): 1-9. Hardin G. 1968. The Tragedy of the Commons. Science 162:1243-1248. Harrison, Glenn W. and J. List. 2004. Field Experiments. Journal of Economic Literature 42(4) :1009-55. Heckathorn, D. D. 1993. Collec ve ac on and group heterogeneity: voluntary provision versus selec ve incen ves. American Sociological Review 58(3): 329-350. Henrich J., R. Boyd, S. Bowles. 2004. Founda ons of human sociality: Economic experiments and ethnographic evidence from fi een small-scale socie es. Oxford: Oxford University Press. Horowitz, J. K., and K. E. McConnell. 2002. A review of WTA/WTP studies. Journal of Environmental Economics and Management 44(3): 426-447. Isaac R. M. and J. M. Walker. 1988. Communica on and free riding behavior: The voluntary contribu on mechanism. Economic Inquiry 26: 585-608. Isaac R. M. and J. M. Walker. 1998. Nash as an organizing principle in the voluntary provision of public goods: Experimental evidence. Experimental Economics 1: 191-206. Isaac, R M, J. M. Walker, and S. Thomas. 1984. Divergent Evidence on Free Riding: An Experimental Examina on of Possible Explana ons. Public Choice 43 (1) : 113-49. Janssen M. A., F. Bousquet, J. C. Cardenas, D. Cas llo, and K. Worrapimphong. 2013. Breaking the elected rules in a field experiment on forestry resources. Ecological Economics 90:132-139. Kahneman, D., J. L. Knetsch, and R. H. Thaler. 1990. Experimental tests of the endowment effect and the Coase theorem. Journal of poli cal Economy 98(6): 1325-1348. Kleemeier E., 2000. The Impact of Par cipa on on Sustainability: An Analysis of the Malawi Rural Piped Scheme Program. World development 28: 929-944. Ledyard J. 1995. 'Public goods: a survey of experimental research'. in J. Kagel and A. Roth (eds). The Handbook of Experimental Economics, Princeton, NJ: Princeton University Press: 111-194. Mahmud M., G. Gupta, P. Maitra, and S. Mitra. 2013. Religious iden ty, Minority Status and Trust: Evidence from Field Experiment. Mailath G. J., 1998. Do people play Nash equilibrium? Lessons from evolu onary game theory. Journal of Economic Literature, 36:1347-1374. Marwell G., P. E. Oliver, and R. Prahl. 1988. Social networks and collec ve ac on: A theory of the cri cal mass. III. American Journal of Sociology 94(3): 502-534. 195 Ministry of Water Resources G. o. B., 1999. Na onal Water Policy. Ministry of Water Resources G. o. B., 2001. Guidelines for Par cipatory Water Management. Mitra S. and G. Gupta, 2009. The logic of community par cipa on: experimental evidence from West Bengal. Economic and poli cal weekly, 44(20): 51-57. Mosse D., 2006. Collec ve ac on, common property, and social capital in South India: An anthropological commentary. Economic Development and Cultural Change, 54:695-724. North D. C., 1990. Ins tu ons, ins tu onal change and economic performance. Cambridge university press. Nowell C. and S. Tinkler, 1994. The influence of gender on the provision of a public good. Journal of Economic Behavior & Organiza on, 25:25-36. Olson M., 1993. Dictatorship, democracy, and development. American Poli cal Science Review 87(3): 567-576. Ostrom E., 1990. Governing the commons: The evolu on of ins tu ons for collec ve ac on. Cambridge University Press, Cambridge. Ostrom E., 1992. Community and the endogenous solu on of commons problems. Journal of Theore cal Poli cs 4:343-351. Ostrom E., R. Gardner, J. Walker. 1994. Rules, Games, and Common-pool Resources. Ann Arbor: University of Michigan Press. Rodriguez-Sickert, C., R. A. Guzmán, and J. C. Cárdenas. 2008. Ins tu ons influence preferences: Evidence from a common pool resource experiment. Journal of Economic Behavior & Organiza on 67(1), 215-227. Sell, J., 1997. Gender, strategies, and contribu ons to public goods. Social Psychology Quarterly 60: 252–265. Sell, J., Griffith, W.I., and R. K. Wilson. 1993. Are women more coopera ve than men in social dilemmas. Social Psychology Quarterly 56: 211–222. Shah T. 1996. Catalysing co-opera on: design of self-governing organisa ons. New Delhi : Sage Publica ons. Shogren, K. A., M. N. Faggella-Luby, S. J. Bae, and M. L. Wehmeyer. 2004. The Effect of Choice-Making as an Interven on for Problem Behavior A Meta-Analysis. Journal of Posi ve Behavior Interven ons 6(4), 228-237. Thye, S. R., and E. J. Lawler. 2009. Altruism and prosocial behavior in groups. Emerald Group Publishing. Travers H., T. Clements, A. Keane, and E. J. Milner-Gulland, 2011. Incen ves for coopera on: The effects of ins tu onal controls on common pool resource extrac on in Cambodia. Ecological Economics 71:151-161. Wade R., 1988. Village Republics: Economic condi ons for collec ve ac on in South India. South Asian Studies. Cambridge: Cambridge University Press. Ward J., J. G. Tisdel, T. Straton, and T. Capon. 2006. An empirical comparison of behavioural responses from field and laboratory trials to ins tu ons to manage water as a common pool resource. IASCP procedings, Bali, Indonesia. Velez, M. A., J. J. Murphy, and J. K. Stranlund. 2010. Centralized and decentralized management of local common pool resources in the developing world: experimental evidence from fishing communi es in Colombia. Economic Inquiry 48(2): 254-265. 196 Determinants of contract choice in groundwater irriga on markets in Bangaldesh M. Saidur Rahman1, M. A. Sa ar Mandal1 and Humnath Bhandari2 1 Dept. of Agricultural Economics, Bangladesh Agricultural University, Bangladesh, saidurbau@yahoo.com, asmandal11@gmail.com 2 Agricultural Economist, Interna onal Rice Research Ins tute, Bangladesh, h.bhandari@irri.org Abstract Boro rice accounts for 57% of Bangladesh’s total annual food grain produc on of 33 million tonnes. It is produced under irrigated condi on using mainly groundwater (80%). Many small farmers buy groundwater from neighbours to irrigate their boro rice. One-fourth crop sharing has been the dominant payment system for irriga on since the 1970s but in recent years new payment systems for irriga on have emerged. This study aims to analyze factors influencing the choice of a par cular payment system in groundwater irriga on markets in Bangladesh. To examine this objec ve, ninety-six villages were randomly selected from five divisions of Bangladesh and primary data were collected from those villages through focus group discussions. A mul nomial probit model was used to determine the factors influencing the choice of a par cular payment system in groundwater irriga on. Results show that crop share, fixed charge and two-part tariff are the major payment systems in groundwater irriga on markets in Bangladesh. The probability of choosing these systems was es mated to be 0.58 for fixed charge, 0.25 for crop share, and 0.18 for two-part tariff. Years in irrigated farming, interest rate of loans, access to electricity, drought, and lowland endowment are the factors that strongly influence the choice of crop share payment scheme. The same factors also significantly affect the choice of fixed charge payment. Drought problems and lowland endowment are the factors that significantly affect the choice of two-part tariff payment scheme. This study concludes that two-part tariff and fixed charge cash payment methods are more preferred by the water buyers than the tradi onal crop share scheme. The newly emerging two-part tariff system should be promoted among irrigator farmers since it is economically beneficial to both buyers and sellers of water and it encourages saving irriga on water in rice farming. Key message: Crop share, fixed cash charge and two-part tariffs are the major payment systems in groundwater irriga on water markets in Bangladesh. The payment systems are shi ing from tradi onal crop share to fixed cash charge and two-part tariff. Policies should promote a two-part tariff payment system to increase economic gains and to save irriga on water in rice farming. Keywords: Crop share, cash payment, two-part tariff, groundwater irriga on, water markets, mul nomial probit 1. Introduc on Groundwater irriga on technologies, mainly shallow tubewells (STWs) and deep tubewells (DTWs), play a crucial role in accelera ng food grain produc on (mainly rice and wheat) in Bangladesh. In the 1980s and the 1990s groundwater irriga on increased rapidly, leading to a rapid expansion of paddy cul va on in the dry season, known as the boro paddy. In 2010/2011, groundwater irriga on accounted for 80 percent of the total irrigated area. In the same year, 1.55 million STWs accounted for 67 percent of groundwater irrigated area and about 66 percent of total area irrigated in the dry season (BBS 2010; BER 2010). The markets for groundwater irriga on technologies have been largely liberalized and priva zed since the early = 1980s. With the expansion of STW irriga on, a rela vely compe ve market for irriga on water has evolved. The main characteris c of the market is that the STW owner irrigates his/her own land and partners’ land41 and sells excess water to irrigate the plots of their neighboring farmers, not only for their neighbor’s interest but 41 Ownership of a STW may have more than one person i.e. a group of farmers from same family kinship may buy a STW to irrigate their land together. They are treated as partner of that STW. 197 also for their own benefit. Payment for irriga on water is made in cash per unit of land, one-fourth crop-share or different modes of rental arrangements. As the irriga on water market matures with an increased number of pumps, different pump owners are a emp ng different strategies for running their water selling business profitably. For example, it has been observed that some farmers jointly own STW enterprises and collec vely raise capital for the purchase and opera on of irriga on equipment (Mandal 1993). The number of irriga on technologies has increased significantly in the past three decades in Bangladesh (Gill 1983; Mandal 1993; BBS 2009; BER 2009; Ali 2010; DAE 2011; Rahman et al. 2011; Majumder et al. 2011). At the ini al stage of groundwater irriga on development when the number of tubewells were few, the seller had a monopoly to sell water to the buyer. The payment system and the water rate proposed by the seller were executed directly. The seller’s preference was stronger than that of the buyer in determining the irriga on contract. As the number of irriga on technologies—par cularly STWs—increased exponen ally over the years, the ability of buyers to select payment systems for irriga on water has become stronger. It has also been observed that some tubewell owners are entering this market only for selling water as a business (Rahman et al. 2011). They mostly agree with the preferences of the users. Past studies and field observa ons show three types of payment systems in groundwater irriga on markets in Bangladesh: crop share, fixed charge payment and two-part tariff (Majumder and Rahaman 2011; Rahaman et al. 2011). In the crop share contract, the water seller provides the required amount of irriga on water for rice produc on and receives one-fourth of the crop output in return. Higher risk in crop produc on causes water buyers to prefer a crop share contract (Kajisa and Sakurai 2005). In this system the water buyer gets as much water as required without extra payment and s/he has less incen ve to invest in produc on because part of the produc on goes to the water seller. This contract system discourages saving irriga on water and leads to economic loss due to sub-op mal produc on. In a fixed charge payment, the water seller provides required irriga on water for rice produc on and receives a fixed amount of cash or non-cash commodity. This system also discourages saving irriga on water but it improves economic gain because the water buyer does not share his/her output with water seller and hence s/he is likely to invest maximum resources to a ain maximum produc on. In the two-part tariff system, the seller only provides the machine, which is paid for either in cash or in kind. The buyer then buys the fuel himself. This system encourages saving irriga on water because the farmer/buyer will minimize the use of fuel and it also improves economic gain as all output is kept by the buyer. Thus, the use of different contract payment systems in water markets has implica ons for water saving and produc on efficiency. It is therefore important to understand the dynamics of contract choice in irriga on water markets. The iden fica on of determinants of contract choice will help in understanding more about the irriga on water markets and develop strategies to popularize appropriate contract systems that benefit buyers, sellers, and society as a whole. Studies on the dynamics of contract choice in groundwater irriga on markets in Bangladesh are limited. This study aims to bridge this knowledge gap. The main objec ve of this paper is to document the major payment systems and determine factors influencing the choice of a par cular payment system in irriga on water markets in Bangladesh. Shallow tubewell irriga on is not common in the coastal zone of Bangladesh but the payment systems for irriga on and its determining factors are similar and important for coastal zone irriga on as well. 2. Methods 2.1 Data This study uses village-level primary data collected from boro rice farmers located across 96 villages in Bangladesh. The unit of analysis is a boro rice farming village. The sample village consists of both water buyer and water seller. Ninety six villages were selected from five divisions— i.e. Rajshahi, Rangpur, Dhaka, Khulna, and Chi agong— of Bangladesh following a mul -stage stra fied random sampling technique. Boro rice produc on, use of groundwater irriga on and prevalence of water markets were the main basis for designing the sampling framework and selec ng study sites. Data were collected from 96 villages using Focus Group Discussions (FGD). The village-level data were derived based on group consensus and the data represent the 198 average situa on of the sample villages. The data collected from 48 upazillas of 31 districts of Bangladesh broadly represent the whole country (Figure 1). A computer assisted personal interview (CAPI) method and Surveybe so ware was used to collect the data. The collected data include biophysical and socioeconomic characteris cs of the village, boro rice yield and details of irriga on water use and the payment system. The survey was conducted from April to September 2013 and the data were collected for the crop year 2013. Fig. 1. Map of the loca ons of the 96 sample villages with GPS coordinates (prepared by GIS lab, IRRI). 2.2 Mul nomial Probit Model Specifica on A discrete choice model is used when a decision maker is confronted with choosing among a set of alterna ves (Train 2007). To fit within a discrete choice framework, the set of alterna ves – the choice set – needs to exhibit three characteris cs: (a) alterna ves need to be mutually exclusive, (b) alterna ves must be exhaus ve, and (c) the number of alterna ves must be finite. Binomial logit or probit discrete choice model is used when the choice set consists of two alterna ves. Mul nomial logit or probit model is used when the choice set consists of more than two alterna ves. In our case, the decision maker (the water buyer) is confronted with the choice of three alterna ve payment methods (crop share, fixed charge and two-part tariff) for irriga on. Therefore, mul nomial logit (MNL) or mul nomial probit (MNP) model is appropriate in our case 42. The MNL or MNP models differ slightly in terms of distribu onal characteris cs but both models give similar parameter es mates. Both MNL or MNP models assume Independence of Irrelevant Alterna ves (IIA) and errors are uncorrelated, i.e. cov(εj, εi)=0. This basically means choice probabili es in the models are es mated using a technique involving random draws and Monte Carlo es ma on. The MNP model es ma on with the “asmprobit” command in Stata so ware uses the simulated Maximum Likelihood technique known as the Geweke-Hajivassiliou-Keane (GHK) algorithm to sa sfy IIA assump ons and it produces highly reliable parameter es mates (Geweke 1991; Keane 1990; Keane 1994; Hajivassiliou and MacFadden 1998; Hajivassiliou et al. 1996; StataCorp 2007; Kropko 2008). Therefore, MNP model is used in our case to examine the choice of alterna ve payment methods for irriga on. The technical details of the mul nomial probit model are presented in Appendix 1. 42 See Green (2008) for detailed specifica on of logit and probit models. 199 2.3 Empirical specifica on of the mul nomial choice model The mul nomial probit model used to select the payment systems in groundwater irriga on is specified below. Yi*,0 = Xi,0β0 + εi,0 Yi*,1 = Xi,1β1 + εi,1 ε i ~ type-1 extreme value distribu on of their error term Yi*,2 = Xi,2β2 + εi,2 Yi = { 0 if crop sharing; Yi*,0 > Y i * ,1 & Yi *,0 > Yi *,2 1 if cash payment (fixed charge); Yi * ,1 > Yi *,0 & Yi *,2 > Yi,2* 2 if cash payment (two part tariff); Yi *,2 > Yi *,0 & Yi * ,2 > Yi *,1 Yi = Dependent variable (payment systems or contract choices) Xi = Vector of independent explanatory variables including the biophysical and socio-demographic factors of the sample villages Table 1. List of dependent and independent variables for empirical mul nomial probit model. Dependent variable (Y i) Independent variables (Xi) Contract choices: 0 = Crop share 1 = fixed charge 2 = two-part tariff X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 Yield (kg/ha) Frequency of user’s plot visit by the seller (no.) Rela on between water seller and water buyer (no. of talks between the seller and the buyer) Chance of denied irriga on by the seller (%) Irriga on cost (Tk./ha) Years of using groundwater irriga on (no.) Interest rate on NGOs loan (Tk.) Flood dura on in the last 10 years (days) Electricity dummy (1= village has electric, 0=otherwise) Drought dummy (1=village is prone to drought, 0=otherwise) Lowland dummy (1=most land in the village is a lowland, 0=otherwise) Distance from Upazila dummy (1= less than or equal to 10 km, 0=otherwise) A mul nomial probit model was used to es mate factors influencing the choice of a par cular payment system in groundwater irriga on. The dependent variable consists of three alterna ve systems of contracts: crop share, fixed charge or two-part tariff. The independent variables included in the model are described in Table 1. The rela onship between the contract choice and yield as well as between the contract choice and the irriga on cost are likely to be endogenous. Theore cally, this endogeneity problem can be solved using an instrumental variable approach. But finding exogenous instruments might be tricky in this case and hence correc ng the endogeneity problem in the specified model is beyond the scope of this paper. Therefore, the poten al endogeneity problem in the model is le uncorrected and it is acknowledged as a limita on of the model. The es mated coefficients of the mul nominal probit model are interpreted rela ve to the reference group. The standard interpreta on of the mul nomial probit model coefficients is that for a unit change in the predictor variable, the rela ve probability of choosing a par cular payment system, as compared to the reference payment system, is expected to change by its respec ve parameter es mate, ceteris paribus. 200 2.4 Inclusion of socio-demographic variables in the model By defini on, a ributes of the decision maker do not vary over the alterna ves. As a result, they can only enter in the model in ways that create differences in u lity over the alterna ves. This can be seen by using three alterna ves: U0=αX0 + βX0 + θ0Z+ ε0 U1=αX1 + βX1 + θ1Z + ε1 U2=αX2 + βX2 + θ2Z + ε2 Where Z is a vector of some socio-demographic variables and θ0, θ1and θ2 capture the effect of this variable on the u lity of alterna ves 1 (u1) and 2 (u2), respec vely. We might expect that θ 0 ≠ θ1, i.e., the effect of the variable differs across alterna ves. The problem is that only differences in u lity ma er and so we cannot es mate the absolute levels of θ 0, θ1 and θ2. Again we need to normalize one of these parameters to 0. Now the model becomes: U0=αX0 + βX0 + ε0 U1=αX1 + βX1 + + v* 1Z + ε1 U2=αX2 + βX2 + v*2Z + ε2 Where θ1= θ 1 - θ0 and θ2= θ2 - θ 1. It is possible to interpret θ1 as the differen al impact of the variable on the u lity of alterna ve 2 (v*2) compared to alterna ve v*1. 3. Water markets in Bangladesh 3.1 Characteris cs of water markets The water market is an emerging market all over the world. Recognizing irriga on as a main input in agriculture, water markets have become a focal point of discussion. The term “water market” refers to localized, village level informal arrangements through which owners of modern wells sell irriga on services to other members of the community (Shah 1989). It is also noted that once the owner finishes irriga ng his own plots, it is to his advantage to sell the surplus water to as many buyers as possible, as long as s/he can cover costs (Kajisa et al. 2005; Shah 1993). The term water market is a general term as water may be li ed from open or tube wells; deep or shallow wells; or from canals, tanks, rivers, drains or other surface sources. It may be transported to the buyer’s field through unlined field channels, plas c pipes, or underground pipeline networks. Where land holdings are fragmented, most sellers of water are also buyers of water from other owners. Most farmers sink wells in one or two of their largest and the best parcels. They o en purchase water for irriga ng their own fields that cannot be irrigated from their own tubewells and also sell water to other farmers under various payment systems. Water markets started as natural oligopolies but large expansion of irriga on equipment markets and support services is leading to increasing compe on in water markets. Nevertheless, a number of specific characteris cs of water inhibit smooth opera on of water markets (Tsegabirhan 2004). 3.2 Mode of exis ng payment systems for irriga on water The payment system for water significantly influences water use and, hence, the performance of STWs and DTWs. As the irriga on water market is maturing with an increased number of pumps different pump owners pursue different strategies in order to run their water selling business profitably. Use of different payment systems for irriga on water is one strategy. In Bangladesh two modes of payments for water are prac ced under DTW and STW: crop share payment and fixed cash payment (Table 2). In the early stage of groundwater irriga on development, choice of a payment system was determined by whether the machine was operated by diesel or electricity. Usually, crop share payment was used for diesel operated machines and fixed cash payment was used for electricity operated machines. Although diesel operated irriga on is more expensive than electricity operated irriga on, farmers are compelled to use diesel pumps due to lack of electricity in the villages (Mukherji 2007). 201 The share of resources for irriga on water supply between the seller and the buyer, flexibility and reliability of water supply, and the amount of money paid for water differ between crop share and fixed charge payment systems. In crop share systems, the tubewell owner supplies water to the user’s plot for the en re irriga on season and receive 25 percent of the harvested crop. STW owners collect their share from the buyer a er the harves ng of boro rice. The cost of boro rice irriga on for water buyers was es mated to be Bangladesh Taka (BDT)18,000–25,000/season/ha for both diesel or electricity operated machines. The seller has an incen ve to provide mely and adequate quan ty of water to his buyers because the seller will receive a good share of harvest if the crop is well irrigated, holding other factors the same. It should be noted that the yield also depends on the users’ prac ces regarding weeding, seedlings, fer lizer and insec cides. Clima c factors such as rainfall, flood, and drought also effect yield. Table 2. Exis ng contracts in irriga on water markets Sl. Opera ng Mode of No. system payment 1. Diesel Crop 2. Diesel Cash 3. Diesel Cash 4. Electricity Cash 5. Electricity Crop Rate of payment ¼ share of harvest Fixed/ha/ season Fixed/ha/ season Fixed/ha/ season ¼ share of harvest Timing of payment A er harvest Beginning the season Beginning the season Beginning the season A er harvest Price range Input (BDT) provider 18000-25000 Seller Popularity Trend Popular Old 15000-18000 Seller Not so popular Most popular Popular Emerging Not so popular Old 12000-13000 Buyer 14000-16000 Seller 18000-25000 Seller Most emerging Emerging Source: Rahman et al. 2011 In fixed charge systems, the seller provides water for the whole crop season and receives a fixed amount of cash or non-cash commodity. Fixed cash payment per unit of land or me is usually determined based on the average cost of supplying water including fuel, management and supervision. Payment is made partly prior to and partly a er harves ng rice. Two types of fixed cash payment systems are prac ced in Bangladesh. In the first, the STW owner provides water to the user for the whole season and receives a fixed amount of cash per unit of land. The cash amount that the STW owner receives depends on local arrangements between owner and user, me of payment, soil type and eleva on of the plot, distance from the STW, and rela on with the owner. In this system, the cost of boro rice irriga on for the water buyer was es mated to be BDT15,000–18,000/ season/ha for diesel operated machines and BDT14,000–16,000/season/ha for electricity operated machines. In the second system, users pay a service charge for using the STW but provide their own diesel and other irriga on needs. The service charge varies depending on loca on and contract between the owner and the user. This system is prac ced only with diesel operated machines and the cost of boro rice irriga on for water buyers was es mated to be much lower (BDT12,000–13,000/season/ha) than the other payment system. This system is newly introduced in the country and it is gradually replacing the tradi onal crop share system and the previously men oned fixed cash per unit of land system. This payment system emerged due to the high price of diesel and engine oil, lack of commitment to agreements by owners and be er access to financial resources by owners. The STW owners take less responsibility in this system and transac on cost are also reduced(Rahman et al. 2011; Aggarwal 2007). The irriga on water payment contract systems in Bangladesh are similar to that of other South Asian countries (Bhandari and Pandey 2006; Kajisa and Sakurai 2005). Kajisa and Sakurai (2005) found three types of contracts in irriga on water markets in India: fixed charge per season, flat charge per applica on of water and output sharing. In a fixed charge contract, a water buyer pays a fixed amount of cash once per season for 202 specified irriga on acreage for the en re season. In a flat charge contract, a buyer irrigates as much as he wants at a given price per acre and pays each me he applies the water. In an output sharing contract a buyer pays for the water by providing a certain por on of his product a er the harvest of the crop. 3.3 Groundwater irriga on payment op ons available at village level The results of the descrip ve analysis show that farmers in 68% of the sample villages have only one payment system op on in groundwater irriga on markets, and farmers in the remaining 32% of the sample villages have more than one payment system op on (Figure 2). 100 90 Freq. & Percent 80 70 60 50 40 30 20 10 0 Frequency Percent 1 65 67.71 2 31 32.29 Total 96 100 Fig. 2. Percentage distribu on of available payment op ons at the village level, 2013. 3.4 Frequency distribu on of payment op ons Table 3 shows that the single choice payment op on is available in 65 sample villages and more than one payment op on is prevailing in 31 sample villages. It is interes ng to look into the exis ng payment combina ons in the villages. Out of 31 villages with more than one payment system, seven villages have crop share and fixed charge payment systems, one village has crop share and two-part tariff payment systems, and the remaining 23 villages have fixed charge and two-part tariff payment systems. Figure 3 provides a clear picture of the prevailing scenario of payment op ons for irriga on water in rural areas of Bangladesh. Fixed charge is the dominant payment op on (51%) in villages with a single payment op on, while a combina on of fixed charge and two-part tariff is dominant (74%) in villages with two op ons. The results also show that two forms of cash payment prevail in the sample villages: fixed charge and two-part tariff (Fig. 3). Payment systems at village level Choice: Single (68%) Crop share (23%) Fixed charge (23%) Choice: Double (32%) Two part tariff (26%) Crop share & fixed chaerge (23%) Crop share & two part tariff (23%) Fixed charge & two part tariff (74%) Figure 3. Payment system op ons at the village level. 203 Table 3. Frequency distribu on of prevailing alterna ve payment contract op ons in the sample villages, 2013 Payment op on Single=65 Double=31 Payment types Crop Fixed Two part share charge tariff 15 33 17 - Payment combina ons Crop share & Crop share & fixed charge two part tariff 7 1 Fixed charge & two part tariff 23 Source: Field survey, 2013 4. Results and discussion Table 4 presents the summary results of the es mated mul nomial probit model. The reference group in the model is crop share payment system. Thus, the results are interpreted in terms of fixed charge payment system versus crop share payment system and two-part tariff payment system versus crop share payment system. A rela vely large likelihood ra o and the sta s cally significant Chi square value indicate that the es mated mul nomial model equa on is well fi ed to the data. Table 4. Summary results from mul nomial probit model es ma on, 2013 Number of observa ons = 96 Wald chi2 (24) = 34.73 Probability > chi2 = 0.0725 Log likelihood = -58.438289 0 Payment system 1 Paddy yield Frequent visit to the user’s plot by the seller Rela on between the seller and the buyer Chance of denied irriga on by the seller Irriga on cost Years of using groundwater irriga on Interest rate of NGOs loans (Tk.) Flood dura on Electricity dummy Drought dummy Low land dummy Distance from village to upazila headquarters dummy Constant term 2 Paddy yield Frequent visit to the user’s plot by the seller Rela on between the seller and the buyer Chance of denied irriga on by the seller Irriga on cost Years of using groundwater irriga on Interest rate of NGOs loans (Tk.) Flood dura on (day) Electricity dummy Drought dummy Low land dummy Distance from village to upazila headquarters dummy Constant term Note: ***, **, * indicate the level of significance at 1%, 5% and 10%, respec vely. 204 Base outcome = Crop share Coefficients Std. Err. Fixed charge 0.00108 *** 0.00044 -0.01912 0.01998 0.00011 0.02269 0.11020 0.07384 -0.00010 ** 0.00005 -0.13750 *** 0.04389 -0.07803 * 0.04276 ** -0.02760 0.01236 1.91391 ** 0.94990 -2.07133 *** 0.72280 1.57627 *** 0.62687 0.45375 0.54716 2.51065 3.01869 Two part tariff 0.00070 0.00046 ** -0.04767 0.02295 0.05144 ** 0.02616 -0.03440 0.08295 -0.00007 0.00005 -0.02453 0.04666 -0.02196 0.04773 -0.00132 0.01327 -0.19003 0.77525 -0.17991 0.76776 -0.21757 0.77941 0.37720 0.62261 -0.80239 3.53152 4.1 Determinants of fixed charge payment rela ve to crop share The maximum likelihood es mates of the mul nomial probit model show that the rela ve probability of choosing a fixed charge payment system versus a crop share system is posi vely related to paddy yield, access to electricity and having lowland. It is nega vely related to cost of irriga on, years of using groundwater irriga on, interest on loan, longer dura on of floods and drought problems. The rela onships with the rest of the variables are insignificant (Table 4). The es mated coefficients indicate that the rela ve probability of choosing a fixed charge payment versus crop share payment increased by 0.001 units for one unit increase in yield, by 1.91 units for those villages with access to electricity as compared to those without electricity, and by 1.58 units for lowlands villages as compared to other villages. The rela ve probability of choosing a fixed charge payment versus crop share payment decreased by 0.0001 units for one unit increase in irriga on cost, by 0.14 units for one unit increase in the years of using groundwater irriga on, by 0.08 units for one unit increase in the interest rate of a loan, by 0.03 units for each unit increase in flood dura on and by 2.07 units for villages affected by drought as compared to those that are not affected. 4.2 Determinants of two-part tariff payment rela ve to crop share The model es mates show that the rela ve probability of choosing a two-part tariff payment system versus crop share system is posi vely related to the rela on between the seller and the buyer and nega vely related to the frequency of visit to the buyer’s plot by the seller, while the rela onship with the rest of variables was insignificant. The es mated coefficients indicate that the rela ve probability of choosing a two-part tariff payment versus crop share payment increased by 0.05 units for one unit increase in the rela onship between the seller and the buyers and it decreased by 0.05 units for one unit increase in the frequency of seller’s visit to the buyer’s field. 4.3 Determinants of choosing a crop share payment system Table 5 presents the marginal effect coefficients of different variables in choosing a crop share payment system for groundwater irriga on. The parameter es mates of the probit model show that the probability of choosing a crop share system is posi vely related to cost of irriga on, years of using groundwater irriga on, flood problems, and drought problems. It was nega vely related to paddy yield and having lowland area. The variables that have a rela vely large effect in choosing a crop share payment are drought dummy and lowland dummy. The marginal effect coefficient of a variable indicates the effect of that variable on the probability of choosing a crop share payment system. The marginal effect coefficient of -0.0002 for paddy yield indicates that a 1 t/ha increase in paddy yield will decrease the probability of choosing a crop share payment method by 0.02%, ceteris paribus. The marginal effect coefficient -0.21723 for the lowland dummy indicates that the villages with lowlands have a 22% higher probability of choosing a crop share payment than those without lowlands, ceteris paribus. The coefficients of other variables can be interpreted in the same fashion. Table 5. Marginal effects of the predicted payment system (crop share), 2013 Variables Yield (kg/ha) Frequent visit (no.) Good rela on (no.) Chance of denied irriga on (%) Irriga on cost (Tk./ha) Years of using irriga on (year) Interest rate of NGOs (Tk.) Flood dura on (day) Electricity dummy Drought dummy Low land dummy Distance from upazila dummy y = Pr(Payment method=0) (predicted outcome of Crop share) = .18634181 dy/dx Std. Err. Expected value (X) -0.00020*** 0.00008 6286.90 0.00515 0.00374 45.66 -0.00232 0.00423 28.62 -0.01586 0.01345 5.62 0.00002** 0.00001 19348.70 0.02280*** 0.00841 31.96 0.01330 0.00831 29.77 0.00442* 0.00234 15.13 -0.27838 0.20389 0.86 0.25437*** 0.08115 0.73 -0.21723** 0.09457 0.42 -0.09292 0.10988 0.68 Note: ***, **, * indicate the level of significance at 1%, 5% and 10%, respec vely. 205 4.4 Determinants of choosing a fixed charge payment system Table 6 presents the marginal effect coefficients of different variables in choosing a fixed charge payment system for groundwater irriga on. The factors that significantly affect the probability of choosing a fixed charge payment system are yield, chance of denied irriga on by the seller, irriga on cost, years of using groundwater irriga on, interest rate on NGO loans, dura on of floods, access to electricity, occurrence of drought, and land type under boro rice. The probability of choosing a fixed charge payment system is posi vely related to yield, chance of denied irriga on, access to electricity, and endowment of lowland. It is nega vely related to irriga on cost, years of using groundwater irriga on, interest rate on NGO loans, longer flood dura on and occurrence of drought. The variables that have a rela vely large effect on choosing a fixed charge payment are chance of denied irriga on, years of using groundwater irriga on, access to electricity, occurrence of drought and endowment of lowlands. The marginal effect of 0.0337 for chance of denied irriga on indicates that a one percent increase in refusing to provide irriga on by the seller will increase the probability of choosing a fixed charge payment system by 3.37%. Likewise, the marginal effect of -0.4138 for the drought dummy indicates that villages prone to drought have a 41.38% lower probability of choosing a fixed charge payment than those villages not prone to drought. The coefficients of other variables can be interpreted in the same fashion. Table 6. Marginal effects of the predicted payment system (fixed charge) Variables Yield (kg/ha) Frequent visit (no.) Good rela on Chance of denied irriga on (%) Irriga on cost (Tk./ha) Years of using irriga on (year) Interest rate of NGOs (Tk.) Flood dura on (day) Electricity dummy Drought dummy Low land dummy Distance from upazila dummy y = Pr(Payment method=1) (predicted, outcome of fixed charge) = .68922906 dy/dx Std. Err. Expected value (X) 0.00021** 0.00010 6286.90 0.00020 0.00468 45.66 0-.00576 0.00553 28.62 0.03367* 0.01760 5.62 -0.00002* 0.00001 19348.70 -0.03442*** 0.01075 31.96 -0.01863* 0.01075 29.77 -0.00732** 0.00317 15.13 0.55805*** 0.19105 0.86 -0.41381*** 0.09850 0.73 0.40726*** 0.11397 0.42 0.08361 0.13574 0.68 Note: ***, **, * indicate the level of significance significant at 1%, 5% and 10%, respec vely. 4.5 Determinants of choosing a two-part tariff payment system Table 7 presents the marginal effect coefficients of different variables in choosing a two-part tariff payment system for groundwater irriga on. There are fewer variables that significantly influence the probability of choosing a two-part tariff payment system. These variables are user’s plot visits by the water seller, conversa on between user and seller, drought problems and land type. The probability of choosing a two-part tariff payment system increased by 0.80% for each increase in conversa ons between the seller and user and by 15.94% for villages prone to drought. In contrast, the probability of choosing a two-part tariff payment system decreased by 0.54% for each increase in visits by seller to users’ plots and by 19.00% for villages located in low lands. 206 Table 7. Marginal effects of the predicted payment system (two-part tariff) Variables y = Pr(Pay_method1=2) (predicted outcome of two part tariff) = .12442913 dy/dx Std. Err. Expected value (X) -0.00001 0.00006 6286.90 -0.00535* 0.00323 45.66 0.00808** 0.00394 28.62 -0.01781 0.01206 5.62 0.00000 0.00001 19348.70 0.01161 0.00716 31.96 0.00533 0.00717 29.77 0.00290 0.00208 15.13 -0.27967 0.18668 0.86 0.15944** 0.07003 0.73 -0.19002** 0.08602 0.42 0.00931 0.08740 0.68 Yield (kg/ha) Frequent visit (no.) Good rela on Chance of denied irriga on (%) Irriga on cost (Tk./ha) Years of using irriga on (year) Interest rate of NGOs (Tk.) Flood dura on (day) Electricity dummy Drought dummy Low land dummy Distance from upazilla dummy Note: ***, **, * indicate the level of significance significant at 1%, 5% and 10%, respec vely. 4.6 Aggregate probability of choosing a payment system from the model The mul nomial probit regression model shows that the probability of choosing fixed charged payment is highest (0.58), followed by crop share (0.25) and two-part tariff (0.18). The es ma ons from the specific mul nomial probit model are similar to the exis ng payment methods actually prac ced by farmers for groundwater irriga on in Bangladesh (Table 8). Table 8. Probability of choosing payment methods for groundwater irriga on markets Payment methods Crop share, p0 Fixed charge, p1 Two-part tariff, p2 Observa on 96 96 96 Mean 0.24588 0.578376 0.175743 Std. Dev. 0.242514 0.341866 0.201567 Min 1.39E-05 0.001017 3.65E-09 Max 0.958907 0.999854 0.817899 The village level survey data show some changes in payment methods during the last 10 years but these changes are not significant. Farmers observed that changes in the payment systems in the short- to medium-term future will be small as users have limited payment op ons to choose. Only a few villages have more than one payment method. The individual household survey data also show that only one payment op on exists in a par cular tubewell command area, thus, no providing no op on to choose. Table 9. Payment methods in irriga on water markets from mul nomial probit model Payment systems Frequency Percent Cumula ve percent Crop share ( 0) 23 23.96 23.96 Fixed charge (1) Two part tariff (2) 56 17 58.33 17.71 82.29 100 Total 96 100 - It can be seen from Tables 3 and 9 that the predicted values and survey-based values do not differ radically. So the effect on the es mated parameters is fairly modest. We can confidently use these parameter es mates to forecast future payment systems based on the significant coefficients from the model es ma on. 207 4.7 Alterna ve-specific mul nomial probit: some limita ons In this analysis, the mprobit model is used to examine why farmer choose a specific payment system and prac ce it over the years. Why do they think that the method is somehow good for them? Why do they change their payment methods? What were the reasons behind the change? What are the characteris cs of the village that determine the choice? Theore cally, there are three op ons for farmers but in a par cular village, there are few op ons. And our collected dataset does not have all relevant informa on, like rice yield of households within the villages for different payment systems. It is therefore not possible to test the asmprobit command. This is considered to be one of the limita ons of the study—limited data due to the prevailing groundwater irriga on system. Asmprobit fits the mul nomial probit model by using maximum simulated likelihood (MSL) implemented by the Geweke-Hajivassiliou-Keane (GHK) algorithm. By es ma ng the variance-covariance parameters of the latent-variable errors, the model allows us to relax the independence of irrelevant alterna ves (IIA) property that is a characteris c of the mul nomial logis c model. Asmprobit requires mul ple observa ons for each decision, where each observa on represents an alterna ve that may be chosen. The cases are iden fied by the variable specific to the case op on, whereas the alterna ves are iden fied by the variable specified in the alterna ve op on. The outcome or chosen alterna ve is iden fied by a value of 1 for the dependent variable, indica ng the alterna ves chosen; only one alterna ve may be chosen for each case. It allows for two types of independent variables: alterna ve-specific variables and case-specific variables. Alterna ve-specific variables vary only across cases and are specified in the case variable op on. 5. Conclusions and recommenda ons There are three payment systems that exist in the groundwater irriga on market in Bangladesh. In the 1970s and 1980s, only a crop share payment system existed. A er the import market liberaliza on of small scale STW engines in the late 1980s, farmers of Bangladesh increased the cul va on of irrigated HYV boro rice. Small and cheap irriga on technologies enabled farmers to invest more on tubewells, leading to more land under irriga on. A er years of experience in irriga on, some farmers adopted other payment systems, i.e. fixed-charge cash payment and two-part tariff payment. The la er is the latest payment scheme wherein the farmer will only pay a service charge for the use of the tubewell. Fuel and other needs are borne by the farmer/buyer himself. The descrip ve analysis shows that the number of fixed charge and two-part tariff are payments have increased over the years and the crop share system has decreased. Depending on the nature of the irriga on water market, users have choices but not full freedom to choose a par cular payment. It is also seen that the seller and the user face some disputes in water transac ons. They have introduced self resolu on protocol at the village level. Sellers and buyers rarely go to court to resolve their disputes regarding irriga on contracts. A er 30 to 40 years of experience in irriga on transac ons they have adopted a sustainable, village-based resolu on system that is unique to the market. There are fewer disputes reported with the use of a two-part tariff payment system compared to the fixed-charge system, which is however more commonly found among farmers. A crop share payment system does not promote water conserva on and discourages water users from saving irriga on water because they get as much water as they requires without extra cost. Also, this system provides less incen ve for water users to invest resources in rice produc on and achieve maximum yield because part of the produc on will be shared with the water seller. This produc on behavior leads to an economic loss. The irriga on cost per unit area for the water user is higher in crop share systems than other payment systems. Therefore, crop share payment systems are less economically beneficial to the seller, the buyer and society as a whole. Compared to crop share systems, fixed charge payment systems are economically beneficial but also do not encourage the water user to save irriga on water. On the other hand, two-part tariff systems are not only economically beneficial to both the buyer and the seller, but also provides incen ves to the user to save irriga on water. Therefore, proper policies and a concerted effortsare needed to change the payment system in irriga on water markets from crop share to fixed charge or two-part tariff payment, depending on the local context. 208 Rural electrifica on is one of the significant factors that increases the probability of choosing a fixed charge payment method. The very high price of paddy in 2007 and 2008 is also believed to be a major cause for the expansion of fixed charge and two-part tariff payment systems. If output price increases without propor onate increase in irriga on cost, that reduces the popularity of crop share payment systems and presents strong economic reasons for shi ing to any other payment system. Expansion of electrifica on in rural areas and con nuous and regular supply of electricity is one way of expanding fixed charge payment systems. If the village is dominated by diesel operated tubewells, promo on of a two-part tariff payment system would be a best bet. It is also noted that cash and credit availability are major issues. It has been observed that credit availability is increasing in rural areas over mes due to the prolifera on of micro-finance ins tu ons. The interest rate of credit from money lenders is also decreasing because more NGOs and micro-finance ins tu ons are opera ng in rural areas. Commercial bank loans are also becoming increasingly more accessible and cheaper for the farmers who are living closer to the upazila main towns and in villages with good transport and communica on infrastructure. Farmer are also becoming more conscious about their ability to access resources through media support, with the excep on of those farmers living in remote areas. Higher educa on, awareness and trainings, opportuni es to diversify income sources and changing livelihoods make farmer more careful about choosing their agricultural prac ces and other ac vi es for improving their welfare. Recently, the government of Bangladesh has provided facili es for loans without collateral, especially for tenant farmers. NGOs provide loans to smallholder farmers and the rural poor without collateral. Grameen Bank and other small-scale credit providers are coming forward to support the agricultural and small-scale business ac vi es of rural poor farmers. Such posi ve efforts makes farmers more capable of bearing irriga on expenses on their own and give them more flexibility to choose a suitable payment system for irriga on water. Addi onal tubewell owners are entering into the water selling business and they are providing more op ons to the users for expanding their tubewell command area. More research is needed to study these emerging op ons and promote a compe ve water market that not only benefits both buyers and sellers but also saves irriga on water by increasing water produc vity. Groundwater is not widely used in south Bangladesh due to salinity problem. But there are some pocket areas in the coastal and inland coastal zone of south Bangladesh where farmers are using non-saline groundwater to irrigate their crops using tubewells. As the adop on of water-intensive improved agricultural technologies gradually increases in coastal and inland coastal zones, farmers are also increasing investment in irriga on, including pumping of groundwater where it is not saline and purchasing irriga on machines to pump surface water from ponds, canals and khals. These changes are likely to expand water markets in south Bangladesh in the future. Different strategies are needed for efficient use of scarce freshwater resources and for increasing water produc vity in south Bangladesh. This study indicates that the promo on of a two-part tariff payment system is one simple and effec ve strategy to save irriga on water in rice produc on and will increase water produc vity. This two-part tariff can be promoted in south Bangladesh for efficient use of freshwater resources. Acknowledgements This paper has been derived from IRRI’s Village Dynamics in South Asia (VDSA) Project, which funded the first author’s PhD disserta on research on “Determinants of water price, contract choice and rice produc on efficiency in groundwater irriga on markets in Bangladesh.” The authors are thankful to IRRI and the VDSA Project for funding this research work. The authors would also would like to thank Valerien O. Pede of IRRI for providing valuable inputs to this paper. 209 References Aggarwal, RM. 2007. Role of risk sharing and transac on costs in contract choice: Theory and evidence from groundwater contracts, Journal of Economic Behavior & Organiza on, 63: 475–496. Ali, M. H. 2010. Fundamentals of Irriga on and On-farm Water Management, volume 1, No. XXII, 556P, 89 illus, 29 in color, Hardcopy, ISBN: 978-1-4419-6334-5. BADC, 2008. Minor Irriga on Survey Report 2007–08. Bangladesh Agricultural Development Corpora on, Ministry of Agriculture, GOB, Dhaka, Bangladesh, p 122. Bangladesh Agricultural University, 1986. Water Markets in Bangladesh: Inefficient and Inequitable? 285p, Mymensingh. BBS (Bangladesh Bureau of Sta s cs), 2001. Yearbook of Agricultural Sta s cs of Bangladesh. Ministry of Planning, The Government of the Peoples Republic of Bangladesh (GOB). BBS, 2009. Sta s cal Year book of Bangladesh, Ministry of Planning, GOB, Dhaka, Bangladesh. BBS, 2009. Sta s cal Year book of Bangladesh, Ministry of Planning, GOB, Dhaka, Bangladesh. BBS, 2010. Sta s cal Year book of Bangladesh, Ministry of Planning, GOB, Dhaka, Bangladesh. BER, 2009. Ministry of Finance, Government of the People's Republic of Bangladesh, Dhaka. BER, 2010. Ministry of Finance, Government of the People's Republic of Bangladesh, Dhaka. Bhandari H, Pandey S., 2006. Economics of groundwater irriga on in Nepal: some farm-level evidences. Journal of Agricultural and Applied Economics, 38 (1): 185-199. DAE, 2011. Annual Report 2010. Department of Agricultural Extension (DAE), Ministry of Agriculture, GoB, Bangladesh. EPADC (East Pakistan Agricultural Corpora on). 1968. Annual report 1966–67, East Pakistan Agricultural Corpora on, Dhaka. Gill, G.J. 1983. The Demand for Tube Well Equipment in Rela on to Groundwater Availability in Bangladesh. AERSS Paper No. 13. Bangladesh Rural Advancement Commi ee, Dhaka. Green, W. H., 2008. Econometric analysis, 6th ed., N.J. (USA): Pren ce Hall. Hajivassiliou, V. and D. McFadden, 1998. The Method of Simulated Scores for the Es ma on of LDV Models. Econometrica. 66: 863-896. Hajivassiliou, V., D. McFadden and P. Ruud, 1996. Simula on of Mull variate Normal Orthant Probabili es: Methods and Programs. Journal of Econometrics, 72:85-134. IBRD (Interna onal Bank for Reconstruc on and Development), 1972. Land and Water Sector Study, The World Bank. Kajisa, K. and Sakurai, T. 2005. Efficiency and equity in groundwater markets: the case of Madhya Pradesh, India. Environment and Development Economics, 10(6): 801-819. Keane, M. P. 1990. Four Essays in Empirical Macro and Labor Economics. PhD Disserta on. Brown University. Keane, M. P. 1994. A Computa onally Prac cal Simula on Es mator of Discrete Choice Models. Econometric Theory. 11:437-483. Kotz, S. and Nadarajah, S. 2000. Extreme Value Distribu ons: Theory and Applica ons. London: Imperial College Press. 210 Kropko, J. 2008. Choosing Between Mul nomial Logit and Mul nomial Probit Models for Analysis of Unordered Choice Data. MS thesis submi ed to the faculty of The University of North Carolina at Chapel Hill. Majumder, M. K. and S. Rahman. 2011. Impact of Introducing LLP on Improving Income of Farm Households in Some Areas of Bhola District in Bangladesh. J. Bangladesh Agril. Univ., 9(2): 233-240. Mandal, M. A. S. 1993. Groundwater Irriga on in Bangladesh: Access, Compe on, and Performance. Friedrich Kahnert and Gilbert Levine edited book on Groundwater Irriga on and the Rural Poor: Op ons for Development in the Gange c Basin, The World Bank, Washington, D.C. Mc Fadden, D. 1973. Condi onal Logit Analysis of Qualita ve Choice Behavior. In Fron ers of Economitrics, Ed. P. Zarembka, Ney Work: Academic Press. Michler, J. D. and Wu, S. 2013. Contrac ng for Groundwater Irriga on: A Principal-Agent Based Approach to Determining Contract Effec veness. Paper presented in the AAEA & CAES joint annual mee ng, Washington, DC, August 4-6, 2013. Mukherji, A. 2004. Groundwater Markets in Ganga-Meghna-Brahmaputra Basin: Theory and Evidence. Economic and Poli cal Weekly, 30(31):3514-3520. Mukherji, A. 2007. The Energy-Irriga on Nexus and Its Impact on Groundwater Markets in Eastern Indo-Gange c Basin: Evidence from West Bengal, India, Energy Policy, 35(12):6413–6430. Rahman, M. S. and A. Angelsen. 2011. Comparison between Irriga on Payment Systems and Probability of Using Water Saving Technology. Journal of Economics and Sustainable Development, ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) www.iiste.org, Vol.2, No.10. Sa ar M. A. 1999. Irriga on Development and Management in Bangladesh. Oxford & IBH, New Delhi, p 120. Shah, T. 1993. Groundwater markets and irriga on development: Poli cal economy and prac cal policy, Oxford University Press, Bombay. StataCorp. 2007. Stata Sta s cal So ware: Release 10. College Sta on, TX: StataCrop LP. Train, K. E. 2007. Discrete Choice Models with Simula on. New York: Cambridge University Press. Tsegabirhan, W. 2004. Es ma ng Willingness to Pay for Irriga on Water: A Con ngent Valua on Case Study on Small Scale Irriga on Schemes, A.A.U., FBE, Department of Economics, Ethiopia. 211 Appendix 1. Technical details of the mul nomial probit model The detailed technical specifica ons of the mul nomial probit model are discussed below. We can say ‘Y’ denotes a random variable taking on the values {0; 1; ...; J} for J a posi ve integer, and let ‘X’ denote a set of condi oning explanatory variables ( NxK vector including a constant). In our irriga on water market case, ‘Y’ would be {j=0: crop share; j=1: fixed charge and j=2: two part tariff} and ‘X’ would be commitment, supervision, good rela on, age, educa on level, rice area, number of tubewell, farm size, etc. It is also assumed that the variables are independent and iden cally distributed (i.i.d.). Here we are interested in how ceteris paribus changes in the elements of x a er the response probabili es, P(y=j|x), j=0, 1, ..., J. Since the probabili es must sum to unity (1), P(y=0|x) will be determined once we know the probabili es for j=1, ..., J. It is assumed here on the mul nomial model as a series of binary models. That is, the probability of the alterna ve j against alterna ve i for every i≠j is evaluated. The model started by considering the binary model, which is as follows: We can derive from the above equa on: Pj=F(Xβj )(Pi+Pj) It needs to be men oned here that: P So we can get the following by using the expression for i / from above: Pi For Pj : 212 To find an explicit form for Pi we only have to subs tute the G(.) by exp(.), and then we obtain the mul nomial probit/logit model: We know that the response probabili es must be sum to 1, we set the probability of the reference response (j=0) to the following: The mul nomial probit (MNP)/mul nomial logit (MNL) is obtained through maximum-likelihood es ma on which we can express as: Mc Fadden (1974) has shown that the log-likelihood func on is globally concave, which makes the maximiza on problem straigh orward. The par al effects for this model are complicated. For con nuous Xk, we can express the following equa on: Where βik is the k-th element of βi even the direc on of the effect is not en rely revealed by βik. Deriving probit/logit based models from the random u lity model where we consider the basic u lity equa on: Uij=Vij+ εij It is known that the logit model is obtained by assuming that each εij is distributed iid extreme value. This distribu on is also called the Gumbel distribu on or a Type-I extreme value distribu on. The density for each unobserved component of u lity is: ƒ(εij) = exp- εij exp(-exp- εij) This is shown in Figure 1. Type-1 Extreme Value PDF 0.4 0.3 0.2 0.1 0 -4 -3 -2 -1 0 X 1 2 3 4 Fig. 1. Type-1 Extreme Value PDF Source: Train 2007 213 The cumula ve distribu on is like: F(εij) = exp(-exp- εij) Type-1 Extreme Value CDF 1 0.75 0.5 0.25 0 -4 -3 -2 -1 0 X 1 2 3 4 Fig.2. Type-1 Extreme value CDF Source: Train 2007 There are 3 types of extreme value distribu ons: i. Type-1, (Gaumbel-type distribu on) where Pr|X≤ x = exp[-e(x-μ)/σ] where μ, σ (>0) are parameters; ii. Type-2 (Frechet-type distribu on); and iii. Type-3 (Weibull-type distribu on) The term “extreme value" is a ached to these distribu ons because they can be obtained as limi ng distribu ons of the greatest value among n independent random variables each having the same con nuous distribu on. By replacing X by −X, limi ng distribu ons of least values are obtained (Kotz and Nadarajah 2000). The variance of the distribu on is π2/σ=1.64. By assuming this, we have normalized the u lity scale. The difference between two extreme value distribu ons is logis c. It is known that the coefficients in logit will now be larger than in a probit model by a factor of √1.64. The probit coefficients can be converted to the scale of the logit coefficients by mul plying them by √1.64 214 Sec on 4 Homestead Produc on Systems 215 Do homestead food produc on systems hold promise for household food security? Empirical evidence from the southwest coastal zone of Bangladesh M. Karim, M.H. Ullah, K.A. Kabir and M. Phillips WorldFish, Bangladesh and Malaysia, m.karim@cgiar.org, hadayet127@gmail.com, k.kabir@cgiar.org, m.phillips@cgiar.org Abstract Aqua c-agricultural systems, comprised of field crop systems (FCS) and homestead food produc on systems (HFS), are an important source of income, food and nutri on for millions of poor people living in southwest Bangladesh. The contribu on of aqua c-agricultural systems to households is predicted to vary across agro-ecological zones. This study aims to determine the rela ve contribu on of homestead and field food produc on systems among three household categories (func onally landless, marginal, small) and agro-ecological zones across a salinity gradient (polders 30, 3-H, 3-L, and 43) in southwest Bangladesh. The results indicate that HFS contributed to the food security status of func onally landless households while the contribu on of FCS to these households was minimal. FCS contributed significantly, though, to the food security status of small and marginal households. Food security ra os associated with the HFS system did not vary largely (>0.05) across households; however the food security ra o associated with FCS in marginal and small households was significantly (<0.05) higher than func onally landless households. Among the polders, HFS had a significantly higher contribu on to household food security status in Polder-3H and Polder-3L while FCS contributed most in Polder-3L. The varia on is largely a ributed to the size of land holding and access to cul vatable land among household categories and polders. Results also suggest that access to produc ve assets, such as land is the major factor influencing food security status among household categories. HFS are very suitable for poor farmers, especially those that are func onally landless and in some instance for marginal farmers. Results also suggest that HFS has greater resilience, in some cases, compared to FCS. The study also suggests that there is scope for further intensifica on of HFS. Therefore, ensuring access to quality inputs, and knowledge and technical skills for farmers should be considered a priority to maximize the benefits of HFS for landless and marginal farmers. 1. Introduc on Adequate produc on and supply of cereals and staple food commodi es is s ll a major challenge to achieving food security in developing countries, though the provision of an adequate nutri onal quality for life func ons is ge ng increasing a en on. Food security is a mul -faceted concept, defined in almost 200 different defini ons (Smith et al. 1992). The most refined one was given by FAO (1996) as, “… a condi on when all people, at all mes, have physical and economic access to sufficient, safe and nutri ous food to meet their dietary needs and food preferences for an ac ve and healthy life”. This defini on overcomes the drawbacks of previous defini ons, which mostly used a calorie-based approach. Households are major components of broader food systems and can themselves be considered as (sub-)systems. Households can therefore be seen as the most appropriate entry point for the analysis of food security (Alinovi et al. 2009). Food security at the household level could be defined as “ the ability of the household to secure, either from its own produc on or through purchases, adequate food for mee ng the dietary needs of all members of the household” (FAO 2014). In developing countries such as Bangladesh, aqua c-agricultural systems (AAS) are an important source of income, food and nutri on for millions of people. Households dependent on these systems commonly have access to two dis nct systems: homestead food produc on systems (HFS) and field crop systems (FCS). HFS is mainly a need-oriented, self-provisioning, integrated, mul -species, economically sustainable and environmentally safe farming system around the house where the soil is enriched by homemade biological formula ons and integrated farming is undertaken (Bha acharya et al. 2013). In constrast, FCS is 216 predominantly a market-driven cropping system that o en depends on external inputs and commonly uses high-yielding and selected crop species to respond to prevailing market demand. These two systems collec vely produce a wide variety of foodstuffs, including cereals, fruits, vegetables, livestock and fish (Bha acharya et al. 2013; John 2014). Agriculture produc on from field crops has long been highlighted as the major source of household income and livelihood (Hossain et al. 2007); the importance of homestead farming has o en been ignored. The United Na ons General Assembly declared the year 2014 as the Interna onal Year of Family Farming, recognizing the importance of this system. In addi on to providing an income, HFS may contribute to improving food security and the nutri onal status of homestead members, and conserving biodiversity. That’s why in recent mes interven ons have been targeted at improving homestead food produc on, most of which have resulted in households increasing their consump on of a greater array of micronutrient rich ingredients (Talukder et al. 2000; Helen Keller Interna onal 2002; Galhena et al. 2013). Southwest Bangladesh encompasses Khulna and Barisal division, with a total popula on of 24 million and an overall density of 660 people/km2 (BBS 2011). The land eleva on rarely exceeds 3 meters (UN 2010) and salinity plays a central role in crop agriculture produc on (Rabbani et al. 2013b). Cropping is generally limited to the monsoon season when rainwater can feed the crops. However, the areas with par cularly close proximity to the ocean are o en inundated for long periods as drainage conges on causes persistent waterlogging of land and can hinder field crop produc on (Roy 2004). This situa on also opens up opportuni es for culturing aqua c species such as the giant freshwater prawn, Macrobrachium rosenbergii, freshwater finfish species and black ger shrimp, Penaeus monodon, the la er in more saline regions. Crop rota onal (fish-paddy or fish-shrimp) systems in ghers43 is normally prac ced in this region (Roy et al. 2013). During the rainy season (June-December) the gher is full of water, and shrimp and fish are produced. From January to May, this low-lying land tends to dry out and crops such as MV boro rice are grown on the mainland while shrimp and fish shelter in the canals. The coastal popula on is exposed to a number of climate-induced hazards such as varia ons in temperature, erra c rainfall pa erns, drought, cyclones and storm surges, flooding, salinity intrusion and rising sea levels (Rabbani et al. 2013a). The landscape is strongly influenced by tributaries of the Ganges River flowing towards the Bay of Bengal. Due to the low-lying nature of land, proximity to the ocean and frequent natural disasters such as tropical cyclones and accompanying storm surges o en bring various hazards to people living in the region. The increasing trend in sea level rise, water logging and salinity intrusion may be further intensified and exacerbated by climate change, which could significantly impact exis ng water resources and larger areas of agriculture land in the future (Shamsuddoha and Chowdhury 2007; Rabbani et al. 2013a). Agricultural ac vi es as well as cropping intensi es in the 31 upazillas of southwest Bangladesh have been reported to have already changed; where farmers can no longer grow mul ple crops in a year the livelihood and food security of poor households has been nega vely affected (Huq and Rabbani 2011; Rabbani et al. 2013a). Moreover, agricultural land (crop agriculture) showed an annual decline of about 0.27 percent from 1976-77 to 2010-11 (SRDI 2013). This trend will further diminish the access of poor people to this natural resource. Poor and marginalized households in par cular need op ons for alterna ve, locally-driven, immediate and longer-term adapta on strategies that can support household food security. HFS can bring a diverse array of plant (vegetables) and animal sourced foods (such as goats, ca le and poultry) to poor and marginalized communi es and support improved household food security status. In recent years several studies have been done on household food security and nutri onal aspects (Faridi and Wadood 2010; Alam et al. 2011; Uddin 2012; Kashem et al. 2013). However, studies that specifically address the rela ve contribu on of HFS and FCS to household food security status in the coastal area of Bangladesh have not been conducted. This research hypothesized that HFS can significantly contribute to improving the household food security status of coastal popula ons and could be considered a more eco-friendly and resilient system compared to FCS, in the context of dynamic geographical se ngs. It specifically aimed to assess the following ques ons: a) what is the rela ve contribu on of HFS and FCS to household food security?; b) does HFS significantly improve household food security status?; and c) is HFS an environmentally sound and resilient system in the context of the coastal area? 43 Ghers are modified paddy fields where elevated dikes surround the main land. A canal that usually takes up approximately 20% of the land is usually constructed around the periphery. 217 2. Materials and Methods 2.1 Study area This study was conducted in three polders: Polder-30 in Ba aghata Upazila44 (Khulna District), Polder-3 in Debhata Upazila and Kaliganj Upazila (Satkhira District) and Polder-43 in Amtali Upazila (Barguna District) (Fig. 1). These polders were selected based on their salinity levels, where Polder-3 was considered a medium to high saline zone (0-28 ppt), Polder-30 represents a low to medium saline area (0-20) and Polder-43 is considered a low saline area (0-6). The areas of Polder-30, Polder-3, Polder-43 were 7,874 hectares, 35,780 hectares, and 4,453 hectares, respec vely. Polder-3 tends to be waterlogged during the month of July to September. Polder-3 is very large and further subdivided into Polder-3H (medium-highland; more than 10 ) and Polder-3L (low-lying land; eleva on between 0 and 10 ). Coastal Zone of Bangladesh L eg en d Distri ct head quarters E xpo sed coast P older 3 P older 30 P older 43 /2F P older 43 /2F Distri ct boun dary River net work N 0 25 50 Kilometers S cale 1:2, 000,000 M ap of Ban gla desh Index Working areas B ay o f B en g al Bay of Bengal 89°0'0''E 90°0'0''E 91°0'0''E 92°0'0''E Fig. 1. Loca on of the study areas. 2.2 Data collec on and analysis The field research for this study was conducted over a two-month period between January and March 2012. A complete list of all the households in each polder was collected from the Union Parishad45 and a total of 21,851 households were recorded in the study area. Each polder, including the subdivisions of Polder-3, was considered as separate strata to perform stra fied random sampling. Sta s cal Package for Social Science, SPSS, version 16 so ware (SPSS, Chicago, IL, USA) was used to randomly select households from each strata. Sampling was conducted at a 5.86% level of interval. Ini ally, all surveyed households were categorized into five different land classes according to HIES (2010), however eventually this study considered only those households who owned land area less than 1 ha (Table 1). This is because this study focused mainly on small landholders and use of their homestead and field resources. Thus a total of 1,119 households from three types of land ownership categories, as defined by the Bangladesh Household and Income Expenditure (HIES) surveys, were sampled and surveyed from the study areas: a) func onally landless households (<0.19 ha); b) marginal households (0.2-0.6 ha); and c) small households (<0.61-1.0 ha). 44 45 Upazila is a sub-district, a smaller administra ve unit in Bangladesh formerly called thana. The smallest administra ve unit of Bangladesh. 218 A semi-structured ques onnaire was used to conduct the survey and collect informa on on land holding size, land lease or rent status, farming prac ce, income, species diversity, shocks and risk involved in different components and resource use pa erns between HFS and FCS. The HFS of this study includes five components: aquaculture, vegetable gardening, fruit gardening, poultry and livestock. The ques onnaire was translated, tested and modified prior to the survey. Ten male and one female temporary employees of WorldFish were assigned the task of visi ng each household and interviewing the household head by asking the ques ons set out in the ques onnaire. Table 1. Distribu on of households according to landholding size (% reported on household number) Land categories Func onally landless Marginal Small Total ha <0.19 0.2-0.6 >0.6-1.0 Dec (~) < 47 48-148 149-247 Polder-30 212 (63) 83 (25) 43 (13) 338 (30) Polder-3-H 148 (70) 46(22) 16(8) 210 (19) Polder-3-L 160 (64) 74 (29) 17 (7) 251 (22) Polder-43 145 (45) 112 (35) 63 (20) 320 (29) Total 665 (59) 315 (28) 139 (12) 1,119 (100) Note: Values in parentheses are percentages The data analysis was conducted using SPSS version 16.0. Descrip ve analysis (frequency distribu on, mean, standard devia on) was performed to present basic homestead and field crop resources, farming prac ces of household categories by loca on, the shocks and risk faced by the households, and integra on and recycling pa erns of resources between systems. Comparisons on income and food security ra o of different household categories and loca ons were made by ANOVA F-test and a two-tailed P<0.05 indicated sta s cally significant differences unless otherwise stated in the text. The level of household food security was computed using quan ta ve methods developed by Bala and Hossain (2010). This method was developed based on the food security concept of USDA (2007). Using this concept, Bala and Hossain defined food security as follows: Food security = [(food available from different sources and also equivalence food from different sources – food requirement)]/food requirement (1) According to this concept, all food aid commodi es were converted into grain equivalents based on calorie content. Yusuf and Islam (2005) however proposed that a dietary composi on for balanced nutri on in Bangladesh should be 2,345 kcal/cap. The proposed 2,345 kcal is equivalent to 1.357 kg of rice based on price. Thus, Bala and Hossain converted all food aid commodi es into grain equivalents based on economic returns (price) in order to compute food security, and these grain equivalents are termed equivalent foods (Majumder et al 2012). Therefore, following Bala and Hossain and Majumder et al. (2012), food security for this study was computed as follows: Household food security from homestead system = [(food available from aquaculture and equivalent food from income of aquaculture + food available from livestock and equivalent food from income of livestock + food available from poultry and equivalent food from income of poultry + food available from vegetable and equivalent food from income of vegetable + food available from fruit and equivalent food from income of fruit = total food requirement)]/total food requirement (2) Household food security from FCS =[(food available from paddy and equivalent food from income of paddy + food available from fish and equivalent food from income of fish + food available from vegetable and equivalent food from income of vegetable = total food requirement)]/total food requirement (3) The beta (β) diversity index, based on presence-absence (0/1) data, was measured to illustrate the compara ve diversity of homestead and field farming systems. The eight measures of beta (β) diversity 219 available and described in Koleff et al. (2003) were computed using PAST (PAleontological Sta s cs) so ware (Hammer et al. 2001). These two system varied in their structure and components. Hence, to measure β diversity index, this study only consider components or species that are hor cultural crops/species. This allowed for a more realis c comparison between the two systems. 3. Results 3.1 Characterizing land categories and func onal groups About 60% of the surveyed households belong to the func onally landless category, followed by small (28.15%) and marginal (12.42%) households (Table 1). Polder-3H had a rela vely high numbers (70%) of func onally landless households, while the lowest numbers of these households was recorded in Polder-43. In contrast, among the marginal households, Polder-43 had the rela vely highest (35%) number. Likewise in the marginal household category, Polder-43 had rela vely higher numbers of small households, while the lowest was recorded in Polder-3L. This suggests Polder-43 had rela vely more resource-rich households. On average, households owned two and a half mes more non-homestead land (field) than homestead land. This difference is greater for the marginal household (almost three mes) and small household (five mes) categories. This difference was most pronounced in Polder-30 where households owned three mes more non-homestead land than homestead land. The mean area of homestead land ranged from 0.03±0.03 (func onally landless; Polder-3-H) to 0.17±0.21 (small; Polder-43) (Table 2). The highest mean land area of non-homestead land was 0.67±0.16 (small; Polder-3L) while the lowest was 0.01±0.03 (func onally landless; Polder-3-H and Polder-3-L). Nevertheless, there was a great deal of varia on in landholding size within household categories and polders. Table 2. Mean homestead land and field crop land area (ha) by household category and loca on Loca on Func onally Landless HL FCL Marginal HL FCL Small HL FCL Overall mean HL FCL Polder-30 0.04 ±0.30 0.02 ±0.04 0.09 ±0.70 0.27 ±0.12 0.09 ±0.70 0.71 ±0.12 0.06 ±0.06 0.17 ±0.24 Polder-3-H 0.03 ±0.03 0.01 ±0.03 0.08 ±0.07 0.29 ±0.11 0.17 ±0.21 0.57 ±0.23 0.05 ±0.08 0.12 ±0.19 Polder-3-L 0.04 ±0.03 0.01 ±0.03 0.11 ±0.11 0.25 ±0.13 0.11 ±0.09 0.67 ±0.16 0.06 ±0.08 0.13 ±0.20 Polder-43 0.04 ±0.04 0.03 ±0.04 0.12 ±0.09 0.27 ±0.14 0.16 ±0.11 0.62 ±0.15 0.09 ±0.09 0.23 ±0.25 Overall mean 0.04a ±0.03 0.02a ±0.04 0.10b ±0.09 0.27b ±0.13 0.13c ±0.12 0.65c ±0.16 0.07 ±0.08 0.17 ±0.23 Note: HL= homestead land, FCL=field crop land. Mean values followed by different superscript le ers indicate significant difference (P < 0.05) based on ANOVA (test was done for HL and FCL separately among the household categories). On average, within homestead land aquaculture ponds occupied approximately 33% of the land, followed by fruit gardens (17%) and dwelling houses (15%) (Fig. 2a). Approximately 36% of the homestead land area of Polder-43 was covered by aquaculture pond followed by Polder-3L (32%). On average, on non-homestead land, crop land was the predominant land cover type (78%) followed by rice-fish plots (6%) (Fig. 2b). Approximately 90% of the non-homestead land area of Polder-43 was covered by crop land, followed by Polder-30 (83%). 220 a) Polder-30 Polder-3-H Polder-3-L Polder-43 0 40 80 120 Yard vegetable garden Tree covered area Poultry Pond Livestock shade Fruit garden/trees 20 60 100 140 Percent Polder-30 b) Polder-3-H Polder-3-L Polder-43 Rice-fish plot Pond (outside homestead) Forest land Fallow Crop land 0 50 100 150 200 250 300 Percent Fig. 2. Frequency distribu on of land use by loca on for a) HFS and b) FCS. 3.2 Household income and food security Most (69.72%) of the func onally landless households depended on HFS for food supply, while 30.28% depended on FCS (Table 3). In contrast, most small households relied on FCS (65.11%) for their food supply. The marginal households category was also more likely to depend on FCS (57.32%) for their food supply. The small households category obtained the highest mean income from HFS while the households of the func onally landless category earnedthe lowest (Table 3). Income obtained from HFS did not vary significantly (>0.05) among the household categories. The income of small households from FCS was, however, significantly higher (<0.05) compared to the other categories. The income of marginal households from FCS was also significantly higher (<0.05) compared to the func onally landless category. The compara vely higher income of small households from both systems allowed them to purchase a greater amount of food staples equivalent of rice (ton). Like income from HFS, the higher food security ra o was also obtained for the small household category, though the ra o did not vary significantly (>0.05) among the categories. In contrast, the food security ra o of small households obtained from FCS was significantly (<0.05) higher than the other two categories. A very high and posi ve (598.83%) food security status, obtained from FCS, was also observed for the small households category while the lowest was recorded for the func onally landless category (123.2%). 221 The mean household income obtained from HFS was significantly (<0.05) higher in Polder-3L and Polder-3H compared to Polder-30 and Polder-43 (Table 4). In contrast, the mean household income obtained from FCS was found to be significantly (<0.05) higher in Polder-3L compared to other household categories. Like household income, the household food security ra o was found to be significantly higher in Polder-3L and Polder-3H compared to other categories. The household food security ra o obtained from FCS was found to be significantly (<0.05) higher in Polder-3L. The household food security status obtained from HFS was found to be posi ve and varied from 219.3% (Polder-30) to 334.22% (Polder-3L). The household food security status obtained from FCS was much higher (370.88%) in Polder-3L, while the lowest (193.69%) was recorded in Polder-43. HFS contributed greatly to the food supply of Polder-3L households while FCS was a greater contributor to the food supply of Polder-30 households. Table 3. The rela ve contribu on of HFS and FCS to food security status by household category Variables Func onally landless (N=665) HFS FCS Income from Homestead Produc on (US$) 698.57 Equivalent to rice (ton) [EC] 1.87 a a a Food security ra o [FSR= EC-RR/RR] 2.78 Food security status (%) [FSR*100] 277.85 HFS 412.65 Required rice (ton/year)[RR] 0.5 Marginal (N=315) a Small (N=139) FCS 694.15 a 888.33 HFS b FCS 727.09 a 1291.99 c 1.11 a 1.86 a 2.38 b 1.95 a 3.46 c 0.5 0.5 0.5 0.5 0.5 1.23 a 2.75 a 3.8 b 2.93 a 5.99 c 123.2 275.46 380.49 293.27 598.83 *Contribu on in HS food (%) 69.72 30.28 42.68 57.32 34.89 65.11 Contribu on in food requirement (%) [EC/RR*100] 223.2 375.46 480.49 393.27 698.83 377.85 Note: *Food from homestead/total food*100 Table 4. The contribu on of HFS and FCS to household food security status by loca on Variables Polder-30 (N=338) Polder--3-H (N=210) Polder-3-L (N=251) Polder-43 (N=320) HFS HFS FCS HFS FCS HFS FCS FCS Income from Homestead Produc on (US$) 592.23 a 625.7a 798.13 b 624.63a 805.37 b 873.37b 677.72a 544.72a Equivalent to rice (ton) [EC] 1.58a 1.67a 2.13b 1.67a 2.15b 2.33b 1.81a 1.46a 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Required rice (ton/year)[RR] 0.5 Food security ra o [FSR= EC-RR/RR] 2.19a 2.37a 3.34b 2.37a 3.30b 3.71b 2.65a 1.94a Food security status (%) [FSR*100] 219.3 237.35 330.31 236.77 334.22 370.88 265.4 193.69 *Contribu on in HS food (%) 53.26 46.74 63.45 36.55 57.59 42.41 58.99 41.01 Contribu on in food requirement (%) [EC/RR*100] 337.35 430.31 336.77 434.22 470.88 365.4 293.69 319.3 Note: *Food from fieldcrop/total food*100. Numbers in parentheses are standard error mean. Mean values followed by different superscript le ers indicate significant difference (P < 0.05) based on ANOVA. 222 Approximately 11% and 52% of func onally landless households depended on HFS and FCS, respec vely, and were food insecure when they solely considered HFS or FCS as their source of food supply (Table 5). This trend is, however, opposite for the small household category where approximately 38% and 9% of small households that depend on HFS and FCS, respec vely, were food insecure. However, when households depended on both HFS and FCS for their food supply, they tended to be less food insecure (1%). Table 5. Households with posi ve and nega ve food security status by household category (% reported on household number) HFS HH types Nega ve FS Posi ve FS Func onally landless 11 89 FCS Nega ve FS Posi ve FS 52 48 Overall Nega ve FS Posi ve FS 2 98 Marginal Small Overall mean 13 9 36 1 0 1 29 38 20 71 62 80 87 91 64 99 100 99 Note: Nega ve FS= nega ve food security status, Posi ve FS= posi ve food security status 3.3 Defining two systems in the context of diversity and resilience 3.3.1 Diversity The higher values of β diversity index in FCS indicates lower similarity among the varie es in different polders (Table 6). By contrast, lower values of β diversity index in HFS indicate higher similarity among the varie es in different polders. There were more than 70 and 40 hor cultural crops/varie es in HFS and FCS, respec vely. Table 6. Global beta diversity index of homestead and FCS Index Whi aker Harrison Cody HFS 1.660 0.550 25.000 FCS 0.314 0.100 24.000 Routledge Wilson-Shmida Mourelle Harrison2 0.260 1.580 0.530 0.070 0.080 0.450 0.150 0.060 Williams 0.166 0.150 3.3.2 Shocks and risk ranking Approximately 55% of households reported diseases as the major risk to HFS, followed by epidemic (27%) and poaching and accident (8.12%) (Table 7). Among the HFS components, poultry was reported to be highly affected by diseases (33.57%) and also epidemics (30%). Aquaculture, one of the major components of HFS, was mostly reported to be vulnerable to flooding. Overall, among the HFS components, vegetable produc on, fruit gardening and aquaculture seemed to be less risky while poultry was considered highly vulnerable. 223 Table 7. Shocks and risk for different components of HFS (% reported on household number; N=554) Rank Shocks and risks Aquaculture Fruit/ tree Livestock Poultry Vegetable Overall mean 1 2 3 4 Disease Epidemic Poaching and accident Disaster 0.54 0.00 0.00 0.00 0.00 0.00 0.00 0.18 20.76 1.99 2.71 1.08 33.57 25.99 5.42 2.35 0.00 0.00 0.00 1.08 54.87 27.98 8.12 4.69 5 6 Flooding Others (Poor water quality etc.) 3.25 1.08 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.25 1.08 Overall mean 4.87 0.18 26.53 67.33 1.08 100.00 Approximately 47% of households reported diseases in rice, due to pest a ack, as a major problem in FCS (Table 8). Flooding (28%) also caused serious problems in FCS. Diseases in fish (fish culture with paddy) were also reported to nega vely affect FCS. Table 8. Shocks and risk for FCS (% reported on household number; N=554) Rank Shocks and risk Percent 1 2 3 4 Diseases outbreak in rice due to pest infesta on Flooding Fish disease Poisoning problem in rice field 46.67 28.10 20.00 1.43 5 6 7 8 Rice damaged due to saline water intrusion Pest a ack in vegetables Fish died due to saline water intrusion Irriga on problem 1.43 0.95 0.48 0.48 9 Vegetable field damaged by excess water 0.48 3.3.3 Environment, integra on and recycling of resources The different level of integra on and recycling of resources in HFS and FCS can be described as inter- and intra-rela onships of these systems, respec vely (Fig. 3). The HFS system worked more as an inter-system where integra on and resource flows were observed among five components. In HFS, 57.51 % of households used pond water for poultry produc on. Similarly, 50.08% of households used pond water for vegetable produc on. Approximately 21.8% of households used poultry and/or livestock manure to produce fruits and vegetables. In addi on, approximately 11% and 8.39% of households fer lized aquaculture ponds for fish produc on through livestock manure and poultry li er, respec vely. Homestead aquaculture ponds were found to be a major source of resource flows among HFS components. In contrast, FCS acted more like an intra-system where rice was the major contributor to integra on and resource flow (Fig. 3). Approximately 44% of households reported that they used straw as fer lizer to fer lize the same field for the next cropping cycle. Likewise, vegetables (mostly cabbage) were also used as fer lizer to fer lize fields. Eighteen percent of households used the same water to produce both rice and fish. Livestock are not a component of FCS, however, they had so much influence on FCS and were very closely linked with rice produc on and thus shown as an overlapping component within FCS. 224 A Water, 50.05% HH Water, 2.1% HH Aquaculture in Homestead Pond Pond Bo om Soil , 3.55% HH Water, 57.51% Vegetabel Byproducts 3.07% HH Manure 11% HH Livestock Produc on Vegetabel Produc on Homestead Produc on Sustem Li ers 8.39% Poultry Produc on Poultry and Livest ock Manure, 21.8% HH Fruit Produc on B Fish Culture wih Paddy Manure, 68 % HH Livestock Water, 18% HH Field crop System Rice Cul va on Straw as far lizer 44% HH Vegetable Produc on Far lizer [cabbage] 7% HH Straw, 71% HH Fig. 3. Integra on and recycling of resources among different components of HFS and FCS. 225 4. Discussion 4.1 HFS for household categories: who benefited most? HFS showed strong poten al to contribute to household food security, especially for func onally landless households in the southwest coast of Bangladesh. The food security ra o did not vary greatly among the household categories in HFS, whilst the food security ra o of small and marginal households was significantly higher than func onally landless households in FCS. This implies that func onally landless households, in general, are likely to explore and u lize more of their homestead resources. Households with smaller land holdings, such as func onally landless households, tend to be less reliant on FCS which is probably due more to the fact that they don’t have sufficient access to field cropland, than to their willingness and ability. Land is a cri cal asset, especially for the rural poor as a means of livelihood through the consump on and sale of crops and other products (Meinzen-Dick et al. 2007). Func onally landless households with limited access to land may seek employment opportuni es from other farmers or non-farm sources; however the growth and stability of such employment is also dependent on the produc on and profitability of homestead farming (Meinzen-Dick et al. 2007). Thus, resource-poor farmers o en tend to depend more on homestead produc on for their food staples and secondary staples than those endowed with a fair amount of assets and resources such as land and capital (Wiersum 2006; Galhena et al. 2013). This is strongly supported by the findings of this study since almost 70% of the food staples of func onally landless groups came from HFS. In recent years, a rapid increase in par cipa on of women, par cularly from resource-poor households, in homestead farming was observed (BBS 2013; Khatun et al. 2014). This was most probably also one of the major reasons of higher dependency of poorer households on HFS. Marginal and small households were however more dependent and focused on FCS even though they owned sufficient homestead land. This is also evident from the rela vely higher (more than 65%) contribu on of FCS to the total household food supply of small households. Marginal households also received a compara vely larger share of total household food supply from FCS. This indicates that marginal and small households are not much interested in bringing more homestead land under cul va on. They seem to invest more in cropland. Rice is the largest and most profitable crop prac ced within FCS and requires addi onal inputs such as high yielding varie es of seeds, chemical fer lizers, irrigated water, pes cides and labor (Quddus 2009). Households with compara vely larger land holdings and higher income, mostly mariginal and small, were thus more likely to focus on FCS. In contrast, HFS requires minimal inputs and labor, most of which are managed by household members and thus mainly contributes to mee ng farmers’ basic needs (John 2014). This in fact indicates that HFS is more suitable for resource-poor or func onally landless households in helping to meet household food security and producing a marketable surplus for the purchase of non-producible items. If resource-rich marginal and small households are involved in more HFS, they can increase their food security status as well as contribute to a steady supply of products for the local market (Meinzen-Dick et al 2007). Thus, HFS can posi vely contribute to food security from the household to the na onal level. 4.2 Does loca on influence the rela ve contribu on of HFS and FCS to household food security? The households of Polder-3H and Polder-3L earned significantly higher than Polder-30 and Polder-43 from HFS and thus became more food secure. This is most probably a ributed to the suitable agroecologial se ngs that offer be er farming prac ce opportuni es among the studied areas. Households in Polder-3H and Polder-3L were also likely ready to explore op mum opportuni es from both the systems. Livestock, vegetable and fruit gardening were reported to be most preferable and profitable homestead farming components, mainly for the subsistence farmers of Bangladesh (Sarma and Ahmed. 2011; Khatun et al, 2014). The higher land eleva on of Polder-3H might be helpful in producing vegetables and fruits even during periods of waterlogging (July to September) in this area. In contrast, households may also use water logging areas for seasonal fish culture (Das et al 2009) and thus can contribute to improving their household food security. The largest share of household food supply in Polder-3H (≈63%) and Polder-3L (≈58%) also came from HFS. This also indicated that resource poor households (≈59% in this study), mostly func onally landless and partly marginal households irrespec ve of loca ons, in the coastal area of Bangladesh predominantly depend on 226 HFS for mee ng a greater por on of their basic food staples. Salinity has been reported to exert a nega ve effect on the farming households and agro-diversity in the coastal zone of Bangladesh (Rahman et al 2011; Rabbani et al 2013b). The higher household food security ra os obtained from HFS however was reported from Polder-3H and Polder-3L which had the highest salinity levels among the study areas. Thus, in fact the results of this study did not demonstrate any nega ve effects of salinity on HFS, which implies a certain resilience to salinity changes in the studied areas. The household food security ra o from FCS did not vary significantly among the polders except Polder-3L. The higher food security ra o in the households of Polder-3L was perhaps due to the higher yield from aman and boro crop. Aman is a rain-fed rice crop (Basak, 2011; WFP, 2014) and covers about 70% of the total rice cropped area in the coastal zone of Bangladesh (Mia and Islam 2005). Polder-3L, as a low lying area, has the highest capacity to store rainwater and thus significantly supports aman crop (Basak, 2011 ). In addi on, Boro rice crop is completely dependent on irriga on (WFP, 2014). Polder-3L is expected to have higher ground water recharge capacity, consequently higher ground water level, and also situated adjacent to the Ichama River thus providing easy access to irriga on. Thus the significantly higher food security ra o in the households of Polder-3L implies that availability and access to freshwater resources, minimizing the nega ve impact of salinity intrusion, may be a crucial factor for FCS. Therefore this study argues that salinity, availability of freshwater and land eleva on may influence the rela ve contribu on of food security both in HFS and FCS. 4.3 Are HFS resilient and environmental sound systems in the context of coastal areas? The agro-ecology of coastal Bangladesh is highly dynamic and significantly influenced by different factors such as changing land use pa erns (Mia and Islam 2005), salinity (Roy 2004; Rahman et al. 2011; Roy et al. 2013), and climate induced hazards including varia ons in temperature, erra c rainfall pa erns, drought, cyclones and storm surges, flooding, and rising sea levels (Shamsuddoha and Chowdhury 2007; Rabbani et al. 2013a ; Rabbani et al. 2013b; SRDI 2013). The resilience of a system may depend on its capacity to maintain func oning, structure, and feedback in the face of disturbance (Folke et al. 2004). The lower similarity, with lower number of vari es, among the hor culture varie es of FCS indicates that these species are not equally capable of adap ng to the dynamic ecological se ngs of the different polders. In contrast, the higher similarity among the hor culture species (more than 70) of HFS indicates that most of the species were present in all four polders and have compara vely higher resilience capability to adapt to different, adverse and dynamic environmental condi ons. This is also supported by previous findings that species in HFS are less suscep ble to environmental factors such as salinity. Addi onally, aquaculture, livestock and poultry components of HFS comprise approximately 25 fish species, four and three animal species, respec vely. A recent study done by Bha acharya et al. (2013) also reported more than 40 local varie es of hor cultural crops and 25 fish species in HFS in West Bengal, India. The higher number of species in HFS also reveals this system’s capacity to consistently support diversified hor cultural crops and may be very helpful against confron ng climate change. A system’s ecological resilience and stability may also focus on a system’s response to shocks and long-term change (Folke et al. 2004; Leslie and Kinzig 2009). This study iden fied six different types of shocks and risks in HFS. The problems in FCS are categorized under nine different key points. Both systems were considerably threatened by diseases. However, among the components in HFS, poultry is severely affected by diseases. The most common diseases are IBD, Aflatoxicosis, Chicken Anemia virus, Egg Drop Syndrome and Avian influenza (bird flu). Despite facing diseases, poultry farming has recently experienced tremendous development in Bangladesh due to its role in poverty allevia on and economic development (Uddin et al 2010). In contrast, flooding stands as a noteworthy factor that can cause huge damage mostly to FCS. Flooding is a large-scale natural disaster and can unpredictably damage huge amounts of crops as it is very rarely possible to protect cropland (Khan et al. 2012). Within HFS, livestock, vegetables and fruit gardening are found to make a larger contribu on to household income. Bha acharya et al. (2013) reported that household income in HFS was comprised of aquaculture (54%) followed by animal husbandry (37%) and hor culture (9%). Since poultry is a risky agribusiness, households have the op on to choose and invest in other HFS components. This will s ll 227 allow them to earn sufficient money to improve their household food security status. Thus, HFS allows year round cul va on of different agricultural products (hor culture, aquaculture, etc.) that can provide either a source of addi onal income to the households or func on as an alterna ve livelihood (Bha acharya et al. 2013). Intelligent management of available resources of a farming system through integra on and recycling is important for op mum u liza on of land, improving food security, be er income genera on, as well as nutri onal security of small and marginal farmers (Jayanthi et al. 2009; Dadabhau and Kisan 2013). The higher level of integra on and recycling of resources in HFS shows that this system has be er poten al to integrate and recycle most of its resources. The components of this system worked more in an inter-rela onship where pond water is considered to be the most important driver of the system. All of the five components of HFS, including aquaculture, somehow depend on pond water during their produc on cycles. This occurs through several ways, such as for washing, making foods, drinking purposes, and irriga ng plants and vegtables, in the form of nutrient supply and finally to produce fish from the pond. The diversity of resource flow is also much higher in HFS than FCS in the form of water, manure, vegetable byproducts and pond bo om soil. The most notable advantage of u lizing these byproducts within the system is that this increases the biological capacity of the system and increases system produc vity. Jayanthi et al. (2009) ar culated that integra on and u liza on of low-cost/no-cost material at the farm level of recycling reduces produc on costs and consequently improves produc vity and income considerably. However, the lower level of integra on in FCS compared to HFS implies that households might explore the poten al of further intensifica on of HFS for increasing produc on and income. This study nevertheless confirms that there are some limita ons, although it has yielded some interes ng and preliminary findings on household food security. The food security model that was used does not consider nutri onal aspects of the households. However, this study assumes that if households produce sufficient food and have money to buy food staples they are likely to maintain the dietary composi on for balanced nutri on proposed by Yusuf and Islam (2005). Addi onally, HFS and FCS are two dis nct systems. Therefore the system characteris cs of these two systems should differ from each other and, in some cases, may not be comparable. The compara ve overview was not specifically done to draw any conclusions about the be er system, but rather to depict some advantages and disadvantages that might be useful for par cular groups. 5. Conclusion HFS has several advantages such as available family labor, mul -crop and diversified species cul va on, use of household refuses, resource flow among the components, low cost or available input and, most importantly, owning a small pice of homestead land. HFS proved to be very suitable for poor farmers, especially those that are func onally landless and in some instances for marginal farmers. In Bangladesh, func onally landless households own <0.19 ha of land and have long been known to depend on a combina on of various forms of agricultural tenancy, laboring and non-agricultural livelihoods (Hossain et al. 2003). HFS could allow most of these household farmers to minimize risk, and increase produc on and profits whilst improving the u liza on of organic wastes, residues and byproducts. HFS has the poten al to improve the household food security of the poor irrespec ve of their land holding size. The poten al and benefits of HFS could be much higher if other household categories, such as small households, also become interested in HFS. This will provide sufficient quan es of food staples at local markets at affordable prices, and will consequently help poorer household meet their household food requirements. However, intensifica on of HFS might be hampered due to an inadequate supply of inputs and adop on of improper management prac ces. Therefore, available supply of quality inputs must be ensured and farmers’ knowledge and technical skills should also be improved through capacity building programs in the form of awareness programs, trainings and workshops. 228 References Alam M.A., M.A. Rahman, M.S. Flora, M.R. Karim, M.P.I. Sharif and A. Ahmad. 2011. Household Food Security and Nutri onal Status of Rural Elderly. Bangladesh Medical Journal. 40(3): 8-11. Alinovi, L., E. Mane and D. Romano. 2009. Measuring household resilience to food insecurity: Applica on to Pales nian households. Food and Agriculture Organiza on, Rome. Bala, B. K. and M.A. Hossain. 2010. Food security and ecological footprint of the coastal zone of Bangladesh. Environment, Development and Sustainability. 12(4): 531-545. Basak, J.K. 2011. Changing Rainfall Pa ern Effects on Water Requirement of T.Aman Cul va on in Bangladesh. Public Journal of Environmental Science. 11(1). Bha acharya, A., A.D. Deo, J.K. Sundaray and A.G. Ponniah. 2013. Impact of homestead farming system on rural economy: A study in South 24 Pargana district of West Bengal. Presented at Interna onal Ecosystem Conserva on, Climate Change and Sustainable Development Conference, held in Kerala, India on 3-5 October 2013. Abstract, 1 p. Dadabhau, A.S. and W.S. Kisan. 2013. Sustainable Rural Livelihood Security Through Integrated Farming Systems – A Review. Agri. Reviews. 34(3): 207-215. Das N.G., M.S. Hossainand M.S. Islam. 2009. Waterlogged Area as New Horizon for Aquaculture Development: A Golden Dream to the Rural Communi es of Begumgonj, Noakhali. Bangladesh Journal of Marine Sciences and Fisheries. 1(1): 47-62. FAO. 2014. Household food security and community nutri on. FAO Agriculture and Consumer Protec on Department. h p://www.fao.org/ag/agn/nutri on/household_en.stm Faridi, R. and S.N. Wadood. 2010. An Econometric Assessment of Household Food Security in Bangladesh. The Bangladesh Development Studies. 33(3): 97-111. Galhena, D., R. Freed and K. Maredia. 2013. Home Gardens: A Promising Approach to Enhance Household Food Security and Wellbeing. Agriculture & Food Security. 2(1): 2-8. Giraldo, D. P., M. J. Betancur and S. Arango. 2008. Food security in developing countries: A systemic perspec ve. Paper presented at the Interna onal Conference of the System Dynamics Society held in Athens, Greece on 20-24 July 2008. Hammer, Ø., D.A.T. Harper and P. D. Ryan. 2001. PAST: Paleontological Sta s cs So ware Packagefor Educa on and Data Analysis. Palaeontologia Electronica. 4(1): 9pp. Hossain, M., D. Lewis and S. Begum. 2007. Rice Research, Technological Progress and Poverty: The Bangladesh Case. Pp. 56-102. In Agricultural Research, Livelihoods and Poverty. Eds M. Adato and R. Meinzen-Dick. Bal more, MD: Johns Hopkins University Press. Huq, S. and G. Rabbani. 2011. Climate Change and Bangladesh: Policy and Ins tu onal Development to reduce vulnerability. JBS. 13: 1–10. Jayanthi, C., C. Vennila, K. Nalini and B. Chandrasekaran. 2009. Sustainable integrated management of crop with allied enterprises: Ensuring livelihood security of small and marginal farmers. The Asia Pacific Tech Monitor. 26(1): 21-27. John, J. 2014. Homestead Farming in Kerala: A Mul -Faceted Land-Use System. Review of Agrarian Studies. 4(1): 80-94. Kashem M.A., M.G. Farouque and P.C. Roy. 2013. Impact of Crop Varie es on Household Food Security in Haor Areas of Bangladesh. Int. J. Agril. Res. Innov. & Tech. 3(2): 7-11. 229 Khan M.N.H., M.Y. Mia and M.R. Hossain. 2012. Impacts of Flood on Crop Produc on in Haor Areas of Two Upazillas in Kishoregonj. J. Environ. Sci. & Natural Resources. 5(1): 193-198. Khatun M.F., M.K. Hossain, K.A. Hossain and M.M. Rahman. 2014. Par cipa on Poten al of Rural Women in Different Homestead Farm Ac vi es. The Agriculturists. 12(1): 57-67. Majumder S., B.K. Bala and M.A. Hossain. 2012. Food Security of the Hill Tracts of Chi agong in Bangladesh. Journal of Natural Resources Policy Research. 4(1): 43-60. Meinzen-Dick, R., P. Kameri-Mbote and H. Markelova. 2007. Property Rights for Poverty Reduc on. 2020 Focus Brief on the World’s Poor and Hungry People. Brief. Interna onal Food Policy Research Ins tute, Washington DC. Quddus, M.A. 2009. Crop produc on growth in different agro-ecological zones of Bangladesh. Journal of Agicultural University. 7(2): 351-360. Rabbani, G., S.H. Rahman and L. Faulkner. 2013a. Impacts of Clima c Hazards on the Small Wetland Ecosystems (ponds): Evidence from Some Selected Areas of Coastal Bangladesh. Sustainability 2013. 5(4): 1510–1521. Rabbani, G., A. Rahman and K. Mainuddin. 2013b. Salinity-induced loss and damage to farming households in coastal Bangladesh. Int. J. Global Warming. 5(4): 400–415. Rahman M.H., T. Lund and I. Bryceson. 2011. Salinity impacts on agro biodiversity in three coastal, rural villages of Bangladesh. Ocean and Coastal Management. 54(6): 455–468. Sarma P.K. and J.U. Ahmed. 2011. An economic study of small scale ca le fa ening enterprise of Rajbari district. J. Bangladesh Agri. Univ. 9(1): 141–146. Shamsuddoha, M. and R.K. Chowdhury. 2007. Climate Change Impact and Disaster Vulnerabili esin the Coastal Areas of Bangladesh. COAST Trust, Dhaka. Soemarwoto, O. 1987. Homegardens: a tradi onal agroforestry system with a promising future. In Agroforestry: A Decade of Development. Eds. H. Steppler and P.K.R. Nair. Nairobi, Kenya: Interna onal Council for Research in Agroforestry. SRDI. 2013. Trends in the availability of agricultural land in Bangladesh. Technical Report. Soil Resource Development Ins tute, Ministry of Agriculture, Bangladesh. Uddin, M.B., S.S.U. Ahmed, M.M. Hassan, S.A. Khan and M.A. Mamun. 2010. Prevalence of poultry diseases at Narsingdi, Bangladesh. Int. J. BioRes. 1(6): 9-13. Uddin, M.E. 2012. Household Food Security Status of Marginal Farmers in Selected Storm Surge Prone Coastal Area of Bangladesh. The Agriculturists. 10(1): 98-103. WFP (World Food Program). 2014. Agricultural Produc on & Na onal Food Balance. Accessed on 7th December 2014. Available at h p://www.foodsecurityatlas.org/bgd/country/availability/agricultural-produc on#sec on-12. Wiersum, K.F. 2006. Diversity and change in homegarden cul va on in Indonesia. In Tropical Homegardens: A Time-Tested Example of Sustainable Agroforestry. Vol 3. Ed. Nair BMKPKR. Dordrecht, The Netherlands: Springer Science. 230 Homestead farming: A biodiverse system to enhance resilience to climate vulnerability J.K. Sundaray1* , A. Bha acharya2, A.G. Ponniah3, T.K. Ghoshal2, A.D. Deo4, J.P. Sharma2 and M. Phillips5 1* Central Ins tute of Freshwater Aquaculture, India, jsundaray@gmail.com 2 Kakdwip Research Centre, CIBA, India 3 Central Ins tute of Brackishwater Aquaculture, India 4 Central Ins tute of Fisheries Educa on, India 5 WorldFish, Malaysia Abstract Climate change will have a profound impact on agriculture and aquaculture in India due to the inunda on of land with saline water and the widespread freshwater shortages that will impact vulnerable rural communi es. Against this backdrop, homestead farming, a resource-rich farming system adjacent to the homestead, is examined as a mechanism to support rural livelihoods and provide resilience against the expected increase in environmental and economic turbulences due to climate change. A study was conducted on 480 households, selected from Kakdwip, Namkhana and Sagar blocks in West Bengal, India to assess the biodiversity in homestead farming systems (HFS) and the poten al for mi ga ng the impact of climate change on rural livelihoods using HFS. A total of 42 crop varie es, 38 aquaculture species and five animal species were recorded in these small (average size of 0.1 ha), complex farming systems including coconut, mango, potato, fish, ca le and poultry. Seventy-six percent of the crops grown were food items (vegetable and fruits), followed by spices (12%), medicinal plants (12%), mber (10%) and fodder (5%). Ninety-two percent of the aquaculture produce was fish and the remaining 8% was shrimp/prawns. By incorpora ng a biodiverse assemblage, a synergy is promoted that enhances farm processes such as the ac va on of soil fer lity through u liza on of organic wastes. This can also result in a 39% increase in income above the na onal poverty line average. Through con nuous produc on and u liza on throughout the year, this farming system helps to mi gate the risks of dependency on a single crop that in some years is lost due to extreme clima c events like heavy rain, cyclones and disease. It also supports the dietary diversity of households and contributes 6%, 24% and 26% of total required energy, protein and fat, respec vely, resul ng in a reasonable degree of nutri onal wellbeing of the household members. The present study has shown that inves ng in this system can help to mi gate the impact of climate variability. Keywords: homestead farming, biodiversity, polyculture, climate change, resilience 1. Introduc on Climate change is a change in the state of the climate that can be iden fied by changes in the variability of its proper es, and that persists for an extended period, typically decades or longer (IPCC 2007). Vulnerability to climate change is the degree to which geophysical, biological and socio-economic systems are suscep ble to and unable to cope with the adverse impacts of climate change (Fussel and Klein 2006). Climate change is projected to intensify the challenges already faced by smallholder farmers globally. Changes in rainfall, rising temperatures and varia ons in soil carbon u liza on by crops are expected to nega vely influence the growing condi ons and the poten al yields of many crops (Amikuzuno and Hathie 2013). Although the effect of changes in climate on crop yields varies greatly from region to region (Cline 2007) the decline in produc on worsens food insecurity and poverty both globally and locally (Al eri and Kooha an 2008). In many countries including India, agriculture is largely rain-fed and farmers commonly cul vate local crop varie es and local livestock and fish breeds with li le resilience to the immediate effects of climate variability such as drought, flood and large temperature fluctua on. Conversely, some local varie es are well adapted to local condi ons and cul va ng a range of local and improved strains of produce can enhance resilience. The severe cyclonic 231 storm, Aila, which hit the Sundarban region in India and Bangladesh in 2009, resulted in extreme devasta on to the agricultural sector (CSE 2012). It is es mated that climate variability will decrease cereal produc vity by 10 to 40% by 2100 and a 10C increase in temperature will reduce wheat produc on by 4 to 5 million tons (Biswal 2008). Climate change adapta on strategies are now a ma er of urgency (FAO 2009). Farming households need to u lize innova ve prac ces, experien al knowledge and locally available resources in order to be self-reliant. In many areas, par cularly in the developing world, farmers prac ce some kind of farming system adapted to the local condi ons that enables them to generate sustainable yields to meet their subsistence needs, despite marginal land endowments, clima c variability and low use of external inputs (Wilken 1987; Denevan 1995). Part of this success is linked to the high levels of biodiversity exhibited by tradi onal farming systems, which in turn posi vely influence agro-ecosystem func on (Vandermeer 2002). Diversifica on is an important farm strategy for managing produc on risk in small farming systems. In tradi onal farming systems the prevalence of complex and diversified cropping systems is of key importance to the stability of peasant farming systems, allowing crops to reach acceptable produc vity levels even under trying environmental condi ons. One of the most suitable coping mechanisms and strategies used by small and marginal holders to enhance resiliency against clima c anomalies is Homestead Farming Systems (HFS). HFS is a dynamic system, containing mul ple species in an ecologically sustainable environment and is a good example of diversified produc on of hor culture, aquaculture and animal husbandry (Bha acharya et al 2013b). Raised beds that do not get water- or saline-logged and the cul va on of local or improved varie es that are well adapted to the microclimate around the homestead provides resilience to HFS. 2. Materials and methods 2.1 Study area To assess the u liza on of HFS resources and the poten al for mi ga ng the impact of climate change on rural livelihoods, a household survey was conducted in three coastal blocks in the Sundarban region of South 24 Pargana District of West Bengal, India during 2011 and 2012. The selected blocks were Kakdwip (21°52’06”N, 88°11’12”E), Namkhana (21°46’00”N, 88°14’00”E) and Sagar (21°38’09”N, 88°07’26”E), all located in the southern extremity of the district. Sca ered through this area are the distributaries of the Ganges River, which are fed by sea des. The average temperature of the district varies from a maximum of around 38°C to a minimum of around 14°C. The annual average rainfall is 1800 cm, more than 75% of which falls during the monsoon season. This area is regularly hit by the bay cyclones (District Disaster Management Plan 2012). 2.2 Sampling procedure and data collec on A random sampling method was adopted to draw representa ve samples of 480 households (HH) from 18 villages in each block. Twenty households were considered from each village (in Sagar, 40 households from each village). Three complementary approaches were adopted for the study: (a) formal interviews with the village-level government body to collect secondary data; (b) direct observa on by project associates/authors of this paper; and (c) a pre-structured ques onnaire conducted with the head or members of the selected households to collect primary informa on. In addi on, some exis ng literature regarding the challenges faced by the households was also reviewed. 232 West Bengal, India South 24 Parganas Kakdwip Namkhana Sagar 2 Gram Panchayat (G.P) 2 Gram Panchayat (G.P) 2 Gram Panchayat (G.P) 3 villages from each G.P 3 villages from each G.P 3 villages from each G.P 20 HH from each village 20 HH from each village 40 HH from each village 480 households (HH) Fig. 1. Sampling frame. 2.3 Data processing and analysis The data from the completed ques onnaires were sta s cally analyzed with Sta s cal Packages and MS-Excel sta s cal program for both quan ta ve and descrip ve explana ons. 3. Results and discussion 3.1 Impact of saliniza on in the agricultural sector Among the studied households almost 10% were facing extreme salinity intrusion that resulted in some human health impacts, reduc on in crop quality and widespread freshwater shortages. The low average produc vity of 2.3 ton per ha in homestead produc on systems in this area is likely due to salinity intrusion (District Sta s cal Handbook 2009). In some areas, salinity has increased beyond the safe threshold for agriculture (CSE 2012), especially for paddy produc on, which is an integral crop for food security and income. The limit for paddy produc on is 4-6 PPT, but in some of the study sites salinity reached 10 PPT. Salinity intrusion is par cularly prominent in the Sagar block where salinity ranged from 5-25 PPT. Salinity penetrated to a depth of ~1.5 m in the study area, severely affec ng paddy cul va on (Anon 2011). As a result farmers were switching from the tradi onal prac ce of paddy cul va on to growing produce in HFS as a climate change adapta on strategy. 3.2 Impact of climate change on the livelihoods of people in the study area In the study site only 43% of households were engaged in natural resource-based ac vi es (such as paddy cul va on, hor culture, wage labor in agricultural fields, wood and honey collec on, fish and shrimp seed collec on) as their major source of income (Table 1). Reports in 2009 demonstrated that agriculture was the primary livelihood for approximately 60% of the popula on (Anon 2009). This drop in the number of people relying on agriculture for their primary livelihood may have been a ributed to a drop in the produc vity of the land due to climate variability or the shortage of land access that resulted in 32% of the rural popula on shi ing to non-farm casual labor (e.g. construc on) as their primary occupa on. Since the employment opportuni es in agriculture are decreasing, seasonal migra on of the earning member of the family has become the standard prac ce of this popula on. For rural households, moving away from agricultural ac vi es can further exacerbate food insecurity by reducing access to fresh and nutri ous produce. This leaves households par cularly vulnerable, especially when facing the risks from climate variability (CSE 2012). Ninety-nine households in the study region were small and marginal farmers. Farmers in this category are the most disadvantaged and are par cularly vulnerable to climate anomalies (Al eri and Kooha an 2008). Severe 233 soil erosion and the pressure of a growing popula on have reduced agricultural land by 21% from 2001 to 2009 (Hazra et al. 2010). The share of households falling below poverty line (BPL) is close to 35%, 48% and 44% in Kakdwip, Namkhana and Sagar blocks, respec vely (CSE 2012), due to a reduc on in livelihood opportuni es for households in the study region. This is a direct indicator of vulnerability of people and highlights the inability of the area to sustain and support the livelihoods of people in the area (Anon 2009). Substan al investment of resources is necessary to widen the livelihood opportuni es for people in the study area with homestead farming systems being one area with poten al for investment. Table 1. Occupa onal distribu on of the households Source of income % share of households Agriculture and all natural resource based ac vi es Non-farm casual labour Business Van/Rickshaw pulling 43 32 10 5 Salaried employee in govt. sector Others 5 5 3.3 Characteris cs of the homestead farm in the study area The homestead farm is a unit where fresh water (rainwater) is conserved in the homestead pond to provide access to fresh water for a range of household ac vi es throughout the year. The homestead farm size ranged from 0.03 ha to 0.5 ha with a mean of 0.1 ha in this region and aquaculture comprised the largest area (Table 2). The HFS is a biodiverse micro-environment that is an integral source of food, fodder, fuel, medicines, spices, construc on materials and income in many countries around the world. HFS is a dynamic, resource-rich system where the composi on and diversity of species are influenced by changes in the socioeconomic circumstances and cultural values of the households that maintain these farms (Eyzaguirre and Watson 2002). The primary objec ve of homestead farming is subsistence produc on and income genera on (Trinh et al. 2003). The surveyed households cul vate a combina on of species with different maturity periods to ensure an uninterrupted supply of produce for their own consump on (Garί 2003) and also for the sale of surplus products to enhance income (High and Shackleton 2000; Murphy 2008; Shackleton et al. 2008). Table 2. Land sharing among various HFS components Components Aquaculture Share of total homestead land 50 Hor culture/agro-forestry Animal husbandry 37 13 3.4 Poten al contribu on of homestead ponds to mi ga ng clima c loss Homestead pond water is not only the key resource in HFS, but it is also an important source of fresh water in the event of inunda on of surrounding land with saline water due to climate varia on. Each component of HFS is directly or indirectly linked with homestead pond water, with 94% of households using it to support hor culture and agroforestry and 77% of households using it for animal rearing (Table 3). 234 Table 3. U liza on of homestead pond water Par culars Hor culture/agroforestry % of households involved 94 Domes c ac vi es Animal rearing 93 77 The effec ve u liza on of homestead pond water by a huge propor on of households helps to mi gate the damage done by saline water intrusion and therefore the pond plays a key role in influencing the economics of HFS which can return up to US$ 3,169 per ha (gross) (Table 4). Table 4. Contribu on of different components in gross return from HFS Components Aquaculture Gross return (US$ per ha land) 1,711 Hor culture Animal husbandry Total 275 1,183 3,169 3.5 Adapta on strategies to climate change: Species composi on in HFS Systems with higher crop diversity and mixed farming (fish-crop-livestock) are thought to be more resilient to the risks of clima c change (FAO 2011). The households in this study displayed a diversified species mix with 31 key species of vegetables, crops, aquaculture and livestock. The diversified species in homestead farms demonstrates that the vegetable crops, medicinal and fodder or firewood plants cul vated in the homestead farms were convenient (Table 5) and enhanced access to food, provided treatments for common ailments, fed domes c animals, and were used as fuel and cash income, respec vely (Akrofi 2012). Along with this, the aquaculture species are the primary source of animal protein. Although not yet common, the introduc on of Piaractus brachypomus, pacu, as an aquaculture species is thought to improve the resilience of the system. 235 Table 5. Frequency of key species in HFS Components Crops (Hor culture/ Agro-forestry) Aquaculture Animal husbandry Scien fic name Common name Frequency Economic usage Abelmoschus esculentus Luffa acutangula Capsicum annum Solanum lycopersicum Solanum tuberosum Allium cepa Cucurbita maxima Colocasia esculenta Ladies finger Ribbed gourd Chili Tomato Potato Onion Pumpkin Arum 40 45 33 45 116 41 57 70 F F F, S F F F F F Solanum melongena Mangifera indica Musa paradisiaca Cocos nucifera Curcuma longa Azadirachta indica Piper betle Brinjal Mango Banana Coconut Turmeric Neem Betel vine 110 134 100 342 17 15 105 F F, FW, FO, T F F, FW, T S M, T C Bambusa tulda Labeo rohita Catla catla Cirrhinus mrigala Pun nus japonicus Labeo bata Oreochromis nilo cus Hypophthalmichthys molitrix Bamboo Rohu Catla Mrigal Japani Pun Bata Tilapia Silver Carp 81 421 420 328 292 118 78 148 T F F F F F F F Mystus gulio Pangasius hypophthalmus Amblypharyngodon mola Piaractus brachypomus Penaeus monodon Macrobrachium rosenbergii Gallus gallus domes cus Tangra Pangus Mola Pacu Tiger shrimp Scampi Hen 13 19 9 13 9 14 32 F F F F F F F Bos indicus/Capra aegagrus Ca le(Cow/Goat) 92 F Note: F- Food item, S- Spices, FW- Firewood, FO- Fodder, T- Timber, C- Cash crop 236 Table 6. Types of varie es with their percentage of occurrence Components Crops (Hor culture/ agro-forestry) Aquaculture Animal husbandry Economic products Food items (vegetable and fruit) Spices Timber Medicinal Fodder/Firewood Others (Ornamental, cash crops etc.) Fish % of total species of each component 76 12 10 12 5 5 92 Shrimp/Prawn Poultry products Livestock products (meat) 8 40 60 A total of 42 crop species belonging to different botanical families were recorded across the surveyed homestead farms. Of the hor cultural crops, 37 species (88%) were used as food-related crops (food items such as vegetable, fruit and spices) and 13 (32%) were non-food crop species (medicinal, mber, firewood or fodder). Among the species 76% were used as food items (vegetable and fruits) followed by spices (12%), medicinal plants (12%), mber (10%), fodder or firewood (5%) and others (5%) such as ornamental and perishable crops (Table 6). Apart from these crops, 38 aquaculture species and five animal species were also encountered in these small (0.1 ha), dynamic systems. Among the aquaculture species, finfish species dominated with 92% of homestead culturing fish in ponds while shrimp/prawns comprised only 8% (Table 5). The average biodiversity index in HFS is high 0.74 (where 1 is the maximum) (Table 7). Table 7. Biodiversity index in an 1 ha area of HFS Average no. of species/variety (a) Average no. of individuals (b) Biodiversity index (b*a) 20 27 0.74 3.6 Impact of assembly of biodiversity in rural economies Biodiversity in all components (e.g. aquaculture, hor culture/agro-forestry, animal husbandry) increases the resilience in changing clima c condi ons and under both bio c and abio c stress. Biologically or gene cally-diverse popula ons and species-rich ecosystems have greater poten al to adapt to climate change (FAO 2007) and systems with many short dura on crops that provide benefits within a few months help to mi gate the risk of a crop being damaged by a natural disaster. The prac ce of cul va ng or culturing indigenous and locally-adapted plants and animals in HFS that are highly resistant to adverse condi ons is also important. Further, the integra on of resources between terrestrial and aqua c subsystems creates a synergy that promotes and enhances beneficial processes such as the ac va on of soil fer lity to improve the produc vity of the system. It was found that the households that integrated more components in their HFS realized 27% higher income than the other households that did not draw on synergies between different components (Bha acharya et al 2013b). This can result in a 39% increase in income above the na onal poverty line average (Table 8). Table 8. Enhancement of household economy by HFS Per Capita Income of BPL households (rural) (Rs/month) (a) 13.61 1 2 Average household size (rural) (b) 4.72 ≈ 5 Total Income of BPL households (Rs/annum) (a*b) 816 Income a er prac cing HFS (Rs/annum) 1133 Income increase 39% Planning Commission 2013 (Converted into US$) NSSO 2010 237 3.7 Seasonality in HFS and the impact on food and nutri onal security Food security in most subsistence communi es in some regions of the world is influenced by seasonality (Savy et al. 2005). In areas where agricultural produc on is mainly rainfed and highly labor-dependent the effect of seasonality on food security is par cularly pronounced (Masanjala 2006; Savy et al. 2005). Rural households engaging in agricultural ac vi es o en face food insecurity in the dry season when reserve food starts to decline (Hesselberg and Yaro 2006). The failure of a single crop due to the extreme clima c events like heavy rain, cyclones, large temperature fluctua on or disease can increase the vulnerability of poor households at certain mes of the year. However, HFS can prevent acute food insecurity including over the dry season. The con nuous produc on and u liza on (73% consump on and 27% marke ng as surplus product) throughout the year makes this system valuable for mi ga ng the risks of dependency on a single food source. Figure 2 demonstrates the important contribu on of HFS to dietary energy, protein and fat requirements. Further, the high species diversity (Table 5) also ensures high dietary diversity and a range of micronutrients. 30 25 Animal husbandry 20 15 Hor culture/ Agro-forestry 10 5 Aquaculture 0 Energy Protein Fat Fig. 2. Contribu on of HFS to nutri onal wellbeing. 4. Way forward for climate variability and mi ga on measures 4.1 Diversifica on of crops The study highlighted that HFS are highly diverse, which can enhance resilience to climate change. The introduc on of more climate- and salt-resilient species including Piaractus brachypomus (pacu), Musa paradisiaca (banana) and Cocos nucifera (coconut) is likely a good op on for adap ng to climate change (Bagamba et al. 2012). 4.2 Culture of brackishwater species in homestead ponds Homestead ponds are regularly subjected to saliniza on making it advisable to culture brackish water species including Scatophagus argus (spo ed scat), Etroplus suratensis (pearl spot), Mystus gulio (Tangra). 4.3 Enhance the prac ce of betel vine Betel vine is a cash crop that acts as a buffer against economic loss because it has a short cropping cycle and can generate US$ 2,500 per ha per year. 4.4 Technical support for improved management prac ces Long-term training and technical support is required to enhance farmers’ knowledge of climate variability and different techniques for adap ng to climate variability. 238 5. Conclusion and recommenda ons HFS, the tradi onal farming prac ce involving local knowledge and techniques adopted by local people, remains the dominant climate change adapta on strategy for poor rural households (Al eri and Kooha an 2008). HFS enhance resilience and help to meet farmers’ subsistence needs by delivering diverse and nutri ous food and addi onal income. This predominantly man-made ecosystem offers an integrated approach to long-term and sustainable development outcomes. Further studies should use indicators to quan fy the impact of effec vely managed homestead farming systems on nutri onal status and wellbeing of household members. Despite being an important element of rural landscapes and playing a pivotal role in the predominantly bioresource-based rural economy, HFS has not received a en on in any policies or programs by the government. Therefore, it is recommended that HFS be included in government policies to enhance the livelihoods of the rural poor by mi ga ng clima c impact. Acknowledgements The authors gratefully acknowledge the CGIAR Challenge Program on Water and Food (CPWF), World Fish and ICAR, New Delhi for providing financial support. References Al eri, M. A. and Kooha an, P. 2008. Extent of tradi onal and family farming systems. Enduring Farms: Climate Change, Smallholders and Tradi onal Farming Communi es. Penang, Malaysia: Third World Network (TWN). 17-19. Amikuzuno, J. and Hathie, I. 2013. Climate Change Implica ons for Smallholder Agriculture and Adapta on in the White Volta Basin of the Upper East Region of Ghana. Paper presented at Impact World 2013, Interna onal Conference on Climate Change Effects, May, 27-30, in Potsdam. Anon. 2009. District Human Development Report, South 24 Parganas. Development and Planning Department, Government of West Bengal. Anon. 2011. Post-Aila salinity defeats Sunderbans farmers. The Hindu. June 8, h p://www.thehindu.com/scitech/agriculture/ar cle2087079.ece. Akrofi, S. 2012. Home Garden: a poten al strategy for food and nutri on security in HIV households: A case study in rural Ghana, PhD thesis. Wageningen, the Netherlands., Wageningen University. Bagamba, F., Bashaasha, B., Claessens, L. and Antle, J. (2012). African Crop Science Journal, Vol. 20. pp. 303 - 316 Bha acharya, A., Sundaray, J. K., Deo. A. D., Ponniah, A. G. 2013b. Resource recycling in homestead farming: An innova on towards the inclusive development in Sunderban of West Bengal. Paper presented at the Na onal Seminar on Innova ons in Science and Technology for inclusive development. Indian Science Congress Associa on (Bhubaneswar Chapter). 23-24 Nov, 2013. Bhubaneswar. Biswal, B. (2008). Climate Change: Impacts on Indian Agriculture. h p://www.slideshare.net/ soumyashree85/global-climate-change-its-impact-on-indian-agriculture-presenta on Cline, W. R. 2007. Global Warming and Agriculture: Impact Es mates by Country. Washington, DC: Center for Global Development. CSE 2012. Living with changing climate. Impact, vulnerability and adapta on challenges in Indian Sundarbans. Centre for Science and Environment. New Delhi. 1-108. Denevan, W.M. 1995. Prehistoric agricultural methods as models for sustainability. Adv. Plant Pathology 11. 21-43. District disaster management plan. 2012 South 24 Parganas: h p://s24pgs.gov.in/pdf/...up/Final_ Disaster%20Management_Plan_2012.pdf. 239 District Sta s cal handbook. 2009: h p://www.s24pgs.gov.in/s24p/DSHB_S24PGS_2009.pdf. Eyzaguirre, P. B and Watson, J.W. 2002. Home gardens and agrobiodiversity: an overview across regions. Paper presented at the Second Interna onal Home Gardens Workshop: Contribu on of home gardens to in situ conserva on of plant gene c resources in farming systems, 17–19 July 2001, Witzenhausen. Federal Republic of Germany. Interna onal Plant Gene c Resources Ins tute, Rome. FAO. 2007. Adapta on to climate change in agriculture, forestry and fisheries: Perspec ve, framework and priori es. Inter-departmental working group on climate change. Rome, Italy: FAO. FAO. 2009. Coping with a changing climate: considera ons for adapta on and mi ga on in agriculture. Environment and Natural Resources Management Series. No. 15. Rome: FAO. Fussel, H. M. and Klein, R.J.T. 2006. Climate change vulnerability assessments: an evolu on of conceptual thinking. Clima c Change, 75, 301-329. Garí, J. A. 2003. Agrobiodiversity strategies to combat food insecurity and HIV/AIDS impact in rural Africa. Advancing grassroots responses for nutri on, health and sustainable livelihoods. Popula on and Development Service, FAO, Rome, Italy (preliminary edi on). Government of India Ministry of Foreign Affairs. 2011. Census of India 2011. h p://censusindia.gov.in Hazra S, Samanta K, Mukhopadhya A and Akhand A. 2010. Temporal Change Detec on (2001-2008) Study of Sundarban. School of Oceanographic Studies, Jadavpur University. Kolkata Hesselberg, J. E. & J. A. Yaro. 2006. An assessment of the extent and causes of food insecurity in northern Ghana using a livelihood vulnerability Framework. Geo Journal 67. 41–55. High, C. & C. M. Shackleton. 2000. The compara ve value of wild and domes c plants in home gardens of a South African rural village. Agroforestry Systems, 48: 141-156. IPCC. 2007. Summary for Policymakers. In Climate Change (2007) Impacts, Adapta on and Vulnerability. Contribu on of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palu kof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK, 7-22. Masanjala, W. H. 2006. HIV/AIDS, Household income and consump on dynamics in Malawi. In S. Gillespie (ed.), AIDS Poverty and Hunger: Challenges and Responses Highlights of the Interna onal Conference on HIV/AIDS and Food and Nutri on security, Durban South Africa. April-14-16 2005 pp. 56-73. Interna onal Food Policy Research Ins tute, Washington D.C. Murphy, L. L. 2008. AIDS and kitchen gardens: insights from a village in Western Kenya. Popula on Environment, 29: 133–161. Savy, M., Y. Mar n-Prevel, P.S. Sawadogo, Y. Kameli & F. Delpeuch. 2005. Use of variety/diversity scores for diet quality measurement: rela on with nutri onal status of women in a rural area in Burkina Faso. European Journal of Clinical Nutri on, 59. 703-716. Shackleton, C. M, F. Paumgarten & M. L. Cocks. 2008. Household a ributes promote diversity of tree holdings in rural areas, South Africa. Agroforestry Systems, 72: 221–230. Sundarban Development Board. 2010-11 Trinh, L. N., J. W. Watson, N. N. Hue, N. N. De, N. V. Minh, P. Chu, B. R. Sthapit & Eyzaguirre, P. B. 2003. Agrobiodiversity conserva on and development in Vietnamese home gardens. Agriculture, Ecosystems and Environment, 97. 317–344. Vandermeer, J. (ed.) 2002. Tropical Agroecosystems. Boca Raton: CRC Press. Wilken, G.C. 1987. Good Farmers: Tradi onal Agricultural Resource Management in Mexico and Guatemala. Berkeley : University of California Press. 240 Homestead produc on systems in Sundarbans region of West Bengal, India – Current status and opportuni es S. Mandal, D. Burman, S. K. Sarangi, B. K. Bandyopadhyay and B. Maji 1 Central Soil Salinity Research Ins tute, Regional Research Sta on, Canning Town, India, subhasis2006@gmail.com, burman.d@gmail.com, sksarangicanning@gmail.com, bimalbkb@gmail.com, b.maji57@gmail.com Abstract Homestead produc on systems (HPS) are an integral part of the daily household ac vi es and produce food (fruits, vegetables, fish and livestock) for household consump on in coastal areas of West Bengal. They contribute significantly towards mee ng daily food and nutri on requirements, generate income when surplus produce is grown and can therefore help to mi gate price or output shocks due to unforeseen events and can help to reduce poverty. A survey was conducted in the Sundarbans region of West Bengal, India to understand the current status of HPS and opportuni es to improve HPS and livelihoods of rural communi es. Aquaculture in homestead ponds (average area 445 m2) and homestead gardens (vegetables and fruits; average area 120-240 m2) were the two key components of the system. The ponds were mostly perennial but some held a limited amount of water during non-monsoon months. On average, 70 to 75% of the total vegetables produced (average total of 340 kg/household) in the HPS were consumed by the households (HH) and this accounted for 30 to 40% of the total household requirement. Part of the harvest (25 to 30%) was marketed every one to two days. Similarly, around 30 to 35% of the fish produced (143 kg/household) in the HPS was consumed by the farm family. Around 50 to 60% of the total fish produced in the HPS were sold. In the case of livestock produc on was minimal, resul ng in 80 to 85% of produc on being consumed by the households. Nevertheless this provided almost 50% of the households’ needs, with a small amount le over to sell at the local markets (CSSRI 2014). Produc vity of HPS could be improved and these systems could provide greater contribu on to the goal of regional food security. To achieve this, farmers need assured supply of quality inputs and training on produc on and management of the main components: fish, vegetables and livestock. Since HPS are a ended to by all members of the household, training programs should focus on the whole family, including men, women and children (12 to 18 years of age). In addi on to technical training farmers need financial advice, and improved support groups and technical resources. Enhancing the produc on level would increase the quan ty of marketable surplus and thereby increase the contribu on of HPS to regional food and nutri on security. Key Message: Homstead produc on systems make substan al contribu ons towards mee ng daily food and nutri on needs, income, employment, mi ga ng price/output shocks due to unforeseen events and figh ng poverty. Keywords: homestead, food security, poten al, resource constraints 1. Introduc on Homestead gardening has the poten al to improve the nutri onal status of household members, but tradi onal gardening prac ces need to be developed so that households produce a wide variety of vegetables and fruits throughout the year (Torlesse et al. 2004). The homestead area in India was defined as the dwelling house together with the courtyard, compound, garden, outhouse, place of worship, family graveyard, guesthouse, shop, workshop, offices for running household enterprises, tanks, wells, latrines, drains and boundary walls annexed to the dwelling house. Some mes gardens, orchards or planta ons, though adjacent to the homestead and lying within the boundary walls, may be located on a clearly dis nct piece of land. In such cases, gardens, orchards or planta ons were not considered as homestead land (Govt. of India 2006). 241 Although there has been no systema c study on homestead produc on systems (HPS) the majority of the households (HH) in the study area (coastal salt affected areas of Sundarban) have some kind of HPS adjacent to their dwelling house, irrespec ve of the land holding size. In coastal areas of West Bengal the opera onal farm holding size is very small (<0.5 ha per HH) and these plots may be further fragmented as families expand. The poor farming communi es are poverty stricken, having very low investment capaci es, and land produc vity is very low (primarily mono-cropped, rice-based farming systems with 2.0 t/ha in wet season) due to acute shortage of irriga on water in non-monsoon months. Therefore, the HPS systems are important for improving livelihoods and a aining household level food security in this region. Access to good quality water is extremely limited in the coastal areas (par cularly during non-monsoon period) and farmers depend primarily on rainwater harvested in monsoon months from their pond(s). Therefore, the homestead pond is an integral component of HPS in the coastal area and each component of the HPS is linked with the availability of water in these ponds. With training interven ons these resources could be used more efficiently and the produc vity of the whole HPS can be enhanced significantly. Families were able to grow 12 different species and improved the homesteads’ food intake by 4 kg of vegetables per week. Homestead plots can provide a ready source of food where it is needed most: in the households of the rural poor (Nielsen et al. 2006). Homestead produc on contributed to reduced prevalence of anemia (63.9 percent) among children in program households in Bangladesh (Talukdar et al. 2010). It is believed that HPS can be u lized more intensively and can poten ally contribute more significantly to rural community livelihoods, although very li le work has been done previously on HPS in West Bengal to prove this. In the study site almost every household possesses a single or mul ple ponds with most being 35 to 40 years old and with very li le management (e.g. de-sil ng, trea ng soil or water). All the economic (at subsistence scale) ac vi es of HPS depend on and revolve around the u liza on of the water resources from the pond. Rearing fish in these ponds is very common but mostly to fulfill the households’ requirement, without depending on any external inputs such as feed or fer lizer. There has not been any significant change in input use or management prac ces, par cularly for homestead aquaculture, despite the knowledge being widely available amongst the scien fic community and extension workers. Besides fisheries, the homestead pond is used for growing vegetable crops, fruit trees, poultry and some mes irriga on of paddy fields. Although the volume of produc on is low, informa on on marketable surplus of these quan es may explore new opportuni es and be er linkage with markets. The system is complex and variable with a range of components and resource recycling pathways connec ng them. There is very li le planning around the methods for HPS so quan fying the inputs and outputs is very difficult. The current study used primary surveys to provide detailed informa on on produc on status, poten al and constraints to enhancing produc on capaci es. Key issues dealt with in this study were: What is the current status of HPS, including socio-economic pa erns of the farmers, resources under the system and produc on pa erns? Quan fying the benefits and iden fying ways to improve the produc on capaci es of the system Exploring possible market linkages with the homestead products and other opportuni es The present study on HPS provides detailed informa on on produc on status, poten al and constraints to enhancing produc on capaci es. Based on this, policy makers can be made be er aware of the grassroot level happenings and challenges, and produce appropriate future strategies for the region. 2. Methods 2.1 Sources of data and sampling design The primary survey was conducted in 2012 and 2013. Detailed socio-economic informa on was collected through a pre-structured and tested survey specifically designed for the homestead produc on system. In addi on, some secondary informa on was collected and used in this report, for example, from the District Sta s cal Handbook 2007, Govt. of West Bengal. 242 The Sundarbans region of coastal West Bengal is comprised of 19 blocks under two districts; South 24 Parganas (13 blocks) and North 24 Pargans (6 blocks). The Sundarbans region is a coastal, salt-affected landscape. The current study focused on North 24 Pargans. ICAR-Central Soil Salinity Research Ins tute, Regional Research Sta on Canning Town carried out the survey. In North 24 Parganas there are six blocks, Haroa, Hasnabad, Hingalganj, Minakhan, Sandeshkhali I and Sandeshkhali II. Out of these, the village level HPS survey was carried out in two blocks: Sandeshkhali I and Sandeshkhali II. These two blocks were selected based on available secondary informa on (Bandyopadhyay et al. 2003), extent of salinity of this region and a er discussion with the scien sts, Assistant Director of Agriculture (ADA), Govt. of West Bengal, working in the fields of agriculture and fisheries, and with local farmers. The mul stage stra fied random sampling design was followed to select the, Gram Panchayats (elected local government at village level) and villages within the Gram Panchayats for the survey. Farm households with at least one farm pond, conduc ng some form of homestead produc on were purposely selected for the survey. The survey was carried out in four Gram Panchyats: Hatgachi in Sandeshkhali I and Bermajur-I, Bermajur-II and Durgamandap in Sandeshkhali II block. The survey covered six villages, Dakhin Kanmari, Semulha , Bermajur, Jupkhali, Daudpur and Durgamandap and was conducted in a total of 240 households in the North 24 Pargana District. 2.2 Analy cal technique Descrip ve sta s cs were primarily used for analyzing the data collected through primary surveys. Farm budge ng technique was used for analyzing annual costs and returns of the enterprises or the system. Financial analysis through discounted method such as Internal Rate of Return (IRR), Benefit Cost Ra o (BCR) and Net Present Value (NPV) were employed to examine the long-term feasibility of the current HPS. These financial criteria account for the me value of money invested on the system and provide be er informa on on making decisions for long-term investments in HPS. The financial analysis was based on some assump ons such as: economic life of the pond was 15 years (although these are used for a longer period, at the 15th year the costs of reparing the ponds become very high, and also for be er accoun ng of the costs and returns); discount rate of 12% (the prevailing maximum rate of interest charged by the bank and sufficient to cover the me value of money); full benefit (first year is the planning period) of the system will be from the second year onwards; major excava on/repair a er 10 to 12 years; and key components of the systems are fish and vegetables. Under this analysis, the contribu on of family labor was included as a cost item. The criteria of the financial analysis (IRR, BCR and NPV) indicated that the system was not genera ng sufficient income for long-term investment if the contribu on of family labor is included as a cost, which normally farmers’ do not consider as a cost item. According to the standard size-classifica on (set by the Indian Planning Commission), more than 95% of the farmers in the study area are classified as marginal farmers (<1 ha of opera onal holdings). We a empted to further classify our farmers based on land holding size within the marginal group as it has been found that produc vity is related to land holding size (Chand 2011). The new classifica on was as follows: landless (0 0.2 ha), marginal 1 (0.2 – 0.4 ha), marginal 2 (0.4 – 1 ha), small (1-2 ha) and others (>2 ha). However, while analyzing the data it was observed that the income genera ng ac vi es of HPS were closely linked with the size of ponds (availability of water) rather than size of total opera onal holdings. In addi on, the ac vi es and produc on under HPS were linked with the family size, alterna ve livelihood op ons and size of ponds within the HPS but not to the total size of opera onal holdings. Under these circumstances, the data analysis was carried out without making any size-classifica on (homogeneous as almost all farmers are marginal), focusing on HPS irrespec ve of the total opera onal holdings of the farmers to avoid any unforeseen confusion over data interpreta on. This will provide useful informa on targeted at marginal farmers and facilitate the future planning of HPS ac vi es in the region. This indicates that future HPS programs should be focused on farmers irrespec ve of their land holding pa erns. 243 3. Results and discussion 3.1 Socio-economic features of homestead farmers The average age of the respondents was skewed to the age group 40 to 60 years, followed by age group 20 to 40 years (31%), 60 to 80 years (12%), over 80 years (2%) and below 20 years (2%), demonstra ng that these HPS ac vi es are mainly carried out by the middle aged popula on. The young popula on either engaged in other non-agricultural ac vi es or migrated in search of alterna ve income sources in distant places. Rural economies are poorly diversified; the vast majority of rural poor are engaged as day laborers in agriculture (72%), a seasonal occupa on that can leave households without income for several months of the year (Van et al. 2009 and Mar n et al. 2008). Average family size of the farm families was 5.23 including children. All family members including the children (below 14 years of age) par cipate in ac vi es of homestead produc on systems on a daily basis, for few minutes to a few hours. Similar par cipa on has been observed under aquaculture opera on also. Mostly the ac vi es under HPS are carried out throughout the year (in both the kharif and rabi seasons). Occupa onal pa erns of the farm households was analyzed on the basis of percentage of respondents devo ng maximum me on a par cular avoca on, as well as percentage of respondents earning maximum income from a par cular avoca on. In terms of me spent, cul va on of crops was the most dominant occupa on (34%) followed by pisciculture plus cul va on (27%), wage labor (21%) and other (18%). In terms of major sources of income, pisciculture plus cul va on provided the largest percentage of income to farm families (39%) followed by cul va on (38%), wage labor (19%) and other (4%). Farm households are engaged in cul va on ac vi es for longer periods of me, but pisciculture plus cul va on provided the major source of income. Major income sources for farmers in the study area were pisciculture, growing vegetables, wage earnings, brackishwater aquaculture, service and others. Both male and female members of the family contributed to all of these ac vi es. Cul va on was primarily dependent on rainfall; lands are mostly mono-cropped with rice in the kharif season. Farmers are engaged in crop cul va on for around 3 to 3.5 months only and earn meager amounts (Rs. 17188/HH/year). Pisciculture (freshwater) provides rela vely be er income (Rs. 23334/HH/year) compared to crop cul va on. The pisciculture ac vi es are dependent on the availability of water in ponds and are feasible for eight to nine months per year. Vegetables are grown mostly by using water from ponds and restricted to a very limited area. On average, produc on of vegetables fetches earnings of Rs 5100/HH/year but this component contributes significantly to household nutri on if nutrient-rich vegetables are grown. A study at Mymensingh, Bangladesh revealed that farmers produced on an average of 244.17 kg of vegetables year round. They consumed most of the vegetables (149.5 kg), distributed a small por on (25.67 kg) and sold a larger por on (69 kg) to meet their daily necessi es (Dey et al. 2012). Wage labor (such as work under the na onal scheme MGNREGA, other public works or in others’ fields) and migra on to other places (nearby towns or distant places) provides good income to the households (Rs 37152 and Rs 38173 per HH per year from wage earnings and migra on, respec vely). But this income stream is not always ensured and very o en fails to provide a decent livelihood op on for the farm households. Brackishwater aquaculture is also prevalent in the region and generates significant income (Rs. 53450/HH/year) but requires high investment and intensive care. Therefore, this system (brackishwater aquaculture) is operated at large scales by wealthy people from outside the region. 3.2 Resources in Homestead Produc on Systems Homestead produc on systems are comprised of several key resources including water, fish, hor cultural crops, livestocks. The pond and the water in the pond is a keystone resource of the HPS as all of the components of HPS are dependent on this water. Out of all enterprises, fisheries and growing vegetables on the dyke area are the most important components as they provide nutri on as well as income to the family. The expansion of area and intensifica on of crops under homestead ac vi es primarily dependent on the homestead ponds (size, depth, distance and availability of water throughout the year) rather the total opera onal holdings size of the farmers. Although rigours studies are not available in the study area, the 244 authors of this paper have observed that farmers’ with lower opera onal holdings are more likely to use the HPS more intensively to maximize the produc on and more dependent on this system as compared to the large farmers because large farmers tend to focus their efforts on field crops. On average, each farm household in the study had more than one pond within or outside the HPS. Some of the households had as many as five ponds. Average age of the ponds was observed to be 58 years, some of which were very old (more than 100 years) and some which were excavated recently (within five to six years). The age of around 56% of the total ponds in the sample was over 60 years, followed by < 20 years (18%), between 40 and 60 years (15%) and 20 to 40 years (11%). Average maintenance cost of ponds (renova on, de-sil ng, etc.) for every 10 to 12 years has been calculated to be Rs. 5025/-46. Sedimenta on occurs in older ponds, which reduce the capacity to hold water, in turn affec ng produc vity. Good management is required but very o en farmers were not so keen for renova ons due to financial constraints. Average size of the ponds was es mated to be 0.04 ha. Maximum pond size was observed to be 0.09 ha. Although the majority of the ponds in the study area were perennial by defini on, water availability in the ponds during dry months was very limited and not sufficient for aquaculture opera on throughout the year. Adequate water was only available in the ponds for aquaculture for eight to ten months, following which me ponds are dried and fish are harvested. This dry period may provide an opportunity remove some of the sediment from the bo om of the pond, provided households have the me and human resources to do so. Maximum water depth in the ponds was eight to nine and in some ponds the depth was as low as 3 , primarily due to lack of renova on work (de-sil ng). Shallow water depth has implica ons for fish health and survival because the temperature and therefore dissolved oxygen fluctua ons are more extreme in shallow water. Most of the ponds are located within the backyard of dwelling houses or very close to the homestead area (within 15 to 20 meters). This is because most ponds were excavated for making the uplands to construct the houses. Daily management of the land is easier when it is closer to the homestead area. Rainfall is the primary source of water in the ponds and is stored during the rainy season. In most of the HPS pond, fish are grown in a very tradi onal way without any care or scien fic management (stocking density, fish composi on or feeding). Only one cycle of fish are grown during the eight to nine month period for which the pond is opera onal each year. Periodic harves ng is done to meet household consump on needs in small quan es and a one- me harvest is done a er complete drying of the ponds. Besides aquaculture in the homestead pond, growing vegetables, fruit and mber trees on the dyke or in homestead gardens was the major ac vity under HPS. A number of vegetables were grown in the homestead gardens including brinjal, okhra (bhindi), potato, cabbage, cauliflower, pumpkin, yam, spinach, colocasia, amaranthus, cucumber, bi er gourd, beet and carrot. The area of vegetables cul va on under the HPS was 200 to 250 m2. Inputs like seeds, pes cides and human labor (mainly family labor) were key cost components in the produc on system. Input cost (excluding imputed value of human labor) and average produc on of vegetables was calculated to be Rs. 450kg/HH and 340 kg/HH, respec vely. The value of vegetables produced in the system was es mated to be Rs. 5100/HH. On average one to two human laborers (mostly family members) were engaged in this opera on for around two to three hours daily. Vegetables in the HPS are quite a profitable enterprise and many farmers undertake their cul va on intensively and with good care. Other non-aquaculture ac vi es under HPS are rearing of ca le and backyard poultry (3 to 4 no per HH), ducks (3 to 4 no per HH) and a few households were rearing sheep (garol breed suitable for the coastal region). Rearing of large ruminants like ca le or buffaloes was not so prevalent in this region. 3.3 Fisheries in homestead produc on systems A number of fish are grown in the homestead ponds as composite fish culture. These are rohu, catla, mrigal, japani pun , silver carp, tangra, vetki, lapia, mourala, prawn, pangas, golden carp, sol, koi and magur. Fish are grown in the ponds without following any scien fic management prac ces like stocking density, composi on of fish, periodic liming of ponds or providing fish feed. Few farmers have undergone training for scien fic fish 46 1 US$ = Rs. 62 (approximately) 245 rearing management and mostly follow tradi onal cul va on prac ces. However some farmers have been observed growing fish with intensive care and for earning profit. These farmers could provide a pla orm around which to build community learning groups and workshops. Since water is available normally for eight to nine months a year, mostly one cycle of fish are grown in the pond. Average size of fish seeds varied from 1.5”to 5” depending on the fish species. Many farmers preferred to use bigger fish seeds (fingerlings) to avoid risk of mortality. Average produc on of fish from the pond was es mated to be 75kg/pond/year and 143 kg/HH/year 47. The average value of fish produced was Rs. 8250/pond/year and Rs. 15730/HH/year. Average weight of fish during harves ng varies across fish species from 75 gm (e.g., prawn) to 400 to 800 gm (e.g., carps). Out of total produc on (143 kg/HH/year) from the homestead ponds, average quan ty of fish sold was es mated to be 107 kg/HH/year and the rest (36 kg) was consumed by the family members. Average consump on of fish per household per year has been es mated to be 84 kg, out of which 43% (i.e., 36 kg) was obtained from homestead produc on. With an avergage family size of 5.23 person in the study area, the per capita fish cosump on (16 kg per annum) was much higher than the Indian average (9.8 kg per annuam, Govt. of India 2011) because fish is extensively grown and consumed by the popula on in this part of India. Value of fish from homestead ponds sold in the market has been es mated to be Rs. 11770 per HH per year. 3.4 Contribu on of HPS to household food security Farm households obtain several food items from their HPS almost daily but the variety depends on seasonal availability. Primarily these are vegetables, fish, fruit and livestock and livestock products (egg, meat, milk, etc.). It was es mated that on an average 70 to 75% of vegetables produced in the HPS were consumed by the family members, which accounted for nearly 30-40% of their total requirement (Table 1). Some part of the harvest (25 to 30%) was being marketed daily or on alternate days. Similarly, around 30 to 35% of fish produced in the HPS were consumed by the farm family, accoun ng for 50 to 60% of their total household fish requirement. Almost 50 to 60% of the total fish produced in the HPS went to market. In the case of livestock, since the produc on quan ty was limited, 80 to 85% were consumed by the HH (accoun ng for almost 50% of their needs). The rest (10 to 15%) were sold to the local markets. Availability of fish, vegetables and livestock products from the HPS were quite small in quan ty, but contributed greatly to daily household requirements, thus reducing external dependence and making the farm family more self-reliant. A study conducted by Lokesh and Hanstad (2004) in Karnataka indicated that 90 to 100% of the vegetables and fruits, and 100% of the milk products consumed by the households in their study were produced on the households’ homestead plots. Table 1. Contribu on of HPS to households food security Items Vegetables Fish Fruits Livestock Contribu on (%) Home consump on of total produce Fulfilling the total requirement Marketed of total produce Average produc on (kg/HH) 70-75 30-35 85-90 80-85 30-40 50-60 50 25-30 60-65 5-10 10-15 340 143 - Source: Annual Report (2013-14) 47 Note: First average is per pond and second average is per household. Most farm households have mul ple ponds with an average of 1.90 ponds/HH. 246 As well as contribu ng to daily household nutri on requirements, HPS also played a cri cal role in mi ga ng the risk of food insecurity associated with natural disasters. There was a devasta ng cyclone (Aila) in the Sundarbans in May 2009 that caused almost all farmland to be inundated with saline water. This hindered crop produc on for more than two years. The HPS was the only area suitable for cul va on due to its high eleva on and the rela vely limited intrusion of saline water. Saline water recedes quickly from the HPS and salinity was washed away a er a few good showers and the plots under HPS became suitable for growing crops. 3.5 Economics of the homestead produc on system The homestead produc on system is complex and diverse and it is difficult to obtain reliable input-output data from the respondent farmers. Fisheries in the homestead ponds are grown in a tradi onal way without much care and scien fic management. But the crops (mainly vegetables) in the homestead garden are grown quite intensively and farmers try to maximize their output from the small area under opera on. Economic analysis of fisheries in ponds, based on opera onal annual costs and return (without the ini al investment and other fixed investment like pump, cast nets, etc.), indicated that the system was profitable. The annual opera onal expenditure components were expenditures on fish seed, fish feed, repairing of cast nets, human labor for feeding and harves ng of fish, human labor for daily supervision/intercultural opera on of vegetable produc on and harves ng, purchasing of vegetables seeds, plant protec on chemicals, medicines and miscellaneous. Annual opera onal costs and return has been calculated to be Rs 7620 and Rs. 14280 for average pond and vegetables produc on area, respec vely, in the study area. To analyze the financial feasibility of long-term investments in the system (including the ini al investment in pond excava on), discounted cash flow measures like Internal Rate of Return (IRR), Net Present Value (NPV) and Benefit-cost Ra o (BCR) were computed (Table 2). The es mated IRR was 11%, i.e., less than the discount rate of 12%. NPV was nega ve (Rs. -2816) and BCR was less than one (0.98). Under current prac ces the long-term investment in HPS was not a financially a rac ve proposi on. However, it has mul ple func ons, u lity and value for HH in the coastal areas under study. Normally the ini al investment needed for pond excava on is financed by various government-sponsored schemes. With technological interven ons and financial support the system can be made more efficient and can contribute significantly to the economy of the region. 247 Table 2. Economics of exis ng HPS Sl no Item A. Fixed cost a. Pond excava on Quan ty Rate 1 550 labour @Rs.150/day 82500 b. c. B. a. b. c. d. Cast net Pump Variable cost Repairing of embankment Repairing of net Fish seed Fish feed 1 1 1250 15000 1250 15000 65 labour @Rs.150/day 6000 70 3000 100 Every 10-12 years Every year e. Supervision/intercultural (human labor) Release of fish seeds, feeding and harves ng (human labour) Miscellaneous Medicine/lime Vegetables (seeds etc) 20 Rs. 150/ - day 3000 Own supervision 5 Rs. 150/- day 750 Own/hired/mutual sharing f. g. h. i. C. Return Sale of fish Sale of vegetables Sale of water 1 2000 Value Remarks 100 150 450 17000 Rs21900 from 3rd year onwards 3000 Financial analysis: IRR = 11% (<12 % the discoun ng rate), NPV = Rs. (-)2816, BCR = 0.98. Note: in Rs per bigha or 33 decimal of pond, 2012; Engagement of human labor in pond management is low apart from the pond excava on. Releasing the fish seeds, feeding and harves ng are the ac vi es done by human labor and thus the quan ty of human labor was low. 4. Harnessing the opportuni es of homestead produc on systems The primary objec ve of HPS in the study area is to supplement the daily food requirement without any commercial mo ve. Food-based strategies, including homestead food produc on through homestead vegetable produc on programs, not only increase food security but also have an impact on reducing micronutrient deficiencies and women’s empowerment as well as their economic security (HKI/IPHN, 2006). The area of opera on under HPS was very small and the quan ty of marketable surplus was limited. Any produce sold was sold at the local market called haat (within a 2 to 3 km radius) on a daily basis. Since the areas under individual crops were very limited, the quan ty and type of produce available to sell was not linked with the changing market prices. Homestead produc on systems in the study area were not market responsive because the capital investment and area under cul va on were limited, so level of produc on and type of commodi es grown could not be easily altered in line with market supply. The aquaculture component appeared to be managed very differently to the vegetable component of HPS. Aquaculture produc on management was very tradi onal, less capital intensive and devoid of any rigorous management, (i.e., stocking densi es, fish composi on, feeding management, phased harves ng, de-sil ng of 248 ponds or fer liza on in ponds to increase the food availability for fish). Farmers were not so driven to maximize the output from their ponds, rather it was their way of life. Conversely, vegetable produc on units were quite intensive. Farmers tried to maximize their output with all-out efforts from this small produc on unit. It was observed that some of the farmers were changing the crop mix and purchasing quality seeds from their markets to grow in their vegetable gardens depending on market demand. Farmers were keen to gain knowledge par cularly on crop protec on and nutrient management. Efforts should be made to enhance farmers’ ability to adopt intensive aquaculture prac ces as well. The HPS were comprised of several crops like vegetables, fruits, medicinal and aroma c plants or mul purpose tress. In a very small area, a number of vegetables were raised. There was no specializa on of crops and individual outputs were quite small. But it was also observed that a few farmers were growing only a few number of vegetables (like brinjal, cabbage, cauliflower, chilly or tomato) and these farmers were earning a be er income from their plots as compared to their fellow farmers. Their produc on systems were also somewhat driven by market demand and they were more conscious about the crop choice/mix to maximize return. However, a larger variety of crops (mix) provided be er food and nutri on to the family, which is actually the primary objec ve of the HPS. Thus a valid ques on arises: should less variety be produced to meet market demand and raise income or should vegetable produc on be diverse to obtain household food and nutri on security, sacrificing the expected higher return from specialized crops? Research is needed to analyze the trade-off between these two situa ons. Based on the farmers' own choice, priori es and socio-economic condi ons, interven ons may be suggested. Extent of produc on from HPS was not directly dependent on the total size of opera onal holdings. Rather, decisions to operate HPS were dependent on the availability of water in the homestead ponds, family labor, family size, family food requirement and finally the ability to earn cash income by selling available marketable surplus. 5. Conclusion and recommenda ons The survey on HPS elaborated their role, importance and contribu on to the farm households in the coastal region. HPS contributed significantly towards mee ng daily food and nutri on needs, engaged en re families and provided cash income, in turn mi ga ng price or output shocks within the market systems. However, resource use under HPS can be made more produc ve and contribute to a aining regional food security. There is a need to enhance the produc on capacity of the system as a whole. Farmers need ensured supply of quality inputs and training on produc on management of all enterprises—fish, vegetables and livestock. Training should include nutri on educa on and behavior change components (Lanno et al. 2009). Since all members of the family par cipate in HPS, training and upscaling programs should target the whole family, including men, women and children (12 to 18 years of age), either through community groups or school-based projects. Farmers also need financial support to enhance their investment capacity and ensure produc ve resources use. Enhanced produc on levels would increase the quan ty of marketable surplus and thereby contribute more to regional produc on. The government may consider implemen ng schemes for the promo on and improvement of exis ng homestead produc on systems for a large number of farmers in the coastal areas. Acknowledgements The authors acknowledge the financial and technical support to this research work by the Indian Council of Agricultural Research (ICAR) and the CGIAR Challenge Program on Water and Food. This paper presents findings from ‘G2 Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food. We would like to profusely thank the reviewers of the paper for their valuable contribu ons. 249 References Annual Report 2013-14. Central soil Salinity Research Ins tute, Karnal - 132001, India, pp. 127-128. Bandyppadhyay, B. K., B. Maji, H. S. Sen, and N. K. Tyagi. 2003. Coastal Soils of West Bengal – their nature, distribu on and characteris cs, Technical Bulle n no 1/2003, Central Soil Salinity Research Ins tute, Regional Research Sta on, Canning Town, West Bengal, India, pp. 62. Chand, Ramesh., P. A. Lakshmiprasann, and Aruna Singh. 2011. Farm Size and Produc vity: Understanding the Strengths of Small Holders and Improving their Livelihoods. Economic and Poli cal Weekly. Supplement. Vl. XLVI. Nos. 26 & 27. pp. 5. Dey, S., U. K. Sarker, and M.A. Awal. 2012. Year round homestead vegetables produc on: reduc on of nutri onal deficiency and income genera on for small households, Bangladesh J. Prog. Sci. & Tech. 10(2): 187-190. HKI/IPHN. 2006. Homestead Food Produc on: The poten al and opportunity to improve the food security and rural livelihood in Barishal Division. Homestead Food Produc on Programme Bulle n No. 3 (April, 2006) 1-4 pp. Govt. of India. 2006. Livestock Ownership Across Opera onal Land Holding Classes in India, 2002-03, NSS 59th Round, Na onal Sample Survey Organisa on, Ministry of Sta s cs & Programme Implementa on, p. 7. Govt. of India. 2011. Manual on Fishery Sta s cs. Ministry of Sta s cs and Programme Implementaion. Central Sta s cs Office. New Delhi.pp. 1. www.mospi.gov.in. Lanno , Lora., Kenda, Cunningham, and Ruel. (2009). Improving Diet Quality and Micronutrient Nutri on Homestead Food Produc on in Bangladesh, IFPRI Discussion paper 00928, pp. 7. Lokesh, S. B., and T. Hanstad. 2004. Reaping the Rewards of Homestead Gardens: Nutri onal and Income Benefits from Homestead Plots in Karnataka. Bangalore, India: Rural Development Ins tute. Ma n, Imran., Munshi. Sulaiman, and Mehnaz. Rabbani. 2008. Cra ing a Gradua on Pathway for the Ultra-Poor: Lessons and Evidence from a BRAC Programme. Dhaka : BRAC Research and Evalua on Division. Nielsen, R., T. Hanstad, and L. Rolfes. 2006. Implemen ng homestead plot programmes - experiences from India, FAO-LSP Working paper 23. Torlesse, H., H. Moestue, A. Hall, S. de Pee, L. Kiess, and M.W. Bloem. 2004. Dietary diversity in Bangladesh: evidence from the nutri onal surveillance project, In Allevia ng malnutri on through agriculture in Bangladesh: biofor fica on and diversifica on as sustainable solu ons : [Dhaka and Gazipur, Bangladesh, April 22–24, 2002] / editors Nanna Roos, Howarth E. Bouis, Nazmul Hassan, and Khandaker Aminul Kabir. pp. 44-49. Talukder, A. N. J. Haselow, A. K. Osei, E. Villate, D. Reario, H. Kroeun, L. SokHoing, A. Uddin, S. Dhungel, and V. Quinn, 2010. Homestead food produc on model contributes to improved household food security and nutri on status of young children and women in poor popula ons - lessons learned from scaling-up programs in Asia (Bangladesh, Cambodia, Nepal and Philippines). Field Ac on Science Report, www.factsreports.org. Van. Hae en, Roberta, and Phil Moses. 2009. Bangladesh Food Security Country Framework FY 2010-2014. Washington, DC : Food and Nutri on Technical Assistance Project (FANTA II), Academy for Educa onal Development (AED). 250 Homestead farming system: compara ve characteriza on and role in resource poor farmers’ livelihood in Bangladesh and West Bengal K.A. Kabir1, J.K. Sundaray2, S. Mandal3, D.A. Deo2, D. Burman3, S.K. Sarangi3, A. Bha acharya2, M. Karim1, M.B. Shahrier1, S. Cas ne 1 and M. Phillips1 1 WorldFish, Bangladesh and Malaysia, k.kabir@cgiar.org, m.shahrier@cgiar.org, m.karim@cgiar.org, sarahcas ne11@gmail.com, m.phillips@cgiar.org 2 Central Ins tute of Brackishwater Aquaculture, India, jsundaray@gmail.com, ashutoshddeo@gmail.com, aninditabha acharya.87@gmail.com 3 Central Soil Salinity Research Ins tute, India, subhasis2006@gmail.com, burman.d@gmail.com, sksarangicanning@gmail.com Abstract Homestead farming is widely prac ced across the land size spectrum in rural areas of the Ganges delta, in Bangladesh and West Bengal, India, by u lizing family labor to produce a diversified range of crops. A survey was carried out from 2012 to 2013 to increase understanding of this small, complex and dynamic farming unit across the Ganges delta. The survey covered 1280 homesteads in Bangladesh and 720 from West Bengal, in high, medium and low salinity areas, with par cular a en on to poorer households with access to less than 1 ha of land. Homestead farming system (HFS) income in Bangladesh ($246) is slightly higher than West Bengal ($224) though household farming land area is higher in the la er. Five major and o en interlinked farming units—pond aquaculture, vegetable, fruit, poultry and livestock produc on—were common in both countries with the excep ons of betel vine which was more prevalent as a cash crop in West Bengal, and a lower diversity of homestead ponds in Bangladesh. Larger differences between the two countries are observed in vegetable and fruit produc on. In Bangladesh, homestead farming is dominated by vegetable yield with a median of 12,350 kg/ha (range=10,292-14,820) whereas fruit is more significant in West Bengal with a median yield of 12,593 kg/ha (range=0-98,800). Products from all the farming units in both countries have a posi ve correla on between produc on and consump on except for dairy milk in Bangladesh and vegetable and poultry meat in West Bengal. The value of inputs (fer lizer, seed, feed) used in HFS is much higher in Bangladesh than in West Bengal with per household expenditure of $71 in Bangladesh compared to just $15 in West Bengal. The contribu on of HFS to household income is higher in Bangladesh (23%) than it is in West Bengal (16%). There are significant differences in HFS produc vity between low and high salinity areas. The major constraint of farming system intensifica on in Bangladesh for aquaculture is a lack of investment and for other crops is disease management, while households in West Bengal priori ze lack of skills for aquaculture and non-aquaculture crops. Key message: Presence of more farming units in HFS provides higher opportunity for food produc on, improving food and nutri onal security and income. Keywords: farming components, food produc on, aquaculture, non-aquaculture 1. Introduc on The coastal delta of the Ganges River system, encompassing large parts of Bangladesh and the Indian State of West Bengal, is one of the most highly populated regions of the world (Amhed 2006). Bangladesh has a popula on density of 1,203 per sq. km of land area (World Bank 2014) and in West Bengal the density is 1,029 people per sq. km (Census of India 2011). Human development indicators for the rural areas indicate that 35% and 26% of the rural popula on fell below the na onal rural poverty line in 2010 in Bangladesh and India, respec vely (World Bank 2014). 251 The Ganges coastal delta comprises diversified aqua c-agricultural farming systems across a complex landscape subjected to wide seasonal fluctua ons in salinity and freshwater availability, and vulnerability to sea level rise, flooding and extreme weather events such as cyclones (Huq and Ayers 2007; Sayeed 2007). The Ganges delta is considered to be an irrigated cropping system with a yield gap (Václavík et al. 2013). An increase in food produc on is required in the Ganges delta and sustainable intensifica on is widely considered necessary for future food requirements (Garne et al. 2013). Recent research in Bangladesh on produc vity improvement in rice suggests significant opportuni es do exist within the region to increase produc vity of field crops (CPWF 2013), though access of poorer households to sufficient land area remains a concern. Small land holdings are common across the Ganges coastal delta, with 80% of farming households in Bangladesh having access to less than 0.5 ha of produc ve land (Bangladesh Bureau of Sta s cs 2011); in West Bengal 59% of the popula on has land holdings of less than 0.5 ha (Agriculture Census 2010-11). Homestead farming systems (HFS) have been developed based on micro-sites and under the exis ng land constraints, it is an important food produc on unit for the poor in Bangladesh (Miah and Hussain 2010) as well as in West Bengal (Nelson et al. 2006). Much research has focused on field crops but less on overall integrated homestead farming systems. Homestead farming is widely prac ced across the land size spectrum in rural areas of Bangladesh and India (Dey et al. 2012 and Bha acharya et al. 2013). HFS comprise opera onal farm units that occupy a por on of the homestead land, u lize family labor, par cularly women, and involve the produc on of various combina ons of aquaculture, vegetables, poultry, livestock and fruit crops (Dash and Misra 2001). HFS have been promoted for family income and home consump on across Bangladesh (Lanno et al. 2009) and West Bengal (Robin et al. 2006) though less a en on has been given to HFS in more challenged coastal deltaic regions of both countries. Understanding of the influence of salinity and water management on field crops is increasing (Huq and Ayers 2007; Sayeed 2007) but less knowledge is available on homestead farming systems within salinity challenged areas. Further, there have been no studies of a compara ve or commonality nature on homestead systems across the two countries despite their shared deltaic agro-ecologies. This paper presents informa on from consulta ons with households in both Bangladesh and West Bengal, and an extensive survey of HFS during 2012 and 2013 in order to compare homestead farming systems in Bangladesh and West Bengal for iden fying opportuni es for cross-country sharing of knowledge and experience across the region. 2. Methods 2.1 Study sites The study was conducted in three regions in southern Bangladesh and five regions in West Bengal. These eight regions varied in water salinity, ranging from approximately 0 to 25 ppt (Table 1). In Bangladesh large (1000s of ha), circular embankments, called polders, which are prevalent throughout the coastal zone of Bangladesh, define the three regions. The individual polders inves gated in the present study are Polder 43, Polder 30 and Polder 3 (Figure 1). In West Bengal, the study was conducted in three coastal blocks called Kakdwip, Namkhana and Sagar under South 24 Parganas, and two blocks in north 24 Parganas called Sandeshkhali I and Sandeshkhali II (Figure 1). 252 “HIGH SALINITY” “MEDIUM SALINITY” •Water “stagnation” 30-50 cm •Water “stagnation” 30-50 cm several weeks in aman several weeks in aman •River water saline Dec-Jul •River water saline mid-Feb-Jun •High soil salinity in dry •Medium soil salinity in dry season season North 24 Parganas “LOW SALINITY” •Water “stagnation” 30-50 cm several weeks in aman •River water fresh 10-11 months •Mild soil salinity in dry season Polder 30 Patuakhali STU Polder 3 South 24 Parganas Polder 43/2/F West Bengal, India South West Bangladesh Fig. 1. Study areas of the lower Ganges basin in Bangladesh and West Bengal, India. The blocks and polders were selected as representa ve of the socio-economic, agricultural prac ce and environmental condi ons in each country. Each loca on was categorized into three zones of low salinity, moderate salinity or high salinity based on the range of water salinity (Table 1). However, salinity levels are variable, driven by seasonal changes, and are naturally heterogeneous even among different plots within a single farmer’s field (Rahman et al. 1993; Fleming 2004; SRDI 2001; Gain et al. 2007; Shamsuddin et al. 2006). Table 1. Survey areas in southwest Bangladesh and West Bengal, India, categorized by water salinity level and total number of households sampled Country Region Water Salinity (ppt) Number of households (HH) sampled Bangladesh Polder 43 Polder 30 Polder 3 Total Block Kakdwip Block Namkhana BlockSandeshkhali I Block Sandeshkhali II Low (0-05) Moderate (0-10) High (05-25 320 338 461 1,119 120 120 120 120 Block Sagar Total High (05-25) India Low (0-05) Moderate ( 0-10) Moderate (0-10) Moderate (0-10) 240 720 253 2.2 Survey households In India, 720 households were surveyed from each block apart from Sagar where 240 households were sampled (Table 1). Sagar is an island with a dynamic brackishwater ecosystem and is vulnerable to soil erosion, dal movement and saliniza on. The rela vely higher number of households selected in Sagar was intended to capture this dynamic and vulnerable environment. Between 320 and 461 households were surveyed in each of the three polders in Bangladesh (Table 1). Complete lists of all the households residing in each of the polders and blocks were collected from the Union Parishad and Panchayat office in Bangladesh and West Bengal, respec vely. Each of the polders and blocks were considered as a separate region to allow for stra fied random sampling. SPSS so ware was used to randomly select households from each region at 5.86% confidence interval. 2.3 Data collec on A household survey was conducted between January and March 2012 in Bangladesh and August 2012 and January 2013 in West Bengal, India. Both open ended and closed ques ons were asked to ensure par cipants could elaborate when suitable. The survey collected informa on on socio-economics, homestead farming system components, yield and its contribu on to household income. 2.4 Data analysis Households having > 1 ha total land were excluded from the analysis as this study was focused on asset-poor households. The sample number from Bangladesh was 1106 and 652 from West Bengal. Data were analyzed by using R – sta s cal package and SPSS. For conversion from local currency to US Dollars we used the conversion rate at the me of the survey, that is 43 Indian Rupees and 75 Bangladeshi Taka to 1 US$. The results were presented at median value due to the lack of normal varia on among households in each country and salinity level. 3. Results 3.1 Farming land, HFS units and yield There are five major components in HFS and ponds cons tute a major por on of homestead area in both countries (Table 2). However, homestead land area in Bangladesh is nearly half that of West Bengal (Figure 2) and thus in terms of area very small farming units are available for the poor. Table 2. Distribu on of HFS units in Bangladesh and West Bengal in rela on to homestead land area Pond Bangladesh (%) 45 West Bengal (%) 50 vegetable garden Livestock shade Poultry shade Fruit trees 15 3 1 36 29 8 6 7 254 1.00 0.75 country Bangladesh 0.50 India 0.25 0.00 Bangladesh India Country Fig. 2. Homestead land area (ha) in Bangladesh and West Bengal (India) based on land owned. There is a significant difference in homestead land size between regions (Kruskal-Wallis chi-squared test = 397.3, df = 7, P = 0.000). Landholdings at Kakdwip, a low salinity area, were the largest with median 0.14 ha, and also had the highest income. Landholding is smallest in Polder 3, a high salinity area, with a median of just 0.03 ha per household. Though five major farming components were iden fied they are not seen in all farming households (Table 3) and betel vine is seen only in West Bengal. Table 3. Presence of farming components in Bangladesh and West Bengal Bangladesh West Bengal Pond Vegetable garden Number of HH 463 310 % of sampled HH 42 28 Number of HH 475 284 % of sampled HH 73 44 Livestock shade Poultry shade Fruit trees Betel Vine 410 536 655 0 37 48 59 0 79 20 25 98 12 3 4 15 Presence of more farming components is significantly related to increased produc on from HFS. In Bangladesh, homestead farming is dominated by vegetables with a median of 12,350 kg/ha (range=10,292-14,820) whereas fruit drives produc on in West Bengal with a median yield of 12,593 kg/ha (range=NA-98,800). Regarding the produc on of animal source food, fish and prawn produc on are similar in both countries, while HFS in West Bengal produce more chicken products and HFS in Bangladesh produce more livestock products (Figure 3). 255 250 200 150 Bangladesh West Bengal 100 50 0 Poultry meat (kg) Poultry egg (n) Fish (kg) Prawn (kg) Dairy milk (liter) Livestock Meat (kg) Vegetables (kg) Fruits (kg) Fig. 3. The annual yield of homestead products per household in Bangladesh and West Bengal. Overall, the low salinity soil type has significantly higher produc vity in terms of the yearly homestead products produced in both Bangladesh and West Bengal (Figure 4). For example, HFS yield value from low salinity areas in West Bengal was much higher (US$324) than from high salinity areas (US$164) of West Bengal. This was also the case in Bangladesh. 1500 soiltype High Salinity 1000 Low Salinity Moderate Salinity country 500 Bangladesh India 0 High Salinity Low Salinity Moderate Salinity Salinity Level Fig. 4. Value of annual yield (in USD) by salinity level. 3.2 Consump on of HFS products Homestead food consump on in Bangladesh is posi vely correlated with produc vity of most of the HFS components (P<0.05). This rela onship is strongest for aquaculture (r = +0.79), poultry meat (r = +0.85), poultry egg (r = +0.89) and fruit (r = +0.76), while it is rela vely weak for vegetable (r = +0.50) and nega ve for livestock milk (r = -0.09). On the other hand, in West Bengal produc vity has strong posi ve correla on (P<0.05) with fruits (r=+0.96), poultry egg (r=+0.88) and livestock milk (r=+0.76) and is weakly related with poultry meat (r=+0.22), vegetable (r=+0.38) and aquaculture (r=+0.48). 256 3.3 Contribu on of HFS to household income Homestead farming plays a key role for poor households, contribu ng between 10-25% of total household income in 2011 (Figure 6). Total contribu on of HFS to household income is higher in Bangladesh (23%) than it is in West Bengal (16%). Non-aquaculture ac vi es contributed more in Bangladesh (median $176) than in West Bengal (median $44) while aquaculture ac vi es contributed more in West Bengal than in Bangladesh, with a median of $180 and $70 per household per year, respec vely (Table 4). Table 4. Household income from different sources and contribu on of HFS to household income Country Total Off Farm Field ($) HFS HFS NonHFS Total ($) HFS contribu on to ($) ($) Aquaculture Aquaculture Total hh income (%) ($) ($) Bangladesh 1079 667 381 70 176 246 23 West Bengal 1622 1103 333 180 44 224 14 3.4 Constraints and challenges for increasing produc vity from HFS Major constraints as reported by households for improving HFS aquaculture in Bangladesh are the lack of investment capital, while in West Bengal it is the lack of skill. For non-aquaculture HFS in Bangladesh, disease was the major problem and in West Bengal it was again lack of skill (Figure 5a and b). 60 50 40 30 20 10 0 Bangladesh West Bengal Fig. 5a. Major constraints and challenges for improvement of HFS aquaculture. 60 50 40 30 20 10 0 Bangladesh West Bengal Fig. 5b. Major constraints and challenges for improvement of HFS non-aquaculture. 257 3.5 Integra on between HFS units: pond plays a key role There is a high prevalence of ponds in Bangladesh and West Bengal, which likely plays an important role as a water source for other farming components and as a key component for integra on of farming components and recycling of resources. In both Bangladesh and West Bengal households with ponds have higher interac on among farming components than in households without ponds. In Bangladesh 50%, 58% and 2% of households use pond water for vegetable produc on, poultry and livestock, respec vely, and 4% use pond bo om soil for vegetable produc on. Twenty-one percent of households use poultry and livestock manure and 3% vegetable by product as feed for fish produc on in ponds. Also 22% of households use poultry and livestock manure in vegetable and fruit produc on. While in West Bengal pond water use is more common, there was varia on between two study areas. In South 24 Parganas, 100%, 83% and 82% of households use pond water for vegetable produc on, poultry and livestock, respec vely. While in North 24 Parganas, 21% and 19% of households use pond water for vegetable produc on and poultry, respec vely. Integra on among other components appears to be not very strong. Only 6% and 4% of households in South 24 Parganas use poultry and livestock manure for fish culture and vegetable and fruit produc on. In North 24 Parganas no other integra on was reported during the survey. In contrast, households without ponds in Bangladesh had limited integra on, with only 10% of households using poultry and livestock manure in vegetable and fruit produc on. No other integra on was observed. The situa on in West Bengal was even more limited; homesteads without ponds reported no integra on or resource reuse between farming components. This finding suggests a strongly influence of ponds in intensifying and integra ng HFS in Bangladesh and South 24 Parganas of West Bengal, and rela vely weak influence on N24 Parganas. 4. Discussion The findings from southern Bangladesh and West Bengal reveal a diversity of produc on systems and essen ally confirm the importance of HFS for income, food and nutri on on homesteads with limited land. The results also suggest opportuni es to increase produc vity of the homestead farming components, increasing their contribu on to income and improved nutri on of poor households. 4.1 Farming land, HFS units and yield HFS in Bangladesh and West Bengal are comparable in terms of their produc on components i.e., fruit, vegetables, fish and small livestock, but differ in terms of their input investment, yield and consump on. Findings suggest poten al for the poor to increase household food security and nutri on if systems are well managed, with access to necessary inputs and skills. Other authors have found that diversified HFS results in improved yields, leading to increased food consump on and improved nutri on in homesteads (Taher et al. 2004). Homestead produc on systems also offer opportuni es for increasing seasonal availability of foods (Bushamuka et al. 2005). High salinity areas appear to be less produc ve. Intrusion of saline water changes soil salinity and restricts agriculture in coastal areas (Faisal and Parveen, 2004; Chowdhury et al., 2006). Further analysis of the datasets is ongoing to further understand the impacts of salinity on household farming systems, and further research is required on saline-adap ve technology development for homestead food produc on. 4.2 Consump on of HFS products Homestead products are marketed and consumed at the household level. Present research finds strong correla on between HFS produc on and household consump on. Weinberger (2013) similarly men oned that HFS is small-scale, non-commercial and mostly for household consump on. HFS has been considered a viable development approach, targe ng nutri on for several decades, as evidenced by the large set of literature available on the topic (Marek et al. 1990; Schipani et al. 2002; Bloem et al. 1996; Marsh 1997; Cameron et al. 2012; Cabalda et al. 2011; Heim et al. 2009, 2011). 258 Benefits of HFS have been reported to include providing enhanced food supply and increased diversity of food, especially for the poor (Bushamuka et al. 2005; Talukder et al. 2010; Olney et al. 2009; Trinh et al. 2003; Cabalda et al. 2011). The current study essen ally confirms these findings in coastal areas in the Ganges delta, and its importance for poor people. Another advantage in HFS is produc on of food close to consump on areas and the reduced need for transporta on and preserva on (Novo and Murphy 2006). Weinberger (2013) also men oned that improved access to and consump on of nutri ous foods, improved social protec on and a healthier environment can be derive from HFS. 4.3 Contribu on of HFS to household income The current study found HFS contributes 10-25% of household income. From a global perspec ve such contribu ons range from 0% (Méndez et al. 2001) to 50% (Trinh et al. 2003). One important lesson from the experiences of Helen Keller Interna onal (HKI) is that the objec ves of households have changed and a focus on agricultural produc on of nutri ous foods must consider both market-oriented income opportuni es as well as household consump on (McDermo et. al., 2013). McDermo (2013) also men oned crea ng food-related skills (poultry shed construc on, vaccina on) and technologies to help resource-poor landless and land-constrained people to earn more money indirectly through HFS or food produc on. This paper did not explore this aspect, but for the pro-poor this opportunity must be be er understood and further studies should be carried out on this topic. 4.4 Challenges and opportuni es HFS interven ons o en aim to improve nutri ous food consump on, income, intra-household equity and child care (McDermo et. al., 2013). This research iden fied lack of skills, par cularly on disease management and technology extension support, as well as lack of investment as key constraints in Bangladesh and West Bengal. The other factors hampering HFS sustainability include, but are not limited to, heavy use of agrochemicals, dependency on hired labor, risk of market fluctua on for commercial products, scarcity of land, high popula on density and impact of urbaniza on (Soemorowoto and Conway 1992; Karyano 1990; Michon and Mary 1994; Kehlenbeck and Maass 2004; Arifin et al. 1998). According to Alam (2011) the most important characteris cs that will ensure future sustainability are the capacity of the farming system to rapidly cope with changing circumstances. Even in the face of social and demographic changes in and around rural landscapes, HFS itself remain rela vely stable (Alam 2011) Lanno et al. (2009) evaluated the Helen Keller Interna onal (HKI)-supported homestead food produc on (HFP) program that was conducted in Bangladesh for two decades in partnership with 80 local NGOs. They found that the ini a ve increased household food produc on, availability and access to food, and improved micro nutrient deficiency, especially when agricultural interven ons came with behavioral change communica on (BCC) that empowered women through increased access to resources. It also improved child care and nutri on. Similar outcomes have been observed in West Bengal (DRCSC, undated). 4.5 Integra on between HFS units: pond plays a key role Ponds play a key role as a water source for other HFS units and influence more integrated produc on, allowing several op ons for resource reuse and recycling. For centuries, a pond has been used as source of domes c water supply and small-scale irriga on (Li le et al., 2007). Prein (2002) reported “on-farm ponds” as emerging foci for agricultural diversifica on. Ponds provide opportuni es for integra on with hor culture on dikes and are prac ced widely in Asia (Ruddle and Zhong, 1988). The pa ern of integra on and resource reuse between pond aquaculture and livestock rearing is also common in this region (Li le and Edwards, 2003). Whilst Micheilsons (2002) noted that individual component yield and input use efficiency can be lower than very intensive fish farming, in total, integra on of aquaculture and agriculture can provide mul ple 259 benefits (Barman 2000; AIT/DOF 2001; Karim 2006) and increase overall HFS produc on (Karim 2011; Karim 2006). This research also indicates that the presence of a pond in a homestead provides posi ve outcomes, suppor ng more farming units within the HFS and making higher contribu ons to household income and nutri on. 5. Conclusions and recommenda ons This study was the first a empt to compare the whole HFS of two countries of the lower Ganges delta. The study iden fies HFS as a valuable component for the poor for food produc on and household consump on in both countries, with several key messages: - Ponds are common in homestead areas, a feature more common in West Bengal than Bangladesh - Presence of a pond plays a key role in improving HFS - HFS produc vity is higher in low salinity areas compared to high saline areas - Presence of more HFS components results in increased produc vity of household food produc on Further research is required on: - Factors that influence yield and consump on - HFS in saline areas - Resource flow within the components for designing an integrated produc on unit on a very small piece of land for the poor - Poten al role of credit in HFS improvements - Mi ga on of diseases - Technology transfer for rural areas of West Bengal Acknowledgements We would like to acknowledge the CGIAR Challenge Program on Water and Food and CGIAR Research Program on Aqua c Agricultural Systems for suppor ng this study. We also thank WorldFish/USAID AIN project for providing addi onal funding for Bangladesh data collec on, entry and analysis. We are grateful to Dr. Khandaker Murshed-E-Jahan, Dr. Manoranjan Mondal and Dr. T P. Tuong for their contribu on in developing the survey ques onnaire and Dr. Rayhan Hayat Sarwer for developing the survey method. We would like to thanks C. W. Chang for his kind support in analyzing this large data set. References Agricultural Census Database. 2010-11. Number and area of holding by size group. h p://agcensus.dacnet.nic.in AIT/DOF, 2001. Fish seed quality in Northwest Bangladesh. State of the System report. Aquaculture Outreach Programme, AIT, Thailand, 37 pp. Alam, M. 2011. Tropical Homegardens in Bangladesh: Characteris cs and Sustainability. E. Lich ouse (ed.), Alterna ve Farming Systems, Biotechnology, Drought Stress and Ecological Fer lisa on, Sustainable Agriculture Reviews 6, DOI 10.1007/978-94-007-0186-1_8, © Springer Science+Business Media B.V. 2011 Ali, M. N., Ghosh, B., Gantait, S., Chakraborty, S. 2014. Selec on of Rice Genotypes for Salinity Tolerance Through Morpho-Biochemical Assessment. Rice Science 21(5): 288–298. 260 Amhed, A.U. 2006. Who is doing what in Bangladesh? Report on the First Mee ng. Comprehensive Disaster Management Programme, Government of Bangladesh, Dhaka. Arifin HS, Sakamoto K, Chiba K (1998) Effects of urbanisa on on the vegeta on structure of home gardens in West Java, Indonesia. Jpn J Trop Agri 42:94–102 Bangladesh Bureau of Sta s cs. 2011. Popula on and housing census: preliminary results. Census of India, 2011. Final Popula on Totals: West Bengal Profile. Barman, B.K., 2000. Assessment of Nile lapia (Oreochromis nilo cus) Seed Produc on and Grow-out Systems for Small-scale Farmers in Northwest Bangladesh. PhD Disserta on. Asian Ins tute of Technology, Thailand, 246 pp. Barua, S., S. Tripathi, A. Chakraborty, S. Ghosh, and K. Chakrabar . 2012. Characteriza on and crop produc on efficiency of diazotrophic bacterial isolates from coastal saline soils. Microbiological Research 167(2): 95-102 (January 20). Bha acharya, A., A.D. Deo, J. K. Sundaray, and A.G. Ponniah. 2013. Impact of homestead farming system on rural economy -A study in South 24 Pargana district of West Bengal. Abstract, Interna onal Ecosystem Conserva on, Climate Change and Sustainable Development Conference, October 3-5, Kerala, India. Bloem, M.W., Huq, N., Gorstein, J., Burger, S., Kahn, T., Islam, N., et al. (1996). Produc on of fruits and vegetables at the homestead is an important source of vitamin A among women in rural Bangladesh. European Journal of Clinical Nutri on, 50 Suppl 3, S62–S67. Bushamuka, V.N., S. de Pee, A. Talukder, L. Kiess, D. Panagides, A. Taher, and M. Bloem. 2005. Impact of a homestead gardening program on household food security and empowerment of women in Bangladesh. Food and Nutri on Bulle n 26 (1): 17-25. Cabalda, A. B., Rayco-Solon, P., Solon, J. A. A., & Solon, F. S. (2011). Home gardening is associated with Filipino preschool children’s dietary diversity. Journal of the American Diete c Associa on, 111(5), 711. Cameron, R. W. F., Blanuaa, T., Taylor, J. E., Salisbury, A., Halstead, A. J., Henricot, B., et al. (2012). The domes c garden–its contribu on to urban green infrastructure. Urban Forestry & Urban Greening, 11(2), 129–137. CPWF. 2013. Cracking the produc vity code: demonstra ng the benefits of community water management and improved crop technologies in the coastal zone of Bangladesh isn’t simple. h p://waterandfood.org/2013/08/15/cracking-the-produc vity-code Chowdhury, M.A., Shivako , G.P., Salequzzaman, M., 2006. A conceptual framework for the sustainability assessment procedures of the shrimp aquaculture industry in coastal Bangladesh. Int. J. Agric. Resour. Govern. Ecol. 5, 162–184. Dash, S.S., and M.K. Misra. 2001. Studies of hill agro-ecosystems of three tribal villages on the Eastern Ghats of Orissa, India. Agriculture, Ecosystem and Environment 86: 287–302. Dey, S, U.K. Sarker, and M.A. Awal. 2012. Bangladesh Year round homestead vegetable produc on: Reduc on of nutrient deficiency and income genera on for small households. J. Prog. Sci. & Tech 10(2): 187-190. DRCSC, undated. h p://www.drcsc.org Faisal, I. M., & Parveen, S. (2004). Food security in the face of climate change, popula on growth, and resource constraints: implica ons for Bangladesh. Environmental Management, 34(4), 487–498. doi:10.1007/s00267-003-3066-7. 261 Fleming, C. 2004. Challenges Facing the Shrimp Industry in Bangladesh. American Interna onal School, Dhaka. Gain, A.K., K.P. Aryal, P. Sana, M.N. Uddin. 2007. Effect of river salinity on crop diversity: a case study of south west coastal region of Bangladesh. Nepal Agriculture Research Journal 8. Garne , T., M.C. Appleby, A. Balmford, I.J. Bateman, T.G. Benton, P. Bloomer, B. Burlingame, M. Dawkins, L. Dolan, D. Fraser, M. Herrero, I. Hoffmann, P. Smith, P.K. Thornton, C. Toulmin, S.J. Vermeulen, and H.C.J. Godfray. 2013. Sustainable intensifica on in agriculture: premises and policies.Science 341: 33–34. Heim, S., Stang, J.,& Ireland,M. (2009). A garden pilot project enhances fruit and vegetable consump on among children. Journal of the American Diete c Associa on, 109(7), 1220. Heim, S., Bauer, K. W., Stang, J., & Ireland, M. (2011). Can a community-based interven on improve the home food environment? Parental perspec ves of the influence of the delicious and nutri ous garden. Journal of Nutri on Educa on and Behavior, 43(2), 130. Huq, S. and J.M. Ayers, (2007) Cri cal list: the 100 na ons most vulnerable to climate change, in IIED Sustainable Development Opinion. London: Interna onal Ins tute of Environment and Development. h p://www.drcsc.org Karim, M., Li le, D.C., Kabir, M.S., Verdegem, M. J.C., Telfer, T., Wahab, M. A., 2011. Enhancing benefits from polycultures including lapia (Oreochromis nilo cus) within integrated pond-dike systems: a par cipatory trial with households of varying socio-economic level in rural and peri-urban areas of Bangladesh. Aquaculture. 314 (1-4): 225–235. Karim, M., 2006 The livelihood impacts of fishponds integrated within farming systems in Mymensingh District, Bangladesh PhD Disserta on, University of S rling, 351 pp. Karyano (1990) Homegardens in Java. Their structure and func on. In: Landauer K, Brazil M (eds) Tropical homegardens. The United Na ons University, Tokyo, Japan, pp 138–146 Kehlenbeck K, Maass BL (2004) Crop diversity and classifica on of of homegardens in Central Sulawesi, Indonesia. Agroforestry Syst 63:53–62 Lanno , L., K. Cunningham and M. Ruel. 2009. Improving Diet Quality and Micronutrient Nutri on: Homestead Food Produc on in Bangladesh. IFPRI Discussion Paper 00928. www.ifpri.org/millionsfed Li le, D.C., Karim, M., Turongruang, D., Morales, E.J., Murray, F.J., Barman, B.K., Haque, M.M., Kundu, N., Belton, B., Faruque, G., Azim, E.M., Ul Islam, F., Pollock, L.J., Verdegem, M.J., Young, J.A., Leschen, W., Wahab, M.A., 2007. Livelihood impacts of ponds in Asia-opportuni es and constraints. In: Zijpp, A.J., van der, J.A.J., Verreth, Le Quang Tri, M.E.F., van Mensvoort, R.H., Bosma, M.C.M. (Eds.), Fishponds in Farming Systems .2006. : Beveridge; Proceedings of a Symposium Held in Can Tho City, 28– 30 April 2006, Organised by Can Tho University, Vietnam and Wageningen University. Wageningen Publishers, Netherlands. Li le, D.C. and P. Edwards, 2003. Integrated Livestock-Fish Farming Systems: The Asian experience and its relevance for other Regions. Inland Water Resources and Aquaculture Service and Animal Produc on Service. Food and Agriculture Organiza on, Italy. 177 pp. Marek, T., Brun, T., & Reynaud, J. (1990). Do home garden projects improve income and nutri onal status? A case study in Senegal. Food and Nutri on Bulle n, 12(1), 20–25. Marsh, R. (1997). Household food security through home gardening– evidence fromBangladesh. In F. Heidhues&A. Fadani (Eds.), Food security and innova ons: Successes and lessons learned. Frankfurt: Peter Lang Press. 262 McDermo , J., Aït-Aïssa, M., Morel, J., Rapando, N., 2013. Agriculture and household nutri on security − development prac ce and research needs. Food Sec. (2013) 5:667−678 Méndez VE, Lok R, Somarriba E (2001) Interdisciplinary analysis of homegardens in Nicaragua: Micro-zona on, plant use and socioeconomic importance. Agroforestry Sys 51:85−96 Miah, G.M. and J.M. Hussain. 2010. Homestead Agroforestry: a Poten al Resource in Bangladesh. In SUSTAINABLE AGRICULTURE REVIEWS 3: Sociology, Organic Farming, Climate Change and Soil Science, ed. Eric Lich ouse and Springer, Chapter 16. Michon G, Mary F (1994) Conversion of tradi onal village gardens and new economic strategies of rural households in the area of Bogor, Indonesia. Agroforestry Syst 25:31–58 Micheilsons, C.G.J., K. Lorenzen, M.J. Philips and R. Gauthier, 2002. Asian carp farming systems: towards a typology and increased resource use efficiency. Aquaculture Research 33, 403-413. Mondal, M.K., S.I. Bhuiyan, and D.T. Franco. 2001. Soil salinity reduc on and predic on of salt dynamics in the coastal ricelands of Bangladesh. Agricultural Water Management 47(1): 9-23. Nielsen, R., T. Hanstad, and L. Rolfes. 2006. Implemen ng homestead plot programmes: Experience from India. LSP Working Paper 23, FAO. Novo, M. G., & Murphy, C. (2006). Urban agriculture in the city of Havana: A popular response to a crisis. In R. V. Veenhuizen (Ed.), Ci es farming for the future, urban agriculture for sustainable ci es (pp. 329–349). Philippines: RUAF Founda on, IDRC and IIRR. Olney, D. K., Talukder, A., Ianno , L. L., Ruel, M. T., & Quinn, V. (2009). Assessing impact and impact pathways of a homestead food produc on program on household and child nutri on in Cambodia. Food and Nutri on Bulle n, 30(4), 355–369. Prein, M., 2002. Integra on of aquaculture into crop-animal systems in Asia. Agricultural Systems 71, 127-146. Rahman, S.A., M.S. Hussain, Z. Parveen, and A.S.M. Mohiuddin. 1993. Iden fying characteris cs of poten al and actual acids sulphate soils. Bangladesh Journal of Soil Science 24:41-49. Rahman, S.M., M.I. Khalil, and M.F. Ahmed. 1995. Yield-water rela ons and nitrogen u liza on by wheat in salt-affected soils of Bangladesh. Agricultural Water Management 28(1): 49–56. Ruane, A.C., C.M. David, H. Yu Winston, A. Mozaharul, Sk.G. Hussain, A.S. Khan, A. Hassan, Md. Bhuiya, T.A. Hossain, R. Goldberg, R.M. Horton, and C. Rosenzweig. 2013. Mul -factor impact analysis of agricultural produc on in Bangladesh with climate change. Global Environmental Change. 23(1):338–350 Ruddle, K. and G.F. Zhong, 1988. Integrated agriculture - aquaculture in south China: The dike-pond system of the Zhujiang Delta. Cambridge University Press. USA, 173 pp. Sayeed, S.K. 2007. Climate change and Bangladesh: A perspec ve on where we are. In The Daily Star, Dhaka. Schipani, S., van der Haar, F., Sinawat, S., & Maleevong, K. (2002). Dietary intake and nutri onal status of young children in families prac cing mixed home gardening in northeast Thailand. Food and Nutri on Bulle n, 23(2), 175–180. Shamsuddin, S., C. Xiaoyong, M.K. Hazarika. 2006. Evalua on of groundwater quality for irriga on in Bangladesh using geographic informa on system. J. Hydrol. Hydromech 54: 3-14. Soemorowoto O, Conway GR (1992) The Javanese homegardens. J Farm Syst Res-Ext 2:95–118 SRDI. 2001. Soil Salinity in Bangladesh. Soil Research Development Ins tute, Ministry of Agriculture, Bangladesh, Dhaka. 263 Taher, A., A. Talukder, N.R. Sarkar, V.N. Bushamuka, A. Hall, S. de Pee, R. Moench-Pfanner, L. Kiess, and M.W. Bloem. 2004. Homestead gardening for comba ng vitamin A deficiency: The Helen Keller Interna onal, Bangladesh, experience. In Allevia ng malnutri on through agriculture in Bangladesh: Biofor fica on and diversifica on as sustainable solu ons, ed. N. Roos, H.E. Bouis, N. Hassan, and K.A. Kabir, Washington, D.C.: Interna onal Food Policy Research Ins tute. Talukder, A.,Haselow,N. J., Osei,A.K.,Villate, E., Reario,D., Kroeun, H., et al. (2010). Homestead food produc on model contributes to improved household food security and nutri on status of young children and women in poor popula ons. Field Ac ons Science Reports. Trinh LN, Watson JW, Hue NN, De NN, Minh NV, Chu P, Sthapit BR, Eyzaguirre PB (2003) Agrobiodiversity conserva on and development in vietnamese home gardens. Agri, Ecosys Env 97:317–344 Václavík, T., S. Lautenbach, T. Kuemmerle, and R. Seppelt. 2013. Mapping global land system archetypes. Global Environmental Change 23(6): 1637-1647. Weinberger, K. 2013. Home and community gardens in Southeast Asia: poten al and opportuni es for contribu ng to nutri on-sensi ve food systems. Food Sec. (2013) 5:847–856 World Bank. 2014. World devel. indicator data. h p://data.worldbank.org/country?display=graph 264 Producing fish in small shaded homestead ponds: finding solu ons with rural women K. A. Kabir, G. Faruque, R. Sarwar, B. Barman, A. Choudhury, M. Hossain, E. Hossain, N. A. Aleem, M. Karim, K. Kamp and M. Phillips 1 WorldFish, Bangladesh and Malaysia, k.kabir@cgiar.org, f.golam@cgiar.org, r.sarwer@cgiar.org, b.barman@cgiar.org, a.choudhury@cgiar.org, md.hossain@cgiar.org, md.e.hossain@cgiar.org, n.aleem@cgiar.org, m.karim@cgiar.org, k.kamp@cgiar.org, m.phillips@cgiar.org Abstract Small household ponds in rural Bangladesh were not constructed for aquaculture and commonly face various challenges for fish produc on, including physical shading caused by plan ng of trees for wood and fruit. Such ponds are considered underu lized with considerable poten al for improvement. WorldFish ini ated a farmer–researcher collabora ve par cipatory ac on research (PAR) project in 2013 to increase fish produc on and consump on from shaded household ponds and to improve women’s decision-making capacity in the use of these resources. The research was carried out in eight villages of southwest Bangladesh covering freshwater and brackishwater areas. There were six polyculture treatments (T) of which three (T1-T3) were in brackishwater and three (T4-T6) in freshwater areas. There were three treatments with four replica ons of each treatment in each village. All treatments were designed in consulta on with women farmers and implemented by the women with assistance from a project team. The results showed no significant varia on in fish produc on across the areas, though average yield was higher in fresh water. Comparison across all treatments showed significantly higher yield in T4, a polyculture treatment combining GIFT Tilapia, Heteropneustes fossilis, Anabas testudineus and Cyprinus carpio. Overall annual fish produc vity increased more than 600% from the baseline; 76% of fish produced were consumed at household level and annual household fish consump on increased nearly 500%. The PAR approach to aquaculture improvement did not create problems for women in daily household ac vi es; rather 25% men oned that lime improved water quality for their household work and 50% indicated that while taking care of fish at the pond they also had the opportunity to water vegetables and fruit trees regularly, contribu ng to increased yield of these crops. Women’s decision-making capacity on daily pond management increased (99%) while not very much on access to markets (8%). New learning as reported by women farmers was associated with technology, research and leadership. A cri cal challenge for women farmers was access to quality inputs and markets. The posi ve outcomes from the PAR approach, including developing farmers’ capacity to analyze research outputs, suggests scope for wider applica on of the approach. Key message: Engaging rural women through a par cipatory ac on research (PAR) approach helped address the challenges associated with shaded homestead pond aquaculture. Outcomes include increased fish produc on, household consump on and some capacity development among women’s decision making regarding use of the pond resource for aquaculture. Keywords: par cipatory ac on research (PAR), women, aquaculture, fish consump on, learning, decision making 1. Introduc on 1.1 Characteris cs of small homestead ponds and challenges of increasing produc vity Many households in rural areas possess a small pond close to their homestead (Huda et al. 2010; Kranzlin 2000; Li le et al. 2007) and altogether approximately 4.27 million household ponds are found in Bangladesh (Belton and Azad 2012). This accounts for 20% of rural households (Jahan et al. 2010) with varia on among 265 different land-classes (Belton et al. 2011). Usually homestead ponds are small in size, ranging between 0.08-0.1 ha. However, there is further varia on in prevalence of ponds and their size in different parts of the country. Studies in the southern Bangladesh polder zone indicate that 59% of households have a pond in their homestead, a propor on that gradually decreases with lower land holding groups (Bloomer 2012). The average size of ponds in this area is 0.039 ha (CAARP-WorldFish 2009). Ponds were excavated to raise the base of their dwelling houses to avoid flooding and fish culture was o en not their primary func on (Belton and Azad 2012). Rather these ponds are used for mul ple purposes, including bathing, washing, watering livestock and irriga on for dike or homestead vegetable produc on (ADB 2004). Aquaculture is growing in Bangladesh and the trend of producing fish from homestead ponds is also increasing (DOF 2013). However, the average annual yield is s ll below 1.75 ton/ha (Jahan et al. 2010) and it drops to less than 1.0 ton/ha for either smaller or poorly managed ponds (ADB 2004). Belton and Azad (2012) made a cri cal analysis of fish produc on from those ponds and concluded that poorer households are less likely to own a pond than their be er-off counterparts, that any pond they do own is likely to be smaller than the average and that they are more likely to be managed at a lower intensity, and hence be rela vely unproduc ve, both in terms of total and per unit area output. Homestead ponds are o en surrounded by home gardens (Jahan 2011), which usually include mber and fruit trees and seasonal vegetables (Kea nge et. al. 2012). These trees o en shade a major surface area of the small ponds and constrain primary produc vity by restric ng sunlight (Brunea et al. 2003; Giovannini and Piedrahita 1994). Indian major carps are commonly grown in these ponds due to cultural preference (Belton and Azad 2012), a choice that contributes to the low produc vity. In some ponds farmers do not stock any fish and harvest is obtained only through rearing of naturally recruited fish that come into those ponds from rice fields or nearby open waters during the monsoon by rainwater flow or flooding. Raising produc vity by cu ng and trimming trees is not preferred by the farmers as the trees provide major energy requirements, usually in the form of firewood (GoB 2000), and many other benefits as well, par cularly for poor households. No specific research has been carried out in this region to explore op ons for management of shaded ponds, par cularly which fish species might grow be er under this shaded condi on and what type of pond management can allow them to grow more fish as well as to use pond water for other uses. 1.2 Finding solu ons with people In Bangladesh, solving local level aquaculture problems by the tradi onal extension service approach is highly constrained by very limited staff capacity of the Department of Fisheries and a produc on-oriented top-down approach (Lewis 1997; Alam and Thomson 2001). NGOs have tended to work with a target group of landless and marginal farmers, and have tried to priori ze poor women in par cular but their interven ons are very much project-bound and o en go with an imposed agenda, not focused on the socioeconomic context, and with limited inter-project learning sharing (Lewis 1997). However, there is a growing awareness that many of the constraints to increased produc on, consump on and incomes for the poor are primarily social; for example, unequal access to both technical and material inputs and other resources (Worby 1994). Also research in this domain is mostly done by the Na onal Agricultural Research Sta ons (NARS) with very limited collabora on with farmers and a tendency to ignore small farmers’ problems while focusing more on commercially important issues. There is also a gap between researchers and extension services, both government and NGOs. The room for innova on and exchange of ideas is thus not func oning well (Lewis 1997). Par cipatory ac on research (PAR) involves collabora ve research, educa on and ac on oriented towards innova on and social change (Kindon et al. 2007). However, in many research programs in Bangladesh people’s par cipa on is s ll limited and examples of farmers directly conduc ng their own research is largely absent. 1.3 Can women make a difference? In Bangladesh, women’s ability to generate income in the agricultural sector is severely constrained by their limited use, ownership, and control of produc ve physical and human capital (Quisumbing and Maluccio 2003) 266 and limited adult educa on (Ahmed et. al. 2007). However, par cipa on of rural women in agriculture has greatly increased in the past two decades, par cularly as labor in post-harvest ac vi es (Rahman 2000) and they are also widely engaged in agricultural work within the homestead area (Begum 1985; Abdullah 1985). There is evidence to show that directly raising women’s incomes can be instrumental in strengthening their status in society (Lewis 1997). However, very li le is known about the impact of engaging them in PAR where the primary goal is not income. Also, the outcomes of women farmer’s research capacity building through PAR for solving local-level farming problems are poorly documented in Bangladesh, especially in aquaculture. Based on these various challenges, the CGIAR Research Program on Aqua c Agricultural Systems, led by WorldFish, designed a program to increase produc vity from those small shaded ponds by conduc ng research with women. The approach is based on a method where farmers and scien st work together to not only solve a problem but also to develop the capacity and confidence of the farmers. This process is thought to empower the overall community and allow sharing of knowledge in a more convincing way to a diverse audience within the community. The primary aim of this research was to develop low input-based management prac ces to increase fish produc on in shaded ponds without interfering with the use of pond water for other household purposes. The overall aim was to improve household nutri on by increasing fish consump on, especially by iden fying suitable fish species composi on, and to increase women’s decision-making capacity for homestead farming. This paper presents preliminary results on the effec veness of this research-in-development approach to farmers’ learning and developing farmers’ capacity to analyze research results and share them with the community for greater impact. 2. Methods 2.1 Study area The study was conducted in eight villages under eight upazilas of seven administra ve districts in the southwestern part of Bangladesh (Table 1). Four villages were in freshwater and four in brackish water areas (Figure 1). Freshwater and brackish water areas were differen ated based on annual river water salinity of that area. River salinity remains at 0 ppt yearround in freshwater areas, whilst in brackish water areas the river water salinity varies considerably during the year, and in different regions. Generally, salinity starts to increase in late February and con nues to increase un l June reaching up to 25 ppt in the highly saline regions (i.e. Satkhira) and 5-10 ppt in low and medium saline parts (i.e. Barguna, Khulna). Salinity starts to drop down with the monsoon rainfall, eventually reaching 0 ppt in low and medium saline regions (i.e. Barguna, Khulna) and 5-7 ppt in highly saline regions. Legend: Freshwater areas Brackishwater areas Fig. 1. Study areas. 267 2.2 Developing a PAR model for this research This study was conducted jointly by two CGIAR Research Programs and three bilateral projects in order to allow for cross-project learning sharing and integra on of an aqua c agricultural system approach for solving common problems with people. PAR can only be successfully achieved through mechanisms that encourage mee ngs, joint reflec ons and the collec ve development of findings and conclusions (Funtowicz and Ravetz 1993; O mann 2005). This research therefore involved three interconnected layers (Figure 2)—scien sts, facilitators and farmer researchers—to ensure effec ve par cipa on of all stakeholders and thus a empt to bridge the chasm between science and people. The scien st team comprised a pool of WorldFish experts from disciplines of aquaculture, fisheries, biology, economics and gender to provide overall technical guidance, advisory support and coordina on of the research. The scien sts’ team interacted directly with the farmer researchers in designing the experiments and farmers’ research capacity building ac vi es. The facilitators were technical staff from the respec ve projects who were involved in daily communica on with the farmer researchers. The farmer researchers, all women, were the real actors of this research. They contributed to the process of research design, implementa on, data collec on, preliminary analysis and sharing the learning within the groups and beyond. -Technology options -A dvisor y sup port -C oordination Scientist Facilitator Plan Act Reflec t Share PM&E Women farmer researcher Fig. 2. PAR model for shaded pond research. The scien st team consulted with each community to iden fy together poten al interven ons for improving produc vity of ponds, with facilitators responsible to assist the fortnightly sharing mee ng organized by farmer researchers. Various events for sharing of learning were planned among and within areas to evaluate farmer research results, develop their capacity to analyze research findings by using PRA tools, to improve their confidence by sharing research findings with broader audiences and also to establish linkages within and with other communi es. 2.3 Pond and farmer selec on Ponds with a surface area of around five decimal and a minimum 60% shading at the sunniest me of the day were selected for this research. Farmer researchers, all women, were selected based on their interest in research, experience in pond management, access to a homestead pond and literacy level. A total of 96 women farmer researchers par cipated in the research (Table 1). 268 Table 1. Number of farmer researchers of the study by area Hub/Upazila Nagorkanda Babugonj District Faridpur Barisal Monirampur Rajapur Amtoli Ba aghata Kaligonj Shyamnagar Jessore Jhalokhathi Barguna Khulna Satkhira Satkhira Region No. of farmers 12 12 Fresh water (0 ppt salinity) 12 12 12 12 12 12 Brackish water 2.4 Farmer consulta on and experimental design At each community, consulta ons with farmers were conducted using different PRA tools. The focus of the early consulta ons was to make homestead resource profiles. Topics included: iden fying major homestead farming components; understanding the role of ponds in the homestead system and its integra on with other farming prac ces; SWOT analysis of pond aquaculture; iden fying farmers’ species preferences; iden fying preference for fish feeding; finding op ons for addressing mul ple use of homestead ponds; annual water calendar and water use from homestead ponds; women’s preferred ac vi es in pond management; current role of women in household decision-making including homestead agriculture; women’s perspec ve about nutri on; women’s level of par cipa on in agriculture input purchase and product marke ng; and possible role of women in par cipatory technology development and knowledge sharing. Consulta on outputs from four hubs of each region were merged together and a summary was prepared by the scien st team. Species selec on and pond management protocols were developed based on the consulta on summary and further shared with farmers before implementa on. Three polyculture treatments were designed for each hub with low stocking density (Table 2). Each treatment had four replica ons. In each treatment, species were selected under the categories of regular uptake, high value, fast growing and cultural preference. Twenty-five gm lapia were strategically stocked to allow for quick recruitment and 100 gm carp species were stocked to allow them to grow to marketable size within an eight-month meframe. Table 2. Selected fish in different treatments Category Regular uptake High value Brackishwater area Treatment 1 Treatment 2 GIFT Tilapia (25) Mystus gulio (50); Macrobrachi um rosenbergii (5) Fast growing Cultural preference GIFT Tilapia (25) Macrobrachi um rosenbergii (5); Heteropneus tes fossilis (25) Treatment 3 Freshwater area Brackishwater area Treatment 4 Treatment 5 Treatment 6 GIFT Tilapia GIFT Tilapia GIFT Tilapia Clarias batrachus (15) Heteropneus tes fossilis (2) Clarias batrachus (15) Anabas testudineus (50) Anabas testudineus (50) Cyprinus carpio (1) Hypophthal michthys molitrix (2) Anabas testudineus (50) Labeo rohita (2) Cyprinus carpio (1) Catla catla (2); Cirrhinus cirrhosus (4) GIFT Tilapia Note: Figures in parentheses indicate the stocking density per decimal. 269 A feeding strategy was developed based on the consulta on summary and was the same across all treatments and regions. Total feed requirement was calculated based on total biomass stocked in each pond. Feed composi on was 25% kitchen waste, 50% homemade feed prepared from locally available raw ingredients according to the guideline provided by the scien sts to keep consistency at all areas and 25% commercial extruded feed. 2.5 Pond management Before fish stocking, ponds were treated as follows. Unwanted fish were removed by using rotenone at 50 gm/dec; liming was conducted using dolomite and zeolite at 750 gm/dec and 25 gm/dec respec vely; and ponds were fer lized by urea, Triple Super Phosphate (TSP) and Muriate of Potash (MoP) at 150 gm, 75 gm and 75 gm per decimal, respec vely. Ponds were fenced by blue nets to avoid entry of predators and to prevent escape of fish during rain or floods. Ponds were stocked with GIFT Tilapia, Mystus gulio, Macrobrachium rosenbergii, Labeo rohita, Heteropneustes fossilis, Anabas testudineus, Clarias batrachus, Cyprinus carpio, Catla catla, Cirrhinus cirrhosis, Hypophthalmichthys molitrix and Barbonymus gonionotus at 25 gm, 1 gm, 5 gm, 100 gm, 5 gm, 2 gm, 5 gm, 100 gm, 100 gm, 100 gm, 100 gm, and 2 gm, respec vely. Fish seed of each species was supplied from a single source and all were collected from hatcheries. Feed was mixed type (i.e. 50% homemade, 25% commercial feed and 25% kitchen waste) with restricted feeding ra on based on weight gain. Feeding frequency was twice a day across all hubs. 2.6 Monitoring, record keeping and data collec on Farmer researchers and scien sts together set the monitoring protocol including daily feed applica on, observa on of feed u liza on, pond water color, water depth, fish breeding, abnormali es of fish health and behavior, fish disease, harves ng, consump on, sales and monthly body weight sampling. Besides, farmer researchers also listed interested visitors and any other challenges and inspira ons they encountered while conduc ng research. Informa on was documented in homestead pond record books and later these data were entered in a MS Access database by the WorldFish scien st team. Fortnightly mee ngs were organized at each hub among all farmers to share progress and challenges. The facilitator monitored the record keeping and guided sessions on technical aspects of aquaculture research, gender and nutri on. This also allowed the farmer researchers to compare between the treatments and developed their skills for sharing their observa ons. All discussions were documented in the mee ng registrar. Qualita ve and quan ta ve data on pond produc vity, learning and social change were collected through surveys towards the end of each farming season. 2.7 Data analysis Par cipatory data analysis was conducted by the farmers during the regional learning sharing workshop. In addi on, data was entered into MS Access so ware from which data were exported to MS Excel sheets for prepara on of different tables. Both descrip ve and sta s cal analysis was done. SPSS so ware was used for sta s cal analysis. To make a comparison among treatments as well as regions, mean and standard error have been calculated. In order to see the significance differences among the treatments and regions, an ANOVA test (Duncan test) was conducted. 270 2.8 Analyzing social changes Social change data was mainly collected through two sessions conducted between May and June (first session) and September and October (second session) of 2014. A simple descrip ve representa on of learning and decision-making has been fi ed into tables based on the response of women farmer researchers on those topics. 2.9 Par cipatory analysis of research findings by farmers At the end of the first year’s culture season two regional (one in each region) learning sharing and analyzing workshops were conducted where different PRA tools were applied to understand farmer researchers’ capacity to analyze research findings, se ng priori es for research outcomes and exercises to build their confidence on sharing their learning to a broader audience. In this paper some comparisons are made between farmer findings and scien st analysis in order to evaluate farmers’ capacity to analyze research findings and also to test the effec veness of the par cipatory process developed for this purpose. 3. Results 3.1 Produc on Fish produc on from shaded ponds in the 2013-14 season increased significantly (p<0.05) from the baseline of 2012-13 (Figure 3). Fish produc on was be er in freshwater regions than brackish water regions in terms of the amount of fish produced per ha though the difference was not sta s cally significant. While comparing among treatments across both regions, fish produc on was significantly (p<0.05) higher in treatment 4. This can be considered the best species composi on. Tilapia cons tuted 44.9% of total fish produc on, ranging from 32% to 77% in different treatments. Tilapia breeding also occurred within ponds, producing 0-4 cohort with varia on between ponds within the treatment and between treatments. 2012-13 40 2013-14 35 30 25 20 15 10 5 0 T1 T2 T3 T4 Treatment s T5 T6 T_mean Fig. 3. Annual fish produc on (kg/hh) from homestead shaded ponds. 3.2 Consump on Household fish consump on both in terms of amount consumed (Figure 4) and frequency of consump on increased significantly in 2013-14 compared to that of 2012-13 across all treatments in both the regions without much varia on. In 2013-14 in the research ponds household members consumed an average 76.9% of the fish produced, ranging from 69.6% to 87.0% in different treatments. Similarly, mean household weekly fish consump on frequency from their own pond increased from 0.09 in 2012-13 to 1.93 in 2013-14. 271 30 25 20 15 2012-13 10 2013-14 5 0 T1 T2 T3 T4 T5 T6 T_meam Treatments Fig. 4. Annual fish consump on (kg/hh) from homestead shaded ponds. 3.3 Pond water use Regarding water use in regular household ac vi es, 75% of households men oned no problem in their pond record book and mee ng registrar, as well as in impact surveys. The remaining 25% of women men oned that due to use of lime, water quality in their ponds had improved compared to previous years and thus there were benefits for household use. While exploring the impact of aquaculture on other homestead farming components 50% of women men oned there was no water use conflict or problem with other farming system components and the rest (50%) men oned that engaging in aquaculture research was rather helpful for them to manage other parts of the homestead farm. According to the la er 50%, women farmer researchers visited ponds more frequently to take care of fish during which me they also could water and take care of their vegetable and fruit gardens, which resulted in increased yields. 3.4 Decision-making capacity The collabora ve nature of this PAR approach greatly increased women farmer’s decision-making capacity on aquaculture management, though rela vely less advancement was observed on other agriculture- and income-related areas. The least progress was made on making decisions for marke ng (Table 2). There was not much varia on among the treatments in building farmers’ decision-making capacity. Only 1% of farmers reported decreased decision-making capacity and that was on another agricultural input selec on. Table 2. Change in farmers’ decision-making capacity on different ac vi es related to homestead farming Decision-making ac vity Increased # % No change # % Daily pond management Harves ng of fish Selling of fish 95 89 36 99.0 92.7 37.5 1 7 60 1.0 7.3 62.5 Income from sold fish Going to market Other agricultural input selec on (especially for homestead agriculture) 31 8 52 32.3 8.3 54.2 65 88 43 67.7 91.7 44.8 272 Decreased # % 0 0 0 0 0 0 1 0 0 1.0 3.5 Learning At the start of the project, farmers had limited knowledge on the context of research and aquaculture management in shaded ponds. At the end of the first cycle, the major learning as a result of conduc ng ac on research in conjunc on with researchers was mainly about shaded pond management, research capacity and leadership (Table 3). There was not much varia on among treatments in the process of learning. Table 3. Different aspects of farmers learning and number of farmers who acquired that learning Area of learning Pond management Research Research Area of learning Number of farmers Importance of stocking quality fish seed Importance of regular feeding Net fencing to prevent escape of fish & prevent entry of predator Selec on of species in shaded pond aquaculture Selec on of stocking density based on pond management 93 92 71 67 53 Removal of predatory fish Advantages of liming in pond water quality management Regular fish harves ng technique for cat fish Sampling for examina on of fish disease Advantage of using light trap Comparison and result analysis Importance of record keeping Sharing new learning with neighbors 38 36 29 11 7 89 88 72 Priori zing problems Planning for solving the problem Explaining result with reason Sharing with community (providing advice) Facing visitors Organizing mee ngs 61 58 49 96 56 31 3.6 Challenges Limited access to markets and inputs were iden fied as major challenges by 88 of the farmers. While 31 farmers men oned an increased workload, they enjoyed the ac vity. One farmer men oned poaching as a threat of increased produc on. Some challenges were more technical and encountered by less than 10 farmers. Those reported included a lack of natural food in the pond, predators (e.g. snake, frog, rep les), falling tree leaves and worsening pond water quality, higher feed price and flooding of the pond. 3.7 Evalua on of farmers’ capacity to analyze research findings Scoring and vo ng tools used during the workshop led farmers to conclude there was not much difference among the treatments; GIFT Tilapia was considered the best species in terms of produc on and recruitment and increased fish consump on was their most important benefit. Learning on how to solve aquaculture problems through PAR and improved importance at the household level was also considered important. By contrast, scien st analysis of the data showed that produc on was significantly higher in T4 and other treatments were similar in terms of fish produc on. Produc on and recruitment was best in GIFT Tilapia. 273 4. Discussion Engaging rural women in shaded pond aquaculture greatly increased fish produc on, household consump on and advanced women’s decision-making capacity in many sectors relevant to aquaculture, and homestead farming more generally. Fish produc on greatly increased from the baseline. This was due to selec on of high yielding GIFT lapia and other fast growing fish with the addi on of regular feedings. Mamun (2007) men oned GIFT can grow well, u lize feed under adverse environments and are stress tolerant. Feeding supplements low natural food in the rearing system and op mum feeding increases aquaculture produc on (Biswas 2012; Bosma 2011; Azim 2002). However, the produc on is s ll below na onal average produc on (Belton and Azad 2012), sugges ng room for further improvement. Most of the fish produced were consumed at the household level, which is expected to have a posi ve impact on human nutri on. Kabir (2015) also reported a strong correla on between homestead pond fish produc on and consump on. Other studies have also shown that increased produc on from pond aquaculture is closely linked to increased household consump on and nutri onal benefits (Bushamuka et al. 2005; Taher et al. 2004). Women’s decision-making capacity regarding aquaculture management greatly increased. Less progress was observed in some areas, such as financial and marke ng aspects (fish sales, income and going to market to buy inputs or sell fish). Women empowerment in the agricultural index under Feed the Future interven ons in southwest Bangladesh shows an advancement of 39% and also indicates several socio-economic factors influencing this advancement (Alkire et al.2013). In the first year of this PAR a larger improvement was no ced. This might be due to the PAR approach as opposed to a conven onal development interven on. This research created an excellent partnership between scien sts, experts of technology extension and farmers, and resulted in mul dimensional learning for farmer researchers and their increased capacity to lead in a community. Other research has also concluded that PAR involves a (rela vely) egalitarian partnership between expert researchers and other par cipants (Greenwood et al. 1993; Wallerstein 1999; Pain 2004). In this PAR the farmers’ research results were quite similar to those derived from the scien sts. This suggests that PAR has the poten al for transforma ve change from a farmer to a farmer researcher; such capacity building can help farmers become more effec ve in solving local farming problems. 5. Conclusions As this research work has illustrated, the philosophy of par cipatory ac on research provides technological and methodological answers for par cipatory and sustainable development. The approach goes through dialec cal processes that involve rural women, facilitators and scien sts and can build new knowledge on topics not yet explored. Such learning can be observed in the next PAR cycle. However, this requires sharing mul ple understandings to co-generate knowledge and developing reciprocal understanding that invokes all forms of ra onality, not just technical approaches. This study has already broadened cross-project learning and involved two-dimensional (technological and social) processes of development. However, sustainability targets can be achieved only a er developing complete knowledge of the process and determinants of social change in addi on to the technological aspects. Acknowledgements We are grateful to all par cipa ng farmer researchers and facilitators for conduc ng this research, keeping records, doing par cipatory analysis and suppor ng further analysis by sharing data and outcome stories. We would like to acknowledge the CGIAR Research Program on Aqua c Agricultural Systems (AAS) for providing the methodical and financial support to organize common learning events with the farmers. USAID-CSISA, the CGIAR Research Program on Aqua c Agricultural System, CGIAR Research Program on Climate Change Agriculture and Food Security, and USAID –AIN funded specific hubs of the research. 274 References Abdullah, T., 1985. Women in rice farming in Bangladesh and how technology programs can reach them. Women in Rice Farming. Proceedings of a Conference on Women in Rice Farming Systems. Interna onal Rice Research Ins tute (IRRI), Philippines. Gower Publishing Co. Ltd., Aldershot, 26-30 September 1983. ADB, 2004. An evalua on of small-scale freshwater rural aquaculture development for poverty reduc on. Opera ons Evalua on Department. Asian Development Bank, Manila, Philippines. Ahmed, A. U., Hill, R. V., Smith, L. C., Wiesmann, D. M., & Frankenberger, T. (2007). The World’s most deprived: Characteris cs and causes of extreme poverty and hunger. 2020 vision for food, agriculture, and the environment discussion paper 43. Alam, M.F., Thomson, K.J., 2001. Current constraints and future possibili es for Bangladesh fisheries. Food Policy 26, pp. 297–313 Alkire, S., Meinzen-Dick, R., Peterman, A., Quisumbing, A., Seymour, G., Vaz, A., 2013. The Women’s Empowerment in Agriculture Index. World Development Vol. 52, pp. 71–91 Azim, M.E., Verdegem, M.C.J., Khatoon, H., Wahab, M.A., Dam, A.A V., Beveridge, M.C.M., 2002. A omparison of fer liza on, feeding and three periphyton substrates for increasing fish produc on in freshwater pond aquaculture in Bangladesh. Aquaculture. Volume 212, Issues 1–4, 23 September 2002, Pages 227–243 Belton, B., Azad, A., 2012. The characteris cs and status of pond aquaculture in Bangladesh. Aquaculture 358–359: pp. 196–204 Belton, B., Karim, M., Thilsted, S., Jahan, K.M., Collis, W., Phillips, M., 2011. Review of aquaculture and fish consump on in Bangladesh. Studies and Reviews 2011–53. The WorldFish Center, Penang, p. 76. Begum, S., 1985. Women and technology: rice processing in Bangladesh. Women in Rice Farming. Proceedings of a Conference on Women in Rice Farming Systems. Interna onal Rice Research Ins tute (IRRI), Philippines. Gower Publishing Co. Ltd., Aldershot, 26-30 September, 1983. Biswas, G., De, D., Thirunavukkarasu, A.R., Natarajan, M., Sundaray, J.K., Kailasam, M., Kumar, P., Ghoshal, T.K., Ponniah, A.G., Sarkar, A., 2012. Effects of stocking density, feeding, fer liza on and combined fer liza on-feeding on the performances of striped grey mullet (Mugil cephalus L.) fingerlings in brackishwater pond rearing systems. Aquaculture. Volumes 338–341, Pages 284–292 Bloomer, J. 2012. Homestead Aquaculture in Bangladesh: Current Status and Future Direc ons. MSc disserta on in Aqua c Resource Management of King’s College London. h p://www.kcl.ac.uk/sspp/ departments/geography/study/masters/disserta ons/Disserta on-2012-Bloomer.pdf Bosma, R.H., Verdegem, M.C.J., 2011. Sustainable aquaculture in ponds: Principles, prac ces and limits. Livestock Science. Volume 139, Issues 1–2, Pages 58–68 Bushamuka, V.N., S. de Pee, A. Talukder, L. Kiess, D. Panagides, A. Taher, and M. Bloem. 2005. Impact of a homestead gardening program on household food security and empowerment of women in Bangladesh. Food and Nutri on Bulle n 26 (1): 17-25. Brunea, D.E. , Schwartza, G. , Eversoleb, A.G. , Colliera, J.A. , Schwedlerb, T.E., 2003. Intensifica on of pond aquaculture and high rate photosynthe c systems. Aquacultural Engineering. Volume 28, CAARP. 2009. Socio-economic profile and aquaculture rehabilita on impact on farming performance and livelihoods of the Cyclone Sidr affected households in the southern region of Bangladesh. WorldFish ReportIssues 1–2,Pages 65–86 275 DOF.2013. Na onal Fish Week 2013 Compendium (In Bengali). Department of Fisheries, Ministry of Fisheries and Livestock, Bangladesh. 144p. Funtowicz, S., Ravetz, J., 1993. Science for the post-normal age. Futures 25 (7), 739e755. Giovannini, P., Piedrahita, R.H., 1994. Modeling photosynthe c produc on op miza on for aquaculture ponds. Aquacultural Engineering. Volume 13, Issue 2, Pages 83–100 GOB (Government of Bangladesh). 2000. Forestry master plan: 3rd forestry forum background paper. TA no. 1355-BAN Asian Development Bank (ADB), Dhaka, Bangladesh Greenwood, D.J., Whyte, W.F., Harkavy, I., 1993. Par cipatory ac on research as a process and as a goal. Human Rela ons 46, 175–192. Huda, K.M.S., Atkins, P.J., Donoghue, D.N.M., Cox, N.J., 2010. Small water bodies in Bangladesh. Area 42 (2), 217–227 Jahan, M.K., Pemsl, D.E. 2011. The impact of integrated aquaculture–agriculture on small- scale farm sustainability and farmers’ livelihoods: Experience from Bangladesh. Agricultural Systems 104: Jahan, M.K., Ahmed, M., Belton, B., 2010. The impacts of aquaculture development on food security: lessons from Bangladesh. Aquaculture Research 41 (4), 481–495. Kabir, K.A., Karim, M., Phillips, M., 2015. Homestead farming system in southern Bangladesh. WLE Technical Report Kindon, S., Pain, R., Kesby, M., 2007. Par cipatory ac on research approaches and methods : connec ng people, par cipa on, and place. Routledge. ISBN 0-203-93367-2 Master e-book ISBN Kea nge, J. D. H., Chadha, M. L., Hughes, J. d.'. A., Easdown, W. J., Holmer, R. J., Tenkouano,A.,Yang, R.-Y.,Mavlyanova, R., Neave, S.,Afari-Sefa, V.,Luther, G.,Ravishankar,M., Ojiewo,C., Belarmino, M., Ebert, A., Wang, J. F., & Lin, L. J. (2012). Vegetable gardens and their impact on the a ainment of the millen- nium development goals. Biological Agriculture and Hor culture, 28(2), 71–85. Kranzlin, I., 2000. Pond management in rural Bangladesh: problems and possibili es in the context of the water supply crisis. Natural Resources Forum 24, 211–223 Lewis, D., 1997. Rethinking aquaculture for resource-poor farmers: perspec ves from Bangladesh. Food Policy, Vol. 22, No. 6, pp. 533–546 Li le, D., Karim, M., Turongruang, D., Morales, E., Murray, F., Barman, B., Haque, M., Kundu, N., Belton, B., Faruque, G., Azim, E., Islam, F., Pollock, L., Verdegem, M., Leschen, W., Wahab, A., 2007. Livelihood impacts of ponds in Asia—opportuni es and constraints. In: van der Zijpp, A., Verreth, J., Tri, L.Q., van Mensvoort, R. (Eds.), Fishponds in Farming Systems.Wageningen Academic Publishers, pp. 177–202. Mamun, S.M., Focken, U., Becker, K., 2007. Comparison of metabolic rates and feed nutrient diges bility in conven onal, gene cally improved (GIFT) and gene cally male (GMNT) Nile lapia, Oreochromis nilo cus (L.). Compara ve Biochemistry and Physiology Part A: Molecular & Integra ve Physiology Volume 148, Issue 1, Pages 214–222392–402 Pain, R., 2004. Social geography: par cipatory research. Progress in Human Geography 28, 652–663. O mann, G., 2005. Agroecología y Sociología histórica desde La noamérica: elementos para el análisis y potenciación del movimiento agroecológico: el caso de la provincia argen na de Santa Fe. Córdoba University, Cordoba. 276 Quisumbing, A. R., & Maluccio, J. A. (2003). Resources at marriage and intrahousehold alloca on: Evidence from Bangladesh, Ethiopia, Indonesia, and South Africa. Oxford Bulle n of Economics and Sta s cs, 65(3), 283–327 Rahman, S., 2000. Women's employment in Bangladesh agriculture: composi on, composi on, determinants and scope. Journal of Rural Studies vol. 16. Pp 497-507 Taher, A., A. Talukder, N.R. Sarkar, V.N. Bushamuka, A. Hall, S. de Pee, R. Moench-Pfanner, L. Kiess, and M.W. Bloem. 2004. Homestead gardening for comba ng vitamin A deficiency: The Helen Keller Interna onal, Bangladesh, experience. In Allevia ng malnutri on through agriculture in Bangladesh: Biofor fica on and diversifica on as sustainable solu ons, ed. N. Roos, H.E. Bouis, N. Hassan, and K.A. Kabir, Washington, D.C.: Interna onal Food Policy Research Ins tute. Wallerstein, N., 1999. Power between the evaluator and the community: research rela onships within New Mexico's healthier communi es. Social Science & Medicine 49, 39–53. Worby, E. (1994) Hi ng hairs and spli ng targets: anthropological perspec ves on fish culture technology transfer through NGOs in Bangladesh. Rockefeller Founda on Conference paper, Addis Ababa 277 278 Sec on 5 Aquaculture 279 Produc vity, diversifica on and resilience of saline aquaculture systems in coastal southern Bangladesh K.A. Kabir 1, S.B. Saha2, M. Karim1, C.A. Meisner1 and M. Phillips3 1 WorldFish, Bangladesh, k.kabir@cgiar.org, m.karim@cgiar.org, c.meisner@cgiar.org 2 Bangladesh Fisheries Research Ins tute, Bangladesh, sbikashsaha@yahoo.com 3 WorldFish, Malaysia, m.phillips@cgiar.org Abstract The CGIAR Challenge Program on Water and Food conducted research in high salinity areas of the lower Ganges delta of Bangladesh in 2012 and 2013 to increase the produc vity, profitability, diversity and resilience of aqua c agricultural farming systems. This paper reports on research on the diversifica on of shrimp-dominated low produc ve gher farming systems by tes ng crop rota on and incorpora on of fish into the farming systems. Farmers’ exis ng prac ce was used as a control treatment and compared to two improved systems: rota onal monoculture and polyculture systems in outdoor ponds (ghers). Tiger shrimp (Penaeus monodon) produc on in 2012 and 2013 in control, monoculture and polyculture treatment systems was 210, 556, 404 and 390, 857, 567 kg/ha, respec vely. Fish produc on in those years was 728, 2,367, 3,322 and 659, 3,308, 3,560 kg/ha, respec vely. Profitability in both of the improved systems was significantly higher than current farmer prac ce. The economic return from rota onal polyculture was significantly higher than monoculture in 2012, but there was no significant difference in 2013. Species combina on of shrimp- lapia in the dry season and carp-ca ish in the wet season appeared most profitable and resilient (as assessed against the risk of disease) providing new opportuni es for diversifica on of farming systems in the area. The level of profitability increased from Bangladesh Taka (BDT) 2,000/ha/year under exis ng farmer prac ce to BDT 248,000/ha/year and BDT 265,000/ha/year in monoculture and polyculture systems, respec vely. Water drainage and maintaining expected water depth were key challenges for implemen ng these improved systems. Key message: Rota onal polyculture offers an important opportunity for sustainable intensifica on of shrimp-dominated farming systems in the high salinity areas of the lower Ganges delta. Diversifica on of brackishwater aquaculture systems led to significant increases in produc vity and profitability. Keywords: shrimp, monoculture, polyculture, crop rota on, profitability, salinity 1. Introduc on Coastal aquaculture is not a recent prac ce in Bangladesh (Swapan & Gavin 2011). For centuries local people have prac ced tradi onal coastal aquaculture to grow shrimp and fish (Ahmed et al. 2002; Alauddin and Akhter Hamid 1997). A number of chronological historical events were associated with the emergence of commercial shrimp farming in Bangladesh (Swapan & Gavin 2011). Favorable ecological and clima c condi ons, availability of wild shrimp post larvae (PL), cheap labor, unused land and plenty of salt water have contributed to keeping shrimp farming investment costs very low. With strong interna onal market demand, shrimp farming has become very popular in the coastal region of Bangladesh. For some, it appeared as a good source for quick money-making for the coastal community. As a result, shrimp farming became one of the key economic ac vi es in the coastal districts of Cox’s Bazar, Khulna, Bagerhat and Satkhira, and has been expanding in other districts too (Ministry of Water Resources 2006). 280 The development of shrimp farming has been accompanied by posi ve and nega ves impacts. Rural households including small and marginal farmers and landless poor have benefited from shrimp produc on due to a wide range of livelihood opportuni es (Ahmed et al. 2010). The favorable domes c policy environment and emerging global market opportuni es reinforced the commercializa on of shrimp culture in Bangladesh (Ahmed et al. 2002). The value of shrimp exports increased from about US$322 million in 2000 to around US$467 million in 2012 (DOF 2013), contribu ng 2% of total na onal export (DOF 2013). Shrimp stands as second largest export item of the country (Azad et al. 2009). However, massive development of shrimp also had nega ve influences on livelihoods and ecological func oning (Paul & Vogl 2011). Several authors have already expressed doubts about the sustainability of current shrimp farming (Hein 2002; Paez-Osuna et al. 2003; Hall 2004; Chowdhury et al. 2006; Azad et al. 2009). In recent years disease outbreak has been recognized as the biggest obstacle to the development of shrimp aquaculture in Bangladesh, causing massive economic loss (Alam et al. 2007; Miah et al. 2010; Karim et al. 2012). Under this situa on farmer field-based research is needed to find ways out to improve the saline aquaculture systems of Bangladesh. The CGIAR Research Program on Water and Food collaborated with farmers and conducted field-based research in high salinity areas of southwestern Bangladesh between 2012 and 2013 to seek ways to make coastal brackishwater aqua c agricultural systems more produc ve, profitable, diversified and resilient. Experiments were designed considering changes in saline regimes to explore op ons to improve exis ng brackish water aquaculture systems, with a focus on rota onal monoculture and polyculture aquaculture farming systems to allow for crop rota on and risk reduc on. The hypotheses underlying this research were: Annual produc vity can be increased by improving management of shrimp-based farming systems. Rota onal polyculture will diversify the system and reduce economic loss due to shrimp disease in the dry season. Species composi ons can be improved to be er u lize different pond niches and contribute to higher yield. 2. Methods 2.1 Study area The study was carried out in the southern part of polder 3 under the administra ve upazila of Kaliganj of Satkhira Districts (Figure 1). This is one of the major shrimp farming areas (Ministry of Water Resources 2006) that is categorized as a high salinity area (SRDI 2010). 281 Fig. 1. Study area. 2.2 Experimental design Three treatments—a control (farmers’ exis ng prac ce), monoculture and polyculture—each with four replica ons were tested in 2012 and 2013. The control treatment was designed based on farmers’ current management prac ce, developed through three consulta on mee ngs with more than 50 shrimp farmers in the southern part of polder 3. The monoculture and polyculture treatments were developed considering crop rota on and species composi on. A randomized concrete block design (Shieh 2004) was followed for distribu on of treatments and replica ons within the experimental area. 2.2.1 Control treatment The control treatment was based on exis ng farmer prac ces. Ponds were maintained at a water depth of 50 cm and the ponds were dried out at the end of each year. Mul ple stocking of ghers was prac ced star ng in late February with Penaeus monodon, Metapenaeus monoceros, Chelon subviridis, Rhinomugil corsula and Oreochromis mossambicus at stocking densi es of 2, 4, 0.25, 0.25 and 0.25 m-2, respec vely, followed by stocking of Penaeus monodon in each month un l September at a density of 0.5 m-2. Selec ve harves ng was carried out from April un l the end of November. At the end of August Labeo rohita, Catla catla, Cyprinus carpio and Cirrhinus mrigala was stocked at a density of 0.25m-2 for each species. 282 2.2.2 Monoculture treatment In this treatment, a greater water depth of 70-100 cm was maintained and the produc on cycle was divided into three crop sub-cycles allowing for crop rota on in response to changes in salinity. In 2013, the first cycle was Penaeus monodon stocked at a density of 3 m-2 followed by Oreochromis nilo cus (monosex) and Macrobrachium rosenbergii at a density of 4 m-2 and 3 m-2 respec vely. Based on the observa on of disease outbreak and shrimp price in local markets in 2013, the dry season was subdivided into two short cycles of shrimp with a target of harves ng small-sized shrimps by growing at a rela vely high density. So, Penaeus monodon was stocked at a density of 5 m-2 for 70 days in the first and second cycles followed by O. nilo cus (monosex) at a density of 5 m-2 from August to November. 2.2.3 Polyculture treatment This treatment was designed to maintain an improved water depth of 70-100 cm and the produc on cycle was divided into 3 crop sub-cycles allowing for crop rota on in response to changes in salinity. The treatment was intended to keep a backup crop if shrimp disease occured, thus reducing the economic loss and also producing fish for local consump on. In 2013 the first cycle was for 105 days star ng from the end of February and was stocked with Penaeus monodon and Oreochromis nilo cus (monosex) at a density of 3 m-2 & 2 m-2, respec vely. The second cycle for 90 days was stocked with Oreochromis nilo cus (monosex) and Mystus gulio at a density of 3 m-2 & 2 m-2, respec vely. This was followed by Macrobrachium rosenbergii with Mystus gulio at a density of 2 m-2 for each. Like the monoculture treatment in 2013, the dry season was subdivided into two short cycles of shrimp with a target of harves ng small-sized shrimps by growing at a rela vely high density. So, Penaeus monodon were stocked at a density of 5 m-2 for 70 days in the first and second cycles followed with O. nilo cus (monosex) at a density of 3 m-2. This was followed by Labeo rohita, Clarius bactracus and Heteropneutes fossilis at densi es of 0.25 m-2, 1 m-2 and 1 m-2, respec vely, from August to the end of November. 2.3 On farm experimental setup A total of twelve ponds with area between 866 and 1,463 m 2 were prepared. Each pond had an independent water inlet. The control treatments were placed in a shallower part of the experimental area than the other eight ponds. Each year at the beginning of the produc on cycle pond bo oms were exposed to sunlight, allowed to sun-dry, plowed later on and treated with bleaching powder to disinfect the pond. Ponds were filled with dal water by sieving through nylon net up to 50 cm for control treatment ponds and up to 100 cm for monoculture and polyculture ponds. Predatory and unwanted fishes were killed using rotenone at 1.5 ppm and removed from the ponds. Then water of the ponds was treated with dolomite at 20 ppm and fermented molasses at 25kg/ha. Water of the ponds was fer lized with urea at 5.00 ppm and TSP at 2.50 ppm. Water of the ponds was treated with dolomite and Zeolite and fer lized with urea and TSP whenever necessary. Evaporated and seepage water was replenished from the dal water in 2012 and with treated water from the reservoir in 2013. A er each culture cycle ponds were dewatered for complete harves ng in monoculture and polyculture ponds and control treatment ponds were dewatered at the end of each year. 2.4 Source of shrimp and fish seed applied Hatchery produced shrimp PL 15 was stocked in control treatment ponds and PCR tested, WSSV nega ve shrimp PL 15 were stocked in monoculture and polyculture treatments. Wild harvested Metapenaeus monoceros, Chelon subviridis and Rhinomugil corsula seed were collected from the local market. Oreochromis mossambicus, Labeo rohita, Catla catla, Cyprinus carpio and Cirrhinus mrigala for control treatment were collected from the local fry traders. The average weight of the carps varied from 80 to 100 g. Mystus gulio and Macrobrachium rosenbergii were obtained from the Bangladesh Fisheries Research Ins tute (BFRI) hatchery with an average weight of 0.33 g and 0.08 g respec vely. Oreochromis nilo cus (monosex) were obtained from a BRAC hatchery and Labeo rohita, Clarius bactracus and Heteropneutes fossilis were collected from Muktesawery Hatchery, Jessore and weighed on average 0.5 g, 100 g, 8 g and 5 g, respec vely. 283 2.5 Nursing In control treatments shrimp and fish were released directly into the grow out ponds without nursing. Shrimp and fish for monoculture and polyculture experiments were nursed either in earthen pond nurseries or an in-pond nursery for a period of 15 to 25 days. Macrobrachium rosenbergii was nursed for 60 days in an outdoor nursery built adjacent to the experimental ponds. 2.6 Feeding management No supplementary feeding was provided in control treatments. Shrimps were fed with commercial shrimp pellets. M. gulio, Clarius bactracus and Heteropneutes fossilis were fed by commercial ca ish feed and lapia by floa ng feed. The quan ty of feed applied was adjusted a er es ma ng the total biomass of shrimp or fish at seven-day intervals. A feeding check tray was placed in shrimp and freshwater prawn ponds for monitoring feed intake and adjustment if needed. Feeds were applied every morning, noon and evening and spread evenly across the water surface. 2.7 Monitoring water quality and fish health and growth Dissolved oxygen and pH were monitored daily morning and evening. Temperature, transparency and water depth were monitored daily at noon and alkalinity on a weekly basis. Sampling of shrimp and fish bodyweight was carried out at seven-day intervals. 2.8 Disease occurrence Disease occurrence was also monitored and three characteris cs were used to define a diseased pond: (i) symptoms of shrimp or fish disease were recorded; (ii) control measures did not work; and (iii) disease caused increasing mortality at more than 10% per day. Ponds mee ng all three characteris cs were declared as diseased ponds. 2.9 Economic data Records of the quan es and costs of all inputs were recorded for each pond. Labor and land leasing cost was calculated based on the actual cost. Average sale value for P. monodon, M. rosenbergii O. nilo cus , M. gulio , C. bactracus, H. fossilis, M. monoceros, C. subviridis, R. corsula, O. mossambicus, L. rohita, C. catla, C. carpio and C. mrigala was respec vely Bangladesh Taka (BDT) 550, 500, 100, 250, 200, 350, 300, 250, 250, 70, 150, 150, 150 and 150. This price was set based on average market price for the study period. All economic data was in BDT (1 BDT = 0.013 USD in December 2014). 2.10 Data analysis Data analysis was conducted using IBM SPSS Sta s cs 19. The first hypothesis was explored by comparing annual system produc on between the treatments and years. A suitable species combina on was determined based on the profitability, risk of crop loss (defined as frequency of disease occurrence in two years study and the economic loss due to the disease), salinity level of the locality, preference of local farmers (as determined by farmer par cipatory consulta on post harvest) and availability of inputs. Profitability was calculated including labor cost and excluding labor cost. The Duncan test, one-way ANOVA and one sample t test were used for sta s cal significance. 2.11 Ini a ve for green ghers Observing the shi in salinity and salinity tolerance of local grass and medicinal plant neem (Azadirachta indica), an ini a ve was taken in the second year (2013) to integrate plant products into the aquaculture 284 produc on facility by plan ng neem and local grass on gher dikes. Vegetables were also grown on dikes in the wet season. This was done in addi on to the experimental research and the data were not analyzed cri cally. This addi onal component provided an interes ng observa on for its poten al to further develop coastal aquaculture farming systems. 3. Results 3.1 System produc vity The experimental results from 2012 and 2013 clearly show that annual yield from brackishwater aquaculture system can be significantly increased by adop ng improved management prac ces (Figure 2). 2012 4500 3000 2500 3500 3000 Shrimp Fish Yield (kg ha-1) Yield (kg ha-1) 3500 2013 4000 4000 2000 1500 Shrimp Fish 2500 2000 1500 1000 1000 500 500 0 0 Farmer’s prac ce Monoculture Polyculture Farmer’s prac ce Monoculture Polyculture Fig. 2. Annual yield (kg ha -1year -1 ) of fish and shrimp in 2012 and 2013. Shi ing from year round shrimp-focused aquaculture to a season-specific annual produc on system not only increased fish produc on but also resulted in dry season-only shrimp produc on that was more produc ve than the yearround control treatments. (Table 1a and 1b). Shrimp together with freshwater prawn in 2012 and shrimp alone in 2013 was significantly higher than that of the control treatment and total fish produc on was also significantly higher in both the years. Table 1a. Cycle-wise yield in kg ha-1 in 2012 Culture pa erns Cycle-1 Shrimp Cycle-2 Cycle-3 Total annual yield Tilapia Tilapia Tengra Prawn Tengra Farmers’ prac ce Monoculture 197.77 - 2367.48 - 363.63 Polyculture 1163.41 1519.78 352.34 231.74 172.64 - Shrimp *209.84 556.40 Fish 728.14 2367.48 286.50 404.38 3322.04 Table 1b. Yield of the different components during each cropping cycle in kg ha -1 (from 2013) Culture pa erns Cycle-1 Shrimp Tilapia Cycle-2 Shrimp Tilapia Cycle-3 Magur + Total annual yield Shrimp Fish Tilapia/ singh rohu 659.41 3307.71 3560.02 Farmers’ prac ce Monoculture 565.62 - 291.77 - - 390.05 *3307.71 857.39 Polyculture 1744.45 193.19 777.90 557.63 **480.05 566.83 373.64 285 3.2 Suitable species combina on Shrimp- lapia in dry season aquaculture and carp-ca ish (singh-magur) in wet season had significantly higher yield (p<0.05) than other combina ons. The farmers also ranked them higher during par cipatory evalua on following harvest. 3.3 Cost benefit analysis Profitability was significantly higher (p<0.05) in recommended treatments (i.e. monoculture and polyculture) than that of the control treatment (i.e. farmers’ prac ce) in both 2012 and 2013. Comparing profitability between the two recommended treatments, monoculture and polyculture, polyculture treatment was significantly higher (p<0.05) than monoculture treatment in 2012 (Figure 3) and there was no significant varia on in 2013. 700 BDT 1000 ha⁻¹ 500 400 Variable cost Variable cost Total return Gross margin 600 Total return Gross margin 300 200 100 500 400 300 200 100 0 -100 2013 700 BDT 1000 ha⁻¹ 600 800 2012 Farmer’s prac ce Monoculture 0 Polyculture -100 Farmer’s prac ce Monoculture Polyculture Fig. 3. Variable cost, total return and gross margin of different treatments in 2012 and 2013 including labor cost. Profitability including labor cost for farmers’ prac ce (i.e. control treatment) was nega ve in 2012, and significantly (p<0.05) increased in 2013. Profitability of control treatment excluding labor cost was 65,000 and 94,000 BDT ha-1year-1 respec vely in 2012 and 2013. 3.4 Environmental considera on for aquaculture Salinity is a major environmental factor for coastal aqua c agricultural systems and one of the determining factors for crop selec on. There is a shi in salinity levels between dry and wet seasons (Figure 4), which provides plenty of opportunity to diversify coastal aquaculture and also provides opportuni es to integrate aquaculture with dike cropping systems. 25.0 20.0 15.0 2012 10.0 2013 5.0 0.0 F M A M J J A Fig. 4. Water salinity from February to November in 2012 and 2013. 286 S O N Water depth and drainage, though cri cal for aquaculture, are difficult to maintain in prac ce for individual plots. In the improved management prac ce of the experimental setup the prepara on was to keep one meter water in the ponds throughout the year; in prac ce from March to early July it was between 70-80 cm and from July to October it was between 90-105 cm. This further reduced in November to 65 cm in both the years. In both years, dissolved oxygen at beginning of each cycle remained between 7-8 ppm and dropped even to <2 ppm in the morning in la er days of the culture period due to increased total biomass in the ponds. Under that limited dissolved oxygen level water exchange was not possible due to limited drainage facili es in the community level water management system. Weekly average pH in fed ponds under monoculture and polyculture was always between 8.5 and 9.5 in the a ernoon, showing highest fluctua ons during the monsoon period. Water temperature in February remains below 15oC and drops again to 20oC in November. This makes aquaculture unfavorable from December to January due to low water temperatures. Local varie es of grass and neem grew well on gher dikes. The grass grown on the gher dikes of 1 ha experimental plots provided sufficient fodder for four to five goats. Neem has poten al for use in the herbal industry as a raw material. Based on rainfall from August to January, one to two crops of bo le gourd, ladies finger and cucumber can be grown on improved gher dikes with higher width and height. These ini al experiments show poten al for further crop produc on and overall system produc vity. 4. Discussion System produc vity and profitability in brackishwater aquaculture systems can be increased significantly through diversifica on and be er management. Rota onal polyculture increases fish produc on for local market and reduces risk of economic loss from dry season crops. Crop rota on helps reduce water salinity. Produc on of shrimp in both monoculture and polyculture systems was two to three mes higher than the na onal average produc on men oned by Belton et al. (2011), which was between 160 and 230 kg/ha/year. However, this level of intensifica on does not risk water pollu on. An extensive system of shrimp produc on up to 1 mt/ha/year can neutralize the effluent through the natural process of the ecosystem (Wahab et al. 2003). Though produc vity of shrimp systems increased it was s ll much lower than that of other Southeast Asian countries who produce 1.2 to 10 mt/ha/yr (FAO 2012). Environmental condi ons including poor water management facility are important limi ng factors for increased produc on and intensifica on of the system. Manjurul et al. (2014) also listed these factors as playing a role in shaping coastal aquaculture produc on including shrimp. However, further research is needed on the poten al limits of sustainable intensifica on within brackishwater aquaculture systems in Bangladesh. Crop rota on and polyculture provide opportuni es for improving the sustainability of the coastal aqua c agricultural system. Other researchers (Nesar et al. 2008, Nesar et al. 2010; Wahab et al. 2012) has also provided similar findings, no ng that such approaches can provide resilience against a changing climate (Nesar 2013). 5. Conclusions and recommenda ons Considerable opportuni es exist for improving produc vity, diversifica on and resilience of brackishwater aquaculture in the coastal saline zone of Bangladesh. Technologies appear to be very “adoptable” by farmers, as indicated by the number of farmers around the experimental plots that were voluntarily adop ng some of the prac ces (e.g. shrimp- lapia polyculture in the dry season, mixed culture of carp and ca ish in the wet season, growing vegetables on dikes). 287 Water management is also a key factor in improving aquaculture within the region, and adop on of community water management could further improve op ons for aquaculture. Preliminary experiments with growth of vegetables and fodder on dikes also suggests it is possible to integrate vegetable produc on on dikes in the wet season and support livestock by growing grass on the dikes. Saline water represents an asset to the country that needs to be more efficiently used, by both aquaculture and aqua c-agricultural farming systems. To meet growing fish demand (World Bank 2013) further intensifica on of such systems is needed in a sustainable way. Be er understanding of market systems, improved access of farmers to quality inputs and capital investment also needs to be explored for scaling of improved farming systems and crea ng development outcomes and impact from this research. Acknowledgements We would like to acknowledge the CGIAR Challenge Program on Water and Food for providing fund for conduc ng this study and the CGIAR Research Program on Aqua c Agricultural Systems for further collabora on and support in analyzing these data. We also thank the WorldFish AIN project for providing addi onal funding for establishing experimental ponds. We are grateful to Mr. Habibur Rahman and Md. Faruk Hossen for par cipa ng in the research with their land, me and skills. References Ahmed, Asad, S., Mallick, D.L., Liaquat, A. M., and Rahman, A., 2002. Policy Research for Sustainable Shrimp Farming in Asia: Literature Review on Bangladesh Shrimp. Bangladesh Center for Advanced Studies, Dhaka. Alam, S., Pokrant, B., Yakupi yage, A., Phillips, M., 2007. Economic returns of disease- affected extensive shrimp farming in southwest Bangladesh. Aquacult. Int. 15, 363-370. Alauddin, Mohammad, Akhter Hamid, M., 1997. Shrimp Culture in Bangladesh with Emphasis on Social and Economic Aspects. Azad, A., Jensen, K., Lin, C., 2009. Coastal aquaculture development in Bangladesh: unsustainable and sustainable Experiences. Environ. Manag. 44, 800-809. Belton, B., Karim, M., Thilsted, S., Jahan, K.M., Collis,W., Phillips,M., 2011. Review of aquaculture and fish consump on in Bangladesh. Studies and Reviews 2011–53The WorldFish Center, Penang 76. Chowdhury, M.A., Shivako , G.P., Salequzzaman, M., 2006. A conceptual framework for the sustainability assessment procedures of the shrimp aquaculture industry in coastal Bangladesh. Int. J. Agric. Resour. Governan. Ecol. 5, 162-184. DOF.2013. Na onal Fish Week 2013 Compendium (In Bengali). Department of Fisheries, Ministry of Fisheries and Livestock, Bangladesh. 144p. FAO, 2012. FAO yearbook. Fishery and Aquaculture Sta s cs. h p://www.fao.org/3/a-i3740t/index.html Hall, D., 2004. Explaining the diversity of Southeast Asian shrimp aquaculture. J. Agrar. Change 4, 315-335 Hein, L., 2002. Toward improved environmental and social management of Indian shrimp farming. Environ. Manag. 29, 349-359. Karim, M., Sarwer, R.H., Brooks, A.C., Gregory, R., Jahan, M.E., Belton, B., 2012. The incidence of suspected white spot syndrome virus in semi-intensive and extensive shrimp farms in Bangladesh: implica ons for management. Aquac. Res. 43 (9), 1357–1371. Karim, M., Sarwer, R.H., Phillips, M. and Belton, B., 2014. Profitability and adop on of improved shrimp farming technologies in aqua c agricultural systems of southwestern Bangladesh. Aquaculture 428-429 (2014) 61-70 288 Miah, M.G., Bari, M.N., Rahman, M.A., 2010. Resource degrada on and livelihood in the coastal region of Bangladesh. Front. Earth Sci. China 4, 427-437. Ministry of Water Resources, 2006. Coastal Development Strategy.Water Resources Planning Organisa on, Dhaka. h p://www.mowr.gov.bd/Documents/COASTAL%20DEVELOPMENT%20STRATEGY%20%28English%29.pdf Nesar, A., Demaine, H., Muir, J.F., 2008. Freshwater prawn farming in Bangladesh: history, present status and future prospects. Aquaculture Research 39, 806e819. Nesar, A., Allison, E.H., Muir, J.F., 2010. Rice-fields to prawn farms: a blue revolu on in southwest Bangladesh? Aquaculture Interna onal 18, 555e574. Nesar, A., 2013. Linking prawn and shrimp farming towards a green economy in Bangladesh: Confron ng climate change. Ocean & Coastal Management 75 (2013) 33-42 Paez-Osuna, F., Gracia, A., Flores-Verdugo, F., Lyle-Fritch, L.P., Alonso-Rodraguez, R., Roque, A., Ruiz-Fernandez, A.C., 2003. Shrimp aquaculture development and the environment in the Gulf of California ecoregion. Mar. Pollut. Bull. 46, 806-815. Paul, B.G. and Vogl, C.R., 2011. Impacts of shrimp farming in Bangladesh: Challenges and alterna ves. Ocean & Coastal Management 54 (2011) 201-21 Shieh, G. 2004. The effec veness of randomized complete block design. Sta s ca Neerlandica. Vol. 58, nr. 1, pp. 111–124 SRDI 2010. Soil Salinity Report 2010 of Bangladesh, Soil Resources Development Ins tute, Dhaka, Bangladesh Swapan, M. S. H., and Gavin, M. 2011. A desert in the delta: Par cipatory assessment of changing livelihoods induced by commercial shrimp farming in Southwest Bangladesh. Ocean & Coastal Management 54 (2011) 45-54 Wahab, M.A., Bergheim, A. and Braaten, B., 2003. Water quality and par al mass budget in extensive shrimp ponds in Bangladesh. Aquaculture 218 (2003) 413-423 Wahab, M.A., Nahid, S.A., Ahmed, N., Haque, M.M., Karim, M., 2012. Current status and prospects of farming the giant river prawn (Macrobrachium rosenbergii) in Bangladesh. Aquaculture Research 43, 970-983. World Bank, 2013. Annual Report. h p://web.worldbank.org/WBSITE/EXTERNAL/EXTABOUTUS/EXTANNREP/EXTANNREP2013/0,,menuPK:93048 95~pagePK:64168427~piPK:64168435~theSitePK:9304888,00.html 289 Community-based fisheries management: Improving fish biodiversity in inland fisheries of Bangladesh M. G. Mustafa WorldFish, Bangladesh, g.mustafa@cgiar.org Abstract The Community Based Fisheries Management (CBFM) approach in Bangladesh aims to promote the sustainable use of inland fisheries by empowering fisher communi es to manage their resources. An inves ga on into the impact of this approach underscored the strong dependence of mul species and abundance on fisheries management. Using data compiled for impact assessment, fish produc on, abundance and biodiversity were examined in detail for different habitats. When comparing CBFM sites with control sites trends in fish produc on, abundance and biodiversity with me were found to be significantly higher (p<0.01) for the former. Of the 64 project sites, 49 of them showed an upward trend in produc on, 46 showed an upward trend in fish abundance and 48 demonstrated an upward trend in biodiversity. Ten of the 16 control sites showed a declining trend in produc on, abundance and biodiversity with me. In 2005, significant differences (p<0.05) in species assemblages were found in two out of five regions for two types of habitats at CBFM and control sites. Species assemblages in floodplain habitats in the north and river habitats in the east regions were found to be richer and more abundant at CBFM sites compared to control sites. Considered together, this evidence suggests that the CBFM approach also benefits wetlands biodiversity and resource sustainability. Key message: The community-based fisheries management approach might be considered to improve fish produc on, abundance and biodiversity in inland fisheries. Keywords: community-based fisheries management, biodiversity, abundance, assemblages, floodplain 1. Introduc on The importance of Bangladesh’s inland fisheries for the livelihoods and food security of the poor and landless is widely acknowledged. Fish from Bangladesh’s vast inland waters are vital to millions of poor people but landings and species diversity are believed to be declining (Ahmed et al. 1997). Fishers and experts have iden fied poten al causes for this decline including habitat degrada on due to silta on and change-over to agriculture, increasing fishing pressure, destruc ve fishing prac ces and an acute shortage of dry season wetland habitat (Hughes et al. 1994; Ali 1997). There are 260 indigenous species of finfish living in and around the freshwater habitats of Bangladesh (Rahman 1989). Rural families in Bangladesh consume up to 73 species of small indigenous fish during the course of a year, harvested mostly by the families (Minkin 1993). According to IUCN (2000) 54 freshwater fish species out of 260 species are threatened in Bangladesh. The short-term leasing prac ce of small water bodies provides li le incen ve to leaseholders to harvest aqua c resources in a sustainable manner and o en acts as an obstacle to access by poorer members of the community (Craig et al. 2004). Co- management and community-based management approaches have long been advocated for as a means to address common failures associated with conven onal ‘top-down’, government-driven approaches to fisheries management but few studies have quan ta vely demonstrated their benefits. The present study employed data collected from 64 project sites and 16 control sites represen ng a range of different habitat types and geographic loca ons, and many river sec ons, floodplains and depressions (beels) located in the lower Indo-Gange c Plains. To assess the status of fisheries resources and to test alterna ve local fishery management prac ces that might increase biodiversity this study covered fishing ac vi es, fisher’s par cipa on, overall fish abundance and biodiversity data analysis. This paper also evaluates 290 management performance and resource sustainability to determine the benefits of community-based fisheries management (CBFM) in the inland fisheries of Bangladesh. It does so by looking at qualita ve indicators in a variety of wetlands (i.e. fish produc on, abundance and biodiversity) and studying their changes with me at project and control sites. 2. Methods The Community Based Fisheries Management (CBFM) project began in September 2001. This five-year project, funded by the UK Government’s Department for Interna onal Development (DFID), was implemented jointly by WorldFish and the Government of Bangladesh’s Department of Fisheries in partnership with 11 Non-Governmental Organiza ons (NGOs). These field-based partner NGOs were responsible for organizing about 23,000 poor fishing households around 120 sites represen ng a range of different habitat types (beel a permanent water body; haor beel - a large, low-lying depression in a floodplain; river sec on) and located in regions throughout Bangladesh (Figure 1). The CBFM project was designed as an ac on research project to test and assess alterna ve local fishery management arrangements that might achieve greater efficiency, equity and sustainability. Partner NGOs have helped the fishers to develop Beel Management Commi ees, River Management Commi ees and Cluster Management Commi ees for fishery resources management. Management commi ees were formed in all project sites through elec on or selec on of members by stakeholders and each site has its own resource management plans and rules. Commi ees have generally adopted simple conserva on-based measures such as reserve harvest (small fish sanctuary), gear regula on, closed season and habitat restora on under CBFM. The present study employed data collected since 2002, represen ng a range of different habitat types and geographic loca ons. This study was conducted by using species data for project and control sites to determine produc vity (catch per unit area - CPUA), catch per day (CPD) and the Shannon-Weiner biodiversity Index (H’) trends with me, species assemblage, abundance and sites’ similari es. 88° 89° 90° 91° 92° Project Control 88° 89° 90° 91° 92° Fig. 1. Loca on of surveyed sites, CBFM projects and control sites in Bangladesh. 291 3. Data Fishing ac vity was observed for four days per month per site. Gear surveys involved a regular spot survey for a sample of gear in opera on, and the total catch from each gear type. A gear census covered the number and types of gear opera ng in the study sites. Species-wise catch sta s cs for each gear type were recorded between 2002 and 2005 (split year, June to May) for produc vity (catch per unit area-CPUA) and abundance (catch per day-CPD) analysis. Catch data from 2002 to 2005 on species and gear was collected from a maximum of 64 project sites (5 closed beel, 14 river sec ons, 21 ppen beel, 6 haor beel, 18 floodplain beel) and 16 control sites (2 closed beel, 4 river sec on, 4 open beel, 3 haor beel, 3 floodplain beel). At least three years worth of data were available for each site and used for biodiversity (H’) and univariate analysis. The monitoring of control sites began in 2002 and the gill net CPUE was es mated for August and September (peak monsoon) to maximize the sample size. The gill net CPUE was considered during August and September to avoid the effects on CPUE of varia on in gear catchability and only similar months were compared between years to get consistent diversity of species for the inland water systems. The species data in this study was examined in two ways: for biodiversity analyses (H’-Shannon biodiversity Index) and trends with me at each site (slope b value); and for determining species assemblage and abundance between project and control sites. The trend (average annual change) in performance indicator (biodiversity) was first es mated for each site using the general linear model (GLM) where the performance indicator formed the dependent variable and me (year) was treated as covariate. The slope coefficient (b) of the linear (regression) model provided an es mate of the magnitude of performance indicator trend and whether it was upward (+ b value) or downward (- b value). Only sites with at least three years of observa on were included. 3.1 Univariate measure of fish diversity Management performance was quan fied using indicators of produc on and resource sustainability. Annual mul species catch per unit area (CPUA) was employed as a measure of produc on at each site: m  Ma y n   Catch s , y ,m ,g CPUAs ,y  m  June g 1 Equation 1 MaxArea s Where Catchs,y,m,g is the es mated mul species catch landed by gear type g, during month m and year y at site s measured in kg ha-1 y-1, and MaxAreas is the mean maximum area of the site. Fish abundance indicated by mul species catch fisher-1 day-1 or ‘catch per day’ (CPD) expressed as kg day-1 was employed as a measure of resource sustainability: CPD s, y  Catch s, y Annual Fishing Days s ,y Equation 2 Where Annual Fishing Dayss,y is the es mated total number of days spent fishing by the fishers at site s during year y, irrespec ve of the gear type employed. The Shannon-Weiner biodiversity Index (H') (Shannon 1948) was employed as a univariate indicator of fish diversity and used for species-wise catch from 2002 to 2005 (Equa on 3). H '   p (ln p j j 292 j ) Equation 3 \Where pj is the propor on of the total biomass arising from the jth species. Here pj was es mated using the average gillnet catch rate for species j between August and September at site s, during year y (Equa on 4): n  GNCPUE GNCPUE j ,s , y  i n j, i , s , y Equation 4 Where GNCPUE j ,i ,s,y  Catch8 9, j ,i ,s, y NetArea8 9,i ,s,y . Hours8 9,i ,s,y . 1000 Equation 5 Where GNCPUE i.s.y is the catch rate of the ith gillnet sampled at site s between August (month 8) and September (month 9) of year y, and NetAreai.s.y is the area of ith net sampled at site s, in year y. Hourei.s.y is the fishing hours. The ra o was mul plied by 1000 because units (kg m-2 hr-1) were typically very small. 3.2 Mul variate comparisons of diversity Mul variate comparisons of fish diversity were performed by comparing abundance indices (gill net catch per unit effort in August and September (GNCPUE8-9) from 2002 to 2005) of species forming the mul species assemblage between CBFM and control sites. Only testable habitat and region combina ons containing control sites were used for mul variate analysis. The year 2005 was selected, as the year with maximum numbers of control site data available from different habitats for mul variate analysis. Data recorded for floodplain beel habitat in the north (FPB-N), haor beel habitat in the east (HB-E), open beel habitats in the north (OB-N) and north-west (OB-NW), and river habitat in the east (R-E) of the country were used for mul variate analysis. Similari es in the species assemblages at CBFM and control sites were summarized in two-dimensional space using nonparametric mul dimensional scaling (MDS) ordina ons (Clarke 1993). The approach aims to construct a map or ordina on of sites (samples) such that their placement reflects the rank similarity of their species assemblages. Sites posi oned in close proximity to each other in the ordina on have very similar species assemblages, whilst sites that are far apart share few common species or have the same species but at very different levels of abundance. A “stress” measure indicates how well the ordina on sa sfies the (dis)similari es between sites. Stress values <0.2 indicate acceptable fits to the data. The null hypothesis [H0: There are no differences in species assemblages between CBFM and control sites] was tested using a nonparametric permuta on test (analysis of similarity or ANOSIM) based on the difference in average rank within and between the CBFM and control site groups (r sta s c). The significance level of the test is calculated by referring the observed value of the r sta s c to its permuta on distribu on generated from randomly sampled sets of permuta ons of site labels. The species most responsible for the site groupings were then determined by compu ng the average contribu on of each species to the overall average dissimilarity between all pairs of intergroup sites. The MDS and ANSOSIM analyses were performed with the PRIMER so ware (Clarke and Gorley 2006) on fourth-root transformed data and employing the Bray-Cur s (Bray and Cur s 1957) similarity coefficient as the measure of similarity between pairs of sites. 4. Results 4.1 Univariate trend analysis The rela ve frequencies of the upward and downward trends indicate that the CBFM ac vi es have significantly (p<0.01) benefited produc on (CPUA), fish abundance (CPD) and biodiversity (H’) at the majority (70-80%) of CBFM sites. At control sites, downward trends in CPUA, CPD and H’ were more frequent than upward trends. 293 Trends in fish produc on (CPUA) through me were upward at 77% of the 64 project sites that were monitored monthly for at least three years without data gap. At these sites annual fish produc on (kg ha-1) increased on average by 13% a year-1. Taking account of habitat type, annual fish produc on ha-1 year-1 increased on average by 22%, 29%, 12%, and 22% for closed beel, floodplain, open beel and river habitats, respec vely. In contrast, produc on decreased by 19% ha-1 year -1 for haor beel habitats. Trends in fish abundance as indicated by the annual average of fishers’ daily catch rates increased at 72% of the 64 project sites, with an average increase of 17% per year-1. When we consider this annual average increase for habitat types we see that closed beel, floodplain, haor beel, open beel and river habitats experienced an increased annual average of daily catch per fisher of 22%, 12%, 9%, 21% and 19%, respec vely. Forty-eight of the 64 project sites for which three or more years’ es mates were available showed an upward trend in biodiversity (H’) with me. Six of the 16 control sites for which three or more es mates were available also showed an upward trend in biodiversity (H’) with me and the remaining ten showed downward trends. Distribu on of b values for trends in biodiversity (H’) with me for habitat types at project sites are shown in Figure 2a and for control sites in Figure 2b. The posi ve values of slope b indicate that biodiversity (H’) increased at those sites and the nega ve values of slope b indicate that biodiversity (H’) decreased with me. Distribu on of slope (b) values (Project sites) 1.200 1.000 0.800 0.600 0.400 0.200 0.000 -0.200 -0.400 -0.600 -0.800 Closed beel River sec on Open beel Haor beel Floodplain Fig. 2a. Distribu on of slope (b) values for trend (up or down) in biodiversity index (H’) with me for project sites. 0.400 Distribu on of slope (b) values (Project sites) 0.300 0.200 0.100 0.000 0.100 0.200 0.300 -0.400 -0.500 -0.600 Closed beel River sec on Open beel Haor beel Floodplain Fig. 2b. Distribu on of slope (b) values for trend (up or down) in biodiversity index (H’) with me for control sites. 294 4.2 Mul variate analysis Differences in species assemblages between CBFM and control sites were tested within the same habitat and region. Significant (p<0.05) differences in species assemblages at CBFM sites and control sites were found only for floodplain beel habitat in the north (FPB-N) in 2004, and river habitat in the east (R-E) in 2004 and 2005 (Table 1 and Figure 3). One-way ANOSIM showed the R-sta s cs, where an R value of +1 indicates that the most similar samples are within the same groups and value of -1 indicates that the most similar samples are outside of the groups. The R sta s cs were higher in 2005 compared to the baseline survey in 2002 for river habitat in the east (R-E), and compared to the survey in 2003 for floodplain beels in the north (FPB-N) and open beels in the north (OB-N). However, R-sta s cs were lower in 2005 compared to the baseline survey in 2002 for haor beels in the east (HB-E) and open beels in the northwest (OB-NW). Only testable habitat and region combina ons containing control sites were used. Table 1. Results showing the one-way ANOSIM between CBFM and control sites Habitat and Regions Floodplain North (FPB-N) Haor beel East (HB-E) OPEN beel North (OB-N) Open beel North-West (OB-NW) River East (R-E) Years 2002 2003 2004 R-Sta s c No control 0.119 -0.187 Permuta ons No control 105 105 Significant Level % No control 25.7 86.7 2005 2002 2003 2004 0.250 -0.148 0.037 -0.111 560 84 35 84 4.5 75 42.9 71.4 2005 2002 2003 -0.021 No control -0.229 35 No control 28 71 No control 85.7 2004 2005 2002 2003 0.154 0.102 -0.393 -0.196 120 91 15 9 20 34 80 77.8 2004 2005 2002 2003 -0.140 -0.246 -0.222 0.037 165 55 35 35 74.5 87 91.4 42.9 2004 2005 0.685 0.824 35 35 2.9 2.9 295 MDS - Axis 1 vs Axis 2 - Floodplain North MDS - Axis 1 vs Axis 2 - Haor beel North 2. 0 3.0 2.0 2.0 1.0 0.0 1. 0 1. 0 1.0 0.0 -1.0 -2.0 -3.0 -1.0 -2.0 -2 -1 0 A xis 1 1 2 MDS - Axis 1 vs Axis 2 - Open beel North -1.0 -2.0 -2 -1 0 1 2 -2 Axis 1 MDS - Axis 1 vs Axis 2 Open beel North-West 0 Axis 1 2 MDS - Axis 1 vs Axis 2 -River East 3. 0 2. 5 2. 0 1. 5 1. 0 0. 5 -0.0 -0.5 -1.0 -1.5 -2.0 2. 0 Project Control 1. 0 0. 0 -1.0 -2.0 -2 -1 0 1 A xis 1 2 -2 -1 0 A xis 1 1 Fig. 3. MDS ordina ons: comparing species assemblages at CBFM and control sites in each habitat or region combina on. 4.2.1 Floodplain beel habitat in the north (FPB-N) region For floodplain-beel habitat in the north region, more than 30 representa ve species were either absent or less abundant at the two control sites compared to the 13 CBFM sites in 2005 (Figure 4). Species are arranged in descending order of their contribu on to the average dissimilarity between the two groups of sites (CBFM or control). Only those species contribu ng to 90% of the cumula ve average dissimilarity are shown. Of the 36 recorded species 10 were more abundant at CBFM sites. These included, in descending order of their contribu on to the average dissimilarity between the two groups of sites: Nandus nandus, Cirrhinus cirrhosus, Labeo rohita, Anabas testudineus, Pun us sophore, Channa punctatus, Pun us chola, Macrobrachium macolmsonii, Heteropneustes fossilis and Macrognathus puncalus. Only six species were more abundant at control sites compared to CBFM sites: Labeo calbasu, Mastacembelus armatus, Cyprinus carpio, Mystus cavascius, Pun us gonionotus and Mytus bleekeri. However, given the poten al difficulty in selec ng comparable control sites, these differences in species assemblages may simply reflect differences in site habitat. 296 Figure 4. Average abundance [gillnet catch per unit effort (kg 1000 m 2 h-1)] of species caught from CBFM and control sites exploi ng floodplain-beel habitat in the north (FPB-N) region of the country. 4.2.2 Haor beel habitat in the east (HB-E) region Species assemblages at the CBFM and control sites were very similar in 2005. Species are arranged in descending order of their contribu on to the average dissimilarity between the two groups of sites (Figure 5). Only those species contribu ng to 91% of the cumula ve average dissimilarity are shown. Of the 16 recorded species, six were more abundant at CBFM sites. These included, in descending order of their contribu on to the average dissimilarity between the two groups of sites: Anabas testudineus, Nandus nandus, Pun us sophore, Heteropneustes fossilis, Mystus tengra and Wallago a u. However eight species were more abundant at the control sites: Pun us sophore, Wallago a u, Nandus nandus, Macrognathus aculeatus, Xenentodon cancilia, Aorichthys seenghala, Rohtee co o and Pun us conchonius. The la er two were absent from the CBFM sites. 297 Fig. 5. Average abundance [gillnet catch per unit effort (kg 1000 m 2 h-1)] of species caught from CBFM and control sites exploi ng haor-beel habitat in the north (HB-E) region of the country. 4.2.3 Open beel habitat in the north (OB-N) and northwest (OB-NW) region Diversity of species assemblages at the CBFM sites was almost double those of the control sites in 2005. Only those species contribu ng to 91% of the cumula ve average dissimilarity in the open beel habitat (OB-N) are shown in Figure 6. Of the 23 recorded species, 10 were more abundant at CBFM sites. These included, in descending order of their contribu on to the average dissimilarity between the two groups of sites: Aanbas testudineus, Aorichthys seenghala, Pun us gonionotus, Channa striatus, Heteropneustes fossilis, Channa punctatus, Macrognathus aculeatus, Nandus nandus, Mystus aor and Hypophthalmiichthys molitrix. Ten species were absent at the control sites. Only two species were more abundant at the control sites: Macrobrachium birmanicum and Pun us cto. Species assemblage at CBFM sites increased with me in two habitats, but remained unchanged in the remainder. Biodiversity at control sites remained unchanged in all habitats. Species assemblages are richer and more abundant at CBFM compared to control sites in floodplain beel and river habitat in the north and east regions of the country, respec vely. Considered together, this evidence suggests that CBFM benefits biodiversity. 298 Fig. 6. Average abundance [gillnet catch per unit effort (kg 1000 m 2 h-1)] of species caught from CBFM and control sites exploi ng open-beel habitat in the northwest (OB-N) region of the country. 4.2.4 River habitat in the east (R-E) region Species assemblages at the CBFM sites comprised almost three mes more species than those of the control sites in 2005. Species are arranged in descending order of their contribu on to the average dissimilarity between the two groups of sites (Figure 7). Only those species contribu ng to 91% of the cumula ve average dissimilarity are shown. Of the 14 recorded species, 11 were present or more abundant at CBFM sites. These included, in descending order of their contribu on to the average dissimilarity between the two groups of sites: Wallago a u, Pama pama, Labeo bata, Pun us sophore, Mastacembelus armatus, Glossogobius guiris, Nandus nandus, Mystus bleekeri, Aorichthys seenghala, Clupisoma garua and Macrognathus pancalus. Only four species were more abundant at the control sites: Aorichthys seenghala, Labeo calbasu, Cluposoma garua and Labeo gonius. The la er three were absent from the CBFM sites. 299 Fig. 7. Average abundance [gillnet catch per unit effort (kg 1000 m 2 h-1)] of species caught from CBFM and control sites exploi ng river habitat in the eastern (R-E) region of the country. 5. Discussion According to the rela ve frequency of upward and downward trends in performance indicators at CBFM and control sites, the CBFM project appears to have benefited fish produc on (CPUA), abundance (CPD) and biodiversity (H’) at par cipa ng sites. Mean annual increases in fish abundance, indicated by CPD, were significantly greater at CBFM compared to control sites. Furthermore, the mean change in fish abundance at control sites was not significantly different from zero. Rates of change in biodiversity (H’) were found to vary significantly among habitat and were on average also significantly greater at CBFM sites than control sites. Improvements in biodiversity at CBFM sites through me were significant in closed and floodplain beel habitat. Significant improvements in biodiversity were also detected for control sites belonging to only floodplain beel habitat. Closed season appears significant but explains only 24% of varia on in biodiversity (Halls and Mustafa 2006; Mustafa and Halls 2007). Habitat restora on may be more appropriate in improving the availability of resources (Payne and Cowan 1998). Unlike the annual perspec ve of the CPD indicator, GNCPUE provides an index of fish abundance only during a two month period during the flood season when gillnets tend to target migratory whitefish species (Welcomme 1985). Improvements in fish abundance were strongly linked to reduc ons in fishing intensity and destruc ve fishing effort (Halls and Mustafa 2006). The co- and community-based management approaches have long been advocated as a means to address the failures associated with conven onal ‘top down’ approaches to management (Pomeroy and Williams 1994; Hoggarth et al. 1999; Wilson et al. 2003), but few studies have quan ta vely demonstrated their benefits. The CBFM project has already demonstrated that Community Based Organiza ons (CBOs) are mo vated to share and disseminate their knowledge and experiences through mee ngs, exchange visits and newsle ers (Halls et al. 2005). 300 The poten al difficulty in selec ng comparable control sites creates uncertainty for determining which management interven ons may be responsible for possible differences between river habitat in the east and floodplain beel habitat in the northern regions. These differences in species assemblages may therefore simply reflect differences in site habitat. 6. Conclusions and recommenda ons The present study clearly reveals that in project sites managed according to the CBFM regime, fishers have over me established appropriate fisheries management regimes. Rates of change in species abundance were found to vary significantly among habitats and were on average also significantly greater at CBFM sites than at control sites. Improvements over me in species abundance at CBFM sites were significant in floodplain and river habitats. On the basis of the results presented here, it is concluded that the prac ces implemented under community-based fisheries management in Bangladesh have improved, or at least sustained, fish abundance and biodiversity without significant loss compared to those at the control sites. However, further research is needed to determine why species abundance is more enhanced in floodplain and river habitats than in haor beel and open beel habitats. The community-based management approach shows considerable promise and should be considered for wider scaling out within sustainable management ini a ves in the Indo-Gange c Plains. Acknowledgements The authors acknowledge Alan Brooks, Robert Arthur, Susana Hervas Avila and Neil Andrew for their valuable sugges ons and helpful comments on the manuscript. Thanks also to Ashley Halls, Mohammod Ilyas, Khalilur Rahman, Susmita Choudhury and Ismat Ara for their valuable contribu ons towards the compila on, analysis and interpreta on of the data. This research and dissemina on was funded by the Department for Interna onal Development (DFID). The views expressed here are not necessarily those of DFID. References Ahmed, M., A.D. Capistrano, and M. Hossain. 1997. Experience of partnership models for the co-management of Bangladesh fisheries. Fisheries Management and Ecology. 4 (3), 233-248. Ali, M. Y. 1997. Fish, Water and People: Reflec on on Inland Open Water Fisheries Resource of Bangladesh. The University Press Limited. Dhaka. Bray JR. and JT. Cur s. 1957. An ordina on of the upland forest communi es of southern Wisconsin. Ecol. Monogr. 27, 325-349. Clarke, K. R. 1993. Non-parametric mul variable analysis of changes in community structure. Aust. J. Ecol. 18, 117-143. Clarke, K. R. and R. N. Gorley. 2006. PRIMER v6: User Manual/Tutorial. PRIMER-E, Plymouth. Craig, JF., A. S. Halls, J. Barr and C.W. Bean. 2004. The Bangladesh Floodplain Fisheries. Fish. Res. 66, 271-286. Halls, A.S. and M.G. Mustafa. 2006. Final Assessment of the Impact of the CBFM Project on Community-Managed Fisheries in Bangladesh. Report to the WorldFish Center, Bangladesh. Halls, A. S., Arthur, R., Bartley, D., Felsing, M., Grainger, R., Hartmann, W., Lamberts, D., Purvis, J; Sultana, P., Thompson, P., Walmsley, S., 2005. Guidelines for Designing Data Collec on and Sharing Systems for Co-Managed Fisheries. Part I: A Prac cal Guide. FAO Fish. Tech. Pap. 494/1&2. 301 Hoggarth, D.D., V.J. Cowan, A.S. Halls, M.T. Aeron, A. J. McGregor, C. A. Garaway, A. I. Panye, and R. R. Welcomme. 1999. Management Guidelines for Asian Floodplain River Fisheries. FAO Fish. Tech. Pap. 384/1&2. Hughes R., S. Adnan, and B. Dalal-Clayton. 1994. Floodplains or flood plans: London: Interna onal Ins tute for Environment and Development, and Research and Advisory Services. IUCN. 2000. Red Book of Threatened Fishes of Bangladesh. The World Conserva on Union (IUCN), Dhaka, Bangladesh, 116p. Minkin, F. 1993. Food control and nutri onal consequence of biodiversity of fisheries. FAP-16. Environmental Study. Mustafa, M.G. and S.A. Halls. 2006. Community Based Fisheries Management: Fisheries yields and sustainability. Malcolm Dickson and Alan Brooks (ed.), Policy Brief 2. The WorldFish Center, Bangladesh and South Asia Office, Dhaka, Bangladesh. Mustafa, M.G. and S.A. Halls. 2007. Impact of the Community-Based Fisheries Management on sustainable use of inland fisheries in Bangladesh. Conference Proceedings 75. Conference paper 3, Dickson M. and A. Brooks (ed.). Interna onal Conference on Community Based Approaches to Fisheries Management, the WorldFish Center. Bangladesh and South Asia Office, Dhaka, Bangladesh. Payne, A.I. and V.J. Cowan. 1998. Review of stock enhancement in the floodplains of Bangladesh. In J. Petr (Ed.). FAO/DFID Expert Consulta on on Inland Fisheries Enhancements. FAO Fisheries Technical Paper 374. Rome, FAO. Pomeroy, R.S. and M.J. Williams. 1994. Fisheries Co-management and small scale fisheries: a policy brief. Interna onal Centre for Living Aqua c Resources Management (ICLARM), Manila. Rahman, A.K.A. 1989. Freshwater fishes of Bangladesh. Zoological Society of Bangladesh, Dhaka, 352pp. Shannon, C.E. 1948. A mathema cal theory of communica on. Bell. System. Tech. J. 27, 379-423. Welcomme, R.L. 1985. River Fisheries. FAO Fish. Tech. Pap. 262. Wilson, D.C., J.R.Nielsen and P. Degnbol. 2003. The Fisheries Co-management Experience: Accomplishments, Challenges and Prospects. Fish & Fisheries Series. Vol: 26, Springer, New York. 302 Sec on 6 Rice-Based System Intensifica on and Diversifica on 303 Promising rice genotypes for the wet and dry seasons in coastal West Bengal S.K. Sarangi1, D. Burman1, S. Mandal1, B. Maji 1, E. Humphreys2, T.P. Tuong2, B. K. Bandyopadhyay1 and D. K. Sharma1 1 Central Soil Salinity Research Ins tute, Regional Research Sta on, Canning Town, West Bengal and Karnal, Haryana, India, sksarangicanning@gmail.com, burman.d@gmail.com, subhasis2006@gmail.com, b.maji57@gmail.com, bimalbkb@gmail.com, dineshksharma@rediffmail.com 2 Interna onal Rice Research Ins tute, Philippines e.humphreys@irri.org, t.tuong@irri.org Abstract Agricultural produc vity of the coastal lands of West Bengal is very low. Rice is grown on most of the agricultural land during the rainy season, but the majority of the land remains fallow during the dry season due to water scarcity and salinity. Due to water stagna on, the average yield of the wet season crop in the coastal zone of West Bengal (< 2.0 t ha-1) is below both the na onal average (2.4 t ha-1in 2011-12) and the average of West Bengal (2.6 t ha-1). Therefore, the development and dissemina on of improved boro (irrigated rice grown during the dry season) and aman (rainy season crop) cul vars is necessary to increase and stabilize rice produc on and alleviate food insecurity and poverty in the coastal region. This study sought to iden fy improved aman and boro rice varie es in on-farm experiments in South and North 24 Parganas Districts of West Bengal, India, from 2011 to 2014. The performance of elite lines and varie es from the Central Soil Salinity Research Ins tute, Interna onal Rice Research Ins tute, and government agencies of India and Bangladesh was evaluated and compared to that of local cul vars. During each dry season, 17 entries were tested in soils with a range of salini es. The Bangladeshi varie es BRRI dhan47 and Binadhan-8, and the variety WGL 20471 (released from Warangal, Andhra Pradesh) produced the highest average grain yields of 5.0, 4.8 and 4.5 t ha-1, respec vely. These yields were 40 to 60% higher than that of the local cul vars (3.2 t ha-1). During the wet seasons, 13 entries were evaluated at loca ons that experienced a range of water depths. The improved aman varie es yielded ~4 t ha-1, about 20% more than the local cul vars (3.4 t ha-1) across all water depths. The best performing entries across all water depths were Amal-Mana, Swarna-Sub1 and CSRC(D) 7-0-4 and CSRC(D) 12-8-12. The performance of the shorter statured Bangladeshi varie es declined as water depth increased. The results demonstrate the great poten al for increasing the produc vity of the coastal zone through adop on of improved rice varie es. The shorter dura on of the improved wet season varie es may also facilitate cropping intensifica on. Some of the best performing varie es in the dry season came from other countries, highligh ng the importance of collabora on and germplasm exchange among neighboring countries. Key message: There is considerable poten al for increasing rice produc on in the coastal zone of West Bengal through replacement of locally grown wet and dry season varie es with exis ng improved, stress tolerant varie es. Keywords: aman, boro, improved varie es, salinity, waterlogging 1. Introduc on The saline coastal zone of India stretches over a length of 8,129 km along the eastern and western fringes of the country, with a total area of about 3.1 Mha (Table 1). The largest area of saline coastal soils is in West Bengal, with 0.82 Mha located in four out of 20 districts (South 24 Parganas, East Midnapore, North 24 Parganas and Howrah). Rice is the predominant crop in the coastal areas of West Bengal during both the wet (kharif) season from June to November, and during the dry (rabi) season from November to April. The dry season crop is almost en rely dependent on irriga on using groundwater. Farmers in the region are predominantly marginal landholders; the rest are either small farmers or landless (ICAR-CSSRI 2014). They prefer to follow a rice-rice cropping system for a range of reasons including mee ng family food requirements, use of rice crop byproducts for domes c purposes, and unfavorable environmental condi ons 304 for growing other crops. During the rainy season, farmers typically grow tradi onal, photoperiod-sensi ve, low yielding aman cul vars, whereas during the dry season high yielding boro varie es with moderate tolerance to salinity are grown. The single largest problem during the aman crop is waterlogging of varying depth and dura on. This problem is also compounded by occasional natural disasters (cyclones, seawater intrusion, drought). The boro crop faces problems of lack of fresh water and soil and water salinity. Table 1. Distribu on of coastal saline soils in India (source: Yadav et al. 1983) State West Bengal Gujarat Area (Mha) 0.82 0.71 Orissa Andhra Pradesh Tamil Nadu Karnataka 0.40 0.28 0.10 0.09 Maharashtra Kerala Goa Andaman & Nicobar Islands 0.06 0.03 0.02 0.02 Pondicherry Total Area under mangroves 0.001 2.52 0.57 Total 3.09 Several high yielding (4-5 t ha -1) aman varie es with improved tolerance to flooding have been developed recently by various Indian ins tu ons including the Central Soil Salinity Research Ins tute (CSSRI) Regional Research Sta on at Canning Town, West Bengal and the Central Rice Research Ins tute (CRRI) at Cu ack, Odisha. Furthermore, the Sub-1 genes for submergence tolerance have been introgressed into some major varie es. These improved varie es have the poten al to substan ally improve and stabilize rice produc on in flood-prone environments (Singh et al. 2009; Sarangi et al. 2015b). For the dry season, boro cul vars of shorter dura on with greater tolerance to salinity are needed (Sarangi et al. 2014). Where fresh water is available, short dura on varie es are preferred by farmers to reduce the water requirement and thus the pumping cost. Salt-tolerant boro varie es developed by the Bangladesh Rice Research Ins tute (BRRI) for the adjacent coastal zone region in Bangladesh could also fit into the cropping system in West Bengal (Hossain 2003). While there has been good progress in the development of improved varie es for the wet and dry seasons in the coastal zone of West Bengal, these varie es have not been evaluated in farmers’ fields. Therefore, the objec ve of the work presented here was to evaluate the performance of promising aman and boro varie es for the wet and dry seasons, respec vely, in the coastal area of West Bengal. 2. Methods 2.1 Study sites On-farm evalua on of aman and boro varie es was conducted at representa ve sites in Gosaba Block in South24 Parganas District (220 08 56 N, 880 48 26 E) and in Sandeshkhali II Block in North 24 Parganas District (220 18 00 N, 880 49 36 E) of West Bengal, India. The region has a tropical monsoon climate with an 305 average annual rainfall of 1830 mm, of which about 89% occurs from June to October. At Sandeshkhali the source of irriga on water during the dry season was groundwater at a depth of 110-116m. At Gosaba the source was rainwater conserved in ponds and ditches. The depth of fresh groundwater (275 m) at Gosaba is too deep to use for irriga on because of the high cost of installa on and pumping of deep tubewells. The soil (10-15 cm layer) was sampled prior to the start of the first boro and aman experiments and analyzed for texture (hydrometer method), salinity (electrical conduc vity of satura on extract, ECe), pH (1:2.5 soil:water), organic carbon (Walkley and Black 1934), available nitrogen (Subbiah and Asija 1956), available phosphorus (Olsen et al. 1954) and available potassium (Hanway and Heidel 1952).The soils at all sites are heavy textured (silty clay loam to silty clay), slightly acidic, with medium organic carbon (0.50–0.75%), low available nitrogen (<280 kg ha-1), high available potassium (>300 kg ha -1), and medium available phosphorus (11-25kg ha-1) (Tables 2 and 3). The soils were moderately saline shortly prior to transplan ng both the boro and aman crops. Table 2. Ini al soil and irriga on water proper es at the boro experimental sites in January 2012 (mean ± s.e.) Soil ECe (dS m-1) 4.1±0.1 Sandesh khali (n=18) Gosaba 5.4±0.1 (n=18) N (kg ha-1) P (kg ha -1) K (kg ha-1) % OC 5.6±0.2 Water EC pH (dS m-1) 1.9±0.2 8.2±0.1 222±5.1 19.0±0.9 450±6.9 0.7±0.1 5.9±0.8 2.5±0.1 185±5.1 20.6±1.7 468±13.2 0.7±0.1 K (kg ha-1) % OC pH 8.7±0.1 Note: OC = Organic Carbon Table 3. Ini al soil proper es at the aman experimental sites in June 2012 (mean ± s.e.) Soil N (kg ha-1) P (kg ha -1) Sandeshkhali (n=9) Gosaba (n=9) EC (dS m-1) 5.8±0.2 6.5±0.1 5.8±0.1 6.7±0.4 198±6.8 189±5.3 18.2±1.1 19.4±1.9 450±6.9 468±13.2 0.7±0.1 0.7±0.1 Basan (n=9) 5.3±0.3 5.5±0.2 203±7.1 22.1±1.3 390±8.3 0.6±0.2 pH Note: OC = Organic Carbon 2.2 Experimental design The experimental design for evalua on of aman and boro varie es was a randomized block with three replicates at each site (farmers’ field) and a gross plot area of 60 m2 (12 m× 5 m). 2.2.1 Boro season Evalua on of eight to ten boro varie es and advanced breeding lines (“entries”) was conducted for three consecu ve boro seasons (2011-12, 2012-13 and 2013-14) (Table 4). The evalua on was done in farmers’ fields at two loca ons: ‘Sandeshkhali’ in Daudpur village (in Sandeshkhali II Block, North 24 Parganas District), and ‘Gosaba’ in Pakhiralay South, Ja rampur, Dulki and Pakhiralay villages (in Gosaba Block, South 24 Parganas District). At each loca on there were one to six sites (farmers’ fields) each year. Desirable rice variety a ributes for the dry season include salinity tolerance, short dura on, fine grain for higher market price and bold grain for home consump on, and high straw yield for domes c uses and animal 306 fodder. Salinity tolerance of the entries varied from moderate to high and all were photoperiod insensi ve, with dura on ranging from about 110 to 150 d. Some of the entries were common over the three years, however, some were deleted and new entries added depending on performance and availability of new promising ones, respec vely. Most entries were developed in India, but three Bangladeshi varie es (BRRI dhan47, BRRI dhan55 and Binadhan-8) and three IRRI lines (IR 87938, NSIC RC 238, NSIC RC 222) were also included. Old and tradi onal varie es o en grown by farmers in the study areas (Canning-7, Parijat, N. Sankar, S. Sankar) were also included in the second year. Table 4. Characteris cs of the entries evaluated during the boro season over three years Entry Salinity tolerance (dSm-1) Plant height (cm) Dura on (days) Grain yield Grain (t ha-1) type1 Year released In India Canning 7 6-8 95-105 130 4.0-4.5 LB 1989 CSR4 WGL 20471 “Lal-minikit” 6-8 6-8 95-100 95-105 135 130 3.5-4.0 5.0-5.5 LB LS 1985 1991 IET 4786 4-6 “Sada-minikit” 80-95 135 3.5-5.6 LS 20003 Annada 4-6 95-105 135 3.5-4.5 SB 1987 Boby Lalat 4-6 2-4 100-105 102 145 135 4.0-4.5 4.0-4.5 LB LS 1988 BRRI dhan472 4-6 105 152 6.0 MB 2007 Binadhan-82 10 95 145 6.0 MB 2010 BRRI dhan552 Bidhan 2 Parijat N. Sankar S. Sankar IR 87938 NSIC RC 238 NSIC RC 222 4-6 6-8 4-6 2-4 2-4 2-4 2-4 2-4 100 105-110 85-95 95-100 95-100 95-100 100-105 100-105 145 135 135 140 140 140 140 140 7.3 4.0-5.0 3.5-4.0 3.0-4.0 3.0-4.0 3.0-4.0 3.0-4.0 4.0-4.5 LS LS MS LS LS MS MS MS 2011 1976 TV4 TV - 1 LB: long bold, MB: medium bold, SB: short bold LS: long slender, MS: medium slender 2 Varie es developed and released in Bangladesh; characteris cs are for condi ons in Bangladesh 3 Year of no fica on 4 TV: tradi onal variety Year evaluated 2011-12 2013-14 2011-12 2011-12, 2012-13, 2013-14 2011-12, 2012-13, 2013-14 2011-12, 2012-13, 2013-14 2011-12 2011-12, 2013-14 2012-13, 2013-14 2012-13, 2013-14 2012-13 2011-12 2012-13 2012-13 2012-13 2013-14 2013-14 2013-14 307 2.2.2 Aman Aman variety evalua on (nine to ten entries) was conducted at three sites in 2012 and 2013 (Table 5). Most (around 85%) of the land is low-lying, therefore varie es that are generally tall, high yielding, stress tolerant (tolerant to flash/stagnant flooding, lodging, diseases and pests) with long slender grains having good cooking quality are preferred during this season. All varie es evaluated had moderate salinity tolerance, and all were tall or very tall and photoperiod sensi ve except for Swarna-Sub1, a submergence-tolerant variety developed by IRRI (Ismail et al., 2013). All varie es were also long dura on except for Swarna-Sub1, which was medium maturity. The CSRC (D) series are promising lines developed at ICAR-CSSRI RRS Canning Town for low-lying coastal lands. BRRI dhan47 and Binadhan-8 are medium statured, salt tolerant and photoperiod insensi ve varie es from Bangladesh. Table 5. Characteris cs of the entries evaluated during the aman season over two years Variety Salinity tolerance (dSm-1) Plant height (cm) Dura on (days) Grain yield (t ha-1) Grain type Year released SR 26B Sabita 4-6 4-6 130-145 180-190 165 210 3.5-4.0 3.5-4.0 MS LS TV1 1986 Geetanjalli Amal-Mana Patnai 23 NC 678 CSRC (D) 7-0-4 CSRC (D) 13-16-9 CSRC (D) 12-8-12 CSRC (D) 2-17-5 4-6 6-8 4-6 4-6 4-6 4-6 4-6 4-6 160-170 160-170 160-170 165-170 160-175 160-175 160-175 155-175 165 165 165 160 170 170 170 165 4.0-4.5 5.0-5.5 3.0-3.5 2.0-2.5 4.0-4.5 4.0-4.5 4.0-4.5 4.0-4.5 LS LS MS MS MS MS MS MS TV 2008 TV TV 2008 2008 2008 2008 Swarna-Sub1 BRRI dhan47 Binadhan-82 4-6 10 95-105 105 95 140 152 145 4.5-5.0 6.0 6.0 MS MB MB 2009 2007 2010 1 TV: tradi onal variety 2.3 Crop management 2.3.1 Boro Nursery sowing was done during the second week of December and 30 d old seedlings were transplanted at a spacing of 20 cm x 10 cm with 1-2 seedlings per hill. A fer lizer dose of 120-20-0 (N:P2O5:K2O) kg ha-1 was applied to each plot as urea, single super phosphate and muriate of potash, respec vely. All of the P and K and 25% of the N were applied prior to leveling. Half of the N was broadcast 21 d a er transplan ng (DAT) and the remaining 25% was broadcast at 60 DAT. Hand weeding was done twice at 20 and 40 DAT to remove all weeds. Chloropyriphos @ 2 ml l-1 water and tricyclazole @ 0.6 g l-1 water were used to control insects and diseases, respec vely, as recommended. The plots were kept flooded (2.5-7.5 cm) throughout the season un l about 20 d before harvest maturity. 2.3.2 Aman Nurseries were sown in mid-June and 40 d old seedlings were transplanted in the last week of July using two seedlings per hill spaced at 15 cm × 15 cm. All plots in the field received similar management, including 5 t ha-1 308 of farmyard manure applied one month before transplan ng, and inorganic fer lizers applied at 50-20-10 (N-P2O5-K2O) kg ha-1. P and K were incorporated before transplan ng, whereas N in the form of urea was applied in three equal splits at 7 DAT, maximum lering (45 DAT) and at ini a on of flowering (75 DAT). Hand weeding was done twice at 30 and 60 DAT to remove all weeds. Chloropyriphos @ 2 ml l-1 water and tricyclazole @ 0.6 g l -1 water were used to control insects and diseases, respec vely, as recommended. 2.4 Monitoring 2.4.1 Water depth The depth of water in the field during each aman season was monitored daily using a meter s ck installed in each plot. The mean data for all the plots was considered as the flood water depth (FWD) of that loca on. 2.4.2 Soil and irriga on water salinity Soil samples were collected from the root zone (10-15 cm) before transplan ng the boro experiments each season, and at monthly intervals during the crop growth period. The samples were processed as per standard procedure (Gupta 2006) and the electrical conduc vity (EC) of satura on extract (ECe) was determined. Water samples were collected from the irriga on sources for EC determina on at the same me as soil sampling. 2.4.3 Crop data The date of physiological maturity was taken as the date when the grains had turned a golden yellow color and there was senescence of lower leaves. Grain and straw yields were determined in a 5m x 2m area at the center of each plot. The grain and straw were weighed, subsamples were dried at 60oC to determine moisture content, and grain yield (14% moisture) and straw yield (dry) were determined. Plant height (from base to p of the panicle) and the number of panicles hill -1 were determined on three hills at each corner of the yield area. Ten panicles were randomly selected from each plot to determine the number of grains panicle -1. The panicles were hand-threshed, filled (grains) and empty spikelets (chaff) were separated by submerging the threshed spikelets (floa ng spikelets considered empty) in tap water, then counted, weighed and dried to determine grain moisture content and 1000-grain weight at 14% moisture content. 2.5 Weather data Daily rainfall, maximum and minimum temperature and bright sunshine hours (BSH) were collected from the meteorological observatory at Canning Town. 2.6 Sta s cal analysis The data were subjected to analysis of variance (ANOVA, Gomez and Gomez 1984) using MS Excel and the STAR sta s cal package developed by IRRI. In the first boro season there were several sites (farmers’ fields with replicated experiments) at both loca ons, and the mean of the treatments within each farmers’ field was considered as one replicate. In the other two boro seasons there was only one replicated experiment at each loca on. During both rainy seasons there were three farmers’ fields with replicated experiments, and each farmer’s field was considered as one replicate. Treatment means were compared using the least significant difference (LSD) at p=0.05 level of significance. The interac ons between variety, year and site were also determined by ANOVA for the four entries in common across all years for both the boro and aman seasons. To determine the rela onship between yield and variety, average soil salinity from transplan ng to maturity was regressed against grain yield for each boro variety grown in more than one year. For the rainy season, average water depth during August (shortly a er transplan ng, when the crop is most sensi ve to deep water) was regressed against yield. The similarity of the regressions for different varie es was determined using the minimum significant difference (MSD) method (Howell 2013). A pair of regression coefficients, bi 309 and bj, are declared significantly different at the experiment wise error rate α if and only if their difference equals or exceeds the cri cal difference MSDij. That is, if [bi-bj]>MSDij, where MSD ij = Q’0.05[k,v] max (sbi, sbj) and Q’ is a value from the studen zed augmented range distribu on, k = no. of items, and v = degrees of freedom. 3. Results and discussion 3.1 Weather Rainfall was generally well distributed during each aman season and li le rain fell during the maturity period (November and December). Rainfall was much higher during the 2013 aman season (1936 mm from June to October) compared with 2012 (1272 mm), and higher than the long term average during this period (1554 mm) (Fig. 1a). The monthly temperature trends were similar in all three years with the excep on of unusually high minimum temperature in January 2012 (Fig. 1b). This anomaly was also observed in the coastal zone of Bangladesh (Mondal et al. 2015). (a) 800 700 600 500 2011-12 400 2012-13 300 2013-14 Longterm 200 100 0 Jun Jul Aug Sep (b) Nov Dec Jan Feb Mar Apr May 40 Max 35 30 Min 25 201 1-12 20 201 2-13 201 3-14 15 Longterm 10 Jun 310 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May 10 2011-12 9 2012-13 8 2013-14 7 Longterm 6 5 4 3 2 1 0 Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Fig. 1. Rainfall (a), maximum (max) and minimum (min) air temperature (b), and bright sunshine hours (BSH) (c) during the experimental period at ICAR-CSSRI RRS Canning Town. Long-term averages are for 1984-85 to 2013-14. 3.2 Boro variety evalua on 3.2.1 Soil and water salinity At Sandeshkhali, ini al soil salinity ranged from 2.1 to 7.9 dS m-1 across fields and years (Table 6). Ini al soil salinity at Gosaba was higher (3.4 to 9.3 dS m -1). At both loca ons, soil salinity increased with me a er transplan ng, by around 3dS m-1 between January and April each year (Fig. 2a). Salinity was consistently higher at Gosaba than at Sandeshkhali. Irriga on water salinity also increased during the boro season, more so at Sandeshkhali where irriga on was from groundwater (Fig. 2b), consistent with the findings of Burman et al. (2015) over several sites and years. Table 6. Ini al and final soil salinity (dS m-1) in boro fields over three years. Values are mean and range (in parentheses) Loca on Sandeshkhali Gosaba 2011-12 2012-13 2013-14 Ini al Final Ini al Final Ini al Final 4.1 (2.1-7.8) 5.4 (3.4–9.9) 7.2 (4.7-9.4) 8.9 (7.8-10.6) 4.5 (3.2-6.1) 6.2 (4.7-9.3) 7.7 (5.3-9.8) 9.7 (6.4-13.3) 4.4 (3.5-7.9) 5.8 (4.1-8.7) 6.8 (4.8-8.4) 8.1 (6.5-10.8) 311 (a) 12 10 8 6 4 2 0 Sandeshkhali 2011 -12 Sandeshkhali 201 2 -13 Sandeshkhali 2013 -14 Gosaba 2011 -12 Gosaba 201 2 -13 Gosaba 2013 -14 Jan (b) 4.0 3.5 Feb Mar Apr Sandeshkhali Gosaba 3.0 2.5 2.0 1.5 1.0 0.5 0.0 February March April Fig. 2. Average soil salinity during the boro crop each year (a) and irriga on water salinity during the 2011-12 boro crops (b), at Gosaba and Sandeshkhali. Ver cal bars indicate standard error. 3.2.2 Performance of boro varie es The dura on of most of the rice entries during the boro season was about 140 d. The Bangladeshi varie es had shorter dura on ranging from 120 d (BRRI dhan55) to 130 d (BINAdhan-8) to 135 d (BRRI dhan47). The Indian variety Canning7 was also slightly shorter dura on (135 d) than most of the Indian varie es, while Lalat was slightly longer dura on than most at 145 d. The interac ons between variety, year and site were determined by analyzing the data for the four entries that were common across all three years (WGL 20471, IET 4786, Annada and Bidhan 2). The interac ons were not significant for any of the measured parameters (Table 7). The effect of year was not significant for parameters like dura on, plant height, panicles hill -1, grains panicle-1 and test weight. Site had a significant effect on these parameters. However, since the varie es were different in different years, the analysis for grain yield was done year-wise and loca on-wise (Table 8). 312 Table 7. Level of significance of crop dura on, plant height, yield and yield components as affected by variety, year and site. Boro results are for the four varie es in common over the three years (2011-12, 2012-13, 2013-14) (WGL 20471, IET 4786, Annada and Bidhan 2). Aman results are for the four varie es in common over the two years (2012 and 2013) (Amal-Mana, Sabita, CSRC (D) 7-0-4 and CSRC (D) 2-17-5) Effects Dura on (days) Plant height (cm) Panicles hill -1 Grains panicle-1 1000 grain weight Boro Variety (V) Year (Y) V and Y interac on Site (S) V and S interac on S and Y interac on V, Y and S interac on P>F 0.981 0.602 0.981 0.954 0.962 0.328 0.594 P>F <0.05 0.621 0.919 0.848 0.886 0.959 0.999 P>F 0.335 0.756 0.960 0.756 0.782 0.536 0.941 P>F <0.05 1.000 1.000 <0.05 0.767 1.000 0.377 P>F <0.05 0.778 0.456 <0.05 0.879 0.925 0.970 Aman Variety (V) Year (Y) V and Y interac on Field water depth (D) V and D interac on D and Y interac on 0.007 0.899 0.909 0.704 0.740 0.527 <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 <0.05 0.729 0.517 <0.05 0.720 0.302 <0.05 0.188 0.805 0.052 0.548 <0.05 <0.05 0.962 0.389 0.277 0.820 0.763 V, Y and D interac on 0.951 <0.05 0.720 <0.05 0.706 Rela ve yield of the entries varied over the three years and two sites (Table 9). The most consistently high yielding variety was WGL 20471, which yielded an average of 4.6 to 5.8 t ha-1 in all three years at both sites, except at Sandeshkhali in 2011-12. In 2011-12, grain yield of all varie es was low at Sandeshkhali due to damage by rodents. The higher yield of WGL 20471 than the other three common varie es over the three years was due to higher grain weight in comparison with the other three varie es, more grains per panicle in comparison with IET 4786 and Annada, and slightly higher panicle density in comparison with Annada and Bidhan2 (Table 8). Yield of BRRI dhan47 was similar to that of WGL 20471 in 2012-13 at both sites, and at Gosaba in 2013-14, with the added benefit of slightly shorter dura on. However, WGL 20471 has the locally preferred grain type (long slender grains), whereas BRRI dhan47 grain is medium bold. Yield of BRRI dhan55 was rela vely low at both sites, probably reflec ng its much shorter dura on. The performance of the local varie es (Parijat, N. Sankar, S. Sankar) was always inferior to that of the improved varie es at both sites. 313 Table 8. Mean maturity dura on, plant height and yield components of boro rice entries Effects WGL 20471 IET 4786 Annada Boby Lalat BRRI dhan 47 BINA dhan 8 BRRI dhan 55 Bidhan 2 Parijat N. Sankar S. Sankar IR 87938 NSIC RC 238 NSIC RC 222 Canning 7 CSR 4 Dura on (days) 140 140 140 140 145 135 130 120 140 140 140 140 140 140 140 135 140 Plant height (cm) 98.4 95.8 97.2 102.4 102.1 100.1 100.5 82.7 105.6 87.8 97.1 97.9 98.6 103.7 101.0 99.8 92.2 Panicles hill -1 16 16 15 13 13 17 14 10 15 13 14 10 15 12 15 14 14 Table 9. Yield of boro entries over three seasons 2011-12 2012-13 Varie es SandeshGosaba SandeshGosaba khali1 khali WGL 20471 3.9ab 5.4a 5.4a 5.8a IET 4786 3.8bc 4.8abc 4.8abc 5.2b Annada 3.6c 4.6abc 4.6abc 4.6cd Bidhan 2 4.0a 4.7abc 4.7abc 4.8bc de a Boby 3.1 5.5 e Lalat 2.9 5.0ab Canning 7 3.6c 3.9bc CSR 4 3.3d 3.7bc BRRI dhan47 5.5a 5.8a BINAdhan-8 5.0ab 5.2b BRRI dhan55 3.9bc 4.2de bc Parijat 3.7 3.8ef N. Sankar 3.6bc 3.3g S. Sankar 3.4c 3.4fg IR 87938 NSIC RC 238 NSIC RC 222 LSD 0.05 0.1 0.1 0.8 0.5 Mean 3.5 4.7 4.5 4.6 1 Rat damage reduced yield 314 Grains panicle -1 95 81 82 93 85 83 81 74 95 70 56 55 58 71 81 94 94 1000 grain weight (g) 20.47 19.28 19.32 20.83 21.87 25.85 25.45 21.57 19.41 19.68 17.06 16.80 21.64 23.22 21.46 19.35 18.98 2013-14 SandeshGosaba khali 4.7a 4.6a 4.0c 4.1b 4.2bc 4.3ab 3.9c 4.5ab 4.0b 4.4ab 4.2bc 4.4ab 4.2ab 4.4ab 3.4d 3.4d 4.4ab 0.8 4.1 3.4c 3.5c 4.4ab 0.5 4.1 The rela onship between soil salinity and grain yield varied between varie es, but there was a trend for yield of most varie es to decline with increasing salinity (Table 10). Lalat was the most sensi ve variety, with yield declining by 336 kg ha-1 per unit increase in average soil salinity, which accounted for 56% of the varia on in yield (P<0.1). The least sensi ve varie es were BRRI dhan47, BINAdhan-8 and Bidhan 2. BRRI dhan47 was not affected by salinity up to the maximum experienced across the sites (season mean ECe 7-8 dS m -1). The boro entries were divided into two groups (those more sensi ve to salinity (group A) and those less sensi ve to salinity (group B)) based on the size of the regression co-efficient, P value and R2 for further analysis. The responses of WGL 20471, IET 4786, Annada, Boby and Lalat were sta s cally similar (MSD 1.9), and the combined regression showed yield declining by 241 kg ha-1 per unit increase in average soil salinity. The response of BRRI dhan47, BINAdhan-8 and Bidhan 2 was also sta s cally similar (MSD 1.1) and for this group there was no significant response to salinity up to the degree experienced across the experimental sites. Thus, these varie es are more suited to higher salinity situa ons, and BINAdhan-8 has the advantage of about 10 d shorter dura on than the less salt tolerant entries. Table 10. Results of the analysis of the regression of average soil salinity against grain yield for individual varie es, and for the groups (A and B) Group A A A A A A B B B B Variety WGL 20471 IET 4786 Annada Boby Lalat ALL Bidhan 2 BRRI dhan47 Binadhan-8 ALL n 9 9 9 6 6 39 7 5 5 17 Reg.coeff. -0.25 -0.22 -0.20 -0.20 -0.34 -0.24 -0.11 -0.01 -0.18 -0.15 SEm 0.31 0.24 0.20 0.19 0.33 0.13 0.16 0.40 0.33 0.18 P>F 0.18 0.14 0.08 0.06 0.09 <0.01 0.29 0.96 0.47 0.15 R2 0.24 0.28 0.38 0.64 0.56 0.25 0.22 0.00 0.19 0.13 3.3.3 Aman variety evalua on The floodwater was much deeper at Basan than at Gosaba or Sandeshkhali in August 2012 (Fig. 3a). However, in 2013, depth was similar at Basan and Gosabi and higher than at Sandeshkhali (Fig. 3b). There were two peaks in floodwater depth in 2012 at all three sites due to heavy rains, with depth reaching 50 cm at Basan . In 2013, depth gradually increased to about 50 cm at Basan and Gosaba, and to about 45 cm at Sandeshkhali. At Sandeshkhali, the crop was submerged from 8 to 13 August 2012 due to a flash flood, with adverse effects. Swarna-Sub1 and Amal-Mana recovered well due to submergence tolerance and performed be er than the other entries. During 2013, there was stagnant flooding of >25 cm for a longer period star ng from mid-August to the end of August, at all loca ons. (a) 60 Sandeshkhali Gosaba Basan 50 40 30 20 10 0 315 (b) 60 50 Sandeshkhali Gosaba Basan 40 30 20 10 0 Fig. 3. Field water depth during August 2012 (a) and 2013 (b). The interac ons between variety, site and year were not significant for any of the parameters monitored except for plant height, for the four entries grown in both years (Table 7). Within site and year, there were significant differences in plant height between varie es (Table 11). Swarna-Sub1, BRRI dhan47 and Binadhan-8 were much shorter than all other varie es, at around 100 cm. Within variety, plant height varied across sites due to differences in water depth. However, the response varied greatly with variety. Regression analysis showed that Amal-Mana, CSRC(D) 2-17-5, CSRC(D) 12-8-12, Sabita and Patnai 23 responded the most strongly (R2>0.9) to field water depth. Table 11. Plant height (cm) of aman rice entries with different water depths and years 2012 Sandeshkhali Gosaba Amal-Mana 139 141 Sabita 128 144 CSRC (D) 7-0-4 128 130 CSRC (D) 2-17-5 145 152 CSRC (D) 12-8-12 130 136 CSRC (D) 13-16-9 146 152 Swarna-Sub 1 95 105 Geetanjali 138 NC 678 141 SR 26 B 130 Patnai 23 BRRI dhan47 Binadhan-8 LSD 0.05 (variety×year×site): 2.7 Entries Basan 168 172 173 167 177 157 107 147 143 138 - 2013 Sandeshkhali Gosaba 162 176 161 173 143 163 165 175 152 184 149 168 140 161 98 102 99 102 Basan 169 178 150 173 176 157 164 101 101 The dura on of most aman varie es was around 160-165 d (Table 12). However, the dura on of Swarna-Sub1 (140 d), BRRI dhan47 (130 d) and BINAdhan-8 (125 d) was much less. Shorter dura on means earlier harvest and the possibility of earlier establishment of the subsequent boro crop, reducing irriga on requirement for land prepara on and resul ng in earlier boro maturity, meaning that the crop is exposed to lower soil salinity (Sarangi et al. 2015a). The highest number of panicles hill -1 (9) and grains panicle-1 (121-128) were observed in Amal-Mana and CSRC (D) 12-8-12. 316 Table 12. Mean maturity dura on and yield components of aman rice entries Entries Amal-Mana Sabita CSRC (D) 7-0-4 CSRC (D) 2-17-5 CSRC (D) 12-8-12 CSRC (D) 13-16-9 Swarna-Sub 1 Geetanjali NC 678 SR 26 B Patnai 23 BRRI dhan47 Binadhan-8 Dura on (days) 155-175 160-210 155-175 155-175 155-175 155-175 140-145 160-175 165-185 160-175 160-175 130-145 125-140 Panicles hill-1 9 7 8 9 9 8 8 8 7 8 7 7 7 Grains panicle-1 121 119 115 112 128 107 89 80 123 102 96 93 93 1000 grain weight (g) 29.00 27.17 26.50 27.17 26.33 27.50 25.50 27.50 28.00 23.50 24.00 26.93 26.38 There were significant differences in yield of the aman varie es within site x year (Table 13). Rela ve yield varied with site and year. Amal-Mana consistently performed well across sites and years, with yield ranging from 4.4 to 5.0 t ha-1. Swarna-Sub1 also performed well across all sites in the one year that it was evaluated (4.2 to 4.4 5.0 t ha-1). The two Bangladeshi varie es performed well at Sandeshkhali where water depth was least, but poorly at the higher water depth sites. The results of regression analysis suggested that the entries fell into three groups. For group A varie es, yield increased with water depth. For group B, there was no effect of water depth on yield, and in group C yield decreased with water depth (Table 14). Table 13. Grain yield (t ha-1) of aman rice entries at different water depths Entries Amal-Mana Sabita CSRC (D) 7-0-4 CSRC (D) 2-17-5 CSRC (D) 12-8-12 CSRC (D) 13-16-9 Swarna-Sub 1 Geetanjali NC 678 SR 26 B Patnai 23 BRRI dhan47 Binadhan-8 LSD 0.05 2012 Sandeshkhali Gosaba 3.80ab 4.55a 2.70c 2.15e 3.70b 3.70cd 3.05c 3.60d b 3.52 4.15bc 3.50b 3.50d 4.15a 4.38ab 3.96c 3.45d 3.42d 0.36 0.35 Basan 4.40b 2.43d 3.70c 3.60c 4.80a 3.50c 4.20b 3.82c 3.50c 3.60c 0.36 2013 Sandeshkhali Gosaba 4.96a 4.51a 3.34d 4.63a 4.86ab 4.36a 4.73ab 4.57a 3.85 3.91b b 4.51 3.85b 4.13c 3.86b 4.26bc 3.31c 4.23bc 3.32c 0.37 0.42 Basan 4.50a 3.01c 4.79a 4.24ab 4.58a 4.12b 3.81b 2.95c 2.91c 0.59 Note: Means with the same le er in a column are not significantly different at the 5% level by LSD 317 Table 14. Results of the analysis of the regression of average field water depth against grain yield of aman varie es for individual varie es, and for the groups (A, B, C) Group A A A A A A B B B B C C C C C Variety CSRC(D) 12-8-12 Sabita Patnai 23 CSRC(D) 2-17-5 Amal-Mana ALL CSRC(D) 13-16-9 BRRI dhan47 CSRC(D) 7-0-4 ALL Swarna-Sub 1 Geetanjali SR 26B NC 678 ALL n 9 18 9 18 18 72 9 9 18 45 9 9 15 15 48 Reg.coeff. 2.4 2.1 2.3 1.2 1.6 0.03 0.46 0.34 1.66 0.01 0.46 0.02 1.26 1.62 0.004 SEm 0.37 0.36 0.10 0.27 0.15 0.16 0.01 0.39 0.23 0.16 0.07 0.09 0.19 0.20 0.09 P>F 0.20 0.40 0.05 0.21 0.39 0.12 0.37 0.02 0.32 0.78 0.93 0.46 0.72 0.31 0.73 R2 0.99 0.98 0.98 0.93 0.91 0.11 0.88 0.79 0.76 0.01 0.60 0.57 0.41 0.39 0.008 4. Conclusions and recommenda ons During the dry season, lack of fresh water and soil salinity are major constraints to rice produc on in the coastal zone of West Bengal. Several improved varie es performed be er than the varie es commonly grown by farmers during the dry season. The most consistently high yielding variety across sites with varying soil salinity was WGL 20471, with yields usually in the range 4.6 to 5.8 t ha-1. However, this line was more sensi ve to soil salinity than the Bangladeshi varie es BRRI dhan47 and BINAdhan-8, and than the Indian variety Bidhan 2. BRRI dhan47 was the least sensi ve variety to soil salinity, with no effect of soil salinity on yield up to the maximum salinity experienced in the trials in farmers’ fields (season average ECe 7-8 dS m-1). During the rainy season, water depth (which reflects land eleva on) is an important criterion for the selec on of suitable varie es. Varie es varied in their response to water depth. Amal-Mana was consistently the best performer (3.8 to 5.0 t ha-1) across all water depths experienced. The short statured Bangladeshi varie es (BRRI dhan47, BINAdhan-8) performed well at the shallowest water depth (which increased gradually from about 15 to 45 cm during the first three weeks of August), but performed poorly with deeper water. Swarna-Sub1, while also short statured, performed as well as Amal-Mana across the range of water depths in the one year that it was evaluated. The performance of several CSRC (D) lines was also encouraging across a range of water depths, although not always as good as Amal-Mana. Acknowledgements This paper presents findings from G2 project on “Produc ve, profitable, and resilient agriculture and aquaculture systems”, a project of the CGIAR Challenge Program on Water and Food. References Burman, D., Mahanta, K.K., Sarangi, S.K., Mandal, Subhasis, Maji, B., Mandal, U.K., Bandyopadhyay, B.K., Humphreys, E. and Sharma, D.K. Effect of groundwater use on groundwater salinity, piezometric level and boro rice yield in the Sundarbans of West Bengal. These proceedings. 318 Gomez, K.A., and Gomez, A.A., 1984. Sta s cal procedures for agricultural research. John Wiley and Sons, New York. Gupta, P.K., 2006. Soil, plant, water and fer lizer analysis. Jodhpur: Agrobios (India), Jodhpur. Hanway, J.J., and H. Heidel. 1952. Soil analyses methods as used in Iowa State College Soil Tes ng Laboratory. Iowa Agric. 57:1-31. Hossain, M., 2003. Development of boro rice cul va on in Bangladesh: Trends and policies.In: Boro Rice. R.K. Singh, M. Hossain and R. Thakur editors, IRRI-India Office, New Delhi, 25-42. Howell, D.C., 2013. Sta s cal methods for pshychology. Wadsworth, Cengage Learning, Belmont, USA. ICAR-CSSRI, IRRI-EC-IFAD, 2014. Final report of IRRI-EC-IFAD project on improved rice crop management for raising produc vity in submergence-prone and salt-affected rainfed lowlands in South Asia. B. Maji, Sukanta K. Sarangi, D. K. Sharma and Sudhanshu Singh editors, ICAR-Central Soil Salinity Research Ins tute (ICAR-CSSRI), Regional Research Sta on (RRS), Canning Town – 743 329, South 24 Parganas, West Bengal, India. 72 p. Ismail, A.M., Singh, U.S., Singh, S., Dar, M.H.,Mackill, D.J., 2013. The contribu on of submergence-tolerant (Sub 1) rice varie es to food security in flood-prone rainfed lowland areas in Asia.Field Crops Res.152, 83-93 h p://dx.doi.org/10.1016.j.fcr.2013.01.007. Mondal, M.K., Saha, N.K., Ritu, S.P., Paul, P.L.C, Sharifullah, A.K.M., Humphreys, E., Tuong, T.P. and Rashid, M.A. 2015. Op mum sowing window for boro cul va on in the coastal zone of Bangladesh. These proceedings. Olsen, S.R., C.V. Cole, F.S. Watanabe,and L.A. Dean. 1954. Es ma on of available phosphorus in soils by extrac on with sodium bicarbonate. USDA Circular 939:1-19, Gov. Prin ngoffice Washington D.C. Sarangi, S.K, Burman, D., Mandal, S., Maji, B., Tuong, T.P, Humphreys, E., Bandyopadhyay, B.K. and Sharma, D.K. 2015a.Reducing irriga on water requirement of dry season rice (boro) in coastal areas using mely seeding and short dura on varie es. These proceedings. Sarangi, S.K., Maji, B., Singh, S., Burman, D., Mandal, S., Sharma, D.K., Singh, U.S., Ismail, A.M., Haefele, S.M. 2015b. Improved nursery management further enhances the produc vity of stress-tolerant rice varie es in coastal rainfed lowlands. Field Crops Res. 174, 61-70. h p://dx.doi.org/10.1016/j.fcr.2015.01.011 Sarangi, S. K., Maji, B., Singh, S., Sharma, D.K., Burman, D., Mandal, S., Ismail, A.M., Haefele, S.M., 2014. Crop establishment and nutrient management for dry season (boro) rice in coastal areas. Agronomy Journal 106(6), 2013-2023. DOI: 10.2134/agronj14.0182. Singh, S., Mackill, D.J., Ismail, A.M., 2009. Response of SUB1 rice introgression lines to submergence in the field: Yield and grain quality. Field Crops Res. 113, 12-23. DOI: 10.1016/j.fcr.2009.04.003. Subbaiah, G.V., Asija, G.L., 1956. A rapid procedure for the determina on of available nitrogen in soils. Current Science 25: 259-260. Walkley, A.J., and I.A. Black. 1934. Es ma on of soil organic carbon by the chromic acid tra on method. Soil Science 37:29-38.doi: 10.1097/00010694-193401000-00003. Yadav, J.S.P., Bandyopadhyay, A.K., Bandyopadhyay, B.K., 1983. Extent of coastal saline soils in India. Journal of the Indian Society of Coastal Agricultural Research 1: 1-6. 319 Performance of improved aman rice varie es in the coastal zone of Bangladesh M.R.A. Sarker1, M.A. Rahman1, 2, N. Sharma1, M.R. Islam2, M.K. Mondal2, G.B. Gregorio2, E. Humphreys2 and T.P. Tuong2 1 Bangladesh Rice Research Ins tute, Bangladesh, mrasbrri@yahoo.com, akhlas08@gmail.com, nirmal_brri@yahoo.com 2 Interna onal Rice Research Ins tute, Bangladesh and Philippines, akhlasur.rahman@irri.org, m.mondal@irri.org, mr.islam@irri.org, g.gregorio@irri.org, e.humphreys@irri.org, t.tuong@irri.org Abstract Varying levels of salinity and water stagna on affect about 1 Mha of cul vable lands in coastal areas of Bangladesh. Most farmers grow tradi onal photoperiod sensi ve, tall aman varie es that are well adapted to water stagna on but with low yield poten al and long dura on, resul ng in late rabi crop establishment and precluding the ability to diversify to high yielding and/or high value rabi crops. Over the past decade, high yielding varie es of aman rice with tolerance to salinity, submergence and water stagna on, and with mild to no photoperiod sensi vity, have been developed. Therefore we sought to iden fy improved aman varie es suited to the coastal zone and their op mum sowing dates. Twelve rice varie es were tested in the aman season at low (polder 43/2F) and medium (polder 30) salinity loca ons in farmers’ fields over three consecu ve years and three seeding dates (24 June, 15 July, 6 August). Water depth across sites and seasons ranged from 0 (soil satura on) to 50 cm and salinity ranged from 0.2 to 2 dS/m. At the low salinity site, the op mum seeding date for most varie es was late June to mid-July, and yield declined with a delay in sowing to early August. With sowing at the op mum me, BRRI dhan52 and BR11-Saltol (5.3-6.0 t/ha) were the most consistent and highest yielding varie es, followed by BRRI dhan51 (4.8 to 5.4 t/ha). The highest yielding varie es had dura on of 130-146 d for the late June and mid-July sowings, meaning that they would reach maturity in mid- to late November. Given that the op mum sowing date for boro crops in the region is mid-November to mid-December the highest yielding varie es would be suitable for use in high yielding aman-boro cropping systems. However, their maturity is too late for aman-rabi systems in which the rabi crops should be sown in early December for maximum yield. The highest yielding aman varie es with short enough maturity for a high yielding aman-rabi system were BRRI dhan53 (~123 d, 4.7-5.1 t/ha) and BRRI dhan33 (110-120 d, 4.2-4.7 t/ha) for sowings on 24 June and 15 July. At the medium salinity site there was li le effect of sowing date on yield, in contrast with the low salinity site. The most consistent higher yielding varie es across years and sowing dates were BR11-Saltol (4.3-5.8 t/ha), BRRI dhan54 (4.1-5.8 t/ha) and BRRI dhan52 (4.1-6.0 t/ha). The dura ons of BR11-Saltol and BRRI dhan52 were suitable for intensifica on to aman-boro. BR11-Saltol and BRRI dhan52 would also be suitable for aman-rabi systems provided that the aman crops were sown early (late June). Yields of the shorter dura on varie es most suited to aman-rabi systems (BBRI dhan53 and 57 and BINAdhan-8) were more variable across years and sowing dates, with yields ranging from 3.1 to 6.3 t/ha. For the 24 June and 15 July sowing dates, yields were generally lower at the medium salinity site than at the low salinity site. This was par cularly so for the long dura on varie es (BR23 and BRRI dhan41), with yields in polder 30 at 73-80% of those in polder 43/2F. Yield of BINAdhan-8, a salt tolerant variety, was similar at both sites for 24 June sowings (4.6-4.7 t/ha) and 15 July sowings (4.7-5.0 t/ha). However, the other salt tolerant varie es generally performed be er at the low salinity site than at the high salinity site. The results demonstrate the great poten al for increasing rice produc on in the coastal zone in the aman season by adop on of improved varie es. However, their earlier maturity meant that the crops were vulnerable to a ack by rats and birds as the surrounding crops matured much later. Thus the adop on of improved varie es requires a synchronized community approach. 320 Key message: BRRI dhan44, BRRI dhan52, BRRI dhan53, BRRI dhan54, BR11-Saltol are the most suitable aman varie es in low and medium salinity areas of the coastal zone, and produced at least 50% higher yield than the popular local varie es when sown in mid-July. Keywords: salinity, polder, synchronized community based approach 1. Introduc on The cul va on of tall (>130 cm), tradi onal, photoperiod sensi ve and late maturing varie es is one of the major constraints to cropping intensifica on and diversifica on in low and medium salinity areas of the coastal zone of Bangladesh. These varie es are adapted to the hydrological condi ons in the region during the rainy season when water stagna on of 20-60 cm prevails over much of the cul vable land area. The long dura on and photoperiod sensi vity of tradi onal varie es means that they are generally unsuitable for cropping system intensifica on. While there has been great progress in the development of modern high yielding varie es of rice (HYV) with earlier maturity, many of the earlier HYV were not suitable for large parts of the coastal zone because of their short stature and thus poor performance as a result of submergence a er transplan ng and water stagna on, and because of high salinity in some areas. Environmental condi ons vary greatly over small distances in the coastal zone, in par cular the incidence and severity of water stagna on and salinity. Furthermore, the preferred grain type varies with loca on. However, moderately salt tolerant (EC = 8 dS m-1) rice varie es (BR23, BRRI dhan40, BRRI dhan41) have limited adaptability in the coastal zone due to short seedling height and sensi vity to prolonged water stagna on (Salam et al. 2010; Rahman et al. 2013). While drainage of water from the fields following excessive rains is possible by systema cally opening the sluice gates at low de (Mondal et al. 2015) the community coordina on and infrastructure improvements required to reduce submergence and water stagna on across the coastal zone are not yet in place. Tolerance to water stagna on and submergence would thus be highly desirable to enable adop on of HYV, and also to reduce the drainage requirement and thus the investment in infrastructure and management. Improved rice varie es have recently been developed which are able to give high yield under condi ons of moderate water stagna on and/or low to medium soil salinity. Some of these varie es have li le or no photoperiod sensi vity, and some are short or medium dura on, and thus more suited to cropping system intensifica on. Moreover, poten al yield of these varie es is roughly double that of tradi onal varie es. Therefore, experiments were undertaken to evaluate improved aman varie es to iden fy those most suited to the coastal zone and their op mum sowing dates to increase cropping intensity and diversity. 2. Methods 2.1 Site descrip on Replicated variety evalua on trials were conducted in farmers’ fields at two sites in low and medium salinity regions of the coastal zone during three aman seasons (2011-2013). The low salinity site was located at Bazarkhali village in polder 43/2F, and the medium salinity site was at Hatba in 2011, but was shi ed to Kismat Fultola village (2012, 2013) in polder 30 due to uncontrolled flooding at Hatba from an open flushing gate. All sites were located adjacent to a khal. Bazarkhali is located in Amtali Upazila in Barguna District, near the Dalachara River, which has low salinity (EC <1 dS/m) throughout the year (Khan and Kamal 2015). Soil salinity (satura on extract, ECe) in Amtali Upazila varies from 2 to 8 dS/m (SRDI 1998). The major cropping pa erns in Amtali Upazila are transplanted aman-fallow-transplanted aus and transplanted aman-grass pea-fallow. The aman crops are grown using tall, photoperiod sensi ve local varie es. Kismat Fultola and Hatba are located in Ba aghata Upazila of Khulna District, beside the Kazibacha River. Soil salinity in Ba aghata Upazila varies from 3 to 12 dS/m and the river water salinity of Kazibacha River increases 321 as the dry season progresses to a peak value of about 20 dS/m in June (Mondal et al. 2006). The river becomes too saline for irriga on (>4 dS/m) in early to mid-February. The major cropping pa ern in Ba aghata is transplanted aman-fallow-sesame using local varie es of both crops. 2.2 Experimental design Ten to eleven modern HYV and advanced lines were evaluated and compared with the popular local variety at each site (Table 1) for three sowing dates. The trial for each sowing date was in a separate block. There were three replicates in a randomized complete block design within each sowing date. Individual plot size was 25 m2 in the first replicate and 12.5 m2 in the second and third replicates. 2.2.1 Variety Several HYV were in common to both sites, however some varie es were selected based on preferred grain type (bolder grain types preferred in Amtali, more slender types in Ba aghata). There was greater inclusion of salt tolerant varie es at Ba aghata where soil salinity is higher. The local check was Morichsail at Hatba , Kumri at Kismat Fultola and Sadamota at Bazarkhali. Table 1. Characteris cs of aman genotypes used in the experiments (BRRI, 2013) Years evaluated in each polder Genotype 43/2F 30 Varietal Characteris cs Dura on Plant (days) height (cm) Yield (t/ha) Photo-period sensi vity Stress tolerances Grain type BR23 2012-2013 2011-2013 150 (L) 120 5.5 Sensi ve Salinity slender Medium BRRI dhan30 2011 145 (L) 120 5.0 Weakly sensi ve - Medium Bold BRRI dhan33 2011-2013 118 (S) 100 4.5 Non-sensi ve - Bold BRRI dhan39 2011 122 (S) 106 4.5 Non-sensi ve - Medium slender BRRI dhan40 2011 145 (L) 110 4.5 Sensi ve Salinity Medium bold BRRI dhan41 2011-2013 2011-2013 148 (L) 115 4.5 Sensi ve Salinity Long bold BRRI dhan44 2011-2013 2011-2013 145 (L) 130 5.5 Sensi ve Tidal submergence Bold BRRI dhan49 2011-2013 2011-2013 135 (M) 100 5.5 Non-sensi ve - Medium slender BRRI dhan51 2011-2013 142 (L) 90 4.5 Weakly-sensi ve Flash flood submergence Medium slender BRRI dhan52 2011-2013 2011-2013 145 (L) 116 5.0 Weakly sensi ve Flash flood submergence Medium Bold BRRI dhan53 2011-2013 2011-2013 125 (S) 105 4.5 Non-sensi ve Salinity Slender BRRI dhan54 2011-2013 2011-2013 135 (M) 115 4.5 Sensi ve Salinity Slender BRRI dhan57 2012-2013 2012-2013 105 (S) 115 4.5 Sensi ve - Long BINAdhan-8 2011-2013 2011-2013 150 (L) 105 6.0 Non-sensi ve Salinity Bold IR84649-120-8-1-B (Saltol+Sub1) 2011 145 (L) 100 4.5 Non-sensi ve Salinity and Flash Bold flood submergence IR84645-34-9-1-B (Saltol+Sub1) 2011 145 (L) 98 4.5 Non-sensi ve Salinity and Flash Medium slender flood submergence 145 (L) 115 5.5 Weakly sensi ve Salinity Medium bold 130 (M) 90 5.5 Non-sensi ve Salinity Medium Slender BR11-Saltol (IR89574-7) 2012-2013 2012-2013 BRRI dhan28-Saltol (IR89573-84) Sadamota (Local) 2012-2013 165 (L) 150 3.5 Sensi ve Tidal submergence Bold Morichsail (Local) 2011 165 (L) 160 3.5 Sensi ve Tidal submergence Bold Kumri (Local) 2012-2013 165 (L) 160 3.5 Sensi ve Tidal submergence Bold 322 2011-2013 2.2.2 Seeding date There were three target seeding dates at 21 d intervals at each loca on (24 June, 15 July, 6 August), but this was not always possible. In 2011 flooding as a result of heavy rain destroyed the seedling nurseries for the second and third sowings at Hatba . Flooding also destroyed the seedbed for the third sowing in 2012 at Kismat Fultola. Table 2. Seeding and transplan ng dates Year Bazarkhali 2011 2012 2013 Hatba /Kismat Fultola 2011 2012 2013 Actual seeding date Transplan ng date Seedling age at transplan ng 24 Jun 15 Jul 16 Aug 24 Jun 15 Jul 7 Aug 24 Jun 15 Jul 6 Aug 15 July 4 Aug 5 Sep 15 July 13 Aug 28 Aug 15 July 4 Aug 27 Aug 21 21 21 21 29 21 21 21 21 1 July 24 Jun 15 Jul 26 Jun 14 Jul 6 Aug 26 July 15 July 6 Aug 16 July 3 Aug 27 Aug 25 21 21 21 21 21 2.3 Management Pre-germinated seeds were sown at 16-20 kg/ha on the seedbed and urea was applied at 30 kg N/ha. Land prepara on involved three to four passes (cross plowing) in wet soil using a power ller powered by a two-wheel tractor, followed by leveling using a ladder. Before land leveling, fer lizer was broadcast at 11:42:18:3.6 kg of P:K:S:Zn per hectare in the forms of triple super phosphate, muriate of potash, gypsum, and zinc sulphate (BRRI 2013). Urea was broadcast at 70 kg N/ha in three equal installments: 7 to 10 d a er transplan ng (DAT), 4 to 5 ller stage, and 5 to 7 d before panicle ini a on. Twenty-one day-old seedlings were transplanted, except for the only sowing at Hatba in 2011, and the second sowing at Bazarkhali in 2012 (Table 2) when transplan ng was delayed by 4 and 8 d, respec vely, because the water was too deep on the target date. The seedlings in the nursery for the third sowing in 2012 at Bazarkhali were partly damaged due to submergence and there were only sufficient viable seedlings to transplant two replicates. Two to three seedlings per hill were transplanted with 20 cm x 15 cm geometry in all experiments. Hand weeding was done three mes, prior to each topdressing with urea. Disease and insect pests were always well controlled using recommended prac ces (BRRI 2013). Early maturing varie es of the first sowing were severely damaged by rats at both sites in 2011 and at Bazarkhali in 2012, a er which a plas c rat barrier was installed. Bird nets were also installed over the experimental field prior to grain filling to protect the crops from bird damage. These protec on measures were needed due to the earlier maturity of most of the varie es in comparison with the farmers’ varie es in surrounding fields. 323 2.4 Monitoring 2.4.1 Crop Growth dura on was calculated as the number of days from soaking of the seeds to physiological maturity (PM). Physiological maturity was taken as the me when 80% of the grains were hard or translucent. Grain yield was determined in a 5.1 m x 2 m area in the middle of each plot, moisture content was determined using a grain moisture meter, and yield was adjusted to 14% moisture content. 2.4.2 Water depth and salinity Water depth was measured daily at 16 staff gauges installed across the experimental field. Water salinity (electrical conduc vity, EC) was measured weekly using a portable EC meter. 3. Results and discussion 3.1 Bazarkhali 3.1.1 Water depth The water depth in the rice field varied widely within and across years (Fig. 1) depending on the rainfall and river des and management of the sluice gate. Generally water depth varied from 1 to 20 cm, except in the first week of August 2011 (data not presented) and September 2012. In August 2011, water depth increased to 40-45 cm 2 to 3 d a er transplan ng of the seedlings of the second sowing. The seedlings were submerged for about 5 d and the crops in replicate 3 and in half of the plots in replicate 2 were destroyed. In September 2012, water depth increased to almost 50 cm due to very high rainfall. This occurred about a week a er transplan ng of the third sowing, as a result of which the seedlings were submerged or almost submerged for about one week. 60 50 40 2011 2012 2013 30 20 10 0 Fig. 1. Water depth in the rice field at Bazarkhali during aman 2011-2013. 3.1.2 Growth dura on Growth dura on varied significantly among the varie es for all sowing dates each year and ranged from 98-105 d for BRRI dhan57 sown on 6 August to 161-181 d for the local variety (Sadamota) sown on 24 June (Table 3). The genotypes varied greatly in photoperiod sensi vity. Delaying sowing of Sadamota from 24 June to 6 August decreased its dura on by 40-45 d, with PM of all sowing dates occurring in mid-December (Fig. 2). BR23, BRRI dhan41 and BRRI dhan54 were also highly photoperiod sensi ve, with dura on decreasing by 324 30-41 d with delay in sowing from 24 June to 6 August (when not affected by submergence). Most of the other HYV had some photoperiod sensi vity, except for BINAdhan-8 and BRRI dhan53. Submergence a er transplan ng of the second sowing in 2011 increased the dura on of most HYV by 1-9 d, and it also increased the dura on of the most photoperiod sensi ve varie es in comparison with what it would have been in the absence of submergence. Table 3. Growth dura on (days) of aman varie es/genotypes at Bazarkhali Variety BR23 BRRI dhan30 BRRI dhan33 BRRI dhan39 BRRI dhan41 BRRI dhan44 BRRI dhan51 BRRI dhan52 BRRI dhan53 BRRI dhan54 BRRI dhan57 BINAdhan-8 BR11-Saltol Sadamota LSD (0.05) 24 Jun 2011 15 Jul1 16 Aug 137 117 118 153 138 140 135 118 140 142 121 121 140 142 141 140 127 135 124 111 113 112 120 134 126 122 132 113 118 118 168 0.9 161 133 0.3 1 Submerged for 5 d shortly a er transplan ng 2 RD=rat damage 3 GF=Germina on failure (a) 2012 15-Jul 143 6-Aug 119 24-Jun 161 2013 15-Jul 139 6-Aug 121 RD2 110 112 121 117 101 153 138 146 144 RD2 142 RD2 RD2 139 165 1.7 128 127 136 129 124 126 105 130 129 148 1.5 GF3 125 131 126 123 112 105 126 131 125 1.8 153 141 143 141 123 136 113 125 143 181 0.3 139 139 143 139 120 125 114 122 139 161 0.3 115 118 121 118 112 107 98 116 118 136 1.3 (b) 180 400 Kumri Maturity date (Julian day) Sadamota 150 Dura on (days) 24-Jun 156 120 90 60 30 0 350 Kumri Sadamota 11& 10 Dec 18& 20 Dec 15 Dec 300 250 200 150 100 50 0 24 Jun 15 July Sowing date 6 Aug 24 Jun 15 July Sowing date 6 Aug Fig. 2. Growth dura on (a) and maturity date (b) of photosensi ve local varie es at Bazarkhali (Sadamota) and Kismat Fultola (Kumri) as affected by seeding date. 3.1.3 Yield There were significant differences in yield between varie es for all sowing dates in all years (Table 4). There was considerable damage of the earliest maturing varie es of the first sowing in 2011 and 2012 due to rats. In 325 the first year, the first sowing also suffered from shading. As a result, yields of the first sowing were much lower than of the first sowing in later years. In 2011, the second sowing was completely submerged from 5-9 August, causing considerable damage. However the crops in replicate 1 and in most of the plots in replicate 2 survived and went on to yield 3.3 to 4.8 t/ha. The best performers were BRRI dhan44, which is tolerant to dal submergence, and BRRI dhan51 and 52, which have the Sub-1 gene and are tolerant to flash flooding for up to two weeks a er transplan ng. Yield of the local variety was consistently lower than yield of the HYVs, except for similar yield with BRRI dhan33 and 57 (both short dura on, 110 and 105 d) in the second sowing in 2012. In 2012 and 2013, yield of Sadamota was 73, 74 and 79% of the yield of the mean of all the HYV for the 24 June, 15 July and 6 August sowings. The data from 2012 and 2013 were used to analyze the effect of sowing date. Yield of the third sowing was consistently lower than yield of the first two sowings (Fig. 3). The effect was least with the shortest dura on varie es (BRRI dhan33 and 57). There was a general trend for slightly higher yield of the second sowing than the first sowing. Mean yield of the HYV for the 24 June, 15 July and 6 August sowings was 4.8, 5.1 and 3.9 t/ha, respec vely. In comparison, mean yield of Sadamota was 3.5, 3.8 and 3.1 t/ha. Table 4. Yield performance of aman varie es/genotypes at Bazarkhali Genotype BR23 BRRI dhan30 BRRI dhan33 BRRI dhan39 BRRI dhan41 BRRI dhan44 BRRI dhan51 BRRI dhan52 BRRI dhan53 BRRI dhan54 BRRI dhan57 BINAdhan-8 BR11-Saltol Sadamota (L.) LSD (0.05) 1 2011 24-Jun 15-Jul 1 16-Aug 3.7 RD2 3.2 2.8 2.9 4.3 2.6 RD2 3.7 3.4 3.9 3.8 3.3 4.8 4.1 4.4 3.9 4.2 3.0 3.5 3.5 3.3 3.2 2.3 3.4 3.2 3.4 RD2 3.9 3.2 1.5 1.0 2.4 1.8 0.5 24-Jun 5.1 2012 15-Jul 5.5 6-Aug 3.7 24-Jun 4.1 2013 15-Jul 4.6 6-Aug 4.9 RD2 4.2 3.9 4.6 4.7 4.5 5.5 5.4 4.8 5.6 RD2 5.1 RD2 RD2 5.5 3.5 0.9 5.3 5.0 5.3 5.3 4.7 5.6 4.2 4.8 5.8 4.2 0.6 GF3 3.5 3.6 3.3 4.0 4.0 2.8 3.9 3.4 2.5 0.8 3.9 4.2 4.9 5.8 5.0 4.6 3.4 4.6 5.3 3.5 0.7 5.2 5.1 5.4 6.0 5.1 5.4 4.2 4.5 6.0 3.3 0.7 3.7 4.1 4.2 4.9 4.3 3.9 3.3 3.8 4.4 3.7 0.3 Submerged for 5 d shortly a er transplan ng, results are for one or two plots only 2 RD=rat damage 3 GF=Germina on failure 326 24-Jun 15-Jul 06-Aug 6 4 2 0 Fig. 3. Effect of sowing date on yield of aman genotypes at Bazarkhali (average of 2012 and 2013). Ver cal capped bars are SE. The results suggest that late June to early July is the op mum sowing window for both HYV and Sadamota at Bazarkhali. With sowing at this me, BRRI dhan52 and BR11-Saltol (5.3-6.0 t/ha) were the most consistent and highest yielding varie es, followed by BRRI dhan51 (4.8 to 5.4 t/ha). BRRI dhan41, 44 and 54 also had yields exceeding 5 t/ha for the first two sowing dates in 2012, and the second sowing date in 2013, while BRRI dhan53 yielded 4.7-5.1 t/ha. The highest yielding varie es had a dura on of around 140 to 150 d for the first sowing, meaning that they would reach maturity in mid- to late November. Given that the op mum sowing date for boro crops in the region is mid-November to mid-December (Mondal et al. 2015), the highest yielding varie es would be suitable for use in an aman-boro cropping system. However, their maturity is too late for aman-rabi systems with high yielding varie es such as maize and sunflower, which need to be sown in early December for maximum yield (Bha acharya et al. 2015; Rashid et al. 2015) and the soil needs to be drained in early to mid-November to enable it to dry sufficiently for llage or dibbling. The highest yielding aman varie es with short enough maturity for a high yielding aman-rabi system were BRRI dhan53 (~123 d, 4.7-5.1 t/ha) and BRRI dhan33 (110-120 d, 4.2-4.7 t/ha) for sowings on 24 June and 15 July. 3.2 Hatba /Kismat Fultola 3.2.1 Water depth Water depth generally ranged from 0 to 20 cm each year (Fig. 4). In 2011, the 1 July sown crop was almost submerged for about 10 d about two weeks a er transplan ng, and the crop also experienced water stagna on during the grain filling period due to an uncontrolled flushing gate higher in the landscape, allowing dal water to enter. 327 45 40 35 30 25 20 15 10 5 0 2011 2012 2013 Fig. 4. Water depth in the rice field at Hatba /Kisma ultola during aman 2011-2013. 3.2.2 Growth dura on Growth dura on of the genotypes varied significantly for all sowing dates in all years, from 91 d for BRRI dhan57 sown on 6 August 2013 to 172 d for the local variety Kumri sown on 24 June 2012 (Table 5). Growth dura on of all varie es was higher in 2011 due to prolonged water stagna on (Fig. 4) and this effect was more pronounced in the shorter dura on varie es (BRRI dhan53, BINAdhan-8). As at Bazarkhali, there was a consistent trend for decline in dura on with delay in sowing, with the highest photoperiod sensi vity in BR23, BRRI dhan41 and 54, and the local variety (Kumri). BR11-Saltol and BRRI dhan53 showed no and li le photoperiod sensi vity, respec vely, and BINAdhan-8 showed no photoperiod sensi vity in 2013. Table 5. Growth dura on* (days) of aman genotypes at Hatba (2011) and Kismat Fultola (2012-13) Genotype BR23 BRRI dhan41 BRRI dhan44 BRRI dhan49 BRRI dhan52 BRRI dhan53 BRRI dhan54 BRRI dhan57 BINA dhan-8 BR11-Saltol BRRI dhan28-Saltol IR84649-120-8-1-B (Saltol+Sub1) IR84645-34-9-1-B (Saltol+Sub1) Morichsail (L.) Kumri (L.) LSD (0.05) 2011 1-Jul 169 154 140 140 140 158 154 158 2012 24-Jun 161 149 130 132 134 120 133 108 131 131 117 15-Jul 143 138 130 131 120 115 131 102 122 135 115 24-Jun 166 155 148 134 133 112 148 107 113 134 107 2013 15-Jul 152 143 132 122 129 111 134 102 115 125 101 166 2 168 1.3 155 0.3 131 1 158 158 169 1 172 2 * Only one sowing in 2011 and third sowing in 2012 damaged by rainfall on seedbed 328 6-Aug 129 121 122 114 111 106 113 91 112 121 93 3.2.3 Yield There were significant differences in yield between varie es for all sowing dates in all years. In 2011 rats damaged the early maturing varie es. Yields in excess of 4 t/ha were achieved by BR23 and BRRI dhan41 and 52 despite the prolonged water stagna on, while the local variety (Morischsail) only yielded 2.7 t/ha (Table 6). In 2012 and 2013, yields of the HYV ranged from 2.8 to 6.3 t/ha across years and sowing dates, while yield of the local variety (Kumri) ranged from 2.4 to 3.8 t/ha. Yield was generally unaffected by or increased with delay in sowing (Fig. 5), despite the general decrease in crop dura on with delay in sowing at both sites. The effect of sowing date on yield was in contrast with the decline in yield of the last sowing at Bazarkhali. At Kismat Fultola, mean yield of the HYV was 4.4, 4.4 and 4.9 t/ha for sowings on 24 June, 15 July and 6 August, respec vely, compared with 3.1, 3.1 and 2.8 t/ha, respec vely, for the local variety. Yields of the 24 June and 15 July sowings were usually much lower in 2012 than 2013 (the excep ons being both sowings of BR23, and the 15 July sowing of BRRI dhan52 and 53 and BR11-Saltol). The most consistent higher yielding varie es over these two years across all sowing dates were BR11-Saltol (4.3-5.8 t/ha), BRRI dhan54 (4.1-5.8 t/ha) and BRRI dhan52 (4.1-6.0 t/ha). The dura ons of BR11-Saltol and BRRI dhan52 were suitable for intensifica on to aman-boro, given that the op mum sowing date for boro in this region is early November (Mondal et al. 2015). BR11-Saltol and BRRI dhan52 would also be suitable for aman-rabi systems provided that the aman crops were sown in late June. Yields of the shorter dura on varie es most suited to aman-rabi systems (BBRI dhan53 and 57 and BINAdhan-8) were more variable across years and sowing dates, with yields ranging from 3.1 to 6.3 t/ha. 24-Jun 15-Jul 06-Aug 6 4 2 0 Fig. 5. Effect of sowing date on yield of aman genotypes at Kismat Fultola (average of 2012 and 2013). Ver cal capped bars are SE. 329 Table 6. Yield of aman genotypes at Hatba (2011) and Kismat Fultola (2012-13) Genotype BR23 BRRI dhan41 BRRI dhan44 BRRI dhan49 BRRI dhan52 BRRI dhan53 BRRI dhan54 BRRI dhan57 BINA dhan-8 BR11-Saltol BRRI dhan28-Saltol IR84649-120-8-1-B (Saltol-Sub1) IR84645-34-9-1-B (Saltol+Sub1) Morichsail (L.) Kumri (L.) LSD (0.05) 1 2011 1-Jul 4.1 4.1 3.2 3.3 4.2 RD1 3.6 RD1 24-Jun 3.9 3.0 3.6 3.7 4.1 3.7 4.1 3.1 3.1 4.3 2.8 2012 15-Jul 3.9 3.2 4.2 4.3 6.0 4.3 4.3 3.4 3.8 5.8 2.9 24-Jun 3.3 4.5 5.4 5.6 5.5 5.1 5.8 5.4 6.3 5.3 4.3 2013 15-Jul 4.0 4.5 4.8 4.7 4.7 4.5 4.9 4.2 6.1 5.0 4.3 6-Aug 4.2 5.4 5.2 4.9 5.4 5.0 4.8 4.2 5.8 4.8 3.9 3.5 0.6 2.4 0.9 2.7 1.0 3.8 1.2 2.8 0.7 RD1 RD1 2.7 0.6 RD=rat damage For the 24 June and 15 July sowing dates, yields were generally lower at Kismat Fultola than at Bazarkhali, within variety (Figs 6a,b). This was par cularly so for the long dura on varie es (BR23 and BRRI dhan41), with yields (average of 2012 and 2013 within sowing date) at Kismat Fultola of 73-80% of those at Bazarkhali. Yield of BINAdhan-8, a salt tolerant variety, was similar at both sites for 24 June sowings (4.6-4.7 t/ha) and 15 July sowings (4.7-5.0 t/ha). However, the other salt tolerant varie es generally performed be er at Bazarkhali than at Kismat Fultola. This suggests that factors other than salinity were the main cause of the poorer crop performance at Kismat Fultola. a. 24 June sowing 6 4 2 0 330 Bazarkhali Kismat Fultola b. 15-July sowing Bazarkhali Kismat Fultola 6 4 2 0 Fig. 6. Comparison of yield of aman genotypes at Bazarkhali and Kismat Fultola for sowing dates of 24 June (a) and 15 July (b) (average of 2012 and 2013). 4. Conclusions and recommenda ons Aman variety evalua on trials conducted in low and medium salinity loca ons of the coastal zone show the feasibility of achieving high yields with modern high yielding varie es and with much earlier maturity than that of local varie es. In addi on to higher yield (by up to two-fold), the earlier maturity of the HYV would allow for cropping intensifica on through mely establishment of boro or high yielding rabi crops. However, earlier maturing crops were vulnerable to a ack by rats and birds as the surrounding farmers’ crops (local varie es) matured much later. Thus, the adop on of improved varie es with earlier maturity would require a synchronized community approach. Acknowledgements This paper presents findings from G2 ‘Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food. We thank A qur Rahman, Swapan Bhadra, Amal Ray, Lincoln Ray, Tanmoy Ray and Mithun Ray for their technical assistance. References Bha acharya, J., Mondal, M.K., Humphreys, E., Saha, N.K., Rashid, M.H., Paul, P.C. and Ritu, S.P. 2015. Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh. These proceedings. BRRI (Bangladesh Rice Research Ins tute) 2013. Adhunik Dhaner Chash (Modern Rice Cul va on). 17th Edi on. Gazipur-1701, Bangladesh. Khan, Z.H. and Kamal, F.A. 2015. The Ganges coastal zone – current condi ons, future projec ons. These proceedings. 331 Mondal, M. K., T. P. Tuong, S. P. Ritu, M. H. K. Choudhury, A. M. Chasi, P. K. Majumder, M. M. Islam and S. K. Adhikary 2006. Coastal water resource use for higher produc vity: Par cipatory research for increasing cropping intensity in Bangladesh. In C. T. Hoanh, T. P. Tuong, J. W. Gowing and B. Hardy (Eds). Environmental and livelihoods in tropical coastal zones: Managing agriculture-fishery-aquaculture conflicts. Comprehensive Assessment of Water Management in Agriculture Series, no. 2. CABI Publishing p 72-85. Mondal, M.K., Saha, N.K., Ritu, S.P., Paul, P.L.C, Sharifullah, A.K.M., Humphreys, E., Tuong, T.P. and Rashid, M.A. 2015. Op mum sowing window for boro cul va on in the coastal zone of Bangladesh. These proceedings. Rahman, M. A., M. R. A. Sarker, N. Sharma, M. R. Islam, G. B. Gregorio and E. Humphreys. 2013. Turning adversity into opportunity. CURE Ma ers 3(1): page 16. Rashid, M.H., Hossain, F., Nath, D.K., Sarker, P.C., Ferdous, A.K. M. and Russel, T. 2015. Oil seed crops in rice based cropping systems in southern Bangladesh. These proceedings. Salam, M.A., M.R. Islam, M.S. Rahman, M.A. Rahman, M.A.R. Bhuiyan, Z.I. Seraj, T.L. Aditya, M.K. Uddin, M.K. Mondal, A.M. Ismail, D.L. Adorada, R.D. Mendoza, E. Tumimbang-Raiz and G.B. Gregorio. 2010. Rice varie es and cultural management prac ces for high and sustained produc vity in the coastal wetlands of Southern Bangladesh. In: Tropical Deltas and Coastal Zones: Food produc on, Communi es and Environment at the Land-Water Interface. CABI Publisher, Uk. Pp 183-198. SRDI (Soil Resources Development Ins tute). 1998. Land and soil resources use guidelines of Amtali upazila, Barguna. 332 Challenges and opportuni es for aman rice cul va on in ghers used for brackish water shrimp produc on M. A. Rahman 1,2, M. R. A. Sarker2, N. Sharma2, M. K. Mondal3, M. R. Islam 1, G. B. Gregorio1, E. Humphreys1 and T. P. Tuong1 1 Interna onal Rice Research Ins tute, Philippines, akhlasur.rahman@irri.org, m.mondal@irri.org, mr.islam@irri.org, g.gregorio@irri.org, e.humphreys@irri.org, t.tuong@irri.org 2 Bangladesh Rice Research Ins tute, Bangladesh, akhlas08@gmail.com, mrasbrri@yahoo.com, nirmal_brri@yahoo.com Abstract Brackish water shrimp produc on in ghers is a major source of livelihoods for many farming families in the more saline areas of the southwest coastal zone of Bangladesh. In some regions, farmers con nue to grow aman rice during the rainy season a er shrimp harvest. However, the produc vity of rice in shrimp ghers is o en hampered by salinity, submergence, and water stagna on as a result of high rainfall and lack of drainage. Therefore, experiments were conducted to evaluate the performance of modern, high yielding varie es with good tolerance to salinity and submergence in farmers’ ghers from 2011 to 2013. The results showed that with current water management produc on of aman in shrimp ghers is highly risky, but that the risk can be reduced to some degree by the use of stress tolerant varie es. The best performing varie es were those with salinity tolerance (BRRI dhan47, 53, 54) or submergence tolerance (BRRI dhan52). With good water (drainage) management, yields of around 4 t/ha were achieved. Key message: Improved drainage management is essen al for stable and high produc vity of aman rice in shrimp ghers. Keywords: drainage, leaching, salinity, water stagna on, submergence, stress tolerance 1. Introduc on The tradi onal cropping system throughout the southwest coastal zone of Bangladesh was a single rice crop during the rainy season. However, in 1980, the introduc on of brackish water shrimp farming in the more saline parts of the coastal zone changed not only cropping pa erns, but also landholding sizes and land tenure systems (Barmon et al. 2004). The shrimp are cul vated in ghers, which are basically land surround by large levees to pond water for aquaculture. Ghers in the southwest coastal zone are typically 0.25 to 1 ha, but there are some very large ghers, up to several thousand hectares in size. Following the introduc on of shrimp farming, cul va on of aman rice ceased in large areas, however, there are s ll significant areas where farmers prefer to grow aman rice in rota on with brackish water shrimp. The produc vity of rice in ghers used for shrimp culture is o en hampered by salinity, submergence and water stagna on as a result of high rainfall and lack of drainage. In addi on, terminal drought stress and/or terminal salinity a er the rains have ended is some mes problema c for late maturing varie es such as BR23. Therefore, many ghers are fallowed during the rainy season. The high yielding BRRI variety BR23 is usually the preferred variety in the ghers of the southwest coastal zone. While BR23 has good yield poten al (5.5 t/ha under favorable condi ons), it is photoperiod sensi ve and has a long dura on (about 150 d) when sown at the op mum me. Also, BR23 has only moderate salinity tolerance and is sensi ve to submergence. The use of recently developed non-photoperiod sensi ve high yielding varie es with tolerance to salinity at 8-10 dS/m (such as BR11-Saltol, BR28-Saltol, BRRI dhan47, BRRI dhan53 and BINAdhan-8) and with flash flood submergence tolerance (such as BRRI dhan52) (BRRI 2013) may provide an opportunity to increase the produc vity of aman crops grown in rota on with brackish water shrimp in ghers. Therefore, experiments were undertaken to evaluate the performance of aman varie es with improved stress tolerance in farmers’ ghers in the southwest coastal zone of Bangladesh. 333 2. Methods 2.1 Sites Evalua ons of high yielding aman varie es and advanced lines were conducted in farmers’ ghers during the rainy seasons of 2011, 2012 and 2013. All ghers had a history of brackish water shrimp during the dry season and rice during the rainy season, but in some years rice was not grown due to the late onset of the monsoon and thus insufficient leaching of the ghers. All sites were in polder 3 at Sehara village in Kaliganj Upazila, Satkhira District. Different ghers owned and managed by different farmers were used each year. 2.2 Experimental design All experiments involved evalua on of 10 or 11 released Bangladeshi varie es and usually one advanced line in farmers’ ghers. There were three replicates in a randomized complete block design. Plot size in replicate 1 (25 m2) was larger than in the other replicates, as these plots were also used for farmer par cipatory varietal evalua on. Plot size in replicates 2 and 3 was 12.5 m2. Each year, there were two or three seeding/transplan ng dates. In 2011, the experiments were repeated in two ghers owned by different farmers. 2.2.1 Varie es The varie es were selected for evalua on each year based on tolerance to salinity, water stagna on and submergence in comparison with BR23, which is commonly grown in the region (Table 1). Table 1. Characteris cs of aman varie es and lines evaluated in ghers in Sehara, Polder 3, Kaliganj, Satkhira in 2011-2013 Varietal Characteris cs Genotype Year BR23 BRRI dhan40 BRRI dhan41 BRRI dhan44 BRRI dhan47 BRRI dhan52 (BR11-Sub1) 2011-2013 2011 2011-2013 2011-2013 2011-2013 Plant height (cm) 120 110 115 130 105 2011-2013 116 145 (L) 5.5 BRRI dhan53 BRRI dhan54 2011-2013 2011-2013 105 115 125 (S) 135 (M) 4.5 4.5 BRRI dhan57 BINA dhan8 BR11-Saltol BR28-Saltol BR8371-4R-2 IR8465-311-5-1-1-3 2012-2013 2011-2013 2012-2013 2012-2013 2012 2011 115 105 115 90 115 105 105 (S) 155 (L) 145 (L) 130 (M) 145 (M) 135 (M) 4.5 6.0 5.5 5.5 5.5 4.5 1 Dura on Yield 2 Photo-period (days)1 (t/ha) sensi vity Stress tolerances 150 (L) 145 (L) 148 (L) 145 (L) 152 (L) 5.5 4.5 4.5 5.5 6.0 Sensi ve Sensi ve Sensi ve Sensi ve Non-sensi ve Salinity Salinity Tidal submergence Salinity Moderately sensi ve Flash flood submergence Non-sensi ve Salinity Sensi ve Salinity and water stagna on Sensi ve Non-sensi ve Salinity Non-sensi ve Salinity Non-sensi ve Salinity Non-sensi ve Salinity Non-sensi ve Salinity L = long, M = medium, S= short dura on; 2 Under favorable condi ons 334 Grain type Medium slender Medium bold Long bold Bold Bold Medium bold Slender Long Long Medium bold Medium bold Medium slender Long Bold 2.2.2 Seeding and transplan ng date Two seeding/transplan ng dates (“sets”) were implemented in 2011, and three in 2012 and 2013 (Table 2). The original inten on was to transplant 21 d-old seedlings, however, transplan ng was always delayed because the water was too deep as a result of rainfall and inability to drain (the surrounding area was also flooded). The age of seedlings at the me of transplan ng ranged from 26 to 64 d. Table 2. Dates of seeding and transplan ng of rice in the ghers 2011 2012 2013 Seeding date 15 Jul 30 Jul 1 Jul 16 Jul 31 Jul 1 Jul 15 Jul 30 Jul Transplan ng date 20 Aug 25 Aug 3 Sep 3 Sep 16 Sep 19 Aug 9 Sep 14 Sep Seedling age (days) 36 26 64 49 47 49 56 46 2.3 Crop management The farmers drained water from the ghers in July a er harves ng the shrimp and fish. A basal dose of fer lizer (BRRI recommended dose for aman) was applied at 11:42:18:3.6 kg of P:K:S:Zn per hectare, using triple super phosphate, muriate of potash, gypsum, and zinc sulphate. The fer lizer was broadcast on the surface of the saturated, so soil and mixed with the soil by hand. A er mixing of fer lizer to soil the plot was leveled by hand. The rice seedlings were transplanted with two to three seedlings/hill and hill spacing was 20 cm x 15 cm. Nitrogen fer lizer (75 kg N/ha) was applied as 1.8 g urea super granules 7 to 10 d a er transplan ng. The granules were placed at a depth of about 8-10 cm below the soil surface, midway between four hills. 2.4 Monitoring 2.4.1 Water depth and salinity Water depth and salinity were monitored weekly in each gher. Water depth was monitored using a ver cal gauge fixed permanently in the floor of the gher. Salinity was monitored using a portable electrical conduc vity meter. 2.4.2 Agronomic, grain yield and yield components At maturity, plant height from the base to the p of the panicle (cm) was determined from ten randomly selected plants for each replica on. Crop dura on was determined as the number of days from sowing of the seedling nursery to maturity. Maturity was taken as the date when 80% of the grains became physiologically mature, hard and translucent. Grain yield was determined by harves ng the en re area of each plot (replica on 1=25 m2 and replica ons 2 and 3=12.5 m2). The grain was weighed, grain moisture content was determined on a sub-sample by grain moisture meter and yield at 14% moisture was calculated. 2.5 Data analysis The height and yield data were analyzed by analysis of variance using MS, and the least significant difference at 5% probability (LSD) was used to compare the means. 335 3. Results and discussion 3.1 2011 Water management varied between the two ghers – one farmer implemented good drainage management to reduce salinity of the gher (“medium salinity gher”) to provide more favorable condi ons for the rice, while the other farmer preferred to retain saline water for longer to try and increase shrimp yield (“high salinity gher”). As a result, salinity was always below 4 dS/m in one gher, and ranged from about 5 to 7 dS/m in the other gher (Fig. 1). 10 36 day-old seedlings 26 day-old seedlings 8 Poor leaching/water mgt Good leaching/water mgt 6 4 2 0 31-Jul-11 30-Aug-11 29-Sep-11 29-Oct-11 28-Nov-11 Fig. 1. Water salinity in medium (good leaching management) and high (poor leaching management) sali nity ghers in 2011. Arrows indicate dates of transplan ng. Crop dura on of the two sowing dates ranged from 114-119 d (BRRI dhan47, BINAdhan-8) and 149-154 d (BR23) in the medium salinity gher (Fig. 2). Dura on of the second sowing was consistently a few days less than dura on of the first sowing. Plant height at maturity in the medium salinity gher ranged from around 90 cm for BRRI dhan47 and BINAdhan-8 to around 110 cm for BRRI dhan41, 44, 52 and 54 (Fig. 3). Plant height of the first and second sowings was similar within variety. 160 15 Jul sowing, 36 day-old seedlings 30 Jul sowing, 26 day-old seedlings lsd 15 Jul 1.7 lsd 30 Jul 1.7 120 80 40 0 BR 23 Bd 40 Bd 41 Bd 44 Bd 47 Bd 52 Bd 53 Bd 54 BINA 8 IRRI line Fig. 2. Dura on of aman varie es in the medium salinity gher in 2011 (Bd = BRRI dhan; BINA = BINAdhan). Lsd values are for comparing varie es within sowing date at p=0.05. 336 15 Jul sowing, 36 day-old seedlings 30 Jul sowing, 26 day-old seedlings 120 lsd 15 Jul 4.0 lsd 30 Jul 4.2 80 40 0 BR 23 Bd 40 Bd 41 Bd 44 Bd 47 Bd 52 Bd 53 Bd 54 BINA 8 IRRI line Fig. 3. Plant height at maturity of aman varie es in the medium salinity gher in 2011 (Bd = BRRI dhan; BINA = BINAdhan). Lsd values are for comparing varie es within sowing date at p=0.05. Yields in the medium salinity gher were similar for the two seeding date/seedling age combina ons and ranged from 2.7-4.3 t/ha (Fig. 4). The highest yielding varie es were BR23, BRRI dhan44 and BRRI dhan52, with yields consistently in the range of 3.9-4.3 t/ha. About 40% of the varia on in yield was due to varietal dura on (longer dura on varie es tended to have higher yields). There was greater varia on in yield in the more saline gher (Fig. 5). Especially notable was the much poorer performance of the younger seedlings (26 d old) that were transplanted just 5 d a er the older seedlings (36 d old). The younger seedlings did not survive the combina on of salinity (5 dS/m) and high (rela ve to the height of the seedlings) water depth (13 cm) at the me of transplan ng and submergence for about a week in mid-September (Fig. 6) about three weeks a er transplan ng. Only four varie es produced any yield from the younger seedlings, the best being BINA dhan8 at almost 2 t/ha. However, with older seedlings, many of the varie es produced in excess of 3 t/ha, the best being BRRI dhan44 at 4.3 t/ha, compared with 3.4 t/ha from BR23. BRRI dhan40 (with good salt tolerance) completely failed in this situa on due to the water being too deep a er transplan ng. 5 15 Jul sowing, 36 day-old seedlings 30 Jul sowing, 26 day-old seedlings lsd 15 Jul 0.2 lsd 30 Jul 0.1 4 3 2 1 0 BR 23 Bd 40 Bd 41 Bd 44 Bd 47 Bd 52 Bd 53 Bd 54 BINA 8 IRRI line Fig. 4. Variety yields in the medium salinity gher with two sowing date/seedling age treatments in 2011 (Bd = BRRI dhan; BINA = BINAdhan). Lsd values are for comparing varie es within sowing date at p=0.05. 337 5 15 Jul sowing, 36 day-old seedlings 30 Jul sowing, 26 day-old seedlings 4 lsd 15 Jul 0.7 lsd 30 Jul 1.6 3 2 1 0 BR 23 Bd 40 Bd 41 Bd 44 Bd 47 Bd 52 Bd 53 Bd 54 BINA 8 IRRI line Fig. 5. Aman variety yields in the higher salinity gher in 2011 as affected by seedling age/transplan ng date (Bd = BRRI dhan; BINA = BINAdhan). Lsd values are for comparing varie es within sowing date at p=0.05. 30 25 20 15 10 5 0 36 day-old seedlings 31-Jul-11 26 day-old seedlings 30-Aug-11 29-Sep-11 29-Oct-11 28-Nov-11 Fig. 6. Water depth in the high salinity gher in 2011. 3.2 2012 In 2012, transplan ng of the first two sets was delayed to 3 September due to high water depth, with 64 and 49 d old seedlings. Water salinity was 6 dS/m at the me of transplan ng (Fig. 7). The trial was completely submerged for 5 d (from 6-10 September as a result of heavy rainfall) star ng 2 d a er transplan ng. Survival was poor for all varie es of both sets, apart from BRRI dhan52. Recovery rate a er de-submergence (plant survival) varied from 10 to 90%. The third transplan ng was done on 16 September with 47 d old seedlings, a er the water depth had receded to about 20 cm, and while salinity was rela vely low (~2.5 dS/m) as a result of the rainfall. Salinity gradually increased to almost 6 dS/m by the end of October and to 6 dS/m by mid-November. 338 8 Sets 1&2 transplan ng 6 Set 3 transplan ng 4 2 0 1-Sep-12 1-Oct-12 31-Oct-12 Fig. 7. Water salinity in the gher in 2012. BRRI dhan41 and 44 did not survive due to submergence for 5 d immediately a er transplan ng. While BRRI dhan41 is moderately salinity tolerant it is sensi ve to submergence. BRRI dhan44 is suitable for non-saline dal water stagna on (diurnal fluctua ons in water level), but is not tolerant to submergence. BRRI dhan52 produced the highest yield at 4.0 t/ha, followed by BRRI dhan47 at 3.4 t/ha (Fig. 8). Yields of all other varie es were between 2.3 and 2.8 t/ha. The good performance of BRRI dhan52 reflects the fact that it combines salinity tolerance with rela vely tall height, conferring greater tolerance to water stagna on. 5 lsd 0.2 4 3 2 1 0 Fig. 8. Aman variety yields in the gher in 2012 (Bd = BRRI dhan; BINA = BINA dhan; BR8371-4R-2). 3.3 2013 In 2013, salinity fluctuated between 3 and 4 dS/m throughout the season (Fig. 9). Water depth was quite high at the me of the first transplan ng of the 49 d-old seedlings on 19 August, and rain shortly a er transplan ng further increased water depth. As a result, the first transplan ng failed. The second transplan ng of 56 d-old seedlings was done on 9 September a er water depth had declined to about 25 cm, but this was s ll too deep 339 and the seedlings were damaged and ul mately died due to the combina on of high water depth (which increased to almost 30 cm about four weeks a er transplan ng as a result of rainfall) and salinity. A third set of varie es was transplanted 5 d a er the second set (14 September using 46 d-old seedlings). Seven of these varie es survived due to their tall stature combined with salinity tolerance and these varie es went on to produce 2.6-3.7 t/ha of grain (Fig. 10). Both the Saltol varie es (BR11-Saltol, BRRI dhan28-Saltol) failed due to their short stature. The best varie es, with yields of 3.6-3.7 t/ha, were BR23, BRRI dhan41, and BR8371-4R-2. 4 40 3 30 Set 1 transplanted Depth EC 20 Set 2 transplanted Set 3 transplanted 2 1 10 0 0 5-Jul-13 4-Aug-13 3-Sep-13 3-Oct-13 2-Nov-13 2-Dec-13 Fig. 9. Water depth and salinity (EC) in the gher in 2013. 5 lsd 0.1 4 3 2 1 0 Fig. 10. Aman variety yields in the gher in 2013 (Bd = BRRI dhan; BINA = BINA dhan; BR8371-4R-2). 4. General discussion The water depth and salinity data demonstrate the considerable challenges to aman produc on in shrimp ghers in the coastal zone of Bangladesh in terms of water depth and salinity. Both water depth and salinity are highly variable depending on drainage management by the farmers at the end of the shrimp season and the incidence 340 and amount of rainfall. As a result, over the three years, there were many constraints to aman produc on including: (i) water too deep for transplan ng at the desired me (all years for almost all target transplan ng dates), (ii) the need to use older seedlings (beyond the op mal age for best crop performance, in all three years) because water depth was too high, (iii) submergence a er transplan ng (at least one set each year), (iv) high salinity at the me of transplan ng (2011 in poorly leached gher, 2012), (v) high salinity during the second half of the season (2012), and (vi) a combina on of moderately high salinity and high water depth (2011 in poorly leached gher, 2012, 2013). The most consistent performers under these difficult condi ons were the varie es with salinity tolerance (BRRI dhan47, 53, 54) or submergence tolerance (BRRI dhan52). However, even these varie es failed when water depth and/or salinity were too high. With current water management, varie es with mul ple tolerances (salinity, submergence, water stagna on) would probably be beneficial. The findings highlight the need for improved drainage management for stability of aman yield in shrimp ghers. The results showed that with good water management, yields of 4 t/ha can be achieved with modern, high yielding varie es. Good water management requires mely drainage (and thus shrimp harvest) to facilitate leaching of salt from the topsoil by the early monsoon rains, and drainage throughout the season as needed to maintain a suitable water depth for the rice plants, with shallower depth a er transplan ng than later on. Implementa on of drainage throughout the rainy season is biophysically feasible as the rivers are dal, with low de levels below land level in most loca ons (Khan et al. 2015). Furthermore, the ghers are connected to the rivers through a network of shallow canals as this is the means for bringing brackish water in from the rivers for shrimp produc on during the dry season. At present, the canals are primarily designed for water intake. However, slight deepening would facilitate drainage and leaching a er shrimp harvest. Implementa on of good drainage management would require a coordinated community approach and synchroniza on of cropping systems. For example, it would be important that all farmers in the same canal network harvest their shrimp around the same me; otherwise there would be conflic ng needs, with some farmers wan ng to bring in more river water to extend the shrimp phase and others wan ng to drain to commence the leaching of salt in prepara on for the rice season. 5. Conclusions With current water management, produc on of aman rice in ghers used for brackish water shrimp produc on is risky due to salinity, submergence and water stagna on. Transplan ng of seedlings much older than the op mum age (three to four weeks) for high yield under favorable condi ons reduces the risk slightly, but also reduces yield poten al. With modern high yielding varie es with salinity or submergence tolerance the risk is reduced somewhat and yields of 2.5 to 4 t/ha are feasible (although replan ng may be required). Improving drainage management is the key to improving yield and yield stability of aman crops in shrimp ghers, to enable be er leaching of salt prior to rice establishment, and to maintain a suitable water depth for the rice crop throughout the season. Acknowledgements We thank Mustafa Kamal, A qur Rahman, Amal K. Roy and Masum Rana for their assistance during experiments. This paper presents findings from ‘G2 Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food. References Barmon, B.K., Kondo, T., Osanami, F. 2004. Labor Demand for Rice Prawn Gher Farming in Bangladesh: A Case Study of Khulna District. The Review of Agricultural Economics, 60: 273-287. BRRI. 2013. Modern Rice Cul va on, 17th Edi on. 80p Khan, Z.H. and Kamal, F.A. 2015. The Ganges coastal zone – current condi ons, future projec ons. These proceedings. 341 Op mum sowing window for boro cul va on in the coastal zone of Bangladesh M.K. Mondal1, N.K. Saha1,2, S.P. Ritu3,4, P.L.C. Paul3, A.K.M. Sharifullah5, E. Humphreys1, T.P. Tuong1, and M.A. Rashid3 Interna onal Rice Research Ins tute, Bangladesh and Philippines m.mondal@irri.org, n.saha@irri.org, e.humphreys@irri.org, t.tuong@irri.org 2 Patuakhali University of Science and Technology, Bangladesh 3 Bangladesh Rice Research Ins tute, Bangladesh sanjidap05@gmail.com, plcpauliwm@yahoo.com, arashidiwm@yahoo.com 4 Current address: Sylhet Agricultural University, Bangladesh 5 Bangladesh Academy of Rural Development, Bangladesh sharifullahak@yahoo.com Abstract Rice produc on in Bangladesh has increased greatly as a result of the expansion of boro cul va on – irrigated high yielding varie es of rice grown during the dry/winter season. However, the coastal zone has missed out on this development. In medium salinity regions such as Khulna District, boro cul va on is hindered by scarcity of fresh water, as the river water is too saline for irriga on from mid-February to mid-June. In other parts of the coastal zone, such as much of Barisal Division, the cul va on of boro rice has been limited by the mispercep on that the river water is too saline for irriga on in the dry season throughout the coastal zone. Therefore, a series of experiments and demonstra ons were conducted to evaluate the feasibility of boro cul va on in low and medium salinity regions of the coastal zone. It was hypothesized that early boro establishment would reduce the requirements for water stored in irriga on canals (stored water) to finish off the crop in regions where river water salinity becomes too high, and that it would enable intensifica on to three rice crops per year in regions with year-round availability of fresh water. On the other hand, it was hypothesized that early sowing would predispose the crop to cold damage and result in reduced yields. Replicated experiments and on-farm demonstra ons were conducted at medium salinity loca ons from 2005 to 2007 and 2011 to 2014 in Khulna District, and at low salinity loca ons from 2011 to 2014 in Patuakhali and Barguna Districts in Barisal Division. Rice variety BRRI dhan28 was sown in seedling nurseries on a range of dates from late October to late December and transplanted when the seedlings had three to four leaves. At both loca ons in Barisal, irriga on water salinity remained below 1 dS/m throughout the season. Under these condi ons, and with effec ve disease control, yields were highest with sowing from mid-November to mid-December, ranging from 6.0 to 7.5 t/ha. In the more saline environment of Khulna, early sowing (22 October, 1 November) reduced the stored water requirement to finish off the crop (by about 50% for 22 October sowing), but yield was greatly reduced due to cold damage during the reproduc ve stage. The op mum sowing window for maximum yield was from 7 to 15 November, much narrower than the window in Barisal, and with lower maximum yield (around 5 t/ha in the absence of disease) than in Barisal. The lower yield at Khulna and the decline in yield for sowings a er mid-November were probably due to the higher salinity at Khulna. The results demonstrate substan al opportuni es for increasing rice produc on in southern Bangladesh by introducing boro cul va on in the south-central coastal zone where the river water remains fresh throughout the year. They also demonstrate the feasibility of achieving yields of 5 t/ha in medium salinity areas by sowing in the second week of November and storing river water in the khals inside the polders before rivers become too saline. Key message: It is possible to achieve high boro yields (up to 7.5 t/ha) in the coastal zone districts of Patuakhali and Barguna (and in similar environments) by sowing from mid-November to mid-December and irriga ng using the river water, which remains fresh (<1 dS/m) throughout the year. Keywords: salinity, temperature, cold, Khulna, Barisal, BRRI dhan28 342 1. Introduc on Bangladesh’s economy is dominated by agricultural output; 73% of the popula on and 48% of the labor force is engaged in agriculture (BBS 2013). Although Bangladesh has achieved self-sufficiency in rice the country faces enormous challenges to maintain food self-sufficiency for its growing popula on, as there is li le scope to further increase cropping system intensity except in the underu lized coastal zone lands. The coastal zone, covering approximately 2.8 Mha (30% of the cul vable land of the country), is the least produc ve region of Bangladesh but has great poten al for increased produc vity (Tuong et al. 2014). The Government of Bangladesh constructed a series of polders (about 1.2 Mha) during the 1960s and 1970s to increase agricultural produc on of the coastal zone, as well as to protect life and property from dal surges and to prevent saline water intrusion during the dry season. Poldering involved building large embankments around the perimeters of the islands formed between the spaghe of rivers in the delta. However, despite huge investment, the polders are home to some of the world’s poorest and most vulnerable people (BBS 2010; Kabir et al. 2014). While much of Bangladesh enjoyed the benefits of greatly increased rice produc on due to the expansion of irrigated dry season rice (boro), the coastal zone missed out. The expansion of boro produc on was the result of the development of high yielding rice varie es and of groundwater irriga on, and the use of fer lizers. In the coastal zone the scope for groundwater development is limited because the upper aquifer is saline and there are concerns about saliniza on of the deep fresh aquifer (Hasan et al. 2013; Dhaka Tribune 2013). However, there is scope for using surface water resources for boro produc on during the dry season. In moderately saline areas of the coastal zone in Khulna District the river water remains suitable for irriga on for the first couple of months of the dry season, un l early to mid-February (Fig. 1) (Khan et al., these proceedings). Mondal et al. (2006, 2010) showed that with ‘early’ ( mely) establishment and proper management a boro crop could be grown a er harvest of the aman crop (wet season rice) in this region. This requires replacement of the tradi onal, photoperiod sensi ve and late harvested aman landraces with a modern, early maturity aman variety, enabling early establishment of the boro crop. The produc on of boro in this situa on involves two dis nct stages of irriga on. The first stage is gravity irriga on while the river water salinity is <4 dS/m (up to mid-February) by le ng water into the polder canal networks at high de through the sluice gates. The second stage involves storage of river water in the canal networks before the river salinity reaches 4 dS/m and pumping water from the canals to finish off the boro crop. The extent to which this technology could be adopted in moderately saline regions of the coastal zone depends on several factors: (i) the water requirement of the boro crop, (ii) the me when the river water becomes too saline for irriga on, and (iii) the storage capacity of the canal networks inside the polder. Mondal et al. (2010) showed that the boro area could be expanded by increasing the storage capacity of the canals and/or by advancing the cropping period. The storage capacity can be increased by digging new canals/ponds and desil ng exis ng ones, but this would require considerable investment. Advancing boro establishment reduced the amount of stored water required to finish the crop, and also the cost of pumping, but advancing it too early exposed the crop to low temperature stress during the reproduc ve stage with large yield loss primarily due to high sterility. While the rivers become saline during the dry season in much of Khulna Division, Khan et al. (these proceedings) showed most of the rivers in Barisal Division remain non-saline (<1 dS/m) throughout the year. Thus, there is plenty of fresh water available for boro produc on in this region and a dense network of khals (small river canals) that can be used to distribute the water across the landscape. In addi on to increasing rice produc on through expanding the boro area, earlier plan ng of boro could also create the opportunity for triple rice cropping (Tuong et al. 2014; Saha et al., these proceedings; Bha acharya et al., these proceedings; Paul et al., these proceedings). Again the ques on is what is the op mum plan ng window for boro, bearing in mind both yield target and irriga on water requirement. 343 24 20 16 12 8 4 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Fig. 1. Long-term monthly average salinity (1990-2008) of Kazibacha River at Ba aghata, Khulna, Bangladesh (ver cal capped bars indicate standard error of 19 monthly values). Given the considerable poten al for expanding boro produc on in the fresh and moderately saline regions of the coastal zone a series of experiments was undertaken over the past ten years in low and medium salinity regions of the coastal zone with three key objec ves: To determine the effect of sowing date on the yield of boro To determine the irriga on water requirement (and the stored water requirement for irriga on in a medium salinity area) as affected by sowing date To determine the op mum sowing date taking into account tradeoffs between yield, irriga on water requirement and stored water requirement 2. Methodology 2.1 Study sites Two replicated experiments and two non-replicated demonstra ons were conducted at three coastal zone loca ons with low or medium salinity (Table 1). Replicated experiments were conducted at Kismat Fultola village (medium salinity) in Khulna Division during the 2005-06 and 2006-07 dry seasons, and at Patuakhali Science and Technology University (PSTU) farm (low salinity) in Barisal Division during 2012-13 and 2013-14. In addi on, non-replicated demonstra ons were conducted in farmers’ fields at Kismat Fultola and Bazarkhali during 2011-12, 2012-13 and 2013-14. Bazarkhali is also in Barisal Division, and located about 32 km south of PSTU. BRRI dhan28 (a high yielding medium dura on (140 d) boro variety) was grown at all loca ons in all years. Details of the replicated experiments are provided in Mondal et al. (2010) and Saha et al. (these proceedings), and details of the demonstra ons in Paul et al. (these proceedings). Only informa on per nent to the analysis in this paper is briefly summarized here. 2.2 Treatments The experiments and demonstra ons were all designed to compare the effect of sowing date on the performance of boro crops (Table 1). Sowing dates at Kismat Fultola ranged from late October to mid-December, while sowing dates at PSTU and Bazarkhali were from mid-November to the end of December. 344 2.3 Crop management All experimental fields were ploughed three to four mes (wet llage) using a ller powered by a two-wheel tractor. Final land leveling was done with a wooden plank drawn by a dra animal. Small bunds (20 cm x 30 cm) were constructed around all fields, and around all the plots in the replicated experiment at Kismat Fultola, and were compacted to minimize seepage between adjacent plots. Seedlings were raised in nurseries in a separate loca on. Rice seeds were soaked for 12 h and incubated for 48-72 h. The pre-germinated seed was sown on the seed beds at 28 g/m2. The seedlings were carefully uprooted and transplanted when they had three to four leaves, with two to four seedlings per hill at a spacing of 20 cm x 20 cm. The age of the seedlings at the me of transplan ng varied from 21 to 35 d (usually it was 21-27 d) depending on temperature at the me of seedling growth and thus the me taken for leaves to develop. Fer lizers were applied based on Bangladesh Rice Research Ins tute (BRRI 2005) recommenda ons for boro cul va on. In both replicated experiments, urea was applied at 120 kg N/ha in four equal splits: before the last leveling, 25 d a er transplan ng (DAT), five to seven d before panicle ini a on and at heading. In the non-replicated experiments, urea was topdressed in three equal splits on 15, 30 and 55 DAT. Other fer lizers were broadcast at 60 kg P2O5, 40 kg K2O, 60 kg CaSO4 and 10 kg ZnSO4 per ha before the last land leveling. Pes cides were applied as per BRRI (2005) recommenda ons and a er consulta on with the local experts of the Department of Agricultural Extension. In addi on, in the replicated experiments, systemic pes cides were applied at the me of each N applica on to minimize crop damage. Manual weeding was done just prior to urea topdressing. At PSTU, post-emergence weedicide (pyrazosulfuron-ethyl @125 g/ha) was also applied three to five DAT into standing water and the water was held on the plots for two to three d a er treatment to control weeds. All experimental sites were surrounded by plas c barriers with traps to prevent damage by rats, and covered by ne ng as the crops approached maturity to prevent bird damage. Protec on from rats and birds was necessary as there were no other crops surrounding the experiments in the dry season. 345 Table 1. Details of the boro sowing date, experimental sites and treatments Loca on Kismat Fultola Bazarkhali Upazila District La tude Longitude Height above sea level Salinity classifica on River water salinity (EC, dS/m)) Topsoil (0-15 cm) salinity (satura on extract EC, dS/m) Soil texture Predominant cropping systems 2005-07 experiments Ba aghata Khulna 22° 41' 00''N 89° 30' 00''E Amtali Barguna 22°11’33”N 90°15’41”E Patuakhali Science and Technology University Dumki Patuakhali 22°27'51"N 90°22'56"E 2-3 m Medium 2-3 m Low 2-3 m Low 0.2-25 <1 dS/m <1 dS/m 2-16 Silty clay loam Sesame-Aman-Fallow Expt 1 2005-06, 2006-07 4 sowing dates 22 October 1 November 7 November 15 November 2-8 Clay loam Fallow-Aman-Grasspea; Aus-Aman-Fallow 2-8 Clay loam Fallow-Aman-Grasspea; Aus-Aman-Fallow 4 replicates Sub-plots 6 m x 8 m Mondal et al. (2010) 2011-14 experiments and demonstra ons Demo 1 2011-12, 2012-13, 2013-14 Demo 2 Expt 2 2011-12, 2012-13, 2013-4 2012-13, 2013-14 3 sowing dates 10 November 30 November 20 December 3 sowing dates 15 November 7 December 30 December Non-replicated Plots 10 m x 20 m Non-replicated Plots 10 m x 20 m Paul et al. (these proceedings) 346 5 sowing dates 15 November 20 November 30 November 05 December 15 December 4 replicates Plots 5 m x 5 m Saha et al. (these proceedings) 2.4 Irriga on water monitoring All crops were grown under fully irrigated condi ons. Irriga on water was pumped from the nearby khal at each loca on using a 4 h.p. centrifugal pump powered by diesel. The crops were irrigated whenever surface water depth fell below 1 cm, and water was added un l water depth reached 5 cm. Irriga on was stopped 10 to 15 d before harvest. The salinity of the irriga on water was monitored regularly using a portable EC meter. The amount of irriga on water applied to each plot in the replicated experiments at Kismat Fultola was determined using a 90o V-Notch weir to determine flow rate and the me to irrigate each plot. 2.5 Crop monitoring Grain yield and yield components were determined in all experiments and demonstra ons. In addi on, the dates of the main crop development stages were determined in the replicated experiments at both loca ons. 2.5.1 Phenology The dates of panicle ini a on (PI), flowering (FL), and physiological maturity (PM) were determined in each plot in the replicated experiments at Kismat Fultola and at PSTU. Daily monitoring of six plants was done for 10 d, star ng 5 d before the expected onset of PI and FL. For determina on of PI, the main ller of each plant was dissected to determine the date when the panicle was visible to the naked eye (i.e. about 1 mm long). PI was determined as the date when the panicle was visible in 50% of the main llers examined. Flowering date was determined as the date when 50% of the llers of 12 randomly selected hills had commenced anthesis. Physiological maturity (PM) was taken as the date when 80% of the grains had turned golden in all experiments and demonstra ons. 2.5.2 Grain yield and yield components In the replicated experiments, grain yield was determined from 4 m x 2 m (Kismat Fultola) and 3 m x 2 m (PSTU) areas in the middle of each plot. Grain yield in the non-replicated trials at Bazarkhali and Kismat Fultola was determined from 5 m2 areas harvested at five loca ons (four towards each of the four corners and one in the middle of each plot), except in 2013-14 at Kismat Fultola when an area of 5 m x 2 m was harvested in the middle of each plot. The samples were threshed, cleaned and dried, and moisture content was determined to enable calcula on of grain yield at 14% moisture content. Yield components (number of panicles/m2, number of florets/panicle, % floret fer lity, 1000 grain weight) were determined at PM from 16 hills, 4 hills at each corner of the harvest area at Kismat Fultola and at PSTU; and from 16 randomly selected hills in 8 loca ons (2 hills per loca on) at Bazarkhali. 2.6 Weather data A standard rain gauge was installed in the vicinity of each experimental site. Daily maximum and minimum temperature data were obtained from the Bangladesh Meteorological Department (BMD) for the weather sta ons at Khepupara in Patuakhali District, and at Khulna city. The Khepupara sta on is about 24 km and 56 km south of the experimental sites at Bazarkhali and PSTU, respec vely; the Khulna sta on is about 8 km north of Kismat Fultola. Long-term (1998 to 2007) daily maximum and minimum temperatures for Patuakhali (NASA 1 degree grid, h p://power.larc.nasa.gov/cgi-bin/cgiwrap/solar/agro.cgi) and Khulna (BMD) were used to calculate 10-d long-term means for the period from November to April. 2.7 Data analysis Crop and irriga on data from the replicated experiments were subjected to analysis of variance (ANOVA). The least significant difference (LSD) was determined at 5% level of probability. 347 3. Results 3.1 Barisal 3.1.1 Irriga on water salinity Salinity of the water in the khals remained below 1 dS/m throughout the boro season (and the en re year) at both Bazarkhali and PSTU (Figs. 2a, b). (a) 1.0 0.8 0.6 0.4 0.2 0.0 (b) 1.0 0.8 0.6 0.4 0.2 0.0 Fig. 2. Irriga on water salinity (water salinity in the khals) at Bazarkhali (a) and PSTU (b) from 2012 to 2014. 3.1.2 Temperature Temperature during the 2011-14 boro seasons was generally similar to the long term mean, with a few notable excep ons. Minimum temperature was unusually high from 1 to 10 January and 1 to 10 March 2012, and both minimum and maximum temperatures were unusually low from 21 to 31 December 2012 (Fig. 3). This was followed by low minimum temperatures throughout January 2013. Temperature from early January to early March 2014 was also generally lower than the long-term average, more so in mid- to late February and early March. 348 40 35 Max 30 25 20 15 Min Long term 2011-12 2012-13 2013-14 10 5 0 Fig. 3. Maximum and minimum temperature at Patuakhali during the 2011-14 boro seasons in comparison with the long-term (1998-2007) mean. Mean values for 10 to 11 d periods. 3.1.3 Crop dura on There was a consistent trend for crop dura on to become shorter by about 10 d as sowing was delayed from 15 November to 30 December at Bazarkhali (Table 2) and PSTU (Table 3). The more detailed phenological data at PSTU show that the decrease in dura on was due to earlier flowering as sowing date was delayed. The effect of sowing date on me to panicle ini a on (PI) and on the dura on of grain filling was inconsistent and the varia on was small. The earlier flowering with delay in sowing was thus primarily due to reduc on in the period from PI to anthesis, due to warmer weather during this period. Table 2. Dura on (days from sowing to physiological maturity) of boro crops as affected by sowing date at Bazarkhali in Barisal Division Year 2011-12 2012-13 2013-14 15 Nov 148 152 149 Sowing date 7 Dec 138 144 139 30 Dec 123 126 126 349 Table 3. Phenological stages (days a er sowing, DAS) of boro crops as affected by sowing date at PSTU in Barisal Division Year 15 Nov Panicle ini a on (PI) 2012-13 91 2013-14 82 Anthesis 2012-13 117 2013-14 116 Physiological maturity 2012-13 140 2013-14 140 Dura on PI-Anthesis 2012-13 26 2013-14 34 Dura on of grain filling 2012-13 23 2013-14 24 20 Nov Sowing date 30 Nov 5 Dec 15 Dec 89 86 86 83 91 82 86 84 115 116 112 112 112 110 106 105 139 139 137 133 136 131 130 128 26 30 26 29 21 28 20 21 24 23 25 21 24 21 24 23 3.1.4 Grain yield Grain yield of the boro crops in Barisal ranged from 4.4 to 7.9 t/ha across loca ons, sowing dates and years (Table 4). The effect of sowing date on yield was inconsistent across years and across loca ons within year. At Bazarkhali, grain yield declined with delay in sowing from mid-November to late December in 2011-12. The very low yield of the 30 December sowing in that year may have been partly due to leaf blight a ack in the seed bed; growth and llering a er transplan ng were poor. In 2012-13, yields of the mid-November and early December sowings were similar, but again decreased with the late December sowing. In 2013-14, the first sown (15 Nov) crop was severely affected by stem borer (es mated as 20% yield loss) and maximum yield occurred with early December sowing. At PSTU, yield increased with delay in sowing from mid-November to early December in both years. Yields at PSTU in 2012-13 were much lower than at Bazarkhali in the same year, probably due to heavy infesta on with brown spot disease at PSTU, with the first two sowings being the most severely affected. This was associated with a large reduc on in panicle density, more so for earlier sowings (data not presented). In 2013-14, yield also increased with delay in sowing to 7 December, and yields at PSTU were higher than at Bazarkhali, where there was stem borer infesta on, more so in the first sowing. 350 Table 4. Grain yield (t/ha, 14% moisture) of boro crops as affected by sowing date at Bazarkhali and PSTU in Barisal Division Year Sowing date 30 Nov 5/7 Dec1 15 Nov 20 Nov 15 Dec 30 Dec Bazarkhali (mean ± standard error) 2011-12 7.9 ±0.1 6.7 ±0.2 4.4±0.2 2012-13 6.7±0.2 7.0±0.2 5.9±0.2 2013-14 5.8±0.2 6.6±0.3 6.1±0.2 PSTU 2012-13 4.9 5.0 5.6 5.9 6.1 2013-14 6.2 7.0 7.3 7.4 7.1 Significant interac on between year and sowing date, lsd for the interac on = 0.3 1 5 Dec sowing at PSTU, 7 Dec sowing at Bazarkhali 3.2 Kismat Fultola 3.2.1 Irriga on water salinity From 2005 to 2007 the boro crops at Kismat Fultola were irrigated from river water un l the river became too saline (early February, Fig. 4a). Therea er, they were irrigated using water stored in the adjacent khal that was filled in the first week of February just prior to the river becoming too saline, a er which the sluice gate was closed. The results from 2006-07 show that the salinity of the irriga on water gradually increased as the season progressed reaching 4.5 dS/m at the me of maturity of the last sown crop (Fig. 4b) (salinity was not monitored in 2005-06, but it can be assumed that it was similar to that in 2006-07; there were no signs of salinity damage in any crop at any stage). Salinity of the satura on extract also gradually increased, with values similar to that of the irriga on water (data not presented). In 2011-12, the salinity of the water in the khal remained below 4 dS/m throughout the boro season. However, in 2012-13 and 2013-14 the community water management group decided to leave the sluice gate partly open during the dry season to try and prevent silta on in the gate intake canal. As a result the salinity of the irriga on water increased as the dry season progressed to about 15 dS/m around the me of maturity of the last sown crop in 2012-13. Salinity data were not collected during 2013-14, however, there was clear evidence of salinity damage, including seedling death in early February in the most recently (late) transplanted crop. (a) 40 36 32 28 24 20 16 12 8 4 0 1-Nov Boro 2005-06 Boro 2006-07 1-Dec 1-Jan 1-Feb 1-Mar 1-Apr 1-May 351 (b) 5 4 From Canal From River 3 2 1 10 February 0 2006 (c) Date 2007 25 20 15 10 5 0 Fig. 4. Salinity of the river water at Kismat Fultola in 2005-06 and 2006-07 (a), and salinity of the irriga on water in 2006-07 (b) and 2012-13 (c). 3.2.2 Temperature Maximum temperature at Khulna was considerably lower in 2006-07 than in 2005-06 from mid-December to mid-February, and also lower than the long-term average (Fig. 5a). Maximum temperature in February 2006 was higher than the long-term mean. Mean minimum temperature during the first 10 d of January and 1-10 March 2012 was much higher than the long-term average (Fig. 5b), as also observed at Patukhali (Fig. 3). Also as at Patuakhali, 2012-13 was colder than normal in late December and January. Minimum and maximum temperatures during the last part of December at Khulna were much lower than the long-term average and minimum temperatures during January also remained lower than normal (Fig. 5b). 352 (a) 40 35 Max 30 25 20 15 Min 10 5 Long term 2005-06 2006-07 0 (b) 40 35 Max 30 25 20 15 Long term 2011-12 10 5 0 Min 2012-13 2013-14 Fig. 5. Maximum and minimum temperature at Khulna during the 2005-07 (a) and 2011-14 (b) boro seasons in comparison with the long term (1998-2007) mean. Mean values for 10-11 d periods. 3.2.3 Crop phenology The me taken to reach PI increased with delay in sowing from around 75 d with sowing on 22 October to around 90 d with sowing on 7 and 15 November in 2005-06 and 2006-07 (Table 5). This was due to lower temperatures during the vegeta ve phase as sowing was delayed (Fig. 5a). The dura on of PI to anthesis (~41 d) was much longer (by 7 to 14 d) with 22 October sowing than with later sowing dates as PI occurred in early January, at the start of the period when temperature was lowest (during the first three weeks of January). The net result was that the effects of me of sowing on the date of anthesis were small and inconsistent. The dura on of grain filling was also about 10 d longer with 22 October sowing than all later sowings due to the produc on of numerous late llers. In 2006-07 it was also probably partly due to 10 d earlier anthesis and thus cooler weather during early grain filling. The net result was that the dura on of 22 October sowings was longer than that of 15 November sowings each year (by 7 to 10 d), with the dura on of the 1 and 7 November sowings somewhere in between, but with no consistent trends. 353 The me to maturity of the crop for the 10 November sowing in 2011-12 and 2013-14 was similar to that of the 15 November sowings in 2005-07 (around 145 d) (Table 6). However, in 2012-13 the dura on was about 20 d longer, presumably due to unusually cold weather in late December 2012 and low minimum temperature in January 2013 (Fig. 5b). In 2012-13, seedlings from the 30 November sowing died a er transplan ng due to cold damage as transplan ng took place in early January. In 2013-14, seedlings from both the 30 November and 20 December sowings died a er transplan ng due to salinity, possibly exacerbated by low temperature in the case of the 30 November sowing. Table 5. Phenological stages of boro crops as affected by sowing date at Kismat Fultola from 2005-07 (adapted from Mondal et al. 2010) Year Panicle ini a on(PI) 2005-06 2006-07 Anthesis 2005-06 2006-07 Physiological maturity 2005-06 2006-07 Dura on PI-Anthesis 2005-06 2006-07 Dura on of grain filling 2005-06 2006-07 Sowing date 22 Oct 1 Nov 7 Nov 15 Nov 74 77 84 94 93 93 93 90 116 117 112 127 121 125 121 118 155 154 140 154 149 152 147 145 42 40 28 33 28 32 28 28 39 37 28 27 28 27 26 27 Table 6. Dura on of phenological stages of boro crops as affected by sowing date at Kismat Fultola from 2011-14 Year Panicle ini a on 2011-12 2012-13 2013-14 Anthesis 2011-12 2012-13 2013-14 Physiological maturity 2011-12 2012-13 2013-14 354 10 Nov Sowing date 30 Nov 20 Dec 95 97 92 89 Crop died 87 79 Crop died Crop died 122 131 129 120 Crop died 116 112 Crop died Crop died 143 166 150 143 Crop died 144 135 Crop died Crop died 3.2.4 Grain yield Grain yield increased with delay in sowing to 7 November in both 2005-06 and 2006-07, and declined with further delay to 15 November in 2006-07 only (Table 7). The low yields of the first sowing in 2005-06 and the first two sowings in 2006-07 were due to high sterility (more so for the 22 October sowing) and fewer florets per panicle, as a result of low temperature during the reproduc ve phase (Mondal et al. 2010). Yields of the 7 and 15 November 2005-06 sowings were reduced due to stem borer a ack, and predicted yield of these two sowings was 5-5.6 t/ha in the absence of stem borer a ack (adjusted yield was calculated using the propor on of white heads [panicles with no filled grains] to total number of panicles within the harvest area). In the 2011-14 boro crops, delaying sowing beyond 10 November greatly reduced grain yield in two out of three years (Table 8). Table 7. Grain yield, water input and stored-water (from canal networks) produc vity of boro crops during 2005-07 as affected by sowing data at Kismat Fultola, Khulna Division (adapted from Mondal et al. 2010) Sowing date Grain yield1 (t/ha) Adjusted grain yield2 (t/ha) 2.7 4.6 3.6 4.3 1.0 2.7 4.9 5.0 5.6 1.2 2005-06 22 Oct 01 Nov 07 Nov 15 Nov LSD (p=0.05) 2006-07 22 Oct 01 Nov 07 Nov 15 Nov LSD (p=0.05) 0.7 2.4 5.0 4.5 0.8 Irriga on amount (mm) Land prepara on water4 (mm) River Canal Total Rainfall (mm) Total water input (mm) Canal water produc vity (kg grain/m3 canal water) 349 269 238 179 91.4 147 202 220 269 26 496 471 458 448 NS3 31 80 130 172 16 16 16 16 543 567 604 636 1.9 2.3 1.7 1.6 0.4 351 323 266 215 54 143 223 268 278 37 494 546 534 493 NS 31 80 130 172 96 96 96 96 621 722 760 761 0.5 1.1 1.9 1.6 0.4 1 Actual yield 2 Adjusted to remove the effect of stem borer damage (Mondal et al. 2010) 3 NS not significant 4 Land prepara on water was same in both years Table 8. Grain yield of boro crops during 2011-14 as affected by sowing data at Kismat Fultola (mean ± s.e.) Year 2011-12 2012-13 2013-14 10 Nov 4.6±0.3 4.9±0.2 4.6 Sowing date 30 Nov 5.9±0.2 0.0 1.7 20 Dec 6.3±0.6 0.0 0.0 355 3.2.5 Irriga on input and water produc vity At Kismat Fultola, irriga on input from the river decreased with delay in sowing from 22 October to 15 November in both years while input from water stored in the canal increased with delay in sowing (Table 7). The net result was that there was no significant effect of me of sowing between 22 October and 15 November on total irriga on input. 4. Discussion Combining the yield data for both experiments (five different years x sowing dates from 22 Oct to 30 Dec) at Kismat Fultola shows that maximum average yield occurred when sowing was during the period 7 to 14 November (Fig. 6a). In the absence of stem borer damage, this trend would have been even more pronounced. Furthermore, yield variability was least for these sowing dates, sugges ng that sowing in the second week of November is op mum for higher yields. The simula ons of Sharifullah et al. (2009) also showed that yield increased with delay in sowing to 10 November, and with no further yield increase beyond that to 20 November (20 November was the latest sowing date tested). The simula ons also showed that the requirement of stored water for irriga on increased steadily as sowing is delayed from 15 October to 20 November, consistent with the findings of the field experiments. In the field experiments, produc vity of the stored water (kg grain/m3) was highest and similar for 1 to 15 November sowings in 2005-06, and for 7 and 15 November sowings in 2006-07 (Table 7). Thus there were tradeoffs between yield, stored water irriga on requirement and produc vity of the stored water, for sowings between 15 October and 20 November. Maximum yield and stored water produc vity were observed for sowings in the second week of November, while stored water requirement was least for the earliest sowings (15 and 22 October in the modeling and field studies, respec vely). The results from the two loca ons in Barisal Division showed li le effect of sowing date from mid-November to mid-December on both grain yield and variability of grain yield (Fig. 6b). During this period, yields were always above 6 t/ha, except for the first two sowings at PSTU, which were affected by stem borer and brown spot. Yields at Kismat Fultola (Khulna District) were always much lower than yields at both sites at Barisal for similar sowing dates. The 10 d mean minimum temperature at Khulna and Patuakhali was similar during each boro season from 2011-12 to 2013-14 (Fig. 7). However, maximum temperature during the coldest part of the year was a couple of degrees lower in Khulna in 2011-12 and 2013-14, but not in 2012-13, the coldest of the three years. The lower temperature at Khulna may partly explain the lower yields at Khulna in 2011-12 and 2013-14, but not in 2012-13. A more likely explana on is the higher salinity at Khulna. While the salinity of the irriga on water in 2005-07 and 2011-12 could be considered acceptable (<4 dS/m), it was much higher than at both sites in Barisal. The threshold for salinity injury in rice (yield and growth) is about 3 dS/m for Indicas and 1.9 dS/m for Japonicas. BRRI dhan65 is an Indica, sugges ng that yield at Khulna was constrained by salinity, even. Salinity stress at these thresholds does not necessarily manifest as visual symptoms, but does reduce yield and growth if salinity exceeds these limits (Ismail et al. 2007). Whether the combina on of marginally high salinity and marginally low temperature for rice would have an antagonis c effect on the plants is not known. The long term data show that both minimum and maximum temperature are lower at Khulna than Patuakhali from late December to mid-February, sugges ng that Khulna is a more risky environment for early planted boro rice than Barisal. 356 8 (a) 6 4 2 n=3 n=8 n=8 n=8 n=8 n=3 n=3 30 Nov 20 Dec 0 22 Oct 01 Nov 07 Nov 10 Nov 15 Nov 8 (b) Bazarkhali PSTU 6 4 n=8 n=3 n=8 2 n=8 n=8 n=3 n=3 n=8 0 15 Nov 20 Nov 30 Nov 05 Dec 15 Dec 30 Dec Sowing date Fig. 6. Effect of sowing date of BRRI dhan28 on grain yield in moderately saline (Kismat Fultola, Khulna district) (a) and low salinity (Dumki, Patuakhali district and Bazarkhali, Barguna district) (b) areas of the coastal zone of Bangladesh. Ver cal bars are standard error. Actual 2005-06 yield data for Kismat Fultola were used. 357 40 35 Max 2011-12 30 25 20 15 M Min 10 5 0 40 35 45 40 35 30 25 20 15 10 5 0 2013-14 Min 40 Max Ma 2012-13 35 30 30 25 25 20 20 15 Max Min 15 10 10 5 5 0 0 Max Min Long term K hulna P atuakha li Fig. 7. Comparison of 10 d mean maximum and minimum temperatures at Khulna and Patuakhali during the 2011-12, 2012-13, 2013-14 boro seasons, and the long-term mean (1998-2007). 5. Conclusions and recommenda ons The study provides strong evidence of the feasibility of high yielding boro cul va on in the south-central (Barisal Division) coastal zone of Bangladesh, where river water salinity remained <1 dS/m throughout the year. Yields of 6 to 7.5 t/ha were achieved with sowing from mid-November to mid-December in the absence of significant disease incidence. In the medium salinity site in Khulna District maximum yields of around 5 t/ha were achieved with sowing in the second week of November. Here, yields appeared to be constrained by the generally higher salinity at this loca on, even when salinity of the irriga on water remained below 4 dS/m throughout the season. Low temperature reduced floret fer lity and yield of earlier sown crops and increased vulnerability to disease (and possibly salinity). Therefore, the development of boro varie es with greater cold tolerance during the vegeta ve and reproduc ve stages is needed, especially considering the narrow sowing window and risk of lower temperatures in some years that occurred during the experiment period. At all sites, infesta on with stem borers was a problem in some years. Our boro crops were the only crops in the field and were surrounded by fallow land, and were thus more vulnerable to a ack by pests and diseases. In other parts of Bangladesh where boro crops are widespread, farmers are able to achieve good control of pests and diseases. 358 Acknowledgements This paper presents findings from PN10 ‘Managing Water and Land Resources for Sustainable Livelihoods at the Interface Between Fresh and Saline Water Environments in Vietnam and Bangladesh’ and G2 ‘Produc ve, profitable, and resilient agriculture and aquaculture systems’, projects of the CGIAR Challenge Program on Water and Food. The authors are grateful for the high quality technical assistance provided by Swapan Bhadra, Akbar Ali, Jahid Hasan, Elias Hossain, Amal Ray, Samir Sarkar, Shak Mondal, Lincoln Ray, Tanmoy Ray and Mithun Ray. References Bangladesh Bureau of Sta s cs (BBS) 2010. Census of agriculture 2008. Ministry of Planning, Dhaka, Bangladesh. Bangladesh Bureau of Sta s cs (BBS) 2013. Sta s cal year book of Bangladesh 2012. Ministry of Planning, Dhaka, Bangladesh. Bangladesh Rice Research Ins tute (BRRI). 2005. Adhunik dhaner chas (Modern rice cul va on - in Bangla). BRRI, Gazipur,Bangladesh. Bha acharya, J., Mondal, M.K, Humphreys, E., Saha, N.K., Rashid, M.H., Paul, P.C. and Ritu, S.P. 2014. Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh. These proceedings Dhaka Tribune 2013. Salinity in coastal aquifers alarming. h p://www.dhakatribune.com/bangladesh/2013/ oct/20/salinity-coastal-aquifers-alarming Hasan, M.R., Shamsuddin, M. and Hossain, A.F.M.A. 2013. Salinity status in groundwater: A study of selected upazilas of southwestern regions in Bangladesh. J. Global Sci.Technol. (1):112-122. Ismail, A.M., Heuer, S., Thomson, M.J. and Wissuwa, M. 2007. Gene c and genomic approaches to develop rice germplasm for problem soils. Plant Mol. Biol. 65:547–570 Kabir, K. A., Karim, M., Shahrier, M.B., Saha, S.B., Chakraborty, S., Jharendu, P. and Phillips, M. J. 2014. Homestead farming systems in southwest Bangladesh: A survey. Water Land and Ecosystems Technical Paper (in press). Khan, Z.H. and Kamal, F.A. The Ganges coastal zone – current condi ons, future projec ons. These proceedings. Mondal, M. K., Tuong, T.P., Sharifullah, A.K.M. and Sa ar, M.A. 2010. Water supply and demand for dry-season rice in the coastal polders of Bangladesh. In C. T. Hoanh, B. W. Szuster, K. Suan-Pheng, A. M. Ismail and A. D. Noble (Eds.) Tropical Deltas and Coastal Zones: Food Produc on, Communi es and Environment at the Land-Water Interface. Comprehensive Assessment of Water Management in Agriculture Series, no. 9. CABI Publishing pp. 264-278. Mondal, M. K., T. P. Tuong, S. P. Ritu, M. H. K. Choudhury, A. M. Chasi, P. K. Majumder, M. M. Islam and S. K. Adhikary 2006. Coastal water resource use for higher produc vity: Par cipatory research for increasing cropping intensity in Bangladesh. In C. T. Hoanh, T. P. Tuong, J. W. Gowing and B. Hardy (Eds). Environmental and livelihoods in tropical coastal zones: Managing agriculture-fishery-aquaculture conflicts. Comprehensive Assessment of Water Management in Agriculture Series, no. 2. CABI Publishing pp. 72-85. Mondal, M.K, Paul, P.L.C., Humphreys, E., Tuong, T.P., Ritu, S.P. and Rashid, M.A. 2014. Opportuni es for cropping system intensifica on in the coastal zone of Bangladesh. These proceedings 359 Saha, N.K., Mondal, M.K., Humphreys, E., Bha acharya, J., Rashid, M.H., Paul, P.L.C. and Ritu, S.P. 2014. Triple rice in a year: Is it a feasible op on for the low salinity areas of the coastal zone of Bangladesh? These proceedings Sharifullah, A., Tuong, T.P., Mondal, M.K. and D.T. Franco. 2009. Assessing water supply and demand for dry season rice in coastal polders of Bangladesh. In: Humphreys, E. and R.S. Bayot (eds). Increasing the produc vity and sustainability of rainfed cropping systems of poor smallholder farmers. Proceedings of the CGIAR Challenge Program on Water and Food Interna onal Workshop on Rainfed Cropping Systems, Tamale, Ghana, 22-25 September 2008. The CGIAR Challenge Program on Water and Food, Colombo, Sri Lanka. Available at: h p://r4d.dfid.gov.uk/PDF/Outputs/WaterfoodCP/CPWF_Proceedings_Rainfed_Workshop%5B1%5D.pdf (last accessed 31 December 2014) Tuong, T. P., Humphreys, E., Khan, Z.H., Nelson, A., Mondal, M.K., Buisson, M.C. and George, P. 2014. Messages from the Ganges basin development challenge: Unlocking the produc on poten al of the polders of the coastal zone of Bangladesh through water management investment and reform. CPWF Research for Development Series 9. Available at: h ps://cgspace.cgiar.org/bitstream/handle/10568/41708/CPWF%20Ganges%20basin% 20messages%20Sept%2014.pdf?sequence=5 (last accessed 30 Dec 2014). 360 An aus-aman system for increasing the produc vity of a moderately saline region of the coastal zone of Bangladesh S.P. Ritu1,2, M.K. Mondal3, T.P. Tuong3, S.U. Talukdar 2 and E. Humphreys3 Formerly: Bangladesh Rice Research Ins tute, Bangladesh, sanjidap05@gmail.com 2 Sylhet Agricultural University, Bangladesh, talukder_iwm@yahoo.com 3 Interna onal Rice Research Ins tute, Bangladesh and Philippines, m.mondal@irri.org, t.tuong@irri.org, e.humphreys@irri.org 1 Abstract Due to lack of readily available fresh water and salinity intrusion in the dry season, most agricultural land in the coastal zone of Bangladesh grows a single rice crop (aman) during the rainy season, mainly using local varie es with low yield poten al. This study aimed to develop a double rice cropping system with an aus crop grown at the onset of the rainy season followed by an aman crop. A field experiment (Experiment 1) was conducted in 2006-2008 in Ba aghata, Khulna District, to test the hypothesis that high and stable system produc vity can be achieved in moderately saline areas by: (i) dry seeding a short-dura on aus variety (BRRI dhan65) to advance crop establishment, (ii) supplemental irriga on during establishment, and (iii) transplan ng a high-yielding aman variety a er harvest of the aus crop. Experimental treatments in the aus season included three water regimes (rainfed, I1; supplemental irriga on, I2; full irriga on, I3) and three sowing dates (D1 = early April; D2 = mid-April; D3 = late April). In the aman season, two high yielding varie es BR11 and BRRI dhan53 were established by puddling and transplan ng about three weeks a er aus harvest. The findings of Experiment 1 led to a supplemental experiment (Experiment 2) in 2009, with the hypotheses that late transplan ng will secure good aus establishment and that high yield of late planted aman can be maintained using a suitable photoperiod insensi ve variety. The aus rice was established by dry seeding (M1) and transplan ng (M2) with two sowing dates - 30 April (D1) and 10 May (D2). A photoperiod sensi ve variety BRRI dhan46 and photoperiod insensi ve variety BRRI dhan49 were transplanted in the aman season. The effects of irriga on and sowing date treatments on yield of aus varied between years, due to varia on in rainfall distribu on. Yields with full and supplemental irriga on were comparable (4 t/ha) and were greater than those of fully rainfed aus (2 t/ha) when there were dry spells a er sowing. Supplemental irriga on required much less irriga on water (100-200 mm) than full irriga on (660-1042 mm) but had similar yield. The average incremental irriga on water produc vity varied from 0 (in a year of adequate rainfall) to 1.3 kg grain/ha/mm. Delaying aus sowing to late-April delayed the plan ng of aman, decreasing the growth dura on of BRRI dhan53 and reducing yield compared to yield of aman following earlier aus seeding. The dura on of BR11 was less strongly affected by sowing date and had similar yield level (> 4 t/ha) to that of BRRI dhan53 with early April sowing. In Experiment 2, dry seeded aus had similar yield for both late April and early May sowings (>4.6 t/ha), but yield of late sown and transplanted aus had lower yield (3.5 t/ha) due to submergence between flowering and the start of grain filling, while yield of late sown dry seeded aus was not affected as the crop was more advanced at the me of flooding. For aman, BRRI dhan49 had higher yield (>4.6 t/ha) than BRRI dhan46 (average 4.3 t/ha) for both establishment dates. In favorable condi ons, aus-aman cropping system yield ranged from 8.0 to 9.6 t/ha/yr. Thus, with mely establishment of aus (sowing near the end of April) using a modern, short dura on aus variety, followed by a medium dura on aman variety, it is possible to produce 8 to 9 t/ha/yr in the moderately saline coastal zone. Key message: It is possible to produce 8 to 9 t/ha of rice per year in medium salinity areas of the coastal zone of Bangladesh by growing a modern, short dura on, salt tolerant aus variety, followed by a modern aman variety. Keywords: cropping system, direct seeding, dry seeding, Khulna, irriga on 361 1. Introduc on The major constraints to produc vity of the agricultural lands in the polders of the coastal zone of Bangladesh are soil and water salinity during the dry season, waterlogging during the monsoon season, and lack of knowledge of modern agricultural interven ons (Karim et al. 1990). Poor soil physical proper es (bulk density, low permeability) and chemical proper es (high pH, low total N, exchangeable ca ons, low available P and micronutrients) also constrain produc vity of rice and non-rice crops. The salinity constraints are more acute and widespread in the south-west coastal zone (SWCZ) in Khulna Division than in the central coastal zone (Barisal Division) (Khan et al. these proceedings). Farmers in the SWCZ usually cul vate low-yielding (2.5 t/ha) tradi onal aman varie es in the monsoon season. Some farmers also cul vate sesame in the dry season, but this is o en damaged by the pre-monsoon rains due to late establishment and thus late maturity. Several a empts have been made over the past 20 years to increase the produc vity of agricultural cropping systems of the SWCZ, either by improving exis ng systems or developing new ones. Mondal et al. (2004) showed that yield of the aman crop could be more than doubled by the introduc on of modern high yielding varie es (HYV), which also reduced growth dura on and advanced the harvest. The earlier aman harvest allowed ‘early’ ( mely) establishment of boro rice (transplan ng in mid-December), meaning that the crop could be irrigated directly using river water for the first two months a er transplan ng. This also meant a lower stored water requirement to finish off the crop once the rivers become too saline (mid-February). Mondal et al. (2006) showed that annual rice yield could be increased to 6.0 to 8.5 t/ha with this HYV aman-boro system at Ba aghata, Khulna. However, the expansion of this cropping system is constrained by the limited amount of water that can be stored in the canal networks to finish off the boro crop, and the risk of salinity if the sluice gates are not properly closed once the river water becomes saline (Mondal et al. 2015). Another op on for increasing the produc vity of the SWCZ could be to intensify to an aus-aman system. This system is prac ced in many parts of Bangladesh (SRDI 2004), but not in the SWCZ. The aus crop needs to be established early enough so that the aman crop can be planted on me for maximum yield. However, early aus establishment usually requires irriga on. In the SWCZ, the river water is too saline for irriga on for early (April-May) aus establishment. However, delaying establishment of the aus crop un l a er the onset of the monsoon rains could lead to late harvest, delayed establishment of the aman crop and reduced aman yield. Dry seeding (direct seeding into dry lled soil) instead of transplan ng into puddled soil may enable early aus establishment. In water scarce areas of other parts of Asia, the use of dry seeding has enabled double cropping of rice during the rainy season, thereby increasing farm produc vity (Tuong and Bouman 2003; Tuong 1999). Dry seeding enables early crop establishment on the pre-monsoon rains, rather than wai ng un l there has been enough rain to puddle the soil for wet seeding or transplan ng. Therefore, the present study was undertaken with the hypothesis that with the use of a short-dura on, dry-seeded HYV aus variety followed by HYV aman, it is possible to intensify to a highly produc ve aus-aman system in the SWCZ. Successful implementa on would require knowledge of the op mum me for establishment of dry seeded aus. Sowing too early could lead to poor establishment due to water deficit stress. In such situa ons, the use of non-saline groundwater could be used to help establish the crop. But the availability of non-saline groundwater in the coastal zone is limited, and pumping is costly, therefore sowing and management prac ces that minimize irriga on water requirement would be needed. On the other hand, sowing too late could delay harvest and establishment of the aman crop beyond the op mum me, which may in turn reduce aman yield. Most commonly available aman varie es in Bangladesh are photoperiod-sensi ve (flowering is delayed un l October); late establishment of photoperiod-sensi ve varie es would limit biomass accumula on prior to anthesis, thus reducing grain yield. One possible solu on is the use of photoperiod-insensi ve aman varie es. Experiments were carried out to test the feasibility of an aus-aman cropping system in the SWCZ, and to develop a technology package that would lead to stable and high yield with minimum irriga on water requirement to establish the aus crop. The specific objec ves were: (i) to evaluate the effects of date and method of aus establishment and aus irriga on management on the growth and yield of both crops in an aus–aman system, (ii) to evaluate aman varie es for the aus-aman system, and (iii) to determine the effects of sowing date and water management on irriga on water requirement of dry seeded aus. 362 2. Materials and Methods 2.1. Experimental site Two experiments were conducted at two sites (one experiment per site) about 4 km apart in Ba aghata Upazila, Khulna District. Experiment 1 (2006-2008) was established in Basurabad village (site 1), and experiment 2 (2009) was in Fultola village (site 2). Both sites had a cropping system history of a transplanted local aman variety (yield about 2-3 t/ha), followed by a late planted, low yielding sesame crop. The soil at site 1 was silty clay and fairly uniform to the sampling depth (90 cm), with 45-53% clay and 44-54% silt (Table 1). The texture of the topsoil at site 2 was similar to that at site 1. The topsoil (0-15 cm) at both sites was slightly alkaline, with 2.0-2.6% organic ma er (Table 2). Total N content of the topsoil was low at 0.1%, while available P and exchangeable K content were adequate. Micronutrient content (S, B, Cu, Fe, Mn and Zn) of the topsoil at both sites was sub-op mal. The soil proper es were typical of soils of the coastal zone; Haque (2006) found that most saline soils of the coastal zone have low to very low organic ma er, N, P and micronutrient (Zn, Cu) contents. Table 1. Bulk density, par cle size analysis and texture of the soil profile at site 1 Depth (cm) 0-15 15-30 30-45 45-60 60-75 75-90 Bulk density (g cm-3) 1.4 1.6 1.5 1.3 1.4 1.4 Sand (%) 1.2 5.2 3.2 3.2 3.2 3.2 Silt (%) 47.8 49.8 53.8 47.8 46.8 43.8 Clay (%) 51.1 45.1 43.1 49.1 50.1 53.1 Texture Silty clay Silty clay Silty clay Silty clay Silty clay Silty clay Table 2. Chemical proper es of the topsoil (0-15 cm) at the experimental sites pH Experiment 1 Mean 8.0 SE1 0.1 Experiment 2 Mean 7.9 SE 0.1 1 Organic Total ma er N (%) (%) Exchangeable K Ca Mg -1 (meq 100 g soil) Avail. P S B Cu Fe Mn Zn (µg g-1) 0.1 0.0 2.0 0.1 0.4 0.0 24.4 1.3 3.6 0.1 13.8 2.0 125 12.8 0.7 0.1 3.6 0.2 20.7 1.5 10.3 0.9 0.7 0.1 0.1 0.0 2.6 0.1 0.5 0.0 17.0 0.8 5.3 0.2 6.7 1.2 71.6 7.2 0.8 0.0 4.8 0.7 95.8 15.3 12.4 0.7 0.8 0.0 SE = standard error of the mean 2.2 Experimental design 2.2.1 Experiment 1 (2006-2008) Experiment 1 was designed to evaluate the effects of aus sowing date, aus irriga on management and aman variety on the performance of the aus-aman cropping system. For the aus crop, a short dura on, salt tolerant variety OM1490 (released in October 2014 as BRRI dhan65) was established by dry seeding, while the aman crops were transplanted about 7 d a er harvest of the aus crops. 363 Aus treatments Main plots: Water regime (I) I1 = rainfed (RF) I2 = supplementary irriga on (SI) to ensure adequate soil moisture during crop establishment up to three weeks a er emergence I3 = full irriga on (FI) throughout the growing season Sub plots: Date of sowing (D) D1 = early April (5-10 April) D2 = mid-April (10-15April) D3 = late April (25-30 April) Aman treatments Main plots: Variety (V) V1 = BR11 (planted in the aus I3 plots) V2 = PVST2 (BR 5778-156-1-3-HR14) (planted in the aus I2 plots) Sub plots: Date of transplan ng (D) D1 = 10 August D2 = 15 August D3 = 30 August BR11 was selected as it is one of the highest yielding aman rice varie es and only slightly photoperiod sensi ve, with a nominal growth dura on of 145 d. The line PVST2 (released as BRRI dhan 53 in 2010) was also selected as it has the same yield poten al as BR11, but with 5 d less growth dura on. This variety also has moderate salinity tolerance. The aman varie es were randomly assigned to the I2 and I3 main plots, and transplan ng dates were randomized within each aman variety main plot. The I1 plots were also planted with aman crops to avoid border effects. The size of each experimental unit (sub plot) in both seasons was 8 m x 6 m. The main plots and internal bunds were 50 cm x 20 cm and 20 cm x 20 cm, respec vely, and were compacted to prevent seepage between adjacent plots. 2.2.2 Experiment 2 (2009) Based on the findings of experiment 1, it was hypothesized that transplan ng aus rice is more suitable than dry seeding if crop establishment is delayed un l rainfall becomes more frequent. Furthermore, it was hypothesized that the use of late planted photoperiod sensi ve varie es in the aman season would reduce yield due to reduced growth dura on in comparison with a suitable photoperiod insensi ve variety. Therefore a cropping system experiment was conducted to compare the effects of two aus establishment methods (dry seeded, transplanted), two aus seeding dates (30 April, 11 May) and two aman varie es (sensi ve and insensi ve to photoperiod) on cropping system produc vity (Table 3). It was planned to have a 10 d interval between the sowing (and thus transplan ng) dates of the aman sowing date treatments. But the seedlings transplanted on 1 September 2009 (sown on 1 August) were destroyed by heavy rain in the first week of 364 September. Therefore, seedlings from a backup seedbed (sown on 5 August) were used and both sowing date treatments were transplanted on 11 September (35 and 30 d a er sowing, DAS). Table 3. Details of treatments in experiment 2 Aus Main plot Establ. method Sub plot Date of seeding Harvest date Main plot Variety Dry seeding 30 Apr 11 May 30 Apr 11 May BR 46 Transplan ng 15 Aug 19 Aug 17 Aug 27 Aug BR 49 Aman Sub plot Date of seeding Harvest date (transplan ng) 5 Aug (11 Sep) 9 Dec 10 Aug (11 Sep) 11 Dec 5 Aug (11 Sep) 9 Dec 10 Aug (11 Sep) 11 Dec The aus crop was established in a strip-plot design with crop establishment method as the main plot in a single long strip, and date of seeding was in subplots with four replicates (two diagonally opposite subplots per replicate). The two aman varie es were planted in two strips within each aus establishment method plot. The size of each aus sowing date x aman variety unit (sub-subplot) was 4 m x 3 m. The main plot and internal bunds were 50 cm x 20 cm and 20 cm x 20 cm, respec vely. All bunds were compacted to prevent seepage between adjacent plots. 2.3. Crop management The land was plowed under dry condi ons for the dry seeded aus, and was puddled for the transplanted aus (2009) and all aman crops. No irriga on was required for puddling as a result of rainfall. The dry seeded aus was sown with 20 cm row spacing and three to four seeds were sown at a spacing of 20 cm within each row (hill density 25 m -2). The seedbeds for the transplanted rice (2009 aus and 2006-09 aman crops) were prepared in separate places outside the experimental fields. The rice seeds were soaked for 12 h, incubated for 48 to 72 h, and sown by broadcas ng at 100 and 50 g m-2 for the transplanted aman and aus, respec vely. The seedlings were raised using rainwater and groundwater during the aman and aus seasons, respec vely. Insec cide and urea were applied to protect the seedlings and to ensure good seedling growth (BRRI 2004). Thirty-day-old seedlings were transplanted a er final land leveling at a spacing of 20 cm x 20 cm in the aman season (BRRI 2009b). Seedlings of 20-day-old varie es were used in the 2009 transplanted aus at a spacing of 20 cm x 20 cm. Basal fer lizers at 60, 40, 10 and 60 kg P2O5, K2O, ZnSO4 and CaSO4 per ha, respec vely, were incorporated into the soil during the last harrowing of both crops. Urea was applied to both crops at 120 kg N ha-1 in four equal splits 15 d a er emergence (DAE), mid- llering (30 DAE), 5 d before panicle ini a on (DBPI, 45 DAE), and at heading in the aus season, and at basal, mid- llering (35 d a er transplan ng, DAT), 5 DBPI (45 DAT), and heading in the aman season. Good weed control was achieved in the dry seeded rice by applying the pre-emergence herbicide Longstar 25 EC @ 0.2 kg ai ha-1 to the moist soil 1 DAS, followed by the post-emergence herbicides Ronstar 25 EC and Longstar 25 EC @ 0.2 kg a.i. ha-1 two weeks a er emergence, plus hand weeding as needed. In the aman crops, hand weeding was done just before each urea top dressing. A acks of insects, pests, and diseases were closely monitored. Granular systemic pes cides diazinon 10G (1.68 kg a.i. ha-1) and carbofuran 5G (0.5 kg a.i. ha-1) were applied at 15-d intervals to protect the crops from stem borer infesta on. Diazinon 10EC (0.1 kg a.i. ha-1) was sprayed during the mid- llering stage at 15-d intervals to prevent infesta on of leaf roller and yellow stem borer each season. Carbendazim 50WP (0.25 kg ai ha -1) and carbosalfan 20EC (0.15 kg ai ha-1), along with copper oxychloride 50 WP (@ 0.1725 kg ai ha-1), were 365 sprayed to control brown spot and leaf blight. Gypsum and a mixture of potassium and sulfur (1:1 ra o), basally applied at 60 kg ha-1, helped to prevent fungal disease infesta on and aided crop recovery along with fungicides. Malathion 57EC was applied at 0.57 kg ai ha-1 during the grain-filling stage to prevent grain spot. Despite these measures, yield loss due to pests and diseases occurred in some treatments in some years (sec on 3.3.2). 2.4 Water management Groundwater (EC<2 dS/m) was used for irriga on in both experiments. In experiment 1, the aus crop was irrigated according to treatment using groundwater. At each irriga on water was added un l the topsoil was saturated. In 2009, both the dry seeded and transplanted aus were irrigated as needed throughout the season. The aman rice crops were mainly rainfed. During excessive rainfall, water was pumped from the experimental field as needed to maintain a suitable water depth for modern, high yielding rice varie es. 2.5 Monitoring 2.5.1 Soil salinity Salinity of the topsoil (0-15 cm) was determined before each aus seeding and at 15-d intervals during the period of aus establishment (April-June), and monthly at other mes. Electrical conduc vity of the satura on extract was determined using the American Society of Agronomy and Soil Science Society of America (1982) method at the Soil Research and Development Ins tute laboratory at Khulna. 2.5.2 Crop performance Establishment of dry seeded aus. A er seeding of the dry seeded aus, the number of hills in which plants had emerged was counted daily from four rows per plot, two rows from two sides of each plot, un l 50% of the hills had emerged plants (mean of four replicates). This was taken as the date of 50% emergence. Crop development. In both seasons, the dates of anthesis and physiological maturity were determined. The date of 50% anthesis was es mated by visual observa on of 10 rows per plot in each replica on. Physiological maturity was taken as the date when 80% of the grains in the whole field had become golden, following the procedures of Cassman (1994) and IRRI (1994). Yield components. Yield components (panicles density, total spikelets per panicle, % filled spikelets, 1000-grain weight on oven dry weight basis) were determined from 16 hills following the procedures of Cassman (1994) and IRRI (1994). Grain yield. Grain yield was determined in 4 m x 2 m and 2 m x 1 m areas in the middle of each sub plot during 2006-2008 and 2009, respec vely. Grain yield was calculated at 14% moisture content (Gomez 1972). 2.5.3 Irriga on amount The discharge rate of the tubewell was determined using the volumetric method (Michael 1978): Q = V x T where Q = discharge rate, m3/s V = volume of the bucket, m3 T = me required to fill the bucket, s The me required to irrigate each plot was recorded, and the depth of applied water was calculated from the product of discharge rate and me divided by the plot area (m2). Total irriga on during the growth of rice was determined by summing the amount of water applied during each irriga on. 366 2.5.4 Incremental water produc vity Incremental water produc vity was calculated during the aus 2006-08 seasons. Incremental water produc vity is the yield increase of the irrigated treatment over yield of the rainfed treatment divided by the amount of irriga on water applied. 2.5.5 Weather data A rain gauge and a USWB Class A evapora on pan were installed near the experimental field and monitored daily at 8 AM. Monthly totals of rainfall and evapora on were calculated. Other weather data including maximum temperature (Max T) and minimum temperature (Min T) were collected from Khulna meteorological sta on, about 8 km north of the experimental sites. 2.6 Sta s cal analysis In experiment 1, the crop and irriga on input data were analyzed using a standard split plot analysis of variance (ANOVA) for each crop. In experiment 2, the aus and aman crop data were analyzed using a strip plot ANOVA. Total system yield was determined by summing the yields of the aus and aman crops. Sta s cal analysis of total system yield (aus plus aman) in 2006-2008 was done for the systems in which the aus was fully irrigated (I3), by two factor ANOVA in a split plot design with aman variety as the main plot and aus sowing data as the sub plot. Total system yield in 2009 was done by three factor strip plot ANOVA with aus establishment method in the strips, and with aus sowing date and aman variety fully randomized sub plots. The least significant difference (LSD) test at 5% probability was used to compare treatment means. 3. Results 3.1 Weather The wet season started in May in all years except 2007, when it started in June, and usually ended between late September and mid-October (Figs 1a, 2). Annual rainfall varied from 1600 mm in 2008 to 2000 mm in 2006, about 80% of which occurred from June to October, compared with the long term (1999 to 2006) average of 1685 mm. Rainfall was unusually high in July 2006, August 2009 and September in all years, and unusually low in April in all years except 2007. Evapora on was usually similar to the long term values, except for unusually high evapora on during the rela vely dry rainy season (May-August) in 2008. Monthly evapora on ranged from 55 to 65 mm in January and December each year to around 180 mm in May to August of 2008. Temperature was favourable for rice throughout the cropping period (Fig. 1b). Mean monthly minimum and maximum temperatures during the aus-aman period were similar to the long term average each year. Mean maximum temperature tended to be highest during the period of establishment of aus (April/May, around 36°C), and fluctuated around 32°C, during the remainder of the aus-aman period each year. Mean monthly minimum temperature was around 20 to 25°C throughout the aus-aman period each year. 367 600 500 400 300 200 2006 RF 2007 RF 2008 RF 2009 RF Long Term RF 2006 EV 2007 EV 2008 EV 2009 EV Long Term EV (a) 100 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 40 (b) 35 30 25 20 15 2006 2007 2008 2009 Long Term 10 5 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Fig. 1. a) Monthly total rainfall (RF) and evapora on (EV) and b) monthly average maximum and minimum temperatures during 2006-2009 in comparison to the long term (1999-2006) data for Khulna City. 368 200 180 160 140 120 100 80 160 Rainfall 2006 D1 D2 D3 BR11 BRRI dhan53 140 120 100 80 60 60 40 20 0 200 180 160 140 120 100 80 60 40 20 0 40 20 0 160 Rainfall 2008 D1 D2 D3 BR11 BRRI dhan53 140 120 100 80 60 40 20 0 200 180 160 140 120 100 80 60 40 20 0 200 180 160 140 120 100 80 60 40 20 0 160 Rainfall 2007 D1 D2 D3 BR11 BRRI dhan53 140 120 100 80 60 40 20 0 160 Rainfall 2009 D1(DS) D2(DS) D1(TP) D2(TP) 140 120 100 BRRI dhan46_D2 BRRI dhan49_D2 80 60 40 20 0 Fig. 2. Daily rainfall during the aus and aman seasons of 2006 to 2009. The dates of seeding, emergence and physiological maturity of fully irrigated (I3) aus (2006-2008), and of dry seeded (DS) and transplanted (TP) aus (2009) are indicated by colored dots. The dates of transplan ng and maturity of the aman varie es are indicated by colored triangles 3.2 Soil salinity Topsoil salinity of the satura on paste extract (ECe) generally varied from about 4 to 6 dS/m, with higher values in April and June 2007 as a result of the late onset of the monsoon rains (Fig. 3). Salinity decreased to 2-3 dS/m during the rainy season in some but not all years. 14 12 10 2006 2007 2008 2009 8 6 4 2 0 Fig. 3. Topsoil (0-15 cm) ECe (dS m -1) of the study area during the study period (2006-2009). Ver cal capped bars are standard errors of the means. 369 3.3 Effect of sowing date and irriga on method and establishment method on aus crop performance 3.3.1 Insect, pest and disease infesta on Pest and disease infesta on were common during both the aus and aman cropping seasons each year. In both seasons leaf rollers a acked during the early vegeta ve stage, stem borers from the vegeta ve to flowering stages, and rice bugs in the grain filling stage; however, these infesta ons were well-controlled and did not affect yield. The main diseases were leaf blight in 2006, and brown spot and grain spot in 2007-2009. These diseases were also well-controlled and not yield-reducing (sec on 2.3). For irrigated dry seeded aus, soil moisture fluctuated between saturated and field capacity most of the me during the early establishment period. This created a environment conducive for soil pests (cutworms and ants). Whenever the soil surface became dry, insects were visible in the field. Cutworms cut the roots of the young plants and the problem only became visible when the hills became yellowish, and ants removed the sown seed. Severe cutworm damage occurred in the irrigated treatments of the early sown (D1) 2007 aus. Severe infesta on also occurred in all sowing date treatments in the 2008 cropping season, mainly affec ng I3. Repeated applica on of systemic insec cides at double the usual rate did not stop the a ack in 2008, probably due to adultera on of the insec cide. In aus 2006, cutworms were not a problem, probably due to the higher rainfall in May. Cutworms and ants caused li le damage in 2009, because of the combina on of irriga on and insec cide provided good control. 3.3.2 Aus establishment and popula on dynamics Fi y percent emergence of dry seeded aus in the irrigated treatments occurred from 6 to 20 DAS depending on the incidence of rainfall and on the irriga on treatment. Emergence of dry seeded rice was faster with irriga on than under rainfed condi ons, except for the early April 2007 sowing when there was op mal rain (10 and 28 mm 1 d before and 2 to 3 d a er the first sowing date, respec vely)(Fig. 4). In most situa ons, final hill density under rainfed condi ons (5-10 hills m-2) was much lower than under irrigated condi ons (14-24 hills m-2). The irrigated treatments did not suffer from water stress during establishment, yet hill density was some mes less than the theore cal maximum. This was due to insect pest damage as described above. De Da a (1981) reported that cutworm is a problem in upland (dry seeded) rice. In the dry seeded plots of experiment 2 in 2009, hill density was the theore cal maximum (25 hills m-2) for both seeding dates because full irriga on enhanced crop establishment and there was no pest damage. 3.3.3 Water depth during the aus crops The experimental field was occasionally flooded by water from surrounding (higher) lands that overtopped the bund surrounding the experimental field following heavy rainfall. In June 2006, the en re crop was submerged for 4 d, when the irrigated crops of D1 and D2 were at the panicle ini a on stage and all the rainfed crops and the irrigated crops of D3 were at mid llering (Table 4). High rainfall in July 2006 again submerged the irrigated rice of D1 and D2 at the flowering stage. In 2009, the late sown transplanted aus (M2 D2) was submerged for several days between flowering and grain filling. 370 Table 4. Rainfall and water levels inside and outside the experimental field that caused par al or complete submergence of the crops, and the dates and stages at which submergence occurred during the aus season Average Water Total field level Submergence Raifall period rainfall water outside Crop phenology1 (treatments)2 affected period level expt field (cm) (cm) (cm) 2006 24 May-10 Jun 32 32 32 10 -14 Jun MT1 (I1D1, I1D2*, I1D3 ; I2D3 I3D3 ) aus PI2 (I2D 1,I2D2* , I3D1, I3D2* ) aus 1–14 Jul 40 40 50 7-14 Jul FL3 ( I2D1,I2D 2* , I3D1, I3D2* ) aus, seedbed aman (D1) 2009 20 Jul-9 Aug 60 80 80 6 -12 Aug FL-GF4 (transplanted D2* aus) 1 MT = maximum llering, 2PI = panicle ini a on, 3FL = flowering, 4GF = grain filling 5 I =Water regimes in the aus season; I1 = rainfed, I2 = supplementary irriga on, and I3 = full irriga on 6 D = date of sowing; D1 = 5-10 Apr, D2 = 15-20Apr, D3 = 25-30 Apr (2006-08); D1= 30 Apr and D2 = 10 May (2009) *Bold fonts indicate severely affected crops 25 25 D1, 2006 D2, 2006 20 20 15 I1 I2 I3 10 10 5 5 0 0 25 20 25 D3, 2006 5 0 D1, 2007 20 15 10 I1 I2 I3 15 15 I1 I2 I3 10 5 I1 I2 I3 0 371 25 25 D2, 2007 D3, 2007 20 20 15 15 10 I1 I2 I3 5 0 25 D1, 2008 15 I1 I2 I3 5 0 0 D3, 2008 20 I1 I2 I3 10 5 25 D2, 2008 20 20 10 5 0 25 15 I1 I2 I3 10 I1 I2 I3 25 20 15 15 10 10 5 0 2009 30-Apr 10-May 5 0 Figure 4. Hill density of dry seeded aus from seeding to physiological maturity (PM) as affected by water regime (I1= rainfed, I2= supplementary irriga on and I3 = full irriga on) and sowing date (D1, D2 and D3) during 2006-2008 cropping seasons, and irrigated dry seeded aus as affected by sowing date (D1 and D2) in 2009. 3.3.4 Aus phenology 2006-2008. Crop dura on varied from around 110 to 130 d across sowing dates and irriga on treatments each year (Table 5). There was a consistent trend for around 10 d shorter dura on as sowing was delayed from early to late April in both rainfed and irrigated treatments. The decrease in dura on was due to a shorter me from sowing to flowering as sowing was delayed, which was probably due to increasing temperature during April/early May (Fig. 5) (Yoshida 1981). There was li le effect of me of sowing on dura on of the grain filling period. 372 Flowering and PM were delayed in the rainfed treatments when establishment was delayed as a result of inadequate rain, as for the first sowing each year, and the second sowings in 2006 and 2007. Physiological maturity was delayed in the irrigated treatments of the second sowing in 2006 as a result of water stagna on during flowering (Table 4). Table 5. Time (days a er sowing, DAS) of flowering and physiological maturity of aus crops as affected by irriga on treatment and date of sowing (2006-2008) and by establishment method and date of seeding (2009) I1 2006 – dry seeded 10 Apr 107 15 Apr 102 30 Apr 88 2007 – dry seeded 7 Apr 98 16 Apr 1042 29 Apr 97 2008 9 Apr 106 16 Apr 95 29 Apr 95 2008 – dry seeded 9 Apr 106 16 Apr 95 29 Apr 95 2009 – dry seeded 30 Apr 11 May 2009 – transplanted 30 Apr 11 May - Flowering I2 I3 Physiological maturity I1 I2 I3 I1 Grain filling I2 I3 881 97 95 97 95 87 129 124 110 117 113 109 117 113 109 22 22 22 29 16 14 20 18 22 98 92 90 98 92 88 123 1302 120 121 120 110 125 117 110 25 26 23 23 28 20 27 25 22 99 92 92 94 86 85 129 119 116 117 112 107 117 112 106 23 24 21 18 20 15 23 26 21 99 92 92 94 86 85 129 119 116 117 112 107 117 112 106 23 24 21 18 20 15 23 26 21 - 75 70 - - 107 100 - - 32 30 - 78 78 - - 109 1193 - - 31 41 1 Delayed and poor establishment (Fig. 3b) 2 Water stagna on during flowering 3 Water stagna on during flowering and early grain filling 373 45 40 35 2006 2007 2008 2009 30 25 20 15 Fig. 5. Mean daily temperature (5-d means) at Khulna in April to May during 2006 to 2009. 2009. Dura on of the 30 April dry seeded and transplanted aus crops was similar to that of the late April sown irrigated crops in the previous three years. Delaying sowing to 11 May further reduced dura on of the dry seeded crop to 100 d, however, dura on of the transplanted crop increased to 119 d (Table 5). The longer dura on of the late sown transplanted crop was due to both later flowering and a longer grain filling period. The longer grain filling period was due to the fact that the crop was flooded (80 cm water depth for about 7 d during flowering and early grain filling) (Table 4). 3.3.5. Yield and yield components Experiment 1. Grain yield of the sowing date by irriga on management treatment combina ons ranged from 0.7 to 4.5 t/ha (Figs 6a-c). Yields in 2008 were generally lower than in the other two years due to pest infesta on and less rainfall. The interac on between water regime and sowing date on all yield components and grain yield was not significant (P=0.05) in 2006 and 2008, but was significant in 2007. In 2006 and 2008, grain yield was significantly affected by sowing date, but not by irriga on treatment. The response to irriga on treatment and sowing date was variable depending on the incidence and amount of rainfall each year, and of flooding. Under favourable condi ons (adequate water, no submergence or water stagna on, no severe disease or pest infesta on), yield was similar for all sowing dates and around 4 t/ha. There was a consistent trend for the irrigated treatments to have higher yields than rainfed treatments in years when the rice was seeded during dry spells, with significant differences in 2007. This was due to poor establishment (Figure 4) and low panicle density (Table 6) of the rainfed treatments in the absence of adequate rainfall. Be er establishment (higher plant density) resulted in higher yield, except when submergence occurred at cri cal stages (in I2 D2 and I3 D2 of 2006) (Table 4) and when pest damage occurred (all irriga on treatments of D1 2007 and D1 2008)(sec on 3.3.1)(Ritu 2012). Submergence of the second sowing of both irrigated treatments in 2006 tended to reduce panicle density and/or the number of spikelets per panicle. Experiment 2. In aus 2009, yields ranged from 3.5 to 4.7 t/ha (Fig. 7). The interac on between establishment method and sowing date on grain yield was significant at p<0.064. The interac on was due to a large decline in yield of transplanted rice for the second sowing due to submergence between flowering and the start of grain filling (Table 4). The panicles were fully submerged, whereas the other treatments were more advanced 374 and the panicles were above the water. The submergence of the later sown transplanted rice resulted in a 23% reduc on in the number of spikelets per panicle, whereas the number of spikelets per panicle in the dry seeded rice increased with the second sowing (Table 7). This is consistent with the findings of Reddy and Mi ra (1985) that submergence of rice at flowering is very detrimental to grain yield. Delaying sowing reduced panicle density of both the dry seeded and transplanted rice, but in the case of the dry seeded rice, this was compensated for by more spikelets per panicle. Within sowing date, panicle density of dry seeded rice was much higher than for transplanted rice, as commonly found (e.g. Sudhir-Yadav et al. 2011). 6000 6000 Mean I 5000 (a) LSD0. 05 4000 3000 Early April Mid April Late April 2007 Grain yield (kg ha ¹) Grain yield (kg ha ¹) 2006 5000 (b) LSD0. 05 a 4000 a a a a ab b 3000 b b 2000 2000 1000 1000 0 I1 0 Early April Grain yield (kg ha ¹) 6000 5000 Mid April 2008 (C) I2 Late April I3 Water regime Mean I LSD0. 05 4000 3000 2000 1000 0 Early April Mid April Late April Fig. 6. Grain yield of aus rice in (a) 2006, (b) 2007 and (c) 2008 as affected by sowing dates (D1 = early April, D 2 = mid April and D 3 = late April) and water regimes (I1 = rainfed, I2 = supplementary irriga on, I 3 = full irriga on). Ver cal bars are the LSD (P=0.05) for seeding dates in 2006 and 2008, and for the interac on between seeding date and irriga on treatment in 2007. 6000 (d) DS LSD0.05 5000 TP a 4000 b 3000 2000 1000 0 D1 Seeding date D2 Fig. 7. Grain yield of aus rice as affected by establishment method and sowing date in 2009. DS = dry seeding, TP = transplan ng, D 1 = sowing on 30 April, D2 = sowing on 11 May. Ver cal bars are the LSD (P=0.05) for the interac on between sowing date and establishment method. 375 Table 6. Effect of sowing date and water regime on panicle density, number of spikelets per panicle, percent filled grain and thousand grain of aus rice (2006-2008) 2006 cropping season I1 I2 I3 Mean D Produc ve panicles ( m-2) D11 138 b4 212 b 224 a D2 140 b 180 ab 198 a D3 279 a 248 b 253 b 2 LSD.05 48.3 Spikelets per panicle (no) D1 78 a 76 a 64 b D2 65 ab 73 a 53 b D3 63 a 66 a 71 a LSD.05 = 12.0 Percent filled spikelet (%) D1 76 73 69 73 y D2 72 73 77 74 y D3 84 81 84 72 x Mean I 78 76 77 Thousand grain weight (gm) D1 21.6 a 21.2 a 20.9 b D2 21.2 a 21.5 a 21.0 a D3 21.5 a 21.6 a 21.4 a LSD.05 = 0.42 2007 cropping season I1 I2 I3 3 2008 cropping season I1 I2 I3 Mean D D1 D2 D3 357 a 323 ab 274 b 318 a 308 a 309 a 131 b 212 a 233 a LSD.05= 48.7 D1 D2 D3 165 112 46b 205 184 129 147 126a 146a LSD.05 =56.0 D1 D2 D3 55 ab 60 a 52 b 73 a 62 b 59 b 79 a 70 b 65 b LSD.05 = 6.8 D1 D2 D3 62 a 59 b 59 b 52 b 57 ab 69 a 60 b 71 a 67 ab LSD.05 = 9.7 D1 D2 D3 79 a 83 a 84 b D1 D2 D3 82 87 79 Mean I D1 D2 D3 20.9 20.6 20.9 Mean I D1 D2 D3 77 ab 64 b 86 a 85 a 88 a 89 a LSD.05 = 5.4 21.5 a 21.0 ab 20.8 b 21.6 a 21.4 a 21.3 a 21.2 b 21.9 a 21.8 a LSD.05 = 5.5 87 88 86 82 87 87 84 87 20.8 21.2 21.3 21.3 21.5 21.7 20.8 a 21.2 a 85 x 87 x 83 y 86 20.9 x 21.1 x 21.4 x 21.4 a 1 D1 = Sowing on 5-10 April, D2 = Sowing on 15-20April, D3 = Sowing on 25-30 April; 2 LSD.05 = LSD values for comparing three seeding dates under the same water regimes at 5% level of significance; 3 I1 = Rainfed, I2 = supplementary irriga on, I3 = full irriga on; in each season and in each row mean values followed by the same le er are not significantly different at 5% level by LSD. Mean values are averaged over four replica ons; 4 In each season and in each column mean values followed by the same le er are not significantly different at 5% level by LSD. Mean values are averaged over four replica ons. 376 Table 7. Effect of sowing date and establishment method on yield components of aus in 2009 Establishment method1 (M) DS TP Mean D Seeding date2 Panicle density (no. m-2) D1 534 D2 403 Mean M 468a Filled spikelets (%) D1 80 D2 84 Mean M 82 a 406 318 362b 80 84 82 a 470a 361b 80 y 84 x DS TP Mean D D1 D2 Spikelets per panicle (no.) 63 b 75 a 72 a 58 b D1 D2 Mean M LSD.05= 11.42 Thousand-grain weight (g) 21.7 21.8 21.7 x 21.7 21.3 21.5 x 21.7 a 21.5 a 1 DS = dry seeded, TP = transplanted; 2 D1 = sowing on 30 April 2009, D2 = sowing on 11 May 2009; 3 In each row, mean values followed by the same le er (a, b) are not significantly different at the 5% level by LSD. Mean values are averaged over four replica ons; 4 In each column, mean values followed by the same le er (x, y) are not significantly different at the 5% level by LSD. Mean values are averaged over four replica ons; 5 6 LSD.05 = LSD for comparing establishment methods under the same seeding date; LSD.05 = LSD for comparing seeding date for same establishment method. 3.4 Effect of transplan ng date and variety on performance of aman rice 3.4.1 Insect and pest infesta on In all four aman seasons, there was li le damage from insects or diseases. The main insect pests were leaf roller and stem borer during the mid llering stage, which were controlled by cypermethrin and carbofuran and/or diazinon. Rice bug a acked during the grain filling stage and this was controlled by propiconazol. 3.4.2 Water depth during the aus growing period Farmers usually use the river water to prepare the land for aman transplan ng from late July. This results in a water depth of about 30 cm in lowland areas such as the experimental site, including the seedling nursery. This high water level together with high rainfall in July/August caused seedbed submergence during 2006, 2008, and 2009. Seedlings of D1 in 2006, D1 and D3 in 2007, all sowing dates in 2008, and D1 in 2009 were fully submerged for 6 d (Table 8). Water in the seedling nurseries was rapidly removed by pumping and there was serious damage to the seedlings due to con nuous rainfall and water stagna on. The aman crop was affected by submergence during the mid llering stage of all three seedings of 2006, and during the mid llering stage of D1 in 2007. Submergence during panicle ini a on occurred in D1 in 2007 and D2 in 2008. Basal fer lizer applied during transplan ng, and top dressed urea at the MT (maximum llering) and PI (panicle ini a on) stages was also lost from the target plots as a result of pumping out excess water or inunda on of the whole experimental area by flood water entering the experimental site from outside by overtopping the bunds. 377 Table 8. Rainfall and water levels inside and outside the experimental field that caused par al or complete submergence of the crops, and the dates and stages at which submergence occurred, during the aman season Raifall period Total rainfall (cm) 1–14 Jul 19–24 Sep 40 47 11–16 Aug 26 Aug – 8 Sep 23–25 Sep 14 21 18 10 - 19 Jul 5– 20 Aug 25–27 Sep 18 20 25 4-9 Sep 27 Average Water field level water outside level expt field (cm) (cm) 2006 40 50 62 70 2007 19 30 30 35 30 33 2008 28 35 30 40 37 45 2009 40 60 Submergence period Crop phenology1 (treatments)2 affected 7-14 Jul 21-30 Sep seedbed (D1) MT1 (D1,D2 and D3) 12-17 Aug 7- 17 Sep 24 -30 Sep Seedling D1 Seedling of D3 and MT D1 PI2 (D1) 14-18 Jul 11-17 Aug 28-30 Sep Seedbed seedbed and main field seedling PI (D2) 4-9 Sep D1 plan ng delayed 1 MT = maximum llering; PI = panicle ini a on; 2 3 D = date of transplan ng D1 =10 Aug, D2 =20 Aug, D3 =30 Aug (2006-08); D1=11 Sept (2009) (30 days old seedling). *Bold fonts indicate severely affected crops and seedlings. 3.4.3 Phenology Growth dura on over the three sowing dates and three years from 2006 to 2008 ranged from 131 to 145 d for BR11, and from 121 to 136 d for BRRI dhan53. In 2009, dura on of BRRI dhan46 ranged from 121 to 126 d, and for BRRI dhan49 the range was 123 to 138 d. BR11, BRRI dhan49 and BRRI dhan53 all exhibited photoperiod sensi vity, with greater sensi vity in BRRI dhan53 and BRRI dhan49. The dura on of BR11 was decreased by 1 to 10 d with delay in sowing from 10 to 30 August during 2006 to 2008 (Table 9). The dura on of BRRI dhan53 decreased by 8 to 14 d with delay in sowing in the same years, while the dura on of BRRI dhan49 decreased by 15 d in 2009 with a delay in sowing of only 5 d. In the case of BR11 and BRRI dhan53, the decrease in dura on with delay in sowing was due to a decrease in the me to flowering. The effect of sowing date on dura on of the grain filling period was small and inconsistent. However, in the case of BRRI dhan49, there was a large decrease (by 13 d) in the dura on of the period from flowering to PM with delayed sowing. 378 Table 9. Effect of sowing date and variety on me (days a er sowing, DAS) of flowering and physiological maturity of aman crops in 2006 to 2009 Year Transpl. date 2006 10 Aug 20 Aug 30 Aug 2007 10 Aug 20 Aug 30 Aug 2008 10 Aug 20 Aug 30 Aug 2009 11 Sept 11Sept 1 Flowering (DAS) Physiological maturity (DAS) Dura on of grain filling (d) BR11 BRRI dhan53 BR11 BRRI dhan53 BR11 BRRI dhan53 109 106 145 134 36 28 108 100 141 136 33 36 106 93 144 126 38 33 111 103 141 133 30 30 106 95 133 125 27 30 101 92 131 119 30 27 106 100 136 133 30 33 98 92 132 124 34 30 102 89 133 121 31 32 BRRI dhan46 BRRI dhan49 BRRI dhan46 BRRI dhan49 BRRI dhan46 BRRI dhan49 93 96 126 138 33 44 90 92 121 123 31 31 1 Seedlings were transplanted 30 DAS, except for the first sowing in 2009 when the seedlings were transplanted 35 DAS (sown on 5 Aug) 3.4.4 Yield Yields of BR11 ranged from 3.5 to 4.9 t/ha, and of BRRI dhan53 from 2.8 to 4.5 t/ha, across the three sowing dates and three years 2006 to 2008 (Figs 8a-c). There was a significant interac on between variety and sowing date on grain yield of aman in 2006 and almost (p<0.06) in 2008, but not in 2007 (Figure 8). Yield of BR11 was significantly higher than yield of BRRI dhan53 in 2007. Yield of BR11 was not affected by sowing date, indica ng that the reduced dura on as sowing was delayed from 10 to 30 August did not affect yield significantly. These results are in contrast with other findings that a growth dura on of 145 d is needed for maximum yield (6.5 t/ha) of BR11 (BRRI 2009a). However there was a consistent trend for yield of BRRI dhan53 to decline with delay in sowing in all three years, with significant differences in 2006 and 2008, probably reflec ng the greater reduc on in dura on of this variety. The reduc on in yield with delayed sowing of BRRI dhan53 was associated with a trend for lower panicle density and grain weight in 2006 (significant) and 2008, and also significantly fewer spikelets per panicle in 2006 (Table 10). Both varie es produced higher yield (>4 t/ha) in 2006 (first two seeding dates) and 2008 (first seeding date) than in 2007 (<4 t/ha) as the aman crop was submerged three mes in 2007. In addi on to the direct effects of submergence on crop processes, submergence prevented proper topdressing of N fer lizer during cri cal stages including panicle ini a on. Widawsky and O’Toole (1990) reported that submergence can cause up to 70% yield loss in rainfed lowland rice. 379 200 6 700 0 600 0 500 0 LSD.05 a a a 200 7a B R11 BRRI d han 53 700 0 Me an D 600 0 a LSD 0.05 500 0 b 400 0 400 0 b 300 0 300 0 200 0 200 0 100 0 100 0 0 10 A ugust 20 A ugust 30 A ugust 0 Date of transpl anting B R11 200 7b 700 0 200 8 Me an V 600 0 500 0 BRRI d han 53 700 0 B R11 BRRI d han 53 600 0 LSD=0.0 5 500 0 400 0 400 0 300 0 300 0 200 0 200 0 100 0 100 0 0 10 A ugust 20 A ugust LSD=0.0 5 0 30 A ugust 10 A ugust Date of transpl anting 20 A ugust 30 A ugust Date of transpl anting Fig. 8. Grain yield of aman as affected by sowing date and variety. Ver cal bars are LSD (P=0.05) for the interac on between seeding date and variety in 2006 and 2008, and for the effects of variety and sowing date in 2007. 2009 7000 Mean D 6000 5000 LSD0.05 a b 4000 3000 2000 1000 0 BRRI dhan46 Fig. 9. Grain yield of aman varie es during the 2009 aman cropping season. 380 BRRI dhan49 In 2009, the interac on between sowing date and variety on grain yield was not significant. Grain yield of BRRI dhan49 (4.7 t/ha) was significantly higher than of BRRI dhan46 (4.3 t/ha). There was no effect of seeding date on grain yield. The higher yield of BRRI dhan49 was associated with significantly more spikelets per panicle and significantly higher spikelet fer lity, although this was partly compensated for by lower grain weight (Table 11). The results suggest that BRRI dhan49 can s ll give acceptable yields, even if sowing is delayed un l 10 August, despite its photoperiod sensi vity. This is consistent with a recent BRRI finding that BRRI dhan49 gave yields of 4.5 t/ha when 35 d-old seedlings were transplanted on 15 September (BRRI 2009a). Table 10. Effect of variety on panicle density, number of spikelets per panicle, percent filled grain, and thousand-grain-weight during different establishment dates of aman rice in 2006 to 2008 Transplan ng date1 2006 cropping season Variety 2 V1 V2 Mean D D1 D2 D3 Mean V LSD.053 Interac on 210 217 234 220 D1 D2 D3 Mean V LSD.05 for D Interac on 115 107 102 108 7.42 D1 D2 D3 Mean V 210 199 198 202 N/S 117 a 97 b 91 b 102 116 a4 102 b 97 c N/S Spikelets per panicle (no.) 131 107 119 x 5 115 103 109 y 122 113 117 x 123a 108b 67 66 77 72 76 69 73 69 LSD.05 for D =7.26 22.0 a 22.2 a 22.2 a 0.42 22.1 a 21.8 a 20.2 b 2008 cropping season Variety V1 V2 Mean D 239 218 223 201 244 188 235 a 202b LSD.05 for v=17.8 N/S 92 a 93 a 98 b 229 212 216 106 a 102 a 112 a 14.9 N/S N/S Interac on D1 D2 D3 Mean V LSD.05 Interac on 210 208 216 2007 cropping season Variety V1 V2 Mean D -2 Panicle density (no. m ) 202 171 186 167 174 171 194 180 187 188 175 67 y 74 x 73 x 5 Filled spikelets (%) 71b 79a 73a 70a 68a 65a 76 a 75 a 75 a LSD.05 = 5.6 LSD.05 = 3.6 N/S Thousand-grain weight (g) 22.3 23.0 22.6 a 22.3 22.3 22.3 a 20.7 20.7 20.7b 21.8 22.0 20.8 21.0 20.4 20.7a 1.06 N/S 74 a 75 a 81 b 21.7 21.5 20.3 21.2 x 21.3 x 20.4 y 21.2b 1 D = transplan ng date, D1 = 10 August, D2 = 20 August, D3 = 30 August; V= rice Varie es, V1 = BR11, V2 = BRRI dhan53; 3 LSD.05 = LSD values for comparing three seeding dates under the same variety at 5% level of significance; 4 In each season and in each row, mean values followed by the same le er (a, b) are not significantly different at the 5% level by LSD. Mean values are averaged over four replica ons; 5 In each season and in each column, mean values followed by the same le er (x, y) are not significantly different at the 5% level by LSD. Mean values are averaged over four replica ons. 2 381 Table 11. Effect of variety and seeding date on yield components of aman rice in 2009 Transplan ng date2 D1 D2 Mean V Interac on LSD.053 Filled spikelets (%) D1 D2 Mean V Interac on LSD .05 Varie es1 V1 V2 Panicle density (no. m-2) 251 293 274 250 263 271 N/S 72 68 70 b4 N/S 81 82 82 a Mean D 272 262 76 75 Varie es V1 V2 Spikelets per panicle (no.) 83 110 83 102 83 b 106 a N/S Thousand-grain weight (g) 23.1 17.6 22.6 17.9 22.9 a 17.7 b N/S Mean D 96 93 20.4 20.2 1 V1 = BRRI dhan 46 (photoperiod-sensi ve), V2 = BRRI dhan49 (photoperiod-insensi ve); 2 D1 = sowing on 5 Aug and transplan ng on 11 Sep (35-day-old seedlings), D2 = sowing on 10 Aug and transplan ng on 11 Sep (30-day-old seedlings); 3 LSD.05 = LSD values for comparing seeding dates for the same variety at the 5% level of significance; In each row, mean values followed by the same le er are not significantly different at the 5% level by LSD. Mean values are averaged over 4 replica ons. 4 3.5 Annual yield of the aus–aman cropping system Total system yield ranged from 5.5 t/ha to 8.0 t/ha over the three years from 2006 to 2008, and was much higher (8.0 to 9.6 t/ha) in 2009. The effects of aus sowing date and aman variety on annual system yield were inconsistent over the three years 2006 to 2008. The interac on between aman variety and aus seeding date was significant (P<0.05) in 2006 and 2008, but not in 2007. With BR11, there was no consistent effect of seeding date, but with BRRI dhan53, total system yield declined with delay in sowing in both 2006 and 2008 (Figs 10a, d). In 2007, system yield was significantly lower with the earliest sowing (Figure 10c). The BRRI dhan65-BR11 aus-aman system produced annual yields of 7 to 8 t/ha for all sowing dates under favourable condi ons (no significant damage from drought, waterlogging or pests). Yields of this system were lower (6 to 7 t/ha) in 2008 because of lower aus yield caused by drought and insect infesta on. Total yield of the aus-aman system in 2009 varied from 7.9 to 9.6 t/ha, and was significantly affected by the interac on between aus sowing date, aus establishment method and aman variety (Fig. 11). With dry seeded aus in the system, total yield (8.8 to 9.6 t/ha) was not affected by aus sowing date x aman variety. However, with late sowing of transplanted aus, system yield was lower with the photoperiod sensi ve variety (BRRI dhan49) (7.9 to 8.3 t/ha) than with the non-photoperiod sensi ve variety (BRRI dhan46) (8.7 to 9.2 t/ha). 382 2006 12000 5-10 Apr 15-20 Apr 25-30 Apr LSD 0.05 10000 a a 8000 2007a LSD = 22.9 8000 b c 6000 6000 4000 4000 2000 2000 0 0 BRRI dhan65-BR11 2007b Mean of sowing dates 10000 a b 12000 BRRI dhan65-BRRI dhan53 Variety 12000 10000 8000 Mean BRRI dhan65-BR11 and BRRI dhan65-BRRI dhan53 LSD 0.05 a b a BRRI dhan65-BR11 2008 12000 10000 6000 6000 4000 4000 2000 2000 5-10 Apr 15-20 Apr Date 25-30 Apr 5-10 Apr 15-20 Apr 25-30 Apr LSD 0.05 8000 0 BRRI dhan65-BRRI dhan53 b b a a b b 0 BRRI dhan65-BR11 BRRI dhan65-BRRI dhan53 Variety Fig. 10. Total grain yield of the aus-aman cropping system as affected by aus seeding date and aman variety and aman variety during (a) 2006, (b-c) 2007 and (d) 2008. Ver cal bars are the LSD (P=0.05) for comparing al l treatment combina ons in 2006 and 2008, and for comparing seeding date and variety means in 2007. 2009 12000 BRRI dhan46 BRRI dhan49 10000 8000 6000 4000 2000 0 30 Apr 11 May BRRI dhan65 (DS in aus) 30 Apr 11 May BRRI dhan65 (TP in aus) Establishment method and date Fig. 11. Total grain yield of the aus-aman cropping system as affected by aus establishment method and sowing date and aman variety in 2009. Ver cal bar is the LSD (P=0.05) for the interac on between between aus establishment method and aman variety x sowing date. 383 3.6 Effect of seeding date and irriga on management on irriga on input to aus During the three aus seasons of 2006 to 2008, rainfall intercepted by each treatment combina on varied from 664 to 1043 mm (Table 12). Irriga on input to I2 ranged from 94 to 188 mm, and to I3 from 87 to 300 mm. There was a significant interac on between sowing date and irriga on treatment (supplementary vs. full) on irriga on input to the aus crops in all three years (2006-2008). Each year, irriga on input to the fully irrigated treatment declined as sowing was delayed, while the effect of sowing date on irriga on input to the supplementary-irrigated treatment was variable. Total water input (rainfall plus irriga on) to aus ranged from about 900 mm in 2008 to about 1300 mm in 2006. In 2009, the aus crops received around 725 mm rain, and irriga on input ranged from 562 to 1040 mm depending on sowing date and establishment method (Table 13). The interac on between seeding date and establishment method on irriga on input was not significant. Irriga on input was significantly higher with early sowing than late sowing, and with dry seeding than with transplan ng, reflec ng the dry weather in May (Fig. 1). Table 12. Total water input from rainfall and irriga on in aus growing season (2006-2008), Basurabad village, Ba aghata, Khulna, Bangladesh I1 2006 I2 Rainfall (mm) D1 1068 996 D2 1068 996 D3 1043 1043 Irriga on (mm) D1 0 116 D2 0 101 D3 0 161 LSD 87 Total water applied (mm) D1 1068 1113 D2 1068 1098 D3 1043 1204 LSD 76 384 I3 I1 2007 I2 996 996 1043 800 851 842 897 913 875 922 957 923 664 664 659 773 715 745 767 738 747 300 279 262 0 0 0 108 100 94 63 251 195 142 0 0 0 188 140 86 56 183 159 87 1297 1276 1305 800 851 842 1005 1014 968 95 1173 1152 1065 664 664 659 961 855 831 86 950 897 834 I3 I1 2008 I2 I3 Table 13. Total water input from rainfall and irriga on to aus in 2009 Dry seeding 30 Apr-15 Aug 11 May – 19 Aug Transplan ng 26 May- 17 Aug 6 June-27 Aug Establishment method means Dry seeding Transplan ng LSD (p=0.05) Sowing date means 30 April 11 May LSD (p=0.05) Rainfall (mm) 723 728 728 728 Irriga on (mm) 1040 723 849 562 Total water applied (mm) 1763 1451 1577 1290 725 728 n/a 882 629 36 1607 1365 n/a 725 728 n/a 868 642 36 1593 1370 n/a 3.7 Incremental water produc vity Under favourable condi ons (no submergence, water stagna on or severe pest or disease infesta on), there was a general trend for higher incremental input water produc vity with supplementary irriga on than full irriga on within seeding date, with bigger differences for earlier seeding, except for D3 of 2008 (Table 14). Incremental water produc vity of both irriga on treatments was higher when the crop faced drought that hampered rainfed rice establishment (e.g. D3 of 2007). On the other hand, there were mes when rainfed rice had a higher yield than irrigated rice because the la er suffered from pest infesta on or submergence. The computa on of incremental water produc vity is not meaningful for the third sowing date in 2006 and for the first two sowing dates in 2007 as irrigated yield was lower than or equal to rainfed yield. Rainfed D1 and D2 of the 2008 cropping season had low yield due to drought during crop establishment, which caused high incremental water produc vity of I2 and I3 in D1 and D2 of 2008. But incremental water produc vity was low in D2 of 2006 compared with that in D1 under both irrigated water regimes because of low yield in D2 due to crop submergence. Table 14. Incremental water produc vity (kg rice m-3 water) in the aus seasons of 2006 to 2008 as affected by sowing date and irriga on treatment 2006 D1 D2 D3 I22 0.77 0.27 na3 2007 I3 0.31 0.008 0.009 I2 Na Na 1.34 2008 I3 Na Na 1.26 I2 0.43 0.63 1.03 1 D1 = seeding on 5-10 April, D2 = seeding on 15-20 April, D3 = seeding on 25-30 April; 2 I2 = supplementary irriga on, I3 = full irriga on; I3 0.04 0.54 1.16 3 na = not applicable; when yield without irriga on is equal to or higher than that with irriga on, indica ng no increment in yield with irriga on compared to without irriga on. 385 4. Conclusions and recommenda ons This study aimed to develop technical packages for an aus–aman cropping system with stable and high yields in areas of the coastal zone with moderate salinity during the dry season. The interac on between aus seeding date and water management varied from year to year due to varia on in rainfall at the beginning of the rainy season. With a short dura on modern aus variety (BRRI dhan65, ~110 d) and supplementary irriga on of less than 100 mm, yields of 3.7 to 4.2 t/ha of dry seeded aus for sowing dates ranging from mid-April to late April were achieved under favourable condi ons (no submergence, water stagna on or severe pest or disease infesta on). Combining this with a modern, high yielding aman variety resulted in total system yields approaching 9 t/ha/yr, much higher than the 2.5 t/ha/yr of rice commonly obtained by farmers who grow a single rice crop (aman) using local varie es. For systems with late sown (early May) aus, the use of a photoperiod sensi ve aman variety reduced aman and total system yield, therefore the use of non-photoperiod sensi ve varie es is recommended. For good establishment and yield stability of a solely rainfed aus crop, seeding should be delayed un l the end of April. The research also highlighted the problem of poor water management in the polders, leading to submergence and water stagna on at mes in both the aus and aman crops. This points to the need for be er water management at a regional scale, including the separa on of lands of higher and lower eleva on with small levees to prevent accumula on of water in lower lying lands, and for the ability to drain. It also points to the need for be er submergence and water stagna on tolerance in modern aus and aman varie es. In areas without suitable water for supplementary irriga on, seeding of aus can be delayed un l 10 May followed by transplan ng during the first week of June. In most years, this would allow field prepara on taking advantage of rainfall or dal flooding. Late crop establishment, however, would expose the newly established aus crop to higher risk of submergence from high rainfall or flooding from surrounding areas. The combina ons of BRRI dhan65 (aus) + BR11 (aman) and BRRI dhan65 (aus) + BRRI dhan46 (aman) gave the best yields under favourable condi ons. The use of photoperiod-insensi ve varie es (such as BRRI dhan49) in the aman season would give greater flexibility to the aus–aman cropping calendar than would a photoperiod-sensi ve variety such as BRRI dhan49. Acknowledgements The paper presents findings from CGIAR Challenge Program on Water and Food (CPWF), CPWF phase 1 project, PN10 en tled “Coastal Resources Management for Improving Livelihoods". Therefore, the authors are grateful to CGIAR Challenge Program on Water and Food (CPWF) for financial support to carry out research work and a fellowship for the PhD of the first author. We are thankful to our field assistant Mr. Amal Krisna Roy for his con nuous support during the work and the farmers for provision of land for field experiments References American Society of Agronomy (AAS) and Soil Science Society of America (SSSA). 1982. Methods of soil analysis, part 2. AAS and SSSA, Madison, Wisconsin, USA. p 225-260. American Society for Tes ng and Materials (ASTM). 1964. Standard method for grain-size analysis of soil (D422-63) (adopted 1963). In: Nomenclature and defini ons, standard methods, suggested methods. 4th ed. ASTM, Philadelphia, Pa. p 95-106. Bangladesh Rice Research Ins tute (BRRI). 2009a. Annual internal research review report. BRRI, Gazipur, Bangladesh. 386 Bangladesh Rice Research Ins tute (BRRI). 2009b. Adhunik dhaner chas [in Bengali]. BRRI, Gazipur, Bangladesh Cassman, K.G. (ed.). 1994. Breaking the yield barrier: proceedings of a workshop on rice yield poten al in favorable environments, IRRI, 29 November-4 December 1993. Interna onal Rice Research Ins tute, Manila, Philippines. De Da a, S.K. 1981. Principles and prac ces of rice produc on. John Wiley & Sons, USA. 618 p. Food and Agriculture Organiza on (FAO). 1998. Crop evapotranspira on: guidelines for compu ng crop water requirements. FAO Irriga on and Drainage Paper 56. FAO, Rome, Italy. 300 p. Gomez, K.A. 1972. Techniques for field experiment with rice. IRRI, Los Banos, Philippines. 46 p. Haque, A.K. 2006. Salinity problems and crop produc on in coastal regions of Bangladesh. Pakistan Journal of Botany 38(5): 1359-1365. Interna onal Rice Research Ins tute (IRRI). 1994. Soil and plant sampling and measurements. IRRI, Los Banos, Laguna, Philippines. 13 p. Karim, Z., S.G. Hussain, and M. Ahmed. 1990. Salinity problems and crop intensifica on in the coastal regions of Bangladesh. Soils Publica on no. 33. Soils Irriga on Division, BARC, Dhaka, Bangladesh. p 1-22. Michael, A.M. 1997. Irriga on: theory and prac ce. Vikas Publishing House Private Ltd., New Delhi, India. 801 p. Mondal, M.K., Saha, N.K., Ritu, S.P., Paul, P.L.C, Sharifullah, A.K.M., Humphreys, E., Tuong, T.P. and Rashid, M.A. 2015. Op mum sowing window for boro cul va on in the coastal zone of Bangladesh. These proceedings. Mondal, M.K, T.P. Tuong, S.P. Ritu, M.H.K. Choudhury, A.M. Chasi, P.K. Majumder, M.M. Islam, and S.K. Adhikary, 2006. Coastal water resource use for higher produc vity: par cipatory research for increasing cropping intensity in Bangladesh. In: C.T. Hoanh, T.P. Tuong, J.W. Gowing, and B. Hardy (eds.). Environment and livelihood in tropical coastal zones: managing agriculture-fishery-aquaculture conflicts. Comprehensive Assessment of Water Management in Agriculture Series. CABI Interna onal, UK. p 72-85. Mondal, M.K, S.P. Ritu, M.H.K. Choudhury, A.M. Chasi, M.A. Mannan, P.K. Majumder, M.M. Islam, and S.K. Adhikary. 2004. Introduc on to soil and water salinity dynamics of the south-west coastal region of Bangladesh. In: Proceedings of the workshop on coastal water management and uptake technologies, Khulna, Bangladesh, 26 June 2004. p 1-7 Nishat, A. 1988. Review of present ac vi es and state of the art of the coastal areas of Bangladesh. In: Coastal area resource development and management ,part II. Coastal Area Resource Development and Management Associa on (CARDMA), Dhaka, Bangladesh. p 23 – 35. Reddy, M.D. and B.N. Mitra. 1985. Response of rice to different forms of urea and phosphorus fer liza on under intermediate deepwater condi on. Plant and Soil 84(3): 431-435. Ritu, S. P. 2012.Crop and water management for increasing rice cropping intensity and produc vity of the coastal region of Bangladesh. Unpublished PhD thesis. Submi ed to Bangladesh Agricultural University. November 2010. Soil Resource Development Ins tute (SRDI). 2004. Land use map of Bangladesh. SRDI, Ministry of Agriculture, Dhaka, Bangladesh Sudhir-Yadav, Gurjeet Gill , E.Humphreys, S.S.Kukul, and U.S.Walia(2011). Effectof water management on dry seeded and puddled transplanted rice Part 1: Crop performance. Field Crops Research 120, 112-122. 387 Tuong, T.P. 1999. Produc ve water use in rice produc on: opportuni es and limita ons. Journal of Crop Produc on 2(2): 241-264. Tuong, T.P. and B.A.M. Bouman. 2003. Rice produc on in water-scarce environments. In: Kline J.W., Barker R., and Molden D. (eds.). Water produc vity in agriculture: limits and opportuni es for improvement. CABI Interna onal, Wallingford, UK. p 53-67. Widawsky, D.A. and J.C. O’Toole. 1990. Priori zing the rice biotechnology research agenda for Eastern India. The Rockfeller Founda on. 388 Rice-sunflower: An alterna ve cropping system for sustained livelihoods in the coastal zone of Bangladesh M. Afsar 1 and T.H. Miah2 1 Exim Bank Agricultural University Bangladesh, Bangladesh, mahnazafsar@yahoo.com 2 Bangladesh Agricultural University, Bangladesh, tofazzal_miah@yahoo.com Abstract Although a large number of crops and vegetables are grown in Bangladesh the country is not yet self-sufficient in food grain produc on. Despite the country’s efforts to increase agricultural produc on, feeding a huge popula on of 150.7 million (Hossain 2014) from cul vable land that is con nuously decreasing in size due to infrastructural development and environmental challenges is the most cri cal issue in Bangladesh. Promo ng agricultural growth is a cri cal element of the government’s strategy aimed at food security and improving the household incomes of rural people. This study assesses the profitability of a modern rice-sunflower cropping pa ern over a tradi onal rice-sesame pa ern and the causes of favoring the new cropping pa ern in Ba aghata Upazila in the Khulna District of Bangladesh. Profitability is considered in terms of income and food security. The analysis is anchored on a sample of 200 farmers; 100 farmers from each of the pa erns were selected. Ac vity budgets, a logit model, and a food security index were employed to achieve the main objec ves. It is evident from the study that the socio-demographic characteris cs of rice-sesame and rice-sunflower farmers differ from each other. The farmers who adopted sunflower in their exis ng pa ern were more experienced as well as older than the rice-sesame farmers. Educa onal status of the two groups was more or less similar and agriculture was the main occupa on of both categories of the farmers; but the par cipa on of the sunflower farmers in off-farm ac vi es was higher than that of sesame farmers. The adopters of the modern rice-sunflower cropping pa ern earned much higher profits than tradi onal rice-sesame growers. In addi on, the average daily per capita calorie intake was rela vely higher for the rice-sunflower adopters than the rice-sesame growers. Finally, this study iden fied some crucial problems in the expansion of sunflower cul va on in coastal Bangladesh and suggests probable solu ons. It is concluded that a modern rice-sunflower pa ern can play a significant role in achieving food security and improving the household income of coastal farmers in the Khulna District of Bangladesh. Key message: The study confirmed that sunflower is a profitable enterprise for farmers in coastal zone of Bangladesh. The farmers in the coastal zone can therefore produce sunflower for be er financial return and improved livelihoods and food security. Keywords: ac vity budgets, net return, food security, sunflower, coastal Bangladesh 1. Introduc on Improved seeds and new cropping pa erns that make use of high yielding varie es of crops are essen al for ensuring the food security of the people of Bangladesh. The economy of Bangladesh largely depends on agriculture. Agriculture has direct linkages to food security through the poten al to supply food grains for consump on. New cropping pa erns with higher yields than exis ng cropping pa ern have the poten al to greatly impact food security, as they enhance produc on and thereby provide greater access to food via both produc on and trade. Bangladesh has achieved near self-sufficiency in the produc on of rice, a staple food in the country. However, self-sufficiency in other food items has yet to be achieved. The poten al of minor crops in Bangladesh has not been fully exploited for contribu on to the goal of sustainable food security. Their produc on and u liza on is s ll very low. Under these circumstances, oilseed crops may be considered as minor crops but are 389 nevertheless important from a nutri onal point of view. Oilseed crop cul va on can also help to achieve important subs tu on by reducing imports of edible oils from abroad. Sunflower has proved to be a rela vely good oilseed crop. The cul va on of sunflower began in Bangladesh in 1975 as a garden plant and has gradually gained popularity as a field crop in the coastal area of Bangladesh (Miah 2014). Fluctua on in economic well-being caused by periodic shocks is a common phenomenon, but its effects are especially pernicious in the coastal zone of Bangladesh (Osmani and Ahmed 2013). In the coastal zone, most lands remain fallow in the rabi season due to a shortage of irriga on water and/or soil salinity constraints. Tropical coastal deltas present one of the most challenging planning and management se ngs given their diverse character and loca on at the land-water interface. The most prominent issue in the recent evolu on of tropical delta systems has been the widespread expansion of shrimp aquaculture, first in Thailand and then throughout the coastal areas of East Asia, Southeast Asia and South Asia (Chuenpagdee and Pauly 2004). This shrimp farming has led to the drama c transforma on of coastal land use and subsequent environmental impacts that include: loss of mangrove habitats, water pollu on, land saliniza on and declining fisheries (Primevera 1997; and Talaue-McManus 2006). In Bangladesh, shrimp farming is profitable from the viewpoint of individual farmers but it is undesirable from the viewpoint of society (Fatema 2012). The cropping pa ern followed in this area is mainly sesame-T. Aman rice. Sesame is a summer crop and is highly suscep ble to water logging. The CSISA-CIMMYT Khulna hub has taken an ini a ve to promote sunflower produc on in this area in collabora on with BARI (Bangladesh Agricultural Research Ins tute) and Khulna University. BRAC (Bangladesh Rural Advancement Commi ee) and some other Non-Government Organiza ons have also tried to introduce a rice-sunflower pa ern. However, the impact of sunflower cul va on on household income and food security is not known due to a lack of available data. Surprisingly only limited literature related to sunflower is available in Bangladesh. No research has yet been conducted to iden fy the role of rice-sunflower cropping pa erns in improving the household income and food security of farmers in the polder area. This study was therefore designed to provide detailed informa on regarding sunflower produc on. Specifically, the study was designed to achieve the following objec ves: to iden fy the socioeconomic backgrounds of rice-sesame and rice-sunflower farmers to es mate the profitability of a rice-sunflower cropping pa ern in terms of household income and food security to assess the causes of favoring the new cropping pa ern to iden fy some problems in the expansion of sunflower cul va on and probable solu ons 2. Research methods 2.1 Sample and data sources This study is based on primary data collected from selected respondents of five villages, namely Amtola, Baruiabad, Titukhali, Debitola, and Boyervanga in Ba aghata Upazila of Khulna District, Bangladesh. A simple random sampling technique was followed for selec ng tradi onal rice-sesame farmers while purposive sampling was employed for selec ng farmers of the new rice-sunflower cropping pa ern. In total 200 farmers, 100 from each of the selected cropping pa erns, were selected for the study. 390 Table 1. Distribu on of sampled farmers Selected villages Amtola Baruiabad Titukhali Debitola Boyervanga Total Number of ricesesame farmers 20 20 20 20 20 100 Number of ricesunflower farmers 20 20 20 20 20 100 Total sample size 40 40 40 40 40 200 Source: adapted from Afsar (2013) The required data were collected in August and September 2012 using a structured interview schedule. Secondary data were also collected from the government and research reports, online materials and periodicals. Descrip ve sta s cs were employed to examine the objec ves (i), (iii) and (iv) whereas an ac vity budget, food security index (Babatunde et al. 2007) and logit model (Babatunde et al. 2007) were used for analyzing objec ve (ii). These specific methods are explained in the following sub-sec ons. 2.2 Ac vity budget Ac vity budgets were calculated for each of the concerned crops. The algebraic equa ons employed to prepare ac vity budgets is as follows: π = TR – TC Or, π = TR – (VC + FC) In these equa ons, π stands for net return, TR is the total return from the sale of produce, TC is the total cost required for the produc on of the concerned crop. This total cost is also the sum of the variable costs (VC) and the fix costs (FC). All these variables are calculated in Bangladesh Taka (BDT) per hectare. 2.3 Food security index and logit model Hoddino (1995) outlines four ways of measuring food security outcomes, namely individual intake, household caloric acquisi on, dietary diversity and indices of household coping strategies. Logis c regression was used by Kidane et al. (2005) and Feleke et al. (2005) to assess the causes of household food insecurity. Babatunde et al. (2007) examined the factors influencing the food security status of rural farming households in Kwara State of Nigeria. Using the calorie intake approach this study revealed that 36 and 64% of the households were food secure and food insecure, respec vely. Nasrin (2011) and Mannaf (2012) checked household food security against a recommended minimum calorie requirement (i.e., 2122 kilocalories per person per day). In the present case, we follow a similar approach where we first measure household food security and then iden fy the causes. In order to measure food security for each household in the sample, we derived a food security index from the daily per capita caloric intake of each household compared to the recommended per capita daily caloric intake. When the caloric intake for the household is greater than or equal to the recommended intake, the household is considered to be food secure and the household food security status Zi takes the value 1. If the intake is lower than the recommended requirement, then the household is considered to be food insecure and Zi is equal to 0. The food security index calculated for each sampled household is then used as the dependent variable of a logit model. The logit model es mated aims to iden fy the determinants of food security and thus, to assess 391 the role of the new rice-sunflower cropping pa ern in improving household food security of farmers in the coastal zone. The implicit form of the model es mated was as follows: Zi = βXi + Ui Where Zi is the food security status of i th household as previously defined Xi is a vector of explanatory variables and Ui is the error term β is the vector of the es mated parameters of interest. The explanatory variables considered in the vector Xi include the size of the household, the age of the household head, the size of the farm (in hectares), the produc on per capita (in BDT), the household income (in BDT), the food expenditures (in BDT) and a dummy equal to 1 if the household is involved in off-farm ac vi es. 3. Results and discussion 3.1 Socioeconomic characteris cs of the farmers First, it is essen al to know the socio-demographic profile of the sample farmers to get a complete picture of rice-sunflower produc on and its impact on income and food security. 3.1.1 Farmer age Table 2 indicates that the highest percentage of selected rice-sesame farmers (32.00%) belonged to the age group of 30 to 40 years while the highest percentage of rice-sunflower farmers (35.00%) belonged to the age group of 50 to 60 years. Table 2. Age distribu on of selected farmers Age group (Years) 20.00 - 30.00 30.01- 40.00 40.01-50.00 50.01- 60.00 Above 60.00 Total Average age (Years) Rice-sesame farmer Number of respondents % of total 9 9.00 32 32.00 29 29.00 20 20.00 10 10.00 100 100.00 45.92 Rice-sunflower farmer Number of respondents % of total 6 6.00 20 20.00 29 29.00 35 35.00 10 10.00 100 100.00 49.2 Source: adapted from Afsar (2013) So it can be stated that the farmers who have adopted sunflower into their cropping pa erns are slightly older and more experienced in farming than rice-sesame farmers. 3.1.2 Farmer occupa on The farmers were engaged in various occupa ons. Agriculture was the main occupa on of both categories of farmers. Some were also engaged in small business, service and even labor selling (Table 3). 392 Table 3. Distribu on of sample farmers according to their main occupa ons Types of occupa on Agriculture Business Labor selling Service Total Rice-sesame farmers Number of respondents % of total 81 81.0 5 5.0 8 8.0 6 6.0 100 100 Rice-sunflower farmers Number of respondents % of total 80 80.0 10 10.0 4 4.0 6 6.0 100 100 Source: adapted from Afsar (2013) The par cipa on of the sunflower farmers in other income genera ng, off-farm ac vi es, especially business, was rela vely higher than that of sesame farmers. 3.1.3. Farmer land use pa erns Table 4 indicates that the average land holdings of rice-sesame and rice-sunflower farmers were 1.061 and 1.360 hectares, respec vely. From these areas, crops (including rice, vegetables, sesame, sunflower and Golda/Bagda) accounted for 0.918 and 1.181 hectares, respec vely, for each group. The areas under sesame produc on were 0.793 hectares by the rice-sesame farmers and 1.006 by the rice-sunflower farmers. The area under sunflower produc on by the rice-sunflower farmers was 0.131 hectare, which was 8.98% of total cul vated land (Table 4). Table 4. Land ownership pa erns of the selected farmers Land type Homestead area Area under crops Area under ponds Total owned area Area under sesame produc on (ha) Area under sunflower produc on (ha) Rice-sesame farmers Area (ha) % of total 0.082 7.74 0.918 86.50 0.061 5.75 1.061 100 0.793 (73.99) - Rice-sunflower farmers Area (ha) % of total 0.095 7.01 1.181 86.82 0.083 6.16 1.360 100 1.006 (69.17) 0.131 (8.98) Note: Figures within the parenthesis indicate percentages of land under sesame and sunflower cul va on out of total cul vable land. Source: adapted from Afsar (2013) The average size of rice-sunflower farmers’ land holdings (1.36 hectares) was much larger than that of rice-sesame farmers (1.06 hectares). This implies rela vely larger-scale farmers (in terms of cul vable land area) adopt sunflower. 3.1.4 Farmer sources of household income The highest contribu on to household income comes from rice farming. It contributed to 21.05% of the total income for rice-sesame farmers. However, for rice-sunflower farmers, 33.55% of income comes from sunflower cul va on and 12.93% from rice. 393 Table 5. Household annual income of selected farmers Sources of household income Rice Sunflower Sesame farming Vegetables produc on Livestock farming Shrimp culture White fish culture Small business Service Labor selling Total Rice-sesame farmer Amount (Tk/year) % of total 20,968.00 21.05 18,866.00 18.95 10,163.00 10.21 7666.00 7.70 18,940.00 19.02 9346.00 9.38 3247.00 3.26 5850.00 5.88 4520.00 4.54 99,566.00 100.00 Rice-sunflower farmer Amount (Tk/year) % of total 20,968.00 12.93 54,417.00 33.55 17,595.00 10.84 9000.00 5.54 6830.00 4.21 19535.00 12.04 16346.00 10.08 8500.00 5.24 6700.00 4.13 2300.00 1.42 162,191.00 100.00 Source: adapted from Afsar (2013) It can therefore be concluded that sunflower and rice farming were the most significant contributors to the household income of rice-sunflower farmers. The average annual income of rice-sunflower farmers was much higher than rice-sesame farmers. From the above-men oned discussions it can be cau ously concluded that the major socioeconomic characteris cs of rice-sesame and rice-sunflower farmers differ from each other. The farmers who adopted sunflower were in rela vely be er economic condi on than rice-sesame farmers. As a result they can more easily purchase seed and bear the produc on costs of sunflower. This also means that not all farmers can afford the high cost of produc on associated with sunflower. 394 395 3.2 Farmer profit 3.2.1 Economics of rice, sesame and sunflower farming For any agricultural produc on, the profit of the concerned crops plays a significant role in the produc on decisions of farmers. To assess the role of a rice-sunflower cropping pa ern in improving household income and food security it is essen al to know the rela ve profitability of rice, sesame and sunflower. The average cost of T. Aman rice seedlings was Tk 3627.00 per hectare, which accounted for 10.28% of total cost (Table 6). The costs of seed per hectare of sesame and sunflower were Tk 594.00 and Tk 10,600.00, respec vely, accoun ng for 2.53% and 21.60% of total costs, respec vely (Table 6). The cost of sunflower seeds was much higher than those for seasame or rice. In the study area, power llers were used three mes on average for land prepara on at a rate of Tk 8.00/decimal for each llage opera on. Per hectare costs of power ller use for land prepara on of T. Aman rice, sesame and sunflower were es mated at Tk 5928.00 (Table 6). In the study area, average wage rate was Tk 300.00 per man-day during the study period. It can be seen from Table 6 that Aman rice growers used 69 man-days per hectare, accoun ng for 58.69% of the total gross costs of T. Aman produc on. Table 6 indicates that sesame and sunflower growers used 55 and 75 man-days per hectare for human labor, respec vely. Sunflower should be planted in rows with row-to-row distance of 50 cm, giving 30cm plant-to-plant distance in a row; this requirement may explain why sunflower cul va on is labor intensive. Since sunflower farmers have to apply fer lizers in their fields and the sowing method of sunflower is different from that of sesame, human labor costs for sunflower cul va on are much higher than sesame. An op mum dose of fer lizer is a major requirement for produc on but in the study area most farmers did not use any type of fer lizer during sesame cul va on, which is followed by rice cul va on. The sample farmers used three kinds of fer lizers, namely, Urea, Triple Super Phosphate (TSP) and Muriate of Potash (MOP) for T. Aman rice and sunflower. Total fer lizer costs of T. Aman and sunflower accounted for 9.50 and 11.51% of total costs, respec vely (Table 6). The selected farmers did not use any insec cide for sesame cul va on while rice-sunflower farmers used it to protect their crops from pest a ack. Per hectare insec cide cost was Tk 1134.00 for sunflower, or 2.31% of total gross costs and Tk 1087.00 for T. Aman rice, or 3.08 % of total gross costs. Finally, irriga on is considered one of the most important inputs of sunflower produc on. Rice and sesame does not require irriga on. Per hectare cost of irriga on for sunflower produc on was Tk 2465.00, which cons tuted 5.02% of total costs (Table 6). The cost of the opera ng capital was es mated at Tk 578.00 per hectare, or 1.64% of total cost of T. Aman rice produc on, and Tk 480.00 and Tk 805.00 for sesame and sunflower farming, respec vely (Table 6). 3.2.2 Selected cropping pa ern profitability In total, per hectare costs of T. Aman rice, sesame and sunflower produc on were es mated at Tk 35,270.00, Tk 23,502.00 and Tk 49,084.00, respec vely (Table 6). Per hectare yield of T. Aman rice was found to be 3,046.00 kg with a per unit price of Tk 17.00 per kg. The gross return of T. Aman was Tk 56,238.00 per hectare (Table 6). Table 6 shows that gross returns per hectare for sesame and sunflower farming were Tk 41,888.00 and Tk 103,501.00, respec vely. Per hectare net returns of T. Aman rice, sesame and sunflower were Tk 20,968.00, Tk 18,386.00 and Tk 54,417.00, respec vely (Table 6). 396 These results indicate that produc on of T. Aman rice, sesame and sunflower were profitable for individual farmers, but with some large differences in profitability among the selected crops. The addi onal income from sunflower produc on contributed significantly to the total household income of sunflower growers. Sesame and sunflower are both profitable crops, but sunflower is more profitable than sesame. Table 7. Comparison of profitability following different cropping pa erns Cropping pa ern Rice - sesame Rice - sunflower Difference Net return (Tk/ha/yr) 39354.00 75385.00 36031.00 Source: adapted from Afsar (2013) Table 7 shows that there is a significant difference in profitability between the two categories of farmers; the rice-sunflower farmers earned much higher profits than the rice-sesame farmers. The main reason was, as men oned earlier, that per hectare yield of sunflower was much higher in the polder area than tradi onal sesame crops. In fact, rice-sunflower farmers received Tk 36,031.00/ha more profit than rice-sesame farmers. If farmers replace sesame with sunflower they can obtain an addi onal Tk 36,031.00 from the same hectare of land. The role of sunflower cul va on for improving farm household income is highly encouraging and likely significant. 3.3 Farmer food security 3.3.1 Household nutri onal status Food security can be viewed from several perspec ves, such as availability of food, access to safe and nutri ous food, and u liza on of food. Household calorie availability was es mated using a food nutrient composi on, which is presented in appendix 2. Table 8 shows the per capita daily calorie intake from different food items by the households. Table 8. Calorie intake from different food items by family members of households under two cropping pa erns Food items Rice Fish Egg Pulses Vegetables Meat Milk Dry fish Soybean oil Sunflower oil Spices and other Total Difference in per capita daily calorie intake Rice – sesame (Kcal/day/capita) 1622.43 40.54 13.26 66.50 98.55 3.72 1.08 2.12 163.65 34.94 2046.80 226.57 Rice – sunflower (Kcal/day/capita) 1728.12 48.26 13.81 81.67 103.16 32.50 6.93 2.21 59.72 161.52 35.47 2273.37 Source: adapted from Afsar (2013) 397 Average per capita calorie intake by households prac cing a rice-sesame cropping pa ern was es mated at 2046.80 kcal as shown in table 8, which is rela vely lower than the recommended daily calorie intake of 2122 kcal per day. Thus, these households lived below the food poverty line and were food insecure from a nutri onal availability viewpoint. On the other hand, average per capita calorie intake was rela vely higher for households prac cing a rice-sunflower cropping pa ern, at 2273.37 kcal. This was above the recommended daily calorie intake. Thus, these farmers were, on an average, food secure. 3.3.2 Logit model and food security In order to assess the role of a rice-sunflower cropping pa ern and other determinants on farm household food security, a logit model was es mated. As men oned earlier, seven explanatory variables were iden fied to be major determinants of food security in this study. The es mated parameters of the logit regression model are presented in Table 9. Table 9. Es ma on of the logis c regression of determinants of the food security status of farming households Variable Coefficient Constant -3.837 Household size -0.726 Age of household head 0.011 Farm size 0.586 Per capita produc on 0.003 Income 0.002 Involvement in off-farm ac vi es 0.920 Food expenditures 0.002 Standard Error 1.533 0.320 0.019 0.851 0.001 0.001 0.509 0.001 Level of Significance 0.012 0.023* 0.544 0.492 0.052** 0.005* 0.070*** 0.003* Exponen al of coefficient or odds ra o 0.022 0.484 1.011 1.796 1.003 1.002 2.510 1.002 Note: * indicates significant at 1% level, ** indicates significant at 5% level, *** indicates significant at 10% level Source: adapted from Afsar (2013) Household size has a nega ve coefficient, which was significant at 1% level. The household size coefficient of -0.726 means there is a nega ve rela onship between household size and food security. An odds ra o of 0.484 means a unit increase in household size will reduce the probability of a household to be food secure by 0.484. Per capita aggregate produc on was posi ve and significant at 5% level. A unit increase in per capita produc on will consequently increase the probability of household to be, food secure by 1.003. Household income was posi ve and significant at 1% level. This indicates that the higher the household income, the higher the probability that the household will be food secure. A unit increase in the level of income will increase the probability of a household to be, food secure by 1.002. The household income of rice-sunflower farmers was much higher and 46.60% of total income came from rice and sunflower cul va on, which was much higher than the percentage share of other sources of income (Table 5). Table 9 shows that involvement in off-farm ac vi es is posi ve and significant at 10%. This indicates that households that were engaged in non-farm ac vi es were nearly 2.51 mes more likely to be food secure than those households that were not engaged in off-farm ac vi es, other things remaining the same. Finally, food expenditures have a low but posi ve coefficient that was significant at 1% level. A unit increase in food expenditures increase the probability of household to be food secure by 1.002. 398 3.3.3 Food security index On average, the food security index for rice-sesame farmers was 1.05 when they were food secure whereas the value of this index for food insecure households from this group was 0.85. Average caloric intake of food insecure households was 1806.15 kcal, lower than the na onal average calorie intake (Table 10). On the other hand, Table 10 shows that farmers who cul vated sunflower in the study area could be classified as food secure since their average food security index was 1 and on average they were able to meet the recommended daily per capita calorie requirement of 2,122 kcal. From the above-men oned discussion, it can be concluded that food security was much more common in the sunflower farmers’ households than among the sesame farmers’ households. Table 10. Food security indicators for farm households under two cropping pa erns Rice-sesame Food security indices Food secure Food insecure All households households Food security index 1.05 0.85 0.95 Per capita daily calorie 2236.97 1806.15 2046.80 avail-ability (kcal) Rice-sunflower Food secure Food insecure All households households 1.07 0.90 1.00 2271.52 1919.18 2273.37 Source: adapted from Afsar (2013) 3.4 Causes for favoring rice-sunflower cropping pa ern A er realizing the profitability of sunflower farming many farmers are willing to choose the rice-sunflower cropping pa ern. Indeed, the current adop on rate of sunflower is, perhaps, far be er than ever before. Appendix Table 1 shows that about 25.00% and 35.00% of farmers were in favor of sunflower and in favor of both sesame and sunflower, respec vely. Table 11 shows the ranking of reasons for adop ng a rice-sunflower cropping pa ern. About 95.00% of the farmers stated that sunflower cul va on was profitable for them. About 80.00, 50.00, and 40.00% of the sunflower farmers argued that they would grow sunflower for their own consump on of edible oil, u liza on of cul vable land and u liza on of family labor, respec vely. Fi een percent of all farmers expressed that they would produce sunflower because it might be harvested earlier and they could consequently move to rice cul va on a er sunflower harvest. Table 11. Reasons for choosing sunflower cul va on Reasons For earning more profit For household consump on of edible oil For be er u liza on of cul vable land For proper u liza on of family labor Early harves ng of sunflower provides an opportunity for growing next crop and hence earning more profits Number of respondents 114 96 60 48 18 % of total 95.00 80.00 50.00 40.00 15.00 Source: adapted from Afsar (2013) Some farmers do not cul vate sunflower due to the high price of its seeds and other inputs like hired labor and fer lizers, the scarcity of financial capital and lack of awareness about the cul va on process and the poten al benefits from sunflower cul va on. These were perhaps some of the important reasons for not adop ng a rice-sunflower cropping pa ern in the study area. 399 3.5 Problems associated with sunflower farming and probable solu ons Sunflower farmers in the study area have been facing some crucial problems that may explain the low rate of adop on despite the poten al of sunflower. These issues are briefly discussed below. 3.5.1 Economic and technical problems Lack of knowledge about cultural prac ces. As sunflower was a newly introduced crop in the study area, 55.00% of farmers (Table 12) reported that due to lack of knowledge about cul va on prac ces they failed to take proper care of the crop and thus they did not gain the expected per hectare yield. Inadequate extension services. Although the selected farmers were producing the sunflower crop that was supplied by BRAC, it was alleged that their extension service was not sufficient for serving the purpose of individual farmers in the study area. Forty-three percent of sunflower growers reported that they were not ge ng adequate extension services (Table 12). Lack of ins tu onal credit facili es. As confirmed by this study, the cost of sunflower cul va on is rela vely higher than the cul va on costs of other crops. Consequently, sunflower requires substan al amount of cash to purchase recommended doses of inputs such as seeds, fer lizers and irriga on water. Many farmers lacked the money to purchase the relevant inputs and it can be seen from Table 12 that 37.00% of sunflower growers men oned that they lacked access to credit during crop cul va on when they would have needed it. Inadequate supply of quality seeds. Availability of improved and quality seeds was another limi ng factor in sunflower produc on, as reported by 84.00% of sunflower growers (Table 12). Higher price of fer lizer, insec cides and water. About 76.00% of the sunflower growers sampled reported that they had to purchase fer lizer and insec cides from the market that they claimed were a higher price than the normal price (Table 12). Parrot and squirrel a acks. Parrots and squirrels eat the seeds of sunflowers and cause great damage to yield. Seventy percent of sunflower growers reported this problem (Table 12). A ack by pest and disease. Few incidences of pest and disease a acks were noted in the produc on of sunflower crop. Only 10.00% of the farmers referenced this problem (Table 12). Flower plucking. Sunflower is not only recognized as an oilseed crop but is also well known as an ornamental crop to both children and adults. People are fascinated by its lovely color and o en pluck the flower. It can be seen from Table 12 that 51.00% of the sunflower farmers sampled reported this problem. 400 Table 12. Problems associated with sunflower farming Problems and Constraints (a) Economic and technical problems i. Lack of knowledge about cultural prac ces ii. Inadequate extension services iii. Lack of ins tu onal credit facili es iv. Inadequate supply of quality seeds v. Higher price of fer lizers, insec cides and water vi. Parrot and squirrel a acks vii. A ack by pest and disease viii. Flower plucking (b) Marke ng problems i. Inadequate demand ii. Lack of proper marke ng channel (c) Problems rela ng to consump on i. Not habituated to eat regularly Number of respondents % of total 55 43 37 84 76 70 10 51 55.00 43.00 37.00 84.00 76.00 70.00 10.00 51.00 67 75 67.00 75.00 43 43.00 Source: adapted from Afsar (2013) 3.5.2 Marke ng problems Inadequate demand. There was no organized market for sunflower seeds and oil in the study area. From Table 13, it can be seen that 67.00% of growers iden fied inadequate demand for sunflower in a local market as a serious problems. Lack of proper marke ng channel. Lack of sunflower oil millers for oil extrac on and lack of proper marke ng channels were iden fied as constraints to adop on. About 75.00% of farmers reported that the marke ng facili es were very poor for sunflower (Table 13). 3.5.3 Problems related to consump on Not habituated to eat regularly. Consump on of sunflower was also an important obstacle. Indeed, 43.00% of sunflower growers stated that they could not sell their product in the local market because the people were not accustomed to regularly consuming sunflower oil. 3.5.4 Sugges ons for improving sunflower produc on The selected farmers provided the following sugges ons for overall improvement of sunflower produc on: Farmers need training about scien fic methods of sunflower cul va on so that they can take ini a ve to reduce the cost of produc on. Ins tu onal credit should be made available on easy terms and condi ons to the concerned producers. Supply of an adequate quan ty of quality seeds should be ensured at the right me in the study area. Fair prices for quality inputs and reasonable price for output should be ensured to farmers. Parrot is the main threat to sunflower cul va on and parrots ate sunflower seeds during sowing and harves ng mes. Extension personnel should work immediately to solve this problem. 401 Widespread extension work is needed to popularize the benefits of produc on and consump on of sunflowers. Farmers should make use of the sunflower byproducts to recoup a por on of their produc on costs. For example, they can use the plant as a house building material similar to straw, as food for domes c animals or as organic manure for the field. If they can sell it in the market (even at a small price) their profit from sunflower cul va on will be rela vely higher. 4. Conclusion and recommenda ons The study confirms that both the rice-sesame and rice-sunflower cropping pa erns are profitable. Although both rice-sesame and rice-sunflower farmers earned profits, produc vity varied amongst the farmers as reflected by the study’s profitability analysis. Households following a rice-sunflower cropping pa ern have higher incomes and benefit from a be er food security status than households cul va ng rice and sesame. There are bright prospects for the expansion of sunflower due to its high nutri onal value and a rising demand elsewhere in the country. Therefore, the following policy recommenda ons have been made for increasing the produc on and financial returns of sunflower produc on: There should be a marke ng policy so that farmers can get fair prices for their products. We no ced that there was great varia on in input use and other farm prac ces in the study areas. Most farmers did not follow the recommended doses of inputs. Extension work should be geared up in the study area so that farmers can improve their knowledge about recommended doses of the inputs used for sunflower cul va on. Seed preserva on techniques should be developed so that farmers can preserve their own seeds and thereby reduce seed cost. Apart from the above-men oned specialized policies for rice and sunflower farmers, some common policies could also be undertaken for the greater interest of farmers and be er management of Polder 30. Specifically, efficient water management and a producers’ coopera ve may solve many of the problems of growers of rice and sunflower in the study area. The concerned authority may formulate the relevant rules and regula ons in this regard and thus, posi ve steps could be taken for the food and nutri on security of the people of the coastal region of Bangladesh. Acknowledgements The authors express their deep gra tude and hear elt indebtedness to Dr Adi Mukherji, Dr Marie Charlo e-Buisson, Interna onal Water Management Ins tute (IWMI), New Delhi, India and to Dr Manoranjan K Mondal, Interna onal Rice Research ins tute (IRRI), Dhaka, Bangladesh for their valuable sugges ons for the comple on of this study. The authors are grateful to IWMI and CPWF for providing the funds for conduc ng this study. The ar cle is based on the MS thesis of the first author, Mahnaz Afsar, submi ed to the Department of Agricultural Economics, Bangladesh Agricultural University (BAU), Mymensingh, Bangladesh in 2013. References Afsar, M. 2013. Impact of rice-sunflower cropping pa ern on income and food security of farm households in Polder 30 of Ba ghata Upazila in Khulna district of Bangladesh. MS Ag Econ thesis, Bangladesh Agricultural University, Mymensingh. Babatunde, R.O., O.A. Omotesho and O.S. Sholotan. 2007. Factor influencing food security status of rural farming households in North Central Nigeria. Agricultural Journal 2(3): 351-57. BBS 2010. Household Income and Expenditure Survey 2010, Bangladesh Bureau of Sta s cs, Ministry of Planning, Dhaka. 402 Chuenpagdee, R. and D. Pauly, 2004. Improving the state of coastal areas in the Asia-Pacific region, Coastal Management 32:3-15. Fatema, K. 2012. An economic inves ga on into dilemma of rice versus shrimp farming in some selected areas of Dacope Upazila (polder no. 30) under Khulna district of Bangladesh. MS Ag Econ thesis, Bangladesh Agricultural University, Mymensingh. Feleke, S. T., R. L. Kilmer and C. H. Gladwin. 2005. Determinants of food security in Sothern Ethiopia at household level. Agricultural Economics Journal 33:351-63. Hoddino , J. 1995. Opera onalizing Household Security and Development Strategies. Interna onal Food Security Policy Research Ins tute (IFPRI), Technical Guideline No. 1, Washington, USA. Hossain, M. 2014. Impact of Price Hike on Food Security in Bangladesh, Lap Lambert Academic Publishing, Deutschland, Germany. Kidane, H., Z. G. Alemu and G. Kundhlande. 2005). Causes of household food security in Koredegaga Peasant Associa on, Oromiya Zone, Ethiopia’, Agricultural Economics Journal 4:543-60. Mannaf, M. 2012. An economic study on maize produc on and its impact on food security in selected areas of Bogra district. MS Ag Econ thesis, Bangladesh Agricultural University, Mymensingh. Miah, T.H. 2014. Economic analyses of on-farm and off-farm income genera ng ac vi es in the project areas of the Coastal Livelihoods Adapta on Project (CLAP). A report prepared for GIZ-CLAP, Dhaka, Bangladesh. Nasrin, M. 2011. Land tenure system and food security in some selected areas of Mymensingh district. MS Ag Econ thesis, Bangladesh Agricultural University, Mymensingh. Osmani, S.R. and M. Ahmed. 2013. Vulnerability to shocks and Coping Strategies in Rural Bangladesh, Working Paper No. 21, Ins tute of Microfinance, Dhaka. Primavera, J.H. 1997. Socio-Economic Impacts of Shrimp Culture. Aquaculture Research 28: 815-27. Talaue-McManus, L. 2006. Pressures on rural coasts in the Asia- Pacific region in Harvey, N. (ed.). Global Change and Integrated Coastal Management: The Asia- Pacific region Springer, Dordrecht, the Netherlands, 197-229. WFP (1988), ‘Agricultural Produc on & Na onal Food Balance’, Bangladesh, Dhaka. 403 Appendix Table 1 - Nutrient Composi on Items Coarse Rice A a (Wheat) Dal Len l Fish Rohi Telapia Mrigal Pangas Sarpu . Energy (per 100 Sol) kcal 365 341 344 343 159 120.55 127.50 98 170.23 161 Items Meat Beef Mu on Chicken Duck Vegetables Potato Brinjal Helencha shalk Data shak Energy (per 100 Sol) kcal 146 136.40 194 125.29 130 48 80.20 42 41 28.66 Egg Milk (Cows) Oil(Soybean) Spices Onion Garlic Chili Turmeric Dry fish Hen Energy (per 100 Sol) kcal 179 70 900 146 50 145 40.71 349 279.79 153 Number of respondents 80 50 70 200 % of total 40.00 25.00 35.00 100.00 Items Note: 1 medium egg is 63 to 73 grams (average 68 grams). Source: WFP (1988) Appendix Table 2 - Farmers’ willingness toward crop produc on Farmers’ opinions regarding crop cul va on In favor of sesame cul va on In favor of sunflower cul va on In favor of both sesame and sunflower cul va on Total Source: adapted from Afsar (2013) 404 Oilseed crops in rice-based cropping systems in southern Bangladesh M.H. Rashid1* , F. Hossain2, D.K. Nath2, P.C. Sarker2, A.K.M. Ferdous2 and T. Russell2 1 2 Bangladesh Rice Research Ins tute, Bangladesh, hrashid67@yahoo.com Interna onal Rice Research Ins tute, Bangladesh, f.hossain@irri.org, d.nath@irri.org, p.sarker@irri.org, a.ferdous@irri.org, t.russell@irri.org Abstract The agro-ecological condi ons and tradi onal rice-based farming prac ces of southern Bangladesh have resulted in lower crop diversity and cropping intensity than in other parts of Bangladesh. The hypothesis of the work here was that there is good scope to intensify the exis ng aman-boro cropping system (prac ced in more favorable areas) through replacement of the long or medium dura on rainy season rice varie es with short dura on, high yielding varie es and the inclusion of oil seed crops between the rice crops. In the less favorable tradi onal aman-fallow areas, the hypothesis was that there is good scope to intensify produc on and increase income through growing sunflower. However, a major constraint to mely establishment of crops a er aman harvest is that soils are too wet for llage. Therefore, field trials and farmer par cipatory demonstra ons on evalua on of short dura on BARI mustard varie es, replacement of farmers’ aman varie es with short dura on aman varie es, inclusion of oilseeds in the cropping sequence, and oil seed crop establishment methods were conducted. High yielding mustard varie es were validated against farmers’ local varie es in aman-mustard-jute and aman-mustard cropping sequences. Establishment of oil seeds with no llage (zero ll) and llage were compared. For mustard, this involved relay sowing by broadcas ng on the wet soil surface prior to aman harvest, or llage a er harvest and prior to seeding. Sunflower was established a er aman harvest by dibbling into non- lled soil, or following conven onal llage, soil moisture at maturity of the aman rice. Early maturing varie es of aman produced similar yields to farmers’ varie es, and allowed for the inclusion of short dura on oil seed mustard without affec ng boro rice produc on. The yield of the short dura on BARI mustard varie es was higher than that of the farmers’ variety. In medium salinity soils, sunflower, a salt tolerant crop, produced yields of 1.9 to 2.6 t ha -1, compared with yields of 2.4 to 3.1 t ha-1 in low salinity soils. Yield of sunflower declined as sowing was delayed beyond 5 December and dibbling into non- lled soil allowed for earlier sowing rather than wai ng for the soil to dry sufficiently for llage. Sunflower and mustard were profitable and both provided farmers with an extra income of USD626 to 653 ha-1. System produc vity was improved and labor costs were reduced by adop ng zero llage systems for both mustard (relay intercropping with rice) and sunflower (dibble plan ng with rice stubble reten on). Finally seed supply, processing and marke ng need to be addressed to enable expansion of oilseed crops. Key message: Relay intercropping of mustard and dibbled plan ng of sunflower are profitable interven ons in rice-based cropping systems in southern Bangladesh. Keywords: mustard, sunflower, llage, dibbling, short dura on 1. Introduc on Cropping systems in the coastal zone of Bangladesh are dominated by rice, as in other parts of Bangladesh, but cropping intensity is low for several reasons. Much of the coastal zone suffers from poor drainage and thus rice is usually transplanted into stagnant water where only the tall, late maturing and low yielding local varie es can survive. Further, the soil remains saturated during the winter season, which delays the sowing of dry season crops. Soil and water salinity increase as the dry season progresses, further limi ng the ability to grow dry season crops. Even in the less saline (<4 dS m-1) areas of the coastal zone, it is difficult to fit oilseed crops such as mustard in the short fallow period between the rainy season rice crop (transplanted aman, T. 405 Aman) and winter season (boro) rice. Prior to the introduc on of boro rice in the 1980s, T. Aman was frequently followed by oilseed crops such as mustard (Brassica juncea) and sesame (Sesamum indicum) or grain legume crops such as len l (Lens culinaris) and mungbean (Vigna radiata). In many areas, boro replaced oilseed and grain legume crops. In addi on, many farmers in southwest Bangladesh cannot grow oilseed crops such as mustard or soybean because the soils are too saline in the la er part of the dry season. Present domes c edible oilseed produc on in Bangladesh is 722,000 tons, which meets only one-third of na onal demand (BBS 2013). Present per capita food oil consump on is only 10 g day-1 compared with the recommended intake of 22 g day-1 (Rahman and Khan 2005). Bangladesh has one of the lowest fat intake levels in the world. Only 12% of total energy requirement is derived from fats. This is far below the recommended 35% for six- to twelve-month old children and the recommended 20 to 35% for pregnant and lacta ng mothers (Michaelsen et al. 2011). Children have high demand in rela on to their size for energy foods. When the body is not supplied with enough energy from carbohydrates it converts whatever proteins are in their food to energy, which can cause protein deficiency and stunted growth. Stun ng (below average height for age) in Bangladesh, at 41% of children under five years, is high (Bangladesh Demographic and Health Survey 2013). Increasing the oil content of children’s foods would be more likely to occur if households grew oilseed crops and produced oil for family consump on. The constraints of oilseed expansion are largely related to use of long dura on aman varie es, and inadequate knowledge of crop management under saline condi ons. The constraints to oil seed produc on need to be overcome by replacing the exis ng aman varie es with short dura on modern rice varie es and by introducing salt tolerant varie es of oilseed crops. The development of short dura on rice varie es such as BRRI dhan33, BRRI dhan39 and BINA dhan7, and of high yielding mustard varie es such as BARI sarisha14 and BARI sarisha15, have created an opportunity to fit mustard between the aman and boro crops (BARC 2001; Islam 2013). This would increase the produc vity of rice-based cropping systems (Rashid et al. 2012b). Also, in Faridpur and Barisal regions, there are areas of aman-mustard-jute and aman-mustard-fallow cropping sequences with low yielding local mustard varie es that could be replaced by high yielding mustard varie es to increase system produc vity. In coastal Bangladesh, plan ng of dry season crops is o en delayed un l January or even February because water remains on the land un l December, at which me the weather becomes cold and foggy, and it is only a er the weather starts to warm up in the la er part of January that significant soil drying starts to take place. Cul va on using a two-wheel, tractor operated power ller is only possible once the topsoil has dried below field capacity. As a result, valuable soil moisture is lost while wai ng for the soil to dry, and the late planted crops are exposed to damaging levels of soil and water salinity in March and April. Further, these late planted crops are at risk of being damaged by pre-monsoon rains that start from early May. Early plan ng, preferably in November and December, is essen al to prevent the crops from exposure to increasing soil salinity and the pre-monsoon rains. In the case of small seeded crops like mustard, early plan ng can be achieved by broadcast sowing shortly prior to rice harvest. The seed then germinates on the moist soil under the rice canopy. In the case of larger seeded crops like sunflower, the seed can be dibbled just a er rice harvest using a small s ck or finger to create a hole for the seed (Rashid et al. 2012a). Another possible approach, yet to be fully tested, is to use power ller operated seeders (PTOS) to directly sow crops into the rice stubble. This approach could be used for all types of oil seed crops but currently has only been tested on sesame. Therefore, farmer par cipatory trials and demonstra ons were conducted to test and develop systems for the inclusion of oilseed crops in rice-based cropping systems in the coastal zone of Bangladesh. 2. Materials and methods 2.1 Study sites Adap ve trials on varietal performance of mustard were conducted in Brahmankanda and Noapara villages under Bhanga Upazila of Faridpur District, on clay loam soils. The sites are strongly influenced by the south-western monsoon. The average annual rainfall is 1546 mm, with more than 80% occurring from 406 mid-June to the end of September. Monthly mean minimum temperature of the area is 12.60C in January and maximum is 35.8°C in May. Another field trial was conducted at Pankhali village under Dacope Upazila of Khulna District to determine the effect of sowing date on dibbled sunflower. Pankhali is about 1 m above sea level and belongs to Agro-Ecological Zone 13. The average annual rainfall of the area is 1710 mm. Monthly mean minimum temperature is 12.50C in January and maximum is 35.5°C in May. The soil of the experimental field is loam in texture. Farmer par cipatory demonstra ons of mustard varie es, alterna ve establishment methods for mustard and sunflower, and cropping system intensifica on, were conducted in several upazilas of Khulna, Satkhira, Bagerhat Districts in Khulna Division, Patuakali District in Barisal Division, and Jhenaidah District in Faridpur Division. Greater Khulna, Barisal and Faridpur are 1 to 2 m above sea level. The average annual rainfall of each region is 1710, 1955 and 1467 mm, respec vely, with monthly mean minimum temperature of 12.5, 12.1 and 11.20C, and mean maximum temperature of 35.5, 35.1 and 37.10C, respec vely. The soil of the demonstra on plots was loam in Kaligonj Upazila in Jhenaidah, Dumuria Upazila of Khulna, Sadar and Kolaroa Upazilas of Satkhira, Fakirhat Upazila of Bagerhat, and Babugnoj Upazila of Barisal. Soils of other loca ons were clay to clay loam in texture. The soils of sites selected for the aman-mustard-boro cropping system and establishment experiments were of low salinity (<2 dS m-1). The demonstra on plots of sunflower in Patuakhali and in Kolaroa and Sadar Upazilas of Satkhira were slightly saline (<4 dS m -1). The peak soil salinity of the other sunflower demonstra on plots ranged from 6 to 9 dS m-1 at crop maturity in April. 2.2 Adap ve trials and demonstra on 2.2.1 Adap ve trials Rice-mustard-jute is one of the major cropping sequences in Brahmankanda and Noapara villages of Bhanga Upazila under Faridpur District. Farmers usually grow T. Aman (BRRI dhan39), Indian varie es of jute (JRO-524) and local mustard (Choitali). Mustard and sunflower trials were carried out in these villages. Mustard. High yielding mustard varie es, BARI sarisha11 and BARI sarisha16 were grown side-by-side with Choitali in six farmers’ fields, each field considered as one replicate. All mustard varie es were sown on the same day, and the farmers’ usual aman and jute varie es were retained in the system. The trial was laid out in a randomized complete block (RCB) design, and plot size was 100 m 2. Sunflower. The effect of sowing me of dibbled planted sunflower was studied with six sowing dates (25 November, 5, 15 and 25 December, and 5 and 15 January) a er aman harvest at a single site, Pankhali of Khulna District. The treatments were assigned in a RCB design with three replica ons and plot size 100 m2. 2.2.2 Farmer par cipatory demonstra ons Demonstra ons of mustard varie es, mustard establishment method (relay sown before harvest or broadcast onto lled soil a er harvest) and the aman-mustard-boro cropping sequence were implemented in plots of 1320 m2 in numerous farmers’ fields in many loca ons from 2012 to 2014 (Table 1). Many demonstra ons of dibbled and lled sunflower were also conducted in plots of 800-1320 m2. 407 Table 1. Demonstrated technologies and loca ons, 2012-2014 Demonstra on technology Name of villages Aman-mustardboro; aman -boro i) Bara a ii) Bashghata Relay sown (zero ll) and lled mustard in amanmustard-boro system District No. of demonstraons/village i) Dumuria ii) Satkhira Sadar i) Khulna ii) Satkhira i) 6 ii) 30 i) Tipna, Chuknagar ii) Parkukrail, Ramer danga, Alipur, Daulatpur iii) Chaltabaria, Sultanpur iv) Magri v) Nagarghata vi) Jaria vii) Mathanpur i) Dumuria ii) Satkhira Sadar iii) Kaligonj iv) Debhata v) Tala vi) Fakirhat vii) Kaligonj i) Khulna ii) Satkhira iii) Satkhira iv) Satkhira v) Satkhira vi) Bagerhat vi) Jhenaidah i) 3, 4 ii) 3, 3, 3, 4 iii) 3, 3 iv) 5 v) 9 vi) 5 vi) 12 Mustard varie es under relay and lled methods of plan ng i) Chaltabaria, Sultanpur, Datpur, Varasimla, Khamarpara ii) Bohera, Kamta, Magria iii) Jafarpur, Jhaudanga, Alaipur, Panikouria iv) Bashghata v) Nagarghata vi) Kundorria i) Kaligonj ii) Debhata iii) Kolaroa iv) Satkhira Sadar v) Tala vi) Ashashuni i) Satkhira ii) Satkhira iii) Satkhira iv) Satkhira v) Satkhira vi) Satkhira i) 7, 9, 2, 6, 2 ii) 18, 2, 10 iii) 21, 9, 19, 10 iv) 4 v) 30 vi) 30 Local and improved mustard variety in rice-mustard sequence Baradi Gorangol Gournadi Barisal 7 408 Upazila Table 1 (cont.). Demonstrated technologies and loca ons, 2012-2014 Demonstra on technology Name of villages No lled and lled dibbled sunflower i) Mohammadpur, Atulia, Patrakhola, Dhumghat, Koikhali, Joyakhali, Bayeshkhali ii) Varasimla, Marka, Khamarpara iii) Parkukrail iv) Naggolghara v) Juge Pukuria, Nagarghata vi) Pankhali, Kholisha vii) Fatehpur viii) Bisot ix) Guptamari, Parsolua, Debitola x) Ponchu, Tipna, Jolerdanga xi) Islampur xii) Char Miazan xiii) Hetalia, Gabura xiv) Manijhuri xv) Hoglapasha xvi) Protappur No lled and lled dibbled sunflower and salinity effect i) Mohammadpur, Patrakhola, Dhumghat ii) Varasimla iii) Nurerchalk Upazila District No. of demonstraons/village i) Shaymnagar ii) Kaligonj iii) Satkhira Sadar iv) Debhata v) Tala vi) Dacope vii) Bagerhat Sadar viii) Kachua ix) Ba aghata x) Dumuria xi) Kolapara xii) Bauphal xiii) Patuakhali Sadar xiv) Amtali xv) Patharghata xvi) Babugonj i) Satkhira ii) Satkhira iii) Satkhira iv) Satkhira v) Satkhira vi) Satkhira vii) Bagerhat viii) Bagerhat ix) Khulna x) Khulna xi) Patuakhali xii) Patuakhali xiii) Patuakhali xiv) Barguna xv) Barguna xvi) Brisal i) 12,11, 18, 8, 7, 7, 9 ii) 15, 10, 6 iii) 10 iv) 11 v) 7, 8 vi) 18 vii) 6 viii) 6 ix) 7, 7, 7 x) 27, 13, 10 xi) 6 xii) 6 xiii) 6, 6 xiv) 6 xv) 6 xvi) 6 i) Shaymnagar ii) Kaligonj iii) Debhata i) Satkhira ii) Satkhira iii) Satkhira i) 6, 7, 7 ii) 15 iii) 7 Aman-mustard-boro. The farmers in the selected villages cul vate medium dura on aman varie es such as BR10, BRRI dhan30 and Swarna, which are harvested in the last week of November. They then sow boro in seedbeds in December that are ready for transplan ng in mid-January to the first week of February. Farmer par cipatory demonstra ons were conducted in which the medium dura on aman variety was replaced with a short dura on variety (BRRI dhan39 or BINA dhan7), and mustard was added to the system during 2012 to 2014 (Table 1). The mustard varie es BARI sarisha14 and BARI sarisha15 were established by broadcas ng and relay sowing seven to ten days prior to aman harvest, or following harvest and llage with a power ller, depending on soil moisture at the me of maturity of the rice. A er mustard harvest, the boro crop was transplanted into puddled soil and harvested in April. Aman-mustard-fallow. In Gournadi, Barisal, farmers cul vate a short dura on, local aman variety (Rajashail) followed by a local mustard variety called Maghi. BARI sarisha15 was compared with Maghi sarisha in seven farmers’ fields. Aman rice was transplanted in mid-August and harvested in the first week of November. Mustard was sown during to the third week of November and harvested in February. Aman-sunflower. Representa ve fields were selected where farmers prac ce an aman-fallow cropping system in villages in Khulna and Barisal. Farmer par cipatory demonstra ons on crop establishment method in 409 sunflower (the hybrid Hysun33) and the effect of soil salinity on sunflower produc on were conducted in 2012-13 and 2013-14. 2.3 Crop management prac ces In the aman-mustard-boro cropping sequence, the short dura on rice varie es and mustard varie es were grown following the BRRI and BARI recommended crop management prac ces (BRRI 2011; BARI 2012). Aman rice was rainfed in both non-saline and saline loca ons. The seeds of high yielding aman varie es (BR10, BR23, BRRI dhan41, BRRI dhan53 and BRRI dhan54) were sown during the first fortnight of July in the saline areas and transplanted in late July/early August. In Khulna region, the sunflower seed was dibbled just a er aman harvest. In Barisal region, sunflower seed was sown into hand-dug furrows made a er harvest and lling using a two-wheel tractor with a power ller. The recommended management prac ces for rice and sunflower were followed (BRRI 2011; Rashid et al. 2012a). No ll dibbled sunflower was sown in the standing rice stubble (20-25 cm high) 20 to 30 days before the lled sunflower, and one to two irriga ons were applied to the sunflower crops. 2.4 Data collec on and analyses Grain yield was es mated by harves ng an area of 10 m2 from the centre of each plot and expressed as t ha-1 at 14, 9 and 9% moisture contents for rice, mustard and sunflower, respec vely. The produc vity of mustard and sunflower was determined as rice equivalent yield (REY), which was calculated as follows: REY  Sunflower or mustard yield ( kg / ha) x sunflower or mustard price (Tk . / kg ) Price of paddy (Tk./kg) Electrical conduc vity (EC) of the topsoil (0-15 cm) was measured at the sites of three upazilas of Satkhira in the lled and non- lled plots of the sunflower demonstra ons. EC was measured in a 1:5 soil: water suspension (EC1:5) and was converted to soil satura on extract EC (ECse) by mul plying by conversion factors of 14 for loam, 10 for clay loam, 9 for light clay and 6 for heavy clay soil following the method of Watling (2007). Analysis of variance was done using CropStat Version 7.2, and differences were considered significant only at P < 0.05. Mul ple linear regression analyses using SPSS version 11.5 were run to determine the effects of soil salinity on seed yield of sunflower. Soil salini es, on the day of seeding (0 DAS) and at 30 and 60 DAS and maturity, were the independent variables and seed yield was the dependent variable. For the economic analyses, the cost of inputs (hire of power ller and thresher rental for land prepara on and threshing, labor for different opera ons, seed, fer lizer, pes cide, irriga on, etc.) and the farm gate prices of outputs (paddy, seeds of mustard or sunflower) were calculated based on their local market price. The cost of labor, including family labor, and all input costs were included in the total variable cost (TVC) (Tables 2 and 3). The gross margin (GM) per hectare for each crop management or cropping sequence op on was calculated by deduc ng the total variable cost from the gross return. The benefit cost ra o (BCR) was calculated by dividing the gross return (GR) by TVC. 410 Table 2. Compiled variable costs used in the financial analysis of technologies demonstrated in 2012-2014 Demonstra on technology Seed cost ($ ha -1) Fer lizer cost ($ ha -1) Pes cide cost ($ ha-1) Aman-mustard-boro; aman-boro A-M-B1 41.7 A-B2 33.5 A-M-B 278 A-B 225 A-M-B 142 Relay sown (zero ll) and lled mustard in amanmustard-boro system in Khulna Relay Tilled Relay Tilled Relay Tilled 42.3 40.8 255 255 141 138 Relay sown (zero ll) and lled mustard in amanmustard-boro system in Jessore Relay Tilled Relay Tilled Relay Tilled 31.7 30.2 427 397 153 91.6 Mustard variety under ll and relay method of plan ng Relay Tilled Relay Tilled Relay Tilled BARI sarisha14 8.30 6.80 63.5 82.9 - - BARI sarisha15 8.30 6.80 78.0 73.1 - - Local and improved mustard varie es in rice-mustard sequence in Barisal Local 4.40 No lled 12.3 Improved 5.80 Tilled 12.3 Local 87.4 No lled 186 Improved 100 Tilled 169 Local No lled - Improved Tilled - A-B 138 No lled dibbled and lled sunflower 1 A-M-B = Aman-Mustard-Boro 2 A-B = Aman-Boro 411 Table 3. Compiled variable costs used in the financial analysis of technologies demonstrated in 2012-2014 Demonstra on technology Tillage cost ($ ha-1) Irriga on cost ($ ha-1) Labour cost ($ ha-1) Aman-mustard-boro; aman-boro A-M-B1 146 A-B2 126 A-M-B 309 A-B 282 A-M-B 901 Relay sown (zero ll) and lled mustard in amanmustard-boro system in Khulna Relay Tilled Relay Tilled Relay Tilled 102 164 303 305 588 611 Relay sown (zero ll) and lled mustard in amanmustard-boro system in Jessore Relay Tilled Relay Tilled Relay Tilled 160 203 411 401 997 1100 Mustard variety under ll and relay method of plan ng Relay Tilled Relay Tilled Relay Tilled BARI sarisha 14 - 53.8 23.2 35.5 110 124 BARI sarisha 15 - 52.8 34.7 34.3 136 118 Local and improved mustard varie es in rice-mustard sequence in Barisal Local Improved 60.2 Tilled - Local 64.6 No lled 77.0 Improved 15.4 Tilled 35.1 Local 48.5 No lled 51.9 Improved 119 135 Tilled 228 319 No lled A-B 749 No lled dibbled and lled sunflower A-M-B = Aman-Mustard-Boro 1 2 A-B = Aman-Boro 3. Results and discussion 3.1 Rice-mustard based cropping systems 3.1.1 System produc vity of aman-fallow-boro and aman-mustard-boro In the aman-mustard-boro cropping system demonstrated in 30 farmers’ fields of Khulna and Satkhira, grain yield of short dura on aman was similar to that of the farmers’ medium dura on varie es with similar crop management at around 5.6 t ha-1 (Table 4). On average, mustard produced 1.59 t ha-1 seed. The yield of boro was similar following mustard or fallow (5.73 and 5.90 t ha-1, respec vely), showing that growing mustard between aman and boro crops did not reduce boro yield. Rice equivalent yield of the aman-mustard-boro was significantly higher (by 43%) than that of the aman-fallow-boro system. The higher REY in aman-mustard-boro also resulted in a higher BCR and gross margin (by 105%) (Table 5). 412 Table 4. Grain, seed yield of T. Aman, mustard and boro rice and rice equivalent yield of aman-mustard-boro and rice-fallow-rice cropping pa ern in Khulna and Satkhira, 2012-13 Cropping sequence Farmer (no.) Aman-mustard-boro Aman-fallow-boro LSD0.05 CV (%) 30 6 - Grain/seed yield (t ha-1) T. Aman Mustard Boro 1 4.71 1.59 5.73 2 4.75 5.90 3 ns ns 2.9 4.2 1 grain yield of short dura on rice varie es (BRRI dhan39 and BINA dhan7) 2 medium dura on rice (BR10) 3 not significant REY of cropping system 15.2 10.65 0.76 7.5 Table 5. Economic return from aman-mustard-boro and aman-fallow-boro cropping sequence in Khulna and Satkhira, 2012-13 Cropping pa ern Aman-mustard-boro Aman-fallow-boro (Farmers’ prac ce) Farmer (no.) Gross return (USD ha-1) TVC (USD ha-1) 30 3038 1816 GM (USD ha-1) 1222 BCR 1.67 6 596 1.38 2051 1554 Note: Price USD t-1: Mustard seed: 584.4, aman paddy: 194.8, boro paddy: 207.8, 1USD = 77 BDT 3.1.2 Rice-mustard-fallow In the aman-mustard-fallow cropping system in Barisal District, BARI sarisha15 gave a significantly higher (by 37%) seed yield than Maghi sarisha (Table 6). This was due to more pods per plant and seeds per pod, and higher seed weight, in BARI sarisha15. Gross margin of BARI sarisha15 was 133% higher than the local variety (Table 7). BARI sarisha15 remained in the field 14 days longer than the local mustard variety. However, the longer dura on of BARI sarisha 15 did not affect establishment of the next crop as the mustard crop is followed by fallow un l establishment of the next aman crop. Table 6. Seed yield and yield component of mustard varie es under rice-mustard-fallow cropping system at Gournadi, Barisal, 2013-14 Variety Farmer (no.) Plant height (cm) BARI sarisha15 Maghi sarisha (FP) LSD0.05 CV (%) 7 7 - 115 64 2 2.1 Plant m Pod plant-1 Seed pod-1 Thousand Seed seed yield weight (g) (t ha-1) Growth dura on (days) 74 106 5 4.2 61.7 32.2 2.7 4.4 21.7 13.7 0.7 3.2 3.05 2.45 0.15 4.0 84 70 2 1.8 -2 1.63 1.10 0.01 4.5 413 Table 7. Economic return of mustard varie es under the aman-mustard-fallow system at Gournadi, Barisal, 2013-14 Variety Farmer (no.) Gross return (USD ha-1) Total variable cost (USD ha-1) Gross margin (USD ha-1) BARI sarisha15 7 849 355 494 Maghi sarisha 7 498 286 212 Note: Seed price (USD t-1): BARI sarisha15 seed: 520.9, Maghi sarisha: 452.7, 1USD= 77 BDT 3.1.3 Produc vity of the aman-mustard-boro cropping system with and without llage in mustard Mustard established in lled soil yielded 1.76 and 1.27 t ha-1 in Khulna and Jessore regions, respec vely, and net return from the aman-mustard-boro system was USD 1240 and 1440 ha-1, respec vely (Tables 8 and 9). Relay-sown mustard produced a lower seed yield of 1.21 and 0.99 t ha-1 in Khulna and Jessore, respec vely, with net return of USD 999 and 1550 ha-1. The higher net return from mustard in Jessore region was due to higher market price for mustard following storage for a longer period by the farmers. Although the relay mustard earned lower net return than the lled mustard, the system was found promising as it ensured early plan ng of mustard and hence early transplan ng of boro rice, which is a concern to the farmers. Table 8. Seed yield and economic return of relay and lled mustard in aman-mustard-boro cropping system, Khulna region, 2013-14 Tillage in mustard Tilled Relay LSD0.05 CV (%) Farmer (no.) 18 27 - Grain/Seed yield (t ha-1) T. Aman 3.94 3.85 ns 10.0 Mustard 1.76 1.21 0.17 20.1 Boro 4.85 4.78 ns 9.3 REY (t ha-1) TVC Gross (USD ha-1) return (USD ha-1) 13.49 1515 2753 11.84 1431 2430 0.65 8.7 - Gross margin (USD ha-1) 1238 999 - Note: Price USD t-1: Mustard seed: 519.5, Aman paddy: 194.8, Boro paddy: 220.8, 1USD = 77 BDT Table 9. Seed yield and economic return of relay and lled mustard in aman-mustard-boro cropping system, Sadar and Kaligonj Upazila, Jhenaidah, 2013-14 Tillage in mustard Tilled Relay LSD0.05 CV (%) Farmer (no.) 12 12 - Grain/Seed yield (t ha-1) T. Aman 5.08 5.05 0.23 5.1 Mustard 1.27 0.99 0.09 9.4 Boro 6.36 6.18 Ns 9.4 REY (t ha-1) TVC Gross Gross (USD ha-1) return margin -1 (USD ha ) (USD ha-1) 15.0 2263 3707 1444 14.2 2180 3729 1549 0.6 4.8 - Note: Price USD t-1: Mustard seed: 649.4, Aman paddy: 246.8, Boro paddy: 279.2, 1USD = 77 BDT 414 3.1.4 Produc vity of mustard varie es under lled and relay method of seeding in aman-mustard-boro cropping sequence The interac on between variety and establishment method was significant in Khulna region (Table 10). In lled soil, yield of BARI sarisha15 was slightly but significantly higher than that of BARI sarisha14. The same trend occurred with relay sowing, but the difference was not significant. Yield with relay plan ng was much lower than yield with llage prior to sowing for both varie es. Table 10. Seed yield of mustard varie es and economic return from relay and lled mustard under aman-mustard-boro cropping sequence in Satkhira, 2013-14 Establishment method of Variety mustard Tilled Relay LSD0.05 CV (%) BARI sarisha14 BARI sarisha15 BARI sarisha14 BARI sarisha15 Est. method (M) Variety (V) MXV Farmer (no.) 59 47 15 58 - Seed yield (t ha-1) 1.64 1.89 1.26 1.31 0.28 ns 0.06 9.0 TVC (USD ha-1) 303 285 204 257 - Gross return (USD ha-1) 944 1092 728 754 - Gross margin (USD ha-1) 641 807 524 497 - BCR 3.32 4.03 3.84 3.49 - Note: Price USD t-1: Mustard seed: 576.9, 1USD = 77 BDT 3.1.5 Mustard variety in aman-mustard-jute cropping system The interac on between variety and loca on was also significant for two loca ons in Faridpur (Table 11). There was a consistent trend for higher seed yield of BARI sarisha11 than BARI sarisha16 at both loca ons; however, the difference was only significant at Brahmankhanda. Seed yield of the local variety Choitali was significantly lower than that of both BARI varie es at both loca ons. The stover yield of all varie es was similar. The growth dura on of BARI sarisha11 was six to eight days longer than that of Choitali. Economic analysis indicated an increase in gross margin of USD 171 and 202 ha -1 by replacing Choitali with BARI sarisha 11 in Noapara and Brahmankanda, respec vely. 415 Table 11. Seed, stover yield and economic return of mustard varie es in an aman-mustard-jute cropping system at two loca ons of Bhanga, Faridpur, 2013-14 Loca on Variety Seed Farmer yield (No.) (t ha-1) Brahman kanda BARI sarisha11 BARI sarisha16 Choitali (FP) BARI sarisha11 BARI sarisha16 Choitali (FP) - 6 6 6 6 6 6 - Noapara LSD 0.05 Variety (V) Loca on (L) (LXV) (CV %) Total Stover Growth Gross Gross variable yield dura on return margin cost (t ha-1) (days) (USD ha-1) (USD ha-1) (USD ha-1) 2.03 1.86 1.66 1.94 1.82 1.62 4.23 4.13 4.00 4.10 3.95 3.82 0.10 0.12 0.14 6.4 ns ns ns 8.6 109 113 101 1206 1182 1016 1162 1169 990 485 500 497 492 489 492 721 682 519 669 680 498 - - - Note: Price (USD t-1): Mustard seed: 519.5, Stover: 39, 1 USD= 77 BDT, ns =not significant 3.2 Aman-sunflower cropping system 3.2.1 Plan ng date of sunflower Sunflower dibbled on 25 November and 5 December without llage produced similar seed yield (~3 t ha-1), significantly higher than yield of all later sowings. There was a trend for yield to decline as sowing was delayed from 5 December to 15 January, with the lowest yield (2.51 t ha-1) for January 15 plan ng. The results indicate that farmers lose 22 kg ha-1 of grain yield worth USD 11.4 ha-1 for every day that plan ng is delayed a er 5 December. While a yield of 2.5 t ha-1 for the 15 January planted sunflower is s ll a good yield, the later planted sunflower would also be at greater risk of damage from early monsoon rains (which did not occur in the study year) as the crop would not be ready for harvest un l the end of April. 3.5 3 2.5 2 1.5 1 0.5 0 25 Nov. 5 Dec. 15 Dec. 25 Dec. 5 Jan. 15 Jan. Fig. 1. Seed yield of sunflower as affected by sowing date (error bars indicate the LSD at 5% level of significance). 416 3.2.2 Establishment method in sunflower and soil salinity effect Data collected from 150 demonstra ons in Khulna region showed that crop establishment method did not affect the seed yield of sunflower (Tables 12 and 13). Seed sown by dibbling gave similar seed yield (2.29 t ha-1) to seeding in lled soil (2.39 t ha-1). But the former method earned 19% higher gross margin due to the cost of land prepara on and earthing up. Moreover, dibbled crops were ready for harvest at least 20 days earlier than the crops planted in lled soil, reducing the risk of damage by early monsoon storms. April and May 2014 were much drier than normal, with no significant pre-monsoon storms. Ini al soil salinity ranged from 0.5 to 2.5 dS m-1 across the sunflower demonstra on sites (Fig. 2). Soil salinity increased with me a er sowing up to maturity (Table 13), and there was strong posi ve linear rela onship (R 2 = 0.72**) between ini al and final soil salinity (Fig. 2). The highest seed yield was achieved in Kaligonj (2.52 to 2.55 t ha -1) where soil salinity at crop maturity (ECse 6.8 dS m-1) was least. Yield was least at Debhata (2.08 to 2.09 t ha -1) where the salinity was highest (9.88 dS m-1 at crop maturity). The results of the mul ple linear regression show that soil salinity at seeding (X1) had a significant and nega ve effect on yield, and the coefficient (-0.347) had the largest magnitude (Table 14). Soil salinity at other stages had no significant effect on seed yield, although the effect of salinity at harvest was almost significant, with a nega ve co-efficient about one-third of that for salinity at sowing. The yield of sunflower seeded a er aman harvest varied from 2.4 to 3.1 t ha-1 across loca ons in Barisal (Table 15). The highest seed yield was in Bauphal (3.08 t ha -1), which was similar to seed yield in Patharghata, Amtali and Kolapara. The lowest seed yield was in Gabua of Patuakhali Sadar Upazila, mainly due to smaller head size and fewer seeds per head due to lack of water for a second irriga on. The BCR and the gross margin showed a similar trend to that of seed yield except for Babugonj (Table 16). Although the seed yield was lower in Babugonj, the gross return and gross margin was higher due to its higher market price in the locality. Table 12. Seed yield and economic return from dibbled and lled sunflower in an aman-sunflower cropping sequence in Khulna region, 2013-14 Establishment method Farmer (no.) Seed yield (t ha -1) TVC (USD ha-1) Gross return (USD ha-1) Gross margin (USD ha-1) Increased GM over lled (%) BCR Dibbled Tilled LSD0.05 CV (%) 150 90 - 2.29 2.39 Ns 20.4 461 629 - 1172 1225 - 711 596 - 19 - 2.75 1.95 - Table 13. Seed yield of dibbled and lled sunflower and soil salinity at different loca ons of Satkhira in an aman-sunflower cropping sequence, 2013-14 Loca on Shaymnagar Kaligonj Debhata LSD0.05 CV (%) Establishment method Dibbled Tilled Dibbled Tilled Dibbled Tilled Est. method (M) Loca on (L) MXL - Farmer (no.) 18 2 11 3 7 - - Seed yield (t ha-1) 2.08 2.09 2.55 2.52 1.89 ns 0.15 0.22 8.6 0 DAS 1.20 1.20 0.94 1.00 1.68 ns ns ns - Soil salinity (dS m-1) 30 DAS 60 DAS Maturity 4.16 5.59 7.22 4.25 6.15 7.25 4.12 5.32 6.80 4.65 6.05 7.73 4.40 6.32 9.88 ns ns 1.02 ns ns 1.19 ns ns 1.72 19.2 417 y = 2.8115x + 4.2108 R2 = 0.7168** Soil salinity at seeding (dS m-1) Fig. 2. Rela onship between soil salinity at seeding and maturity stages of sunflower in Satkhira. Table 14. Standardized coefficients for explaining the effect of salinity on seed yield of sunflower in the sunflower establishment method farmer par cipatory demonstra ons in Satkhira (N=42) Soil salinity Symbol Unstandardized coefficients B (Constant) Intercept, a 2.374 At 0 DAS X1 -0.347 At 30 DAS X2 0.034 At 60 DAS X3 0.158 At maturity X4 -0.105 Std. Error 0.240 0.164 0.097 0.111 0.058 Standardized coefficients t value Significance 9.896 -2.120 0.347 1.423 -1.819 0.000 0.042 0.731 0.165 0.079 Beta -0.588 0.110 0.658 -0.592 Table 15. Seed and stalk yield and yield component of sunflower under an aman-sunflower-fallow cropping system at different loca ons of Barisal, 2013-14 Loca on Kolapara Bauphal Hetalia (Patuakhali Sadar) Amtali Patharghata Gabua (Patuakhali Sadar) Babugonj LSD0.05 CV (%) 418 Plant height (cm) Plant m-2 Head Seed diameter head-1 (cm) Thousand Seed seed yield weight (t ha-1) (g) 126 176 164 3.27 3.40 3.67 18.5 19.2 17.5 1169 1231 1044 78.8 79.7 74.5 2.82 3.08 2.70 4.17 5.10 3.85 110 117 114 173 146 163 3.60 3.40 3.63 17.8 18.5 15.9 1115 1192 914 74.6 78.2 74.5 2.91 3.01 2.37 4.02 4.35 3.75 112 111 113 191 6 3.4 3.47 0.29 7.0 17.2 1.1 5.2 1100 122 9.4 76.2 2.2 2.5 2.62 0.28 8.7 3.55 0.59 12.2 123 2 1.7 Stalk yield (t ha-1) Growth dura on (days) Table 16. Economic return of sunflower under an aman-sunflower cropping system at different loca ons of Barisal, 2013-14 Loca on Kolapara Bauphal Hetalia (Patuakhali Sadar) Amtali Patharghata Gabua, Patuakhali Sadar Babugonj TVC (USD ha-1) 698 662 654 716 699 658 743 Gross return (USD ha-1) 1286 1438 1207 1311 1368 1082 1569 Gross margin (USD ha-1) 588 776 553 595 669 424 826 BCR 1.84 2.17 1.85 1.83 1.96 1.64 2.11 4. Conclusions and recommenda ons Oil seed crops can be introduced within the exis ng rice-based cropping systems of the coastal zone of Bangladesh without compromising produc vity of other crops in the rota on. Mustard can be grown between aman and boro crops by replacing medium dura on aman varie es with short dura on modern varie es. New mustard varie es of BARI Sarisha 11 and BARI Sarisha 15 provide significantly higher seed yield than local varie es. The income of growers can be increased and with higher BCR by growing oilseed crops. Mustard marke ng and processing is well established in coastal Bangladesh, but the areas where it can be grown are rela vely limited due to soil salinity. Seed of new mustard varie es is not readily available and what is available is mixed. Produc on of mustard seed in isola on seems to be difficult to achieve. Maintaining proper isola on distance between seed produc on farms and produc on of seed of the most preferred variety in an area can solve this problem. Early planted sunflower gives a significantly higher grain yield than January sown sunflower and its earlier harvest reduces the risk of damage from early monsoon rain. Sunflower can be grown even a er medium dura on aman varie es such as BRRI dhan41 or BRRI hhan 54 in soils where salinity increases to >7 dS m1 for the second half of the crop. Soil salinity at sowing is a major determinant of sunflower yield and so plan ng sunflower in fields with lower soil salinity at seeding should be encouraged. A simple test for farmers to determine soil salinity and thus suitability for cul va on of sunflower is needed. Sunflower has a promising market in urban areas where it can be marketed under the low cholesterol label. Both sunflower and mustard have poten al to increase farmers’ income but lack of quality seed, higher cost of seed of improved varie es, poor oil extrac on technology and defec ve links to major market players are currently constraints to wide scale adop on. Acknowledgements The authors thank the United States Agency for Interna onal Development (USAID) for the support to carry out this study through Cereal System Ini a ve for South Asia-Bangladesh Project (CSISA-BD). Thanks are extended to the community who supported to evaluate the trials and demonstra ons in their fields. 419 References Bangladesh Demographic and Health Survey, January 2013. BARC (Bangladesh Agricultural Research Council). 2001. A compendium: Packages of Technologies. In: A handbook for farming systems development. ed M.F. Haque, M.A. Razzaque and Abu Akteruzzaman, p.12. BARC, Dhaka. BARI. 2011. BARI technology (commodity). h p://www.bari.gov.bd/index.php?op on=com_advancedsearch&view=advancedsearch BBS 2013. Sta s cal Yearbook of Bangladesh. 2012. Bureau of Sta s cs. Sta s cs and Informa cs Division. Ministry of Planning, Government of the People’s Republic of Bangladesh, Dhaka. Pp. 140-142. BRRI (Bangladesh Rice Research Ins tute). 2011. Modern rice cul va on. 16th edi on. p. 72. Islam, M. R. 2013. Maximiza on of crop yield in t. aman-mustard-boro cropping pa ern by agronomic manipula on. Half Yearly Progress Report, KGF, BARC. Michaelsen, K.F., G.K. Dewey, A.B. Perez-Exposito, M. Nurhasan, I. Lauritzen, and N. Roos. 2011. Food sources and intake of n-6 and n-3 fa y acids in low-income countries with emphasis on infants, young children (6-24 months), and pregnant and lacta ng women. Maternal and Child Nutri on. Blackwell Publishing Ltd Maternal and Child Nutri on (2011), 7 (Suppl. 2), pp. 124–140. Rahman M. M. and S. I. Khan. 2005. Food security in Bangladesh: Food availability. In: ed. Government of the People's Repubpic of Bangladesh and World Food Programme-Bangladesh. p.9. Workshop held at IDB Bhavan, Agargaon, Dhaka, Bangladesh, 19-20 October, 2005. Rashid, M. H., S. Nasrin, M. K.I. Rony, D. Mahalder, S.Begum. 2012a. Sunflower cul va on a er harves ng of t. aman rice in saline area (In Bangla). Folded brochure. Interna onal Rice research Ins tute, 2/2 Banani, Dhaka 1213, Bangladesh. Rashid, M.H., M.K.I. Rony, S. Nasrin. 2012b. Increasing produc vity of rice-rice cropping system adop ng short dura on rice and mustard and relay cropping. Paper presented in the Interna onal Conference on Environment, Agriculture and Food Sciences, held in Phuket, Thailand on 11-12 August, 2012. pp. 13-16. Planetary Scien fic Research Center. Rony, M.K.I., M.H. Rashid, S. Nasrin, M.A. Saleque. 2013. Effect of zero llage on Boro rice cul va on in ghers of south-western Bangladesh. Bangladesh Journal of Agronomy. 16(1): 105-107. Watling, K. 2007. Measuring salinity. Fact Sheet. Department of Natural Resources and Water. Queensland Government, Australia. L 137. h p://joomla.speedweb.com.au/soil2015/images/sampledata/publica ons_tab/schoolresources/factsheets/ 07_measuring-salinity.pdf 420 Triple rice in a year: Is it a feasible op on for the low salinity areas of the coastal zone of Bangladesh? N. K. Saha1,2, M. K. Mondal1, E. Humphreys1, J. Bha acharya1,2, M. H. Rashid2, P. C. Paul3 and S. P. Ritu3,4 1 Interna onal Rice Research Ins tute, Philippines, n.saha@irri.org, m.mondal@irri.org, e.humphreys@irri.org, j.bha acharya@irri.org 2 Patuakhali Science and Technology University, Bangladesh, mhrashid_pstu@yahoo.com 3 Bangladesh Rice Research Ins tute, Bangladesh, plcpauliwm@yahoo.com 4 Current address: Sylhet Agricultural University, Bangladesh, sanjidap05@gmail.com Abstract The cropping intensity and produc vity of the coastal zone of Bangladesh is much lower than that of the country as a whole. In the coastal zone most farmers grow a single, low yielding, late maturity aman variety during the rainy season, and a large por on of the land is fallow for much of the dry season. There is a general percep on that during the dry season the river water is saline throughout the coastal zone; however, most of the rivers in the Barisal Division remain non-saline and suitable for rice irriga on throughout most or all of the year. Furthermore, there is a percep on that modern, high yielding and early maturity aman varie es cannot be grown in the coastal zone due to water stagna on. However, with separa on of lands of different eleva on using small levees, and by draining excess water from the fields at low de, modern aman varie es can be successfully grown. Given this and the recent availability of the first high yielding, short dura on aus variety for Bangladesh, it should be possible to intensify produc on to two and three high yielding crops per year in much of Barisal Division. Increasing produc vity of the coastal zone is now a high priority of the Government of Bangladesh. Studies were therefore conducted in Patuakhali District to determine the feasibility of growing three rice crops per year and to evaluate plan ng date and aman variety op ons. The experiment was conducted for two years (2012-14) with three aus sowing dates (10, 25 April and 1 May), four aman varie es (BRRI dhan33, BRRI dhan49, BRRI dhan52, BRRI dhan53), and five boro sowing dates (15, 20, 30 November and 05, 15 December) in 12 aus-aman-boro cropping system combina ons. Total in-field dura on of the systems ranged from 277 to 312 d, showing the feasibility of growing three rice crops per year. Total system produc vity ranged from 13.4 to 17.2 t/ha/yr; two to four mes that of current farmer prac ce. The results show that cul va on of three rice crops per year in the low salinity coastal zone is a feasible technology for greatly increasing produc vity of the coastal zone and contribu ng to the future food security of Bangladesh. Key message: Triple rice cropping with total system produc on of around 16 t/ha is possible in the coastal zone of Bangladesh in areas where there is year-round fresh water availability, separa on of lands of different eleva on (e.g. using large bunds) and ability to drain at low de. Keywords: cropping system intensifica on, Patuakhali, aus, aman, boro 1. Introduc on Rice is the main food staple and the dominant crop in Bangladesh, occupying nearly 80% of the total cropped area. Rice is grown throughout the year. The crop grown during April to July is known as aus, rice grown during July to December is called aman, and that grown from November to May is known as boro. Boro is grown under irrigated condi ons while the produc on of aus and aman is mostly rainfed. The main rainy or monsoon season (kharif 2) occurs from July to September/October, with significant pre-monsoon rain from May to June (kharif 1). There is very li le rain from November to March/April, the dry season. 421 While Bangladesh as a whole is currently self-sufficient in rice produc on this is not the case for the coastal zone (Tuong et. al 2014). Rivers are dal in the coastal zone with diurnal fluctua ons of two to three m. The salinity of the rivers increases during the dry season, more so closer to the coastline and in the southwest. However, most of the agricultural lands of the coastal zone are protected from flooding and saline water intrusion as a result of the construc on of polders. Despite this cropping intensity in this region is much lower than in other parts of the country. Over 50% of land remaining fallow during the dry season (BARC 2008) because of the lack (or perceived lack) of fresh water for irriga on, and because of soil salinity which increases as the dry season progresses. Most farmers in the coastal zone grow a single aman crop using tradi onal, tall varie es typically yielding only 2 to 3.5 t/ha. Old seedlings (up to 70 d) are typically transplanted in late July or August. Tall seedlings are required to survive stagnant flooding which occurs in the polders primarily as a result of rainfall, and this prac ce is common throughout the coastal zone. As the local varie es are photoperiod sensi ve, they are not ready for harvest un l December. The aman crop is o en followed by a low yielding crop such as keshari (grasspea) or sesame. In reality, there is abundant fresh water in the rivers throughout the year in significant areas of the coastal zone. This is par cularly true in much of Barisal Division and it is predicted that this will con nue to be the case even in the climate change scenario with a 22 cm mean sea level rise and a moderate precipita on change (Khan et al. these proceedings). Furthermore, there is a dense network of former river distributaries (khals or canals) inside the polders that can be used to deliver fresh water from the large rivers surrounding the polders. Despite this, the area under aus (0.27 Mha) and boro (0.14 Mha) cul va on is very low when compared with the area under aman (0.71 Mha) (N.K. Saha, pers. comm.). A major constraint to cropping system intensifica on is the low level of adop on of non-photoperiod sensi ve, high yielding aus and aman varie es with shorter dura on than tradi onal varie es. Modern aus and aman varie es cannot tolerate water stagna on, hence their low level of adop on. Thus while much of Bangladesh experienced a rapid increase in produc on as a result of the introduc on of input responsive, high yielding crop varie es, the development of irriga on (mostly groundwater) and cropping system intensifica on to two to four high yielding crops per year, the coastal zone missed out on this ‘Green Revolu on’. It is possible to grow modern aus and aman varie es with systema c opera on of the sluice gates to drain out excess water at low de, so in theory it should be possible to grow three rice crops per year in parts of the coastal zone where fresh water is available year-round. Ritu et al. (these proceedings) have demonstrated the possibility of intensifying to an aus-aman system using a short dura on, high yielding aus variety (HYV aus) and non-photoperiod sensi ve HYV aman without jeopardizing the yield of the aman crop. Furthermore, Sharifullah et al. (2009) have shown the feasibility of an HYV aman-boro system in the coastal zone, with sowing of the boro crop (mid-November) advanced in comparison with the common prac ce in northern Bangladesh to reduce irriga on water requirements. However, the feasibility of combining all three crops in a year has not been shown in the coastal zone. The general objec ve of the research presented here was to explore sowing date and aman variety combina ons for fi ng three rice crops together in a single year. The main hypotheses were that: (i) with the use of short dura on HYV aus, non-photoperiod sensi ve HYV aman, and a medium dura on boro variety it is possible to implement aus-aman-boro cropping systems with total system yield in excess of 15 t/ha, and (ii) boro yield and thus system produc vity will be higher with earlier establishment. 2. Methods 2.1 Experimental site A field experiment was conducted on the experimental farm of Patuakhali Science and Technology University, Dumki, Patuakhali-8602, Bangladesh from 2012 to 2014. The field was located at 22027'51.063"N la tude, 90022'56.873"E longitude and 3 m above mean sea level. The soil was a clay loam and the site had a cropping history of T.aman-fallow/khesari/mungbean-fallow. The field was in a low part of the farm and about 50 m from a dal canal whose water remained non-saline throughout the year. The field was surrounded with 422 bunds that were raised to 70 cm high and 50 cm wide to provide protec on from flooding that can occur when high rainfall results in runoff from surrounding lands. 2.2 Experimental design The experiment was designed to compare op ons for triple rice (aus-aman-boro) cropping systems. Aus crops were grown from April to July, aman from July to November and boro from November to April star ng with the 2012 aman crop. The experiment had an unbalanced factorial design with three aus establishment prac ces, four aman varie es and two boro sowing dates in a split-split plot design with four replicates. The experiment compared 12 cropping system (CS) op ons (Table 1, Fig. 1). The details of the treatments are: Three aus establishment prac ces (E) in main plots (10 m x 10 m): E1: Sown on 10 April, transplanted 1 May E2: Sown on 25 April, transplanted 15 May E3: Direct (wet) seeded on 1 May The Vietnamese variety OM1490 (moderately salt tolerant, short dura on - 100 d seed to seed) was used in all aus treatments. This variety was released in Bangladesh as BRRI dhan65 in October 2014. Two aman varie es (V) in sub plots (5 m x 10 m): For the first two aus treatments (E1 and E2) two medium dura on aman varie es were grown V1: BRRI dhan49 (dura on 135 d) V2: BRRI dhan52 (dura on 145 d, submergence tolerant) For the wet seeded aus (E3) two short dura on aman varie es were grown V3: BRRI dhan33 (dura on 118 d) V4: BRRI dhan53 (dura on 125 d) Transplan ng of the aman crops was not permi ed less than 10 d a er harvest of each aus treatment because farmers are busy at harvest and it takes me to prepare their fields for establishment of the next crop. Thus aman crops were planted within 12 to 18 d a er aus harvest. Two boro sowing dates (D) in sub-sub plots (5 m x 5 m): Within each aus-aman treatment combina on, the first boro sowing date was selected to ensure that transplan ng did not take place less than 10 d a er harvest of the aman crop, and not earlier than 15 November as previous research in Khulna (Mondal et al. 2010) had shown that earlier sowing led to serious yield loss due to cold damage during the reproduc ve stage. The second boro sowing date was 15 d a er the first. As a result, there were a total of five boro sowing dates ranging from 15 November to 15 December (Table 1). 423 Table 1. Cropping system treatments Aus establishment method/sowing date (E)1 Aman variety (V) V1: BRRI dhan49 E1: 10 April V2: BRRI dhan52 V1: BRRI dhan49 E2: 25 April V2: BRRI dhan52 V3: BRRI dhan33 E3: 1 May V4: BRRI dhan53 Boro sowing date (D) D1= 15 November D2 = 30 November D1= 15 November D2 = 30 November D1= 30 November D2= 15 December D1= 30 November D2= 15 December D1= 20 November D2= 05 December D1= 20 November D2= 05 December Cropping system (CS) CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 CS11 CS12 1 E1, E2 transplanted, E3 direct (wet) seeded Fig. 1. Proposed aus-aman-boro crop calendar. 2.3 Tillage and crop establishment All plots were plowed three to four mes (wet llage) using a power ller powered by a two-wheel tractor. Final land leveling with a wooden plank drawn by an ox was done a er basal applica on of fer lizer. Rice seeds were soaked for 12 to 24 h and incubated for 48 to 72 h for germina on. For the transplanted crops, the pre-germinated seeds were sown on the seed bed at 28 g/m2. Ten days a er emergence, urea was applied to the seedbed at 32 kg N/ha (BRRI 2011). Transplan ng was done when the seedlings had four leaves (usually about 21 d a er sowing, DAS), with two to three seedlings per hill, and a hill spacing of 20 cm x 20 cm in all crops. In the direct seeded aus, pre-germinated rice seeds were hand sown in lines on the surface of the puddled soil at a row spacing of 20 cm and a seed rate of 40kg/ha. 424 2.4 Fer lizer applica on Urea, triple superphosphate, muriate of potash, gypsum and zinc sulphate were applied to each crop as per BRRI (2010) recommenda ons. Urea was applied at 120 kg N ha -1 to the boro crops and at 100 kg N ha-1 to the aman crops in four equal splits: before the last leveling, 25 d a er transplan ng (DAT), 5 to 7 d before panicle ini a on (DBPI) and at heading. For the transplanted aus, urea was applied at 100 kg N ha-1 in three equal splits before the last leveling, 5 to 7 DBPI and at heading. For the direct seeded aus, urea was also applied at 100 kg N ha-1 but in four equal splits at 10 to 15 d a er emergence (DAE), 30 to 35 DAE, 45 to 50 DAE (5 to 7 DBPI) and 60 to 65 DAE (heading). All other fer lizers were applied at 60 kg P2O5, 40 kg K2O, 60 kg CaSO4 and 10 kg ZnSO4 per ha before the last leveling. 2.5 Water management The aus crops were established using supplemental irriga on as needed and therea er grown on rainfall, as with the establishment and growth of the aman crops. At mes of excessive rainfall during the aus and aman crops, water was drained from the experimental field by gravity and/or by pumping. Water was drained whenever the depth exceeded the plant height and the depth was lowered to 10 to 25 cm depending on the height of the crop. All boro crops were fully irrigated. Water was pumped from the nearby khal using a 4 horsepower pump powered by diesel. The boro rice was irrigated whenever water depth fell below 1 cm and water was added un l the ponded water depth reached 5 cm. Irriga on was stopped 10 to 15 d before harvest. The salinity of the irriga on water was always <1.0 dS/m. 2.6 Weed, pest and disease control Systemic pes cides (BRRI, 2011) were applied during each N applica on to all crops to prevent crop damage. When infesta ons occurred, recommended pes cides were sprayed. As a result, pest and disease control were good in all crops except for the early sown boro 2012-13 and all aman 2013 crops, which were heavily infested by brown spot disease. Brown spot was controlled in later sowings of boro 2012-13 and in all 2013-14 boro crops using recommended pes cides; however, it could not be controlled in the 2013 aman crops. Good weed control was achieved by applying a post-emergence weedicide (pyrazosulfuron-ethyl@125 g/ha) 3 to 5 DAT into standing water and the water was held on the plots for 2 to 3 d a er treatment. Manual spot weeding was also done prior to urea topdressing. The experimental site was surrounded by a plas c barrier with traps to prevent damage by rats, and covered by ne ng each season when the earliest crop approached maturity to prevent bird damage. Prior to installing the bird nets there was yield loss in the earliest sowing of the first aus crop due to bird a ack at the milky dough stage. 2.7 Data collec on Grain and straw yield of each plot were determined in a 3 m x 2 m (150 hills) area in the middle of each sub-sub-plot. The samples were threshed, cleaned and the grain and straw weighed. Grain moisture content was determined using a moisture meter and straw sub-samples were weighed and dried at 70oC to determine moisture content. Grain yield is presented at 14% moisture content and straw yield at 0% moisture content. 2.8 Data analysis The effect of the treatments on yield of the aus, aman and boro crops, and of the total cropping system, was compared using one-way ANOVA (12 cropping systems or “treatments”). For the first two aus establishment methods, a split-split plot design ANOVA was also used to determine if there were interac ons between aus establishment method, aman variety and boro sowing date. Least significant difference (lsd) at 5% probability was used to test for differences between means. The analyses were done using the STAR, version 2.0.1 (Sta s cal Tool for Agricultural Research) program developed by IRRI (STAR 2014). 425 3. Results and discussion 3.1 Dura on of crops in the main field: The dura on of each crop in the main field determines the number of crops that can be accommodated annually in the crop sequence. The crop dura on reported for all transplanted crops is from transplan ng to physiological maturity (PM, 80% of grains golden), in cases where seedlings are raised in a separate nursery. The crop dura on for the direct seeded aus crops is from seeding to PM. 3.1.1 Aman The dura on of BRRI dhan49 and 52 was usually in the range of 96 to 102 d, except for the first sowing in 2012 when dura on was 114 to 116 d (Table 2). The longer dura on of the first sowing in 2012 was probably because the crop was submerged twice (from 5 to 12 DAT and from 36 to 39 DAT) as a result of heavy rainfall and flooding from surrounding areas prior to raising the height of the surrounding bund to 70 cm. The second sowing of these varie es was also submerged from 22 to 25 DAT in 2012, which may explain the fact that dura on was 5 d longer in the first year than the second year. There was no submergence of the third sowing in 2012 and no submergence of any aman crops in 2013. The dura on of BRRI dhan33 sown on 30 July was only 86 and 83 d in the first and second years, respec vely, 9 d less than that of BRRI dhan53 sown on the same date each year. The reason for the slightly shorter dura on (by 3 to 4 d) of the short dura on varie es in the second year is not known. Mean daily temperature from sowing to transplan ng was similar throughout the growing season each year. The dura on of all aman varie es in the absence of submergence was generally similar to that observed by others in the region. Roy et al. (2003) found 120 d dura on of BRRI dhan33 transplanted as 30 d old seedlings in August in Barisal district, about a week longer than in our study. However, this work was done at BRRI Regional Sta on at Barisal, which is affected by dal flooding. The dura on of BRRI dhan52 ranged from 111 to 121 d in the studies of Sharma et al. (2013) at Amtali (Barguna District), similar to the values in our study. For BRRI dhan53, dura on was about 10 d earlier than that observed by others (118 to 127 d) (Sharma et al. 2013; Islam and Gregorio 2013). Possible reasons for the longer dura on in other studies include periods of submergence or water stagna on (cri cal informa on which is o en not reported in variety evalua on trials) and the method used to determine dura on, in par cular maturity date (physiological maturity as in our study, or “harvest maturity” when the grain is considered dry enough for the farmers to harvest), informa on which is also seldom reported. 426 Table 2. Dura on of crops from establishment to PM during 2012-13 and 2013-14 (days) 2012-13 Cropping Treatment System combina ons1,2 (CS) Aman Boro CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 CS11 CS12 119 116 119 116 116 109 116 109 116 115 116 115 E1V1D1 E1V1D2 E1V2D1 E1V2D2 E2V1D1 E2V1D2 E2V2D1 E2V2D2 E3V3D1 E3V3D2 E3V4D1 E3V4D2 116 114 101 102 86 95 2013-14 Aus 77 78 96 Total system Aman dura on Boro 312 309 310 307 295 288 296 289 298 297 307 306 119 109 119 109 109 105 109 105 118 106 118 106 101 101 96 97 82 91 Aus 74 75 91 Total system dura on 294 284 294 284 280 276 281 277 291 279 300 288 1 E = aus sowing date (E1: 10 April, E2: 25 April, E3: 1 May); V = aman variety (V1: BRRI dhan49, V2: BRRI dhan52, V3: BRRI dhan33, V4: BRRI dhan53); D = boro sowing date (D1: 10 d a er aman harvest, D2: 25 d a er aman harvest) 2 E1 and E2 transplanted, E3 wet seeded 3.1.2 Boro Boro crop dura on was in the range of 105 to 119 d and declined with delay in sowing, more so in the second year (Fig. 2). The decrease in dura on with delay in sowing was due to higher temperatures during the vegeta ve period. The dura on of the 30 Nov/early Dec sowings was 7 to 9 d longer in 2012-13 than in 2013-14 and this was associated with much lower temperatures from late November to mid-January in the second year (Fig. 3). In a saline loca on (BRRI Regional Sta on in Satkhira), Salam et. al. (2010) found that dura on of BRRI dhan 28 was longer (136 d) than in our study. Islam and Gregorio (2013) reported even longer dura on (140 d) of BRRI dhan28 in par cipatory variety selec on (PVS) trials at several sites in salinity prone coastal zone districts, but this was during the wet season and the method for determina on of dura on was not reported. 140 120 100 80 60 2012-13 2013-14 40 20 0 0 5 10 15 20 25 30 35 40 45 50 Sowing date (Days a er 01 November) Fig. 2. Effect of sowing date on boro crop dura on. 427 35 30 25 20 Boro 2012-13 15 Boro 2013-14 10 2-Nov 22-Nov 12-Dec 1-Jan 21-Jan 10-Feb 2-Mar 22-Mar 11-Apr 1-May 21-May Fig. 3. Daily mean temperature during boro 2012-13 and 2013-14. 3.1.3 Aus There was li le varia on in the dura on of the aus crops, regardless of seeding date, establishment method and year (Table 2). Transplanted crops took 98 and 95 d to mature (seed to seed) in 2013 and 2014, respec vely, whereas direct seeded treatments took 96 and 91 d, respec vely. The slightly longer dura on in 2013 was probably due to slightly lower temperatures during the grain filling stage than in 2014. Dura on of the same variety (OM1490) in the Mekong delta of Vietnam was reported to be shorter, at 85 to 90 d (Lang et al. 2010). 3.1.4 Total crop dura on in the main field The total dura on of crops in the main field ranged from 288 to 312 d in 2012-13, and from 277 to 300 d in 2013-14 (Table 2). However, it is usually be er for farmers to allow their crops to dry in the field for a few days (the me required depending on weather at the me of PM) before harves ng. Therefore, a more realis c es mate of dura on in farmers’ fields would be from about 290 to 330 d. This allows for an average turn-around between crops of about 12 to 25 d, which suggests that all sowing date x variety combina ons evaluated in this experiment are feasible for farmers. The shortest in-field dura on each year was with transplanted aus sown on 25 April followed by BRRI dhan49 or 52, and with the later boro sowing (15 December), systems CS6 and 8. There were no consistent trends for the longest dura on cropping system, partly because of submergence of the early aman crops in 2012. 3.2 Grain yield 3.2.1 Aman Aman yields ranged from 5 to 6 t/ha in 2012 and from 4.5 to 5.3 t/h in 2013. Yields of all sowing date x variety combina ons were lower in 2013 than 2012 due to a severe a ack of brown spot. In 2012, there was a consistent trend for higher yield of earlier sown BRRI dhan49 and 52 (CS1, 3) than for 15 d later sowings (CS2, 4), with significantly higher yield of early BRRI dhan49 than both varie es sown later (Table 3). However, yields of both these variety x sowing date combina ons were similar in 2013. Yield of BRRI dhan33 sown on 1 August was significantly lower than yield of all other variety x sowing date combina ons in 2012, and of BRRI dhan49 and 52 for both sowings in 2013. High yields of BRRI dhan49 and 52 were achieved despite inunda on twice for 4 to 8 d within the first 10 weeks a er the first transplan ng in 2012, and for 4 d during the third week a er the second transplan ng. Yields were similar to those achieved with these varie es under favourable condi ons at Mymensingh, where 428 Shormy (2012) reported a yield of 5.5 t/h for BRRI dhan49, and Uddin (2013) reported a yield of 5.4 t/ha for BRRI dhan52. The results demonstrate the feasibility of achieving high yields of modern aman varie es provided that excess water is drained rapidly following inunda on as a result of heavy rainfall. The results also demonstrate the importance of separa ng lower and higher lands to protect lower lands from inunda on from surrounding lands. Yields of the shorter dura on variety BRRI dhan53 in our experiment (4.7 to 5.4 t/ha) were similar to or higher than mean yields reported from PVS trials in salinity prone coastal districts (4.8 t/ha, Islam and Gregorio 2013) and in a low salinity region of the coastal zone (4.7 t/ha, Sharma et al. 2013). Yields of an even shorter dura on variety (BRRI dhan33, 4.5 to 5.0 t/ha) tended to be lower than those of BRRI dhan53, with some significant differences each year. Yields of BRRI dhan33 were similar to yields obtained in on-sta on experiments in favourable environments at Mymensingh (5.0 t/ha, Rahman et al. 2004), Madaripur (5.4 t/ha, Quddus et al. 2012) and Dhaka (4.6 t/ha, Masum et al. 2013). Table 3. Grain yield of component crops and the total system during 2012-13 and 2013-14, as affected by aus establishment method (E), aman variety (V) and boro sowing date (D) Cropping System (CS) CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 CS11 CS12 LSD (p=0.05) C.V. (%) Treatment combinaons1,2 E1V1D1 E1V1D2 E1V2D1 E1V2D2 E2V1D1 E2V1D2 E2V2D1 E2V2D2 E3V3D1 E3V3D2 E3V4D1 E3V4D2 2012-13 yield (t/ha) Aman 6.0 5.8 5.6 5.5 5.0 5.4 0.4 4.6 Boro Aus 4.9 5.6 5.0 5.4 6.0 6.3 5.6 5.9 4.9 5.9 5.1 5.9 0.4 5.2 4.1 4.1 4.0 3.8 3.8 3.6 3.7 3.8 3.6 3.5 3.5 3.5 NS 9.1 1st year system yield (t/ha/yr) 15.0 15.7 14.7 15.0 15.3 15.4 14.9 15.2 13.4 14.4 14.0 14.8 0.9 4.2 2013-14 yield (t/ha) Aman Boro Aus 5.2 5.1 4.9 5.1 5.2 5.3 5.2 5.0 4.6 4.5 4.7 4.8 0.3 4.0 6.1 7.3 6.2 7.2 7.4 6.8 7.4 7.4 7.1 7.5 7.0 7.3 0.5 5.3 3.7 3.4 3.9 3.6 4.3 4.5 4.6 4.4 4.4 4.2 4.2 4.2 0.5 9.2 2nd year system yield (t/ha/yr) 15.1 15.8 15.1 15.8 16.9 16.6 17.2 16.8 16.1 16.2 15.8 16.3 0.9 3.9 1 E = aus sowing date (E1: 10 April, E2: 25: April, E3 1 May); V = aman variety (V1: BRRI dhan49, V2: BRRI dhan52, V3: BRRI dhan33, V4: BRRI dhan53); D = boro sowing date (D1: 10 d a er aman harvest, D2: 25 d a er aman harvest) 2 E1 and E2 transplanted, E3 wet seeded 3.2.2 Boro There was a consistent trend for increasing yield with delay in sowing of boro from mid-November to late November/early December (Fig. 4). Yield was much lower in the first year than the second year for all sowing dates because of the long cold period and serious a ack of brown spot disease. It is a common observa on that stressed plants are highly vulnerable to brown spot. In the second year brown spot was not a problem. The higher yield in the second year was associated with higher panicle density for all but the last (15 Dec) 429 sowing, and more so for earlier sowings. The higher yield in the second year was also associated with more florets per panicle (for all but the first sowing), and more so for later sowings. Boro yields were in the range of those observed by others. Salam et. al. (2010) reported yield of BRRI dhan28 of 5.7 t/ha at a saline loca on in Satkhira in 2006/7, while Mondal et al. (2010) obtained 4.6 and 5.0 t/ha for 15 and 7 November sowings in the same year, at a saline site in Khulna. In the previous year at the same site, yields ranged from 5.0 to 5.6 t/ha for the same sowing dates. Islam and Gregorio (2013) reported a yield of 7.6 t/ha in PVS trials in the southwest coastal zone during the 2009 and 2010 wet seasons. 8 T 6 LSD 0.05 4 2012-13 2013-14 2 0 0 5 10 15 20 25 30 35 40 45 50 Sowing date (Days a er 01 November) Fig. 4. Effect of sowing date on yield of boro (ver cal bar is lsd (p=0.05) for the interac on between year and sowing date). 3.2.3 Aus Aus yield ranged from 3.5 to 4.1 t/ha (Table 3), similar to the yields of BRRI dhan65 obtained by Ritu et al. (these proceedings) in a medium salinity area of Khulna when water was not limi ng. In Vietnam, Lang et al. (2010) reported a yield of 5.6 t/ha despite the shorter dura on in that environment. Das (2008) reported the yield of OM 1490 (BRRI dhan65) was 3.6 t/ha with the growth dura on of 91 d at Kismat Fultola and recommended to incorporate this variety for increasing cropping intensity in the coastal zone. In the first year of the present study, there was a trend for aus yield to decline as sowing was delayed, but with no significant differences. In 2014, yield of the first sowing was significantly lower than yields of the other two sowing dates because of bird a ack at the milky dough stage (despite the use of net, small birds finds ways to ensure their food security), resul ng in low floret fer lity and low grain weight. 3.2.4 Total system yield Total system yield ranged from 13.4 to 15.7 t/ha/yr in 2012-13 and from 15.1 to 17.2 t/ha/yr in 2013-14 (Table 3). There was no correla on between yield of respec ve systems in the first and second years. However, yield of systems with transplanted aus seeded on 25 April (CS5-8) was significantly higher than yield of most other systems in the second year and among the highest yielding systems in the first year. In 2012-13 there were no interac ons between aus establishment treatment, aman variety (BRRI dhan49, BRRI dhan52) and boro sowing date on system yield. However, in 2013-14 there was a significant aus X boro treatment interac on (Table 4). With early aus sowing (10 April), system yield was less with early boro sowing than sowing that occurred 15 d later. However, with later aus sowing (25 April) system yield was not affected by boro sowing date. 430 Table 4. Interac on between aus establishment prac ces (E) with boro sowing dates (D) of the eight cropping systems during 2013-14 Aus establishment method (E) E1 E2 L.S.D. (P=0.05) Boro sowing date (D) Grain yield (t/ha/yr) Straw yield (t/ha/yr) D1 D2 D1 D2 15.1 15.8 13.7 12.5 17.1 16.7 12.1 12.3 0.6 0.8 3.3 Straw yield 3.3.1 Aman Straw yield ranged from 4.8 to 6.2 t/ha in 2012, and from 3.1 to 5.0 t/ha in 2013 when the crop was affected by brown spot disease. In 2012 straw yield of BRRI dhan49 for both sowings and BRRI dhan52 for the second sowing was significantly higher than that of most other aman variety x sowing date combina ons (Table 5). In the first sowing, a er submergence for 8 d, addi onal muriate of potash (20 kg K2O/ha) was applied. This may be the cause of the higher straw yield of BRRI dhan52 for the first sowing, as evidenced by higher llering of the first sowing than the second sowing (data not presented). In 2013, straw yield of the later sown shorter dura on varie es (BRRI dhan33 and 53) was significantly lower than that of all other treatments. Aman straw yield of all varie es was lower in 2013 than 2012 by up to 2 t/ha, probably due to brown spot infesta on. In the favourable environment of Mymensingh, Shormy (2012) found straw yield of BRRI dhan49 to be 4.3 to 5.8 t/ha, slightly lower than the range in our experiment (4.6 to 6.2 t/ha). The higher straw yields in the first year of our experiment were probably due to submergence leading to taller plant height and longer growth dura on. Also at Mymensingh, Uddin (2013) reported straw yield of BRRI dhan52 of 5.2 t/ha, compared with values of 4.8 to 6.6 t/ha in 2012 and 4.3 to 5.1 t/ha in 2013 in our experiment. Straw yield of BRRI dhan33 in 2012 was 5.2 t/ha, and only 3.1 to 3.4 t/ha in 2013 due to brown spot disease. Reports of straw yield of BRRI dhan33 in the literature also range widely, from 3.4 to 5.9 t/ha in the favourable on-sta on environments at Madaripur and Mymensingh (Rahman et al. 2004; Quddus et al. 2012) 431 Table 5. Straw yield of component crops and the total system during 2012-13 and 2013-14, as affected by aus establishment method, aman variety and boro sowing date Cropping System (CS) Treatment combinaons1,2 CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 CS11 CS12 LSD (p=0.05) C.V. (%) E1V1D1 E1V1D2 E1V2D1 E1V2D2 E2V1D1 E2V1D2 E2V2D1 E2V2D2 E3V3D1 E3V3D2 E3V4D1 E3V4D2 2012-13 Straw yield (t/ha) Aman Boro Aus 6.0 4.2 4.5 4.2 3.7 4.0 4.5 4.0 4.5 3.8 3.9 4.1 4.3 NS 15.1 3.6 3.3 3.7 3.3 4.0 3.3 4.0 3.5 2.9 3.5 3.2 3.5 NS 15.5 6.6 6.2 4.8 5.2 4.9 1.0 12.1 2013-14 Straw yield (t/ha) 1st year 2nd year system system straw yield Aman straw yield Boro Aus (t/ha/yr) (t/ha/yr) 13.8 13.7 14.5 13.6 14.2 14.1 12.8 12.8 11.9 12.6 12.3 12.7 1.6 8.3 4.6 4.9 4.3 5.0 4.7 5.0 5.1 4.3 3.4 3.1 3.6 3.5 0.8 12.3 5.7 4.6 6.0 4.3 4.5 4.3 4.0 4.7 4.9 4.1 4.8 3.6 0.9 13.1 3.3 3.0 3.6 3.1 3.1 3.1 2.9 3.1 4.0 3.6 3.3 3.4 NS 20.0 13.5 12.5 13.9 12.4 12.3 12.4 12.0 12.1 12.3 10.7 11.7 10.4 1.5 8.4 1 E = aus sowing date (E1: 10 April, E2: 25: April, E3 1 May); V = aman variety (V1: BRRI dhan49, V2: BRRI dhan52, V3: BRRI dhan33, V4: BRRI dhan53); D = boro sowing date (D1: 10 d a er aman harvest, D2: 25 d a er aman harvest) 2 E1 and E2 transplanted, E3 wet seeded 3.3.2 Boro There was a significant interac on between year and sowing date on straw yield of boro (Fig. 5). In 2012-13, straw yield of boro was similar for all sowing dates. However, in 2013-14, straw yield of early (15 November) sown BRRI dhan28 was significantly higher than that of all other boro sowing dates in that year, and of all sowing dates in the previous year. 7 LSD 0.05 6 5 4 3 2 2012-13 1 2013-14 0 0 10 20 30 Sowing date (Days a er 01 November) 40 50 Fig. 5. Effect of sowing date on straw yield of boro (ver cal bar is lsd (p=0.05) for the interac ons between year and sowing date. 432 3.3.3 Aus Straw yield of aus was not affected by sowing date or establishment method in either year, and was similar across years (Table 5). 3.3.4 Total system straw yield Total system straw yield ranged from 11.9 to 14.5 t/ha in 2012-13, and from 10.4 to 13.9 t/ha in 2013-14 (Table 5). In both years, BRRI dhan49 (both sowing dates) tended to have the highest straw yield, followed by BRRI dhan52 for the first sowing. In 2012-13, there were no interac ons between aus establishment treatment (E1, E2), aman variety (BRRI dhan49, BRRI dhan52) and boro sowing date on system straw yield. In 2013-14, there was a significant interac on between aus establishment method and boro sowing date on straw yield (Table 4). Highest yield was obtained for early sown aus (10 April) combined with early sown boro than for all other combina ons. With later aus sowing, system straw yield was not affected by boro sowing date. 4. Conclusions and recommenda ons The results show that aus-aman-boro cropping systems using HYV are feasible in areas of the coastal zone where there is year-round availability of fresh water, provided that there is separa on of lower and higher lands, and that drainage of the aman and aus crops can be implemented as needed. Total system yield ranged from 13.4 to 17. 2 t/ha in the 12 systems over two years, but most treatment combina ons yielded in excess of 15 t/ha each year. Thus, there can be considerable flexibility in sowing date and aman variety in successfully implemen ng a triple rice system. However, successful implementa on requires community coordina on; both for water management and for cropping system synchroniza on, as isolated crops and small fields maturing earlier than other crops in the landscape are highly vulnerable to a ack from pests such as rats and birds. Op miza on of the systems for different regions requires further agronomic trials in a few strategic loca ons, combined with crop modeling and par cipatory farmer evalua on. This will need to include development of environmentally friendly and sustainable methods of disease and pest control. Acknowledgements The work presented in this paper was part of G2 ‘Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food (CPWF) and the CSISA Bangladesh project, funded by USAID. We are grateful to Akbar Ali, Jahid Hasan and Elias Hossain (BRAC staff) for their assistance in conduc ng the study in the field and laboratory, and the farm manager (Mr. Moniruzzaman) of PSTU for necessary help in field prepara on. References Bangladesh Agricultural Research Council (BARC). 2008. Es ma on of Fallow Land Area and Crop Produc on Plan for Barisal, Pirojpur, Jhalaka , Bhola, Patuakhali and Bagerhat Districts. A report submi ed to the ministry of agriculture, govt. of Bangladesh. May 2008. (In Bangla). Bangladesh Bureau of Sta s cs (BBS). 2008. Yearbook of Agricultural Sta s cs of Bangladesh. Ministry of Planning, Govt. of People's Republic of Bangladesh, Dhaka. Bangladesh Bureau of Sta s cs (BBS). 2009. Yearbook of Agricultural Sta s cs of Bangladesh. Ministry of Planning, Govt. of People's Republic of Bangladesh, Dhaka. BRRI (Bangladesh Rice Research Ins tute). 2011. Adhunik Dhaner Chash. Dhaka: MRI prin ng press. 433 Cereal Systems Ini a ve for South Asia (CSISA) in Bangladesh. 2011. Annual progress report. Submi ed to United States Agency for Interna onal Development (USAID) Dhaka, Bangladesh. Submi ed by Interna onal Rice Research Ins tute (IRRI), WorldFish Center (WF) and Interna onal Maize and Wheat Improvement Center (CIMMYT). November 2011. Das, D. 2008. Yield performance of six Vietnam rice varie es in the coastal saline environments of Khulna district. MS diss., Khulna University, Bangladesh. Hamid A. 2010. Rice produc on in dal ecosystem: farmers fail to harness the benefits of green revolua on, pp 1-7. In Crop Produc on in the Coastal Ecosystem - Challenges and Opportuni es. (ed. M. Jafar Ullah). Agrarian Research Founda on, Dhaka. November 2010. Islam M.R. and G.B. Gregorio. 2013. Progress of salinity tolerant rice variety development in Bangladesh. SABRAO J. Breed. Genet 45 (1): 21-30. Jasim uddin Md,. 2013. Fer lizer management in three modern T. Aman rice varie es. MS diss., Bangladesh Agricultural University, Mymensingh. Lang N.T., B.C. Buu, N.V. Vite and A.M. Ismail. 2010. Strategies for improving and stabilizing rice produc vity in the coastal zones of the Mekong Delta, Vietnam. Tropical deltas and coastal zones: food produc on, communi es and environment at the land-water interface. ed. Hoanh C. T., Szuster B. W., Suan-Pheng K., Inmail A. M. and Noble A. D., 9:209-222. UK: CPI Antony Rowe Ltd. Masum, S.M., M.H. Ali, M.S.H. Mandal, I.F. Chowdhury and K. Parveen. 2013. The effect of nitrogen and zinc applica on on yield and some agronomic characters of rice cv. BRRI dhan33. Intl. Res. J. Appl. Basic. Sci. 4 (8): 2256-2263 Mondal M.K., T.P. Tuong, A.K.M. Sharifullah and M.A. Sa ar. 2010. Water supply and demand for dry-season rice in the coastal polders of Bangladesh. Tropical deltas and coastal zones: food produc on, communi es and environment at the land-water interface. ed. Hoanh C.T., Szuster B.W., Suan-Pheng K., Inmail A.M. and Noble A.D., 9: 264-278. UK: CPI Antony Rowe Ltd. Mondal, M.K., T.P. Tuong, A.K.M. Sharifullah and M.A. Sa ar. 2010. Water Supply and Demand for Dry-season Rice in the Coastal Polders of Bangladesh. pp. 264-265. Quddus M. A., M. H. Rashid, M. A. Hossain, H. M. Naser and J. Abedin Mian. 2012. Integrated nutrient management for sustaining soil fer lity through chickpea-mungbean-t.aman cropping pa ern at madaripur region. Bangladesh J. Agr. Res. 37(2): 251-262. Rahman M. S., M. A. Haque and M. A. Salam. 2004. Effect of different llage prac ces on growth and yield contribu ng characters of transplanted Aman Rice (BRRI dhan33). J. Agron. 3(2): 103-110. Rahman Md. M. 2012. Enhancement of resilience of coastal community in Bangladesh through crop diversifica on in adapta on to climate change impacts. Masters diss. BRAC University. Ritu, S.P., M.K. Mondal, T.P.Tuong, S.U. Talukdar, E. Humphreys. 2015. An aus-aman system for increasing produc vity of a moderately saline region of the coastal zone of Bangladesh. In “Revitalizing the Ganges Coastal Zone”. Eds E. Humphreys et al. CPWF Conference Proceedings, 21-23 October 2014, Dhaka (these proceedings). Salam M.A., M.R. Islam, M. S. Rahman, M.A. Rahman, M.A.R. Bhuiyan, Z.I. Seraj, T.L. Aditya, M.K. Uddin, M.K. Mondal, A.M. Islam, D.L. Adoarda, R.D. Mendoza, E.B. Tumimbang-Raiz and G.B. Gregorio. 2010. Rice varie es and cultural management prac ces for high and sustained produc vity in coastal wet lands of southern Bangladesh. Tropical deltas and coastal zones: food produc on, communi es and environment at the land-water interface. ed. Hoanh C. T., Szuster B.W., Suan-Pheng K., Inmail A.M. and Noble A.D., 9: 183-198. UK: CPI Antony Rowe Ltd. 434 Sharifullah, A., Tuong, T.P., Mondal, M.K., and D.T. Franco. 2009. Assessing water supply and demand for dry season rice in coastal polders of Bangladesh. In: Humphreys, E. and R.S. Bayot (eds.). Increasing the produc vity and sustainability of rainfed cropping systems of poor smallholder farmers. Proceedings of the CGIAR Challenge Program on Water and Food Interna onal Workshop on Rainfed Cropping Systems, Tamale, Ghana, 22-25 September 2008. The CGIAR Challenge Program on Water and Food, Colombo, Sri Lanka. Sharma N., M.R.A. Sarkar, M.A. Rahman and M.R. Islam. 2013. Varietal evalua on of rice for improving produc vity in southern Bangladesh. Interna onal journal of Bioresearch 15(4): 7-13. Shormy, S.A.S .2012. Organic nutrient management on yield, nutrient content and their uptake by T. Aman rice. MS diss.,Bangladesh Agricultural University, Mymensingh. SRDI (Soil Resources Development Ins tute). 2010. Saline soils of Bangladesh. SRMAF project, Ministry of Agriculture, Govt. of Bangladesh. Dhaka: SFSDP-SRDI Publica on. STAR. 2014. Biometrics and Breeding Informa cs, PBGB Division, Interna onal Rice Research Ins tute, Los Baños, Laguna. Available at h p://bbi.irri.org/. Tuong T.P., E. Humphreys, Z.H. Khan, A. Nelson, M. Mondal. M. Buisson and P. George. 2014. Messages from the Ganges Basin Development Challenge: Unlocking the Produc on Poten al of the Polders of the Coastal Zone of Bangladesh through Water Management Investment and Reform. The CGIAR program on Challenge Program for Water and Food. 435 Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh J. Bha acharya1,2, M. K. Mondal2, E. Humphreys2, N. K. Saha1,2, M. H. Rashid1, P. C. Paul3 and S. P. Ritu3,4 Patuakhali Science and Technology University, Bangladesh mhrashid_pstu@yahoo.com Interna onal Rice Research Ins tute, Bangladesh and Philippines j.bha acharya@irri.org, m.mondal@irri.org, e.humphreys@irri.org, n.saha@irri.org 3 Bangladesh Rice Research Ins tute, Bangladesh plcpauliwm@yahoo.com 4 Current address: Sylhet Agricultural University, Bangladesh sanjidap05@gmail.com 1 2 Abstract Cropping intensity in the coastal zone of Bangladesh (173%) is lower than the na onal average of 199%. In the coastal zone, farmers typically grow a single crop of rice (aman) during the rainy season using low yielding and late maturing local varie es, some mes followed by a low input and low yielding pulse or sesame crop. The possibility of increasing produc vity by a growing high yielding, short dura on and salt tolerant aus variety prior to establishing high yielding aman varie es, without jeopardizing yield of the aman crop, has been previously shown. The work here sought to build on this by evalua ng the feasibility of intensifying to an aus-aman-rabi cropping system using a short dura on aus variety (BRRI dhan65), a medium dura on, high yielding aman variety (BRRI dhan44), and high yielding or high value rabi crops (maize - Pacific984; sunflower Hysun33). This involved conduc ng a replicated field experiment for two years at Patuakhali with the objec ves of evalua ng the effects of aus sowing date (20 March, 5 and 20 April, 5 May), rabi crop species (maize, sunflower) and mulch (0, 5 t/ha of rice straw during the rabi crops) on the feasibility and the produc vity of the aus-aman-rabi system. The longest total system in-field crop dura on was about 330 d. This allowed for an average turn-around me between crops of about 12 d, thus demonstra ng the feasibility of intensifica on to an aus-aman-rabi system for a range of crop establishment dates. Total rice equivalent yield ranged from 13.9 to 19.3 t/ha/yr. Rabi crop yield was the main determinant of trends in total system rice equivalent yield, which declined as establishment date of the rabi crops was delayed beyond 15-30 December. The rabi crops required irriga on, with water pumped from a nearby dal khal in which salinity of the water remained below 1 dS/m throughout the two years. The aus and aman crops were grown on rainfall throughout the year except for the 20 March sown aus crop, which required irriga on for puddling and transplan ng. This work shows that it is possible to implement highly produc ve aus-aman-rabi cropping systems in areas of the coastal zone of Bangladesh protected from dal flooding (as in the polders), where there is fresh water in the rivers year-round, and where the sluice gates are managed to drain excess water at low de as needed during the rainy season to enable the produc on of an early maturity, high yielding aman variety. Key message: It is possible to intensify to rice-rice-rabi systems producing around 18 t/ha/yr of rice equivalent yield in low salinity regions of the coastal zone of Bangladesh where drainage is possible during the rainy season. Keywords: intensifica on, drainage, water management, maize, sunflower 1. Introduc on More than 30% of the cul vable land in Bangladesh is in the coastal zone (SRDI, 2010). However, cropping intensity, at 173%, is lower than the na onal average of 199% (BBS, 2011). In the coastal zone farmers typically grow a single crop of rice (aman) during the rainy season using tall, late planted and low yielding local varie es that are photoperiod sensi ve and thus do not mature un l December. Unlike the rest of 436 Bangladesh, farmers in the coastal zone have not widely adopted modern high yielding rice varie es (HYV) because their shorter stature makes them unsuitable for the high water depths that o en prevail in the coastal zone. While about 1 Mha of the coastal zone is protected from diurnal dal flooding by embankments, much of this land experiences water stagna on (prolonged periods of flooding at depths of 20 to 50 cm) during the rainy season as a result of heavy rainfall and/or mis-management of the sluice gates connec ng the canals inside the polders to the dal rivers surrounding them. Further, the lack of separa on of higher and lower lands within the polders exacerbates the problem of water stagna on in the lowest lands. In the coastal zone the local aman crop is some mes followed by a low input, low yielding relay-sown legume (common in parts of Barisal Division) or a late-sown (mid-February to early March) sesame or mungbean crop (common in parts of Khulna Division). The late-sown sesame and mungbean crops are o en damaged by the early kharif rains and cyclonic events that are common (every three to four years) in May. The late harvest of the local aman varie es also prevents the cul va on of high yielding rabi crops such as maize, sunflower and wheat. Thus large areas of land (approximately 810,000 ha) lie fallow during the dry season (Hasan et al. 2013) due to the late aman harvest, the lack of ready access to fresh water, and increasing soil salinity as the dry season progresses. In reality, fresh water is abundant in much of the south-central coastal zone (Barisal Division) throughout the year (Khan et al., these proceedings). Here, the dominant cropping pa erns are: pulse-fallow-T.Aman (55%), fallow-fallow-T. Aman (20%) and pulse-T. Aus-T. Aman (25%) (BRRI, 2012). Most farmers use local aman varie es yielding 2-3 t/ha of rice and those who grow aus harvest an addi onal 3-4 t/ha/yr. The main hypothesis of the work presented in this paper is that with good water management (especially drainage, but also irriga on during the dry season) the use of modern HYVs with earlier maturity than local rice varie es, it is possible to greatly increase cropping system produc vity in low salinity regions of the coastal zone. More specifically, we hypothesized that it is possible to implement highly produc ve aus-aman-rabi cropping systems using a short dura on, salt tolerant high yielding aus variety, a medium dura on, high yielding aman variety, and high yielding hybrid maize or sunflower crops. We also hypothesized that the produc vity of an aus-aman-rabi system is likely to be affected by the establishment date of each crop, and in par cular the establishment date of the rabi crops, as yield of some rabi crops declines with delay in sowing a er mid-November/early December (Shahadat and Rahman 2012; Rashid et al. 2014). However, earlier sowing of rabi crops also means earlier sowing dates for all other crops in the system. Therefore, our third hypothesis was that high yield could be maintained with early establishment of aus and aman crops. Our fourth hypothesis was that the later the establishment of the rabi crop, the greater the requirement for irriga on during the dry season, and the greater the risk of damage from early kharif rains or cyclones. It is well established that mulching reduces soil evapora on and thus the rate of soil drying so we also hypothesized that this would reduce the irriga on requirement for the rabi crops. Therefore, we conducted a cropping system experiment with the objec ve of evalua ng the effects of aus sowing date, rabi crop species and mulching on: (i) the feasibility (of successfully growing three crops in succession in a single year), and (ii) the produc vity of an aus-aman-rabi system in the south-central coastal zone of Bangladesh. 2. Methodology 2.1. Experimental site An aus-aman-rabi field experiment was conducted on the research farm of Patuakhali Science and Technology University, Bangladesh, star ng with the aman 2012 crop and ending with the aus 2014 crop. The experimental field was located at 22 46‘38.02‘’ N la tude and 90 38‘74.5’’ E longitude at an al tude of 3 m above sea level. The field was situated about 50 m from a canal connected to the river Payra, which enabled drainage of excess water at low de and provided a source of irriga on water during the dry season. Salinity of the canal water was always <1 dS/m. The soil at the experimental site is a clay loam and the site had a cropping history of T.aman-fallow/khesari/mungbean-fallow. 437 2.2. Experimental design The experiment compared aus-aman-rabi cropping system op ons. Four aus sowing dates (main plots), two rabi crops (sub-plots), and two rabi mulch treatments (sub-sub-plots) were evaluated in a split plot design with four replicates. Thus a total of 16 cropping system combina ons were compared (Table 1, Fig. 1). Sub-sub-plot size was 6 m x 5 m. The treatments were: Aus sowing date A1 = 20 March A2 = 05 April A3 = 20 April A4 = 05 May Rabi crop species C1 = maize C2 = sunflower Rabi mulch treatment M1= rice straw mulch at 5 t/ha M2 = no mulch The aus variety was OM 1490, a short dura on (~100 d) high yielding and salt tolerant Vietnamese rice variety that was recently (October 2014) released in Bangladesh as BRRI dhan65. The aman variety was BRRI dhan44, with a growth dura on of ~145 d. The aman crops were planted not less than 10 d a er harvest of each aus crop, as farmers are busy at the me of harvest and it takes me to prepare the land for establishment of the next crop. This resulted in aman sowing dates of 20 June, 05 and 20 July, and 05 August in A1-A4, respec vely (Table 1). Hybrid maize (Pacific-984, dura on 120-130 d) and sunflower (Hysun33, dura on 100-110 d) were grown in the rabi season. The maize and sunflower were established not less than 10 d a er harvest of the aman crop, which resulted in target sowing dates of 30 November, 15 and 30 December and 15 January in A1-A4, respec vely (Fig. 1). However, in the first year, the first and second rabi sowings were delayed to 15 and 20 December due to waterlogging at the desired me of sowing. 438 Table 1. The 16 aus-aman-rabi cropping system combina ons Cropping system (CS) CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 CS11 CS12 CS13 CS14 CS15 CS16 1 Rabi Aus sowing date (A) Aman Sowing date A1 20 March >10 d a er aus harvest >10 d a er aman harvest A2 05 April >10 d a er aus harvest >10 d a er aman harvest A3 20 April >10 d a er aus harvest A4 05 May C1 Maize >10 d a er aus harvest M1 M2 M1 M2 M1 M2 M1 M2 M1 M2 M1 M2 M1 M2 M1 M2 C2 Sunflower C1 Maize C2 Sunflower C1 Maize C2 Sunflower >10 d a er aman harvest C1 Maize >10 d a er aman harvest Treatment combina ons1 Mulch tmt (M) Species (C) C2 Sunflower A1C1M1 A1C1M2 A1C2M1 A1C2M2 A2C1M1 A2C1M2 A2C2M1 A2C2M2 A3C1M1 A3C1M2 A3C2M1 A3C2M2 A4C1M1 A4C1M2 A4C2M1 A4C2M2 A1, 2, 3, 4 = aus sown on 20 March, 05 April, 20 April and 05 May, respec vely; C1 = maize, C2 = sunflower; M1 = rice straw mulch (5 t/ha), M2 = no mulch A M J J A S O N 30 June D F 20 Mar M 30 Nov 30 Nov 15 July T. Aus 15 Apr Rabi T. Aman 05 Jun 15 Dec 30 Jul 15 Dec T. Aman T. Aus 30 Apr Rabi 30 Dec 20 Jun 20 Apr A Rabi 20 Jun 05 Apr M 30 Mar T. Aman T. Aus 15 Aug 30 Dec T. Aman T. Aus 05 May J 15 Nov 05 Aug 15 May Rabi 15 Jan Fig. 1. Proposed crop calendar for the 16 aus-aman-rabi cropping system treatment combina ons. 2.3 Experimental layout The experimental area was surrounded by a large bund (50 cm high and 20 cm wide) to protect the site from flooding from adjacent lands. Each sub-sub-plot was surrounded by bunds 20 cm high and 20 cm wide to minimize water flows between adjacent plots. 439 2.4 Cultural Prac ces 2.4.1 Land prepara on For rice, the flooded soil was lled three to four mes using a two-wheel tractor with a power ller, followed by one leveling (‘laddering’) using a wooden ladder that was manually pulled. Final land leveling with a hand pulled plank was done a er basal applica on of fer lizer. For rabi crops, the land was prepared by a power ller powered by a 2-wheel tractor once the soil moisture had declined to about field capacity, followed by one leveling using a wooden ladder pulled manually. Basal fer lizer was applied at the me of final land prepara on. 2.4.2 Crop establishment For rice, a seedbed was prepared in a separate place near the experimental field. The seed was soaked in water for 12 h and incubated for 48-72 h and then the germinated seeds were sown on the seedbed at a rate of 30 kg/ha. Urea (32 kg N/ha; BRRI, 2011) was applied 10 d a er sowing. Twenty-one day-old seedlings were transplanted with two to three seedlings per hill and a hill spacing 20 cm x 20 cm for both the aus and aman crops. Seeds of maize (15 kg/ha) and sunflower (5 kg/ha) were sown in 2-4 cm deep furrows made using a hand furrower. Plan ng geometry for maize was 75 cm x 25 cm and geometry for sunflower was 60 cm x 45 cm. 2.4.3 Fer lizer applica on Urea, triple superphosphate, muriate of potash, gypsum and zinc sulphate were applied to the aus and aman crops as per BRRI recommenda ons (BRRI, 2011) at 100 kg N, 60 kg P2O5, 40 kg K2O, 60 kg CaSO4 and 10 kg ZnSO4 per hectare to each crop. All the fer lizer and one-fourth of the N were applied prior to the final land leveling. The remaining N was top-dressed in three equal splits at 25 d a er transplan ng (DAT), 5-7 d before PI and at heading in the aman crops. For the aus crops, all fer lizer and one-third of the N were applied prior to leveling, and the remaining N was applied 5-7 d before PI and at heading. Fer lizer applica on to maize and sunflower was as per BARI recommenda ons (BARI, 2011). For maize, N, P2O5, K2O, CaSO4, ZnSO4 and B were applied at 253-144-168-33-4-2 kg/ha, respec vely. These nutrients were supplied as urea, triple superphosphate, muriate of potash, gypsum and zinc sulphate. One-third of the N and all the other fer lizers were broadcast prior to the final leveling, and the remaining N was applied in two equal splits just prior to the first (about 25 DAS) and second (about 50 DAS) irriga ons. For sunflower N, P2O5, K2O, CaSO4, ZnSO4 and B were applied at 92-84-90-27-2-2 kg/ha, respec vely. Half of the N and all of the other fer lizers were broadcast prior to the final land prepara on. The remaining N was top dressed at the flower primordial stage (45 DAS). Fer lizer (basal and top dress) was broadcast in the first year and in bands 15 cm from the seed row in the second year. 2.4.4 Water management The aus and aman crops were established and grown on rainfall, with the excep on of the 20 March sown aus crops for which irriga on from the khal was required for puddling and transplan ng. Water was drained from the experiment field when rainfall raised the water depth beyond about 30 cm, by gravity and/or by pumping. The rabi crops were grown using residual soil moisture and irriga on water pumped from the nearby canal as needed. The water was sprayed across the soil surface using a handheld hose. Pre-sowing irriga on was provided for later sowings when the soil was too dry for good germina on and crop establishment. Vacuum gauge tensiometers were installed to guide irriga on scheduling. The tensiometers were installed in ght fi ng holes that were sealed at the top, with the ceramic cups at 37.5 cm. The tensiometers were located mid-way between four plants. For maize and sunflower, irriga on was applied when soil tension had increased to 50 kPa and 70 kPa, respec vely. 440 2.4.5 Weeding Weed control was excellent in all crops. The aman and aus crops were weeded manually as needed prior to urea topdressings. For maize and sunflower, hand weeding was done at 20 to 25 days a er emergence (DAE) and 40 to 50 DAE (prior to urea topdressing and earthing up). 2.4.6 Insect pest and disease control Disease and insect pests in the rice and rabi crops were controlled by following BRRI and BARI recommenda ons (BRRI, 2011; BARI, 2011) when necessary. In general pests and diseases were well-controlled, apart from an outbreak of blast disease during the first aman crop that led to some yield loss from the first and second sowings. The field was surrounded by a plas c barrier with rat traps to prevent rat damage, and was covered at suscep ble stages with a net to prevent bird damage. 2.5 Data collec on 2.5.1 Yield For rice, grain yield was determined in a 3 m x 2 m area in the middle of each sub-sub-plot. The samples were threshed and cleaned, and sub-samples of the grain and straw were oven dried at 70 C to calculate moisture content. Dry grain yield was converted to 14% moisture content, while straw yield was reported dry. Seed yield of maize and sunflower were determined in 3 m x 2 m and 2.25 m x 1.75 m areas, respec vely, in middle of each plot. Grain moisture content was determined, and yield was reported at 12% moisture content for both crops. Rice equivalent yield (REY, t/ha) of rabi crops was calculated as: REY = yield of rabi crop (t/ha) x price of rabi crop (Tk/kg) / price of rice (Tk/kg). 2.5.2 Water Irriga on water. The discharge rate of the pump was calculated from the me taken to fill a 180 L drum. The depth (mm) of irriga on provided to the rabi crops was determined from the dura on of irriga on of each plot, the discharge rate, and plot area. Total irriga on during the rabi crops was determined. Field water depth. The depth of standing water in the rice plots was monitored using seepage scales (measuring tapes) that were installed in each plot. 2.6 Analysis of data The data were analyzed using ANOVA with a 3-factor split plot design for all but the first aman crop, for which a single factor ANOVA was used as at this stage the only treatment was sowing date. Comparison of means was done using the least significance difference (LSD) at 5% probability. The analyses were done using CropStat version 7.2 (IRRI, 2007). 3. Results and discussion 3.1 Crop dura on 3.1.1 Aman Each year, the dura on of the aman crop from transplan ng to PM decreased with delay in sowing, slightly in 2012 (from 113 to 109 d) and considerably in 2013 (from 112 to 96 d) (Table 1). Dura on of the first sowing was similar each year, but shorter for later sowings in 2013, the difference increasing up to 13 d with delay in sowing. The results for 2013 are consistent with the fact that BRRI dhan44 is a photoperiod sensi ve variety. 441 The apparent absence of photoperiod sensi vity in 2012 may be due to the fact that the third and fourth sowings were inundated for 7 and 5 d, respec vely, two to four weeks a er transplan ng, but there was no prolonged inunda on in 2013. 3.1.2 Rabi crops The dura on of maize from sowing to harvest ranged from 124 to 132 d, and of sunflower from 107 to 116 d (Table 1). There was a slight decline in dura on with delay in sowing, which was more evident in the second year when there was a greater range of sowing dates. In that year dura on of maize declined by 6 d between sowing on 30 November and 15 January, while dura on of sunflower declined by 8 d over the same period. The decline in dura on with delay in sowing is consistent with the warmer weather experienced by the later sown crops. The use of straw mulch consistently delayed maturity, but only by a couple of days. Table 2. Field dura on of each crop and the total cropping system Cropping System (CS) CS 1 CS 2 CS 3 CS 4 CS 5 CS 6 CS 7 CS 8 CS 9 CS 10 CS 11 CS 12 CS 13 CS 14 CS 15 CS 16 Treatment1 A1C1M1 A1C1M2 A1C2M1 A1C2M2 A2C1M1 A2C1M2 A2C2M1 A2C2M2 A3C1M1 A3C1M2 A3C2M1 A3C2M2 A4C1M1 A4C1M2 A4C2M1 A4C2M2 Aman 113 113 111 109 2012-13 Boro Aus 129 77 126 115 111 129 79 128 114 110 128 78 126 112 109 127 76 125 110 108 Total 319 316 305 301 321 320 306 302 317 315 301 298 312 310 295 293 Aman 112 108 102 96 2013-14 Boro Aus 132 77 130 116 115 129 77 128 115 113 128 76 126 112 111 126 73 124 108 107 Total 321 319 305 304 314 313 300 298 306 304 290 289 295 293 277 276 1 A1, 2, 3, 4 = aus sown on 20 March, 05 April, 20 April and 05 May respec vely; C1 = maize, C2 = sunflower; M1 = rice straw mulch (5 t/ha), M2 = no mulch 3.1.3 Aus The dura on of aus from transplan ng to PM ranged from 73 to 79 d over the four sowing dates (from 20 March to 5 May) and two years (Table 2). There was li le effect of me of sowing on dura on of aus, except for slightly reduced dura on (by 3 to 4 d) with the last sowing date. 3.1.4 Total system in-field dura on Total in-field dura on of the three crops ranged from 293 to 321 d with maize in the system, and from 276 to 305 d with sunflower in the system. The longest dura on systems allowed an average of 15 d between each crop. Given that the aus and aman crops were harvested at PM, and that there will be a delay of a few days between PM and harvest of aus and aman, this means an in-field dura on of about 330 d for the longest dura on systems, and an average of about 12 d between harvest of one crop and transplan ng of the next if all crops are harvested on me. 442 3.2 Grain yield 3.2.1 Aman Grain yield of aman ranged from 3.2 to 5.4 t/ha over the two years, similar to the observa ons of others (Rashid and Khan, 2006; Masum et al. 2008; CCC, 2009; Biswas et al., 2011). Sa ar and Abedin (2012) and Sharma et al. (2013) suggested that BRRI dhan 44 was a suitable variety for increasing produc vity in dal southern Bangladesh. There were no significant interac ons between sowing date, rabi crop and mulch treatment on the yield of aman. However, in both years there was a significant effect of sowing date on yield. In 2012, yield of aman was least for earlier sowings (Fig. 2). In contrast, in the second year, yield of the first three sowings was similar and yield of the last sowing was significantly lower than yield of the first sowing. The much lower yields of the first two sowings in 2012 were probably due to inunda on for several days (twice) within the first four weeks a er transplan ng and to heavy infesta on of brown spot and blast diseases which started during panicle ini a on to flowering. Infesta on of the later crops was well-controlled by spraying. Raising of the bund around the experimental site and drainage at low de prevented prolonged inunda on and saved the earlier crops in 2012, demonstra ng the feasibility of growing high yielding aman varie es in this region provided there is separa on of lower and higher lands by small levees and removal of excess water as a result of heavy rainfall by systema cally draining at low de. 6 20 June 05 July 20 July 05 August 5 4 3 2 1 0 2012 2013 Fig. 2. Grain yield of aman as affected by sowing date (ver cal bars are lsd0.05 for comparing means within year). 3.2.3 Rabi crops Yield of maize and sunflower ranged from 6.9 to 8.8 and 2.3 to 3.7 t/ha, respec vely, over the two years. These results are similar to the findings of others in the coastal zone (Rahman, 2012) and in other parts of Bangladesh (Ali et al., 2008; Alom et al., 2009; Sarker et al., 2014). There were no significant interac ons between sowing date and mulch treatment on yield of maize or sunflower. Each year, yield of maize was slightly but significantly decreased with mulch, whereas yield of sunflower was significantly increased with mulch (Figs 3a,b). The reasons for this are not known, but one possibility is the reduced rate of drying of the soil under mulch, and thus the tendency for greater waterlogging, together with the fact that sunflower is more tolerant to waterlogging and extracts more soil moisture than maize (Schmidt, 1995). Sowing date also significantly affected the yield of maize and sunflower each year, with similar trends for both crops but the trends were slightly different across years (Figs 5a,b). In 2012-13, maximum yield of maize and sunflower occurred with 20 and 30 December sowings, while 15 December and 15 January sowings had significantly lower yield. In contrast, in 2013-14, yields were largest with 15 December sowing and declined significantly with later sowings. In the second year, the decline in yield of sunflower with the last sowing was 443 much greater than in the first year or for maize in both years. CIMMYT (2005) also reported that yield of hybrid maize decreased with delay in sowing a er 20 December, while Rashid et al. (2014) also found that yield of Hysun33 decreased with late rabi sowing in a coastal saline area of Bangladesh. The lower yield of the first sowing date in our experiment was probably due to waterlogging due to over-irriga on a er sowing in an a empt to restrict soil cracking. 10 Mulch 4 No mulch 8 Mulch No mulch 3 6 2 4 1 2 0 0 2012-2013 2013-2014 Fig. 3a. Effect of mulch treatment on yield of maize (ver cal bars are lsd0.05 for comparing means within year). 10 30-Nov 30-Dec 15-Dec 15-Jan 2012-2013 Fig. 3b. Effect of mulch treatment on yield of sunflower (ver cal bars are lsd0.05 for comparing means within year). 20-Dec 4 8 2013-2014 30-Nov 30-Dec 15-Dec 15-Jan 20-Dec 3 6 2 4 1 2 0 0 2012-2013 2013-2014 Fig. 4a. Effect of aus sowing date on yield of maize (ver cal bars are lsd0.05 for comparing means within year). (Aus sowing date treatments resulted in the range of rabi sowing dates indicated in the legend.) 2012-2013 2013-2014 Fig. 4b. Effect of aus sowing date on yield of sunflower (ver cal bars are lsd0.05 for comparing means within year). (Aus sowing date treatments resulted in the range of rabi sowing dates indicated in the legend.) 3.2.3 Aus Grain yield of aus treatments ranged from 3.8 to 4.6 t/ha over the two years, similar to the yields observed by Ritu (2011) in a medium salinity area of the coastal zone. There were no significant interac ons between sowing date, rabi crop and mulch treatment on yield of aus each year. There was a consistent trend for highest aus yield with the first sowing (20 March), with significantly higher yield than most or all other sowing dates (Fig. 5). 444 6 20-Mar 5-Apr 20-Apr 5-May 5 4 3 2 1 0 2013 2014 Fig. 5. Effect of sowing date on grain yield of aus (ver cal bars are lsd 0.05 for comparing means within year). 3.2.4 System yield Total system rice equivalent yield (REY) ranged from 13.9 to 19.3 t/ha/yr over the two years. Each year, there was a significant interac on between rabi crop and mulch treatment. Mulch increased the yield of the sunflower systems by 0.8 to 1 t/ha each year, whereas there was a small but significant decline in yield of the maize systems (Figs 6a,b). In the first year, REY increased with delay in sowing date to 30 December (Fig. 7a). In the second year there was also significant interac on between rabi crop and sowing date on system yield. Yield was highest for systems in which rabi crops were sown on the first two dates, and declined significantly as sowing was delayed to the third and fourth dates (Fig. 7b). However, the decline was greater with the last sowing date of sunflower. The trends in rabi yield had a strong influence on system yield because REY of the rabi crops was more than 50% of total system yield. No Mulch 20 Mulch 16 16 12 12 8 8 4 4 0 0 Maize Sunflower Fig. 6a. Interac on of rabi species and mulch treatment on system yield in 2012-2013 (ver cal bar is lsd0.05 for comparing all treatment combina ons). No Mulch 20 Maize Mulch Sunflower Fig. 6b. Interac on of rabi species and mulch treatment on system yield in 2013-2014 (ver cal bar is lsd0.05 for comparing all treatment combina ons) . 445 20 30-Nov 15-Dec 30-Dec 15-Jan 20 16 16 12 12 8 8 4 4 0 20 Mar 05 Apr 20 Apr 05 May Sowing dates Fig. 7a. Effect of aus sowing date on system yield in 2012-2013 (ver cal bars are lsd0.05 for comparing means within year). 0 Maize Sunflower Fig. 7b. Interac on between aus sowing date and rabi species on system yield in 2013 (ver cal bars are lsd0.05 for comparing means within year). (Aus sowing date treatments resulted in the range of rabi sowing dates indicated in the legend.) 4. Discussion The results clearly demonstrate the biological feasibility of intensifying to high yielding aus-aman-rabi cropping systems in the coastal zone of Bangladesh in regions where fresh water is available year-round to enable irriga on of rabi crops, and for establishment and early growth of the aus crop in years when the early kharif rains are inadequate. The other key requisite is the ability to drain excess water in order to enable the cul va on of modern non-photoperiod sensi ve short to medium dura on aman varie es and to enable drainage of water a couple of weeks prior to aman harvest. This is essen al to allow the soil to dry sufficiently for mely establishment of the rabi crops, the yield of which declines as sowing is delayed beyond mid November/early December (Shahadat and Rahman 2012; Rashid et al. 2014). 5. Conclusion and recommenda ons This work shows that it is possible to implement highly produc ve aus-aman-rabi cropping systems in areas of the coastal zone of Bangladesh protected from dal flooding (as in the polders) and where there is fresh water in the rivers year-round. We successfully implanted highly produc ve aus-aman-rabi cropping systems by using modern, high yielding rice and rabi varie es and draining excess water at low de when needed during the rainy season. Total rice equivalent yield of the systems tested ranged from 13.9 to 19.3 t/ha/yr. This was achieved using a short dura on, high yielding aus variety (BRRI dhan65), a medium dura on, high yielding aman variety (BRRI dhan44), and hybrid varie es of maize (Pacific 984) and sunflower (HySun33). Rabi crop yield was the main determinant of trends in total system rice equivalent yield, which declined as establishment date of the rabi crops was delayed beyond 15-30 December. Acknowledgements This paper presents findings from G2 ‘Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food (CPWF). We extend our apprecia on to Akbar Ali, Jahid Hasan and Elias Hossain (BRAC staff) for their contribu ons to this study. 446 References Ali, M.Y., S.R. Waddington, D. Hodson, J. Timsina, and J. Dixon. 2008. Maize-rice cropping systems in Bangladesh: Status and research opportuni es. Mexico, D.F.: CIMMYT. Alom M. S., N. K. Paul and M. A. Quayyum. 2009. Performances of different hybrid maize (Zea mays l.) varie es under intercropping systems with groundnut (Arachis hypogaea l.). Bangladesh journal of agriculture research 34(4): 585-95. BARI (Bangladesh Agricultural Research Ins tute). 2011. Krishi Projuk Hatboi. Dhaka: Bengal Com prin ng press. Biswas, K. K., S. Ahmed and M. R. Karim. 2011. Effect of nitrogen levels and spacing on the performance of BRRI dhan44. Eco-friendly agriculture journal 4(11): 726-29. BRRI (Bangladesh Rice Research Ins tute). 2011. Adhunik Dhaner Chash. Dhaka: MRI prin ng press. BRRI (Bangladesh Rice Research Ins tute). 2012. Climate Change Impacts, Vulnerability and Adapta on: Sustaining Rice Produc on in Bangladesh (Bioforsk-BRRI-CEGIS Collabora ve Project) Report on Farmers Group Discussion (FGD) in Salinity Study Area of Patuakhali and Barguna District. CCC (Climate Change Cell). 2009. Impact Assessment of Climate Change and Sea Level Rise on Monsoon. Climate Change Cell, Department of Environment, Ministry of Environment and Forest, Dhaka, Bangladesh. CIMMYT (Interna onal Maize and Wheat Improvement Center). 2005. Food security in Bangladesh: Improving wheat, maize and papaya produc on and impacts of arsenic contamina on. Annual report (July 2004-June 2005). CPWF (Challenge Program for Water and Food). 2011. The Ganges Basin Development Challenges: Increasing the Resilience of Agriculture and Aquaculture Systems in the Coastal Zone of the Ganges Delta. h p://sites.google.com/site/cpw dceoi/ganges-bdc-research. Hasan Md. N., M. S. Hossain, Md. R. Islam, and Md. A. Bari. 2013. Trend in the availability of agricultural land in Bangladesh. Na onal food policy capacity strengthening programme. Soil resources development ins tute, Dhaka. IRRI, 2007. CropStat for Windows, Version 7.2 Metro Manila, Philippines. Masum, S. M., M. H. Ali and M. J. Ullah. 2008. Growth and yield of two T. aman rice varie es as affected by seedling number hill-1 and urea super granules. Journal of agriculture educa on technology 11(1&2): 51-58. Rahman Md. M. 2012. Enhancement of resilience of coastal community in Bangladesh through crop diversifica on in adapta on to climate change impacts. Masters diss. BRAC University. Rashid M. H. and M. A. H. Khan. 2006. Tillering Dynamics and Produc vity of BRRI dhan44 as Influenced by Spacing and Nitrogen Management Technique. Journal of Agriculture and Rural Development 4(1&2): 47-52. Rashid M. H., S. Nasrin, and D. Mahalder. 2014. Zero lled dibbled sunflower enables plan ng earlier and harvests more in the coastal saline area of Bangladesh. Interna onal journal of environmental science and development 5(3): 260-64. Ritu S. P. 2011. Crop and water management for increasing rice cropping intensity and produc vity of the coastal region of Bangladesh. PhD diss. Bangladesh Agriculture University. 447 Sarker Md. A. Z., Md. A. Alam, A. Hossain, Md. and A. Mannaf. 2014. Agro-economic performance of crop diversifica on in rice based cropping systems of northwest Bangladesh. Agriculture, forestry and fisheries 3(4): 264-70. Sa ar S. A. and M. Z. Abedin. 2012. Op ons for coastal farmers of Bangladesh adap ng to impacts of climate change. Paper presented at the interna onal conference on environment, agriculture and food sciences, August 11-12, in Phuket, Thailand. Schmidt H. Walter. 1995. Single crop sunflower produc on. Agronomy facts index (AGF-107-95). Ohio State University Extension. Available at h p://ohioline.osu.edu/agf-fact/0107.html. Shahdat M. K., and M. A. Rahman. 2012. Effect of sowing me on yield of hybrid maize in Barisal region. Bangladesh agricultural research ins tute, agronomy division, RARS, Barisal, Bangladesh. Sharma N., M. R. A. Sarkar, M. A. Rahman and M. R. Islam. 2013. Varietal evalua on of rice for improving produc vity in southern Bangladesh. Interna onal journal of Bioresearch 15(4): 7-13. SRDI (Soil Resources Development Ins tute). 2010. Saline soils of Bangladesh. SRMAF project, Ministry of Agriculture, Govt. of Bangladesh. Dhaka: SFSDP-SRDI Publica on. 448 Opportuni es for cropping system intensifica on in the coastal zone of Bangladesh M.K. Mondal1, P.L.C. Paul2, E. Humphreys1, T.P. Tuong1, S.P. Ritu2,3 and M.A. Rashid2 1 Interna onal Rice Research Ins tute, Bangladesh and Philippines m.mondal@irri.org, e.humphreys@irri.org, t.tuong@irri.org 2 Bangladesh Rice Research Ins tute, Bangladesh plcpauliwm@yahoo.com, arashidiwm@yahoo.com 3 Current address: Sylhet Agricultural University, Bangladesh sanjidap05@gmail.com Abstract Despite huge investment to protect the land of coastal Bangladesh by construc ng polders and con nuous implementa on of development projects, agricultural produc vity of the region remains very low. Most farmers grow a single, low yielding (2-3.5 t/ha) aman crop using tradi onal, late maturing varie es and much of the land lies fallow for several months each year. There has been li le adop on of modern, high yielding aman varie es (HYV) and where rabi crops are grown a er aman harvest yields are less than 1 t/ha. The present study was conducted to demonstrate the poten al for raising produc vity through cropping system intensifica on and diversifica on in low and medium salinity regions of the coastal zone. The study selected Polder 43/2F in Amtali Upazila, Barguna District as its low salinity site and Polder 30 in Ba aghata Upazila, Khulna District as its medium salinity site. At Amtali, HYV aus-aman-boro and aus-aman-rabi cropping systems with total system rice equivalent yield of 14 to 20 t/ha/yr were successfully implemented over three years from July 2011 to June 2014. At Ba aghata, where fresh water scarcity is a major limita on during the dry season, HYV aman-rabi and aman-boro systems were implemented over the same period. The aman-boro systems consistently yielded 9 to 10 t/ha/yr provided that the boro crops were sown in early November. Here, tradi onal crops (sesame, mungbean) were grown a er aman harvest. While improved management (line sowing, use of fer lizer, irriga on and improved varie es) improved rabi yield, yields never exceeded 1 t/ha. For stability of rabi yield, ‘early’ ( mely) establishment is needed to avoid crop damage or destruc on as a result of early kharif rains and cyclones in May, as happened in 2013. Results show that key ingredients for cropping system intensifica on and increasing crop produc vity in the coastal zone include the use of early maturity (non-photoperiod sensi ve) HYV aman, drainage during the aman crop following excessive rainfall, drainage in early November (if water is present) to allow mely aman harvest and thus mely establishment of the rabi and boro crops, and use of available freshwater for irriga on of the dry season crops. Key message: Agricultural cropping system produc vity can be more than doubled in the coastal zone of Bangladesh by using improved varie es and management prac ces, cropping system intensifica on and diversifica on, available water resources and good water management (especially drainage). Keywords: rice, rabi, yield, Khulna, Barisal, Patuakhali 1. Introduc on The coastal zone of Bangladesh comprises low-lying lands within a dense network of rivers and canals. The rivers are dal with diurnal fluctua ons of 2 to 3 m and salinity of the rivers increases during the dry season, more so closer to the coastline and in the southwest (Khan and Kamal 2015). More than 30% of the cul vable land of the country is in the coastal zone (SRDI 2010). Of the 2.85 Mha of coastal and offshore lands, about 1.2 Mha are protected from flooding and saline water intrusion as a result of the construc on of polders. Despite this, cropping intensity and produc vity in this region is much lower than in other parts of the country (BBS 2011), with over 50% of the land remaining fallow during the dry season due to the lack of freshwater for irriga on, and soil salinity, which increases as the dry season progresses (BARC 2008). 449 There is a general percep on that during the dry season the river water is saline throughout the coastal zone. In reality, most of the rivers in the south-central coastal zone (Barisal Division) remain non-saline throughout the year (Khan and Kamal 2015). It is predicted that this will con nue to be the case, even in the climate change scenario with a 22 cm mean sea level rise and a moderate precipita on change (Khan et al. 2015). On the other hand, the river water in medium salinity areas such as Khulna District only remains suitable for irriga on from July to un l early to mid-February, i.e., for the first couple of months of the rabi season (Mondal et al. 2006). Although Bangladesh as a whole is currently self-sufficient in rice produc on, this is not the case for the coastal zone (MoA-FAO 2013; Tuong et al. 2014). The country faces enormous challenges to maintaining food self-sufficiency for its growing popula on, as there is li le scope to further increase cropping system intensity, except in the underu lized coastal zone lands. Despite huge investment in water resources development and the availability of improved varie es of rice and rabi crops and inputs, which have greatly benefited other parts of Bangladesh, produc vity in the coastal zone remains low. Thus, the polders are home to millions of very poor and vulnerable people whose livelihoods and food security depend on agriculture (BBS 2010; Kabir et al. 2014). Most farmers in this region grow a single aman crop using tall, late planted and low yielding local varie es that are photoperiod sensi ve and thus do not mature un l December. Tall seedlings are required in the rainy season to survive the stagnant flooding that occurs in the polders as a result of both high rainfall and poor management of the sluice gates connec ng the canals inside the polders to the dal rivers surrounding them. Further, the lack of separa on of higher and lower lands within the polders exacerbates the problem of water stagna on in the lower lands. Unlike the rest of Bangladesh, farmers in the coastal zone have not widely adopted modern high yielding rice varie es (HYV) because their shorter stature makes them unsuitable for the high water depths that o en prevail during the rainy season. In Barisal Division, the aman crop is some mes followed by a low input, low yielding relay-sown grasspea crop and/or preceded by a nearly rainy season aus rice crop. In Khulna District the aman crop is some mes followed by a late-sown (mid-February to early March) sesame or mungbean crop. The sesame and mungbean crops are o en damaged by the early kharif rains and by cyclones that tend to occur in May every three to four years. Moreover, many fields (in the low lands) remain flooded un l December, delaying harvest of aman and establishment of rabi crops. The late harvest of the local aman varie es and water logged soil prevent the cul va on of high yielding rabi crops such as maize, sunflower and wheat which need to be established in early December for maximum yield. Mondal (1997) showed the feasibility of increasing produc vity in the coastal zone by intensifying to an HYV aman-rabi system. The feasibility of a high yielding aus-aman system using a short dura on high yielding aus variety and a non-photoperiod sensi ve HYV aman variety has also been shown in medium salinity areas of the coastal zone (Ritu et al. 2015). Furthermore, Mondal et al. (2006) and Sharifullah et al. (2009) showed the feasibility of an HYV aman-boro system in the moderately saline coastal zone. This involved ‘early’ ( mely) sowing of the boro crop (mid-November), irriga on directly from the river un l salinity increased to about 4 dS/m, and finishing the crop on river water stored in the polder canal system prior to the river becoming too saline. The present study was designed to take the next step in cropping system intensifica on - to demonstrate the poten al in medium and low salinity areas of the coastal zone of Bangladesh for more produc ve double (aman-boro/rabi) and triple (aus-aman-boro/rabi) cropping systems, respec vely. 2. Methodology Cropping system intensifica on strategies were evaluated in non-replicated demonstra ons in farmers’ fields. The demonstra ons were implemented at two loca ons: (i) a medium salinity loca on in Khulna Division (Hatba village in 2011-12 and Kismat Fultola village in 2012-14, at nearby loca ons in polder 30), and; (ii) a low salinity site in Barisal Division (Bazarkhali village in polder 43/2/F) (Table 1). At Bazarkhali, fresh water is available year round. Here, triple cropping systems (aus-aman-boro and aus-aman-rabi) were tested. At Hatba and Kismat Fultola fresh water becomes very scarce in the second half of the dry season and double systems of aman-boro and aman-rabi were evaluated. Field size at Bazarkhali was 20 m x 120 m for the 450 aus-aman-rabi demonstra ons, and 60 m x 30m for the aus-aman-boro. At Hatba /Kismat Fultola, field size for the aman-rabi was 20 m x 80 m and 60 m x 30 m for the aman-boro. Thus, field size for all systems was typical of the size of farmers’ fields. However, the fields were sub-divided to evaluate different management prac ces for the aman and boro crops. 2.1 Site characteris cs All sites were located adjacent to a khal connected to the surrounding river via a sluice gate in the polder embankment in order to facilitate drainage during the rainy season and to provide a source of water for irriga on during the dry season. Soil type was silty clay to clay loam. Further site details are provided in Table 1. The trials in polder 30 were shi ed from Hatba to Kismat Fultola in 2012 due to flooding from surrounding (higher) lands at Hatba . Table 1. Site details for the cropping system intensifica on trials, 2011-2014 Village Upazila District Polder La tude Longitude Height above sea level Salinity classifica on River water salinity (EC, dS/m)) Topsoil (0-15 cm) salinity (satura on extract EC, dS/m) Soil texture Predominant cropping systems Cropping systems evaluated Site details Hatba /Kismat Fultola Ba aghata Khulna 30 220 41' 00''N 890 30' 00''E 1-3 m Medium 0.2-24.0 2-16 Silty clay Sesame-aman-fallow aman-boro aman-rabi Bazarkhali Amtali Barguna 43/2/F 22o11’33”N 90o15’41”E 1-3 m Low 0.2-2.4 2-8 Clay loam Fallow-aman-grasspea Aus-aman-fallow aus-aman-boro aus-aman-rabi 2.2 Variety selec on 2.2.1 Rice Modern, high yielding varie es (HYV) of rice were used in all systems (Table 2). To fit three crops per year, short dura on varie es of aus, medium dura on, non-photoperiod sensi ve varie es of aman, and a medium dura on boro variety were used. Some of the aman varie es were also known to have some tolerance to water stagna on. Local grain type preference for the aman varie es was also considered – bold in Amtali and slender in Ba aghata. 451 Table 2. Characteris cs of the varie es used in the demonstra ons at each loca on (BRRI 2010, BRRI 2013) Variety Dur- PhotoYield Height Stagnant flooding a on1 period (t/ha) (cm) tolerance (days) sensi vity Aus BRRI dhan42 100 N 3.5 100 BRRI dhan 43 100 N 3.5 100 BRRI dhan 48 BRRI dhan65 (OM 1490) Aman BRRI dhan 33 BRRI dhan49 BRRI dhan53 BRRI dhan54 BINAdhan-7 Boro BRRI dhan28 N 5.5 105 3.5 90 110 100 118 135 125 135 120 N N N Y N 4.5 5.5 4.5 4.5 5.5 100 100 105 115 90 140 N 6.0 90 Salinity/ drought tolerance Suitable for higher Drought rainfall areas Suitable for higher Drought rainfall areas Suitable for lowland Drought 1 Dura on from sowing to physiological maturity 2 BK = Bazarkhali; KF = Kismat Fultola/Hatba Y Y Salinity Salinity N N Grain type Site 2 Medium BK Medium BK Medium bold BK Long slender BK Bold Medium slender Medium slender Medium slender Slender BK KF BK BK, KF BK Medium slender BK, KF 2.2.2 Rabi crops Hybrid and high yielding varie es of several rabi crops were evaluated in aus-aman-rabi and aman-rabi cropping systems (Table 3). As fresh water is available throughout the year at Bazarkhali, a range of longer dura on cereals/oilseed and shorter dura on high value rabi hor cultural crops was evaluated. At Kismat Fultola/Hatba , tradi onal crops grown on residual soil moisture, with limited irriga on with marginal quality and without irriga on, were evaluated. Table 3. Rabi crop/variety characteris cs Species Maize Sunflower Watermelon Chilli Sesame Mungbean Variety/ hybrid Hybrid Hybrid Hybrid Hybrid HYV HYV Name Pacific 984 Hysun 33 Asian2 Debgree BARI Til4 BARI mug6 1 Dura on from sowing to harvest maturity 2 BK = Bazarkhali; KF = Kismat Fultola/Hatba Sources: BARI 2006 and BRAC 2011 452 Dura on1 (days) 125-135 90-110 70-90 120-150 90-100 60-65 Yield (t/ha) 8-10 3-4 40-55 3-4 1.0-1.4 1.0-1.5 Stress tolerance/ Disease resistance Drought Drought Yellow mosaic virus Site2 BK BK BK BK BK, KF BK, KF 2.3 Rice crop management 2.3.1 Sowing date The sowing date of aman was chosen to ensure that the rice would be ready for harvest by mid-November so that the field could be drained in early November (if water was present). This would enable the soil to start to dry for ease of harves ng and enable soil prepara on for rabi crops to be advanced. At Bazarkhali, sowing dates were late July/early August (Table 4). At Kismat Fultola/Hatba , where longer dura on aman varie es were grown, the sowing date was earlier (1 July). The sowing date for boro in Bangladesh varies widely, from November to January, depending on farmer prac ce and loca on. As our objec ves were to fit three crops per year at Bazarkhali and to minimize the stored water requirement to finish off the boro crop at Kismat Fultola, it was important to establish the boro rice as early as possible. However, plan ng too early can lead to cold damage during the reproduc ve stage and greatly reduce yield (Mondal et al. 2010). The la er authors showed that for sowings from 22 October to 15 November, the op mal sowing date at Kismat Fultola was in the second week of November. However, later sowing dates were not evaluated. Therefore, boro sowing dates from early November to late December were evaluated in this study (Table 4). All aus crops were sown in mid-April and transplanted in early May, except in 2012 when the second sowing was done on 7 May and transplanted in the last week of May (Table 4). Ritu et al. (2015) had previously shown the feasibility of high yielding aus-aman systems with aus sowing in May. 453 454 2.3.2 Rice crop management All rice crops were managed according to BRRI recommenda ons (Table 5) (BRRI 2010). The lands were ploughed three to four mes (wet llage) using a power ller powered by a two-wheel tractor. Final land leveling with a bamboo ladder drawn by the two-wheel tractor was done a er basal applica on of fer lizer. Rice seeds were soaked for 12 to 24 h and incubated for 48 to 72 h for germina on. The pre-germinated seeds were sown on the seed bed at 28 g/m2. Ten days a er emergence, urea was applied to the seedbed at 32 kg N/ha. Transplan ng was done when the seedlings had four leaves, with two to three seedlings per hill, and a hill spacing of 20 cm x 20 cm in all crops, except in aman 2011 at Bazarkhali and boro 2011-12 at Hatba , which had 25 cm x 15 cm spacing. All fer lizers other than urea were broadcast just prior to the final land levelling. Urea was broadcast in equal splits at various mes a er transplan ng (Table 5). Excess water from the aus and aman fields was drained before topdressing of urea. The aus and aman crops were primarily grown on rainfall, with gravity irriga on by le ng in river water at high de during dry spells. The boro crops were fully irrigated by pumping water from the khal to maintain a shallow water depth (2 to 5 cm). Table 5. Details of rice crop management at Bazarkhali and Kismat Fultola/Hatba Aus Aman Boro Seedling age at transplan ng (days) 20-23 (2 cases 25-27d) 21-24 21-27 (a few cases extended to 29-35 d) (4 leaves) Number of seedlings per hill 2-3 2-3 2-3 Hill spacing 20 cm x 20 cm 20 cm x 20 cm 25 cm x 15 cm 20 cm x 20 cm 25 cm x 15 cm Fer lizer rate (kg/ha) (N, P, K, Ca, Zn) 83-20-35-11-2 92-20-35-11-2 102-20-35-11-2 Fer lizer sources (kg/ha) Urea Triple superphosphate (TSP) Muriate of potash (MoP) Gypsum Zinc sulphate 180 100 70 60 10 200 100 70 60 10 220 100 70 60 10 Time of urea splits (DAT) 15, 30, 45 15, 30, 45/55 (short/med vars) 15, 30, 55 Insect control Basudin Diazinon Virtako Basudin Synothrene Virtako Basudin at me of each urea applica on 2.3.3 Rice crop monitoring The dura on of each rice crop was determined as the number of days from the day of sowing on the seed bed to physiological maturity. Physiological maturity was taken as the stage when 80% of the grains had turned golden. Grain yield was determined by harves ng a 5 m2 area at five loca ons (four towards the corners and one in the middle) of each plot at Bazarkhali. At Hatba /Kisma ultola, rice was harvested from a 2 m x 5 m area in the middle of each plot. The plants were manually threshed and the grain was cleaned and weighed. Grain moisture content was determined on a subsample using a grain moisture meter. Grain yield was calculated at 14% moisture content. 455 2.4 Rabi crops 2.4.1 Rabi crop treatments Kismat Fultola/Hatba . The mungbean and sesame plots were split into subplots with six management treatments in 4 m x 5 m sub-plots with two replicates: M1 = Farmer prac ce (farmers’ variety, broadcast onto dry cul vated soil, no fer lizer, manual weeding) M2 = M1 + line sowing M3 = M2 + fer lizer + irriga on (no irriga on in 2013-14 due to fresh water scarcity) M4 = M3 + mulch (5 t/ha of rice straw) M5 = M4 + improved variety (BARI mung 6, BARI l 4) M6 = M5 no mulch (in 2012-13 and 2013-14 only) Bazakhali. At Bazarkhali the plots were split into sub-plots with and without mulch (5 t/ha rice straw). Sub-plot size was 10 m x 20 m and there was one replicate of each rabi crop x mulching treatment. 2.4.2 Sowing date Sowing date (Table 6) of the rabi crops varied among crops, seasons and sites depending on soil moisture a er aman harvest and the op mum sowing me for each crop, but all were sown within the recommended period (BARI 2006). Harvest dates were influenced by variety, sowing date, weather (temperature) and soil moisture (the growth dura on of mungbean increased in 2013 at Kisma ultola due to rainfall). Table 6. Sowing and harves ng dates of rabi crops at Bazarkahli and Kisma ultola/Hatba from 2011 to 2014 Crop Chilli Maize Sunflower Watermelon Sesame Mungbean Bazarkahli Sowing date Harvest date 10 Dec – 17 Jan 10 Apr – 10 May 16 Dec – 9 Jan 27 Apr – 13 May 16 Dec – 9 Jan 9- 25 Apr 13 – 30 Dec 1 – 15 Apr 12 – 24 Jan 25 Apr – 3 May 11 -24 Jan 30 Mar-13 Apr Kismat Fultola/Hatba Sowing date Harvest date 08 Feb – 16 Mar 06 Feb – 06 Mar 16 May – 12 Jun 30 Apr-29 May 2.4.3 Rabi crop management The land was lled by a power ller powered by a two-wheel tractor when the soil had dried to around field capacity, followed by one leveling using a wooden ladder pulled by the tractor. The excep on was land prepara on for watermelon, for which 60 cm x 60 cm x 60 cm pits were dug at 2 m x 2 m spacing. BARI recommended management was applied for all rabi crops (Table 7) (BARI 2006). Chili, mungbean, sesame, maize and sunflower were sown in 2-4 cm deep furrows made using a hand furrower, and watermelon was sown in pits. Chili was sown at 50 cm row spacing, mungbean and sesame at 30 cm spacing, and thinning was done two weeks a er emergence to give 5 cm spacing between plants. Plan ng geometry was 75 cm x 25 cm for maize, 60 cm x 45 cm for sunflower, and 2 m x 2 m for watermelon. One seed per hill was sown for maize and sunflower, and two seeds per pit for watermelon. Earthing up was done in maize and sunflower 50-60 d a er sowing (DAS). 456 Table 7. Sowing and fer lizer management in rabi crops at Bazarkhali and Kismat Fultola/Hatba Crops Chilli Maize Mungbean Sesame Sunflower Watermelon Seed rate (kg/ha) 0.8-1.0 15-20 25-30 5.5-6.5 5 0.5-0.6 Spacing (row x plant) 50 x 5 75 x 25 25 x 5 30 x 5 60 x 45 200 x 200 Fer lizer (kg/ha) (cm x cm) N 92 258 50 23 55 97 P 35 58 22 16 26 66 K 75 140 54 20 25 100 S 27 33 2 9 20 20 Zn 2 4 2 1 1 0 B 2 2 2 0 2 0 Fer lizer management. At Bazarkhali, fer lizer was applied to all the rabi crops, but at Hatba /Kisma ultola, it was applied in M3, M4, M5 and M6 treatments only (Table 7). At Bazarkhali, one-third of the N (50% in mungbean) and all the P, K, S, Zn and B were broadcast just prior to the final llage opera on, 1 d before seeding, with the excep on of watermelon for which the fer lizer was applied to the pits 5 d before sowing. The rest of the urea was topdressed thrice in chili and watermelon, twice in maize, sesame and sunflower, and once in mungbean. At Hatba /Kisma ultola, half of the N and the full dose of P, K, S, Zn and B were applied during final land prepara on. The remaining N was topdressed once or twice depending on rainfall and salinity of the khal water (once at Hatba in 2012 just a er rainfall on 10 April, and at Kisma ultola on 6 March 2013 a er irriga on of mungbean, on 25 April 2013 a er rainfall in sesame, and in 2014 a er rainfall on 17 March. Water management. At Bazarkhali, the khal was filled with river water by opening the sluice gate at high de when needed, and then the gate was closed to store the water in the khal for irriga on. All rabi crops were irrigated by pumping water from the khal. Irriga on method varied depending on the crop and stage of growth, topdressing of N fer lizer, and soil moisture, and included watering of small seedlings with a watering can, spraying with a hose, and flood irriga on (Table 8). Chili and watermelon were irrigated five mes using a hose and spray nozzle, and the other rabi crops were flood irrigated two to three mes. Mulched and non-mulched plots were irrigated on the same day. At Hatba /Kismat Fultola the rabi crops were mainly grown using residual soil moisture and rainfall due to scarcity of good quality water. Two irriga ons were applied to mungbean in 2013. Irriga on was not applied in 2012 due to mely rainfall (Fig. 1), and there was no irriga on in 2014 due to the high salinity of the water in the khal (Mondal et al. 2015). 457 Table 8. Irriga on of rabi crops at Bazarkhali and Kismat Fultola in 2011-12 to 2013-14. Crop Chili Maize Mungbean Sesame Sunflower Watermelon Stage and me of irriga on(IR) (days a er sowing, DAS) IR1 IR2 IR3 25-40 50-55 65-70 6-8 leaf stage 10-12 leaf stage Tasseling & silking (30-40 DAS) (55-70 DAS) stage (85-95 DAS) 4-6 leaf stage Pod forma on (15-30 DAS) (50-55 DAS) 4-6 leaf stage Pod forma on (15-30 DAS) (50-70 DAS) 6-8 leaf stage Flower bud ini a on Flowering (30-40 DAS) (50-60 DAS) (70-80 DAS) 10-20 30-40 45-55 IR4 80-85 IR5 95-100 65-75 85-95 Weeding. In non-mulched plots, weeding was done twice by hand at 20-25 and 40-50 d a er emergence (DAE), prior to urea topdressing and earthing up in maize and sunflower. In other crops weeding was done during thinning. In mulched plots, weeding was not generally required, but sporadic hand removal was done when weeds appeared. Pest and disease management. Disease and insect pest infesta ons were controlled well by spraying pes cides and insec cides when necessary. At Bazarkahli, Instar and Admire were sprayed once to control aphid in mungbean and Savin was applied five to six mes to control red pumpkin beetle in watermelon. At Kismat Fultola, Bevis n and Nitro were sprayed to control yellow mosaic virus and Virtako to control semilooper in mungbean. No biocides were applied to other rabi crops. 2.4.4 Rabi crop monitoring Most rabi crops were harvested when the leaves turned dry, brown and/or yellow. Sesame and mungbean were harvested when the pods were black, and chili when the fruit turned reddish. Several pickings were required for mungbean, chili and watermelon. Grain or fruit yield of the rabi crops was determined in a 4 m x 5 m harvest area in the middle of each plot for all crops except watermelon at Bazarkhali, and it was done for the whole plot (5 m x 10 m) at Hatba /Kismat Fultola. The number of plants in the harvest area was also determined. For grain/seed yield of maize, sunflower, mungbean and sesame, the samples were threshed, cleaned and sun dried before weighing and determining moisture content using a grain moisture meter. Yield of maize, sunflower and mungbean was converted to 14% moisture content. The yield of chili and watermelon was determined immediately a er harvest (“green” weight). Rice equivalent yield (REY) of the rabi crops was calculated as: REY = rabi yield (t/ha) * rabi price (Tk/kg) / rice price (Tk/kg). The price of rice was Tk 16.25/kg at Amtali and Tk 18.75/kg at Ba aghata. Prices of rabi crops were the same at both loca ons: chili, Tk 25/kg; maize, Tk 16.25/kg; mungbean, Tk 82.5/kg; sesame, Tk 50/kg; sunflower, Tk 33.75, and; watermelon, Tk 5.71/kg. 2.5 Water monitoring The depth of water in the paddy fields was measured daily using ver cal scales installed in the plots at several loca ons. Salinity of the canal water was also measured daily using a portable EC meter. 458 2.6 Weather Daily rainfall was measured at the field sites at Bazarkhali and Kismat Fultola. Daily rainfall, maximum and minimum temperature, and sunshine hours at Khepupara (in Patuakhali District) and Khulna were collected from the Bangladesh Meteorological Department for the period 2011 to 2014. The weather sta on at Khulna is about 8 km north of the experimental sites at Hatba and Kismat Fultola, while the sta on at Khepupara is about 24 km south of the experimental site at Bazarkhali. 3. Results 3.1 Weather 3.1.1 Rainfall Monthly rainfall totals at Khepupara and Khulna meteorological sta ons were generally similar to the long-term averages, with a few excep ons (Figs. 1a,b). At both loca ons rainfall was extremely high in August 2011, with totals in excess of 600 mm at Khulna and 900 mm at Khepupara. The monsoon season ended late in 2013, with unusually high rainfall at both sites in October (301 mm at Khulna, 457 mm at Khepupara). Pre-kharif rainfall was also unusually high in May 2013 at both loca ons due to cyclone Mohasen (430 mm at Khulna, 518 mm at Khepupara), and in June 2014 at Khulna only (555 mm). Khepupara 2011-12 2012-13 2013-14 Oct Dec LT (1985-2010) 1000 900 800 700 600 500 400 300 200 100 0 Jul Khulna Aug Sep 2011-12 Nov 2012-13 Jan 2013-14 Feb Mar Apr May Jun LT (1985-2010) 700 600 500 400 300 200 100 0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Fig. 1. Monthly rainfall at Khepupara (a) and Khulna (b) from July 2011 to June 2014 compared with the long-term average (1985-2010). 459 3.1.2 Temperature Trends in temperature across the years were similar within and across sites (Figs 2a,b). Temperatures were generally favourable for growing rice throughout the year, apart from low minimum temperatures (around 10oC) in late December/early January each year and high temperatures (approaching 40oC in May). The mean monthly temperatures at Khepupara and Khulna during the study period were generally similar to the long-term average temperatures, except for higher than average temperatures in April 2014, in May 2012 and 2014, and in June 2012 (Figs 3a,b). Based on the long-term data, the temperature regime is slightly more extreme at Khulna than at Khepupara. Maximum temperature is higher at Khulna during summer and the rainy season, while minimum temperature is lower at Khulna in the winter (Fig. 4). 50 (a) Patuakhali T max T min 40 30 20 10 0 50 T max (b) Khulna T min 40 30 20 10 0 Fig. 2. Daily maximum and minimum temperatures at Khepupara (a) and Khulna (b) from July 2011 to June 2014. 460 40 (a) Khepupara 2011-12 2012-13 2013-14 LT (1985-2010) 30 20 10 0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun 40 (b) Khulna 2011-12 2012-13 2013-14 LT (1985-2010) 30 20 10 0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Fig. 3. Mean monthly temperature at Khepupara (a) and Khulna (b) from July 2011 to June 2014, and the long-term (1985-2010) averages. 40 35 30 25 20 15 10 5 Tmax-Khepupara Tmin-Khepupara Tmax-Khulna Tmin-Khulna 0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Fig. 4. Long-term (1985-2010) monthly average maximum and minimum temperatures at Khulna and Khepupara. 461 3.1.3 Sunshine hours During the monsoon season (June to September) the amount of bright sunshine (3 to 5 h per day) was lower than at other mes of the year (Fig. 5a). Daily sunshine hours were highest in November and February to May. The long-term data suggest that the amount of bright sunshine at Khulna is higher than at Khepupara throughout the year (Fig. 5b). (a) Sunshine, 2011-2014 10 9 8 7 6 5 4 3 Khepupara 2 1 0 Khulna (b) Sunshine, 1985-2010 10 9 8 7 6 5 Khepupara 4 Khulna 3 2 1 0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Fig. 5. Monthly average sunshine hour of Khulna and Khepupara during 2011-2014 (a) and long-term means (Khulna: 1985-2010 and Khepupara 1988-2010) (b). 462 3.2 Water depth in rice Water depth in the paddy fields varied widely within and across seasons and years (Figs. 6-8), mainly depending on rainfall and management of the sluice gates during the rainy season. 3.2.1 Bazarkhali Aman. Water depth in the aman crops was generally between 5 and 20 cm, except in early September 2012 (second or third week a er transplan ng) when it increased to 50 cm due to high rainfall (Fig. 6). The crops were thus submerged for about 5 d. Boro. Water depth in the boro crops fluctuated between 1 and 5 cm throughout the season as a result of careful irriga on management (data not presented). Aus. Water depth during the aus season generally fluctuated between 0 and 10 cm because the high de river water level from May to July was only slightly higher than the land level (Fig. 7). However, in mid-May 2013, water depth increased rapidly to about 60 cm due to excessive rainfall from cyclone Mohasen. As a result, the crop was submerged for about one week, about two weeks a er transplan ng. 60 50 40 30 20 10 0 2011 2012 2013 Fig. 6. Paddy water depth in aman 2011 to 2013 at Bazarkhali. 70 60 2013 2014 50 40 30 20 10 0 Fig. 7. Paddy water depth in aus 2013 and 2014 at Bazarkhali. 463 3.2.2 Kisma ultola/Hatba Aman. Water depth during the aman crops generally fluctuated between 5 and 25 cm (Fig. 8). However, in 2011, the crop at Hatba was inundated for about a week star ng two weeks a er transplan ng due to excessive rainfall coupled with the intake of water though the sluice gates by the community for land prepara on and transplan ng of the higher lands. The 2011 crop also experienced water stagna on throughout the grain filling period, from mid-October to maturity, as a result of further intake of water by the community to flood higher lands in the landscape. Water depth in 2012 and 2013 was within 5-20 cm during the aman season at Kismat Fultola. Boro. Water depth during the boro crops fluctuated from 0 (saturated soil) to 6-10 cm throughout the season each year as a result of careful irriga on management (data not presented). 50 Aman 2011 40 Aman 2012 Aman 2013 30 20 10 0 Fig. 8. Paddy water depth in the aman crops at Hatba (2011) and Kisma ultola (2012, 2013). 3.3 Khal water salinity 3.3.1 Bazarkhali Salinity of the water in the khal at Bazarkhali reached a maximum of about 0.6 dS/m towards the end of each dry season, and declined to about 0.1 dS/m during the rainy season (Fig. 9). Thus the water in the khal was always suitable for irriga on. 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Fig. 9. Khal water salinity at Bazarkhali. 464 3.3.2 Kismat Fultola/Hatba Salinity of the water in the khal at Kismat Fultola was much higher than that at Bazarkhali, and increased as the dry season progressed (Fig. 10). Maximum EC ranged from 15 dS/m in 2012 to 21 dS/m in 2013. EC declined to about 0.2 dS/m during the rainy season. In the 2012 dry season, the khal was filled with river water by opening the sluice gate at high de un l first week of February, and the water was stored for irriga on by closing the sluice gate. The water remained suitable for irriga on (<4 dS/m) un l early May 2012. But in 2013 and 2014 the Water Users’ Organiza on kept the sluice gate partly open during the dry season to try and prevent silta on in the gate intake canal. Thus the khal water became unsuitable for irriga on by late February in 2013 and 2014. 25 20 15 10 5 0 Fig. 10. Khal water salinity at Kismat Fultola. 3.4 Crop performance at Bazarkhali 3.4.1 Aman BRRI dhan33 had the shortest dura on each year (105 to 115 d), while BRRI dhan54 had the longest dura on (120 to 125 d) (Table 9). Dura on of all varie es other than BRRI dhan54 was 6 to 20 d longer in 2012 than in the other two years, probably due to inunda on for about one week shortly a er transplan ng on 28 August 2012 (Fig. 6). BRRI dhan54 was transplanted 10 d earlier than the other varie es and is taller and thus more tolerant to water stagna on. The BINAdhan-7 was more heavily damaged because of its very short stature, hence its 20 d longer dura on in 2012 than in 2013 due to delayed llering. Yield was always in excess of 4.1 t/ha except for BRRI dhan33 in 2011. Over the three years, average yield of BRRI dhan53, BRRI dhan54 and BINAdhan-7 was similar (4.6-4.7 t/ha), while BRRI dhan33 averaged 4.2 t/ha. The inunda on in 2012 did not appear to affect yield. 465 Table 9. Dura on, grain weight and yield (mean ± s.e.) of aman crops at Bazarkhali Year Dura on (d) 2011 2012 2013 Mean Grain yield (t/ha) 2011 2012 2013 Mean Aus-aman-boro BRRI dhan54 BRRI dhan33 Aus-aman-rabi BRRI dhan53 BINAdhan-7 126 125 120 124 105 115 107 109 118 124 112 118 127 107 117 4.1±0.1 4.8±0.1 4.7±0.03 4.6±0.2 3.8±0.1 4.3±0.1 4.5±0.3 4.2±0.2 4.3±0.1 4.5±0.1 5.3±0.2 4.7±0.3 4.6±0.1 4.7±0.04 4.6±0.02 3.4.2 Boro Dura on of the boro crops ranged from 123 to 152 d (Table 10). There was a consistent trend for declining crop dura on as sowing date was delayed, due to the warmer weather experienced by later sown crops. Grain yields ranged from 5.8 to 7.9 t/ha, except for the 30 December sowing in 2011 that yielded only 4.4 t/ha, probably due to leaf blight a ack in the seed bed which led to poor growth and llering a er transplan ng. In the first two seasons yield declined with delay in sowing, however in 2013-14, yield of the first sowing was lower than yield of later sowings, due to severe stem borer a ack (est. 20% yield loss). Table 10. Effect of sowing date on dura on and yield (mean ±s.e.) of boro crops (BRRI dhan28) at Bazarkhali Year Dura on (d) 2011-12 2012-13 2013-14 Mean Grain yield (t/ha) 2011-12 2012-13 2013-14 Mean 15 Nov Sowing date 7 Dec 30 Dec 148 152 149 150 138 144 139 140 123 126 126 125 7.9±0.1 6.7±0.2 5.8±0.2 6.8±0.6 6.7±0.2 7.0±0.2 6.6±0.3 6.8±0.1 4.4±0.2 5.9±0.2 6.1±0.2 5.5±0.6 3.4.3 Aus Dura on of all aus varie es was ~100 d except in 2013 when dura on was longer (Table 11), probably because the crops were inundated for about a week shortly a er transplan ng (Fig. 7). Yield ranged from 4.6 to 5.2 t/ha in 2012, 3.8 to 4.3 t/ha in 2013 and 3.6 to 4.1 t/ha in 2014. The lower yield in 2014 may be due to lower solar radia on at flowering (data not presented) than in other years, as temperature was similar each aus season (data not presented). 466 Table 11. Dura on, grain weight and yield (mean ± s.e.) of aus crops at Bazarkhali Year Dura on (d) 2012 2013 2014 Mean Grain yield (t/ha) 2012 2013 2014 Mean Aus-aman-boro BRRI dhan 48 BRRI dhan 42 Aus-aman-rabi BRRI dhan 43 BRRI dhan65 (OM 1490) 102 117 100 106 103 108 100 104 103 108 100 104 108 103 106 4.6±0.0 4.3±0.1 4.1±0.1 4.3±0.2 5.1±0.2 4.2±0.2 3.6±0.0 4.3±0.4 5.2±0.2 4.0±0.2 3.6±0.1 4.3±0.4 3.8±0.1 3.6±0.1 3.7±0.1 3.4.4 Rabi The maize and sunflower crops performed well each year, with maize yields ranging from 8.7 to 9.5 t/ha, and sunflower from 3.6 to 3.8 t/ha (Table 12). Yields of the mulched watermelon (42 and 63 t/ha) and chili (2.3 and 3.6 t/ha) were also good. The effects of mulch on maize and sunflower yield were inconsistent across crops and years. In all three years there was li le effect of mulch on sunflower yield, and the same was true for maize in 2013-14. In 2011-12, mulching reduced maize yield by 0.8 t/ha, while the reverse occurred in 2012-13. The high yields and lack of effect of mulch on maize and sunflower suggests that soil moisture was adequate both with and without mulch as a result of mely flood irriga on. In contrast, at a nearby site (at Patuakhali), Bha acharya et al. (2015) found a small (~0.5 t/ha) but significant increase in yield of maize with mulching in 2012-13 and 2013-13, but a significant decline (~0.5 t/ha) in yield of sunflower with mulching. There was a consistent trend for slightly higher yield of sesame and mungbean with mulch, and for much higher yield of watermelon with mulch. This may have been due to lower soil water availability in the non-mulched plots, as all these crops were only lightly irrigated by watering can or sprinkler. Yield of mungbean was disappoin ng, especially in the first and third years, due to poor germina on for unknown reasons (possibly poor quality seed), and the chili crop stand was poor in the third year (many plants died a er emergence). Table 12. Yield of rabi crops at Bazarkhali Crop Season Mulching 1 Maize Sunflower Mungbean Sesame Watermelon Chili Mulch No-Mulch Mulch No-Mulch Mulch No-Mulch Mulch No-Mulch Mulch No-Mulch Mulch No-Mulch 2011-12 8.7 9.5 3.6 3.7 0.40 0.35 Yield (t/ha) 2012-13 2013-14 9.3 9.2 8.7 9.1 3.6 3.8 3.8 3.7 0.85 0.46 0.63 0.25 0.80 0.57 0.56 0.49 63 42 33 12 2 3.6 2.3 2.0 Mean ± s.e. 9.1±0.1 9.1±0.2 3.7±0.1 3.8±0.1 0.57±0.1 0.41±0.1 0.68±0.1 0.53±0.0 52±11 23±10 2.9±0.7 467 1 Sesame crop damaged by 112 mm rainfall prior to harvest, yield of watermelon was not recorded 2 Yield a er the first harvest, combined for both mulched and non-mulched plots (only one harvest due to damage by 500 mm rainfall in May 2013) The growth dura on of the rabi crops varied from 68 to 140 d and there was no effect of mulch on crop dura on (Table 13). The crops matured one to two weeks earlier in 2011-12 than in later years, probably because of late sowing (second half of January, two to three weeks later than in 2012-13 and 2013-14) and thus warmer growing condi ons in 2011-12. All crops were harvested by the end of April, well before the beginning of the pre-monsoon and cyclone seasons (Table 14). Table 13. Growth dura on of rabi crops at Bazarkhali Crop Maize Sunflower Mungbean Sesame Watermelon Chili Dura on (d) 2011-12 125 106 68 92 99 2012-13 132 115 86 101 105 140 2013-14 132 114 92 111 116 129 Mean ± s.e. 130±2 112±3 82±7 101±6 107±5 135±6 Table 14. Growth period (sowing-harvest) of rabi crops at Bazarkhali Crop Maize Sunflower Mungbean Sesame Watermelon Chili 2011-12 09 Jan 12 – 13 May 12 09 Jan 12 – 25 Apr 12 24 Jan 12 – 02 Apr 12 24 Jan 12 – 25 Apr 12 30 Dec 11 – 06 Apr 12 17 Jan 12 – crop damage 2012-13 24 Dec 12 – 05 May 13 24 Dec 12 – 17 Apr 13 16 Jan 13 – 10 Apr 13 17 Jan 13 – 28 Apr 13 30 Dec 12 – 15 Apr 13 22 Dec 12 – 10 May 13 2013-14 16 Dec 13 – 27 Apr 14 16 Dec 13 – 09 Apr 14 11 Jan 14 – 10 Apr 14 12 Jan 14 – 03 May 14 13 Dec 13 – 10 Apr 14 10 Dec 13 – 18 Apr 14 3.5 Crop performance at Kismat Fultola/Hatba 3.5.1 Aman The dura on of BRRI dhan49 ranged from 135 to 147 d, and that of BRRI dhan54 from 140 to 146 d (Table 15). Dura on of BRRI dhan54 was longer than at Bazarkhali, where the crop was sown 24 d later, reflec ng the fact that this variety is sensi ve to photoperiod. Yields ranged from 4.2 to 4.7 t/ha except for BRRI dhan49 in 2013, which yielded 5.7 t/ha. 468 Table 15. Dura on, grain weight and yield components of aman at Kisma ultola/Hatba Cropping Pa ern Variety Aman-Boro BRRI dhan49 Aman-Rabi BRRI dhan54 Year 2011 2012 2013 Mean 2011 2012 2013 Mean Growth dura on (d) 135 139 140 138 140 145 146 144 1000 grain wt (g) 20.2 19.9 22.4 20.8 27.7 26.4 27.0 Yield (t/ha) 4.2±0.1 4.7±0.2 5. 7±0.2 4.9±0.4 4.5±0.3 4.3±0.2 4.7±0.1 4.5±0.1 3.5.2 Boro Some of the seedlings of the 10 November 2011 sowing died a er transplan ng, presumably due to the combined stresses of low temperature and mild salinity. Gap filling was done using spare seedlings and llers from surviving hills. In 2012-13 the 30 November and 20 December sowings died a er transplan ng, and likewise the 20 December sowing in 2013-14. There was a trend for decreased dura on with delay in sowing, as at Bazarkhali (Table 16). Yield of the 10 November sowing was similar each year, at 4.6 to 4.9 t/ha. In 2011-12, yield increased with delay in sowing from 10 November to 20 December when the crop was irrigated with low salinity water (Fig. 10). The reason for low yield of the 10 November sowing is probably due to the seedling death and gap filling described above. In the next two years, there was no or very low yield for later sowings. Yields at Kismat Fultola/Hatba were much lower than at Bazarkhali (except for the last sowing in 2011-12 when the seedling nursery at Bazarkhali was affected by leaf blight). Temperatures in both regions were similar during the boro season (Figs 2a,b); therefore, we suspect that the cause of the generally much lower yields at Kismat Fultola/Hatba was the combina on of salinity and low temperature. Table 16. Effect of sowing date on dura on, yield components and yield of boro (BRRI dhan28) at Kismat Fultola/Hatba Year Dura on (d) 2011-12 2012-13 2013-14 Mean Grain yield (t/ha) 2011-12 2012-13 2013-14 Mean 10 Nov Sowing date 30 Nov 20 Dec 143 166 150 153 143 Cold damage 144 144 135 Cold damage Cold damage 135 4.57 4.89 4.57 4.68 5.86 0.00 1.73 2.53 6.31 0.00 0.00 2.10 469 3.5.3 Rabi In 2012 and 2014, maximum mungbean yields of 0.9 to 1 t/ha were achieved, but yields of sesame were generally poor (usually <0.5 t/ha). Mungbean yield tended to increase with improved management and highest yields were achieved with treatments M5 (line sowing plus fer lizer plus irriga on plus mulching plus improved variety BARI Mung 6) and M6 (as for M5 but without mulch) (Table 17). In 2012, yield was increased from 0.3 to 1.0 t/ha with line sowing, fer lizer and irriga on. However, in 2014 there was no benefit of line sowing and fer lizer, probably due to water deficit, as irriga on was not provided due to the high salinity of the khal water (Fig. 10). Mulching suppressed yield of sesame as the mulch impaired establishment, probably because the rows were covered by the straw. All rabi crops in 2013 were destroyed by waterlogging as a result of cyclone Mohasen in mid-May, which occurred shortly before crop maturity. In that year, sowing of the rabi crops did not take place un l 8 February 2013. Table 17. Yield of rabi crops Hatba /Kisma ultola Crop season1 Crop 2 2012 2014 Mungbean Sesame Mungbean Sesame M1 0.32 0.28 0.67 0.44 M2 0.46 0.47 0.69 0.3 M3 0.61 1.04 0.81 0.39 Yield (t/ha) M43 0.87 0.58 0.65 -3 M5 1.01 0.49 0.9 0.42 M6 0.88 0.37 Mean 0.65±0.10 0.57±0.10 0.77±0.04 0.38±0.02 1 All 2012-13 rabi crops destroyed by cyclone in May, due to late establishment 2 M1 = Farmer prac ce (farmers’ variety, broadcast onto dry cul vated soil, no fer lizer, manual weeding) M2 = M1 + line sowing M3 = M2 + fer lizer + irriga on (no irriga on in 2013-14 due to fresh water scarcity) M4 = M3 + mulch (5 t/ha of rice straw) M5 = M4 + improved variety (BARI mung 6, BARI l4) M6 = M5 no mulch (in 2012-13 and 2013-14 only) 3 Sesame plants died due to excess soil moisture caused by seepage from the boro rice field 3.6 Total system produc on 3.6.1 Bazarkhali Aus-aman-boro. Total system yield ranged from 13.2 to 16.6 t/ha over the three years and three boro sowing dates (Table 18). The lowest system yield was due to severe stem borer a ack of the earliest boro sowing. All other systems yielded more than 14 t/ha. 470 Table 18. Annual aman-boro-aus system yield at Bazarkhali Year Aman (t/ha) Boro (t/ha) Aus (t/ha) 15 Nov 7 Dec 20 Dec 7.9 6.7 5.8 4.6 6.7 7.0 6.6 4.3 4.42 5.9 6.1 4.1 6.3 6.5 6.2 4.3 1 2011-12 2012-13 2013-14 Mean 4.1 4.8 4.7 4.6 1 Boro sowing date 2 Severe stem borer a ack in 15 Nov sown boro crop Total rice yield (t/ha/yr) 15 Nov 7 Dec 20 Dec 16.6±1.2 15.4±0.8 14.5±0.5 15.8±0.7 16.1±0.8 15.8±0.7 13.22±0.2 14.7±0.5 14.9±0.6 15.22±0.5 15.4±0.1 15.1±0.1 Mean 15.5±0.6 15.9±0.1 14.2±0.5 15.2±0.4 Aus-aman-rabi. Total system yield (REY) ranged from 8.5 to 30.7 t/ha over the three years with six rabi crops grown under mulched and no-mulch prac ces (Table 19). Inclusion of watermelon generally resulted in much higher REY than in the aus-aman-boro system, due to the high yield of watermelon. Inclusion of maize and sunflower gave comparable or slightly higher REY (16 to 19 t/ha) than the triple rice system, due to the higher yield of maize and the higher value of sunflower seed. System yield with mungbean, sesame and chili in the system (10 to 13 t/ha in years when the crops were planted on me) was always considerably lower than that of the triple rice system. Mulching greatly increased system yield of watermelon, but yield variability across years was high in both mulched and non-mulched treatments. System yield with maize and sunflower was more stable over the three years than with the other rabi crops. Table 19. Annual aus-aman-rabi system yield at Bazarkhali Rabi crop 2011-12 Chili Maize Mungbean Sesame Sunflower 2012-13 Chili Maize Mungbean Sesame Sunflower Watermelon 2013-14 Chili Maize Mungbean Sesame Sunflower Watermelon Aman1 (t/ha) Rabi (rice equivalent yield, t/ha)2 Mulch No mulch 4.0±0.2 0.0 8.7 2.0 0.0 7.5 0.0 9.5 1.8 0.0 7.7 4.5±0.1 5.5 9.3 4.3 2.5 7.5 22.2 3.5 9.2 2.3 1.8 7.9 14.8 4.8±0.2 Aus1 (t/ha) Total system rice equivalent yield (t/ha) Mulch No mulch 5.2±0.04 9.2±0.8 17.9±2.4 11.2±1.6 9.2±0.8 16.7±1.8 9.2±0.8 18.7±2.9 11.0±1.7 9.2±0.8 16.9±1.9 n/a 8.7 3.2 1.7 7.9 11.6 4.0±0.1 14.0±0.8 17.8±2.9 12.8±0.3 11.0±1.1 16.0±1.9 30.7±10.3 17.2±2.6 11.7±0.7 10.0±1.5 16.4±2.1 20.1±4.3 3.1 9.1 1.3 1.5 7.7 4.2 3.6±0.1 11.9±0.7 17.6±2.9 10.7±1.2 10.2±1.5 16.3±2.2 23.3±6.1 11.5±0.9 17.5±2.9 9.7±1.8 9.9±1.7 16.1±2.1 12.6±0.6 471 1 Mean of three varie es 2 In 2011-12 chili and sesame were damaged due to excess rainfall; in 2012-13 chili yield in no-mulch plots was not recorded 3.6.2 Kisma ultola/Hatba Aman-boro. Annual system produc on ranged from 4.7 to 10.5 t/ha and was strongly affected by boro sowing date (Table 20). The most consistent system was with boro sown on 10 November, with system yield ranging from 8.8 to 10.2 t/ha over the three years. Table 20. Annual aman-boro-aus system yield at Hatba /Kisma ultola Year 2011-12 2012-13 2013-14 Mean Aman (t/ha) 4.2 4.7 5.7 4.9 Boro (t/ha) 10-Nov1 4.6 4.9 4.6 4.7 30-Nov 5.9 0.02 1.72 2.5 Total rice yield (t/ha/yr) 20-Dec 10-Nov 30-Nov 20-Dec 6.3 8.8±0.2 10.1±0.9 10.5±1.1 0.02 9.6±0.1 4.7±2.4 4.7±2.4 2 0.0 10.3±0.6 7.4±2.0 5.7±2.9 2.1 9.6±0.1 7.4±1.2 7.0±1.4 1 Boro sowing date 2 Crop seriously affected/destroyed due to low temperature and high irriga on water salinity Mean 9.8±0.5 6.3±1.6 7.8±1.3 8.0±0.8 Aman-rabi. Annual system REY ranged from 5.2 to 8.9 t/ha, and systems with mungbean (5.9 to 8.9 t/ha) had slightly higher yield than with sesame (5.2 to 7.3 t/ha) (Table 21). Thus REY of the best aman-rabi systems was slightly lower than that of the best aman-boro systems. Table 21. Annual aus-aman-rabi system yield at Hatba /Kisma ultola Rabi crop1 Aman (t/ha) 2011-12 Mungbean Sesame 2013-14 Mungbean Sesame Rabi rice equivalent yield (t/ha) Minimum Maximum Total system rice equivalent yield (t/ha) Minimum Maximum 4.5 1.4 (M12) 0.7 (M1) 4.4 (M5) 2.8 (M3) 5.9±1.5 5.2±1.9 8.9±0.03 7.3±0.9 4.7 2.9 (M1,2,4) 0.8 (M2) 4.0 (M5) 1.2 (M1,5) 7.6±0.9 5.5±2.0 8.7±0.4 5.9±1.8 All 2012-13 rabi crops destroyed by cyclone in May, due to late establishment 1 2 M1 = Farmer prac ce (farmers’ variety, broadcast onto dry cul vated soil, no fer lizer, manual weeding) M2 = M1 + line sowing M3 = M2 + fer lizer + irriga on (no irriga on in 2013-14 due to fresh water scarcity) M4 = M3 + mulch (5 t/ha of rice straw) M5 = M4 + improved variety (BARI mung 6, BARI l4) M6 = M5 no mulch (in 2012-13 and 2013-14 only) 472 4. Discussion Cropping system intensifica on in the coastal zone of Bangladesh has been limited by the mispercep on that the river water is too saline for irriga on throughout the coastal zone. In reality, most of the rivers in the south-central coastal zone (Barisal Division) remain non-saline throughout the year, while the rivers in Khulna District are suitable for irriga on from July to mid-February i.e., for the first couple of months of the dry season (Khan and Kamal 2015). Our findings on salinity in the khals in polders 43/2/F and 30 are consistent with this. There is also the mispercep on that modern, high yielding aman varie es (HYV) cannot be grown in the coastal zone, whereas there are significant areas of medium-high lands on which flooding depth is not too high for HYV. Furthermore, in the polders, systema c opera on of the sluice gates to enable drainage of excess water at low de and separa on of lands of different eleva on could enable the produc on of HYV aman over large areas (Mondal et al. 2015). The results of the work presented above clearly demonstrate the feasibility of intensifying to high yielding aus-aman-boro and aus-aman-rabi cropping systems in the south-central coastal zone of Bangladesh. Here, triple rice cropping can produce about 15 t/ha/yr in comparison with current farmer produc on of 3 to 6 t/ha of rice, and 0-1 t/ha of rabi crops such as grass pea. Alterna vely, high yielding or high value rabi crops such as maize and sunflower can be grown instead of boro, with total rice equivalent yield (REY) similar to or slightly higher than that of the triple rice system. Intensifica on and diversifica on to a wide range of rabi crops provides further more produc ve op ons than current prac ce. The results also demonstrate the feasibility of adop on of high yielding aman-boro cropping systems in areas where fresh water is limited in the dry season (such as Khulna District). In such areas river water can be brought in through the sluice gates at high de for irriga on directly from the river from November un l early to mid-February. The final intake of river water should take place in early February, a er which the gates should be closed to store the water for irriga on by pumping from the khals. In the medium salinity area, ‘early’ (10 Nov) sowing of boro was cri cal for consistently high boro yields and thus system yields of 9 to 10 t/ha, compared with current farmer produc on of about 3 t/ha of aman rice and 0-1 t/ha of sesame (or mungbean). Use of improved produc on techniques (‘early’ ( mely) sowing, sowing in lines, and use of improved varie es, fer lizer, irriga on) increased the produc vity of sesame and mungbean, but not beyond 1 t/ha. However, others (Rahman et al. 2015) have also shown the viability of high yielding or high value crops such as maize and sunflower in this medium salinity area, which would further increase rice equivalent system yield. A key requisite to enable the cul va on of many high yielding and high value rabi crops is ‘early’ ( mely, mid-November) harvest of the aman crop and drainage in early November if there is s ll water present. Early drainage is essen al to allow suitable condi ons for harvest of the aman crop and to allow the soil to dry sufficiently for mely rabi crop establishment. Early aman harvest means growing modern non-photoperiod sensi ve short to medium dura on aman varie es instead of the current prac ce of late maturing, photoperiod sensi ve tradi onal varie es. This in turn means the need for drainage of water at low de when field water depth is too high for HYV (following excessive rainfall), and prior to topdressing N fer lizer. Late establishment results in damage or destruc on of the tradi onal rabi crops by pre-monsoon rainfall and cyclones in May, as happened in 2013. Similarly, impaired produc vity and destruc on of sunflower and maize can occur when establishment is delayed (Rahman et al. 2015). Successful produc on of tradi onal rabi crops (sesame and mungbean) in medium salinity areas like Kismat Fultola currently depends very much on rainfall – too li le, or too much at the wrong mes—and thus produc on is highly risky and varies greatly from year to year. Be er management to allow mely establishment and capitalize on stored soil moisture, and storage of fresh water for irriga on later on, would be beneficial. 473 5. Conclusions The results show that it is possible to greatly increase cropping system produc vity in low and medium salinity regions of the coastal zone, using improved rice and rabi varie es and good water management – in par cular, drainage a er excessive rainfall during the monsoon season to enable the cul va on of high yielding, early maturity aman varie es. In the south-central coastal zone where fresh water was available throughout the year, triple rice system produc vity of 13 to 17 t/ha was achieved in farmers’ fields using short dura on, high yielding aus varie es followed by short to medium dura on HYV aman and a medium dura on boro variety sown in mid-November to mid-December. Similar or higher system rice equivalent yields were achieved by replacing the boro crop with maize, sunflower and watermelon sown in early December In moderately saline areas in Khulna District, freshwater is limited to the first couple of months during the dry season. Here, 9 to 10 t/ha of rice were produced in an HYV aman-boro system with sowing of the boro crop in early November. Since cropping intensity and produc vity in other parts of Bangladesh are already high, the under-u lized agricultural lands of the coastal zone may well be the only region where significant gains in food produc on can be made to address future challenges to the food security of Bangladesh. Acknowledgements This paper presents findings from ‘G2: Produc ve, profitable and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food. The authors are grateful for high quality technical assistance from Swapan Bhadra, Amal Ray, Lincoln Ray, Tanmoy Ray and Mithun Ray. References Bangladesh Agricultural Research Council (BARC) 2008. Es ma on of Fallow Land Area and Crop Produc on Plan for Barisal, Pirojpur, Jhalaka , Bhola, Patuakhali and Bagerhat Districts. A report submi ed to the Ministry of Agriculture, Govt. of Bangladesh. May 2008. (In Bangla). Bangladesh Agricultural Research Council (BARC) 2012. Climate and land resources informa on database. h p://www.barcapps.gov.bd/dbs/index.php?t=ym_temph p://www.barcapps.gov.bd/dbs/index.php?t=ym_t emp (last accessed 28 December 2014) Bangladesh Agricultural Research Ins tute (BARI) 2006. Krishi projuc hatboi (Handbook of Agro-Technology in Bangla) 4th edi on. BARi, Gazipur, Bangladesh Bangladesh Bureau of Sta s cs (BBS) 2010.Census of agriculture 2008. Ministry of Planning, Dhaka, Bangladesh. Bangladesh Bureau of Sta s cs (BBS) 2011. Sta s cal year book of Bangladesh 2012. Ministry of Planning, Dhaka, Bangladesh. Bangladesh Rice Research Ins tute (BRRI) 2010. Adhunik dhaner chas (Modern rice cul va on - in Bangla). BRRI, Gazipur,Bangladesh. Bangladesh Rice Research Ins tute (BRRI) 2013. Adhunik dhaner chas (Modern rice cul va on - in Bangla). BRRI, Gazipur,Bangladesh. Bha acharya, J., Mondal, M.K., Humphreys, E., Saha, N.K., Rashid, M.H., Paul, P.C. and Ritu, S.P. 2015. 474 Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh. These proceedings. BRAC 2011. Crop intensifica on for achieving food self-sufficiency in the coastal regions of Bangladesh. Final Narra ve Report (January 2010-October 2011) submi ed to the Eureopean Union (EU), December 2011. Kabir, K. A., Karim, M., Shahrier, M.B., Saha, S.B., Chakraborty, S., Jharendu, P. and Phillips, M.J. 2014. Homestead farming systems in southwest Bangladesh: A survey. CPWF Working Paper Series (submi ed). Khan, Z.H. and Kamal, F.A. 2015. The Ganges coastal zone – current condi ons, future projec ons. These proceedings. Khan, Z.H., Kamal, F.A., Khan, M.S.H., Rahman, M., Khan, M.S.A., Khan, N.A.A.K. and Sharma, B.R. 2015. External drivers of change, strategies and future projec on on surface water resources of the Ganges coastal zone of Bangladesh. These proceedings. Ministry of Agriculture-Food and Agriculture Organiza on. 2013. Master Plan for agricultural development in the southern region of Bangladesh. Prepared by the Ministry of Agriculture of the Government of Bangladesh and Food and Agriculture Organiza on of the United Na ons, Dhaka, Bangladesh, March 2013. Mondal, M.K. 1997. Management of soil and water resources for higher produc vity of coastal saline ricelands in Bangladesh. Unpublished PhD disserta on, University of the Philippines Los Baños, Philippines. Mondal, M.K., Tuong, T.P., Ritu, S.P., Choudhury, M H.K., Chasi, A. M., Majumder, P.K., Islam, M.M. and Adhikary, S.K. 2006. Coastal water resource use for higher produc vity: Par cipatory research for increasing cropping intensity in Bangladesh. In C. T. Hoanh, T. P. Tuong, J. W. Gowing and B. Hardy (Eds). Environmental and livelihoods in tropical coastal zones: Managing agriculture-fishery-aquaculture conflicts. Comprehensive Assessment of Water Management in Agriculture Series, no. 2. CABI Publishing p 72-85. Mondal, M.K., Tuong, T.P., Sharifullah, A.K.M. and Sa ar, M.A. 2010. Water supply and demand for dry-season rice in the coastal polders of Bangladesh. In C. T. Hoanh, B. W. Szuster, K. Suan-Pheng, A. M. Ismail and A. D. Noble (Eds.) Tropical Deltas and Coastal Zones: Food Produc on, Communi es and Environment at the Land-Water Interface Comprehensive assessment of water management in agriculture series, no. 9. CABI publica on: 264-278. Mondal, M.K., Humphreys, E., Tuong, T.P., Rahman, M.N. and Islam, M.K. 2015. Community water management and cropping system synchroniza on: The key to unlocking the produc on poten al of polder ecosystems in Bangladesh. These proceedings. Rahman, M.N., Amin, M.G.A., Mondal, M.K. and Humphreys, E. 2015. Residual soil moisture u liza on techniques for improving land produc vity in the coastal zone of Bangladesh. These proceedings. Ritu, S.P., Mondal, M.K., Tuong, T.P., Talukdar, S.U. and Humphreys, E. 2015. An aus-aman system for increasing the produc vity of a moderately saline region of the coastal zone of Bangladesh. These proceedings. Sharifullah, A., Tuong, T.P., Mondal, M.K. and D.T. Franco. 2009. Assessing water supply and demand for dry season rice in coastal polders of Bangladesh. In: Humphreys, E. and R.S. Bayot (eds). Increasing the produc vity and sustainability of rainfed cropping systems of poor smallholder farmers. Proceedings of the CGIAR Challenge Program on Water and Food Interna onal Workshop on Rainfed Cropping Systems, Tamale, Ghana, 22-25 September 2008. The CGIAR Challenge Program on Water and Food, Colombo, Sri Lanka. 475 Available at: h p://r4d.dfid.gov.uk/PDF/Outputs/WaterfoodCP/CPWF_Proceedings_Rainfed_Workshop%5B1%5D.pdf (last accessed 31 December 2014) Soil Resources Development Ins tute (SRDI). 2010. Saline soils of Bangladesh. SRMAF project, Ministry of Agriculture, Govt. of Bangladesh. Dhaka: SFSDP-SRDI Publica on. Tuong, T. P., Humphreys, E., Khan, Z.H., Nelson, A., Mondal, M., Buisson, M.C. and George, P. 2014. Messages from the Ganges basin development challenge: Unlocking the produc on poten al of the polders of the coastal zone of Bangladesh through water management investment and reform. CPWF Research for Development Series 9. Available at: h ps://cgspace.cgiar.org/bitstream/handle/10568/41708/CPWF%20Ganges%20basin%20messages%20Sept% 2014.pdf?sequence=5 (last accessed 30 Dec 2014). 476 Op mizing use of fresh and saline water for irriga on of boro rice in salt affected areas of Bangladesh using the crop model ORYZA v3 A.M. Radanielson1, O. Angeles1, T. Li 1, A.K. Rahman2 and D. Gaydon 3 1 Interna onal Rice Research Ins tute, Philippines a.radanielson@irri.org, o.angeles@irri.org, t.li@irri.org 2 Bangladesh Agricultural Research Ins tute, Bangladesh a ksobari26@yahoo.com 3 CSIRO Agriculture Flagship, Australia don.gaydon@csiro.au Abstract Salinity is es mated to affect more than 30% of the cul vated rice area in Bangladesh. With climate change, the salt-affected area is likely to expand with increasingly detrimental impact on the region’s produc vity. Adap ve strategies based on conjunc ve use of fresh and saline water for irriga on were evaluated. Simula ons of different crop management combina ons were performed with a modified version of the rice crop model ORYZA v3, which included a crop salinity-response module. The model was calibrated and validated with field experimental data collected between 2012 and 2014 in Satkhira, Bangladesh and Infanta, Quezon, Philippines. The factors considered were ming and amount of fresh and saline irriga on water, date of sowing and crop growth dura on. Clima c variability and changes in soil salinity were represented using historical data from 2000 to 2014 for Satkhira. ORYZA v3 demonstrated good accuracy in simula ng observed rice yields. Weekly irriga on, alterna ng between the use of fresh and saline water (with prevailing dynamics of salinity at the experimental site in Satkhira), reduced soil salinity and increased yield in comparison with sole use of saline water. A sensi vity analysis of simulated yield suggested that: i) early sowing to escape higher salinity at flowering me resulted in higher yield; ii) using fresh water for irriga on for two weeks alterna ng with saline water for one week significantly reduced yield loss due to salinity compared with con nuous irriga on with saline water; and iii) the use of an improved, salt-tolerant variety such as BRRI dhan47 with medium growth dura on irrigated with alternate two weeks fresh water and one week saline water would result in similar yield to that of BRRI dhan47 irrigated with fresh water. Further simula ons with future climate data and in other salt-affected rice growing regions would be desirable to explore the generality of the results. These could be valuable to inform the development of priori es for investment and research for effec ve strategies to increase rice produc on in salt-affected areas. Key message: Irriga on with a mixture of fresh and saline water and the use of salt-tolerant varie es are feasible strategies to increase rice produc on in salt-affected areas with limited availability of fresh water. Key words: modeling, phenology, rice, salinity, variety, water produc vity 1. Introduc on Soil salinity is among the environmental factors limi ng rice crop produc on. It affects land in coastal areas and causes loss of cul vated land of about 1.6 million ha per year worldwide (FAO, 2013). Salinity is currently es mated to affect more than 30% of the cul vated rice area in Bangladesh. With climate change, this is likely to expand with increasingly detrimental impact on the region’s produc vity (Wassmann et al. 2009). Improving rice produc vity to address the challenge of mee ng and maintaining food security for the country must effec vely address the cropping systems management challenges in these areas. Among the technologies developed to alleviate the effect of salinity is the use of tolerant varie es, which could significantly increase produc vity using suitable crop management (Ismail et al. 2007). Soil salinity dynamics are among the biophysical factors driving the cropping calendar under salt-stressed condi ons in Bangladesh (Gaydon et al. 2014). On one hand, the establishment of wet season rice depends on the availability of enough rain to flush the accumulated salt deeper into the soil, beyond crop roots. On the 477 other hand, the start of dry season cropping must be as early as possible for the crop to avoid the high peak of soil salinity later in the dry season. Studies have reported that alternate use of fresh and saline water makes be er use of resources under salt-stressed condi ons (Beltran 1999; Flowers et al. 2005). Furthermore, with climate change, a shi in plan ng dates would be expected to improve rice produc vity for regions such Bangladesh (Li et al., 2015). Temporal and spa al variability of salinity adds complexity in cropping management and needs suitable tools to assist in effec ve decision making. Modeling provides a powerful tool for formula on of hypotheses and quan fica on of cropping system performance. It can provide quan ta ve descrip ons and insights into agricultural systems. It helps to define areas where knowledge is lacking and to design more adequate and effec ve experiments (France and Thornley 1984). Physically-based mechanis c models or empirical models that account for meteorological condi ons can give either unrealis c or realis c results, depending on the condi ons under which they were developed (Eatherall 1997). Crop modeling with the ability to represent soil salinity dynamics and the variability of the rice crop response to salinity throughout its crop development could be a useful tool to efficiently study the problem of rice produc on in salt-affected areas and to iden fy op mized irriga on management. Fresh water availability is the most limi ng factor in leaching salt which has accumulated in the root zone and in maintaining soil salinity at levels suitable for crop produc on (Crescimanno and Garofalo 2006). Strategies for maximizing the produc vity of the available fresh water would increase yield and/or allow for an increased area of land under cul va on. Such strategies require an appropriate cropping calendar and adapted cul vars, managed to avoid the high salinity peak that occurs in the later stages of crop growth during the dry season. Conven onal field experiments are limited in their ability to explore different combina ons of these factors. In this work, we evaluate the conjunc ve use of fresh and saline water for irriga on through modeling, aiming to reduce yield loss due to salinity and to increase fresh water produc vity. The objec ves are to: i) iden fy combina ons of sowing date and irriga on water management (based on the use of alternate fresh and saline water) that maintain acceptable yield levels in salt-affected areas, and ii) evaluate the effects of growth dura on of a salt-tolerant variety (with the strategies from (i)) on water produc vity and rice produc on. 2. Materials and methods 2.1 Field experiments 2.1.1 Field experiment design Four field experiments were performed with the genotype BRRI dhan47 during the dry seasons of 2013 and 2014, respec vely. Experiments (Expts) 1 and 2 were conducted in Infanta, Quezon Philippines (14045’N, 121041’E) and Expts 3 and 4 in Satkhira, Bangladesh (24012’N’, 90012’E) (Table 1). BRRI dhan47 is one of several recently released salt tolerant varie es used by farmers in Bangladesh, with high yield under stressed condi ons (Islam et al. 2008). Each experiment had four irriga on water management treatments in a randomized block design with three replicates. 478 Table 1. Experiments for model calibra on and valida on 1 Experiment Season Site Expt 1 Expt 2 Expt 3 Expt 4 Infanta Infanta Satkhira Satkhira 2013 Dry 2014 Dry Boro 2013 Boro 2014 Sowing date Transplan ng date January 26 2013 February 16 2013 January 15 2014 February 4 2014 December 20 2012 February 2 2013 January 1 2014 2 February 2014 Flowering date1 April 22 2013 March 31 2014 April 13 2013 April 10 2014 Physiological maturity1 May 21 2013 April 22 2014 May 9 2013 May 6 2014 Dates are for treatments irrigated with fresh water only In Expts 1 and 2, the rice was grown under four different water management condi ons: i) con nuous irriga on with saline water (SW); (ii) con nuous irriga on with fresh water (FW); (iii) weekly alterna on between fresh and saline water (FW:SW); and two weeks fresh and then one week saline water (2FW:SW). Saline water was pumped from the river canal. The fresh water was pumped from groundwater. Bulk soil salinity at 15 cm depth from the soil surface was determined hourly using 5TE sensors (Decagon Devices, USA) installed in each plot and connected to an automa c data-logger. In Expts 3 and 4, there were also four water management treatments. Treatments (i) and (ii) were the same as in Expts 1 and 2 (FW, SW). In treatments (iii) and (iv) fresh and saline water were mixed in different propor ons, rather than alterna ng, as follows: (iii) equal volume of saline and fresh water (FW: SW), and (iv) 2:1 volume of fresh water and saline water (2FW:SW). Mixture of the saline and fresh water was achieved by irriga ng the field with saline water the first half (or two thirds) of the dura on of the irriga on and then with fresh water. Soil salinity was measured with a portable EC meter for one replicate per treatment. In both experiments, the plots were irrigated regularly to maintain a ponded water depth of about 2 to 3 cm minimum and 10 cm maximum. The salinity of both fresh water (from farm dam) and saline water (from river or canal) varied throughout the season (Table 2). In Expts 1 and 2, water salinity was about 0.49 and 5.5 dS m-1 respec vely for the fresh and saline water. In Expts 3 and 4, salinity of the fresh water ranged from 0.65 to 1.88 dS m-1 with no significant varia on during the crop period. Salinity of the saline water ranged from 1.96 to 8.13 dS m-1 and increased with me, reaching a maximum around flowering me (Table 2). Table 2. Salinity of the fresh and saline water in Expts 3 and 4 Water Fresh water Saline water Experiment Expt3 Expt4 Expt3 Expt4 Sowing me 0.65 0.85 1.36 1.96 Transplan ng 0.82 0.92 2.12 2.02 Panicle Ini a on 1.36 1.17 5.03 4.24 Flowering 1.88 1.48 8.13 6.43 2.1.2 Field experiment management The fields were puddled prior to transplan ng. The age of seedlings at the me of transplan ng was 21, 35, 44 d for Expts 1 and 2, Expt 3, and Expt 4, respec vely (Table 1). Two to three and three to five seedlings per hill were transplanted at a spacing of 20 cm x 20 cm in Expts 1 and 2 and in Expts 3 and 4, respec vely. Fer lizers were applied at the recommended rates and mes for each site. The plots were kept weed free and pests and diseases were well controlled in all experiments. 479 2.1.3 Field experiment monitoring Water salinity and soil salinity were determined before and a er each irriga on event, and otherwise weekly. An automated weather sta on was installed near the fields to record rainfall, air temperature, rela ve air humidity and solar radia on on an hourly basis. Biomass sampling was performed at key stages according to the data requirement of ORYZA v3 for model calibra on and valida on (Bouman et al. 2001). Crop grain yield and total above-ground biomass (WAGT) were measured from two sampling loca ons (six hills per loca on, total sampled area about 0.48m2) in each plot. Grain yield was determined on an area of 5 m2 harvested from the middle of each plot, and is reported at 14% moisture content. 2.2 Crop model calibra on A version of the rice crop model ORYZA v3 (IRRI 2014) that includes a new module to account for the effects of salinity on rice growth was used in this study (Radanielson et al. 2015). ORYZA v3 is a rice simula on model offering the poten al to iden fy produc on constraints and op mal management regimes related to nitrogen applica on, water management and crop scheduling (Bouman et al. 2001). Parameters related to salinity responses are variety-specific as described in Radanielson et al. (2013). The two components of salinity effect (osmo c stress and ion toxicity stress) are considered in the model. Osmo c stress is taken into account through an equa on conver ng soil salinity into soil osmo c poten al, reducing water uptake by the crop. Responses to salt accumula on into the plant are represented through a stress factor described with a logis c func on with two genotypic parameters (Radanielson et al. 2013). The factor is applied to the maximum plant photosynthesis rate and the transpira on rate of the plant. The model parameters of BRRI dhan47 for crop phenological development rate were computed using the phenology recorded from each salinity treatment in all four experiments. Other crop parameters for the model were calibrated using data from the treatments con nuously irrigated with fresh water for the two years using the auto-calibra on applica on of ORYZA v3 (IRRI 2014), assuming non-limited environmental growth condi ons at each site (Table 2). Sta s cal criteria for the calibra on were set aiming to minimize the devia on between the simula on outputs and the observed values for the variables “total above-ground biomass (WAGT)” and “grain yield” for the non-stressed (FW) condi ons. As we assumed that salinity responses are variety specific, parameters related to salinity responses were not calibrated but given as inputs to the model. These parameters were es mated from greenhouse experiments characterizing the salinity responses of BRRI dhan47 (Radanielson et al., 2013). The auto-calibra on tool for ORYZA v3 was set to calibrate parameters related to biomass par oning and leaf area expansion. Table 3. Data use for model valida on Calibra on Experiment Expts 1, 2, 3, 4 Treatment FW Variety BRRI dhan 47 Valida on Expts 1, 2, 3, 4 SW, FW:SW, 2FW:SW FW, SW, FW:SW, 2FW:SW BRRI dhan 47 Simula on 480 BRRI dhan47 with Long dura on growth, BRRI dhan47 with medium dura on growth, BRRI dhan47 with short dura on growth Site Infanta Satkhira Infanta Satkhira Satkhira Years 2013-2014 2013-2014 2000-2014 Note: SW, irriga on with saline water; FW:SW, irriga on with alternate one week fresh and one week saline water in Expts 1 & 2 and irriga on with a mixture of 1:1 volume of fresh and saline water in Expts 3 & 4; 2FW:SW, irriga on with two weeks fresh water and one week saline water in Expts 1& 2 and a mixture of volume 2:1 of fresh and saline water in Expts 3 & 4. 2.3 Crop model valida on The ability of the model to represent the observed crop behavior in Expts 1 to 4 was evaluated by comparing the simulated and observed values of total above ground biomass (WAGT) and rice yield of the saline water treatments at all sites (Table 3). Input values of the model parameters for development rate were different for each simulated treatment to inform the simula on of the crop phenology variability under salt stress in ORYZA v3. Similarly, measured values of soil EC in all four experiments were used as input data in the model. Linear regression was used to compare paired data points for measured and simulated above-ground biomass and grain yield. The slope (α), intercept (β), and coefficient of correla on (R2) of the linear regression were computed using the ORYZA Analysis Tools (IRRI 2014). The model performance was assessed using the Student’s t test of means assuming unequal variance P(t) and using the absolute square normalized root of the mean squared error, RMSE n, which was calculated as follows:  (S  O ) i RMSE n  i 2 i1, n (3) n where Si and Oi are simulated and observed values, respec vely, and n is the number of pairs; µ, the overall mean of the observed values. The index of model agreement (ID) was also used as a measure of the models performance, calculated as follows: (4) where Si and Oi are simulated and observed values, respec vely; S’i and O’i are the difference between the simulated and observed values with the overall mean of the observed values; n is the number of pairs. 2.4 Model simula on scenarios Simula ons of different crop management combina ons were performed with the modified version of the rice crop Model ORYZA v3 integra ng a salinity module (Radanielson et al. 2013). Eleven dates of sowing were imposed at seven-day intervals, from December 1 to February 8. The model valida on was carried out with experiments established between the end December and early January. This window of cropping accounted for the usual late December harves ng of the preceding T. Aman rice and allowed sufficient drainage of the field. Six virtual varie es were used. They were characterized with the poten al rice produc on of the variety BRRI dhan47, however each with different phenology parameters to create short, medium and long dura on growth types with the salinity tolerance characteris cs of BRRI dhan47. Irriga on was managed as per the field experiments. The impact of clima c variability was evaluated using historical climate data from 2000 to 2014 for Satkhira. 2.5 Data analysis Simula on outputs considering virtual salt tolerant varie es, use of fresh and saline water, and date of sowing, were analysed with a general linear regression model using R so ware (R Development Core Team 2008). Mean and standard devia on values over the 15 years of simula on were computed to evaluate variability among factors. 481 WAGT, yield, transpired water produc vity and irriga on water produc vity were the variables considered. Transpired and irriga on water produc vity were calculated as the ra o of grain yield to the amount of water transpired or applied, respec vely (kg ha-1 mm-1). Table 4. Sta s cs for comparison of observed and simulated WAGT and yield for the two sites Sites Calibration Infanta Satkhira Validation Infanta Satkhira Variables (kg ha-1) WAGT Yield WAGT Yield WAGT Yield WAGT Yield n P(t*) β 10 4 12 6 11 3 19 9 0.93 0.17 0.88 0.54 0.69 0.82 0.38 0.79 -64.34 454.07 -97.26 -75.02 12.77 1530 -233.48 158.66 R2 1.01 0.97 1.08 1.08 1.15 0.72 0.89 0.97 0.97 0.95 0.99 0.76 0.98 1 0.81 0.85 RMSE RMSEn % 573 14.96 527 20.19 839 13.88 697 13.10 806 24.77 176 3.18 1116 31.84 265 11.85 ID 0.98 0.91 0.98 0.99 0.92 0.93 0.68 0.79 Note: n, number of data pairs; P(t* ), significance of Student’s paired t-test assuming non-equal variances; α, slope of linear regression between simulated and measured values; β, y-intercept of linear regression between simulated and measured values; R2, square of linear correla on coefficient between simulated and measured values; RMSE, absolute root mean squared error; RMSEn, RMSE normalized by Xobs as a percentage; ID, model index of agreement; WAGT, above-ground biomass. 3. Results and discussion 3.1 Model valida on for crop growth BRRI dhan47 was characterized with dura on of 90 d a er transplan ng (DAT) of 29 d old seedlings, and a flowering me of 64 DAT. The crop dura on in Expt 1 was longer than in Expt 2 (Table 1) due to uncontrolled flooding for 5 to 7 days star ng the day a er transplan ng. In Expt 3, the total crop growth dura on is the longest due to the use of old seedlings of about 44 d. Crop growth dura on was reduced by 4 to 6 d under the stressed condi ons with salinity up to 8 dS m-1, similar to the findings of Cas llo et al. (2007). Simulated and observed values for WAGT were generally in good agreement under con nuous fresh water irriga on (Table 4). The model index of agreement was higher than 0.95 with RMSE ranging from 527 to 697 kg ha-1. Table 5. Analysis of variance for simulated yield by ORYZA v3 for the site in Satkhira. Factors Sowing Irriga on Variety Year Sowing x Irriga on Variety x Irriga on Error 482 Df 10 E3 2 13 30 6 4816 SSE 1987544038 2447742453 867936955 334952216 181603652 347892173 5047780741 MSE 198754404 815914151 433968478 25765555 6053455 57982029 1048127 Fvalue 189.63 *** 778.45 *** 414.04 *** 24.58 *** 5.77 *** 55.32 *** Root mean square errors for yield (421 to 785 kg ha-1) were in the range of the error of the observed values, sugges ng acceptable performance of the model. A similar range of RMSE values was reported for the ORYZA2000 rice crop model in a broader analysis (Bouman and van Laar 2006). The model presented good accuracy in simula ng the observed rice WAGT and yields for BRRI dhan47 a er calibra on of the model crop parameters related to leaf growth and biomass par oning. Model performance in simula ng rice crop development and growth under saline condi ons was also generally acceptable, with an RMSE of 806 to 1710 kg ha-1 for above-ground biomass and 176 to 286 kg ha-1 for yield. Over all sites and seasons, an RMSEn of 12.3% was obtained for yield. Simulated above-ground biomass in Expts 3 and 4 had a much higher RMSEn (31.8%) than at Infanta due to underes ma on of the salt stress at this site. Ini al soil salinity was not considered as the model was run under no salt stress during calibra on. For irriga on with various amounts of saline water, the regression of simulated against observed grain yield had an r2 of 0.61 (P<0.001). Overall, the results presented acceptable ability of the model to simulate rice growth and yield in response to water irriga on management op ons that affect soil salinity (Table 4, Fig.1). The results were comparable with the results of simula ons in the Mekong delta using ORYZA2000 (Tuong et al. 2003), in which the authors represented the effect of salinity by reducing water availability for the crop. By considering the effect of salinity on water availability and crop biomass produc on a significant improvement in model accuracy was obtained. 8 Rice Yields 6 FW:SW expt1 2FW:SW expt1 FW:SW expt2 4 2FW:SW expt2 FW:SW extp3 SW extp3 FW:SW expt4 2FW:SW expt4 2 SW expt4 0 0 2 4 6 8 Measured (t/ha) Fig. 1. Simulated and observed rice biomass under salt stress. Each point represents measured and simulated values from irriga on treatments in Expts 1 to 4. The line represents the regressed linear rela onship between observed and simulated values. SW, irriga on with saline water; FW:SW, irriga on with alternate one week fresh and one week saline water in Expts 1 and 2 and irriga on with a 1:1 mixture of fresh and saline water in Expts 3 and 4; 2FW:SW, irriga on for two weeks with fresh water and one week with saline water in Expts 1 and 2 and a mixture (by volume) of 2:1 of fresh:saline water in Expts 3 and 4. 3.2 Effect of irriga on management on soil salinity and yield reduc on Yields were highest with con nuous irriga on with fresh water, and least with con nuous irriga on with saline water (Fig. 3). The use of alternate irriga on with fresh and saline water had a significant effect on soil salinity at each site in comparison with irriga on with fresh or saline water alone. Alterna ng one week each of fresh and saline water reduced soil salinity by 30% to 40% compared with irriga on only with saline water at Infanta (Figs 2a,b). At Satkhira, a 2:1 mixture of fresh and saline water reduced salinity by 45% and delayed 483 the occurrence of salinity >4 dS m-1 by 4 to 5 d compared to the use of con nuously saline water (Figs 2c,d). Using a combina on of fresh and saline water has previously been shown to mi gate the build-up in soil salinity in crop produc on (Malash et al., 2005; Rezaei et al., 2011). In Satkhira, the con nuously increasing soil salinity was a result of the increasing salinity of the river water as the dry season progress. 10 15 8 10 6 4 5 2 0 0 50 80 110 Days a er sowing 140 16 30 60 90 Days a er sowing 120 16 12 12 8 8 2FW:SW FW:SW 4 4 0 0 SW FW 0 50 100 Days a er sowing 150 0 50 100 150 Days a er sowing Fig. 2. Soil salinity dynamics measured at 15 cm of soil depth. Each point represents the mean value of three measurements taken at different points from the plots of the field in Expt 1 (a), Expt 2 (b), Expt3 (c) and Expt 4 (d). FW, irriga on with fresh water; SW, irriga on with saline water; FW:SW, irriga on with alternate one week fresh and one week saline water in Expts 1 and 2 and irriga on with a mixture of 1:1 volume of fresh and saline water in Expts 3 and 4; 2FW:SW, irriga on with two weeks fresh water and one week saline water in Expts 1 and 2 and a mixture of volume of 2:1 of fresh and saline water in Expts 3 and 4. The mean observed yield over the two seasons was about 3500 kg ha- 1 and significant differences were observed between seasons and the sites. Yield variability was driven by variability in both soil salinity and weather condi on (data not presented). 484 5000 4000 3000 2000 1000 0 D ate of sowing 5000 4000 3000 2000 1000 0 Date of sowing SW 5000 FW:SW 4000 2FW:SW FW 3000 2000 1000 0 Date of sowing Fig. 3. Variability (2000-2014) of simulated yield among varie es, date of sowing and water irriga o n management for Satkhira, Bangladesh. Each bar represents an average value of the outputs simulated by ORYZA v3 over the 15 years under the 11 sowing dates and three groups of virtual varie es: a) variety with short growth dura on; b) variety with medium growth dura on; c) variety with long growth dura on; SW, irrigated with saline water; FW, irrigated with fresh water; FW:SW, irrigated with alternate one week fresh water to one week saline water; 2FW:SW, irrigated with alternate two weeks fresh water to one week saline water. 3.3 Op mized cropping calendar and irriga on management for yield improvement at Satkhira Simulated grain yield from 2000 to 2014 at Satkhira presented large variability with an o-efficient of varia on of about 67% over the 15 years of simula on and the factors considered (Fig. 1). The effects of irriga on management, date of sowing and variety (crop growth dura on) on simulated yield were significant (Table 4). The interac on between irriga on management, date of sowing and variety was also significant, sugges ng 485 that a combina on of these factors to op mize rice produc vity would be possible. Minimum and maximum mean yields were 492 kg ha-1 and 5228 kg ha-1 respec vely, corresponding to a short dura on variety sown on February 1 and irrigated with saline water, and a medium dura on variety sown on December 1 and irrigated with fresh water (Fig. 3). With December 1 sowing and irriga on with fresh water (FW), simulated yields of all varie es were similar (Fig 3). Yield of the long dura on variety sown on this date and irrigated with a 2:1 mixture of fresh and saline water (2FW:SW) was also similar to yield of the former group. The combina on of December 1 sowing x long dura on variety x irriga on with 2FW:SW could be an op on to minimize yield loss under shortage of fresh water in salt affected areas. The highest simulated yield with con nuous irriga on with saline water (SW) occurred with a medium dura on variety sown on the same date (Fig. 3). The recommended date for higher yield for boro rice in the coastal zone is earlier than this date considering the boro crop alone (Mondal et al. 2010, 2015). However, the feasibility of earlier sowing would depend on harvest date of the previous crop. Further simula on in cropping sequence would be needed to confirm the earliest date in op mizing annual system rice produc on. Earlier mes of sowing allowed avoidance of high soil salinity and cold temperature stress at flowering me, which resulted in higher grain yield. The use of alterna ng fresh and saline water for irriga on (two weeks to one week) or the 2:1 mixture reduced the yield loss due to salinity significantly. The use of a tolerant variety with medium growth dura on in this condi on would result in similar yield as the non-stressed condi on with a long dura on crop while reducing the use of fresh water for irriga on by 50%. Yield increased with crop dura on within sowing date x irriga on management combina ons (Fig.3). 3.4 Water produc vity In term of transpired water produc vity (WP), the medium dura on variety performed the best, with a value of 16.26 kg ha-1mm-1 for crops sown on February 8 and irrigated con nuously with fresh water (Fig. 4). Under irriga on with con nuously saline water, transpired WP of the long dura on variety sown on the same date presented a higher value of 28.37 kg ha-1mm-1. Under salt stress, plant responses to the stress corresponded firstly to a reduc on in transpira on rate occurring at a lower level of salinity. Plant photosynthesis rate decreases secondly a er accumula on of salt in the plant and at higher salinity (Munns et al. 1995). This conserva on strategy increased the produc vity of the transpired water by the crop under stress and would explain the simulated higher values of transpired WP under irriga on with con nuously saline water. Crops growing under fresh water had the highest value of water produc vity due to higher yield, par cularly for crops sown a er January 11 (Fig. 4). Lowest transpired water produc vity was observed for the short dura on variety sown on January 25 and con nuously irrigated with saline water (4.72 kg ha-1mm-1). This combina on also had the lowest grain yield due to increasing soil salinity between the end of February and early March. This period corresponds to the panicle ini a on stage. Soil salinity during this period exceeded 4 dS m-1, the threshold salinity above which rice growth and yield are impaired (Zeng and Shannon 2000; Gaydon et al. 2014). Con nuous irriga on with fresh water resulted in much lower soil salinity during this period, preven ng yield loss. Yield of the crops irrigated with a mixture of fresh and saline water was significantly lower as soil salinity remained moderate to high, affec ng the crop growth during the sensi ve reproduc ve stage. In our simula on the use of older seedlings reduced the sensi vity of the crops to salinity during early stages. Crops sown a er January 11 were exposed to higher salinity during seedling stage. Late sowing (a er January 25) reduced irriga on requirement by about 30% compared to early sowing (before December 8) with, respec vely, means of 1051 mm and 730 mm over all irriga on water management, varie es and years. This was associated with an increase in in-season rainfall as sowing was delayed. Crop dura on and irriga on treatment also had significant effects on the amount of irriga on water applied. The long dura on crop irrigated con nuously with fresh water had the highest irriga on requirement (1024 mm). The lowest value was observed for the same variety under irriga on with con nuously saline water (790 mm). 486 These values are in the range of amounts reported for rice under non-stressed and stressed condi ons (Zwart and Bas aanssen 2004; Yadav et al., 2012). The three factors had significant effect on the variability of irriga on water produc vity. There was no significant interac on between crop dura on and irriga on management on irriga on water produc vity. Irriga on water produc vity of the short dura on variety was highest with saline water irriga on and sowing on February 8. Irriga on water produc vity considering the amount of fresh water applied has presented the same trends with a mean value of 2.67 kg ha-1 mm-1. Higher irriga on water produc vity was observed for long dura on crop sown on and a er January 25 and irrigated with a mixture of 1:1 ra o of saline and fresh water, with the maximum value of 6.31 kg ha -1mm-1 (fig. 5). The minimum value, observed for the short growth dura on variety irrigated with a mixture volume of 2:1 of fresh and saline water, was about 0.95 kg ha -1mm-1 sown on January 25 (Fig. 5). This strategy of minimizing salinity effect through the use of fresh water and saline water in irriga on could represent an opportunity to expand the cropped area in salt-affected regions. Produc on of cropping systems in these condi ons could be raised by 30 to 40% as observed in the experiments with be er use of fresh water. 487 20 16 12 8 4 0 Date of sowing 20 15 10 5 0 Date of sowing 40 32 24 SW 16 FW:SW 2FW:SW 8 FW 0 Date of sowing Fig. 4. Variability (2000-2014) of transpired water produc vity among varie es, date of sowing and water irriga on management for Satkhira, Bangladesh. Each bar represents an average value of the outputs simulated by ORYZA v3 over the 15 years under the 11 sowing date and three groups of virtual varie es: a) variety with short growth dura on; b) variety with medium growth dura on; and c) variety with long growth dura on; SW, irrigated with saline water; FW, irrigated with fresh water; FW:SW, irrigated with alternate one week fresh water to one week saline water; 2FW:SW, irrigated with alternate two weeks fresh water to one week saline water. 488 5 4 3 2 1 0 Date of sowing 5 4 3 2 1 0 Date of sowing 5 4 3 2 1 0 Date of sowing Fig. 5. Variability (2000-2014) of simulated irriga on water produc vity, among date of sowing and irriga on management for Satkhira, Bangladesh. Each bar represents an average value of the outputs simulated by ORYZA v3 over the 15 years under the 11 sowing date and three groups of virtual varie es: a) variety with short growth dura on; b) variety with medium growth dura on; and c) variety with long growth dura on; FW, irrigated with fresh water; FW:SW, irrigated with alternate one week fresh water to one week saline water; 2FW:SW, irrigated with alternate two weeks fresh water to one week saline water. 4. Conclusion The modified version of ORYZA v3, accoun ng for rice salt stress response, demonstrated acceptable performance in simula ng the growth and yield of the rice cul var BRRI dhan47. The model was able to capture the variability observed in different seasons and sites under a range of salt-affected condi ons. 489 The subsequent scenario analysis has highlighted the opportunity of alterna ng fresh and saline water use in irriga on to make be er use of available fresh water. For sowing from late December to early January, higher yields were obtained with a long dura on variety with the salt-tolerance characteris cs of BRRI dhan47. However, with the medium dura on variety, higher irriga on water produc vity was obtained considering the fresh water applied with acceptable yield. Soil salinity increased con nuously during the crop season in Satkhira. This must be considered in evalua ng the sustainability of the system, par cularly the effect of the saline water on soil degrada on and saliniza on of the ground water. The study would need to then incorporate historical soil salinity trends to capture the actual growing condi ons and also the common prac ces of farmers. Further simula ons with future climate data and in different salt-affected rice growing regions would also be desirable to confirm the study findings and its perspec ve in climate change adapta on. These could be valuable to establish priori es for investments and research orienta on for effec ve strategies to increase rice produc on. Acknowledgments This work was supported in part by the Global Rice Science Partnership (GRiSP), Stress Tolerant Rice Varie es for Africa and South Asia (STRASA), and the Australian Centre for Interna onal Agricultural Research (ACIAR – via project LWR/2008/019 “Developing mul -scale climate change adapta on strategies for farming communi es in Cambodia, Laos, Bangladesh and India (ACCA)”). References Beltrán, J.M.1999. Irriga on with saline water: benefits and environmental impact. Agricultural Water Management 40, 183-194. Bouman, B.A.M., and H.H. van Laar. 2006. Descrip on and evalua on of the rice growth model ORYZA2000 under nitrogen-limited condi ons, Agricultural Systems 87, 249–273. Bouman, B.A.M., M.J. Kropff, T.P. Tuong, M.C.S. Wopereis, H.F.M. ten Berge, and H.H. van Laar. 2001. ORYZA2000: modeling lowland rice. Interna onal Rice Research Ins tute, Los Baños, Philippines, and Wageningen University and Research Centre, Wageningen, Netherlands. Cas llo, E.G., T.P. Tuong, A.M. Ismail and K. Inubushi. 2007. Response to salinity in rice: Compara ve effects of osmo c and ionic stresses. Plant Produc on Science 10, 159–170. Crescimanno, G. and P. Garofalo. 2006. Management of irriga on with saline water in cracking clay soils. Soil Science Society of America Journal 70, 1774-1787. Eatherall, A. 1997. Modelling climate change impacts on ecosystems using linked models and a GIS. Clima c Change 35, 17-34. FAO. 2013. Land and Water Management h p:// h p://www.fao.org/nr/aboutnr/nrl/en/ Accessed 24 May 2013. Flowers, T.J., R. Ragab, N. Malash, G. Abdel Gawad, J. Cuartero and A. Arslan. 2005. Sustainable strategies for irriga on in salt-prone Mediterranean: SALTMED. Agricultural Water Management 78, 3-14. France J. and J.H.M., Thornley.1984. Mathema cal models in Agriculture, Bu erworths, London press. Gaydon, D.S., A.M. Radanielson, O. Angeles, M.M.R. Khan, A.K. Chaki, M.A. Rahman and T. Li. 2014. Inves ga ng op ons to increase fresh water produc vity in Boro rice for salt affected areas of Bangladesh, using crop modeling, Paper presented at the IRC Interna onal Rice Congress 2014, October 26-30, in Bangkok, Thailand. IRRI, 2014. ORYZA2000 : A crop growth simula on model for rice. h ps://sites.google.com/a/irri.org/oryza2000/scien fic-documenta on. Accessed September 2014. 490 Islam, M.R., M.A. Salam, M.A.R. Bhuiyan, M.A. Rahman, R. Yasmeen, M.S. Rahman, M.K. Uddin, G.B. Gregorio and A.M. Ismail. 2008. BRRI dhan 47: A salt tolerant rice variety for Boro season isolated through par cipatory variety selec on. Interna onal Journal of Biological Research 5, 1: 1-6. Ismail, A. M., S. Heuer, M. Thomson and M. Wissuwa. 2007. Gene c and genomic approaches to develop rice germplasm for problem soils. Plant Molecular Biology 65, 547–570. Li., T., O. Angeles, A.M. Radanielson, M. Marcaida III and E. Manalo. 2015. Drought stress impacts of climate change on rainfed rice in South Asia. Climate Change. Under review. Malash N., T.J. Flowers, R. Ragab. 2005. Effect of irriga on systems and water management prac ces using saline and non-saline water on tomato produc on. Agricultural Water Management 78, 25-38. Mondal, M.K., T.P. Tuong, A.K.M. Shariffulah and M.A. Sa ar. 2010. Water supply and demand for dry-season rice in the coastal polders of Bangladesh. In C. T. Hoanh, B. W. Szuster, K. Suan-Pheng, A. M. Ismail and A. D. Noble (Eds.) Tropical Deltas and Coastal Zones: Food Produc on, Communi es and Environment at the Land-Water Interface (CABI publica on): 264-278. Mondal, M.K, N.K. Saha, S.P. Ritu, P.L.C. Paul, A.K.M. Sharifullah, E. Humphreys, T.P. Tuong, and M.A Rashid. 2015. Op mum sowing window for boro cul va on in the coastal zone of Bangladesh. These proceedings Munns, R., Schachtman, D. P., and A. G. Condon, 1995: The signifiance of a 2 phase growth-response to salinity in wheat and barley. Australian Journal of Plant Physiology 22, 561-569. R Development Core Team, 2008. R: A language and environment for sta s cal compu ng. R Founda on for Sta s cal Compu ng, Vienna, Austria. ISBN 3-900051-07-0, URL h p://www.R-project.org. Radanielson, M. A., O. Angeles, T. Li. 2013. Salinity effect on rice biomass: seasonal and transient modelling of plant responses. Paper presented at the Interna onal Society for Ecological Modelling Conference 2013, October 27-31 in Toulouse, France. Radanielson, M.A, D.S.Gaydon, T.Li and O. Angeles. 2015. Modeling salinity effect on rice growth and rice yield with ORYZA v3 and APSIM-Oryza. Ecological Modelling. In prepara on. Rezaei, M., N. Davatgar, A. Ashrafzad, N. Pirmoradian, M.R. Khaledian, M. Kavosi, E. Amiri and M. Vazifedost.2011. Intermi ent irriga on: a procedure to use saline water in rice cul va on. Paper presented at the ICID 21st Interna onal Congress on Irriga on and Drainage, October 15-23, in Tehran, Iran. Sudhir-Yadav, E. Humphreys, T. Li, G. Gill, and S.S. Kukal.2012. Evalua on of tradeoffs in land and water produc vity of dry seeded rice as affected by irriga on schedule. Field Crops Research 128, 180-190. Tuong, T. P., N.D. Phong, and B.A.M. Bouman. 2003. Assessing rice yield in rice–shrimp systems in the Mekong Delta, Vietnam: a modeling approach. In Rice–shrimp farming in the Mekong Delta: biophysical and socioeconomic issues. ACIAR Technical Reports 52e,102-110. Wassmann, R., S. V. K. Jagadish, K. Sumfleth, H. Pathak, G. Howell, A. Ismail, R. Serraj, E. Redona, R. K. Singh, and S. Heuer. 2009. Regional vulnerability of climate change impacts on asian rice produc on and scope for adapta on. Advances in Agronomy, 102, 91-133. Zwart, S.J. and, W.G.M. Bas aanssen. 2004. Review of measured crop water produc vity values for irrigated wheat, rice, co on and maize. Agricultural Water Management 69, 2: 115–133. Zeng, L. and M. C. Shannon. 2000. Salinity effects on the seedling growth and yield components of rice. Crop Science 40,4 :996–1003. 491 Response of wheat, mustard and watermelon to irriga on in saline soils A.R. Akanda1, S.K. Biswas1, K.K. Sarker1, M.S. Mondal2, A.F. Saleh2, M.M. Rahman2 and A.Z.M. Mosleuddin3 1 Bangladesh Agricultural Research Ins tute, Bangladesh razzaquebari@gmail.com, sujitbari@yahoo.com, ksarkerwrc@gmail.com 2 Ins tute of Water and Flood Management, Bangladesh University of Engineering and Technology, Bangladesh mshahjahanmondal@iwm.buet.ac.bd, saleh@iwfm.buet.ac.bd, mmrahman@iwfm.buet.ac.bd 3 Bangladesh Agricultural University, Bangladesh abunajia@yahoo.com Abstract Proper water management is necessary to sustain irrigated agriculture in areas with saline soil and saline water in Bangladesh. Therefore, a study was carried out to inves gate the effects of irriga on on wheat, mustard and watermelon in saline soils of coastal areas. The experiments were conducted in farmers’ fields in Debhata, Satkhira District and Amtali, Barguna District during the rabi season of 2013-2014. The experiments were laid out in a randomized block design with four irriga on treatments of wheat (BARI Gom 25), mustard (BARI Sarisha 14) and watermelon (Hybrid-Big family), replicated three mes. Soil salinity (in situ electrical conduc vity) increased from 2 to 5 dS/m at the me of sowing to 8 to 10 dS/m at the me of harvest with farmers’ prac ce irriga on management. Increasing irriga on frequency from one to two or three in wheat and mustard did not affect soil salinity at any stage throughout the season. However, increasing irriga on frequency from one to two had significant posi ve effects on the growth, yield contribu ng characters and yield of wheat, but li le effect on mustard yield or yield components. Wheat yield increased from about 3 t/ha with one irriga on during the vegeta ve stage to 3.9 t/ha with irriga on at the crown root ini a on (CRI) and boo ng stages. Providing a third irriga on at grain filling did not increase yield further. Increasing irriga on frequency of watermelon from 25 to 15, 10 and 5 d intervals greatly reduced the build up in soil salinity during the second half of the season, and increased watermelon yield by more than 50%. The highest fruit yield of watermelon was around 35 t/ha achieved using 5 and 10 d irriga on intervals, while the farmers’ prac ce produced the lowest yield (20.7 t/ha) at a 25 d irriga on interval. The highest total water produc vity of wheat (2.6 kg/m3) and mustard (1.45 kg/m3) occurred with a single irriga on at the CRI and preflowering stages, respec vely. Water produc vity of watermelon was the highest (8.9 kg/m3) at 10 and 15 d irriga on intervals and lowest (7.8 kg/m3) at 25 d intervals. The results show that high yields of wheat (~4 t/ha), mustard (~1.5 t/ha) and watermelon (~35 t/ha) can be achieved on moderately saline soils of the coastal zone using moderately salt tolerant varie es and proper irriga on management, similar to achievable yields in non-saline regions. Key message: Proper irriga on prac ces can increase the yield of rabi crops by avoiding water deficit stress and suppressing the build up of soil salinity in coastal areas of Bangladesh. Keywords: water produc vity, soil salinity, coastal zone, Bangladesh 1. Introduc on Soil salinity is one of the serious abio c stresses that reduces plant growth, development and produc vity worldwide (Siringam et al. 2011). More than 800 Mha of land throughout the world are affected by varying degrees of salinity (FAO 2008). About 33% of all irrigated lands worldwide are affected by salinity and sodicity. The coastal areas of Bangladesh cover about 20% of the country and comprise more than 30% of the cul vable lands of the country. About 53% of the lands of the coastal areas are affected by varying degrees of salinity. Agricultural land use in these areas is very poor compared to the country’s average cropping intensity of 191% (Haque 2006; BBS 2011). The factors that contribute significantly to the development of saline soil are dal flooding during the wet season (June to October), direct inunda on by saline water and the upward or lateral movement of saline groundwater during the dry season (November to May). The severity of the 492 salinity problem in Bangladesh increases with the desicca on of the soil. The effects of salinity depend on the degree of salinity at the cri cal stages of crop growth, which reduces yield, in severe cases to zero. The dominant crop grown in the saline areas is transplanted aman rice which is grown during the rainy season using tradi onal, low yielding varie es. The cropping pa erns followed in the coastal areas are mainly Fallow-Fallow-Transplanted aman. But recently the cul va on of a wide range of crops such as wheat, mustard, sunflower and vegetables a er the aman harvest has been expanding around some surface water sources and shallow wells with low salinity water (e.g. electrical conduc vity (EC) ranging from 1.7 to 3+ dS/m). The global produc on of watermelon is greater than that of any other cucurbits (Robinson and Decker-Walters 1997), and world produc on has expanded from 2.9 to 3.7 x 106 tons in the period from 1999 to 2004 (FAOSTAT 2004). In Bangladesh too, watermelon produc on is increasing every year as farmers are ge ng good returns. A major share of this crop is cul vated in coastal districts like Bhola, Barguna, Cox's bazar and Patuakhali. It is also a major economic crop in coastal areas of Bangladesh. However, the sustainability of produc on systems based on crops that are not specifically tolerant to salinity requires proper management of both water and salt. With careful water management it is possible to sustain irrigated agriculture in areas with saline soil and saline groundwater with and without subsurface drainage. Many studies have reported substan al increases in crop yields as a result of suitable irriga on management, including studies under saline condi ons (Batra 1990; Ayars et al. 1991; Minhas 1996; Zhang et al. 2004; Malash et al. 2005; Jalota et al. 2006; Ali et al. 2007). In the absence of sufficient rainfall for natural leaching, irrigated farming in arid lands is exposed to accumula on of salts in the soils. Both the quan ty and quality of irriga on water used and their effects on farm produc vity need to be precisely known. Considerable research has been directed towards defining the effects of salinity on crop growth and development (Maas 1990; Shalhevet 1994; Shannon and Grieve 1999). The impact of irriga on with slightly saline water on crops (wheat, mustard and watermelon) and soil has not been studied in coastal regions of Bangladesh. The present study was therefore carried out to iden fy the best irriga on strategy for the cul va on of wheat, mustard and watermelon in coastal saline soils irrigated with low salininity water. The specific objec ves were: (i) to generate physiological data at different growth stages, (ii) to find out the cri cal growth stages for salinity stress, (iii) to find out the effect of salinity on crop yield, and (iv) to develop guidelines for irriga on of rabi crops in the coastal zone of Bangladesh. 2. Materials and methods 2.1 Study area Three experiments were conducted in farmers’ fields in Satkhira and Barguna Districts in Bangladesh during the rabi season of 2013-2014. Field experiments on wheat and mustard were carried out in farmers’ fields in Kulia village, Debhata Upazila, Satkhira District. An experiment on watermelon was conducted in a farmer’s field in Kalibiri village, Amtali Upazila, Barguna District. The soils were clay loam with an average field capacity of 27% (gravimetric water content) and mean bulk density of 1.34 g/cm over the 0-60 cm soil profile (15 cm layers). 2.2 Experimental design Three field experiments were laid out in a randomized complete block design with four irriga on treatments for wheat, mustard and watermelon, and replicated three mes. Each experiment was conducted in a farmers’ field. The unit plot size for wheat, mustard and watermelon was 4 m × 3 m, 5 m × 3m and 3.5 m × 3.5 m, respec vely. The irriga on treatments were as follows: 493 Wheat T1 = Farmers’ prac ce (irriga on during early vegeta ve stage at 35-40 d DAS) T2 = One irriga on at the crown root ini a on (CRI) stage (17-21 DAS) T3 = Two irriga ons, at CRI and boo ng stages (55-60 DAS) T4 = Three irriga ons, at CRI, boo ng and grain filling stages (75-80 DAS) Mustard T1 = Farmers’ prac ce (irriga on during early vegeta ve stage (20–25 DAS) T2 = One irriga on at preflowering stage (30-35 DAS) T3 = One irriga on at siliqua filling stage (45-50 DAS) T4 = Two irriga ons at preflowering and siliqua filling stages Watermelon T1 = Irriga on at 5 d intervals from emergence to fruit se ng T2 = Irriga on at 10 d intervals from emergence to fruit se ng T3 = Irriga on at 15 d intervals from emergence to fruit se ng T4 = Three irriga ons, at 25, 50 and 75 d a er emergence (DAE) (farmers’ prac ce) The experimental blocks were separated by a distance of 2 m and the plots in each block were separated by a buffer of 1 m to prevent lateral movement of water from one plot to another plot. 2.3 Crop management 2.3.1 Wheat and mustard Wheat (BARI Gom 25), a medium salt tolerant variety (no yield penalty up to 6-7 dS/m of soil salinity), was sown at 140 kg/ha on 28 November 2013 with a row spacing of 20 cm. Mustard (BARI Sarisha 14), a low salt tolerant variety (no yield penalty up to 3.5-6.5 dS/m of soil salinity), was sown at 7 kg/ha on 15 December 2013 with a row spacing of 30 cm (BARI 2011). Fer lizer (N120P30K50S20B1Zn4.5 kg/ha) was applied in the forms of urea, triple super phosphate, muriate of potash, gypsum, borax, and zinc sulphate, respec vely. Cow dung at the rate of 5 t/ha was applied before final land prepara on. Two-thirds of the urea and all other fer lizers were applied at the me of final land prepara on, while the remaining urea was applied before the first irriga on. One manual weeding was done for wheat and mustard at 30 and 33 DAS, respec vely. The mustard crops were sprayed with Rovral-50wp at 0.2% at 30 DAS for preven on against diseases. Wheat and mustard were harvested on 11 and 3 March 2014, respec vely. There was no significant weed, pest or disease infesta on in the experimental plots. 2.3.2 Watermelon Watermelon (Hybrid-Big family: no yield penalty up to 7-8.5 dS/m) was planted in pits spaced at 1.5 m x 1.5 m on 17 January 2014, one seed per pit (AHT 2011). Fer lizer (N83P24K40S16kg/ha) was applied in the forms of urea, triple super phosphate, muriate of potash and gypsum, respec vely. Decomposed cow dung was applied at 5 t/ha. Half of the P and K, and the full dose of S and cowdung were applied during pit prepara on. The remaining P and K was applied at the branching stage (30 DAE). The N was applied into four equal splits at 15, 30, 45 and 60 DAE. There were two manual weedings at 23 and 40 DAE. Watermelons were harvested from 15-25 April 2014 as the fruits matured. 494 2.4 Monitoring 2.4.1 Wheat The number of spikes, number of grains per spike, 1000 grain weight, and grain and straw yield of wheat were determined at harvest. Harvest index was calculated as the ra o of dry grain yield to total dry grain plus straw yield. The sample size of the harvested area was one square meter for determining grain and straw yield. Sub-samples (30 plants) from each plot were randomly selected to detemine yield contribu ng characters. 2.4.2 Mustard Plant popula on, number of siliqua per plant, seed per siliqua, 1000 seed weight, and grain and straw yield of wheat were determined at harvest. Harvest index was calculated as the ra o of dry grain yield to total dry grain plus straw yield. The size of the harvested area was one square meter for determining grain and straw yield. Sub-samples (10 plants) from each of the plots were randomly selected to determine yield contribu ng characters. 2.4.3 Watermelon The number of watermelon fruits per plot, average weight of fruit, fruit yield and above-ground biomass (except fruit) were determined at harvest. The number of days to flower ini a on, fruit se ng and days to fruit maturity were also recorded. 2.4.4 Soil water content and irriga on water Soil moisture before each irriga on was monitored by the researchers. Gravimetric soil water content was determined at the mes of sowing and harves ng and before each irriga on. The soil samples were collected considering a root zone depth of 30 cm (ini al stage), 45 cm (vegeta ve stage) and 60 cm (flowering and grain filling stage). The soil was sampled in 15 cm increments, well-mixed, subsampled, weighed, dried at 105oC, and reweighed to determine gravimetric moisture content. The irriga on water requirement was calculated by the following formula (Mandal and Du a 1992; Michael 1978): d = PwAsD 100 Pw = FC – RL where, d = depth of irriga on water to be applied (cm) As = apparent specific gravity of soil D = depth of soil profile to be irrigated (cm) FC = soil moisture content at field capacity (%, g water/100 g soil) RL = residual soil moisture level before each irriga on (%, g water/100 g soil) The amount of applied irriga on water was the depth of water needed to refill the soil profile (0-30, 45 or 60 cm depending on growth stage) to field capacity. Field capacity was determined by ponding water method on the soil surface as suggested by Michael (1978). Total water use (TWU) was calculated as the sum of irriga on input, rainfall and soil water contribu on (SWC) between sowing and harvest. The amount of irriga on water was determined by volumetric measurement and supplied to the experimental plots using a polythene hose pipe. The plots were irrigated with shallow tubewell water with electrical conduc vity (EC) ranging from 1.7 to 3.4 dS/m using groundwater for wheat and mustard, and medium salinity canal water with EC ranging from 4.5 to 6.5 dS/m for watermelon. The rate of ou low of the shallow tubewell was also calculated by volumetric measurement. 495 2.4.5 Soil salinity The soil salinity was monitored in situ at different growth stages of wheat and mustard and at 10 d intervals in watermelon using a portable HI 993310 (Hanna Instruments, Woonsocket, R.I.) water conduc vity and soil salinity meter with steel probes (HI 76305) that can be inserted directly into the soil. The HI 76305 is an amperometric steel probe that measures the total conduc vity of the soil, that is, the combined conduc vity of air, water and soil. Soil salinity was measured from 0-15 and 15-30 cm at at least three spots in each plot. 2.5 Data analysis Crops data were analyzed by ANOVA using MSTAT-C. The least significant difference (LSD) at 5% probability was used to iden fy significant differences between treatments. 3. Results and discussion 3.1 Soil salinity At the me of sowing, soil salinity ranged from about 2.3 dS/m in watermelon to about 4.9 dS/m in wheat. Salinity increased as the dry season progressed in all treatments in all crops, to maximum values at harvest of around 10 and 8 dS/m in wheat (Fig.1a) and mustard (Fig.1b), respec vely, and to around 7 dS/m in watermelon. There was a consistent trend for lower salinity with higher irriga on frequency, but the differences were small and not significant in wheat and mustard throughout the season, and in watermelon un l about half-way through the season. The effect of irriga on treatment on soil salinity in watermelon increased during the la er half of the season, and at the me of harvest salinity ranged from about 8.5 dS/m with farmer prac ce to about 5.5 dS/m with irriga on at 5 d intervals (Fig.1c). Based on observed data, the soils can be classified as moderately saline soils (S2) that can be used for moderately salt-sensi ve crops using proper irriga on management (Majumdar 2004). 496 T1 12 10 8 6 4 2 0 T2 T3 T4 (a) Growth stages of wheat 10 T1 8 T2 T3 T4 6 4 2 0 At sowing Vegetative Flowering Siliqua Grain formation filling (b) Growth stages of mustard Maturity 10 T1 T2 T3 T4 8 6 4 2 0 0 10 20 30 40 50 60 70 80 90 (c) D after planting (DAP) for watermelon Fig. 1. Soil salinity at different growth stages of wheat (a), mustard (b) and at 10 d intervals for watermelon (c) as affected by irriga on treatment. Error bars indicate the standard error of the means. 497 3.2 Response of wheat, mustard and watermelon to irriga on 3.2.1 Growth, yield and yield contribu ng parameters of wheat Irriga on had a significant posi ve effect on the growth, yield and yield contribu ng parameters of wheat (Table 1). One irriga on at CRI did not significantly increase grain or straw yield compared with one irriga on about 20 d later (farmers’ prac ce). Increasing the number of irriga ons from one to two significantly increased grain yield by increasing spike density and grain weight. However, there was no further increase in grain yield (nor in any yield components) by applying a third irriga on during grain filling. However, the third irriga on resulted in a significant increase in straw yield over all other treatments. Applying two or three irriga ons produced a grain yield of around 4 t/ha, which is a good yield in non-saline environments. The effects of the irriga on treatments on soil salinity were small, sugges ng that responses to irriga on were the result of reduced water deficit, and not due to differen al effects of the irriga on treatments on soil salinity. Mridha et al. (2001) reported that op mum wheat yield (3-4.5 t/ha) requires three to four irriga ons for a range of soil types and agroclima c condi ons across Bangladesh. Table 1. Yield and yield contribu ng parameters of wheat under different irriga on levels a Treatment T1 T2 T3 T4 d LSD0.05 f CV(%) Spike/m 2 (no.) 438 478 502 508 30 10.1 Grains/spike (no.) 18.0 21.3 22.3 22.7 e ns 9.5 1000 grain weight (g) 54.3 55.3 57.4 57. 6 2.4 5.6 b Grain yield (t/ha) 3.02 3.35 3.90 4.08 0.52 7.9 Straw yield (t/ha) 5.24 5.28 5.42 6.06 0.71 8.3 c HI 0.37 0.41 0.42 0.40 ns 5.4 a Treatments: T1 = Farmers’ prac ce (irriga on at ini al vegeta ve stage); T2 = One irriga on at CRI; T 3 = Two irriga ons, at CRI and boo ng; T4 = Three irriga ons, at CRI, boo ng and grain filling b Grain yield at 12% moisture content c HI harvest index LSD least significant difference at 5% level d e ns non-significant f CV coefficient of varia on 3.2.2 Yield and yield contribu ng parameters of mustard There were very few effects of irriga on treatment on the performance of mustard, with yield of all treatments around 1.4 to 1.5 t/ha (Table 2). The results are supported by the findings of Sardar et al. (2014) who reported similar mustard yields in moderately saline (1 to 1.6 t/ha) and non-saline (1.4-1.6 t/ha) areas of Bangladesh. There were no effects on plant density, number of seeds/siliqua, seed yield or harvest index (HI). However, plant height, and the number of siliqua/plant was lower with farmers’ irriga on management than for other treatments. Farmers’ prac ce straw yield was significantly lower than yield with two irriga ons. 498 Table 2. Growth, yield and yield contribu ng parameters of mustard under different irriga on levels in saline soil a Treatment T1 T2 T3 T4 b LSD0.05 d CV(%) Plant /m2 82 79 81 92 c ns 9.3 Siliqua/ plant 75.0 93.0 87.7 97.0 13.3 7.8 Seed/ siliqua 36 38 37 40 ns 8.1 1000 seed weight (g) 3.53 3.58 3.58 3.59 ns 5.5 Seed yield (t/ha) 1.42 1.49 1.44 1.52 ns 7.5 Straw yield (t/ha) 9.63 9.94 10.2 10.5 0.76 11.0 HI 0.16 0.16 0.15 0.15 ns 11.1 a Treatments: T 1 = Farmers’ prac ce (Irriga on at ini al vegeta ve stage); T2 = One irriga on at preflowering stage; T 3 = One irriga on at siliqua filling stage; T4 = Two irriga ons each at preflowering and siliqua filling stages b c LSD least significant difference at 5% level ns non-significant d CV coefficient of varia on 3.2.3 Yield and yield contribu ng parameters of watermelon There were consistent trends for faster plant development with reduced irriga on frequency, but the differences were small (maximum 5 d at maturity) and never significant. Highest yield was around 35 t/ha with 5 and 10 d irriga on frequency (Table 3). Sadar et al. (2013) reported that yields of 52 t/ha were possible through proper irriga on in coastal saline areas. There were large effects of irriga on frequency on fruit yield, with significantly lower yield of farmers’ prac ce than all other treatments. Yield with 5 d irriga on interval was significantly higher than with 15 d irriga on interval. Unit fruit weight also increased with irriga on frequency, with significantly higher values for 5 and 10 d intervals than with less frequent irriga on. Table 3. Yield components, yield and biomass of watermelon under different irriga on regimes Flowering Treatment Ini a on (DAS) Days to 80% flowering T1 T2 T3 T4 b LSD0.05 d CV(%) 65 65 63 63 ns 7.4 a 58 58 55 54 c ns 9.1 Days to fruit se ng 72 72 70 68 ns 8.0 Days to maturity 98 98 96 93 ns 10.1 Unit fruit weight (kg) 3.32 3.12 2.62 2.16 0.63 5. 7 Yield (t/ha) ADM (except fruit) 37.7 33.7 28.4 20.5 6.1 5.9 2.40 2.21 1.81 1.61 0.47 5.1 a Treatments: T 1 = Irriga on at 5 d interval a er emergence up to fruit se ng; T2 = Irriga on at 10 d interval a er emergence up to fruit se ng; T3 = Irriga on at 15 d interval a er emergence up to fruit se ng; T4 = Three irriga ons each at 25, 50 and 75 d a er emergence b c LSD least significant difference at 5% level ns non-significant d CV coefficient of varia on 499 3.3 Water input and water produc vity of wheat, mustard and watermelon 3.3.1 Water use and water produc vity of wheat Total water use was similar for farmers’ prac ce and one irriga on at CRI (~134 mm) and increased as the number of irriga ons increased (Table 4). Soil water deple on at harvest decreased slightly as the frequency of irriga on increased. The highest water produc vity (2.60 kg/m3) was obtained with one irriga on at CRI while the lowest (1.93 kg/m3) was obtained with three irriga ons. Table 4. Water use and water produc vity of wheat under different irriga on water regimes a Treatment T1 T2 T3 T4 Total irriga on water applied (mm) 53 41 92 138 Rainfall (mm) 32 32 32 32 b SMC (mm) 54 56 48 41 c TWU (mm) 139 129 172 211 Grain yield (t/ha) 3.02 3.35 3.90 4.08 Water produc vity (kg/m3) 2.17 2.60 2.27 1.93 Note: aTreatments: T 1 = Farmers prac ce (Irriga on at ini al vegeta ve stage); T2 = One irriga on at CRI stage; T3 = Two irriga ons each at CRI and boo ng stages; T 4 = Three irriga ons each at CRI, boo ng and grain filling stages b SMC soil moisture contribu on c TWU total water use 3.3.2 Water use and water produc vity of mustard As with wheat, soil water deple on in mustard decreased as irriga on frequency increased (Table 5). As a result, there was li le varia on in total water use across the four treatments. Water produc vity was highest with a single preflowering irriga on. Table 5. Irriga on water applied and irriga on water produc vity of mustard grown underdifferent irriga on regimes a Treatment T1 T2 T3 T4 a Total irriga on water applied (mm) 60 42 56 89 Rainfall (mm) - SMC (mm) TWU (mm) 54 61 46 39 114 103 112 128 Seed yield (t/ha) 1.42 1.49 1.44 1.52 Water produc vity (kg/m3) 1.25 1.45 1.29 1.19 Treatments: T1 = Farmers prac ce (Irriga on at ini al vegeta ve stage); T2 = One irriga on at preflowering stage; T3 = One irriga on at siliqua filling stage; T4 = Two irriga ons each at preflowering and siliqua filling stages 500 3.3.3 Water use and water produc vity of watermelon Total water use varied greatly across the four treatments, from 261 mm with farmers’ prac ce to 455 mm with a 5 d irriga on interval (Table 6). The varia on in total water use was due to the large varia on in irriga on input, from 187 to 391 mm. The seasonal use of water by the highest yielding treatment was 455 mm, which is within the range of 400-600 mm reported by Doorenboss and Kassam (1979) for watermelon cul va on. Water produc vity was highest with 10 and 15-d irriga on intervals, at around 8.8 kg/m3. Table 6. Water use and water produc vity of watermelon under different irriga on water regimes a Treatment T1 T2 T3 T4 Total water applied (mm) 391 311 250 187 Rainfall (mm) 21 21 21 21 SMC (mm) 43 48 52 53 TWU (mm) 455 380 323 261 Yield (t/ha) 37.70 33.67 28.41 20.47 Water produc vity (kg/m3) 8.29 8.86 8.79 7.84 a Treatments: T 1 = Irriga on at 5 d interval a er emergence up to fruit se ng; T2 = Irriga on at 10 d interval a er emergence up to fruit se ng; T3 = Irriga on at 15 d interval a er emergence up to fruit se ng; T4 = Three irriga ons each at 25, 50 and 75 DAE (farmers’ prac ce) 4. Conclusions and recommenda ons Increasing the irriga on frequency of wheat from one to three, and of mustard from one to two, had only a small effect on the development of soil salinity at two sites with ini al salini es of about 5 and 2.5 dS/m, respec vely. With one irriga on, soil salinity in wheat and mustard increased to 10 and 8 dS/m at harvest. Yield of wheat increased from around 3 t/ha with one irriga on to around 4 t/ha with two or three irriga ons. The results suggest that the yield response to increased irriga on frequency from one to two was due to increased water availability, and demonstrate the good tolerance of BARI Gom 25 to soil salinity. Similarly, there was li le effect of increasing irriga on frequency from one to two on soil salinity in the mustard, nor on crop performance. However, increasing irriga on frequency of watermelon from 25 to 15, 10 and 5 d intervals greatly reduced the buildup in soil salinity during the second half of the season and increased watermelon yield by more than 50%. The highest fruit yield of watermelon was around 35 t/ha at 5 and 10 d irriga on intervals, while the farmers’ prac ce (25 d intervals) produced the lowest yield (20.7 t/ha). The reduc on in yield at lower irriga on frequency may have been due to higher salinity and/or water deficit. The results demonstrate that good yields of rabi crops can be achieved on moderately saline soils of the coastal zone with the standard irriga on management for salinity tolerant varie es of wheat and mustard. For watermelon, 10 d irriga on intervals also give high water produc vity, which might be recommended for irriga on management on moderately saline soils in coastal areas of Bangladesh. Acknowledgements The paper is based on the project research of ESPA funded by DFID. The authors would like to acknowledge DFID, BARI, Gazipur, BUET, Dhaka for providing all facili es and sharing knowledge for sustaining irrigated agriculture in coastal areas of Bangladesh. Finally, we acknowledge the CGIAR Challenge Program on Water and Food for invi ng us to present the findings of this paper at the Revitalizing the Ganges Coastal Zone Conference held on 21 to 23 October 2014, Dhaka, Bangladesh. 501 References Ali, M.H., M.R. Hoque, ,A. A. Hassan, , M.A. Khair. 2007. Effects of deficit irriga on on yield, water produc vity, and economic returns of wheat. Agric. Water Manage. 92: 151–161. Ayars, J.E., Hutmacher, R.B., Steiner, J.J., Mantel, A.B., Vail, S.S. 1991. Irriga on of carrots on a sandy loam soil. Irrig. Sci. 12: 193–198. Akanda, M.A.R., Islam, M.S., Khan, B.R., Mondal, M.R.I.1994. Response of mustard to irriga on at different stages of growth. Annual Research Report 1994, IWM Division, Bangladesh Agricultural Research Ins ue, Gazipur-1701. ATH, 2011. Agriculture Technology Handbook (ATH): Part 1, Published by Bangladesh Agricultural Research Ins tute (BARI), Joydebpur, Gazipur. Batra, B.R. 1990. Effect of different levels of irriga on and fer liza on on growth and yield of carrot (Daucuscarota L.) for root produc on. Vegetable Sci. 17 (2): 127–139. BBS (Bangladesh Bureau of Sta s cs). 2011. Sta s cal year book of Bangladesh. Bangladesh Bureau of sta s cs, Sta s cs Division, Ministry of planning. Government of the people's Republic of Bangladesh. BARI, 2011. BARI Annual Report 2010-2011, Published by Bangladesh Agricultural Research Ins tute (BARI), Joydebpur, Gazipur. Doorenboss, J. and Kassam, A.H. 1979. Yield response to water. FAO Irriga on and Drainage Paper No. 33, Rome. FAO, 2008. Land and Plant Nutri on Management Service. Retrieved from:h p://www.fao.org/ag b/agl /agll/spush/. FAOSTAT, 2004. FAO Sta s cal Yearbooks, World Food and Agriculture Organiza on of the United Na ons. Sta s c Database, Produc on Sta s cs, www.fao.org Haque, A.A. 2006. Salinity Problems and Crop Produc on in Coastal Regions of Bangladesh.Pak. J. Bot., 38(5): 1359-1365.Jalota, S.K., Sood, A., Chahal, G.B.S., Choudhury, B.U., 2006. Crop water produc vity of co on-wheat system as influenced by deficit irriga on, soil texture and precipita on. Agric. Water Manage. 84: 137–146. Maas, E.V. 1990. Crop salt tolerance. In: Tanji, K.K. (Ed.), Agricultural Salinity Assessment and Management, ASCE Manuals and Reports on Engineering Prac ce No. 71. American Society of Civil Engineers, New York, pp. 262–304. Malash, N. Flowers, T.J., Ragab, R., 2005. Effect of irriga on systems and water management prac ces using saline and nonsaline water on tomato produc on. Agric. Water Manage. 78: 25–38. Mandal, M.A.S. and Du a, S.C. 1995. Crop Diversifica on: Findings from a field research programme. Published by The University Press Limited, Red Crescent Building, 114 Mo jheel C.A., Dhaka 1000. Minhas, P.S. 1996. Saline water management for irriga on in India. Agric. Water Manage. 30: 1–24. Michael, A.M. 1978. Irriga on: Theory and Prac ce. 1st Edi on. Vikash Publishing House Pvt. Ltd. New Delhi. 801 p. 502 Mridha, M.A.K., Rashid, M.H., Islam, M.S., and Alam, M.S. 2001. Irriga on requirement of wheat under different soil types and agro-clima c condi ons, Annual Research Report 2001, IWM Division, Bangladesh Agricultural Research Ins ue, Gazipur. Majumdar, D.K. 2004. Irriga on water management: Principles and Prac ce, 3rd prin ng, Prac ce Hill Company of India Private Limited, New Delhi 11001, India. Pasternak, D., De Malach, Y., 1995. Crop irriga on with saline water. In: Pessarakli, M. (Ed.), Handbook of Plant and Crop Stress. Marcel Dekker, Inc., NY, pp. 599–622. Robinson, R.W. and Decker-Walters, D.S. 1997. Curbits (Crop Produc on Science in Hor culture), CABI, ISBN 0851991335. www.cuke.hort.ncsu.edu/curbit/wmelon. Siringam, K., Juntawong, N., Cha-um, S. and Kirdmanee C.2011. Salt stress induced ion accumula on, ion homeostasis, membrane injury and sugar contents in salt-sensi ve rice (Oryza sa va L. spp. indica) roots under isoosmo c condi ons. Afr. J. Biotech 10: 1340-1346. Shalhevet, J. 1994. Using water of marginal quality for crop produc on: major issues. Agric. Water Manage. 25: 233–269. Shannon, M.C., Grieve, C.M. 1999. Tolerance of vegetable crops to salinity. Sci. Hort. 78: 5–38. Sardar, P.K., Hossain, M.M. and Howlader, M.M. 2013. Produc on of high value curbit crops. A way of adapta on to the coastal saline areas of Khulna. On Farm Division, Bangladesh Agricultural Research Ins tute, Daulatpur, Khulna. Sardar, P.K., Hossain, M.M. and Howlader, M.M. 2013.OFRD Khulna: Ac vi es and Achievements, Bangladesh Agricultural Research Ins tute, Daulatpur, Khulna. Yesim, E., Yuksel,A.N. and Orta,A.H. 2001. The Effects of Deficit Irriga on on Watermelon Yield, Water Use and Quality Characteris cs. Pakistan Journal of Biological Sciences, 4: 785-789. Zhang, Y., Kendy, E., Qiang, Y., Changming, L., Yanjun, S., Hongyong, S. 2004. Effect of soil water deficit on evapotranspira on, crop yield, and water use efficiency in NorthChina Plain. Agric. Water Manage. 64: 107–122. 503 Rabi crop establishment methods for increasing land produc vity in the coastal zone of Bangladesh M.N. Rahman1, M.G.M. Amin1, M.K. Mondal2 and E. Humphreys2 1 Bangladesh Agricultural University, Bangladesh, nefaur25@gmail.com, maminbau@yahoo.com Interna onal Rice Research Ins tute, Bangladesh and Philippines, m.mondal@irri.org, e.humphreys@irri.org 2 Abstract Cropping intensity in the coastal zone of Bangladesh is low, with much land lying fallow during the dry season. While many farmers cul vate rabi crops such as sesame, mungbean and grass pea, yields are low (0.5-1 t/ha). Furthermore, sesame and mungbean are not sown un l mid-February to early March and as a result early kharif rains and cyclonic storms o en cause crop damage. Early establishment (December) of rabi crops would reduce the risk of damage by rains and cyclones, and allow for the growth of high yielding and high value crops such as sunflower and maize. However, soil moisture content in December is o en too high for llage. Therefore, no ll methods of crop establishment are needed. Tradi onal prac ce for crops such as maize and sunflower is to sow into cul vated soil, and to hill up the crops to prevent lodging. An experiment was therefore conducted from December 2013 to May 2014 in Khulna District to evaluate crop establishment methods for sunflower and maize. Dibbling shortly a er aman harvest (in early December) was compared with sowing a er llage (a er the soil had dried sufficiently for use of a two-wheel tractor operated power ller (farmers’ prac ce, T1)). There were three dibbled treatments: T2: dibbling in non- lled, almost saturated soil with 5 t ha-1 of rice straw mulch applied immediately a er sowing; T3: as for T2 but without mulch; and T4: as for T3 but with llage (soil spading) and earthing 30 d a er sowing. The dibbled treatments were sown on 10 December 2013, while the lled treatment was sown on 4 February 2014. Early sown sunflower and maize produced grain yields of about 2 and 6 t ha-1, respec vely, despite the lack of irriga on (one irriga on of only 9 mm a er urea topdressing 55 d a er emergence). Delaying sowing to 4 February reduced sunflower yield to 2.2 t ha-1, while the late sown maize was destroyed by the early kharif rains. The results demonstrate the high poten al for increasing crop produc on in the coastal zone by early establishment of rabi crops such as maize and sunflower. Key message: Sunflower and maize can be established by dibbling several weeks earlier than farmers’ prac ce ( llage), with greatly reduced risk of damage from early kharif rains and cyclones. Keywords: dibbling, mulch, sunflower, maize, soil cracks 1. Introduc on The coastal zone of Bangladesh, the most vulnerable area of the eastern Ganges basin, occupies about 30% of the country’s land area and is home to some of the world’s poorest people. The livelihoods of the people of the coastal zone depend primarily on agriculture. Of the 2.86 Mha of coastal and offshore lands of Bangladesh, about 1.06 Mha of arable land are affected by varying degrees (0.3 to 70 dS m-1) of salinity (SRDI 2012). Cropping intensity in the coastal area is low due to soil salinity, unavailability of quality irriga on water in the dry season (Rahman and Ahsan 2011) and the late harvest of the aman crop, which prevents mely establishment of rabi crops (Mondal et al. 2015). Peak salinity of the river water during the dry season exceeds 5 dS m-1 across about 59% of the coastal zone, while river water salinity is less than 2 dS m-1 in about 39% of the region (SRDI 2012). Other challenges to agricultural produc vity include excessive flooding during the rainy season, severe cyclonic storms, and dal surges throughout the year. The major cropping pa erns in medium salinity areas of southwest Bangladesh are aman-rabi and aman-fallow. The predominant cropping prac ce in the study area, polder 30 in Ba aghata Upazila, is 504 tradi onal rainy season aman rice (July to December), followed by sesame (on about 80% of the land area). The sesame is not planted un l mid-February to early March because the soil is too wet for llage prior to this (Ritu and Mondal 2006). As a result, the sesame crops are o en damaged and some mes completely destroyed by early rainfall in March and April and by cyclonic events in May (Mondal et al. 2015). The soil is too wet for llage un l February because the farmers grow photoperiod sensi ve local rice varie es that are not harvested un l December, by which me the weather is cold and foggy, with li le or no soil evapora on un l the weather starts to warm up at the end of January. Furthermore, water drains very slowly from the silty clay soils that predominate in the region. Replacement of the local aman varie es with modern high yielding aman varie es (HYV) with earlier maturity would allow for drainage of the fields in early November, harvest in mid- to late November, and thus vacant fields for rabi crop establishment in December. However, excessive water depth during the rainy season is a major constraint to the use of HYV. Several recent studies have shown that, given the ability to drain excess water, HYV can be successfully grown during the rainy season with yields of 4 to 5 t ha-1 (Bha acharya et al. 2015; Mondal et al. 2015; Saha et al. 2015). Furthermore, Mondal et al. (2015) showed that it is possible to manipulate the sluice gate in a sub-polder to enable effec ve drainage, and to harvest HYV before the end of November. The early aman harvest then creates the opportunity to grow high yielding and high value crops such as maize and sunflower. Early plan ng of these crops is desirable for many reasons including: (i) to make be er use of residual soil moisture, (ii) to avoid exposing the crop to high soil salinity, which increases as the dry season progresses, (iii) to avoid damage from the early kharif rains and cyclones in May, and (iv) to avoid grain filling during the ho est me of the year. However, soil moisture may be too high for llage shortly a er rice harvest, especially on the heavy textured soils of the coastal zone. Therefore, no or reduced ll methods of crop establishment are needed. The present study was therefore undertaken to evaluate crop establishment techniques for maize and sunflower in a medium salinity region of the coastal zone. 2. Materials and methods 2.1 Site A field experiment evalua ng establishment methods of sunflower and maize was conducted from December 2013 to May 2014 in polder 30 at Kismat Fultola village (22.7263°N, 89.5163°E) in Ba aghata Upazila, Khulna District. The area belongs to the agro-ecological zone of the Ganges dal floodplain. The climate is subtropical, with average annual rainfall of 1,850 mm, a cool short winter from December to February with monthly mean minimum temperatures of 12–15°C, and a hot summer in April to June (monthly mean maximum temperatures of 37–38°C). The soil in the experimental field was a silty clay with a bulk density of 1.45 g cm-3 at 15 cm (Mondal et al. 2006). Salinity of the topsoil (0–15 cm) varies from 4.0 to 12.5 dS m-1 (electrical conduc vity of the satura on extract) in the dry season and remains below 4.0 dS m-1 in the wet season (Mondal et al. 2006). River water salinity remains below 1 dS m-1 from July to December and gradually increases from January to a peak of around 20 dS m-1 in April or May (Mondal et al. 2006). The salinity of the groundwater at 40 m depth at the study site is about 4 dS m-1 throughout the year. The experimental site has a long history of transplanted aman using tradi onal varie es, followed by rabi crops, usually sesame. In the two years preceding the experiment high yielding varie es of aman were grown. 2.2 Experimental design The experiment compared two rabi crops (sunflower, maize) and four establishment methods (see below) in a randomized complete block design with five replicates. Plot size was 5 m × 10 m and there was a buffer of 0.5 m between plots. The establishment methods were: 505 T1: dibbling a er plowing with a power ller powered by a two-wheel tractor (farmers’ prac ce) T2: dibbling in non- lled, saturated soil with 5 t ha-1 (~100 mm thick) of rice straw mulch applied immediately a er sowing, between the rows T3: as for T2 but no mulch T4: as for T3 but with llage (using a spade) to loosen the soil followed by earthing up of the plants once soil moisture had decreased to field capacity (30 d a er sowing, DAS) 2.3 Crop management The aman crop was drained in early November 2013 and harvested in the third week of November, with the straw removed at ground level. The non- lled treatments (T2-4) were sown by dibbling into the saturated soil on 10 December, while sowing of T1 was delayed un l the soil was dry enough for llage using a ller powered by a two-wheel tractor on 2 February followed by sowing on 4 February. Crop management was according to BARI recommenda ons (Hossain et al. 2006). Maize (Pacific-984; dura on 125–135 d) and sunflower (Hysun-33; dura on 110–120 d) were sown on a rectangular grid with two seeds per intersect (“hill”) at a depth of 2 cm. Hill spacing of sunflower and maize was 60 cm x 45 cm (3.7 hills m-2) and 45 cm x 30 cm (7.1 hills m-2), respec vely. Plants were thinned to one plant per hill 15 d a er emergence (DAE). The dose of N-P-K-S-Zn-B was 258-58-140-33-4-2 kg ha-1 for maize, and 55-26-25-20-1-2 kg ha-1 for sunflower, as per BARI recommenda ons (Hossain et al. 2006). All the fer lizer except for 50% of the urea was applied immediately a er sowing by removing a chunk (15 cm x 10 cm x 10 cm, LxWxD) of soil about 7 cm to the side of each seed posi on, applying the correct amount of fer lizer to the hole using a calibrated dispenser, and replacing the soil. The rest of the urea was broadcast 55 d a er DAE followed by a light irriga on (~9 mm, using a hand held hose) to wash the urea into the soil. There were no other irriga ons because of the increasing salinity (from February onwards) of the water in the khal adjacent to the experimental site (Mondal et al. 2015). Disease and insect pests were effec vely controlled when necessary as per BARI recommenda ons (BARI 2011). Leaf blight of sunflower was controlled by applying Al ma Plus 40 WP (emamec n benzoate + thiamethoxam) at 75 g/ha on 5 March 2014. Hairy caterpillar of sunflower was controlled by applying Knowine 50 WG (carbendazim 50%) on 11 March 2014 at 1.0 kg/ha. Preventa ve spraying of the insec cide Virtako 40 WG (chlorantraniliprole 20% + thiamethoxam 20%) was done at 75 g/ha on 26 January and 9 February 2014. 2.4 Crop monitoring Seven randomly selected (non-border) plants in each plot were tagged and monitored from 45 DAS to harvest to monitor crop development, height (every 15 to 18 d) and yield components at maturity. Crop development stages were taken as the date when four of the seven tagged plants had reached each stage. The start of flower bud ini a on in sunflower was taken as the date when the terminal buds formed a miniature floral head at the apex of the stem. The start of flowering was determined as the date when all disk florets were open. The date when 80% of the seeds had become hard in each flower head was also recorded. Physiological maturity was taken as the date when all the seeds had turned black and shiny. At maturity, sunflower head diameter, the number of seeds and empty florets per head, and the weight of 1000 seeds were determined. Head diameter was measured with a 30 cm stainless steel scale then the heads were sun dried for three days and threshed manually prior to coun ng and weighing 1000 seeds. In maize, tasseling was taken as the stage when all branches of the tassel were visible, and silking when the silks were visible outside the husks. The milky stage was the date when all the kernels were yellow and filled 506 with white (“milky”) fluid. Physiological maturity was defined as the date when the kernels had turned shiny and golden. At maturity, plant density, the number of cobs per plant, cob length, cob circumference at the widest part, the number of kernels cob-1 and 1000-kernel weight were determined from the seven tagged plants. Seed/grain yield of both crops was determined by harves ng and manually threshing all the plants in each plot for both crops. The sunflower of T2, T3 and T4 were harvested on 10 April 2014, while T1 was harvested on 25 May 2014. Treatments 2-4 of maize were harvested on 5 May 2014. Most of the maize plants in T1 were destroyed by heavy rainfall on 2 May 2014, when this treatment was at the silking stage. The sunflower heads and cobs were sun dried for 3 d prior to manual threshing and weighing of the grain. Grain moisture content at the me of weighing was not determined. The crop data were analyzed for significance of difference between the treatment means at 95% probability using the Sta s cal Tool for Agricultural Research (STAR) so ware (STAR 2014). 2.5 Water monitoring 2.5.1 Irriga on Irriga on amount was determined from the pump discharge rate and the me taken to irrigate. The discharge rate was determined by measuring the me taken to fill a 50 L drum (average of three determina ons). 2.5.2 Soil water tension Soil water tension was measured daily using tensiometers installed at 37.5 cm depth in one replicate of T2, T3 and T4 in both the maize and sunflower. The tensiometers were installed on 21 January 2014 and monitored un l the soil became too dry for their use (late March). 2.6 Weather A standard rain gauge and a USWB Class A evapora on pan were installed about 200 m from the experimental field and monitored each morning at 9 am. Daily maximum and minimum temperature were collected from the Khulna meteorological sta on, about 8 km north of the experimental site. 3. Results and discussion 3.1 Weather Rainfall was low throughout the growth period of all crops (Fig. 1). There was a mely fall of 23 mm on 16-17 February, just before flower bud ini a on of the sunflower, and 13 mm on 23-25 March during the la er part of seed filling of the sunflower. No other rain was received during the growth period of T2-4 of both crops. The late sown sunflower (T1) received 40 mm of rain on 2-4 May at the beginning of seed development. 507 T1 lodged Maize T2-4 Maize T1 Sunflower T1 Sunflower T2-4 1 21 41 61 81 101 121 Days a er 30 Nov. 2013 141 161 181 Fig. 1 Daily rainfall throughout the cropping period (horizontal lines show the cropping period from sowing to harvest for each treatment; maize T1 was destroyed by a storm at silking, which caused lodging). Monthly pan evapora on increased from 60 to 80 mm from December to February, and then increased more rapidly to a peak of 199 mm in April (Fig. 2). Monthly mean daily maximum temperature varied from 26°C (January) to 36°C (April), while mean minimum temperature varied from 15°C (in January) to 26°C (in May). Mean daily sunshine hours were least in January (5.4 h) and highest in April (8.8 h). 260 240 220 200 180 160 140 120 100 40 Rai nfall Ev apora on Tmin Tmax Sunshine hour 35 30 25 20 15 80 60 40 10 5 20 0 0 Fig. 2. Monthly total rainfall and evapora on at the study site, and monthly means of daily sunshine hours, and maximum (Tmax) and minimum (Tmin) temperatures at Khulna weather sta on, from December 2013 to May 2014. 3.2 Soil water tension Soil water tension at 37.5 cm increased a er sowing un l a light irriga on was applied on 7 February (Fig. 3). This small amount of irriga on decreased soil tension in the non-mulched treatments because the soil was cracked (2-4 cm wide cracks to a depth of 20 to 35 cm) and the irriga on water flowed down the cracks and wet the soil at depth. The irriga on did not affect soil water tension in the mulched treatments as, while there were also large soil cracks, the width and depth of cracking was less in these treatments at this stage. Soil water tension at 37.5 cm was much lower in the mulched treatment (T2) than in the non-mulched treatments (T3 and T4) during the first three months in the sunflower, and during the first two months in the maize (Fig. 508 3). The slower soil drying in the mulched treatments was probably due to reduced soil evapora on – it is well established that mulch reduces soil evapora on, more so when the plant canopy is small (Bond and Willis 1969). As the season progressed, soil drying in the mulched treatments increased so that soil water tension became similar to that in the non-mulched treatments in early March (maize) and in late March (sunflower). As there was s ll a cover of mulch on the soil surface up to maturity, the greater drying in the mulched treatments at 37.5 cm as the season progressed probably reflects extrac on of soil water by the roots. The more rapid soil drying during the first two months in the mulched maize than in the mulched sunflower suggests faster water uptake by the maize at that depth during the vegeta ve stage. 100 80 R/I T2 T3 T4 Sunflower 25 20 60 15 40 10 20 5 0 0 100 80 60 R/I T2 T3 T4 Maize 25 20 15 40 10 20 5 0 0 Fig. 3. Soil water tension at 37.5 cm depth for different establishment methods of sunflower and maize (R: rainfall, black bars; I: irriga on, blue bar). 509 3.3 Performance of sunflower 3.3.1 Sunflower development Growth dura on of the early sown sunflower (T2-4) was 117-121 d, compared with 110 d for the late sown sunflower (T1) (Table 1). Flower bud ini a on and physiological maturity of T2-4 occurred 68-70 and 109-113 DAS, respec vely. There was a small but consistent trend for mulch to delay development (by 3-4 d at physiological maturity, PM). Crop dura on of the later sown sunflower (T1) was about 10 d less than that of T2-4, due to a shorter vegeta ve phase (shorter me from sowing to the start of flower bud ini a on). In the same year, Bha acharya et al. (2015) also found a reduc on in dura on of sunflower with delay in sowing from mid-December (from 113-115 d) to mid-January (107-108 d), and a small delay (1-4 d) in maturity with mulching, at Patuakhali. Table 1. Development stages of sunflower (DAS) Treatment T1 – cul vated, sown 4 Feb T2 – ZT, mulch, sown 10 Dec T3 – ZT, no-mulch, sown 10 Dec T4 – ZT, no-mulch, sown 10 Dec, spading & earthing up 30 DAS Start of flower bud ini a on 58 70 68 70 Start of flowering 74 84 82 83 80% seeds hard 89 102 100 100 Physiol. maturity 104 113 109 110 3.3.2 Growth parameters of sunflower There was a consistent trend for mulch to increase plant height in comparison with all other treatments, with significant differences at maturity (Fig. 4, Table 2). Plant height at maturity with conven onal prac ce (T1) and in the spaded treatment (T4) was significantly lower than in both the mulched and non-mulched zero llage treatments, by around 10 cm. The mulched sunflower also had significantly larger head diameter than in all other treatments, while head diameter with conven onal prac ce was significantly lower than in all other treatments. 250 T1 200 T2 T3 150 T4 100 50 0 Fig. 4. Plant height of sunflower during the growing period as affected by establishment method. Ver cal bars (barely visible) are standard errors of the means. 510 3.3.3 Yield and yield components of sunflower The number of heads per m2 was the same in all treatments (3.7 m-2). The number of florets per head in the early sown treatments (1400-1480) was significantly higher than in the late sown treatment T1 (1200) (Table 1). Floret fer lity was slightly but significantly higher in the late sown treatment than in all other treatments except T3 (dibbled, no mulch), and significantly lower in the mulched treatment than in all other treatments. Average seed weight was significantly lower with late sowing (by about 20%) than in the early sown treatments. There appears to be a systema c error in seed weight, as the 1000 seed weight of Hi-Sun33 is typically 59-69 g observed by Rahman (2012) in the coastal region of Bangladesh and 52-57 g recorded by Nasim et al. (2012) in Pakistan, however, the rela ve seed weight of the treatments appears sensible. The high 1000 seed weight is probably due to lack of proper drying of the seed. Seed yield of all early planted treatments was similar (2.9-3.2 t ha -1) and about 50% higher than that of the late sown treatment, mainly due to lower seed weight, and partly due to fewer florets per head (Table 2). The poorer performance of the late sown crop is probably due to a combina on of factors: lower residual soil water content at the me of sowing, higher evapora ve demand throughout the crop growth period and higher soil salinity. The lower seed weight indicates stress during the grain filling period, most likely water deficit stress due to the lack of rainfall and the higher evapora ve demand experienced by this treatment. Yields of 3 t ha -1 of the early sown treatments were very good given the lack of irriga on, reflec ng the ability of the crop to make use of residual soil water to depth, and the mely rain at flower bud ini a on. It is likely that the yields are overes mated by about 50% due to insufficient drying. BRAC (2010) reported an average yield of 3 t ha-1 of sunflower in the region during the rabi season using two to three irriga ons. Rashid et al. (2014) reported a yield of 3.1 t ha-1 for 14 January plan ng and 2.5 t ha-1 for late plan ng (5 March) in the coastal zone. Rashid et al. (2015) showed that yield declined by 22 kg ha-1 for every day delay in sowing a er 5 December. Table 2. Plant height at maturity, yield and yield a ributes of sunflower as affected by establishment method (the number of heads per m2 was 3.7 in all treatments) 1 Treatment Plant height (cm) T1 T2 T3 T4 LSD 0.05 CV % 135 148 144 136 7.6 11.4 Head diameter (cm) 17.8 23.5 21.1 21.9 1.2 12.5 No of florets1 per head 1200 1480 1460 1400 92 14.1 Floret fer lity (%) 1000-seed wt (g) Seed yield (t ha-1) 97.4 90.7 94.9 93.9 2.6 5.8 69.6 93.7 87.3 87.6 6.8 17.09 2.2 3.2 2.9 3.1 0.4 9.4 Disk flowers 3.4 Performance of maize 3.4.1 Maize development The early sown maize (T2-4) reached the tasselling stage 91-94 DAS, and physiological maturity at 139-143 DAS (Table 3). Tasseling of the late sown maize (T1) occurred 7-10 d earlier and silking occurred 7-11 d earlier than of the earlier sowing crops. However, the late sown maize was destroyed by lodging as a result of strong winds and 35 mm rain in early May (Fig. 1) when the crop was at the silking stage. There was a small but consistent trend for mulch to delay tasseling (by 2-3 d), silking stage (by 3-4 d), the milky stage (by 3-6 d), and physiological maturity (by 3-4 d). 511 Table 3. Development stages of maize (DAS) Treatment Tasseling Silking T1 – cul vated, sown 4 Feb T2 – ZT, mulch, sown 10 Dec T3 – ZT, no-mulch, sown 10 Dec T4 – ZT, no-mulch, sown 10 Dec, spading & earthing up 30 DAS 84 94 91 92 87 98 94 95 Milky kernel stage destroyed1 119 113 116 Physiol. maturity destroyed 143 139 140 Crop lodged as a result of rain and wind at the silking stage 1 3.4.2 Maize growth parameters There was a consistent trend for significantly taller maize plants in the mulched treatment than in other treatments sown at the same me, with significant differences during the second half of the season (Fig. 5, Table 4). At maturity, the mulched plants were significantly taller than the other early sown treatments, by 30-40 cm. But spading and earthing up gave the longest cobs (19.8 cm). T1 200 T2 T3 150 T4 100 50 0 Fig. 5. Plant height of maize during the growing period as affected by establishment method (T4 lodged in early May as a result of rain and strong wind). Ver cal bars are standard errors of the means. 3.4.3 Yield and yield a ributes of maize Plant density was 7.1 m-2 in all plots and very few plants had more than one harvestable cob per plant. There were no significant treatment effects on the number of kernels per cob (mean 385) (Table 4). Average kernel weight was significantly higher with spading and earthing up than in the no ll treatments. Maize yield ranged from 5.7 to 6.2 t ha-1, with no significant differences between treatments. In well-irrigated maize at Patuakhali, Bha acharya et al. (2015) achieved higher yields (about 9 t ha-1) in the same season, with 500 to 600 kernels per cob and 1000 kernel weight of 350 to 370 g (J. Bha acharya pers. comm.). In the same region, BRAC (2010) obtained maize yield of 8 t ha-1 using two to four irriga ons. 512 Table 4. Growth, yield and yield a ributes of maize as affected by establishment method (the number of cobs m-2 was 7.1 in all treatments) Treatment Plant height (cm) T2 T3 T4 LSD 0.05 CV % 213 174 183 10 11 Cob length (cm) 17.6 14.8 19.8 1.7 20.5 Cob perimeter (cm) 12.4 12 12.9 ns 12.9 No of kernels cob-1 395 373 388 ns 37.5 1000 kernel wt (g) Seed yield (t ha -1) 237 248 282 27 22.7 5.7 5.7 6.2 ns 12. 4. General discussion Dibbling of sunflower and maize into moist soil (close to saturated) on 10 December, shortly a er aman harvest, produced yields of around 3 and 6 t ha -1, respec vely. These yields were achieved in the absence of irriga on (apart from 9 mm applied a er urea topdressing), making use of residual soil moisture and a small amount of in-season rainfall (36 mm in two events). There was a 50% reduc on in yield of the late sown sunflower following conven onal llage once the soil had dried sufficiently. Conven onal llage delayed crop establishment by almost two months, and the reduc on in yield was probably partly due to soil water deficit during the la er part of the season, resul ng in fewer florets per head and lower seed weight. However, there is also evidence that delaying sowing of sunflower beyond mid-December also reduces yield of well-irrigated crops (Rashid et al. 2015). The late sown maize was destroyed by wind and rain at the silking stage, which caused lodging. If it had not been destroyed at that stage, it would probably have been damaged by the 204 mm of rain that fell in the last week of May. The late sown sunflower matured just before this rain and thus serious damage was avoided. The work demonstrates that early establishment of sunflower and maize is possible and can produce good yields even in the absence of llage and irriga on. Furthermore, early sowing greatly reduces the risk of crop damage as a result of strong wind and rain that occur in April in some years, and the risk of damage from cyclonic storms that tend to occur in May. The shorter dura on of sunflower is also an advantage in terms of crop water requirement and risk of damage from early rains. Mulch was effec ve in conserving soil moisture during the vegeta ve stage of the crops (as evidenced by the soil tension data, and the be er growth and visual appearance of the mulched crops), but this did not translate into yield. In the case of sunflower, yields of the early sown sunflower were similar to those expected of well-irrigated crops, sugges ng that these crops did not suffer from water deficit stress, despite the lack of irriga on. This may have been because of three main factors: (i) the high plant available water on the silty clay soil (~200 mm in the top 60 cm, Michael 1985), which was saturated to depth at the me of seeding, (ii) the strong tap root in sunflower and its ability to extract residual soil water to about 1 m depth (Jensen, undated: h p://homeguides.sfgate.com/kinds-roots-sunflowers-have-60427.html), and (iii) the low evapora ve demand during the first two months a er sowing (Fig. 2). In the case of maize, yields were about two-thirds of those of a well-irrigated maize crop in the region, sugges ng increasing soil water deficit stress as the maize season progressed, as evidenced by fewer kernels per cob and lower kernel weight than in high yielding crops of this variety. The lack of effect of mulch in the maize may have been due to the fact that the soil in all treatments ul mately dried to the degree that T2-4 experienced similar levels of water deficit stress during the second half of the maize season. 513 5. Conclusions This study shows the feasibility of growing sunflower and maize in a medium salinity region of the coastal zone, with crop establishment in almost saturated soil shortly a er aman harvest. Early establishment (early December) by dibbling in non- lled soil gave much higher yield than late establishment a er the soil had dried sufficiently for llage to take place. Late establishment resulted in destruc on of the maize crop by early rains. On the silty clay soil at this loca on, respectable yields of sunflower (est. 2 t/ha) and maize (6 t/ha) were achieved with only one light irriga on. Mulching conserved soil moisture, reduced soil cracking and improved crop growth, but did not increase yield, probably because all crops eventually ran out of water. Further studies on the use of mulch and standing stubble, soil water dynamics and establishment method are needed across a range of situa ons (soil type, land eleva on) to fine tune the guidelines for the produc on of high yielding rabi crops in the polders of the coastal zone of Bangladesh, and to determine the profitability of produc on. Acknowledgements This paper presents findings from G2 ‘Produc ve, profitable, and resilient agriculture and aquaculture systems’, a project of the CGIAR Challenge Program on Water and Food. The authors are grateful for high quality technical assistance from Amal Ray, Lincoln Ray, Tanmoy Ray and Mithun Ray. References Agele, S.O., Olaore, J.B. and Kinbode, F.A.A. 2010. Effect of some mulch materials on soil physical proper es, growth and yield of sunflower (Helianthus annuus, L). Advan. Environ. Biol., 4(3):368–375. BARI (Bangladesh Agricultural Research Ins tute). 2011. Regional yield trial of sunflower varie es/lines. Regional review paper. Jessore. pp.136–138. Bha acharya, J., Mondal, M.K., Humphreys, E., Saha, N.K., Rashid, M.H., Paul, P.C. and Ritu, S.P. 2015. Rice-rice-rabi cropping systems for increasing the produc vity of low salinity regions of the coastal zone of Bangladesh. These proceedings. Bond, J.J. and Willis, W.O. 1969. Soil water evapora on – surface residue rate and placement effects. Soil Sci. Soc. Am. Proc. 33: 445–448. BRAC. 2010. Crop intensifica on for achieving food self-sufficiency in the coastal regions of Bangladesh. Dhaka. Bangladesh. Hansen, V.E., Israelsen, O.W. and Stringham, G.E. 1980. Irriga on principles and prac ces. John Willey and Sons Inc. USA, p 44. Haque, S.A. 2006. Salinity problems and crop produc on in coastal regions of Bangladesh. Review ar cle. Pak. J. Bot., 38(5):1359–1365. Hossain, M.H, M.M. Rahman, M.H. Rashid, A.T.M. Farid, M.A. Quyum, M. Ahmed, M.S. Alam, and K.M. Sahahuddin. 2006. Krishi Projuk Hatboi (Handbook on agro-technology), fourth edi on, Bangladesh Agricultural Research Ins tute, Gazipur 1701, Bangladesh. Jensen, L. What kind of roots do sunflower have? Available at: h p://homeguides.sfgate.com/kinds-roots-sunflowers-have-60427.html Michael, A.M. 1985. Irriga on theory and prac ce. Vikash Publishing House Pvt. Ltd, New Delhi, India. pp. 478-482. 514 Mondal, M.K. 1997. Management of soil and water resources for higher produc vity of the coastal saline rice lands of Bangladesh. PhD thesis, University of the Philippines LosBanos, Philippines. Mondal, M.K., S.P. Ritu, M.H.K. Choudhury, A.M. Chasi, M.A. Mannan, P.K. Majumder, M.M. Islam, and S.K. Adhikary. 2004a. Appropriate water management strategy for boro rice cul va on in the coastal saline environment. In proceedings of the workshop on Coastal water management and uptake technologies held at Hotel Castle Salam, Khulna, Bangladesh on 26 June 2004. pp.28–37. Mondal, M.K., T.P. Tuong, S.P. Ritu, M.H.K. Choudhury, A.M. Chasi, P.K. Majumder, M.M. Islam, and S.K. Adhikary. 2006. Coastal water resource use for higher produc vity: par cipatory research for increasing cropping intensity in Bangladesh. CAB Interna onal 2006. Environment and Livelihoods in Tropical 72 Coastal Zones, pp.72–85 (eds C.T. Hoanh, T.P. Tuong, J.W. Gowing and B. Hardy). Mondal, M.K., Humphreys, E., Tuong, T.P., Rahman, M.N. and Islam. M.K. 2015. Community water management and cropping system synchroniza on: The keys to unlocking the produc on poten al of the polder ecosystems in Bangladesh. These proceedings. Nasim, W., A. Ahmad, A. Bano, R. Ola nwo, M. Usman, T. Khaliq, A. Wajid, H.M. Hammad, M. Mubeen and M. Hussain. 2012. Effect of nitrogen on yield and oil quality of sunflower (Helianthus annuuas L) hybrids under sub humid condi ons of Pakistan. American J. Plant Science (3):243-251. Rahman, M.M. 2012. Enhancement of resilience of coastal community in Bangladesh through crop diversifica on in adapta on to climate change impacts. Unpublished thesis of Masters in Disaster Management, BRAC University, Dhaka, Bangladesh. Rahman, M.M. and M. Ahsan. 2011. Salinity constraints and agricultural produc vity in coastal saline area of Bangladesh. Soil Resources in Bangladesh: Assessment and U liza on. Rashid, M.H., S. Nasrin, and D. Mahalder. 2014. Zero lled dibbled sunflower enables plan ng earlier and harvests more in the coastal saline area of Bangladesh. Int. J. Environ. Sci. Develop., 5(3):260–264. Rashid, M.H., Hossain, F., Nath, D.K., Sarker, P.C., Ferdous, A.K. M. and Russel, T. 2015. Oil seed crops in rice based cropping systems in southern Bangladesh. These proceedings. Ritu, S.P. and M.K. Mondal. 2006. Soil and water salinity rela ons in the southwest coastal region of Bangladesh. J. Progres. Agric., 17(1):249–256. SRDI (Soil Resources Development Ins tute). 2012. Saline soils of Bangladesh. Soil fer lity assessment, soil degrada on and its impact on agriculture program of SRDI, Mri ka Bhaban, Krishikhamar Sarak, Farmgate, Dhaka 1215. Saha, N.K., Mondal, M.K., Humphreys, E., Bha acharya, J., Rashid, M.H., Paul, P.L.C. and Ritu, S.P. 2015. Triple rice in a year: Is it a feasible op on for the low salinity areas of the coastal zone of Bangladesh? These proceedings. STAR. 2014. Biometrics and Breeding Informa cs, PBGB Division, Interna onal Rice Research Ins tute, Los Baños, Laguna. Available at h p://bbi.irri.org/ 515 Screening of watermelon varie es for the coastal area of Khulna M.M. Hossain, S.M. Zaman, P.K. Sardar and M.M. Howlader Bangladesh Agricultural Research Ins tute, Bangladesh mosharraf74@yahoo.com, sm_zaman60@yahoo.com, dr.proshanta74@yahoo.com, mohsinhawlader@yahoo.com Abstract A field experiment was conducted at a mul -loca on trial site, Dacope, during the 2011-12 and 2012-13 rabi seasons. The aim of the study was to evaluate the performance of watermelon varie es in a moderately saline area of the coastal zone. The design of the field experiment was a randomized complete block with four replicates. Four commercial watermelon varie es viz. World Queen, Field Master, Hunter and Winner, were evaluated. Significant varia on was found in yield and yield contribu ng characters among the varie es. In both years Hunter (52.2 t/ha) and Winner (49.2 t/ha) had the highest fruit yield, while the lowest fruit yield (44.7 t/ha) was from Field Master. Key message: Hunter was the best performing watermelon variety at Khulna in terms of both yield and sweetness. Keywords: yield, brix, salinity 1. Introduc on Watermelon originates from the Middle East to the Mediterranean and belongs to the family Cucurbitaceae (Maynard et al. 2001). There is remarkable diversity of watermelons, each having unique flavor, texture and appearance (Norman 1992). According to FAOSTAT (2009), melon is grown on 1.3 million ha producing 26.7 million tons. China is the leading country in produc on of melon (51%), followed by Turkey (6%) and the United States of America and Spain (4%). Melons are consumed as dessert, fresh-cut fruit and juice (Sa ner and Lester 2009). The refreshing pulp, high nutri onal value, sweet pleasant aroma, bright color, firm fresh texture, high sugar content (>10%) and good shape are important traits that make melons unique and a refreshing treat in Japan (Gusmimi and Whener 2005; De Melo et al. 2000; Long 2005; Seko 2004) with high profitability (Best 2001). Watermelon is now one of the popular summer fruits and an important labor-intensive cash crop in Bangladesh. Generally, it is grown in Chi agong, Comilla, Jessore, Faridpur, Rajshahi, Pabna and Natore districts. The total produc on area is 12,000 ha and total produc on is 100,000 tons (BBS 2005). Recently, commercially cul va on has taken place in the coastal areas of Khulna. But the yield of watermelon in the southern belt is less than that of other areas in Bangladesh. Suitable varie es, soil, climate, fer liza on and other management prac ces are important for achieving poten al yield of any crop. Among them, variety selec on is one of the most important factors for obtaining higher yield. Suitable varie es are needed for the coastal region. Therefore, an experiment was undertaken to evaluate watermelon varie es in Khulna region. 2. Materials and methods The experiment was conducted at the BARI mul -loca on site at Dacope during the 2011-12 and 2012-13 rabi seasons. The experiment was laid out in a randomized complete block design with four replicates and a unit plot size of 6 m × 6 m. The treatments were four watermelon varie es viz. World Queen, Field Master, Hunter and Winner. The land was fer lized with 92-35-125-18-1-1 kg/ha of N-P-K-S-Zn-B in the forms of urea, triple superphosphate, muriate of potash, gypsum, zinc sulphate and boric acid. Half of the nitrogen and all the phosphorus, potassium, sulphur, zinc and boron were applied during final land prepara on. The remaining nitrogen was applied 30 d a er sowing as a side dressing about 20 to 30 cm from the base the plants. Seeds 516 were sown on 4 February 2012 and 8 February 2013 with a spacing of 2 m x 2 m. A 12 cm-thick straw mulch cover was applied immediately a er sowing. Irriga on, weeding and other intercultural opera ons were done as and when required. The crop was harvested from 3 April to 15 May 2012 and 12 April to 20 May 2013. Fruit yield and yield contribu ng characters were determined on ten randomly selected plants from each plot, and the data were analyzed sta s cally. Mean differences were calculated by Duncan’s Mul ple Range Test (Gomez and Gomez 1984). Soil salinity of the experimental plot and rainfall were recorded during the crop growing period. 2.1 Soil salinity Soil samples were collected and air dried, crushed and ground with a hammer and foreign ma ers and plant materials were removed. Then a soil solu on was made by adding one part soil with five parts dis lled water. The soil solu on was shaken well for 30 minutes and then allowed to se le for one minute. EC was determined using a HANNA Model Hl 8033 EC meter, which was calibrated with 0.01M KCI at 25o C (Peterson 2002). 3. Results and discussion 3.1 Soil salinity Soil salinity increased as the season progressed to maximum values of 8 to 10 dS/m (Fig. 1). 12 12 10 10 8 8 6 6 4 4 2 2 0 01-Feb-12 0 02-Mar-12 01-Apr-12 01-May-12 31-May-12 01-Feb-13 03-Mar-13 02-Apr-13 02-May-13 01-Jun-13 Fig. 1. Soil salinity during the watermelon season in 2012 (a) and 2013 (b). 3.2 Crop performance Length and diameter of fruit, number of fruits per plant, individual fruit weight and fruit yield differed significantly among the varie es. Days to maturity varied slightly and ranged from 60 to 62 d for World Queen to 64 to 68 d for Hunter. Adelberg et al. (1997) stated that some cul vars were early while others were intermediate or late due to gene c and varietal differences. Mortality ranged from 23.9% (World Queen) to 11.6% (Hunter). No significant differences were found in the number of fruit per plant. Field Master and World Queen had the longest fruit length (~22 cm) and highest fruit diameter (~18.5 cm). No significant varia on was observed in individual fruit weight. Hunter and Winner had the highest yield (49-52 t/ha) due to lower mortality. The findings of significant differences in fruit yield were consistent with the observa ons of other researchers (Izge et al. 2007; Sana et al. 2003; Tsegaye et al. 2007). 517 Table 1. Yield and yield a ributes of watermelon varie es at the MLT site, Dacope, Khulna during 2011-12 and 2012-13 (pooled data) Variety Days to maturity World Queen Field Master Hunter Winner LSD (0.05) CV (%) 60-62 60-63 64-68 60-64 - Mortality (%) 23.9 19.2 11.6 14.5 3.5 10.2 Fruit/plant (no.) 1.30 1.25 1.34 1.30 NS 3.80 Fruit length (cm) 31.3 32.3 28.6 30.8 1.7 2.7 Fruit diameter (cm) 18.5 18.4 17.2 18.2 0.2 0.6 Individual fruit wt. (kg) 3.87 3.69 3.68 3.70 0.24 NS Fruit yield (t/ha) 45.9 44.7 52.2 49.2 3.8 3.9 Brix (%) also varied between the watermelon varie es. Highest Brix (12.5%) was measured in Hunter and the lowest (6%) from Winner (Fig. 2). According to Hosoki et al. (1990), sugar content in melon can be categorized into three classes. Fruit with Brix below 10 can be classified as class 1; 10-12, class; and above 12, class 3. 14 12 10 8 6 4 2 0 12.5 10.8 9 6 World Field Hunter Winner queen master Fig. 2. Brix of the watermelon varie es. 4. Conclusions and recommenda ons Out of four watermelon varie es, Hunter had the longest field dura on (64-68 d), highest yield (52.2 t/ha), and highest Brix (12.5%). So Hunter may be recommended for fresh cut fruit and supplied to the local market due to its sweet taste as well as its high yield. Acknowledgements Thankful apprecia on is due to the authority of the Sponsored Public Goods Resources (SPGR) project for support in terms of funds and support of officers and SPGR staff. Thanks is extended to Dr. Jalal Uddin Sarker, Chief Scien fic Officer (former), On-Farm Research Division, Bangladesh Agricultural Research Ins tute, Joydebpur, Gazipur who gave us valuable sugges ons and instruc on. The authors are also grateful to all officers, Scien fic Assistant (SA) and field staff for their cordial coopera on and inspira on to complete the project successfully. 518 References Adelberg, J.W.; X.P Zhang and B.B. Rhodes. 1997. Micro Propaga on of Water melon (Citrullus lanatus). In Biotechnology in Agriculture and Forestry 39. HighTech and Micropropaga on V, Bajaj, Y.P.S. (Eds.). Springer Publica on, New Delhi, Pages. 60. Bangladesh Bureau of Sta s cs (BBS). 2005. Sta s cal yearbook of Bangladesh. Sta s cal division, Ministry of Planing, Dhaka 1000, Bangladesh. Best, K. 2001. Economic profitability of four muskmelon varie es grown under field condi ons. J. Agric. Sci. , 5: 21-25. De Melo, M.L.S.; N. Narain and P.S. Bora. 2000. Characteriza on of some nutri onal cons tuents of melon (Cucumis melo hybrid AF-522) seeds. Food Chem., 68:411-414 FAOSTAT.2009. Web page of the food and agriculture organiza on of the United Na ons. Sta s c Database. Produc on Sta s cs. Gomez, K.A. and A.A. Gomez.1984. Sta s cal procedures for Agricultural Research. 2nd Edn. John Willey and Sons, New York. 207-208. Gusmimi, G and T.C. Whener. 2005. Founda on of yield improvement in watermelon. 45: 141-146. Hosoki, T.; A. Ishibashi; H. Kitamura; N. Kai; M. Hamada and T.Chota,1990. Classifica on of oriental melons based on morphological, ecological and physiological differences. J. Jap. Soc. Hort. Sci., 58: 959-970. Izge, A.U. ; A.M. Kadams and A.A. Sajo, 2007. Agronomic performance of selected cul vars of pearl millet (Pennisetum glaucum L.R. Br.) and their hybrids in North-Eastern Nigeria. J. Agron., 6: 344-349. Long, R.L. 2005. Improving fruit soluble solid contents in melon (Cucumis melo L.) (Re culates group) in Australian produc on system. Ph.D. Thesis, Central Queensland University. Maynard, D.N.; A.M. Dunlap and B.J. Sidori. 2001. Cantaloupe variety evalua on. Scient. Hor cult., 8: 16-20. Norman, J.C. 1992. Tropical Vegetable Crops. Stockwell Ltd. New York, ISBN-13: 978-0722325957, Page:252. Peterson, L. 2002. Analy cal Methods Soil, Water, Plant Material, Fer lizer. Soil Resource Development Ins tute. Dhaka. 45-48. Sa ner, R.A. and G.E. Lester. 2009. Sensory and analy cal characteris cs of a novel hybrid muskmelon fruit intended for the fresh-cut industry. Postharvest Biol. Technol., 51: 327-333. Sana, M.; A. Ali; M.A. Malik; M.F. Saleem and M.Rafiq. 2003. Compara ve yield poten al and oil contents of different canola cul vars (Brassica napus L.). J.Agron., 2: 1-7. Seko, T. 2004. Characteris cs and quality of melon plant. Vegetable Hort. Melon, 2: 129-145. Tsegaye, E; D. Nigussie and E.V. Devakara Sastry. 2007. Gene c variability for yield and other agronomic trats in sweet potato. J.Agron., 6: 94-99. 519 520 Sec on 7 Outscaling 521 Targe ng improved cropping systems in the coastal zone of Bangladesh: A decision tree approach for mapping recommenda on domains P.K. Chandna1, A. Nelson1, M.Z.H. Khan2, M.M. Hossain3, M.S. Rana4, M. Mondal 1, S. Mohanty1, L. Humphrey1, F. Rashid5 and T.P. Tuong1 1 Interna onal Rice Research Ins tute, Bangladesh and Philippines p.k.chandna@irri.org, a.nelson@irri.org, m.mondal@irri.org, s.mohanty@irri.org, e.humphreys@irri.org, t.tuong@irri.org 2 Ins tute of Water Modeling, Bangladesh zhk@iwmbd.org 3 Soil Research Development Ins tute, Bangladesh moqbul_h@yahoo.com 4 Local Government Engineering Department, Bangladesh sohel_lged76@yahoo.com 5 Bangladesh Water Development Board, Bangladesh fazlur64@gmail.com Abstract Sa sfying the food demands of a rising popula on, conserving natural resources and improving livelihoods are major challenges in the food-insecure coastal zone of Bangladesh. A large area remains either uncul vated or underused following the harvest of the aman crop due to excess or scarce water, salinity, inequitable water management and insufficient water infrastructure. Increasing cropping intensity on these lands could substan ally improve the food supply and enhance livelihoods in the region. This paper describes a decision tree-based mapping methodology that iden fies the recommenda on domains of four exis ng and seven improved cropping systems for targe ng improved agronomic and water management prac ces in underu lized lands of Barisal Division. Firstly, an extensive literature review and expert group discussions, taking into account new findings from the CGIAR Challenge Program on Water and Food’s Ganges Basin Development Challenge, were conducted to iden fy land use requirements, their quan ta ve threshold values, and their logical integra on into the decision tree. Spa al datasets represen ng each requirement were developed and combined following the decision tree logic into cropping system suitability maps. The suitability maps were assessed through focus group discussions in four polders across the zone to determine the validity of the recommenda ons. The analysis iden fied about 0.5 million ha areas in six districts of Barisal Division where cropping intensity could be increased from 100% to up to as much as 300% using locally tested, improved cropping systems. The methodologies and technologies used in the study are applicable to most of the salinity-prone coastal zones of South and South East Asia. Keywords: water and soil salinity, spa al analysis, decision trees, cropping systems 1. Introduc on Increased food security and improved livelihoods in poor rural areas are major development goals for South Asian countries including Bangladesh, India, Nepal and Pakistan. Low cropping intensity is one factor that limits the achievement of both. In South Asia approximately 14 million ha (mha) of poten ally produc ve land remains fallow during the winter (rabi) season a er the monsoon (aman) rice crop (Subbarao et al. 2001; Padmanabhan 2008). This is due to four major causal factors: (1) lack of irriga on water crea ng in-season drought; (2) flash and stagnant floods causing waterlogging in low land areas (Chandna et al. 2010); (3) higher level of salinity in soil or water; and (4) late harvest of aman crop or excessively moist soils leading to late plan ng and lower produc vity of rabi crops or the decision to leave land fallow (Choudhary et al. 2008; Pandey et al. 2010; Chandna et al. 2012). In Bangladesh more than 2.65 mha are affected by flash and stagnant flooding, of which about 1.6 m ha are inundated almost every year (Mackill et al. 2006; Bailey-Serres et al. 2010; Mirza 2011; Mackill et al. 2012; Ismail et al. 2013). Around 1.1 m ha are affected by coastal salinity, which is exacerbated by sea level rise and 522 cyclones (SRDI 2010). In addi on, about 2.3 and 1.2 m ha of aman and rabi crops, respec vely, are exposed to drought every year (IOP, 2009; Dey et al. 2011). In short, there is a prevailing mul -stress environment for agriculture in Bangladesh, and more par cularly in the coastal polder zone (CPZ), where soil and water salinity has increased the impact of the detrimental effects of floods and droughts. The CPZ of Bangladesh, a fer le deltaic plain, is one of the most densely populated agricultural regions in the country. Rice (Oryza sa va) is a major staple food crop and supports about 8 million people in the CPZ, of which 80% live below the country’s poverty line (US$1.25/day/person) (Chandna et al. 2014). The main source of food security in the CPZ is a single, low yielding crop of rice grown during the rainy (aman rice) season. Resembling the other mul -stress prone areas of South Asia, a large area (Table 1) of poten ally produc ve land in Barisal Division remains uncul vated in the rabi and aus (pre-monsoon) seasons following the harvest of the aman rice crop, mainly due to the four major causal factors men oned above. Another major causal factor in the CPZ is deferred maintenance of water infrastructure (sluice gates, drainage canals, etc.) leading to drainage conges on and waterlogging. Stagnant flooding (0.3-0.5 m for more than two weeks) in low lying areas during the rainy season is another major causal factor (Ismail et al. 2013; Islam et al. 2014). It not only restricts the adop on of high yielding varie es (HYV) in the aman season, but also affects the plan ng of subsequent rabi crops, par cularly where drainage is poor (Chandna et al. 2012). Furthermore, high levels of salinity (> 4 ds/m) decrease the termina ve energy and germina on rate of plants and reduce produc vity (Rashid et al. 2004). For example, Ali (2005) showed that a decline of rice produc vity of about 69% in a village of Satkhira District between 1985 and 2003 was associated with salinity. In spite of its increased vulnerability to various agricultural stresses and natural disasters (Milimen et al. 1989; World Bank 2009; Brammer 2010; Islam et al. 2012; Mirza 2013; UNDP 2013) the CPZ is also known as a "region of opportuni es and high poten al" due to: its fer le alluvial soil; availability of freshwater – as yet largely untapped –in the Padma, Brahmaputra and Megna River systems for most part of the year; a dense irriga on canal network (khals); the availability of short dura on improved cul vars; and improved knowledge of agronomy and water management aspects. In addi on to these factors, semi-diurnal dal movement (low and high de, twice a day) in the CPZ provides an opportunity for low-cost gravity drainage and irriga on. At low de the river level is generally lower than the land level within the polders, crea ng opportuni es for gravity drainage of excess water (as a result of heavy rainfall) to a level that would allow for good growth and yield of HYV aman rice. Drainage of the earlier maturing HYV shortly prior to harvest would allow the soil to dry sufficiently for mely establishment of rabi crops. Achieving good water management within a polder requires an integrated approach of improved agronomy and water and infrastructure management (Golder et al. 2013). Recent studies by the CGIAR Challenge Program on Water and Food (Mondal et al. these proceedings) have shown that it is possible to have three crops per year (triple rice or double rice plus one upland crop) in “low salinity” zones (e.g. Barisal Division); and two crops per year (double rice or one rice plus upland crop) in “moderate salinity” zones (e.g. Khulna Division). Even in the “high salinity” zones (e.g. Satkhira Division), farmers can prac ce shrimp-rice systems by cul va ng shrimp in the dry season followed by rice in the wet season (Nuruzzamam et al. 2007; Quader et al. 2010). These innova ve cropping systems can substan ally increase agricultural produc vity and improve farmers’ livelihoods. However, any single improved prac ce may not be suitable for all situa ons, sugges ng that technologies should be targeted to their most appropriate niche (Graaff et al. 2008; Chandna et al. 2012). Assessing target zones for promising technologies facilitates out-scaling in a fast and cost-effec ve manner (Chandna et al. 2011; Erenstein et al. 2010). 523 Previous studies have u lized crop-based characteriza on and targe ng methods. In such methods, the land use requirements of individual crops are assessed (eg. targe ng of high yielding varie es) for a par cular me of the year, mainly based on soil and clima c parameters (Hodson and White 2007; Behzad et al. 2009; Mamun et al. 2011; Halder 2013; Samanta et al. 2013; ESA 2014). This method may not be adequate for assessing cropping system suitability in mul -stress, lowland, coastal environments where resource profiles (e.g. salinity, submergence, inunda on depth and water availability) vary both temporally and spa ally. We have observed that most of the cri cal biological parameters (eg. water salinity, gravity drainage) and seasonal aspects ( me, season and intensity) of CPZ environments have not been fully accounted for in many past studies on crop-based characteriza on (Hossain et al. 2007; Jafari et al. 2010; Mamun et al. 2011; Halder 2013;). Therefore, we need to consider a method to generate ‘recommenda on domains’ for CPZ cropping system suitability (Getnet and MacAlister 2012; Bussel et al. 2013) that accounts for the spa al and temporal complexity of cropping system parameters for coastal saline zones. Recommenda on domain analysis (RDA) is a methodology for iden fying geographical areas that are suitable for the adop on of improved cropping system prac ces on the basis of well-defined criteria that account for each crop as part of the cropping system. In other words the criteria for a triple cropping system is not simply the suitable area for crop A overlaid on the suitable area for crop B and then crop C, but rather the criteria for realis cally adop ng the full cropping system accoun ng for cri cal ming of plan ng and harves ng, or residual soil moisture, for example, such that a second or third crop can be realis cally and sustainably included in the system. This paper presents RDA, as applied to the Coastal Zone of Bangladesh, for developing cropping system suitability maps for current and improved cropping systems in low, medium and high salinity zones in the Barisal region. It describes the characteris cs of the region, the cropping systems under considera on, the parameters required for those cropping systems and the decision trees for suitability mapping. The resul ng maps are presented and discussed alongside the outcomes of focus group discussions in the CPZ on the feasibility of the cropping systems assessed. 2. Material and Methods 2.1 Study Area The study area, Barisal Division, is situated in the southern part of Bangladesh, between 21°49’41” to 23°05’36” N and 89°53’18” to 91°01’ E (Figure 1). It occupies an area of 13,297km2 with a popula on of over 8 million. The total number of farm holdings in Barisal Division is about 0.5 million, out of which 85% belong to small farmers (BBS 2011). The dominant soils are loamy, sandy loam, clay loam, and clay (SRDI 2010). Mean annual rainfall in the study area is 1955 mm and maximum and minimum annual average temperature is 12.1°C and 35.1°C, respec vely. 524 N Bangladesh A Dhaka Coastal Polder Zone Barisal Division Khulna Division Bangladesh 30 0 30 Kilometers Fig. 1. Map A (le ) depicts the loca on of the Coastal Polder Zone and Barisal Division within Bangladesh. Map B (right) shows the districts within Barisal Division. The monsoon or rainy season (June to October) is characterized by high rainfall, high humidity, and high cloud cover. Sediment load and water levels of the area also increase during this period. The salinity in soil and water is moderately high along the coast and gradually reduces northward due to upstream freshwater flow. Occasional thunderstorms, cyclones and storm surges occur during the monsoon season. The winter season (November to February), with a northeast monsoon wind, is characterized by dry, cool, and sunny weather with occasional rain. The lowest temperatures occur in December and January and the highest during the months of May and June. Sunshine hours are lowest during the rainy and winter seasons and highest in the summer season. The hydrology of coastal Bangladesh is different to many coastal delta regions of South and South East Asia because of its ‘polders’ that control saline water intrusion. A polder is a low-lying tract area, enclosed by embankments (levees) that forms an ar ficial, separate hydrological en ty. A polder has no open connec on with sea or upstream river water, except that water is allowed to enter or drain through manually operated sluice gates built at riverbanks or irriga on canals (khals). These sluice gates help maintain the flow of fresh or saline water through gravity irriga on or drainage during low and high des. Salinity concentra on (in both soil and water) decreases in the inland direc on and increases in the seaward direc on of the CPZ. During the wet season, direct rainfall and an increased upstream flow have a dilu on effect on salinity concentra on. In Barisal, surface water in upstream rivers remains fresh for most of the year (Zahir et al. these proceedings), leading to an opportunity to increase cropping intensity in single cropped areas (Table 1). The most dominant cropping systems in the Barisal Region are aman-fallow-fallow, T. Aman-T. Aus, T. Aman-rabi and T. Aman-T. boro. The aman crop covers over 0.90 m ha, or 65% of the net cul vated area (NCA), whereas T. Boro and T. Aus season crops are cul vated on 0.23 and 0.37 m ha (27% of the NCA), respec vely (BBS 2011). Pulses and oilseeds (which are grown during the winter season) are the dominant rabi crops, extending over 0.3 m ha (22% of the NCA). Other rabi crops (potato, sugarcane, jute, wheat and maize) account for 0.05 m ha (4% of the NCA). More than 50% of the area in Barisal Region remains fallow a er a low yielding aman crop. The opportunity to increase produc vity through increased cropping intensity is much higher in central and southern parts of Barisal Region (Table 1). 525 Table 1. An overview of land use in Barisal Division and Bangladesh - 2010 Region Bangladesh (All districts) Barisal (Northern district of Barisal Division) Patuakhali (Southern district of Barisal Division) Cropped Area (000 ha) Single Double 2577 (33)* 3934 (50) 202 (44) 194 (42) 184 (61) 88 (29) Triple 1282 (16) 67 (14) 28 (9) Note: Figures in parentheses indicate the percentage of total cropped area Source: BBS, 2011, Yearbook of Agricultural Sta s cs of Bangladesh 2.2 Methodology We first iden fied the exis ng and improved cropping systems to be mapped as well as their requirements through literature reviews and roundtable discussions with experts on agronomy, fisheries, soil and water. These requirements were quan fied and related to the best available spa al data such that there was one spa al data layer for every unique requirement. A decision tree approach was chosen to combine these spa al data layers into suitability maps that delineated the recommenda on domains of the cropping systems. These maps were assessed through expert opinion and through focus group discussions in the CPZ. 2.2.1 Current and Improved cropping systems in the CPZ The descrip on, environment and other details of these cropping systems are given in Tables 2 and 3. These descrip ons or ‘narra ves’ are the basis for the development of spa al databases of thema c layers to characterize the requirements of each cropping system, and for the decision trees that relate the combina on of requirements in a given loca on to a specific suitability score. Table 2. Descrip on and environment of exis ng cropping systems Extensive year round brackish water polyculture Shrimp post larvae (PL) are stocked in February, when brackish water has adequate salinity and suitable temperature; brackish water fish/shrimp are stocked few weeks later. Water is replenished as needed. Harvest starts a er two months un l the end of November. Brackish water shrimp – Aman Rice (tradi onal) Shrimp PL are stocked in February when brackish water has adequate salinity and suitable temperature. Phased stocking and harvests are carried out throughout the dry season, un l the end of July. Soil salinity is flushed/leached out of the ghers by rainfall and intensive drainage at low de. Tradi onal, photoperiod sensi ve rice is seeded in July, transplanted at the end of August and harvested in December and January. T. Aus high yielding varie es (HYV) - T. Aman (tradi onal) Aus HYV is seeded in late April and harvested by the end of August. T. Aman rice is transplanted by the first week of September to be harvested by the end of December. Aus crop seedbeds and land prepara on are irrigated with river water when it is fresh, or with groundwater or pond water. (Farmers make use of rainfall to reduce pumping costs, but T. Aus can be prac ced only when there is a good source of irriga on water.) Aman rice is completely rain-fed, and local varie es that can withstand flood depths of up to 0.8 m are used. T. Aman (Tradi onal) - Rabi Tradi onal, photoperiod sensi ve Aman rice is seeded in June, transplanted in August to be harvested by end of first week of January. Rabi upland crop (e.g. sesame, and mungbean) is seeded in Feb-March, to be harvested in May-June. Aman rice is completely rainfed, local varie es that can withstand flood depth up to 0.8 m. Rabi crop is grown with residual soil moisture and supplemented with irriga on when groundwater or (stored) canal/pond water is available. Rainfall at the end of the Rabi season (April - May) nega vely affects the Rabi crop, good surface drainage is important 526 A total of 11 cropping systems were selected for developing recommenda on domains in Barisal. Four of the cropping systems (Table 2) belong to exis ng prac ces and seven cropping systems (Table 3) are ‘improved packages’ (Mondal et al. these proceedings) with improved cul vars, and agronomic and water management prac ces that have been successfully tested on farmer par cipatory sites at polders 3, 30 and 43/2F (which represent high, medium and low salinity zones, respec vely). Table 3. Descrip on and environment of improved cropping systems T.Aman HYV – T.Boro HYV Short dura on, non-photoperiod sensi ve Aman HYV rice is transplanted in July to August, to be harvested by the end of November (moderate salinity zone) or December (low salinty zone). Boro rice is seeded around 15 November to 15 December. Aman rice is rainfed. Its performance depends greatly on maximum inunda on depth/land topography. When inunda on depth >0.4 m drainage capacity must be considered. This in turn depends on the difference between land eleva on and the water levels at low des and on the distance from canals/rivers. Boro rice is irrigated with groundwater (GW) or river water (when fresh) or with water stored in canal networks. When river water is fresh, the ability to irrigate by gravity is considered. Storage capacity is considered when stored water is used. Distance from water source is considered in both cases. T. Aus HYV - T. Aman HYV Aus HYV is seeded in late April, to be harvested by the end of August. HYV Aman rice is transplanted by the first week of September and harvested by December. Aman rice can be photoperiod sensi ve, and can be of medium dura on. Aus seedbeds and land prepara on are irrigated with GW or from stored canal water when there is good source of irriga on water. In this case, capacity of gravity irriga on is considered. Aman HYV rice is mainly rainfed but may need some irriga on in December. Its performance depends greatly on maximum inunda on depth/land topography. When inunda on depth >0.4 m drainage capacity must be considered. This in turn depends on the difference between land eleva on and the water levels at low des and distance from canals/rivers. T. Aus HYV - T. Aman HYV - T. Boro HYV A short dura on Aus HYV is seeded in early April, transplanted in late April and harvested by the end of July. Short dura on, non-photoperiod sensi ve HYV Aman rice is transplanted by the first week of August and harvested by the end of November. Boro rice is transplanted before mid-December and harvested by mid-April. Aus seedbeds and land prepara on are irrigated with GW or from canal water when there is a good source of irriga on water. In this case, capacity of gravity irriga on is considered. Aman HYV rice is completely rainfed. Its performance depends greatly on maximum inunda on depth/land topography. When inunda on depth >0.4 m drainage capacity must be considered. This in turn depends on the difference between land eleva on and the water levels at low des and distance from canals/rivers. Boro rice is irrigated with GW or river water (when fresh) or with water stored in canal networks. When river water is fresh, the ability to irrigate by gravity is considered. Storage capacity is considered when stored water is used. Distance from water source is considered in both cases. T. Aman HYV - Rabi (winter crop) HYV Aman rice is transplanted by the first week of August and harvested by the end of November to early December. Rabi upland crop is seeded during December and harvest in early April. Aman rice can be photoperiod sensi ve and can be of medium dura on. Aman HYV rice is completely rainfed. Its performance depends greatly on maximum inunda on depth/land topography. When inunda on depth >0.4 m drainage capacity must be considered. This in turn depends on the difference between land eleva on and the water levels at low de and distance from canals/rivers. For mely establishment of rabi crop, land must be drained free of standing water by early December. Since we may not have adequate data on land evalua on the "proxies" for drainability will be the maximum inunda on depth and proximity to canals/rivers. Rabi crop is irrigated with GW or river water (when fresh) or with water stored in canal networks. When river water is fresh the ability to irrigate by gravity is considered. Storage capacity is considered when stored water is used. Distance from water source is considered in both cases. 527 T Aus HYV - T. Aman HYV- Rabi A short dura on Aus HYV is seeded in early April, transplanted in late April and harvested by the end of July. Short dura on, non-photoperiod sensi ve HYV Aman rice is transplanted by the first week of August and harvested by the end of November. Rabi upland crop is seeded in early December and harvested in early April. Aus seedbeds and land prepara on are irrigated with GW or from canal water when there is a good source of irriga on water. In this case, capacity of gravity irriga on is considered. Aman HYV rice is completely rainfed. Its performance depends greatly on maximum inunda on depth/land topography. When inunda on depth >0.4 m drainage capacity must be considered. This in turn depends on the difference between land eleva on and the water levels at low des and distance from canals/rivers. For mely establishment of rabi crop land must be drained free of standing water by early December. Since we may not have adequate data on land evalua on, the "proxies" for drainability will be the maximum inunda on depth and proximity to canals/rivers. Rabi crop is irrigated with GW or river water (when fresh) or with water stored in canal networks. When river water is fresh the ability to irrigate by gravity is considered. Storage capacity is considered when stored water is used. Distance from water source is considered in both cases. Brackish water shrimp - HYV Aman Rice Shrimp PL is stocked in February when brackish water has adequate salinity and suitable temperature. Water is replenished as needed. Phased stocking and harvests are carried out throughout the dry season, un l the end of July. Soil salinity is leached/flushed out of the ghers by rainfall and intensive drainage at low de. HYV rice is seeded in August, transplanted at the end of August or beginning of September and harvested in December. When inunda on depth >0.4 m, drainage capacity must be considered. This in turn depends on the difference between land eleva on and the water levels at low des and distance from canals/rivers. 2.2.2 Parameters that define the cropping system requirements The requirements of each crop and cropping system are defined by one or more parameters. The 29 parameters are listed below (Table 4), grouped into seven factor groups. Some parameters are es mated on weekly or monthly me steps (i.e. latest month when river water conduc vity (EC) <3 dS/m, minimum salinity at PL stocking, weekly air minimum temperature at stocking). Thus the number of data layers required to assess the suitability across all systems is much larger than 29. Some examples of the spa al data sets are shown in Figure 2. 528 Table 4. Parameters in the spa al database used for cropping system suitability analysis Factor (f) Parameter Irriga on with GW 1 Fresh (EC < 4dS/m) groundwater availability 2 Ground water pumping depth Irriga on with 3 Latest month when river water EC <3 dS/m surface water (SW) 4 Maximum river water EC/salinity in April/May 5 Maximum river water EC/salinity in August 6 Difference in high water level in April and land surface for gravity irriga on 7 Difference in high water level in March and land surface for gravity irriga on 8 Storage capacity 9 Proximity to river, canal, ponds for irriga on Drainability 10 Proximity to river, canal for drainage 11 Maximum inunda on depth/land type 12 Maximum inunda on depth for > three consecu ve days in May 13 Maximum inunda on depth in September/October 14 Maximum inunda on depth for > one week in September/October 15 Maximum inunda on depth for > two consecu ve weeks in September/October 16 Difference in land surface and low water level in May for drainage 17 Difference in land surface and low water level in September/October for drainage Soil 18 Soil texture 19 Soil pH 20 Soil salinity Field or gher water 21 Minimum salinity/EC at PL stocking 22 Maximum salinity/EC 23 Inunda on depth/land type Climate 24 Weekly air minimum temperature at stocking 25 Weekly mean minimum air temperature in January 26 Two-week mean air temperature in December and January 27 Cumula ve rainfall in July and August Social and 28 Livelihood/asset index economic 29 Technology adop on index N Mean Minimum Temperature (° C) in 2ⁿ week of February Soil EC (ds/m) 0 -2 2 -4 4 -6 6 -8 8 - 12 12 - 15 > 15 5 - 10 10 - 15 15 - 20 20 - 25 25 - 30 30 0 30 60 Kilometers Fig. 2. Example layers; minimum temperature in February (le ) and soil EC values (right). 529 2.2.3 Decision trees, weigh ng and spa al analysis of suitability classes Using a decision tree approach (Park and Lee, 2014) in ArcGIS 10.1 so ware, we developed a logical model for determining suitability classes based on 'IF', 'IF NOT' and ‘AND’ statements. Separate decision rules were prepared for each cropping system as per their land use requirement. The method and process followed to integrate the land use parameters with decision rules for prepara on of recommenda on maps is given below. 1. Each cropping system is composed of a sequence of crops, where this is one crop per cropping season and one or more cropping seasons per year. 2. The suitability of any given loca on to cul vate a crop is assessed by a number (n) of factors (f, where f=1 to n) that are determined by expert opinion and literature. 3. Each factor may be quan fied by one parameter (e.g. if soil salinity is a factor, it depends solely on a measurement of soil salinity) or by more than one parameter (e.g. if “drainability” is a factor, it depends on (i) topography, (ii) the difference between land surface and water level at low des, and (iii) proximity to canals/rivers). These quan fiable parameters and their values are determined by expert opinion. 4. Since each factor is quan fiable, we relied on expert opinion and literature to classify each factor into ranges and assign each range a suitability score(s). To reduce complexity we limited the suitability score to four values: most suitable, s=3; suitable, s=2; marginally suitable, s=1; and not suitable (SN), s=0 (Figure 3). For example, if soil pH is factor 1 (s1) for a T. Aus HYV crop then expert opinion may group soil pH into four classes: most suitable (score s1=3 if 5.5 < pH < 8.5), suitable (score s1=2 if pH is in between 5.0 and 5.4 or between 8.6 and 9.0), marginally suitable (score s1 = 1 if pH in between 4.0 and 4.9 or between 9.1 and 10.0) and not suitable (score s1 = 0 if pH > 10.0 or pH < 4.0). An example of the scoring per layer is shown in Figure 3. N Inunda on Depth Drainability Not Suitable (SN) Marginally Suitable (S1) Suitable (S2) Most Suitable (S3) Not Suitable (SN) Marginally Suitable (S1) Suitable (S2) Most Suitable (S3) 10 0 10 20 Kilometers Fig. 3. Illustra ng the examples of inunda on depth (le ) and soil drainability (right) a er quan ta ve ranges are assigned with a suitability score for each individual layer. 5. Factors are not necessarily of equal importance in determining the suitability of the crop. Expert opinion was used to further refine the importance of each suitability score by mul plying it by a weight (0 ≤ w ≤ 1), resul ng in a weighted suitability score for each factor (w1.s1 ,w2.s2 ,… wn.sn). Since the maximum weight is 1 and the maximum suitability score is 3, then the maximum weighted suitability score for a factor is 1 x 3 = 3. Conversely the lowest weighted suitability score for a factor is 0 for any factor with a suitability score of 0 regardless of its assigned weight. 530 6. The suitability class for the crop (Sc) is defined as follows: S1= marginally suitable, S2=suitable, S3= most suitable and SN=not suitable. This Sc class is derived by first assigning a suitability score (Ss) to the crop as the sum of the weighted suitability scores of all factors divided by the sum of the weights of all factors: f n wf s f f 1 f n Ss   wf f 1 and then assigning a suitability class to the suitability score as follows: Sc = S1 (marginally suitable) if 0 < Ss < 1.5 Sc = S2 (suitable) if 1.5 ≤ Ss < 2.5 Sc = S3 (most suitable) if Ss ≥ 2.5 Sc = SN (not suitable) if any one weighted score (wfsf) is 0 7. The suitability class for the cropping system (Scs) is derived by compu ng the average of the suitability scores (Ss) from the component crops in that cropping system and then applying the same S1, S2, S3 and SN classifica on. For example, at a given loca on individual crops of the triple rice system Aus HYV - Aman HYV - Boro HYV may have suitability scores Ss of 3.0, 2.4 and 2.0. The average of these is 2.47 so the cropping system suitability score Scs is S2. The result is a map for each crop system with suitability classes S1, S2, S3 or SN. An example decision tree for the water requirements of a boro (dry) season rice crop are shown in Figure 4 where the tree first considers the availability and depth of fresh groundwater. If no groundwater is available the lower part of the tree considers the dura on of fresh surface water during the season and the availability and access to stored fresh surface water. Ground water (GW) availability/criteria Fresh groundwater (< 4dS/m) Shallow or deep tube well ? Yes Shallow Deep No Depth of preha c surface (m) Surface water (SW) availability/crite ria Month when SW salinity is s ll < 3 dS/m Storage capacity for SW (ML/ha) Proximit y to fresh SW source (m) Suitability for a dry season rice crop S3 <6 S3 7 - 20 S2 > 20 S1 S3 Mar Feb 2.5 - 5 1- 2.5 Jan 2.5 - 5 1- 2.5 < 100 S3 > 100 S2 < 100 S2 > 100 S1 < 100 S2 > 100 S1 < 100 S1 > 100 SN Dec SN S1 = Marginally s uitable; Sr = Suitable; S3 = Most suitable and SN = Not suitable Fig. 4. Decision tree for water availability parameters for boro rice. 531 2.2.4 Focus Group Discussions An independent evaluator conducted focus group discussions (FGD) between September and November 2014 in four polders P-39/2, P-43/2F, P-44, and P-55/1 in order to evaluate and validate the innova ve cropping systems maps for the CPZ. FGDs were conducted at randomly selected loca ons within each of the selected polders. The independent evaluator ini ally made a reconnaissance visit to familiarize himself with the present situa on of the polders including water management infrastructure, cropping systems and agricultural prac ces. A simple three-page survey was prepared in order to conduct a FGD in one loca on in each polder. About 20 polder-level maps (about four to six maps of different cropping systems for each selected polder) of different improved cropping systems were provided to the evaluator to elicit comments from the group. The FGDs included more than 60 farmers (including 14 females) in four polders where the domain map method was explained and the resul ng maps were presented as a means to elicit responses and opinions on the feasibility of each improved cropping system domain as shown in the maps. The discussion first captured the current cropping pa erns and current limita ons and challenges faced by the farmers. Then, using the printed maps to focus the discussion, the feasibility/acceptance of the improved systems was assessed and the farmers discussed how current limita ons would also limit the feasibility of the improved systems and what needed to be done in the polder to improve current and future cropping system produc vity. 3. Results and discussion 3.1 Recommenda on domains for exis ng systems The suitable areas for exis ng cropping system are shown in Table 5. The maps of the suitability classes for exis ng cropping systems are shown in Figure 5. The results for each system are discussed in turn below. Table 5. District wise area (1000 ha) of the recommenda on domains for exis ng cropping systems in Barisal Division Cropping system Transplanted Aus-Aman Sc S1 S2 Transplanted Aman-Rabi S1 S2 Brakish Water Shrimp - T Aman S1 S2 Year Round Aquaculture S1 S2 Barguna 73 34 108 * 5 47 * 46 Barisal 93 45 135 3 1 3 * 4 Bhola 76 1 77 * 66 5 47 32 Jhalokha 45 15 57 2 * * * * Patuakhali 97 67 165 * 15 37 6 55 Pirojpur 40 11 49 2 * 10 * 8 Total 424 174 591 8 88 102 54 144 Note: * = No suitable area; Sc = ‘Suitability Class’, where S1 represent marginal suitable; S2 – Suitable and S3 represent most suitable area; Areas less than 1000 ha are not shown in the table 532 A. T.Aus-Aman (tradi onal) B. T. Aman-Rabi C. Shrimp (brackish) - Aman D. Year Round Aquaculture Not Suitable Marginally Suitable Suitable Most Suitable Fig. 5. Recommenda on domain maps for the four current cropping systems in Barisal. 3.1.1 T.Aus (HYV) – T.Aman (tradi onal) system The results of the RDA revealed that about 0.60 m ha (44%) of the NCA is suitable for T.Aus-Aman system cul va on (Figure 5A), of which 31% and 13% of the area was found under moderately suitable (S1) and suitable (S2) categories, respec vely. Barisal, Patuakhali and Berguna districts were primarily found suitable for a T.Aus-Aman system. In terms of individual crop suitability, about 0.7 m ha (51% of the NCA) was found suitable for a tradi onal aman crop under the marginally suitable category. It is es mated that farmers are growing tradi onal aman on about 0.89 m ha—about 20% more than our suitable area figure. In the recommenda on domain analysis, we excluded areas where the inunda on depth is higher than 0.8 m, whereas in reality farmers grow tradi onal aman crop varie es in fields where flood depth is as high as 1.0 m or more. This is one major 533 reason (along with drainage conges on) why a large area of tradi onal aman crop is damaged by stagnant floods almost every year during the rainy season. RDA analysis also revealed that about 0.54 m ha (39% of the NCA) is individually suitable for T.Aus, of which about 30% and 9% were found under the S1 and S2 categories, respec vely. 3.1.2 Tradi onal Transplanted Aman - Rabi (T.Aman-Rabi) About 0.60 m ha (44% of the NCA) is suitable for a T. Aman - Rabi crop system, of which 43% and 1% were found under S1 and S2 categories, respec vely, mainly in Patukhali, Barisal, Barguna and Bhola districts (Figure 5B). This suitable area is almost double the area (around 0.35 m ha) presently under T. Aman – Rabi cropping systems. 3.1.3 Extensive year-round brackish water polyculture system In the southern part of Barisal Division, water becomes saline during the dry season between February and June. Saline water (par cularly with EC > 6 ds/m) plays a pivotal role in brackish water shrimp aquaculture and makes it highly favorable for bagda shrimp farming. About 0.20 m ha (14%) of the net cul vable area (1.38 m ha) is suitable for year round brackish water aquaculture, of which 4% and 10% of the area was found under marginally suitable (S1) and suitable (S2) categories, respec vely, in the southern districts of Barisal Division. Barisal, Pirojpur and Jhaloka districts are mostly unsuitable for extensive year-round brackish water polyculture systems due to low water salinity in adjacent river systems, as brackish water is a major prerequisite for brackish water shrimp (Fig. 5C). 3.1.4 Brackish water shrimp - Tradi onal Aman Rice The results for extensive a year-round brackish water polyculture system and brackish water shrimp Tradi onal Aman rice system are the same due to their analogous land use requirements. About 0.20 mha (14%) of the NCA is suitable for a brackish water shrimp - Tradi onal Aman rice system, of which 4% and 10% of the area was under marginally suitable (S1) and suitable (S2) categories, respec vely, in the southern districts of Barisal Division. About 90% of the total suitable area for Brackish water shrimp - Tradi onal Aman rice is confined to the southern parts of Bhola, Berguna and Patuakhali districts due to the favorable environment for both shrimp culture and tradi onal aman rice crop (Figure 5D). 3.2 Recommenda on domains - improved cropping systems The suitable areas for improved cropping systems are shown in Table 6. The maps of the suitable areas for improved cropping systems are shown in Figure 6. The results for each system are discussed in turn below. Table 6. District wise area (1000 ha) under recommenda on domain for improved cropping systems in Barisal division Cropping system Shrimp - T.Aman (HYV) Sc S1 S2 T.Aman (HYV) - T.Boro (HYV) S1 T.Aus (HYV) - T.Aman (HYV) S1 T.Aman (HYV) – Rabi (Long) S1 T.Aus (HYV) - T.Aman (HYV) - T.Boro (HYV) S1 T.Aus (HYV) - T.Aman (HYV) - Rabi S1 Year Round Aquaculture S1 S2 Barguna 5 43 80 79 80 79 79 * 46 Barisal 1 3 108 112 108 106 106 * 4 Bhola 66 5 70 68 70 71 71 47 32 Jhalokha * * 44 43 44 43 43 * * Patuakhali 10 32 132 128 132 130 130 6 55 Pirojpur * 8 36 37 36 36 36 * 8 Total 82 92 468 467 468 465 465 54 144 Note: * = No suitable area; Sc = ‘Suitability Class’, where S1 represent marginal suitable; S2 – Suitable and S3 represent most suitable area. Areas less than 1000 ha are not shown in the table. 534 A. T.Aus (HYV)T.Aman(HYV) B. T. Aman(HYV)T.Boro(HYV) C. T.Aman(HYV)Rabi (Long) D. T.Aus(HYV)-T.Aman (HYV)-T.Boro(HYV) E. T.Aus(HYV)T.Aman(HYV)-Rabi F. ShrimpT.Aman(HYV) Not Suitable Marginally Suitable Suitable Most Suitable Fig. 6. Recommenda on domain maps for improved cropping systems in Barisal. 3.2.1 T.Aus (HYV) – T.Aman (HYV) Although the T.Aus -T.Aman system is the third major cropping system of Barisal Division a er Aman-Fallow-Fallow and T.Aman-T.Boro, nearly 83% and 92% land area is under tradi onal low yielding aus and aman crops, respec vely, in comparison to the na onal averages of 48% and 59% (BBS, 2011), respec vely. Nearly 0.54 m ha (under S1 category) were found suitable for cul va on of Aus (HYV) - Aman (HYV) in Barisal Division (Table 6, Figure 6A). This indicates that there is great poten al for conver ng a large area under tradi onal low yielding T.Aus – T.Aman to HYV. 3.2.2 T.Aman HYV - T.Boro (HYV) T.Aman (HYV) - T.Boro (HYV) is the most popular cropping system in Barisal region. The yield of T.Aman (HYV) - T.Boro (HYV) is generally 30-40% higher than T.Aus (HYV)-T. Aman (HYV) systems. A total of 0.90 and 0.23 m 535 ha area were under aman and boro rice (;ocal and HYVs together) in Barisal division. Out of this total, only 7% and 74% of the area was recorded under HYVs in aman and boro seasons, respec vely. Barisal and Bhola Districts contributed about 80% of the total area under boro rice (BBS 2011). Nearly 0.47 m ha area (34% of the NCA) is suitable for cul va on of T.Aman (HYV) - T.Boro (HYV), categorized under the marginally suitable class (Figure 6B). T.Boro (HYV) has huge poten al in Patuakhali, Barisal, Bhola, and Berguna Districts as an individual crop as well as at cropping system level. Currently, very li le area is under boro rice in Patuakhali and Berguna Districts, while the RDA revealed that 0.22 m ha of single crop area in these two districts can be converted into double crop through introduc on of improved crop and water management prac ces in the boro rice season. 3.2.3 Aus (HYV) - T.Aman (HYV) – T.Boro (HYV) Nearly 0.47 m ha (34% of the NCA) is under the marginally suitable (S1) category for cul va on of T.Aman (HYV)- T.Boro (HYV). About two-thirds of the total suitable area is found in Patuakhali, Barisal and Berguna Districts (Figure 6D). 3.2.4 Aus (HYV) - T.Aman (HYV) - Rabi The map of recommenda on domains for T.Aus (HYV) - T.Aman (HYV) - rabi (Figure 6E) is the same as T.Aus (HYV) - T.Aman (HYV) - T. Boro (HYV) (Figure 6D) because of similar land use requirements. Weekly mean minimum temperature during the early stages (December and January) of boro rice is the only addi onal criteria for a T. Boro (HYV) crop in comparison to rabi (long). Insignificant spa al varia on in temperature (< 1.5˚C) within the Barisal region during these months did not reveal any major difference between these two systems in terms of our RDA. The same applies to T. Aman (HYV) – Boro (HYV) and T.Aman (HYV)– rabi (Long) systems (Figures 6B and 6C, respec vely). 3.2.5 Shrimp -T.Aman(HYV) Nearly 0.17 m ha (7% of the NCA) is suitable for the cul va on of shrimp-T.Aman (HYV) in Barisal Division (Figure 6F). About 0.08 and 0.09 m ha were es mated to be under the marginally suitable (S1) and suitable (S2) categories, respec vely. About 90% of the total suitable area for a shrimp-aman (HYV) system is confined to the southern parts of Bhola and some parts of Berguna and Patuakhali Districts due to favorable environments for shrimp in the dry season (saline water) and tradi onal aman rice (freshwater) in the wet season. 3.2.6 Improved yea- round brackish water polyculture system About 0.20 m ha (14% of the NCA) is suitable for imrpvoed yearround brackish water aquaculture, of which 4% and 10% of the area was found under marginally suitable (S1) and suitable (S2) categories, respec vely, in the southern districts of Barisal Division. Areas of Barisal, Pirojpur and Jhaloka Districts are mostly unsuitable for improved year-round brackish water polyculture systems due to lower levels of water salinity in adjacent river systems (same as Figure 5C). 3.3 Farmer group discussions and valida on of recommenda on maps The results of the FGD in four polders in Barisal division are shown in Table 7. The FGDs revealed that most of the improved cropping systems are acceptable to farmers in low saline zone polders, except the shrimp-aman (HYV) system. In well managed polders (eg. P-43/2F), IS5 to IS9 are mostly acceptable. IS5, IS7 and IS9 are more acceptable in P-44 and P-55/1 and IS5, IS6 & IS8 in polder 39/2D. In general, IS7 and IS9 are rela vely less acceptable due to issues related to water availability. Note that FDG results in medium and high salinity polders are not shown in this short conference paper. 536 Some general observa ons were also drawn from the FGDS. Firstly, that all improved cropping systems will be acceptable to the farmers in the low salinity zones, except IS11. Secondly, the recent trend of producing cost effec ve rabi crops need to be taken into considera on. Thirdly, the adop on of improved systems in many areas requires: regular maintenance of water management infrastructure; upgrading of drainage and flushing capacity by adding draining/flushing infrastructure (sluice/inlet/canals); and ensuring availability of good quality agricultural inputs and modern equipment. Finally, demonstra ons of the systems at polder or sub-polder level will expedite acceptance and adop on. Table 7. FGD polder characteris cs and summarized responses on area deemed suitable per improved cropping system per polder Polder characteris cs Polder People name Area Cul va M F Salinity (000) ha table (%) Suitable area (%) per improved cropping system (targeted salinity zone in brackets) IS-6 IS-7 IS-8 IS-5 IS-9 IS-10 IS-11 (L)* (L) (L) (L/M) (M) (H) (H) 39/2D 43/2F 44 55/1 15 59 9 7 18 13 5 3 3 3 Low (L) Low (L) Low (L) Low (L) 10 4 18 10 75 79 71 70 39 43 35 64 59 28 18 64 59 53 35 16 18 53 9 8 7 7 Note: M and F in the table represents number of male and female par cipants in the FGD, respec vely. * IS-5. Aman HYV - Boro HYV; IS-6. Aus HYV – Aman HYV; IS-7. T.Aus HYV – Aman HYV – Boro HYV; IS-8. T.Aus HYV – Aman HYV – Rabi; IS-9. Aman HYV – Rabi (long); IS-10. Improved year-round aquaculture; IS-11. Shrimp – HYV Aman 4. Conclusions and recommenda ons The recommenda on domain analysis approach resulted in detailed, cost-effec ve delinea on and mapping of cropping system suitability maps for the Barisal region of Bangladesh. Approximately 0.5 m ha of land was iden fied on which double or triple crop system could be adopted. A further 0.20, 0.60 and 0.59 m ha were found suitable for year-round brackish water aquaculture (and brackish water shrimp), T.Aus-T.Aman (tradi onal) and T.Aman (tradi onal)-rabi crop systems, respec vely. Approximately 4.7 m ha were found suitable for improved cropping systems including T.Aus (HYV) - T.Aman (HYV), T.Aman (HYV) - T.Boro (HYV), T.Aman (HYV)-rabi (long) and T.Aus (HYV) - T.Aman (HYV) - T.Boro (HYV) systems under various suitability classes. T.Boro (HYV) or T.Aus (HYV) or rabi crops/systems have enormous poten al in Patuakhali and Berguna Districts. Approximate 60% of the cropped area remains fallow a er a single tradi onal aman crop in Patuakhali and Berguna Districts. The RDA revealed that 0.22 m ha of single crop area in these two districts can be converted into double or triple crop systems through the introduc on of improved crop and water management prac ces. Specific improved and produc vity-enhancing cropping systems have been described that can improve produc vity in low salinity zones of Barisal Division. Given that proven technologies exist to ameliorate many of the problems encountered in these stress prone areas, the work undertaken in this study has the poten al to increase the efficiency of technology transfer and targe ng with limited resources. 537 Further benefits of RDA include the iden fica on of areas for targe ng next genera on stress tolerant cul vars (salt, submergence and drought tolerant) or new cropping systems in coastal and inland zones of South and South East Asia. 4.1 Limita ons of the study Several limi ng factors including technical and biophysical factors must be taken into account because they may impede the smooth and successful transfer of technologies. Though the study was conducted using the best available datasets and techniques, not all layers were of the same quality or detail and a few low resolu on, old or proxy layers were used to complete the datasets. The study has used the following breakpoints between classes (< 1.5 = marginal suitable (S1); 1-5-2.5 = suitable (S2), and; > 2.5 is most suitable (S3)), however this choice could be revised based on a sensi vity analysis using different breakpoints. Gravity Irriga on (an input parameter for RDA) is unlikely to play a bigger role in the future considering the increasing number of investments from donor funded projects, banks and service providers to buy small sized pumping sets in Barisal region. Given the limited me, the FGDs were conducted at one place in each of the selected polders, which may not be enough to pick up representa ve opinions of farmers, par cularly for very large polders. As men oned in sec on 3.3, results for medium and high salinity zones were not presented in this paper. We also omi ed the socio-economic indices (poten al technology adop on index and livelihood index) that also play a role in the full RDA methodology. These other areas and factors will be described in future papers. Acknowledgements We acknowledge the funding support provided by the CGIAR Challenge Program for Water and Food under the Ganges Basin Development Challenge program for conduc ng this research project. We also acknowledge BBS, BMD and BWDB (Na onal Ins tutes of Govt. of Bangladesh) for sharing research and census data for conduc ng this study. We also thank Mr. AHM Kausher (an independent evaluator and consultant for the FGD part of this study) for his dedicated efforts in conduc ng FGDs in selected polders of CPZ. References Ali, A. M. S. 2006. Rice to shrimp: Land use/land cover changes and soil degrada on in southwestern Bangladesh. Land use Policy 23(4): 421–35. Bailey-Serres, J., Fukao, T., Ronald, P.C., Ismail, A.M., Heuer, S., Mackill, D., 2010. Submergence tolerant rice: SUB1’s journey from landrace to modern cul var. Rice 3: 138–147. BBS. 2011. Yearbook of Agricultural Sta s cs of Bangladesh. Bangladesh Bureau of Sta s cs (BBS), Sta s c Division, Ministry of Planning, Government of People Republic of Bangladesh, Dhaka. Behzad, M., Albaji, A., Papan, P., Nasab, S.B., Naseri, A.A. and Bavi, A., 2009. Qualita ve Evalua on of Land Suitability for Prinicpal Crops in the Gargar Region, Khuzestan Province, Southwest Iran. Asian Journal of Plant Sciences 8: 28-34. Brammer, H., 2010. A er the Bangladesh Flood Ac on Plan: looking to the future. Environ. Hazards 9 (1): 118–130. 538 Bussel, L.G.J. van; Wart, J. van; Wolf, J.; Grassini, P.; Boogaard, H.L.; Groot, H.L.E. de; Claessens, L.; Yang, H.; I ersum, M.K. van; Cassman, K.G 2013. Climate Zona ons As Extrapola on Domains for Yield Gap Assessments. In: Proceedings of ASA, CSSA and SSSA Interna onal Annual Mee ngs on Water, Food, Energy & Innova on for a Sustainable World, Tampa, Florida, USA, 3-6 November 2013. Chandna, P., Ladha, J. K., Singh, U. P., Punia, M., Erenstein, O., Pathak, H., et al. 2009. Technology targe ng in underused lands in the Gange c Plains of South Asia. pp. 335e354. In J. K. Ladha, Y. Singh, O. Erenstein, & HardyB.. (Eds.), Integrated crop and resource management in the rice-wheat system of South Asia (pp. 395). Los Banos (Philippines): Interna onal Rice Research Ins tute. Chandna, P., Singh, S. S., Erenstein, O., Khan, A. R., Ladha, J. K., Gupta, R. K., et al.2010. Geo-spa al techniques for assessing natutal resource and targe ng resource conserving technologies. In Khan., et al. (Eds.), Resource conserva on technologies for food security and rural livelihood. Rice wheat consor um for the Indo- Gange c plains, P 527 Chandna, P., Khurana, M. L., Ladha, J. K., Punia, M., Mehla, R. S., & Gupta, R. 2011. Spa al and seasonal distribu on of nitrate-N in groundwater beneath the rice wheat cropping system of India: a geospa al analysis. Environmental Monitoring and Assessment Vol- 178: 545-562. Chandna, P., Ladha, J. K., Singh, U. P., Punia, M., Gupta, R., Sidhu B. S. 2012. Using remote sensing technologies to enhance resource conserva on and agricultural produc vity in underu lized lands of South Asia. Applied Geography Vol-32: 757-765. Chandna, p., Nelson A., Toung P., Khan Z., Hossain M., Rana S., and Rashid F. 2014. Extrapola on domains for efficient technology targe ng in coastal zones of Bangladesh. Rice Today. Vol.13-2: 20-21. Chowdary, P. K., Chandran, R. V., Nee , N., Bothale, R. V., Srivastava, Y. K., Lingle, P., et al. 2008. Assessment surface and subsurface waterlogged area in irriga on command area of Bihar state using remote sensing and GIS. Agricultural Water Management Vol-95: 754-766. Dey, N. C., M. S. Alam, A. K. Sajjan, M A. Bhuiyan, L. Ghose, Y. Ibaraki and F. Karim. 2011. Assessing Environmental and Health Impact of Drought in the Northwest Bangladesh. J. Environ. Sci. & Natural Resources Vol- 4(2): 89-97. Erenstein O., Hellin J., Chandna P., 2010. Poverty mapping based on livelihood assets: A meso-level applica on in the Indo-Gange c Plains, India. Applied Geography Vol-30: 112–125 Esa, E., 2014. Land Suitability Assessment for Sorghum and Maize Crops Using a SLA and GIS Approach in Dera Wereda, ANRS, Ethiopia. ERJSSH 1, 119-139. Forkuo, E.K., Nke a, A.K., 2011. Digital Soil Mapping in GIS Environment for Crop-Land Suitability Analysis. Interna onal Journal of Geoma cs and Geosciences 2: 133-146. Golder, P.C., Sastry, R.K., Srinivas, K., 2013. Research Priori es in Bangladesh: Analysis of Crop Produc on Trends. SAARC J. Agri. 11: 53-70. Getnet K., MacAlister C., 2012. Integrated innova ons and recommenda on domains: Paradigm for developing, scaling-out, and targe ng rainwater management innova ons, Ecological Economics, Vol -76: 34–41. Graaff, J., Amsalu, A., Bodnár, F., Kessler, A., Posthumus, H., & Tenge, A. 2008. Factors influencing adop on and con nued use of long-term soil and water conserva on measures in five developing countries. Applied Geography Vol-28: 271-280. Hodson. D., White. J., 2007. Use of spa al analyses for global characteriza on of wheat-based produc on systems. Journal of Agricultural Science Cambridge Journal of Agricultural Science 145 (1):115-125. 539 Halder, J.C., 2013. Land Suitability Assessment for Crop Cul va on by Using Remote Sensing and GIS. Journal of Geography and Geology Vol. 5-3:65-74. IOP, 2009. Adap ve measures for coping with increased floods and droughts in Bangladesh, IOP Conf. Series: Earth and Environmental Science 6: 292001. Islam, K.M.N., Islam, M.N., 2014. Spa al assessment of Livelihood, Poverty and Water Vulnerability. Bangladesh Integrated Water Resources Assessment Project Report: 1-28. Islam, M.S., Rahman, M.A., Sultana, N., Nath, B., Paul., A., 2012. Using Geospa al Techniques to assess the Salinity Impact on Agricultural Landuse: a study on Shyamnagar Upazila, Satkhira. Journal of Agriculture and Environment for Interna onal Development – 106: 157-169. Ismail A.M., Singh U.S., Singh S., Dar M., Mackill D., 2013. The contribu on of submergence-tolerant (Sub1) rice varie es to food security in flood-prone rainfed lowland areas in Asia. Field Crop Research Vol. 152: 83-93. Jafari, S., Zaredar, N., 2010. Land Suitability Analysis using Mul A ribute Decision Making Approach. Interna onal Journal of Environmental Science and Development 1, 441-445. Mackill, D.J., Coffman, W.R., Garrity, D.P., 1996. Rainfed Lowland Rice Improvement. Interna onal Rice Research Ins tute, Los Ba˜nos, Philippines, 242 pp. Mackill, D.J., Ismail, A.M., Singh, U.S., Labios, R.V., Paris, T.R., 2012. Development and rapid adop on of submergence-tolerant (Sub1) rice varie es. Adv. Agron. 115: 303–356. Mamun S.A., Rahman F., Yeasmin F., Islam M.A. 2011, Suitability of the Physical Proper es of Soil for Crop Produc on: A Study in Tangail Sadar. J. Environ. Sci. & Natural Resources, 4(2): 121-125. Milliman, J.D., Broadus, J.M. and Frank G. (1989). Environmental and Economic Impact of Rising Sea Level and Subsiding Deltas: The Nile and Bengal Examples. In Bangladesh Quest. Vol-1:11-12. Mirza, M.M.Q., 2011. Climate Change, Flooding in South Asia and Implica ons. Regional Environmental Change Vol 11- 1: 95- 107. Mondal, M.K., Paul, P.L.C., Humphreys, E., Tuong, T.P., Ritu, S.P. and Rashid, M.A. 2015. Opportuni es for cropping system intensifica on in the coastal zone of Bangladesh. These proceedings. Nuruzzamam, Md., 2006. Dynamics and Diversity of Shrimp Farming in Bangladesh: Technical Aspects. Bangladesh Centre for Advanced Studies (BCAS) and University Press Ltd. 431-460. Padmanabhan, G. (2008). UN situa on report no. 8. on Bihar, U ar Pradesh and Orissa. At. www.un.org.in/UNDMT/sitrep/260908_SIT_REP.pdf Accessed 17.01.14 Pandey, A. C., Singh, S. K., & Nathawat, M. S. (2010). Waterlogging and flood hazards vulnerability and risk assessment in Indo Gange c plain. Natural Hazards. 55-2: 273-289. Park I., Lee S., 2014. Spa al predic on of landslide suscep bility using a decision tree approach: a case study of the Pyeongchang area, Korea. Interna onal Journal of Remote Sensing Vol 35 -16: 6089-6112 Quader, O., Islam, Z., Rahman, H., Rahman, M., Sarkar, M.H., Khan, A.S., 2010. Suitable Site Selec on of Shrimp Farming in the Coastal Areas of Bangladesh using Remote Sensing Techniques (4 S Model). Proc. of Interna onal Conference on Environmental Aspects of Bangladesh. Rahman, Md. R., Ando, K., Takeda, S., 2013. Effect of Shrimp-Based Cropping Systems on Salinity and Soil Fer lity in a Coastal Area of Bangladesh: A Village-Level Study. Journal of Agricultural Science 5: 1-10. 540 Rashid, M.M., A.K.F. Hoque and M.S. I ekhar, 2004. Salt tolerances of some mul purpose tree species as determined by seed germina on. J. Biological Sci. 4: 288-292. Salam, M.A., Islam, S.M.M., Gan, J., Ross, L.G., 2012. Crab culture poten al in southwestern Bangladesh: alterna ve to shrimp culture for climate change adap on. IRJALS 1: 15-31. Samanta, S., Pal, B., Pal, D.K., 2011. Land suitability analysis for rice cul va on based on mul -criteria decision approach through GIS, Interna onal Journal of Science & Emerging Technologies. Interna onal Journal of Science Vol 2-1: 12-20. SRDI, 2010. Coastal saline soils of Bangladesh. Soil Report published by Soil Resources Development Ins tute. Ministry of Agriculture, Dhaka, Bangladesh. pp.96. Subbarao, G. V., Kumar, Rao, J. V. D. K., Kumar, J., Johansen, C., Deb, U. K., Ahmed, I., et al. 2001. Spa al distribu on and quan fica on of rice-fallows in South Asia: Poten al for legumes. Interna onal crops research ins tute for semi arid tropics, na onal remote sensing agency and DFID plant sciences research programme. Patancheru, Andhra Pradesh, India: ICRISAT UNDP, 2013. Policy Study on The Probable Impacts of Climate Change on Poverty and Economic Growth and the Op ons of Coping with Adverse Effect of Climate Change in Bangladesh. Support To Monitoring PRS And MDGs in Bangladesh, General Economics Division, Planning Commission, Government of the People's Republic & UNDP Bangladesh. (h p://fpd-bd.com/wp-content/uploads/2013/04/ The-probable-impacts-of-climatechange-on-poverty-and-economic-growth-andthe-op ons-of-coping-w.pdf ). The World Bank, 2009, South Asia: Shared Views on Development and Climate Change, Chapter 8: Natural Disasters, 2009, p. 117 541 Decentralized surface water irriga on as a pathway for sustainable intensifica on in southern Bangladesh: On how much land can the drop be brought to the crop? U. Schulthess, T.J. Krupnik, Z.U. Ahmed, and A.J. McDonald Interna onal Maize and Wheat Improvement Center, Mexico, Bangladesh and Nepal, u.schulthess@cgiar.org, t.krupnik@cgiar.org, z.ahmed@cgiar.org, a.mcdonald@cgiar.org Abstract Bangladesh faces threats to the sustained intensifica on of crop produc on, but opportuni es also exist. Declining groundwater tables and escala ng irriga on prices in northern boro rice-producing districts have focused a en on on the south where surface water is perceived as abundant and the opportunity for crop intensifica on is considered to be great. The Government of Bangladesh recently requested more than $7 billion of donor funds to develop surface water irriga on (SWI) to boost dry season (rabi) intensifica on in the south. This region is agro-ecologically complex due to temporal water and soil salinity dynamics, and there is poor informa on on current land-use intensity. We describe the opportuni es and constraints encountered in developing a procedure to es mate the land area for which SWI could be successfully deployed to intensify fallow land in 3.375 Mha and bring it into double (aman-rabi) cropping, as well as to boost yields and/or enable the use of alterna ve crops on land currently under low-yielding, rainfed, and non-intensive rabi cropping pa erns. Using Landsat 5 and 7 scenes for a segment-based classifica on with the random forest algorithm, we isolated current cropland and waterways in the area south of Bangladesh’s major rivers, excluding Sunderbans. Based on Landsat 7 and 8 scenes we extracted maximum rabi season enhanced vegeta on index values, which we classified into fallow, low-, and high-intensity use of cropland for the last three years. A 385 m buffer was applied to waterways carrying water in late March to es mate the command area serviceable by decentralized irriga on water sellers using independent pump sets. We inves gated the poten al for SWI on fallow and low-intensity land by applying a cropping risk matrix to address the twin threats of soil and water salinity. Our analysis indicates that there are at least 20,000 ha of fallow land under the low-risk category, while more than 100,000 ha of low-intensity cropland could poten ally be be brought into intensified produc on using SWI. Focusing on Bangladesh’s south-central hydrological zone, in which considerable volumes of surface water are available, our results indicate about 15,000 and 63,000 ha of fallow and low-intensity crop land could be irrigated with SWI with li le risk to soil or water quality. This informa on can aid in technology targe ng for the efficient deployment of SWI as a tool for intensifica on. Key message: We iden fied about 120,000 ha of cropland suitable for moving from dry season fallows to surface water irrigated cropland. Detailed maps are available and will facilitate technology targe ng. Keywords: remote sensing, technology targe ng, geo-spa al analysis, crop intensifica on, surface water irriga on 1. Introduc on Popula on growth projec ons and increases in per capita income indicate that global food requirements will con nue to expand for at least four more decades before they plateau, with es mates indica ng that a doubling of current staple crop produc on is required by 2050 (Godfray et al. 2010; Tilman et al. 2011). Cereals produc on could be boosted by expanding the land area devoted to cropping, rather than by raising yield poten al alone, though this will entail nega ve environmental externali es (e.g. reduced biodiversity) that should be avoided (Tilman et al. 2011). Sustainable agricultural intensifica on, defined as use of sound agronomy and purposeful manipula on of ecological processes to achieve increased produc vity while minimizing land expansion and environmental degrada on, has been proposed as a poten al solu on to these issues (Garne et al. 2013; Godfray et al. 2010). A key strategy for sustainable intensifica on is mul ple 542 cropping, whereby at least two crops are grown per year on the same piece of land. Currently, 59, 39, 47 and 93% of the arable land in the Indo-Gange c Plains of Bangladesh, India, the Nepali Terai, and Pakistan, respec vely, are irrigated (AQUASTAT 2013). Combined with the predominance of flooded rice as the stable crop, these figures have led to the percep on that most poten ally irrigable land in South Asia has already been brought into intensified produc on (see for example de Fraiture and Wichelns 2010; Godfray et al. 2010). This situa on raises the ques on: What agricultural environments of South Asia offer the greatest opportuni es for sustainable intensifica on? Rather than focusing on already irrigated environments, the answer may lie in more “game-changing” strategies to transform agricultural produc vity in the remaining rainfed or par ally irrigated environments where water resources are available yet land-use intensity is currently low, and where low-cost investments in surface water irriga on (SWI) could enable the move from single to double cropping. In these marginal environments, many of which lie in the under-developed eastern Indo-Gange c plain states of coastal Bangladesh and West Bengal in India, water resources development is low, and farmers typically grow only one rainfed monsoon season rice crop per year. Access to shallow groundwater is limited because upper aquifers are saline and may be restricted due to silty-clay in upper soil strata (MOA and FAO 2012). Moreover, increases in salinity of exis ng aquifers can be expected as sea levels will rise in the future due to climate change. This will further threaten food security. Deep tube well expansion is similarly problema c because of the poten al for saliniza on resul ng from the draw-down of upper water layers by industrial and domes c users (Brammer 2010) in addi on to concerns arising from the natural contamina on of groundwater with ground rock derived arsenic (Hossain 2006). SWI could help to ameliorate these problems. New low-li SWI pumps that increase the efficiency of water delivery per unit of fuel have recently become available, which could help to lower costs and poten ally encourage intensifica on (Santos Valle et al. 2014). However, where SWI is unplanned and poorly targeted, less than op mal performance may also be expected and social and environmental problems may arise from over-extrac on and water compe on. It is therefore important to approach decentralized SWI development intelligently, by iden fying and targe ng appropriate blocks of low-intensity cropping and fallow land for irriga on that can benefit from irriga on, and by insuring that sufficient water supply is available to sustain adequate crop growth without undesirable levels of surface water deple on. What is lacking, however, is up-to-date and precise informa on on how and where to target SWI efforts while making the best use of available freshwater resources without exhaus ng supply or degrading water quality. Southwestern Bangladesh hosts two of the country’s three administra ve divisions with the most people living below the poverty line (World Bank 2010). Only 50% of the region’s 3.4 million farming households grow more than one aman rice crop per year (MOA and FAO 2012). These farmers typically fallow their land during the dry season, while those that do manage a second crop usually cul vate low-input and -output legumes such as grasspea (Lathyrus sa vus), len l (Lens culinaris) and mungbean (Vigna radiata) using residual soil moisture. Only 15% of the region’s farmers have access to groundwater to grow dry season boro rice (MOA and FAO 2012). Using the southwest of Bangladesh (Fig. 1) as a case study this paper shows how remote sensing and GIS technologies can be used to assess land suitability for intensifica on through decentralized SWI in deltaic environments. Because of the regional specificity of the agronomic constraints that farmers face in southwestern Bangladesh, addi onal informa on on soil and water salinity, and dry season plan ng dates were also employed. The basic analy cal steps described in this paper can be modified and adapted to assess similar ques ons pertaining to fallow or low-produc vity land intensifica on and SWI in similar deltaic environments, thereby providing a tool for more effec ve technology targe ng to mobilize sustainable intensifica on and SWI development interven ons. 543 N 24°0'N 23°30'N Legend 23°0'N Major town Hydrological zone South Central 22°30'N South West Waterways Polder Area 22°0'N 0 88°30'F 25 50 89°0'F 100 Kilo me ters 89°30'F 90°0'F 90°30'F 91°0'F 91°30'F Fig. 1. Overview of study area, located in the southwest of Bangladesh. It contains two hydrological zones: south-central and southwest. Most of the land near the coast is enclosed by polders. 2. Material and methods 2.1 Data sources This analysis is based on Landsat 5, 7 and 8 satellite imagery (Level 1T), available for free from h p://earthexplorer.usgs.gov. We used the blue, green, red, near infrared (NIR) and both short wave infrared (SWIR) bands, all of which have a resolu on of 30 m. The western side of the study area is covered by Landsat path 138, row 44 and the eastern side is covered by path 137, rows 44 and 45. Data on surface water salinity covering the period from 2002 to 2012 had been obtained from the Bangladesh Water Development Board. Shape files of the most recent and reliable land eleva on and soil salinity classes were collected from the Bangladesh Country Almanac (BCA 2006) and Soil Resource Development Ins tute (SRDI 2000), respec vely. The BCA landtype shape file contains inunda on classes including Highland, Medium-Highland 1, Medium-Highland 2, Medium-Lowland, Lowland, and Very Lowland, corresponding to the depths at which floodwater is encountered during the monsoon season, as a marker for eleva on class, i.e., no consistent floodwater, <90 cm, 90–180 cm, 180 -275 cm, and more. 2.2 Cropland iden fica on Cropland was iden fied using a set of Landsat 5 scenes acquired on either 21 or 31 January 2010. In late January cropland could be easily separated from forest since vegeta on cover on cropland is generally low at 544 that me. The images were classified into two categories: cropland and “other” which included water, forest, urban areas and land used for aquaculture. In order to avoid poten al misclassifica on due to calibra on errors, raw images from the 2 Landsat paths were classified separately. We first created segments with eCogni on 9 (Trimble Naviga on Ltd., Westminster, CO). Segments are image regions that are more homogeneous within themselves than with nearby regions and represent discrete objects or areas in the image. Each image region then becomes a unit analysis for which a number of a ributes, on top of spectral a ributes, can be measured and used during the classifica on (Carleer et al. 2005). The en re study area measured more than 3 million ha and systema c sampling of ground truth data for the cropland iden fica on would have been a big endeavor. We therefore relied on high resolu on background satellite imagery available in ArcGIS 10.1 (ESRI, Redlands, CA) and visually classified more than 250 segments for each of the two classes to create a training data set. High resolu on satellite imagery contains much more detailed informa on than 30 m Landsat images or segments. Therefore, we chose those segments for training for which the corresponding pixels in the high resolu on images showed uniformity. This was made simpler by the fact that we had to iden fy just two classes, cropland and non-cropland. For each segment the following a ributes were used for classifica on: mean of the digital numbers of bands 1–5 as well as texture (all direc ons) (Haralick 1973) and the normalized difference vegeta on index (NDVI). Addi onally, we calculated the ra o of the NIR band to the visible ones (Equa on 1) as follows: (1) Subsequently, the Random Forest Classifier algorithm in WEKA (see Hall et al. 2009) was used to generate the classifica on rules. Machine learning algorithms do not depend on normal data distribu on assump ons and allow for lumping together dis nct classes such as forest, water, urban, etc. This reduces the effort needed to create dis nct training classes. It also automa cally chooses the relevant variables and discards the other ones. Once the classifier was trained it was used to classify the remaining segments. Subsequently, a visual quality control of the automa cally classified segments was conducted, again using high resolu on background imagery as a reference. Wrongly classified segments were manually assigned to the other class. 2.3 Iden fica on of waterways and surface water dura on We used Landsat 5 images acquired on 26 October 2009 and 8 November 2011, coinciding roughly with the end of the monsoon when waterways are at their maximum extent, to iden fy them. The same methodological approach used for the classifica on of cropland was employed. Some waterways in the study area are ephemeral. We therefore checked for the presence of water in rivers, canals and creeks using the Automated Water Extrac on Index (AWEI; Feyisa et al. 2014) with atmospherically corrected Landsat 8 images from 21 and 30 March 21 2014. AWEIsh was chosen because of its effec veness in improving water extrac on accuracy despite the presence of shadows resul ng from trees lining rivers, canals and water bodies, following Equa on 2: (2) where is the value for reflectance of the respec ve Landsat 8 imagery bands. Using the same threshold as described in the Feyisa et al. (2014) paper, values above 0 were assumed to be water pixels and values sh below 0, nonwater pixels. 2.4 Assessment of land-use intensity Land-use intensity was determined on the basis of a total of 44 Landsat 7 and 8 images acquired between 31 December and 10 April of 2011-12, 2012-13, and 2013-14. The Landsat 7 scenes had already been calibrated to surface reflectance by the United States Geological Survey (USGS). The Landsat 8 images were first calibrated to reflectance using the TOA-DOS approach (Chavez 1996). Since the NIR band of Landsat 8 has 545 different spectral proper es than Landsat 7, the Landsat 8 data were cross-calibrated using Landsat 7 imagery acquired within eight days before and a er the respec ve scene analyzed. The enhanced vegeta on index (EVI), as described by Huete et al. (2002), is a direct measure of the quan ty of light intercepted for photosynthesis (Equa on 3): (3) where is the reflectance surface a er atmospheric correc on, C1 and C2 are coefficients of the aerosol resistance term using the blue band to correct for aerosol caused errors in the red band, and L is the canopy background adjustment to rec fy differen al, nonlinear radiant red and NIR transfer through the crop canopy. When tracked during the course of a cropping season and used to determine maximum light intercep on, which typically corresponds to the peak of a crop’s leaf area index (LAI; Huete et al. 2002), EVI can be a good indicator of the produc vity of a crop community (Schulthess et al. 2012). We therefore measured the intensity of crop produc vity by quan fying the maximum EVI reached by the most widely grown field crops in the study area including lathyrus, fallow, wheat, mustard, mung bean, boro rice and maize. We extracted EVI trends from 10 or more known fields for each of the above crops in each of the three years. The use of repe ve and sequen al observa on is cri cal to capture maximum EVI because of the heterogeneous nature of agriculture in the study region, resul ng in divergent crop phenology both within and across crop species. Following extrac on, EVI values for each of the main crop types were plo ed (Fig. 2) as a func on of the number of days before or a er 1 January un l the 100th day of the year upon which the observa on in ques on was made, corresponding roughly to the first two-thirds of the rabi dry season. We grouped each of the cropland types into three intensity classes: (1) fallow land; (2) low-intensity cropland, comprised of lathyrus, len l and mungbean, neither of which are typically fer lized, weeded or irrigated, and which are broadcasted resul ng in sub-op mal crop stands and poor yields (Dalgliesh and Poulton 2011), and; (3) high-intensity cropland, including wheat, boro rice, maize and mustard, all of which are more intensively grown with higher levels of fer lizer applica on, weeding, pest management and irriga on than in the case of the first three. A er checking for normality and homoscedas city following Sokal and Rohlf (1995), we subjected data from the date upon which the maximum EVI value (corresponding to maximum LAI as a measure of peak produc vity) was observed in each class to a one-way ANOVA using JMP 8.0.2 (SAS Ins tute Inc., Cary, NC) for the 2011-12, 2012-13 and 2013-14 dry seasons. The F-test indicated significance (P<0.001) between classes in each of the three seasons analyzed. Separa on of means with the Tukey-Kramer’s range test at α = 0.05 showed that the fallow, low-intensity and high-intensity classes were consistently different and independent in each season. Because of the significant differences between cropland use intensity classes we then set thresholds to separate classes to be used for all subsequent EVI analyses. Thresholds were set as the mid-distance point between the lower boundary for the standard devia on of the lowest maximum EVI observa on for the high-intensity cropland types, and the uppermost boundary of the standard error for the highest EVI observa on for the low-intensity crop types. For example, in the 2012-13 season maize exhibited the lowest maximum EVI within the high-intensity crop use class at 83 days a er January 1, while the EVI of lathyrus peaked as the highest observa on within the low-intensity class at 27 days. The threshold between high- and low-intensity crop was therefore set as the mid-distance between the lower and upper boundaries for the standard devia ons of these observa ons, respec vely. This conserva ve process was used to dis nguish the low-intensity and fallow crop classes for each season studied. The last step in this analysis consisted of the extrac on of the maximum EVI value for each pixel of the calibrated Landsat scenes for the en re study area in order to broadly map the three land-use intensity classes for cropland (Fig. 3). 546 2.5 Crea on of a buffer area around rivers, canals and creeks Since the efficiency of axial flow pumps decreases with li height and because they can only push water horizontally without gravity feed within a limited distance (Santos Valle et al. 2014), we created a 400 m buffer around those waterways in which water was present in late March. The 400 m width of the buffer was chosen as an empirical value, assumed reachable under most circumstances given feedback from irriga on service providers using the pumps in tandem with flexible hose piping. Intensive agricultural prac ces can result in sedimenta on and nutrient loading of watercourses. Riparian buffers planted with species capable of ameliora ng these problems could aid in mi ga ng the nega ve effects of crop intensifica on. We consequently reduced the 400 m buffer further, excluding a 15 m strip adjacent to rivers and canals from cropping. This resulted in a 385 m wide buffer, which we deemed poten ally suitable for SWI. 80 2011-12 Lathyrus Fallow Wheat Musta rd 70 60 Mungbe an Boro rice Maize Len l α 50 40 β 30 20 γ F- sta s c for maxiumum EVI land use comparisons Tukey-Kramer HSD Level (α =0.05) F186.96*** High Intensity Low Intensity Fallow Land 10 0 80 2012-13 70 α 60 50 40 β 30 20 γ 10 F- sta s c for maxiumum EVI land use comparisons FTukey-Kramer HSD Level (α =0.05) 168.21*** High Intensity Low Intensity Fallow Land 0 80 2013-14 70 α 60 50 40 β 30 20 γ F- sta s c for maxiumum EVI land use comparisons 10 FTukey-Kramer HSD Level (α =0.05) 233.61*** High Intensity Low Intensity Fallow Land 0 -10 0 10 20 30 40 50 60 70 80 90 100 Number of days before or a er January 1 Fig. 2. Dynamices of the Enhanced Vegeta on Index (EVI) derived from Landsat 7 and 8 images collected over three winter seasons in southern Bangladesh for fallow land ( ), low- (lathyrus, len l, and mungbean, indicated by ) and high-intensity (wheat, maize, mustard, and Boro rice, indicated by ) crops. 547 2.6 Interpola on and temporal evalua on of surface water salinity dynamics Salinity concentra ons in the Bangladesh dal estuary vary in me, with salinity typically increasing as the dry season progresses. This results from the gradual reduc on of southward river, canal and creek water flow following the monsoon season (Brammer 2013), with important ramifica ons for irriga on water quality. To account for temporal changes in water salinity we created four datasets based on the median of the observed data from the second halves of the months January to April over the 11-year period (2002–2012). Each dataset was interpolated using Indicator Kriging to create a surface map of salinity. Salinity of river water is measured at sta ons on the main rivers only. No data exist for the other water bodies. Hence, kriging was deemed to give a good approxima on of the salinity levels of smaller rivers, canals and creeks. Those maps were then classified into three water salinity classes: 0–2 dS m–1 (high–quality), 2–4 dS m–1 (medium–quality) and >4 dS m–1 (low–quality). Water salinity tolerance varies greatly among crops. While maize is rather sensi ve, wheat is much more tolerant. In Australia a 10% yield reduc on for maize at 1.7 dS m–1 was reported, while for wheat that threshold was 4.7 dS m–1 (Evans 2006). 2.7 Reclassifica on and applica on of soil salinity and inunda on land types shape files The publically available soil salinity map provided by SRDI (2000) comes with various classes, some of them being “mixed”, i.e., a polygon may belong predominantly to one class, but may also contain data from another class. To simplify the analysis we reclassified all data into three classes, <2, 2–4, and >4 dS m–1 , by assigning the highest reported value in each class as the iden fier for the new class. 2.8 Matrix of land suitability based on soil and surface water salinity Since either high soil and/or surface water salinity are severe constraints for crop produc on we created a matrix as shown in Table 1 as a heuris c tool to simplify the analysis. These thresholds take into account crops that are rather salt intolerant, such as maize. Crop species and even cul vars within a crop species can vary greatly in their ability to withstand soil salinity (Ayers and Westcot 1989). 2.9 Intersec on of the layers and suitability analysis Cropland, EVI, surface water salinity, soil salinity, hydrozone and landtype layers were intersected to assess the suitability of cropland for sustainable intensifica on. Lastly, a subset of the land within the 385 m buffer was created. This resulted in a geospa al database that can be queried for extrac on of descrip ve sta s cs. Table 1. Salinity thresholds of soil and surface water used to determine the suitability classes for agricultural intensifica on and surface water irriga on. Soil Salinity [dS m-1] 0 -2 2-4 >4 0 -2 Highly suitability Medium suitability Non suitable Water Salinity (dS m-1) 2-4 Medium suitability Low suitability Non suitable >4 Non suitable Non suitable Non suitable 3. Results and discussion The study area covered 3.375 million ha of which 57% or 1.926 million ha were iden fied as cropland. Cropland coverage was rather evenly distributed, except for the southwest where large tracks of land with high surface water and soil salinity levels are being used for aquaculture. The network of waterways is much denser in the south-central hydrozone than in the south-west hydrozone. Most waterways in the la er hydrozone carry no or rela vely small amounts of water in the dry season. Water recharge in that part of 548 Bangladesh has been dras cally reduced since the 1976 comple on of the Faraka Dam in West Bengal and the plume of saline water in the Khulna-Sathkira region grows steadily in the winter months. Intermediate soil salinity is an issue in the coastal zones of the south where salinity can range from 2-4 dS m -1 and some mes can be even higher. Unfortunately, there are no regularly reported surface water salinity data available for the stretch of land south of Amtali to the coastline. Hence, actual water salinity levels for that area are not known and the interpolated data may not be en rely representa ve for this region. Land-use intensity levels for all cropland in the study area are summarized in Table 2. Our analysis revealed that in the three years analyzed fallow land area ranged between 219,000 and 271,000 ha or between 11 and 14% of the total cropland area. The Ministry of Agriculture and FAO (2012) placed the number for fallow land at 136,000 ha, while the Bangladesh Bureau of Sta s cs es mated 240,000 ha (2011). Hence our numbers are largely in agreement with the Bangladesh Bureau of Sta s cs but are about double as high as those of the Ministry of Agriculture and FAO. Different defini ons of what cons tutes fallow land and slight differences in the land area assessed may contribute to this discrepancy. Table 2. Land-use intensity levels of cropland during the rabi season in the south of Bangladesh in ha. Data were derived based on an analysis of Landsat 7 and 8 images and ground truth data collected by agronomists. Land-use intensity Fallow land Low-intensity High-intensity Total 2012 271,078 779,095 876,338 1,926,511 2013 218,806 915,548 790,732 1,925,086 2014 230,824 906,382 789,735 1,926,941 Other es mates of fallow land and land that is suitable for intensifica on are higher. Using both remotely sensed and administra ve data, Rawson et al. (2011) es mated that 800,000 ha of rabi season fallow or underu lized land suitable for cropping was available in southwestern and southeastern Bangladesh. The Bangladesh Agricultural Development Corpora on es mated that 634,000 hectares are regularly fallowed or under low levels of produc vity in Khulna and Barisal Divisions during the rabi season (2010), exclusively in the southwest. Taking the average of 2012 to 2014, we es mate that about 867,000 ha are under low-intensity cropping. Our results are slightly higher than those other es mates. No ceable are the intensive produc on levels of crops grown in the northern half of the study area (Fig. 3). Presumably, most of these fields are planted with boro rice, wheat or maize, and irrigated with groundwater. In the south-central hydrozone, a clear gradient of declining produc on intensity from the north to the south can be no ced. Groundwater in that region has high salinity levels, hence irriga on in the winter months is not commonly prac ced. The large tracks of land being cropped at the intermediate and low intensity levels indicate that there is substan al poten al to increase food produc on in the southern part of the south-central hydrozone. Most of that land is actually enclosed by polders. Hence, with appropriate management of the sluice gates it might be possible to “harvest” water with low salinity levels for irriga on. 549 N Legend Fallow Low Intensity High In ten sity Sur face Water 0 25 88°50'0''E 50 89°10'0''E 100 Kilometers 89°30'0''E 89°50'0''E 90°10'0''E 90°30'0''E 90°50'0''E 90°20'0''E 90°22'0''E 90°24'0''E 90°26'0''E Fig. 3. Land use intensity of cropland during the 2014 rabi season in southern Bangladesh. The analysis was based on a series of Landsat 8 images acquired between 31 December 2013 and late March 2014. At the lower right, a detailed view of the cropping intensity levels within a 385 m buffer is given. Since our main objec ve was to iden fy areas suitable for technology targe ng with axial flow and similar low-li , surface water pumps, we created a 385 m wide buffer within a 15 m distance adjacent to the waterbodies for which water could be detected with Landsat images. They have a resolu on of 30 m and therefore a large por on of waterways went undetected. But we might also have included false posi ves, i.e., land that is adjacent to surface water bodies that are shallow or have a low flow rate and therefore are not a reliable source for irriga on water; we could only detect whether surface water is present or not. When comparing surface water maps for the months of January to April we no ced a remarkable reduc on in the southwest hydrological zone. Hence, due to limited availability of water it may not be possible to address large tracks of land with SWI in that hydrological zone (see Fig. 1). The data reported in Table 3 are preliminary indica ons only of the area of land that might be addressed. A more detailed study especially for the south-central hydrological zone, which contains most of the fallow and low-intensity land, is needed to accurately determine the poten al for SWI. High surface water salinity may pose a severe limita on to intensifica on in the south only. Almost 80% of the land within the 385 m buffer zone next to waterways carrying water in March is in areas where soil and water salinity are below 2 dS/m, whereas only 13% is in areas that are not suitable. In addi on to salinity levels, land eleva on is another constraint limi ng usability of land. However, we determined that when analyzing the land types of the buffer land, a total of only 3% of the land is in areas that are either Lowland or Very Lowland. Hence, land eleva on is not a major constraint. Based on these constraints, our ini al data in this hydrological zone point to approximately 15,300 and 62,600 hectares of fallow and low-intensity land, respec vely, in the south-central zone that is not too saline for cropping and that could receive quality surface water irriga on. 550 Table 3. Current land-use intensity of cropland and its suitability for surface water irriga on in the delta region of Bangladesh. Suitability classes are the result of an intersec on of soil and water salinity levels as defined in Table 1. Numbers indicate the area in ha of land that are within a 385 m buffer adjacent to water bodies on which water was detectable in late March of 2014. Land in the Lowland and Very Lowland classes was excluded from these sta s cs. Values in parentheses indicate land area specific to the south-central hydrological zone, for fallow and low-intensity land, respec vely. Land-use intensity Fallow land Low-intensity High-intensity Total Highly suitable 14,403 (8,711) 86,159 (47,939) 66,562 167,124 Medium suitability 6,866 (6,361) 17,262 (14,688) 6,524 30,652 Low suitability 2,144 (2,144) 6,640 (6,640) 999 9,783 Non suitable 23,653 (6,730) 22,049 (11,206) 7,382 53,084 4. Conclusions and recommenda ons In order to assess the poten al of using surface water irriga on in the delta area of Bangladesh, we used a series of Landsat images to analyze current cropping intensi es and assess the poten al for sustainable intensifica on through surface water irriga on, considering various constraints such as surface water availability, land eleva on and salinity. In the western hydrological zone, there is only very limited poten al for intensifica on due to a lack of good quality surface water. However, in the southern part of the south-central hydrological zone there is a lot of land that is currently cropped at low-intensity levels or le fallow. The detailed maps derived in this study will allow for a targeted introduc on of axial flow pumps into the delta area of Bangladesh. Field trials will be required to determine op mal irriga on schedules for various crops that can be grown in these areas. Acknowledgements This research was conducted under the Cereal Systems Ini a ve for South Asia - Mechaniza on and Irriga on (CSISA-MI) project funded by the USAID Mission in Bangladesh. The contents and opinions expressed herein are those of the authors and do not necessarily reflect the views of USAID or the United States Government and shall not be used for adver sing or product endorsement purposes. We thank the Soil Resources Development Ins tute and Bangladesh Water Development Board for access to soil and water quality informa on. References AQUASTAT. 2013. AQUASTAT main country database. Food and Agriculture Organiza on of the United Na ons. Ayers, R.S. and D.W. Westcot. 1989. Salinity problems. In: FAO, editor Water quality for agriculture. FAO, Rome, Italy. p. 1-32. BADC. 2010. Minor irriga on survey report. 2009-2010. BADC Bangladesh Agricultural Development Corpora on (BADC), Dhaka. BBS. 2011. Yearbook of Agricultural Sta s cs of Bangladesh 2011. Bangladesh Bureau of Sta s cs (BBS), Dhaka. p. P. 305. Brammer, H. 2010. A er the Bangladesh Flood Ac on Plan: Looking to the future. Environmental Hazards 9: 118-130. Brammer, H. 2013. The Physical Geography of Bangladesh University Press Limited, Dhaka. Dalgliesh, N.P. and P.L. Poulton. 2011. Poten al for mungbean in sequence with Rabi-season wheat in southern Bangladesh. In: H. M. Raws, editor Sustainable intensifica on of Rabi cropping in southern 551 Bangladesh using wheat and mungbean. Australian Centre for Interna onal Agricultural Research, Canberra. p. 256. de Fraiture, C. and D. Wichelns. 2010. Sa sfying future water demands for agriculture. Agric. Water Manage. 97: 502-511. Evans, L. 2006. Salinity tolerance in irrigated crops. NSW Department of Primary Industries, available online at h p://www. dpi. nsw. gov. au/agriculture/resources/soils/salinity/crops/tolerance-irrigated. Feyisa, G.L., H. Meilby, R. Fensholt, R. Simon and S.R. Proud. 2014. Automated Water Extrac on Index: A new technique for surface water mapping using Landsat imagery. Remote Sensing of Environment 140: 23-35. Garne , T., M.C. Appleby, A. Balmford, I.J. Bateman, T.G. Benton, P. Bloomer, et al. 2013. Sustainable intensifica on in agriculture - premises and policies. Science 341: 33-34. Godfray, H.C.J., J.R. Beddington, I.R. Crute, L. Haddad, D. Lawrence, J.F. Muir, et al. 2010. Food security: the challenge of feeding 9 billion people. Science 327: 812-818. Godfray, H.C.J., I.R. Crute, L. Haddad, D. Lawrence, J.F. Muir, N. Nisbe , et al. 2010. The future of the global food system. Philosophical transac ons of the Royal Society of London. Series B, Biological sciences 365: 2769-2777. Hall, M., E. Frank, G. Holmes, B. Pfahringer, P. Reutemann and I.H. Wi en. 2009. The WEKA data mining so ware: An Update. . SIGKDD Explora ons 11: 1. Haralick, R.M., K. Shanmugam and I. Dinstein. 1973. Textural features for image classifica on. IEE Transac ons on Systems, Man, and Cyberne cs SMC-3: 610–621. Hossain, M.F. 2006. Arsenic contamina on in Bangladesh—An overview. Agr. Ecosyst. Environ. 113: 1-16. Huete, A., K. Didan, T. Miura, E.P. Rodriguez, X. Gao and L.G. Ferreira. 2002. Overview of the radiometric and biophysical performance of the MODIS vegeta on indices. Remote Sensing of Environment 83: 195–213. MOA and FAO. 2012. Master Plan for Agricultural Development in the Southern Region of Bangladesh. Ministry of Agriculture (MoA, Government of Bangladesh) and United Na ons Food and Agriculture Organiza on, Dhaka, Bangladesh. p. 122. Rawson, H.M.E. 2011. Sustainable intensifica on of Rabi cropping in southern Bangladesh using wheat and mungbean. In: R. H.M., editor ACIAR Technical Reports No. 78. Australian Centre for Interna onal Agricultural Research:. Canberra. p. 256 pp. Santos Valle, S., Qureshi, A.S., Islam, M.S., Hossain, M.A., Gathala, M.K., Krupnik, T.J., 2014. Axial flow pumps can reduce energy use and costs for low-li surface water irriga on in Bangladesh. , Cereal Systems Ini a ve for South Asia Mechaniza on and Irriga on (CSISA-MI) Project, Research Report No. 1. CIMMYT, Dhaka, Bangladesh. Schulthess, U., J. Timsina, J.M. Herrera and A. McDonald. 2012. Mapping field-scale yield gaps for maize: An example from Bangladesh. Field Crop. Res. 143: 151-156. Sokal, R.R. and F.J. Rohlf. 1995. Biometry. W.H. Freeman and Company, New York. SRDI. 2000. Soil Salinity Bangladesh. Soil Resources Development Ins tute., Dhaka, Bangladesh. Tilman, D., C. Balzer, J. Hill and B.L. Befort. 2011. Global food demand and the sustainable intensifica on of agriculture. Proceedings of the Na onal Academy of Sciences of the United States of America 108: 20260-20264. World Bank. 2010. Upda ng Poverty Maps of Bangladesh. The World Bank, Bangladesh Bureau of Sta s cs, World Food Program, Dhaka. p. 17. 552 Poten al for expansion of surface water irriga on through axial flow pumps to increase cropping intensifica on in southern Bangladesh A.S. Qureshi 1,2, S. Yasmin1, N.C. Howlader1, K. Hossain1 and T.J. Krupnik1 1 Interna onal Maize and Wheat Improvement Center, Bangladesh, a.qureshi@biosaline.org.ae, s.yasmin@cgiar.org, t.krupnik@cgiar.org, n.howlader@cgiar.org 2 Current address: Interna onal Center for Biosaline Agriculture, UAE Abstract In southern Bangladesh, the lack of irriga on and drainage facili es has resulted in poor land use, low crop produc vity, and increasing poverty. Minor irriga on is dependent on poor quality groundwater extrac on, whereas surface water irriga on is deficient due to lack of irriga on infrastructure. Limited availability of suitable water li ing devices is one of the major bo lenecks. As a result, an es mated 17,000 ha of land are le fallow in the dry season, whereas another 95,000 ha have lost their produc on poten al. To unlock the produc on poten al of this region, improvements in surface water irriga on services are required. Through the Cereal Systems Ini a ve for South-Asia – Mechaniza on and Irriga on project (CSISA-MI), the Interna onal Maize and Wheat Improvement Center (CIMMYT) is approaching these issues from a value chain and targeted technology transfer perspec ve to move towards the sustainable intensifica on of agriculture in this area. The centrifugal pumps widely used to li water from irriga on canals are technically less efficient and have high opera onal costs. As an alterna ve, CSISA-MI introduced imported axial flow pumps (AFPs) to farmers on a trial basis. During this study, AFPs have produced higher water discharge than centrifugal pumps while also reducing fuel use, and thus opera onal costs. Due to reduced me required to irrigate boro rice, an AFP would save about USD70 per season at 1-m head to USD38 at 3-m head, rela ve to centrifugal pumps. CSISA-MI is now working with domes c manufacturers to modify pump design and material to make AFPs more suitable for local condi ons and to reduce costs. During the next three years, the project’s target is to make AFPs economically affordable to the maximum number of irriga on service providers who can supply water to smallholder farmers in south-central Bangladesh. Among other on-going ini a ves (canal rehabilita on, drainage improvements, introduc on of new crop varie es and agricultural machinery), increased access to surface water irriga on is a step towards increasing agricultural produc vity in this region. Key message: Hydro-economic superiority of axial flow pumps over centrifugal pumps makes them a poten al technology for the expansion of irrigated dry season and double cropping in southern Bangladesh. Keywords: low li pump, value chain, irriga on infrastructure, rabi crops 1. Introduc on Agriculture is the major user of water in Bangladesh. Rice (Oryza sa va) is the main staple food and is grown on 75% of the total cul vated land, cons tu ng 90% of total grain produc on in the country (BADC 2013). Due to its compara vely high yield (na onal average 3.4 tons ha -1) compared to early summer aus (1.6 tons ha-1) and monsoon aman (2.0 tons ha-1), dry season winter boro rice produc on has expanded in the last two decades (Talukder et al. 2008). Boro is currently cul vated on an area of 4.8 Mha and contributes about 55% of the overall rice produc on in Bangladesh. The unprecedented increase in boro produc on has helped Bangladesh to increase its total rice produc on from 18.3 million tons in 1991 to 33.8 million tons in 2013, achieving near rice self-sufficiency (BBS 2013). Most areas of boro are in the northern parts of the country. The expansion of boro in the southern parts of the country has been slow. 553 Currently, about 80% of Bangladesh’s groundwater is used for irriga on, of which 73% is es mated to be used exclusively for boro cul va on (Rehman and Ahmed 2008). However, groundwater development problems (i.e. declining water tables, deteriora ng water quality, and increasing energy costs and carbon emissions) threaten the sustainability of irrigated agriculture in Bangladesh. Increasing uncertain es regarding rainfall pa erns due to climate change (which may impact groundwater recharge) also suggest that dependence on groundwater use for irriga on should be reduced and surface water resources must be developed to meet future crop water requirements. The extension of irriga on services in the areas where surface and groundwater resources are rela vely less developed (such as southern Bangladesh) could help considerably in reducing the pressure on areas where land and water resources are already stressed. Realizing this, the Government of Bangladesh (GoB) has developed the “Master Plan for the Southern Region”, to a ract foreign investments of over USD7 billion with a large focus on the expansion of exis ng surface water infrastructure to increase cropping intensity and produc vity of poorly produc ve land (FAO and MoA 2012). Historically, agricultural development in southern Bangladesh, which accounts for 27% of the country’s area and 21% of its popula on, has been largely ignored. This region now lags behind the north, with about 15% of the total cul vable land either fallow and/or not being used for cul va on due to soil and water salinity, waterlogging, and lack of access to surface water for irriga on (Bala and Hossain 2010). Surface water is perceived as abundant in parts of the South where river and canal networks have perennial flow, and where salinity levels do not cross crop-damaging thresholds. Conversely, saline shallow aquifers are common and prohibit the use of shallow groundwater for irriga on. One of the major bo lenecks in expanding irriga on facili es, despite abundant surface water availability, is the lack of surface water irriga on infrastructure and the limited availability of appropriate water li ing devices. For this reason, an es mated 50% of southern Bangladesh’s farmers currently grow only one rain-fed rice crop per year (FAO and MoA 2012). Currently, centrifugal pumps (also known as low li pumps, LLPs) are occasionally used for li ing water from canals for irriga ng crops in the southern region. Centrifugal pumps usually require “priming” before opera on by manually adding water through the outlet un l the en re tube and interior pump system is completely filled to avoid efficiency losses resul ng from air pockets in the suc on system. At least partly due to these technical difficul es and higher opera onal and maintenance costs, wide scale adop on of centrifugal pumps by farmers remains limited for dry season irriga on in southern Bangladesh. Through the Cereal Systems Ini a ve for South-Asia – Mechaniza on and Irriga on (CSISA-MI) project, the Interna onal Maize and Wheat Improvement Center (CIMMYT) is working on a strategy to develop public-private partnerships for the deployment of axial flow pumps (AFPs) imported from Thailand, with the longer-term aim of developing domes c produc on capacity at scale. AFPs are widely used in Thailand and Vietnam where irriga on li requirements are low and where large volumes of water need to be li ed at low pressure (Biggs 2011). This situa on is similar to that found in the southern delta of Bangladesh. The AFP is not a new technology. Use of the AFP in Thailand enabled many farmers to move from single to double rice cropping (Chinsuwan and Cochran 1986) and there exists a rela vely mature AFP manufacturing industry in Thailand and Vietnam. However, AFPs remain rela vely unknown in Bangladesh despite the country’s similar deltaic geo-morphology and poten al for surface water irriga on (SWI). Based on their research in the 1980s, the Interna onal Rice Research Ins tute (IRRI) also found AFPs well suited to low-lying, deltaic environments such as those found in the southwest and south-central zones of Bangladesh. However, before CSISA-MI’s work, they had not yet been introduced on a large scale within Bangladesh. The private sector has shown considerable interest in AFP technologies and is driving this ini a ve in collabora on with CSISA-MI. This paper discusses this unique public-private approach for mo va ng irriga on service providers and farmers to adopt AFPs to increase agricultural produc vity through improved surface water irriga on. The paper also highlights other interven ons that are necessary for boos ng agricultural produc on in this region of Bangladesh. 554 2. Methodology and approach We tested the technical efficiency and social acceptability of axial flow pumps before introducing them to farming communi es. For this purpose, a two- er approach was used. The first approach was related to the tes ng of technical aspects of axial flow and centrifugal pumps, whereas the second was associated with the field scale performance assessment of both pumps. The methodology used for both performance assessments is discussed below. 2.1 Hydro-economic performance assessment of axial and centrifugal flow pumps Since no informa on on the characteriza on and performance of axial flow and centrifugal pump under different condi ons was available, it was necessary to collect this data through tests before developing recommenda ons for farmers about the use of these pumps. Such data could provide sound basis for extension officers and ins tu ons to educate farmers about the efficient use of these pumps under different field condi ons. The hydraulic and economic efficiency performance tests of axial flow and centrifugal pumps were conducted from April through May of 2013 at the Bangladesh Agricultural Research Ins tute (BARI) located in Gazipur, Bangladesh (for details of these experiments and results, please refer to the report by Santos-Valee et al. (2014)). Four locally made prototype AFPs (AFP1-4) based on Thai designs were compared with two commonly used centrifugal pumps. The pumps used for on-farm demonstra on were similar to AFP1-4, but were of different sizes and manufacturers (Table 1). This paper reports the averages of for each kind of pump, which were compared using ANOVA and post hoc LS means planned contrast tests. The economic performance of the pumps was compared through ex-ante analyses (see Santos-Valee et al. 2014 for details). Parameters recorded during the tests included engine fuel consump on (l/h), and pump discharge (l/s). Hydraulic performance was assessed using the standard head verses discharge (H/Q) rela onship for each pump, employing calcula ons of pump discharge efficiency and determina on of the theore cal and actual pump discharges. Economic analysis included both fixed and variable costs. The former primarily comprised key capital outlays (e.g., costs of full pump set, engine, V-belts, etc.), with costs collected from local markets. The fixed cost per year was calculated from the sum of deprecia on and interest on investment. 555 Table 1. Specifica on of pumps demonstrated and tested during the rabi season of 2013-14 Pump ID AFP1 AFP2 AFP3 AFP4 AFP5 AFP6 AFP7 AFP8 AFP9 AFP10 AFP11 AFP12 AFP13 AFP14 AFP15 AFP16 AFP17 AFP18 AFP19 AFP20 CEN1 CEN2 Thai manufacturer Pa anakarnkol Pa anakarnkol Pa anakarnkol Pa anakarnkol Somphonlohamachine Somphonlohamachine Somphonlohamachine Somphonlohamachine Ruapatanakarn Chang Ruapatanakarn Chang Ruapatanakarn Chang Ruapatanakarn Chang Vichakarnkon Vichakarnkon Vichakarnkon Vichakarnkon SuthamKarn Chang SuthamKarn Chang SuthamKarn Chang SuthamKarn Chang Centrifugal pump (4m head) Centrifugal pump (4m head) Diameter (cm) 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 12.7 15.24 10 12.7 Length (m) 4.3 4.3 6.1 6.1 4.3 4.3 6.1 6.1 4.3 4.3 6.1 6.1 4.3 4.3 6.1 6.1 4.3 4.3 6.1 6.1 N/A N/A Impeller (cm) 20.3 20.3 20.3 20.3 25.4 25.4 25.4 25.4 20.3 20.3 20.3 20.3 25.4 25.4 25.4 25.4 ------- 2.2 Field-scale performance assessment of axial and centrifugal flow pumps The CSISA-MI project is being implemented in partnership with Interna onal Development Enterprises (iDE), which, through CSISA-MI, works closely with Rangpur Foundry Limited (RFL), a large scale manufacturer in Bangladesh and sales agent of imported AFPs. In 2013-14, field demonstra ons of AFPs were therefore conducted in collabora on with iDE and RFL. During the demonstra ons, CIMMYT and iDE led in explaining technical aspects of the pumps, whereas RFL looked a er the marke ng and accessibility and opera onal issues such as pricing, a er sale services and spare parts. Field-tes ng of axial flow pumps was carried out at 90 loca ons in Barisal Region, where quality surface water is rela vely plen ful. These short demonstra ons were arranged with the collabora on of RFL, and were used to encourage purchase of imported Thai pumps. The loca ons for pump demonstra ons were determined a er discussions with RFL dealers, water user associa ons and several farmer groups in different Upazillas (sub-districts). These brief demonstra ons were followed by several focus group discussions to select interested local service providers (LSPs) and poten al farmers for: (1) season-long demonstra ons including the use of normally fallow lands for rabi crops, and; (2) short roaming promo onal demonstra ons where LSPs installed AFPs for at least three weeks to deliver water to farmers. In total, 70 LSPs were selected and each LSP was tasked to bring 5 ha of fallow land under cul va on during the dry season by providing surface water irriga on. Seventy-eight percent of the LSPs irrigated boro rice crops, 15% irrigated maize, 4% irrigated mixed maize and wheat parcels, and 4.5% irrigated wheat parcels. To iden fy poten al LSPs for the demonstra on of AFPs, the following criteria were used: 556 Irriga on service provider/Block Manager Power ller engine owner Exis ng centrifugal pump owner Iden fied irriga on LSP according to the Department of Extension’s (DAE) records Has current service provider business and business tendency Literate with at least a Secondary School cer ficate Recognized as a social leader by farmers Technically proficient Willing to provide services, and willing to experiment with new technologies. Based on the above, 70 pumps were distributed to LSPs for longer-term field demonstra ons. Most LSPs replaced their centrifugal pumps with the AFPs, while others opted to establish newly cropped areas (Table 2 and Figure 1). Table 2. Distribu on of AFPs for demonstra on in three districts of Barisal Division Demos managed by CIMMYT iDE Total Barisal 13 15 28 Bhola 7 14 21 Patuakhali 7 14 21 Crop Maize Mixed Rice and Maize Rice Wheat Fig. 1. Spa al distribu on of AFP demonstra ons in southern Bangladesh (largely in Barisal Division). In each loca on, more than one pump may have been demonstrated. Note that because of the proximity of loca ons, some dots overlap. 557 To assess the field performance of the AFPs, 19 LSPs and 28 irriga on service recipient farmers belonging to Barisal, Bhola, and Patuakhali Districts were interviewed. The primary data for this assessment was collected by filling out two different sets of semi-structured ques onnaires with Lead Farmers/LSPs, as well as two service recipient farmers per Lead Farmer/LSP. The subjects for which recipient farmers’ and LSPs’ responses were recorded included: Demographic informa on (e.g. basic profile, household’s involvement with non- farm ac vi es, access to credit, assets, specifica on of the machinery and investment) Past and present agricultural volume and associated income Technical performance of pumps (ease of opera on, discharge, fuel consump on, efficiency of the pump, etc.) Opera onal problems and level of sa sfac on Customer sa sfac on A er sales service delivery Service received through AFPs (e.g. dura on, cul vated land area, service charges, accessibility, mely availability, level of sa sfac on regarding service) The selected LSPs and lead farmers were trained in the use and technical aspects of axial flow pumps before they used them in the field. The training also included book keeping for recording data on the size of irriga on blocks, opera ng me of each pump, fuel used, opera ng cost, and any technical and opera onal problems encountered. The trainings were held in all three districts both for female and male members of the farming community (Table 3). Table 3. Details of trainings on AFP under Barisal region District Bhola Barisal Patuakhali Interven on AFP + Mechanical repair + Business skill development AFP + Mechanical repair + Business skill development AFP + Business skill development Male 100 261 42 Female 0 6 1 Total 100 267 43 3. Results and discussion 3.1 Hydro-economic performance assessment of axial and centrifugal flow pumps The tes ng results indicate that the AFPs produced higher water discharge than centrifugal pumps at low heads (Table 4), consuming less fuel which, in turn, increases the poten al for service providers to save fuel costs and boost profits (Santos-Valee et al. 2014). The hydraulic performance (discharge rate) of AFPs was higher at low li s (i.e. 1–2 m), and dropped significantly at heads exceeding 2.8 m. On the other hand, centrifugal pumps produced low discharge but with consistent discharge rate at 1–3 m heads (Santos-Valee et al. 2014). For example, at 1-m head, average discharge of AFPs was 72% higher than of centrifugal pumps, whereas at 2 m and 3 m heads, AFP discharge was 55% and 28% higher than centrifugal pumps, respec vely. Although the discharge rates obtained by AFPs at 3-m heads were significantly lower than at 1-m head, they remained higher than with centrifugal pumps (Table 4). Water delivery per unit of fuel (m3/l) was highest with AFPs at 1 m head, and gradually decreased as head increased. This was not the case for centrifugal pumps, where water delivery per unit of fuel used remained almost the same at all heads. The water delivery per unit of fuel for AFPs at 1-m head was 112.4 m3/l, which declined to 91 m3/l and 69 m3/l at 2-m and 3-m head, respec vely. Water delivery per unit of fuel for centrifugal pumps at 1-m, 2-m and 3-m heads was 75 m3/l, 75 m3/l and 73 m3/l, respec vely. The average 558 water delivery per unit of fuel for AFPs was found to be 41% higher than centrifugal pumps at 1-m head, though it declined therea er, as indicated by the LS Planned contrasts test (Table 4). Table 4. Water discharge (m3/h) of different pumps at different head levels (adapted from Santos Valle et al, 2014) 1 AFP CEN 2 F-values 1 Water discharge (m3/h) 1-m 2-m 215 a 187 a 125 b 121 b 1428** 1124** 3-m 149 a 116 b 1106** Indicates Axial Flow Pump. 2 Indicates Centrifugal Pump. Values in columns for the Least Squares (LS) Planned Means Contrasts for Horsepower and Pump Type are significantly different at =0.05 according to the Student’s T Test. ANOVA results with a * indicate significance at P≤ 0.05, and ** indicates significance at P≤ 0.001. The four prototype AFPs used propor onally less fuel per unit of water delivered up to a head of 2.8 m compared to centrifugal pumps. A er 2.8 m of head, AFPs con nue to deliver more water than centrifugal pumps but consumed more fuel per unit of water delivered. The higher discharge rate of the AFPs can make significant reduc ons in the me needed to irrigate a crop. The maximum me required to irrigate one hectare of boro rice by an AFP was es mated at 86 h compared to 110 h for a centrifugal pump at 3-m li height (Santos-Valee et al. 2014). At 1-m and 2-m heads, 42 and 36 h less pumping me would be needed. When compared to centrifugal pumps, an AFP would save about USD70 per season at 1-m head and USD38 per season at 3-m head for irriga ng boro rice. 3.2 Field-scale performance assessment of axial and centrifugal flow pumps Local Service Providers also found the field performance of AFPs to be be er than centrifugal pumps in terms of discharge, fuel consump on, and opera onal cost. The average opera onal cost of AFPs (labour, maintenance and fuel costs) was much lower than that of centrifugal pumps. The LSPs confirmed that AFPs can save up to BDT200/ha (USD 2.6/ha) per irriga on for boro rice, which is significant for a crop that may require 15 to 30 irriga ons depending on field water depth, soil type, and proximity of the perched ground water table, indica ng the poten al for large aggregate savings (USD39 to 78/ha) when considered throughout the en re cropping season. Due to their higher discharge rate, AFPs required less irriga on me per hectare, which allowed irriga on service providers to provide water to more farmers than with centrifugal pumps, indica ng the poten al to scale up use of the AFPs to service more farmers while saving costs for irriga on water sellers. The LSPs also expressed sa sfac on with the ease of opera on (no priming is required with AFPs), water li ing capacity, service charge recovery, maintenance and repair. However, due to heavy weight, pump mobility and associated labor requirements were reported as primary issues. LSPs favoured AFPs due to technical backstopping by RFL and CSISA-MI, and reduced fuel costs (Figure 2). Large tracts of agricultural land in southwest Bangladesh can easily be brought under rabi season cul va on by extending surface water irriga on through the adop on of AFPs. Flexible hosepipes (FHP) can also be used to transport irriga on water up to 500 m from canals to fields. With their superior hydraulic performance, AFPs can help in irriga ng more area per unit of me while saving fuel. Following the field demonstra ons, 173 AFPs were purchased commercially by farmers in south-central Bangladesh to replace their centrifugal pumps. The AFPs were purchased commercially through RFL, with CSISA-MI’s facilita on. 559 Despite these benefits, large scale adop on of AFPs could be constrained by minor technical problems encountered during the field-tes ng, and the economic limita ons of farmers to buy imported pumps. The technical problems can be par ally a ributed to the lack of knowledge of LSPs and farmers about the proper installa on and opera on of these pumps under Bangladeshi condi ons. Therefore, solving technical problems and reducing costs would be the key for large-scale adop on of AFPs by farmers to intensify agriculture in this region. Overall Sa sfac on Mobilty No need priming Physical Structure Fuel cost Labor Cost involvement Client farmers’ sa sfac on Service charge recovery Seasonal demand for water Quality of technical services Availability of technical services Availability of a er sales service Availability of spare parts Ease of opera on Water li ing capacity 0% Sa sfactory 20% 40% Moderately Sa sfactory 60% 80% 100% Not Sa sfactory Fig. 2. Level of sa sfac on of LSPs with the axial flow pumps (n = 19). 4. Improving crop intensifica on in southwestern Bangladesh The southwest and south-central zone of Bangladesh is coastal and is prone to cyclones, storm surges, and extreme weather events including violent early monsoon season storms that can have devasta ng effects on crops grown in unprotected coastal areas. Since 1960, a total of 34 major cyclones have affected this region (BBS 2013). In southern Bangladesh (Khulna, Bagerhat and Satkhira Districts in par cular), insufficient southward freshwater flow has resulted in seawater intrusion during the dry season, which affects the salinity of groundwater and creates waterlogged condi ons (MoA 2012). Considerable efforts were made during the 1960s and 1970s to improve drainage condi ons and build embankments for the early release of floodwater, and to protect agricultural land from the sea. As a result 145 polders were developed covering more than 70% of the coastal area (MoA 2012). Due to persistent neglect and poor maintenance, at many places embankments are damaged, canals are silted, sluice gates are missing and the exis ng ones are not operated systema cally to regulate flow of water. Infrastructure in the polders is therefore o en inappropriate for effec ve use of water for irriga on. This is one of the major reasons why large tracts of land are le fallow during the dry winter season, which reduces the cropping intensity of cul vated lands. For this reason cropping intensity of this region is much lower than the rest of the country, par cularly in Khulna, Patuakhali, and Barisal districts, the la er two of which are well-suited for the use of surface water irriga on (Figure 3). 560 250 200 150 100 50 0 Fig. 3. Cropping intensity (2009-10) of selected districts of Bangladesh, excluding Chi agong (southeast) and Sylhet (northeast) (BBS 2013). The do ed line presents average cropping intensity. In the areas where service providers prac ce surface irriga on during the dry season, low-li water (centrifugal) pumps are usually rented from local organiza ons that are managing minor irriga on works in the region. However, the lack of widespread pump markets restricts the availability of pumps in the densi es needed to encourage uptake of surface water irriga on prac ces. There are many other steps that need to be taken simultaneously into account if the poten al of AFPs are to be fully realized. 4.1 Increase access to canal water Southern Bangladesh has a network of rivers and canals. A considerable amount of fresh surface water is available in the eastern part of this zone (especially in Barisal District) throughout the year apart from the dry, winter months (rabi). However, water availability between April to June (pre-Kharif) is rela vely limited (Schulthess et al. 2015). Most of the rivers in Barisal, Patuakhali and Pirojpur Districts are influenced by dal water movement that increases water salinity, especially in the second half of the rabi season when it some mes goes beyond the tolerance level of most crops. This presents a challenge to increasing dry season cropping. The geographical heterogeneity of water availability also poses a problem. This is par cularly the case where canals are silted up due to restricted southward water flow (resul ng in large part from trans-boundary water management and barraging in India), which raises riverbed levels (MoA 2012). For improved agricultural produc on in this region, irriga on and drainage infrastructure need to be redesigned, repaired, or even newly constructed. Dredging (de-sil ng) of targeted canals to lower their bed level could help bring more fresh water from the rivers into the polders. These interven ons should be supported by the repair and construc on of water control structures and pump houses (to bring water in during low de periods), and backed by the development of coopera ve water users associa ons. Water structures such as rubber dams/cross dams and regulators can be of great importance for the development of minor irriga on (MoA and FAO 2012). The water infrastructure also needs to be designed for equitable and op mal use of water by different stakeholders. This can result in significant water savings and increases in crop and aquaculture produc vity in the coastal region (MoA 2012). The Government of Bangladesh is already working on a large-scale canal dredging and community polder management plan with the financial and technical assistance of the Dutch government (Blue Gold: Program Document, 2012). The “Blue Gold” project focuses on polders in three districts—Patuakhali, Khulna and Satkhira—and presents opportuni es for partnerships with CSISA-MI and similar programs. This project aims to cover 160,000 ha with an es mated 150,000 households as direct beneficiaries. The scope of this work needs to be further extended for the benefit of large communi es of farmers. 561 4.2 Increase farmer par cipa on in water management To ensure sustainable management of irriga on and drainage infrastructure, par cipa on of farmers must be encouraged. This process can be best ini ated by favoring community ownership of water management infrastructure, backed by educa onal programs to encourage coopera ve and effec ve management. In many South Asian and La n American countries where large-scale surface water irriga on is prac ced (such as Pakistan, India, China, and Mexico) water user associa ons have been established to collect addi onal water charges beyond those given to LSPs. These fees are employed to maintain irriga on and drainage infrastructure at the ter ary level (the downstream part of the canal system from where farmers take control of water). In the Philippines, farmers are even involved in the planning, design and opera on of irriga on systems (Meinzen-Dick et al. 1995). Levine et al. (1998) have shown that joint management by farmers and government agencies have been successful in implemen ng water alloca on and cropping plans in Mexico. In Bangladesh, a mixed model approach may be most useful. The ‘irriga on boom’ (sensu, Shah et al. 2009) in Bangladesh and parts of South Asia has been built on the installa on of shallow tube wells and small irriga on command areas – o en less than 20 ha – in which LSPs play an ac ve role in canal maintenance (Chowdhury 2012). Shared care of irriga on infrastructure between water users associa ons and LSPs may prove to be promising. This decentralized approach has been successful in other countries and farmers have shown interest in taking over water management responsibili es if community ownership is made explicit (La f and Pomee 2009). Plusquellec et al. (1994) noted that opera on and management goals can be achieved if proper infrastructure and modernized irriga on systems based on the service concept are introduced. Hassan et al. (2004) argue that farmers are willing to pay an irriga on service fee if addi onal benefits are assured. 4.3 Ra onalizing cropping pa erns Despite limited surface irriga on water availability and high produc on costs, many farmers – including those surveyed in this study – are growing a water intensive boro crop. Since surface water availability in some of the canal systems in south-central Bangladesh is limited, farmers can be encouraged to switch to less water demanding crops (such as oilseeds, wheat and maize, and even more profitable hor cultural opera ons). In Bangladesh, boro crops are s ll con nuously flooded for most of the season. This prac ce wastes scarce water resources and increases irriga on costs (Alam et al. 2009). BRRI (2000) has es mated that, on average, ~4.0 m3 of water is used to produce one kg of boro rice in farmers’ fields, compared to 2.0 m3 used in research trials without compromising crop yields. The seasonal irriga on water requirement of boro crop varied from 4,840 to 5,720 m3 ha-1 in Dhaka District, and from 6,000 to 7,100 m3 ha-1 in the higher eleva on Barind area (Karim et al. 2009; Rashid et al. 2009). In contrast, water applica on for boro rice in Mymensingh is 12,800 m3 ha-1 (Sarkar and Ali 2010), and for the north-western region it varies from 12,000 to 13,500 m3 ha-1 for light and heavy soils, respec vely (Dey et al. 2013). Rashid et al. (2009) showed that of the total applied irriga on water only 55% is used for ET and the remaining amount is lost as seepage and percola on. The cost of this excess pumped water is ul mately borne by farmers and ranges from USD26 to 65 ha -1 for electric deep tube wells, and up to USD90 ha-1 for diesel operated STWs (Dey et al. 2013). Due to shallow water tables, a significant amount of crop water requirements could be met through the use of residual soil water at the end of the rainy season and capillary rise from the shallow water table in southern Bangladesh. Es mated irriga on requirements for wheat are approximately 100 mm (Poulton and Rawson 2011) in these environments. The actual irriga on requirements for Bhola and Barisal are es mated to be 29 and 65 mm, respec vely (Poulton and Rawson 2011), because of contribu on via capillary rise. Field studies indicate that rela vely high yielding wheat (up to 4.77 t ha–1) can be grown in south-western Bangladesh using only two appropriately med irriga ons, though yield will decline in environments prone to soil and water salinity (Krupnik et al. 2014). 562 4.4 Increase land produc vity Increased access to surface water through the rehabilita on of irriga on and drainage infrastructure, provision of axial flow pumps to li water from canals and improved on-farm water management prac ces could all help to ameliorate the constraints to increased cropping intensity in southern Bangladesh. However, increases in annual land produc vity will also require the availability and ra onal use of inputs (seed, fer lizer, weed control, etc.), appropriate agricultural technologies and agronomic prac ces. Considerable quan es of perishable and high value agricultural products produced in the coastal zone are wasted during storage or en-route to the market due to lack of proper storage facili es, processing and post-harvest opera ons; for fruit and vegetables losses are es mated at 20 to 45%, for cereals 15%, and 12% of fish and milk products are es mated to go to waste (MoA 2012). Therefore, there is a need to develop storage facili es at appropriate loca ons to avoid these losses. Increased agricultural produc on and diversifica on in cropping systems is also hypothesized to help create addi onal employment in the area. 4.5 Involvement of the private sector In Bangladesh, where the reach of public projects is limited, the role of the private sector in the development of agricultural produc on becomes very important. While much of the farm produc on in the coastal zone is subsistent in nature, farmers depend on the private sector for the provision of inputs and machinery/equipment, and in par cular on post-harvest LSPs for milling rice grain. In parts of south-western Bangladesh, where farmers are increasingly becoming involved in the marke ng of their crops, recent increases in maize produc on were mainly successful due to the establishment of feed mills, which generated market demand. The provision of crop reapers, land prepara on equipment, planters and threshers in Bangladesh is also driven primarily by the private sector, though o en with government subsidies. CSISA-MI therefore works with the private sector to make AFPs available to LSPs and farmers at affordable rates. In the long term, private sector investment in the development of the commercial AFP market is envisioned as a key means of assuring the sustainable deployment of pumps in the southwest. 5. Conclusions The extent of surface water irriga on is currently rather small due to limited availability of water li ing devices (such as low li pumps), farmers’ reluctance to intensify due to poor knowledge of best-bet agronomic prac ces, perceived risk (especially where tenure is insecure), and poorly developed input and output markets. These condi ons have resulted in poor land use, low crop produc vity, and increasing poverty in the area. To promote surface water irriga on, access to water li ing devices needs to be increased. Our work showed that axial flow pumps (AFPs) are a be er alterna ve to exis ng centrifugal pumps because of their hydro-economic superiority. They produce higher discharge rates, consuming less fuel at low li s, and can irrigate more area in a given me than centrifugal pumps. Their large-scale adop on, however, will require that related technical and opera onal problems are resolved and that local manufacturing is facilitated to reduce costs and make the pumps affordable to service providers. The widely used approach of local service provision, through which pump owners sell water to farmer-clients, can also be a workable solu on if a stronger trust rela onship is developed between LSPs and farmers. Efforts are underway to develop equitable and profitable business models that balance LSP’s needs for income with farmers’ needs for affordable irriga on. Work to facilitate the manufacturing and commercial availability of domes cally produced pumps is also expected to lower costs. This will make AFPs more affordable to small farmers and service providers, an important component in efforts to extend surface water irriga on in southern Bangladesh. The diversifica on of agricultural produc on systems and improved water management are also prerequisites to improving food and income security in this otherwise overlooked region. 563 In addi on to increased availability of surface water irriga on access to agricultural technologies, the ra onal use of inputs and best-bet agronomic prac ces, farmer educa on, and proper linkage to output markets is of paramount importance for encouraging crop intensifica on in this region. Farmer and LSP par cipa on in the maintenance of irriga on and drainage infrastructure is also necessary if intensifica on is to be achieved. Acknowledgements This research was supported by the USAID Mission in Bangladesh through the Cereal Systems Ini a ve for South Asia – Mechaniza on and Irriga on (CSISA-MI) project. The contents and opinions expressed herein are those of the author(s) and can in no way be taken to reflect the views of USAID and the United States Government. References Alam, M.S., Islam, M.A., Islam, M.A., 2009. Economics of alternate we ng and drying method of irriga on: Evidences from farm level study. The Agriculturists 7(1&2); P: 82-89. BADC (Bangladesh Agricultural Development Corpora on), 2013. Minor irriga on survey report 2012-2013. Bangladesh Agricultural Development Corpora on, Ministry of Agriculture, Government of Bangladesh. Bala, B. K., Hossain, M.A., 2010. Food security and ecological foot print of the coastal zone of Bangladesh. Env. Dev. Sust. 12: 531-545. BBS, 2013. Sta s cal Yearbook of Bangladesh. Bangladesh Bureau of Sta s cs (BBS) 2012. Dhaka, Bangladesh. P 554. Biggs, D., 2011. Small machines in the garden: Everyday technology and revolu on in the Mekong Delta. Modern Asian Studies. 46(1): 47–70. BRRI (Bangladesh Rice Research Ins tute), 2000. In: BRRI Annual Research Review for 1999. XIX. Rice Farming Systems. October 9-12, 2000. Bangladesh Rice Res. Inst. Gazipur, Bangladesh. Chowdhury, N.T., 2012. Irriga on Ins tu ons of Bangladesh : Some Lessons, Problems, perspec ves and challenges of agricultural water management. InTech, Rijeka, Croa a, pp. 113-132. Chinsuwan, W., Cochran, B.J., 1986. The axial flow low li pump in Thailand in small farm equipment for developing countries. Proceedings of the Interna onal Conference on Small Farm Equipment for Developing Countries. Interna onal Rice Research Center, Los Banos, Philippines. 2-6 September. Pp. 195- 203. Dey, N.C., Bala, S.K., Saif-ul-Islam, A.K.M., Rashid, M,A., 2013. Sustainability of groundwater use for irriga on in northwest Bangladesh. Policy Report prepared under the Na onal Food Policy Capacity Strengthening Programme (NFPCSP). Dhaka, Bangladesh. 89 pp. Government of the Netherlands, 2012. Blue Gold: Program Document for Integrated sustainable Economic Development by improving the Water and Produc ve Sectors in selected Polders. Embassy of the Kingdom of the Netherlands, Dhaka, Bangladesh. Hassan, M., Starkloff, R., Nizamedinkhodjayeva, N., 2004. Inadequacies in the water reforms in the Kyrgyz Republic: an ins tu onal analysis. IWMI Research Report No. 81. Colombo, Sri Lanka: Interna onal Water Management Ins tute (IWMI). Karim, N.N., Talukder, M.S.U., Hassan, A.A., Khair, M.A., 2009. Climate change and its impacts on actual crop evapotranspira on of boro rice in north central hydrological region of Bangladesh. Journal of Agricultural Engineering, The Ins tu on of Engineers, Bangladesh, Vol. 37/AE; P: 73-82. 564 La f, M., Tariq, J.A., 2009. Performance assessment of irriga on management transfer from Government to farmer-managed irriga on system: a case study. Irriga on and Drainage 58: 275-286. Levine, G., Cruz Galvan, A., Garcés-Restrepo, C., Johnson III, S., 1998. Performance of two transferred modules in the Lagunera region: water rela ons. IWMI Research Report No. 23. Colombo, Sri Lanka: Interna onal Water Management Ins tute (IWMI). Meinzen-Dick, R., Reidinger, R., Manzardo, A., 1995. Par cipa on in Irriga on. Paper No. 003. Social Policy and Rese lement Division, Washington, D.C. MoA (Ministry of Agriculture), 2012. Master Plan for the Agricultural Development of Southern Bangladesh. 6th Five Year Plan, Dhaka, Bangladesh. MOA and FAO, 2012. Master Plan for Agricultural Development in the Southern Region of Bangladesh. Ministry of Agriculture (MoA, Government of Bangladesh) and United Na ons Food and Agriculture Organiza on, Dhaka, Bangladesh, P. 122. Plusquellec, H., Burt, C.M., Wolter, H.W., 1994. Modern water control in irriga on: concepts, issues and applica on. World Bank Technical Paper No. 246. Rawson, H.M., (ed) 2011. Sustainable intensifica on of Rabi cropping in Southern Bangladesh using wheat and mungbean. ACIAR Technical Reports No. 78. Australian Centre for Interna onal Agricultural Research: Canberra. 256 pp. Kabir, M.A. Khan, W., Saleh, L.R., Khair, M.A., 2009. Es ma on of water Loss from irrigated rice fields. SAARC Agri., 7(1), P: 29-42. Krupnik, T.J., Ahmed, Z.U., Timsina, J., Shahjahan, M., Kurishi, A.S. M.A., Miah, A.A., Rahman, B.M.S., Gathala, M.K., McDonald, A.J., 2014. Forgoing the Fallow in Bangladesh’s Stress-Prone Coastal Deltaic Environments: Effect of Sowing Date, Nitrogen, and Genotype on Wheat Yield in Farmers’ Fields. Field Crop. Res. 170, 7–20. Schulthess, U., Krupnik, T.J., Ahmed, Z.U. and McDonald, A.J. 2015. Decentralized surface water irriga on as a pathway for sustainable intensifica on in southern Bangladesh: On how much land can the drop be brought to the crop? These proceedings. Shah, T., 2009. Climate change and groundwater: India’s opportuni es for mi ga on and adapta on. Environmental Research Le ers 4, 035005. Santos Valle, S., Qureshi, A.S., Islam, M.S., Hossain M.A., Gathala, M.K., Krupnik, T.J., 2014. Axial flow pumps can reduce energy use and costs for low-li surface water irriga on in Bangladesh. Research Report No. 1. Interna onal Maize and Wheat Improvement Center, Dhaka, Bangladesh. 36 pp. Sarkar, A.A., Ali, M.H., 2010. Irriga on Management for Op mizing Rice Yield and Nitrate leaching. Proceedings, paper Meet-2010. Agricultural Engineering Division, Ins tu on of Engineers, Bangladesh. pp:17-28. Talukder, R.K., 2008. “Food security in Bangladesh: Na onal and global perspec ves,” In Proceedings BKAS 13th Na onal Conference and Seminar on Climate Changes: Food Security in Bangladesh, Vol. 13, Dhaka, Bangladesh. 565 Increasing agricultural and aquacultural produc vity in the coastal zone of Bangladesh M. Sirajul Islam, S.K. Biswas, D. Gain, M.A. Kabir and T.A. Quarashi BRAC, Bangladesh sirajul.i@brac.net, shankar.biswas@brac.net, dhiman.gain@brac.net, abid.km@brac.net, tausif.qurashi@brac.net Abstract The coastal zone of the Ganges delta is very suscep ble to salinity intrusion and seasonal cyclonic storms. The poverty of farming families in the polders of the coastal zone is extreme. Physical and social factors have prevented many farmers from taking advantage of the technological improvements in rice farming and aquaculture that could increase profitability, produc vity and resilience in the polder areas. Coastal zone farmers usually grow a single rice crop using low yielding tradi onal varie es. Furthermore, cropping intensity is low with the result that produc vity of this region is also low. There are great opportuni es to increase produc vity in the coastal zone through crop intensifica on and diversifica on. In light of this, BRAC is scaling out suitable technologies and varie es across a wide range of people throughout the coastal zone. The program described here focused on increasing farming system produc vity and profitability to improve livelihoods. A par cipatory approach was applied to facilitate the scaling out process from varietal selec on to technology adop on. Hybrid variety trials and cropping system trials demonstrated be er produc on in the polders than was previously possible. Several promising short dura on rice varie es as well as salt-tolerant non-rice (rabi) varie es were introduced and were widely accepted by the farmers and are becoming very popular. All the single crop land in the demonstra on blocks is now being converted to double or triple crop land. Rice-fish integrated culture with hor culture on dykes is now well established among the par cipa ng farmers. Addi onally, we piloted community-based brackish water aquaculture to improve produc vity. Results of valida on ac vi es found double the produc on both in agriculture and aquaculture, enabling farmers to obtain more profit than in the previous year. The results of the study will be useful for further scaling out of endeavors in another regions. This type of par cipatory valida on work will have a great future impact on agricultural extension work as well as research work in Bangladesh. Keywords: rice varie es, aquaculture, improved technology, scaling out 1. Introduc on Bangladesh is a small country with an area of about 147,570 km2 and a large popula on of about 160 million. The coastal zone covers about 20% of the country and over 30% of the net cul vable area. As an agrarian country, agriculture is the key economic driver in Bangladesh. More than 50% of the people are directly involved and 30% indirectly involved in this sector. It has been es mated that the popula on of Bangladesh will be 194 million by 2050, when the total rice demand will be 49 million tons per year. As a result it is essen al to ensure increasing crop produc on at a rate that will guarantee food security in the near future. The coastal zone of Bangladesh largely missed the benefits of the green revolu on. Many polders were created in the coastal zone during the early 1960s. The primary func ons of the polders were the protec on of the land from dal flooding and salinity intrusion. This enabled the cul va on of tradi onal aman rice crop varie es, which were long dura on and low yielding. A er harves ng of rice the majority of lands within the polders were le fallow, or sown with a low input, low yielding legume crop in some loca ons. Produc on of high yielding or high value rabi crops was not possible because of the late harves ng of aman rice. Most of the lands in the polders are not suitable for growing the improved high-yielding varie es of rice because the water is too deep for their shorter stature. Shrimp and fish are being farmed in many coastal polders but yields are low and well below poten al. 566 Keeping these challenges in view, BRAC, the largest non-governmental organiza on of the world, has worked to validate and scale out climate smart agricultural and aquacultural technologies to farmers in the coastal zone of Bangladesh since 2012 with the following objec ves: To enhance system produc vity by increasing aqua-cultural and agricultural cropping intensity To ensure food security and to improve the livelihood status of the coastal communi es through dissemina ng aquacultural and agricultural technologies To develop farmer’s capacity in increasing climate change adapta on and resilience. In this paper we present the ac vi es and findings of the 2013-14 program. 2. Methodology BRAC Agriculture and Food Security Program (AFSP) disseminates agricultural technologies through farmers’ par cipatory large scale block demonstra ons. It is considered that large scale demonstra on of improved technologies in crop fields and ghers inspires the neighboring non-par cipant farmers to adopt the improved and modern technologies in their fields. The technology dissemina on strategy is to convert single crop areas to double or triple cropped areas. This is done by introducing stress tolerant agricultural rice and fish varie es to the cropping systems, and by incorpora ng high value rabi crops in the rice-based cropping systems through the use of shorter maturing rice varie es to enable ‘early’ ( mely) rabi crop establishment. BRAC AFSP organizes groups of 40 to 50 marginal farmers farming in a con guous area (demonstra on block) and provides them with par al support to cul vate modern varie es of crops, using improved produc on technologies and prac ces. The farmers are provided with training and the latest informa on for increasing produc on from their field. It has been es mated that the farmers provide about 55% of the rice produc on cost, mainly in terms of labor, fer lizer and irriga on, and the program provides about 45% of the produc on cost in the form of cash to purchase inputs. At present, AFSP is opera ng its extension ac vi es in 50 sub-districts of 12 districts of Bangladesh. Most of the opera onal sites are disaster and stress prone areas of the coastal area where the target is to cover around 60,000 direct par cipants with improved technologies by 2015. The target group is mainly poor and marginal farmers in coastal communi es. Results from the 2013-14 seasons are reported in this paper. Yield data were taken from an area of 20 m2 in five randomly selected farmers’ fields in each demonstra on block. Grain moisture content was determined using a grain moisture meter and yield is presented at 14% moisture content for cereal crops and at 10% for oil seed crops. 3. Ac vi es and results of the 2013-14 program 3.1 Extending improved rice varie es High yielding rice varie es, both hybrid and inbred, were demonstrated during the aus, aman and boro cropping seasons. Hybrid rice as well as salt tolerant boro inbred rice were demonstrated in the aus and boro seasons, while high yielding varie es of inbred rice were demonstrated during the aman season. 3.1.1 Aus Hands-on training for rice cul va on was given to the par cipants in the 2013 aus season and they were advised to cul vate hybrid rice. The farmers cul vated hybrid rice (Alloran, Shak -2 and Sathi) from April to August 2013 in Barisal and Khulna regions. They applied irriga on at the beginning of the cropping season, usually up to May. Later on the fields were inundated with dal water that came naturally, and which was sufficient to meet crop water requirements in most cases. 567 Eighty-eight farmers cul vated hybrid dhan Shak -2, 80 cul vated hybrid dhan Sathi and 179 cul vated hybrid dhan Alloran. With iden cal growth dura on, all varie es had similar average yield (7.1-7.6t/ha). The average hybrid Aus produc on was 7.32 t/ha (Table 1). Average net return was 63,000 to 90,000Tk/ha (Fig. 1). Table 1. Block number, farmer number, seedling age, growth dura on, yield and yield components of hybrid rice in aus season 2013 Variety No. of blocks No. of farmers Hybrid Shak -2 20 Hybrid Sathi 16 Hybrid Alloran 37 Average yield: 7.32 t/ha Seedling age (d) 88 80 179 20 23 22 Growth dura on (d) 119 118 120 Effec ve 1000 llers/ grain hill weight (g) 9 25 11 26 10 28 % filled grains 86 93 87 Yield at 14% (t/ha) 7.29 7.58 7.08 139188 140000 123878 112813 120000 89582 100000 80000 66938 56940 60000 62723 49606 50090 Produc on cost Income Profit 40000 20000 0 Sak 2 Sathi Alloran Fig. 1. Average produc on cost, income and profit of three hybrids during aus season 2013 (Tk/ha). 3.1.2 Aman Nine aman varie es were cul vated in 430 blocks of 12 districts by over 2,000 par cipants. Five crop cuts (1 m2) were taken in four loca ons in five fields in each block for yield determina on. Average variety yields in the main block ranged from 5.2 t/ha (BRRI dhan39) to 7.1 t/ha (BRRI dhan54) (Table 2). The na onal average aman yield for 2011-2012 has been es mated at 2.29 t/ha, less than half the average produc on in the demonstra on blocks (5.84 t/ha). Average cost of produc on was 43,000 Tk/ha, which brought an average net profit of 66,000 Tk/ha (Table 3). In main blocks farmers followed the improved cul va on technology where the yield of the aman varie es was 0.5 to 1.6 t/ha higher than yield of the same varie es grown by farmers outside the demonstra on blocks (“non-block farmers”), who usually follow tradi onal cultural prac ces. Yield of BRRI dhan54 in the non-blocks (6.2 t/ha) was much higher than yield of all the other improved varie es (4.2-5.0 t/ha) and of the local varie es (2.1-5.4 t/ha). 568 Table 2. Number of demonstra on blocks and farmers, seedling age at me of transplan ng, growth dura on, yield and yield components of HYV aman rice in 2013 in demonstra on blocks (“Main Block”) and fields outside the blocks (“Non-block”) Variety 1.0 Main Block BINA dhan7 BRRI dhan33 BRRI dhan39 BRRI dhan41 BRRI dhan49 BRRI dhan51 BRRI dhan54 BR22 BR11 Total/mean 2.0 Non-block BINA dhan7 BRRI dhan33 BRRI dhan39 BRRI dhan41 BRRI dhan49 BRRI dhan51 BRRI dhan54 BR23 BR22 BR11 Jamaibabu (local) Duthkalam (local) Sadamota (local) Sworna (local) Growth Seedling dura on age (d) (d) Yield and yield components No. of blocks No. of farmers 86 61 6 10 107 22 2 21 115 430 430 302 30 50 532 110 10 105 570 2139 21 22 24 25 23 23 24 23 23 115 117 119 145 134 147 171 144 139 12 11 9 10 12 10 16 9 12 90 88 83 92 89 90 95 82 88 24 24 22 22 23 23 26 23 25 6.06 5.47 5.15 5.44 5.84 5.40 7.08 5.81 6.27 5.84 - 88 36 4 10 102 12 4 50 4 165 23 51 13 30 32 32 30 32 32 37 38 38 35 36 35 36 44 47 127 125 122 145 140 152 173 156 140 147 122 116 154 165 11 10 10 10 12 9 12 8 9 11 10 8 9 9 86 82 80 83 86 81 87 76 74 84 78 82 78 85 23 24 23 23 24 24 25 24 22 24 22 25 25 24 4.93 4.72 4.65 4.43 4.96 4.25 6.15 4.14 4.22 4.63 5.40 2.67 2.10 4.59 Effec ve % filled llers/ grains hill 1000 grain weight (g) Yield at 14% (t/ha) 569 Table 3. Total produc on cost, gross income, profit and BCR of HYV rice in aman season 2013 Variety BINA dhan7 BRRI dhan33 BRRI dhan39 BRRI dhan41 BRRI dhan49 BRRI dhan51 BRRI dhan54 BR22 BR11 Average Total produc on cost (Tk./ha x 1000) 45.6 43.2 44.2 42.3 43.4 42.3 42.8 44.8 42.3 43.3 Gross income (Tk/ha x 1000) 118 106 103 106 111 106 111 113 113 110 Profit (Tk/ha x 1000) 73.2 62.9 59.3 63.5 67.7 63.9 68.4 67.7 70.4 66.3 BCR 2.70 2.49 2.41 2.55 2.60 2.52 2.62 2.51 2.66 2.56 3.1.3 Boro In the 2013-2014 boro season around 1200 farmers cul vated a range of hybrids and inbreds in 262 blocks. The highest average yield of ~9.5 t/ha was observed with the hybrids Sak -2 and Sathi. Average non-block yields of respec ve hybrids were 1.2 to 1.9 t/ha lower than yields in the demonstra on blocks. In non-block fields, the average yield of hybrid Shak -2 was about 3 t/ha higher yield than yield of the local variety Vojon (Table 4). Table 4. Number of demonstra on blocks and farmers, yield and yield components of rice in the 2013-14 boro season. Yield and yield components Variety 1.0 Main Block Hybrid Shak -2 Hybrid Alloran Hybrid Sathi BRRI dhan28 Total 2.0 Non-block Hybrid Shak -2 Hybrid Alloran Hybrid Sathi BRRI dhan28 ACI HERA Vojon (local) 570 No. of blocks No. of farmers 42 27 93 100 262 210 135 430 420 1195 13 10 13 12 83 92 93 85 27 28 30 24 9.45 7.71 9.54 6.70 34 10 54 332 10 34 24 13 6 11 10 11 14 9 87 88 87 82 86 66 85 27 25 28 24 27 27 26 7.93 5.78 8.32 5.51 8.60 7.30 4.83 Effec ve llers/ hill % filled grains 1000 grain weight (g) Yield at 14% (t/ha) 3.2 Extension of non-rice crops 3.2.1 Maize During the 2013-14 rabi season, more than ten thousand (10,147) farmers cul vated maize hybrids (Pacific 984, and Pacific 999 super) on 2,572 ha. Yield and yield component data were taken from 742 farmers across 188 main and satellite blocks (areas where a technology is only demonstrated for one season). Crop cuts were taken from the fields of five par cipants per main block and from two farmers’ fields per satellite block. Average yield of maize was 8.5t/ha and varied from 6.3 t/ha in Jhalokathi District to 11.2 t/ha in Bogra District (Table 5). Average yields in excess of 10 t/ha were also achieved in Rangpur, Lalmanirhat and Kurigram Districts. Yield in the medium salinity coastal districts (Khulna, Bagerhat and Sathkhira) were lower at 8 to 9 t/ha. Table 5. Yield and yield components of maize in different loca ons including the coastal region of Bangladesh during rabi season 2013-14 District 1. Main Block Barguna Jhalokathi Pirojpur Bagerhat Khulna Satkhira Bogra Rangpur Lalmonirhat Kurigram Average 2. Satellite Block Barguna Patuakhali Jhalokathi Pirojpur Bagerhat Khulna Satkhira Rangpur Lalmonirhat Average Overall average Plants per m2 Yield and plant characters Grains per cob % of filled grains 1000 grain weight (gm) Yield at 10% (t/ha)a 4.69 5.26 4.82 4.81 5.75 5.70 7.02 6.76 6.94 7.45 5.92 463 508 665 561 352 640 129 119 125 107 367 98 96 99 99 97 99 99 99 99 99 98 307 288 312 346 320 369 351 344 363 451 345 8.25 6.26 8.21 8.11 8.33 9.09 11.23 10.79 10.75 10.67 9.17 4.73 4.76 5.03 4.66 4.76 5.36 5.98 6.04 7.24 5.39 5.66 663 546 575 642 521 531 573 105 122 475 421 90 95 93 98 98 98 97 97 99 96 97 300 325 310 309 329 343 351 322 320 323 334 7.73 8.04 6.92 7.55 7.25 8.28 6.55 8.34 9.03 7.74 8.46 571 Fig. 2 Maize cul va on in the 2013-14 rabi season in Pirojpur district. 3.2.2 Sunflower In the 2013-14 rabi season, cri cal inputs were provided to 11,299 farmers for hybrid sunflower (Hi-sun33) cul va on on 2,844 ha in both coastal and northern regions of Bangladesh. Yield and yield component data were collected from 961 farmers’ fields across 260 blocks (both main and satellite). Five crop cuts were taken from each main block and two from each satellite block. Average yield of sunflower was 2.79 t/ha and average district yield varied from 2.01 to 3.26 t/ha. Highest grain yield (3.26 t/ha) of sunflower was observed in Barguna District (Table 6). Fig. 3. Sunflower in the 2013-14 Rabi season in Bagerhat 572 Table 6. Yield and plant characters of sunflower in rabi season 2013-14 (average of five crop cuts per main block and two per satellite block) Yield and plant characters District 1. Main Block Barguna Patuakhali Jhalokathi Pirojpur Bagerhat Khulna Satkhira Bogra Rangpur Lalmanirhat Kurigram Average 2. Satellite Block Barguna Patuakhali Jhalokathi Pirojpur Bagerhat Khulna Satkhira Gopalgonj Average Overall average Popula on (plants/m2) Flower Seed per diameter (cm) flower % of filled seeds 1000 seed weight (g) Yield a 10% (t/ha) 2.80 2.78 2.99 2.98 2.94 2.87 3.60 2.41 3.05 3.09 2.825 2.94 68 75 15 21 18 21 19 20 26 21 23 30 708 1,220 635 581 985 1,341 1,152 179 249 235 243 685 93 87 85 96 88 86 92 84 80 86 72 86 76 86 94 108 99 92 76 53 71 67 79 82 3.26 3.15 2.78 2.84 2.85 3.04 2.88 1.88 2.18 2.23 2.01 2.64 2.81 2.82 3.09 2.60 2.89 2.75 3.60 2.75 2.91 2.93 23 22 20 23 18 14 19 10 19 24 781 945 569 636 1,191 1,082 1,152 1,544 987 836 93 90 79 89 89 87 92 92 89 88 75 82 107 213 94 91 76 75 102 92 3.14 2.98 2.72 2.89 2.70 3.03 2.88 3.17 2.94 2.79 3.3 Aquaculture in ghers Gher aquaculture is a year-round ac vity and profitable business for the farmers of southern Bangladesh. The par cipants were given experience in a range of improved management prac ces including feed formula on and fish-rice-vegetable integra on. Fish produc on data were collected from 266 farmers who were engaged in fish cul va on in a total of 37 ha of ghers across 15 upazilas. Produc on cost was highest in Tala sub-district of Sathkhira (319 thousand Tk/ha) compared with the average cost of 244 thousand Tk/ha (Table 7). Morrelgonj sub-district of Bagerhat District showed very good performance in both total income (over 1 million Tk/ha) and net profit (almost 800 thousand Tk/ha). Average total income and net profit were 417 and 173 thousand Tk/ha, respec vely. Average fish produc on of the 15 upazilas was 1,325 kg/ha, while the highest produc on of 3,873 kg/ha was obtained at Morrelgonj. Produc on was lower in areas affected by salinity. 573 Table 7. Produc vity and profitability of fish in ghers during 2013 Sl. no Name of upazila/ sub-district 1 Morrelgonj 2 Fakirhat 3 Mollahat 4 Mongla 5 Rampal 6 Satkhira Sadar 7 Tala 8 Kaligonj 9 Assasuni 10 Shyamnagar 11 Kalaroa 12 Debhata 13 Dacope 14 Koyra 15 Paikgacha Total Average 1 No. of farmers 25 12 14 26 14 17 25 14 12 12 19 14 27 16 19 266 Total water body (bigha)1 Produc on cost (tk/ha x 1000) 26 11 38 17 13 17 23 12 12 11 18 13 32 14 15 273 Fish produc on (kg/ha) 3,873 1,318 3,207 372 2,620 788 707 908 1,072 2,090 2,271 230 143 280 - 1,087 786 975 279 713 240 214 264 356 803 294 24 48 170 - 799 470 684 88 452 31 (105) 38 160 594 78 (263) (160) (79) (200) 3.78 2.49 3.35 1.46 2.74 1.15 0.67 1.17 1.82 3.84 1.36 0.08 0.23 0.68 - 244 1,325 417 173 1.71 b Fig. 4. Successful gher ac vity in (a) Bagerhat and (b) Satkhira Districts during 2013. 574 BCR 288 316 291 191 261 209 319 226 196 209 216 287 208 249 200 1 ha = 7.47 bigha a Income Net profit (tk/ha x (tk/ha x 1000) 1000) 4. Conclusions Increasing produc on, produc vity and income from farm holdings is possible through increased cropping intensity, shi ing from local crop varie es to improved varie es and diversifica on. By scaling out the results obtained from valida on ac vi es, produc on of both agriculture and aquaculture can be doubled. The results of the study will be useful for further research and extension work. An integrated approach focusing on training, block level demonstra ons, farm level advisory support, financial support and field days ensured successful demonstra on and promo on of new knowledge among farmers. Acknowledgements BRAC acknowledges the support of the CGIAR Challenge Program on Water and Food’s Ganges Basin Development Challenge. 575 Agricultural machinery ownership and intensifica on in South Asia: What can we learn from Bangladesh? K.A. Mo aleb and T.J. Krupnik Interna onal Maize and Wheat Improvement Center, Bangladesh k.mo aleb@cgiar.org, t.krupnik@cgiar.org Abstract Use of scale-appropriate farm machinery is an important component of sustainable intensifica on efforts to boost farm produc vity and conserve natural resources. Where appropriate machinery is employed, farmers can benefit through increased returns to labor and some mes to land. Investment in machinery, however, can entail substan al costs. Consequently, rela vely few households can invest in farm machinery in developing countries. But in South Asia farmers from a range of economic groups are able to purchase or ‘rent’ services from machinery owners under a widely established system of service provision. This model has driven agricultural mechaniza on in Bangladesh in par cular. But in order to expand the use of scale-appropriate machinery in areas that are currently underserved it is important to understand what socioeconomic characteris cs dis nguish machinery owners and service providers from the general farming populace. Isola on of these characteris cs and the over-arching condi ons that favor the spread of appropriate plan ng, crop management, and harves ng equipment has important policy and ac vity targe ng relevance for development programs aimed at expanding farmers’ access to produc vity enhancing technologies. Using census data from 814,058 farm households we applied a mul nomial probit model es ma on approach to characterize households that invest in irriga on pumps, post-harvest threshers, and two-wheeled tractor driven power llers. Econometric results indicate that households endowed with more physical assets (e.g., land, livestock and ponds) are more likely to invest in agricultural machinery (all P < 0.001). Basic civil infrastructure, for example the availability of electricity, par cularly for irriga on pumps, the provision of paved and gravel road, and access to loan facili es also significantly and posi vely affect machinery ownership. Our results indicate that policy planners and development projects aimed at increasing farmers’ access to agricultural machinery should consider these prerequisite issues, which may underpin the agricultural service provision economy in Bangladesh. Where prerequisite condi ons are met, policy planners and development projects are most likely to be successful; where they are not, such as in coastal regions – broader preliminary interven ons to assure access to credit and civil infrastructure may be needed to boost agricultural mechaniza on. Keywords: scale-appropriate machinery, irriga on pump, thresher, two-wheeled tractor driven power ller, ownership, infrastructure, sustainable intensifica on 1. Introduc on By 2050 the world’s popula on will reach 9.1 billion, 34% higher than in 2009 (FAO, 2009). The largest propor on of this popula on increase will take place in South Asia and Sub-Saharan Africa. This growth is expected to increase the consump on of staple cereals along with other agricultural products, specifically meat and fish – which rely largely on cereal- and silage-based feeds (Cassman et al. 2003; Godfray et al. 2010; Tilman et al. 2011; Mueller et al. 2012).) To ensure only cereal food security in 2050, growth requirement es mates indicate the need for an addi onal three billion tonnes from 2.1 billion tonnes– a 43% increase over 2009 levels (FAO 2009). However, boos ng aggregate produc on by expanding land area is not a sustainable op on (Miah and Sarma 2000; Pimental and Wilson 2004). This is par cularly the case in densely populated South Asia, where farm size is already small and per capita arable land has been declining steadily (World Bank 2014). In Bangladesh and in India, for example, per capita arable land was 0.16 and 0.34 hectares, respec vely, in 1961, though in 2009 these figures fell to 0.05 and 0.14 hectares respec vely (World Bank 2014). 576 An important avenue to lower produc on costs and poten ally enhance yields is to produce more food with increasing efficiency, thereby raising returns to labor and ideally land (e.g. Sison et al. 1985). The use of scale-appropriate farm machinery has an important role to play in this process. Here we dis nguish scale-appropriate machinery from farm machinery generally, as large-scale machinery commonplace in developed countries may not always be appropriate for the small field sizes and farming enterprises encountered in South Asia (Krupnik et al. 2013). A er replenishing capital investments, use of scale-appropriate machinery can help to further reduce farmers’ produc on costs by replacing arduous manual labor and tradi onal tools with increasingly efficient equipment designed for op mal performance (World Bank 2007; Kienzle et al. 2013). In Bangladesh, machinery appropriate for precise direct seeding, fuel-efficient irriga on, and bed plan ng is becoming more readily available (Krupnik et al. 2013). The ques on arises, however, as to how to facilitate farmers’ access to scale-appropriate machinery in Bangladesh’s and South Asia’s underserved regions. A clear understanding of the characteris cs of farm households that invest in farm machinery and provide services to other smallholders is a prerequisite for development programs and farm policies seeking to boost the use of scale-appropriate machinery. Studies of this kind can bring insight into the factors that facilitate or limit investments in machinery by farm households, thereby helping development planners and policy makers to target investments appropriately. Crucially, there are a wide-range of socioeconomic factors that may limit uptake of farm machinery. Where these factors present formidable barriers, investments and policies that focus on overcoming these barriers may be necessary before ini a ng programs promo ng the commercializa on and adop on of farm machinery. To our knowledge research on this area is scant. In response, this research iden fies the factors that facilitate or limit the adop on of the most common agricultural machinery used in Bangladesh. This includes irriga on pumps, threshers and two-wheeled tractor driven power llers, of which at least 420,000 and poten ally 700,000 are opera ng in Bangladesh, depending on the source cited (iDE 2013; Jus ce and Biggs 2013; Ahmmed 2014). As a case study, Bangladesh is noteworthy because of the rapid pace at which mechanized irriga on, land prepara on, and post-harvest ac vi es have taken place in the last 25 years compared to the rest of South Asia, although development has been geographically heterogeneous (Quayum and Ali 2012; Jus ce and Biggs 2013). Thus, analysis of the condi ons that foster the ownership of machinery and supply of machinery services can provide important lessons for development projects and policy makers – both within Bangladesh but also across South Asia – to help target ac vi es and investments aimed at sustainable intensifica on. Using the BBS (Bangladesh Bureau of Sta s cs) Agricultural Census from 2008, we applied an econometric es ma on approach to determine if farmers endowed with more physical assets (e.g., land, livestock, ponds) are more likely to invest in irriga on pumps, rice and wheat threshers, or two-wheeled, tractor driven power llers. We also examine if and how civil infrastructure (rural electrifica on and the provision of paved and gravel roads) also affects the expansion of mechaniza on through increased household machinery ownership. We begin by describing the growth of agricultural mechaniza on in Bangladesh since the 1970s, focusing mainly on government policy liberalizing the farm machinery sector and underwri ng it with subsidy programs. We next describe the survey data set, a er which we specify the econometric model and present major findings. We conclude with generalized policy recommenda ons. 2. The growth of agricultural mechaniza on in Bangladesh Agricultural mechaniza on has been a historical priority issue for the Government of Bangladesh (GoB). Eighty-two percent of the country’s 153 million people live in rural areas where agriculture is the principal livelihood source. The GoB has also given considerable subsidies to support intensifica on (GoB 1999). The GoB has historically pursued agricultural mechaniza on, specifically for irriga on pumping and land prepara on by two-wheeled, tractor driven power llers, in order to move towards rice self-sufficiency. For example, irriga on pumps were first introduced by the public sector, but later by the private sector following 577 the GoB’s voluntary liberaliza on of the machinery market and relaxa on of machinery import tariffs from 1988-1995 (Hossain 2009). Bangladesh’s green revolu on started with the introduc on of short-dura on, irriga on- and fer liser-responsive semi-dwarf high yielding rice varie es (HYV) in the late 1960s (Mo aleb et al. 2014). Ini ally the GoB ac vely introduced four-wheel tractors (Ahmmed 2014), which are arguably scale-inappropriate in parts of Bangladesh given the small average field size of 0.53 hectares within the country (Hossain et al. 2007). The government also introduced centralized irriga on facili es by establishing deep tube wells (DTWs) and supplied low-li pumps (LLPs) to farmers on a rental basis from the Bangladesh Agricultural Development Corpora on (BADC). The GoB also supplied fuel at a 75% subsidized rate to pump owners through BADC un l the 1970s. By 1978, BADC had managed and rented out a total of 9,000 DTWs and 35,000 LLPs, respec vely (iDE 2012). The management of 44,000 irriga on pumps under nearly complete government control, however, was a large logis cal and financial burden for BADC. Eight years following independence, in 1979, Bangladesh undertook liberaliza on policies and as a result the government gradually changed from state-led support of mechaniza on and started priva zing the irriga on and fer lizer markets. BADC started sales to liquidate DTWs and LLPs (both new and recondi oned), first to farmers coopera ves, and later to individual farmers, many of whom became service providers. The priva za on of irriga on markets only gained momentum a er the removal of a number of tariff and non-tariff barriers on the import of irriga on and diesel engines and two-wheeled, tractor driven power llers. Nearly a decade later, a cyclone with wind speeds of more than 150 kilometers hit Bagerhat, Barguna, Bhola, Jessore, Khulna, Patuakhali, Pirojpur, Satkhira Districts and the Sundarban areas of Bangladesh on November 29, 1988 (UNDRO 1988). The cyclone took a major toll on human lives and also dras cally reduced the number of draught oxen and buffalos used for land prepara on. The deficiency was es mated at approximately 5.8 million draught animals, which was equivalent to 132,000 mechanical llers (GoB 1989). The average na onal draught power energy requirement was 0.373 kW per hectare, while the average available draught power was es mated at 0.271 kW per hectare. The resul ng deficit was about 27% less than the total es mated requirement (GoB 1989). The GoB made a quick decision to facilitate mechanized llage by encouraging imports to provide the draught power needed to prepare land immediately following the cyclone. By this me, Bangladesh had ini ated preliminary liberaliza on policies, although import tariffs were s ll in place for some sectors. The GoB’s Standardized Commi ee of Bangladesh was responsible for the quality of imported machinery, including agricultural equipment. The commi ee mainly prescribed the import of high-cost Japanese tractors, pumps, and engines, but discouraged more affordable Chinese made machinery that they deemed to be of low quality (Jus ce and Biggs 2013). The urgency of the cyclone and threat of food and hence poli cal insecurity prompted the GoB under then President Hussain Muhammad Ershad to take a number of steps. In 1988, most major import taxes on standardized diesel engine and two-wheeled, tractor driven power llers were voluntarily eliminated. The following year, Ershad disbanded the Standards Commi ee, which had ini ally upheld tariff restric ons following the cyclone, to facilitate the rapid import of cheap diesel engines and two-wheeled, tractor driven power llers from China that were more appropriate for the small fields encountered in Bangladesh than four-wheeled tractors, at a frac on of the cost (Jus ce and Biggs 2013). Six years later, the import of two-wheeled, tractor driven power llers and two-wheeled tractors was made completely duty free (iDE 2012). 578 Table 1. Import of agricultural machinery to Bangladesh from 2004 to 2013 Year 2004-05 2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 a Number Two-wheeled tractor Value of other Seeders and transplanters of driven power llersb parts and machinery importers Number Pricea per unit (USD) Number Price per unit (USD) (million USD) 109 54,675 639.77 56 578.95 0.87 99 52,863 729.87 69 1812.13 1.13 77 37,606 733.40 66 563.69 1.89 106 56,460 765.55 173 586.63 4.10 103 55,604 949.93 45 1460.98 5.16 119 44,872 894.90 195 581.38 12.27 113 70,843 991.85 425 667.48 7.13 116 51,266 1105.65 662 483.31 10.39 112 30,771 1135.64 246 419.62 2.54 Price is reported as import value plus 2.01% import tax. b Other parts and machinery include agricultural, hor cultural, or forestry machinery for soil prepara on or cul va on; lawn or sports-ground rollers, excep ng two wheeled tractor driven power llers and seeders, or transplanters. Note: 1US$= BDT 77.2 Source: GoB (2014a). The combined elimina on of the Standards Commi ee and market liberaliza on resulted in a drama c increase in imports of small diesel engines for irriga on pumps and two-wheeled, tractor driven power llers. The number of shallow tube wells (STWs) increased from 93,000 in 1982 to 260,000 in 1990 (iDE 2012). Presently, more than 420,027 two-wheeled, tractor driven power llers—the vast majority of which are made in China—are engaged in preparing 80% of Bangladesh’s cropland (iDE 2012). In Bangladesh, a total of 1.63 million STWs, DTWs and LLPs are currently engaged in irriga ng nearly 55% of all cropland (BBS 2011). During the 1990s use of post-harvest crop threshers and crop management equipment increased significantly. Because the import of selected agricultural machinery has been encouraged through low or negligible tariffs for over a decade, traders in Bangladesh now ac vely import from abroad (Table 1). In 2013, the most recent year for which GoB data is available, 112 importers imported 30,771 two-wheeled, tractor driven power llers with an average price USD 1,135 per piece. In the same year Bangladesh imported USD2.54 million worth of spare parts and other agricultural machinery from abroad. Most imported equipment is for llage alone, though nearly two thousand transplanters and addi onal implements that can be a ached to two-wheeled, tractor driven power llers for direct seeding have been imported since 2004 (GOB 2014a). In most cases the owners of the agricultural machinery also work as service providers by ren ng or selling mechanized land prepara on and plan ng opera on services to other farmers (iDE 2012). As a result, even rela vely small farm households have been able to access rela vely affordable machinery services through custom hiring systems (iDE 2012). However, the characteris cs that determine adop on and ownership of machinery – which is a crucial prerequisite for such service provision models – remain unclear. Using a large na onally representa ve data set, the present study characterizes farm households who invest in agricultural machinery and who provide custom services, while also iden fying the civil infrastructural factors that appear to influence the ownership of agricultural machinery and provision of services in Bangladesh. 579 3. Data descrip on This study relies on two data sets made available by the Bangladesh Bureau of Sta s cs (BBS) housed within the Ministry of Planning, and Local Government Engineering Department (LGED), under the Ministry of Local Government, Rural Development and Coopera ves of Bangladesh. The 2008 Agricultural Census is the fourth such census in Bangladesh and was deployed in the month of May. Prior to 2008, previous censuses were conducted in 1996, 1983-84, and 1977. An addi onal Agriculture Sample Survey was conducted in 2005. In its broadest form, the census sought to iden fy the structure and characteris cs of agricultural pursuits managed by household farms. A total of 28.69 million farm households were surveyed in 2008, of which 25.35 million were from rural areas, the remainder being from urban farms. Of the sampled households 1.73 million were from Barisal, 4.88 million were from Chi agong, 9.46 million were from Dhaka, 3.43 million were from Khulna, 7.66 million were from Rajshahi and Rangpur Divisions, and 1.53 million were from Sylhet Division in the northeast. Data on farm machinery ownership focused on irriga on pumps, two-wheeled, tractor driven power llers, and/or threshing machines. Family wealth measured by land size, pond and livestock ownership, the number of family members and gender of the household head were also documented in the present study. Using the same dataset, we also generated sub-district level informa on on electricity availability for agricultural machinery (mostly irriga on pumps) and the provision of formal bank loans accessed by farm households. Data on the extent of paved or gravel roads at the sub-district level were also collected from the spa al data division of LGED via ins tu onal request. Although the 2008 Census covered a representa ve sample of Bangladesh’s farm households, the BBS provides access to only five percent of the en re census data, including 1,042,595 households in 64 districts in all seven divisions: Barisal, Chi agong, Dhaka, Khulna, Rajshahi, Rangpur and Sylhet. This study sought to understand the characteris cs of rural farm households that invest in agricultural machinery. We therefore considered rural farm households only. As a result, we removed urban farm households from our sample resul ng in 814,058 sampled households from 476 sub-Districts in 64 districts in seven divisions of Bangladesh, each of which had different machinery and other endowments (Table 2). The machinery considered included irriga on pumps, threshers and two-wheeled, tractor driven power llers. Table 2. Informa on on ownership status of the sampled agriculture machinery, land and other resources owned by rural households by division in 2008 Division Number of households % owned of a sampled machine % owned an irriga on pump % owned a thresher % owned a two-wheeled tractor driven power ller Land owned (acres) % household owned a pond Number of cows and buffalos owned Barisal 57,727 0.82 Khulna Dhaka Sylhet Chi agong Rajshahi Rangpur All 100,414 241,069 55,629 147,116 138,855 73,248 814,058 6.52 2.59 1.71 3.37 4.27 2.57 3.32 0.28 0.50 0.21 4.33 3.55 0.62 1.96 0.87 0.30 0.99 0.80 0.35 0.91 58.57 1.06 0.89 25.50 0.99 0.72 13.46 0.75 1.07 0.68 34.58 45.27 1.17 0.64 0.98 2.06 0.14 3.05 2.06 0.70 2.05 0.76 0.38 2.08 1.68 0.38 0.73 8.39 0.88 0.60 7.12 1.06 0.76 23.90 1.78 Note: Irriga on pump includes deep tubewells, shallow tubewells and low-li pumps Source: BBS (2010) 580 4. Model specifica on and es ma on To examine how household-level asset endowments and characteris cs, and sub-district civil level infrastructure affect agricultural machinery ownership at the household level, we formulate and es mate equa on 1: 6 Yi   0  ( HHC i )  ( SUBDIS j ) d 1 d ( DDd )  i (1) where Yi is a vector of dependent variables that includes a base value zero if a household did not own one of the specific sampled agricultural machineries in 2008, an irriga on pump ownership dummy that assumes a value of one if a household owned an irriga on pump, a thresher machine ownership dummy that assumes a value of two if a household owned a thresher machine in 2008 and a two-wheeled, tractor driven power ller ownership dummy that assumes a value of three if a household owned a two-wheeled, tractor driven power ller in 2008. Among the explanatory variables, HHCi is a vector of independent variables that includes a gender dummy that assumes a value of one if a household head is female (zero otherwise), a dummy equal to one if the household is an owner operator who does not work as an agricultural laborer in other farms (zero otherwise), the total number of adult family members who are more than 15 years old and a dummy for farm household land size that assumes a value of one if a household owned more than 1.01 hectares of land (zero otherwise). According to the defini on of the BBS (2011a), small farm households owned up to 1.01 hectares of land, medium farm households owned 1.01-3.03 hectares of land and large farm households owned more than 3.03 hectares of land. To clarify, the farm size dummy that we have included in our model is the dummy for medium and large farm households who owned more than one hectare of land. The explanatory variables also include a pond ownership dummy (one if yes, zero if no) and total number of livestock owned (only cows and buffalos) by a household. In Bangladesh pond ownership can play a major role in genera ng income through the provision of aquacultural resources. SUBDISj is a vector of independent variables that includes sub-district-level informa on on the length of paved or gravel road in kilometres, households that operate at least one agricultural machine - most commonly irriga on pumps - using electricity (%), and the propor on of households who received formal loans from banks or formal credit agencies. To understand whether sub-district level infrastructure has different effects in terms of machinery adop on on medium and large farm households versus small farm households we also included three mul plica ve dummies in which we mul plied the medium and large farm household dummy with the length of paved or gravel road at the sub-district level, the propor on of electricity connected to agricultural machinery and the propor on of credit and loan services at the sub-district level. DDij represents six division dummies for seven divisions, where Chi agong division is the base; is the scalar parameter; , and are the vectors of parameters; i stands for household; j stands for the sub-district, where the household is located; and is the random error term. Note that the most common agricultural machine owned by farm households is an irriga on pump; however, a household can also own a thresher machine, or a two-wheeled, tractor driven power ller (Table 2). Considering this issue, to es mate mul ple machinery ownership, we applied a mul nomial probit model, where we assigned value zero as the base if a sampled farmer does not own any of the sampled machines, and a value of one if a farmer owned an irriga on pump, two if a farmer owned a thresher machine, and three if a farmer owned a two-wheeled, tractor driven power ller. Zero was applied if the answer to any of these points was no. The applica on of a mul nomial non-linear model in explaining farmers’ adop on of different agricultural technologies is not new. For example, Mo aleb et al. (2014) also applied a mul nomial logit model in es ma ng use of hybrid and inbred rice seed by Bangladeshi farmers. Quayam and Ali (2012) also applied single equa on logit model in es ma ng the adop on of two-wheeled, tractor driven power llers in Bangladesh. 5. Results On average a sampled household was endowed with 0.31 hectares of land, nearly two cows and water buffalos, and 24% owned a pond (Table 2). A total of 3.3% of the sampled farm households owned at least one of the machines of interest including irriga on pumps, threshers, and/or two-wheeled, tractor driven 581 power llers (Table 2). Of the 3.3%, 2.1% owned an irriga on machine, 1.7% owned a thresher and 0.38% owned a two-wheeled, tractor driven power ller. Because of the pervasiveness of the service provision system in Bangladesh, most of these owners not only serve themselves but also serve other clients farmers on a custom hire basis. This model has been successful in genera ng a large number of service providers providing access to irriga on services and two-, and more recently, four-wheel tractors in select areas, in addi on to post-harvest threshing and shucking services which can be found in Bangladesh (iDE 2012; Quayum and Ali 2012; Jus ce and Biggs 2013). Farm households in Barisal division, however, are less likely to own the sampled agricultural machinery when compared to the na onal average and also to the households in other divisions (Table 2). Households in Khulna, Rajshahi and Chi agong Divisions, conversely, are more likely to own the sampled agricultural machinery than the households in other divisions. In Barisal, 0.82% households owned one or more piece of agricultural machinery. Of this group, 0.28% owned an irriga on pump, 0.50% owned a thresher, and 0.21% owned a two-wheeled, tractor driven power ller. The cropping intensity in Barisal is typically not high (MoA and FAO 2012), owing partly to the difficulty of establishing a dry season crop where dal flooding and salinity are concerns, and also due to farmers’ lack of investment incen ves where produc on risk is perceived (Mo aleb et al. 2013; MoA and FAO 2012). Consequently, only a few farm households invest in agricultural machinery or intensified cropping in this region, for which a logical star ng point to boost cropping intensity would be to put into place measures to lower farmers’ produc on risks as part-and-parcel of any mechaniza on efforts. Crop insurance schemes, stress resistant varie es and agronomic prac ces could be useful in this regard. Of the total dataset, less than 4% of the households were female headed and nearly 65% of all householders are owners and operators of the machinery in ques on (Table 3). More than 90% are small farm holders who were endowed with less than one hectare of land. On average, a sampled household was endowed with more than three family members of which more than one was an adult who was older than 15 years (Table 3). Table 3. Basic demographic and other characteris cs of the sampled households by division in 2008 Division % female headed household % household is an owner operator who does not work as an agriculture laborer Total members in the family No. of adult family members (>15 years old) % small farm households (<1.01hectare) % medium farm households (1.01-3.03 hectares) % large farm households (>3.03 hectares) Barisal 2.66 72.59 Khulna 2.70 58.38 Dhaka 3.39 67.65 Chi agong Sylhet Rajshahi Rangpur All 5.46 4.81 2.44 3.44 3.57 70.64 68.24 60.05 50.42 64.59 3.27 1.67 2.81 1.54 3.09 1.57 3.78 1.81 3.99 1.80 90.70 89.82 92.53 93.74 8.02 8.91 6.60 1.27 1.27 0.86 2.81 1.49 2.76 1.38 3.18 1.60 88.05 91.88 93.49 91.95 9.58 5.52 7.03 5.37 6.96 2.36 0.74 1.09 1.13 1.09 Source: BBS (2010) On average the sampled sub-districts were equipped with 164 kilometres of paved or gravel road. Nearly 62% of the sampled households operate at least one agricultural machine using electricity, primarily for irriga on pumps that require connec ons for deep and shallow tube well extrac on in northern Bangladesh. Nearly 21% of the sampled households received loans or credit (Table 4). Sub-district civil infrastructure such as roads and the availability of electrical connec ons for pumping could affect the decision to purchase 582 agricultural machineries by impac ng overall opera on and transac on costs. For example, the availability of electricity might inspire a household to purchase a motor pump for irriga on, as subsidized electricity can be used as an inexpensive and easy to use source of energy compared to diesel, which requires pre-season purchases of fuel and entails transac on costs for transpor ng fuel from the pumping sta on. Similarly, the extent of paved or gravel roads at the sub-district level could affect the decision to purchase a two-wheeled, tractor driven power ller at the household level, as the extent of paved or gravel roads can expand the chance of mul ple uses for a two-wheeled, tractor driven power ller (e.g. human or goods hauling). Table 4. Basic sub-district level informa on in 2008 Division Length of paved or gravel roads at the sub-district level (00 km)a % household opera ng at least one sampled agriculture machine by electricity b % household who received formal bank loans at the sub-district levelb Barisal Chi agong Dhaka 1.91 1.62 1.48 Khulna 2.42 Rajshahi Rangpur Sylhet 1.66 1.16 1.37 All 1.64 13.67 55.27 88.54 86.86 56.35 60.22 23.40 61.97 28.03 13.15 19.07 28.46 23.94 26.18 12.05 20.67 Sources: a GOB (2014a) and bBBS (2010) 0 2000 4000 6000 8000 10000 To depict the distribu on of physical assets among households with the sampled machines and households without any of the sampled machines, we developed quan tle-quan le plots illustra ng the distribu on of land and livestock among the sampled machine owner and non-owner households (Figures 1 and 2, respec vely). In the graph, if the resul ng distribu on points lie roughly on a line with the sample slope then it indicates that land and livestock are evenly distributed between two groups of households. However, Figures 1 and 2 clearly show that households owning the sampled agricultural machinery also propor onally own more land (Figure 1) and livestock (cows and buffalos) than those who do not (Figure 2). Both are a logical indica on of wealth status of the household. 0 2000 4000 6000 8000 10000 Households with a sampled machine Figure 1. Distribu on of land owned by the sampled rural households by ownership of agricultural machinery including irriga on pumps, threshers, or two-wheeled, tractor driven power llers (source: BBS 2010). 583 100 50 0 0 20 40 60 Households with a sampled machine Figure 2. Distribu on of cows and buffalos owned by the sampled rural households by ownership of agri cultural machinery including irriga on pumps, threshers, or two-wheeled, tractor driven power llers (source: BBS 2010). Table 5 presents the es mated func on explaining sampled agricultural machinery ownership by the sampled households in Bangladesh. All of the household level variables considered in this study were highly sta s cally significant with expected signs in explaining ownership of irriga on pump, thresher, and power ller ownership at the household level. The dummies for agriculture owner operator household, the number of adult family members, the pond ownership dummy, the livestock number, and the dummy for the medium and large farm household size posi vely and significantly influence ownership of the sampled agricultural machinery. The marginal effects model shows that on average, an owner-operator of agricultural machinery tends to have a higher probability of having a sampled machine, ranging from between P < 0.001 at the lowest to P = 0.004 at the highest. Conversely, the female-headed household dummy variable was nega ve and highly sta s cally significant across the es mated func ons for each type of machinery ownership (P<0.001). This reflects conven on in Bangladesh that women headed households are less likely to own produc ve agricultural assets than male-headed households (Table 5). The lack of machinery ownership among women is an important finding for policy planners and development organiza ons concerned with increasing gender equity and even development in Bangladesh, as solu ons are clearly needed to boost women’s access to produc ve assets. While Table 5 shows posi ve asser on between farm size, and livestock number and the adop on of sampled agricultural machinery, a farmer may have logically used a purchased machine to derive income through service provision to invest in more on livestock and/or land a er purchase, rather than before. Without longitudinal data, it is difficult to dis nguish between such cause and effect in the current study, and follow up studies with a representa ve sub-set of the popula on are therefore needed. The es mated func on and corresponding marginal effects nonetheless show strong correla ons among physical assets and machinery ownership by a farm household. The medium and large farm household dummy (owned more than one hectare of land =1) is posi ve and highly sta s cally significant in explaining ownership of all three machines inves gated. Medium and large farm households tend to have a 1 to 3% higher probability of having a sampled agricultural machine compared to a small farm household, reflec ng the more resource endowed nature of larger households which tend in this case to have ownership over more produc ve agricultural assets than smaller households. This is an important point for development and policy planners that focus on extending use of scale-appropriate machinery to small households. Where projects seek to facilitate the commercial purchase of equipment, even at somewhat subsidized rates, our data suggest that poorer 584 households may s ll not be willing to invest. As such, programs focused on machinery should seek to extend access to machinery use by small farm households through service provision models whereby those who do own machinery can sell use of the machine to those unable to purchase through custom hiring processes. Experience has shown that this is the main mechanism by which farmers at large are able to access use of irriga on pumps, threshers, and two-wheeled, tractor driven power llers in South Asia (Jus ce and Biggs 2013), underscoring that not all farmers must own a machine to mul ply farmers’ use of mechanized services. Among the sub-district-level variables, the length of paved or gravel roads, the percentage of agricultural households that operate at least one agricultural machine (usually irriga on pumps) using electricity, and the propor on of households who have received financial loans significantly and posi vely affect the ownership of irriga on pumps (P values ranging between 0.05 to 0.001), thresher (P<0.001) and two-wheeled, tractor driven power llers (P<0.001) (Table 5). Thus where such services and civil infrastructure exist, systems of machinery service provision are more likely to develop. An improved transporta on system can contribute to agricultural machinery adop on by reducing transac ons costs. For example the transporta on costs associated with use of bringing inputs, and post-harvest products to and from the farm, par cularly where the two-wheeled tractors that are used to drive two-wheeled, tractor driven power llers are used to haul materials in a achable flatbed trailers prior to or a er land prepara on is completed (Jus ce and Biggs 2013). 585 Table 5. Es mated func ons applying a mul nomial probit model explaining ownership of pumps, threshers and two-wheeled, tractor driven power llers by farm households in Bangladesh Es mated func on Es mated func on Irriga on pump Thresher Two wheeled tractor driven power ller Female-headed household dummy (yes=1) -0.25*** (-6.38) -0.29*** (-7.83) -0.30*** (-3.95) -0.003*** (-7.34) -0.01*** (9.43) -0.001*** (-4.77) A dummy for household is an owner operator who does not work as an agriculture laborer 0.22*** (17.47) 0.13*** (10.32) 0.19*** (8.78) 0.004*** (17.75) 0.002*** (9.52) 0.001*** (8.23) No. of adult family members (older than 15 years) 0.11*** (21.16) 0.14*** (21.29) 0.14*** (15.51) 0.002*** (20.34) 0.002*** (21.18) 0.001*** (14.54) Pond ownership dummy (yes=1) 0.34*** (25.52) 0.48*** (39.42) 0.37*** (17.51) 0.006*** (20.08) 0.009*** (30.68) 0.002*** (12.55) No. of cows and buffalos 0.10*** (33.10) 0.11*** (32.78) 0.09*** (30.00) 0.002*** (31.88) 0.002*** (31.79) 0.003*** (23.56) Medium and large farm household dummy (>2.49 acres) yes=1 0.84*** (18.97) 0.40*** (9.07) 0.86*** (12.70) 0.03*** (11.52) 0.01*** (5.82) 0.01*** (6.54) Length of paved or gravel road at the sub-district level (KM) 0.0002** (2.20) 0.001*** (5.99) 0.001*** (3.11) 0.000002* (1.85) 0.00001*** (5.83) 0.000001** (2.87) % Households using electricity to run agricultural machinery at the sub-district level 0.18*** (23.82) 0.36*** (52.44) 0.18*** (13.06) 0.003*** (21.13) 0.006*** (49.30) 0.001*** (10.89) Propor on of households receiving formal bank loans at the sub-district level 0.35*** (3.90) 0.53*** (5.59) 1.19*** (7.43) 0.01*** (3.42) 0.008*** (5.22) 0.005*** (7.21) Length of paved or gravel roads at the sub-district level X Medium and large farm household dummy -0.001*** (-3.18) 0.001*** (4.87) -0.0001 (-0.32) -0.00001*** (-3.46) 0.00001*** (5.09) -0.0000003 (-0.39) Propor on of household receiving formal bank loan at the sub-district level X Medium and large farm household dummy 0.531*** (3.49) 0.68*** (4.53) 0.59*** (2.81) 0.008*** (3.24) 0.01*** (4.33) 0.002** (2.45) 0.01 (0.29) 0.13*** (8.54) 0.11*** (4.91) -.0001 (-0.22) 0.002*** (8.43) 0.001*** (4.50) -0.89*** (-16.77) -1.13*** (-29.06) -0.31*** (-5.07) -0.01*** (-34.0) -0.01*** (-59.52) -0.01*** (-4.65) Name of the machine % households using electricity to run agricultural machinery at the sub-district level X Medium and large farm household dummy Barisal division dummy 586 Irriga on pump Thresher Two wheeled tractor driven power ller Table 5 cont. Marginal effects (dy/dx) Es mated func on Irriga on pump Thresher Two wheeled tractor driven power ller Khulna division dummy 0.68*** (26.40) 0.002 (0.07) 0.44*** (10.01) 0.02*** (18.17) -0.001** (-2.68) 0.002*** (6.60) Dhaka division dummy 0.40*** (19.91) -0.66*** (-36.66) 0.22*** (5.71) 0.01*** (18.43) -0.01*** (-44.45) 0.001*** (5.38) Sylhet division dummy -0.16*** (-4.65) -0.72*** (-23.25) 0.17*** (3.38) -0.002*** (-4.49) -0.007*** (-40.81) 0.001*** (3.54) Rajshahi division dummy 0.60*** (26.59) -0.09*** (-4.66) 0.70*** (18.03) 0.02*** (19.62) -0.002*** (-8.46) 0.01*** (11.58) Rangpur division dummy 0.46*** (17.18) -0.60*** (-20.65) 0.36*** (7.31) 0.01*** (13.47) -0.007*** (-34.84) 0.002*** (5.46) Constant -4.39*** (-90.21) -4.08*** (-92.49) -5.41*** (-57.96) N 814058 Name of the machine Wald chi2(54) Prob > chi2 Log pseudo likelihood Irriga on pump Thresher Two wheeled tractor driven power ller 31267.29 0.00 -124123.71 Note: Numbers in parentheses are z-sta s cs based on standard error that allow for intragroup correla on at the household level. *Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1% level The prevalence of paved roads and highways can also enhance the flow of informa on, as farmers can easily visit government extension offices in the ci es, discuss and exchange with each other in markets, and agricultural extension workers can easily travel to and from the office to the farm. In addi on, an improved transporta on system can allow households to use agricultural machinery for alterna ve income genera ng ac vi es – as noted above, many two- and also four-wheeled tractor owners rent their machines for hauling materials before and a er land prepara on is done, thereby extending the use of their machines for more days of the calendar year. Similar to our findings, Mo aleb et al. (2014) also depicted the significantly posi ve influence of sub-district level infrastructure such as paved or gravel roads, on the adop on of agricultural technology such as hybrid rice. They conclude that an improved transporta on system can contribute to adop on of modern agricultural technology in two ways. First, the transporta on costs of carrying seeds, inputs and products to and from the farm and nearest markets and ci es tend to be lower in the presence of improved roads and highways. It makes investment in agricultural technology more profitable. Second, roads and highways also enhance the flow of useful informa on, as farmers can easily visit government extension offices in the ci es and agricultural extension workers can easily travel to and from the office to the farm. Furthermore, similar to our findings, Quayum and Ali (2012) also found that credit availability can significantly and posi vely affect the adop on of two-wheeled, tractor driven power llers in Bangladesh. The mul plica ve dummy in which we mul plied medium and large farm household dummy variables with sub-district level infrastructure variables further elucidates how sub-district level civil infrastructure may affect differently the adop on of agricultural machinery by households of different farm sizes. The es mated 587 func on and the corresponding marginal effects show that the length of paved or gravel road nega vely and significantly affects the adop on of irriga on machines by the medium and large farm households, but posi vely and significantly affects the adop on of a threshers machine and has no influence on the adop on of a two-wheeled, tractor driven power ller. By contrast, households’ access to formal loans posi vely and significantly affects the ownership of all of the sampled machinery (P<0.001) by the medium and large households compared to the small ones (Table 5). Access to electricity to run agricultural machinery, however, posi vely and significantly (P<0.001) affects the adop on of thresher machines and two-wheeled, tractor driven power llers by medium and large farm households compared to small farm households. The extent of adop on of agricultural machinery in Barisal and Sylhet Divisions is low compared to Chi agong division (which is the base division in our model; Table 5). The challenges in the en re Barisal division in terms of agriculture are diverse and extreme, include poverty, and households have a low investment capacity and are averse to risk. Out-migra on by agricultural laborers is also common (MoA and FAO 2012), which one might expect to posi vely affect agricultural machinery ownership, although it appears that farmers’ sense of risk and low investment capacity, in addi on to the civil infrastructural variables noted above, limit uptake of machinery. As such, adop on of machinery is compara vely low in Barisal division. Adop on is also limited in Chi agong division, though this is not necessarily surprising given that much of this region is hilly in nature, with very low groundwater tables, and remains under considerable forest cover. As such, agricultural intensifica on – and the poten al for machinery – remains limited. Our data suggests that GoB and donor agencies may wish to consider special programs to encourage scale-appropriate mechaniza on in Barisal, where farmers could be encouraged to use irriga on and rapid land prepara on services to move from single to double cropping (MoA and FAO 2012) in order to spur the intensifica on of the agricultural sector. However, we underscore that adequate a en on must also be paid to basic civil infrastructural and to the mi ga on of investment risk – which could be par ally ameliorated through low interest and low risk credit programs – that are likely needed in advance of efforts to develop strong agricultural machinery value chains in areas like Barisal division. 6. Conclusion and policy implica ons Scale-appropriate agricultural mechaniza on can play an important role in enhancing labor produc vity in rapidly developing agricultural economies. In select cases – for example, where farmers adopt irriga on or precision sowing and harves ng equipment is used—farmers may experience increased returns from their investment in land. Development and policy planners can benefit from informa on on the socioeconomic and civil infrastructural condi ons that are prerequisite for rural households to purchase farm machinery, and make use of them through service provision models, thereby benefi ng farmers more widely. To our knowledge, there is rela vely li le recent research on these issues, and no previous empirical studies employ sufficiently large scale data sets to examine household characteris cs as they correlate with machinery ownership in developing na ons. To fill this crucial knowledge gap we iden fied some of the factors that appear to facilitate or limit the ownership and adop on of common agricultural machinery in Bangladesh, including irriga on pumps, threshers, and two-wheeled, tractor driven power llers. Our results indicate that farm size and wealth endowment of the farm households significantly and posi vely affect the ownership of agricultural machinery at the household level. Households endowed with more physical asset such as land, livestock, and ponds are more likely to invest in agricultural machinery, and are thereby more likely to serve smaller-scale farmers by providing custom hiring services to help repay investments in machinery and to generate addi onal income. We also demonstrated that sub-district level civil infrastructure such as the availability of electricity (specifically for irriga on pumps), and the provision of paved and gravel roads, in addi on to the availability of loan and credit facili es, also significantly and posi vely affects the ownership of agricultural machinery at the household level. These findings underscore that the provision of basic infrastructure and loan facili es in Bangladesh’s rural areas is prerequisite to agricultural machinery ownership by farm households, and therefore to the development of an agricultural 588 machinery service provision economy. This provision economy is integral to insuring that smaller and less wealthy households can benefit from access to machinery through custom hiring on an as-need basis. As such, government, donors, and development planners should place equal focus on facilita ng these necessary precondi ons before and/or in tandem with efforts encouraging the adop on of appropriate agricultural machinery in areas where uptake is currently limited. Acknowledgements This study was made possible through the support provided by the United States Agency for Interna onal Development (USAID) to the Cereal Systems Ini a ve for South Asia – Mechaniza on and Irriga on (CSISA-MI) project. The contents and opinions expressed herein are those of the author(s) and do not necessarily reflect the views of the US Agency for Interna onal Development or the United States Government and shall not be used for adver sing or product endorsement purposes. References Ahmmed, S. 2014. Present Status, Prospects and Challenges of Mechaniza on in Bangladesh. In Md. Ayum Hossain, Nazmun Nahar Karim, Shoed Hassan, and Sultan Ahmed in Use of Farm Machinery and Efficient Irriga on System Management Training Manual 2014. Gazipur: Farm Machinery and Postharvest Process Engineering Division, Bangladesh Agricultural Research Ins tute (BARI). BBS (Bangladesh Bureau of Sta s cs). 2010. Census of Agriculture 2008. Dhaka: Ministry of Planning. BBS (Bangladesh Bureau of Sta s cs). 2011. Bangladesh Bureau of Sta s cs 2010. Dhaka: Ministry of Planning. BBS (Bangladesh Bureau of Sta s cs). 2011a. Census of Agriculture 2008: Zila Series Patuakhali. Dhaka: Ministry of Planning. Cassman, K. G., A.Dobermann, D. T. Walters, and H. Yang. 2003. Mee ng Cereal Demand While Protec ng Natural Resources and Improving Environmental Quality. Annual Review of Environment and Resources, Vol. 28: 315-358. FAO (Food and Agriculture Organiza on of the United Na ons). 2009. How to Feed the World 2050. Rome: FAO. Online: h p://www.fao.org/fileadmin/templates/wsfs/docs/Issues_papers/HLEF2050_Global_Agr iculture.pdf GOB (Government of Bangladesh). 1999. Na onal Agriculture Policy. Dhaka: Ministry of Agriculture. Online: h p://www.dae.gov.bd/Dae_Policy/NAP.pdf, accessed September 09, 2014. GOB (Government of Bangladesh). 2014a. Automated System for Customs Data. Dhaka: Na onal Board of Revenue (NBR). GOB (Government of Bangladesh). 1989. Mid Term Review of the Third Five Year Plan 1985-90. Dhaka: Planning Commission, Ministry of Planning: 84. GOB (Government of Bangladesh). 2014. GIS Unit. Local Government Engineering Department: Dhaka: Ministry of Local Government, Rural Development and Coopera ves. Godfray, H. C. J., J. R. Beddington, I. R. Crute, L. Haddad, D. Lawrence, J. F. Muir, J. Pre y, S. Robinson, S. M. Thomas, C. Toulmin. 2010. Food Security: The Challenge of Feeding 9 Billion People. Science, Vol. 327: 812-818. 589 Hossain, Mahabub. 2009. The impact of shallow tubewells and boro rice on food security in Bangladesh. Washington D.C.: Interna onal Food Policy Research Ins tute (IFPRI). Online: h p://www.ifpri.org/sites/default/files/publica ons/ifpridp00917.pdf, accessed December 22, 2014 Hossain, M., Lewis, D., Bose, M.L., & Chowdhury, A. (2007). Rice research, technological progress, and poverty: The Bangladesh case. In Adato, M. and Meinzen-Dick, R. (Eds.). Agricultural research, livelihoods and poverty: Studies of economic and social impacts in six countries. Bal more: The Johns Hopkins University Press and Interna onal Food Policy Research Ins tute. iDE (Interna onal Development Enterprises). 2012. Commercializa on of Selected Agriculture Machineries in Bangladesh. Dhaka: Interna onal Development Enterprises (iDE). Online: h p://repository.cimmyt.org/xmlui/bitstream/handle/10883/3394/98527.pdf, accessed December 15, 2014. Jus ce, S. and S. Biggs. 2013. “Rural and Agricultural Mechaniza on in Bangladesh and Nepal: Status, Processes and Outcomes. In Mechaniza on for Rural Development: A Review of Pa erns and Progress from Around the World, eds. Josef Kienzle, John E. Ashburner, and Brian G. Sims, 67-98. Rome: Plan Produc on and Protec on Division, Food and Agriculture Organiza on of the United Na ons (FAO). Kienzle, J., J.E. Ashburner, and B.G. Sims 2013. Mechaniza on for Rural Development: A Review of Pa erns and Progress from Around the World. Rome: Plan Produc on and Protec on Division, Food and Agriculture Organiza on of the United Na ons (FAO). Krupnik, Timothy J., San ago Santo Valle, Andrew J. McDonald, Sco Jus ce, Israil Hossain and Mahesh K. Gathala.2013. Made in Bangladesh: Scale-Appropriate Machinery for Agricultural Resource Conserva on. Mexico, D.F.: Interna onal Maize and Wheat Improvement Centre (CIMMYT). Miah, M.A.H. and N.P. Sarma (2000). Hybrid rice: progress, issues and outlook for Bangladesh. Interna onal Rice Commission Newsle er 49, 65–72. MOA and FAO, 2012. Master Plan for Agricultural Development in the Southern Region of Bangladesh. Ministry of Agriculture (MoA, Government of Bangladesh) and United Na ons Food and Agriculture Organiza on, Dhaka, Bangladesh, p. 122. Mo aleb, K. A., S. Mohanty and A. Nelson. 2014. “Factors influencing hybrid rice adop on: a Bangladesh case”. Australian Journal of Agricultural and Resource Economics, Vol. 58:1–17. doi: 10.1111/1467-8489.12060. Mo aleb, K. A., S. Mohanty, H. T. K. Hoang and R. M. Rejesus. 2013. The Effects of Natural Disasters on Farm Household Income and Expenditures: A Study on Rice Farmers in Bangladesh.” Agricultural Systems 121: 43–52. Mueller, N. D.,M. Johnston, D. K. Ray, N. Ramanku y, J. A.Foley. 2012. Closing yield gaps through nutrient and water management. Nature, Vol. 490: 254-257. Pimentel, David, and Anne Wilson. 2004. World Popula on, Agriculture and Malnutri on. In Popula on and Its Discontents, World Watch Magazine, Vol.17, No.5: 22-25. Online: h p://www.worldwatch.org/system/files/EP175D.pdf, accessed November 26, 2014. Quayum, M. A. and A. M. Ali. 2012. Adop on and Diffusion of Two wheeled tractor driven power llers in Bangladesh. Bangladesh Journal of Agricultural Research, Vol 37, No 2: 307-325. Online: h p://www.banglajol.info/bd/index.php/BJAR/ar cle/view/11234/8209, accessed December 15, 2014. Sison, J.F., R.W. Herdt and D. Duff. 1985. “The Effects of Small Farm Mechaniza on on Employment and Output in Selected Rice-Growing Areas in Nueva Ecija, Philippines.” Journal of Philippine Development, Vol. XII, No., 1: 29-82. 590 Tilman, D., C. Balzer, J. Hill, and B. L. Befort. 2011. Global food demand and the sustainable intensifica on of agriculture. Proceedings of the Na onal Academy of Science of the United States of America, Vol. 108(50): 20260–20264. UNDRO (United Na ons Disaster Relief Organiza on). Bangladesh - Cyclone Nov 1988 UNDRO Informa on Reports 1-6. Online: h p://reliefweb.int/report/bangladesh/bangladeshcyclone-nov-1988-undro-informa on-reports-1-6, accessed on October 02, 2014. World Bank. 2007. Agriculture for Development, World Development Report 2008. Washington DC: The Interna onal Bank for Reconstruc on and Development / The World Bank. World Bank. 2014. World Development Indicators 2014: Arable land (hectares per person). Washington DC: World Bank. Online: h p://databank.worldbank.org/data/views/variableSelec on/selectvariables.aspx?source= world-development-indicators#, accessed September 09, 2014. 591 592 Conference Attendees Shelina Afroza, MoFL Mahnaz Afsar Rayhan Afsar Zafar Ahmed, MoWR Md. Masud Ahmed, Blue Gold Zia Ahmed, CIMMYT Nazneen Ahmed, WorldFish Bulbul Ahmed, BARI Masum Ahmed, Bangladesh Betar Sultan Ahmmed, BARC Mirja Kaizer Ahmmed, WorldFish Md. Aktaruzzaman, BWDB Hamida Lipi Akter, WorldFish Shanjida Akter, WorldFish Afroza Akter, WorldFish Md. Jahangir Alam, MoWR Saiful Alam, WARPO Md. Jahangir Alam, BSMRAU Md. Shariful Alam, MoPA Mohammad Alamgir, WARPO A.K.M. Abdullah Al-Amin, BAU Mir Emdad Ali, ATN Olivyn Angeles, IRRI Gaur Pada Bachar, Coastal Zone Community Mustafa Bakuluzzaman, Shushilan Randy Barker, Cornell/IRRI Kanteswar Barman, CCRIP, LGED Benoy Barman, WorldFish Alak Barman, SO Tahmina Begum, DAE Shawkat Begum , CIP Jayanta Bha acharya, PSTU-IRRI Md. Mokter Hossain Bhuiyan, BARI Shankar Biswas, BRAC Ja sh C. Biswas, BRRI Panchanan Biswas, Coastal Zone Community Mihir Biswas, Coastal Zone Community Altaf Boya , Coastal Zone Community Marie-Charlo e Buisson, IWMI Dhiman Burman, CSSRI Canning Town Binay Kumar Chakraborty, DOF Parvesh Kr Chandna, IRRI Giasuddin Choudhury, Delta Plan Afrina Choudhury, WorldFish Mary Archana Costa, IRRI Arijit Das, IWMI Uzzal Das, Coastal Zone Community Sumona Das, Blue Gold Hasan Dastagir, New Age Bijoy Bhusan Debnath, WorldFish Camelia Dewan, University of London Nepal Dey, BRAC Golam Faruque, WorldFish Paul Fox, IRRI Mamun Gazi, Coastal Zone Community Pamela George, WorldFish T.K. Ghoshal, CIBA ASM Golam Hafeez, BAU Suklal Halder, Coastal Zone Community AKM Fazlul Haque, MoFL Joadder Md. Anowarul Haque, DoF Sadika Haque, BAU Alifa Haque, NSU Md. Enamul Haque, iDE Minam T. Haque, WorldFish Rezaul Hasan, IWM Rokibul Hasan, LGED Mehedy Hasan, ETV Md. A Hashem Abdul Haque Hawlader, Coastal Zone Community Md. Nazmul Hoq, BAU A.K.M. Ariful Hoque, BARI AFM Afzal Hossain, IWM Md. Abul Hossain, DAE Md. Aowlad Hossain, Blue Gold Md. Golger Hossain, DoF Md. Khaled Hossain, CIMMYT Md. Mokarram Hossain, WorldFish Md. Monowar Hossain, IWM Md. Rifatul Hossain, DAE Moqbul Hossain, SRDI Mosharraf Hossain, BARI Shahadat Hossain Md. Kawsar Hossin 593 Md. Houssain, BWDB Shamsul Huda Liz Humphries, IRRI Mirza Islam, BINA Azizul Islam, Coastal Zone Community Mahmudul Islam, Sylhet Agricultural University Md. Kamrul Islam, WorldFish Md. Nazrul Islam, SO Md. Saidul Islam, Shushilan Md Shafiqul Islam, Coastal Zone Community Monirul Islam, BARI Rafiqul Islam, UNB Saiful Islam, Coastal Zone Community Wais Kabir, FAO Md. Pervez Kabir, KU E. Kabir, Bangladesh Betar Kevin T. Kamp, WorldFish S M Kamruzzaman, Coastal Zone Community Manjurul Karim, World Fish Swarna Kazi, World Bank Nandish Kenia, IWMI Deepak Khadka, iDE Md. Khaleduzzaman, Dutch Embassy Sarafat Hossain Khan,CEIP, BWDB Zahir ul Haque Khan, IWM Abu Saleh Khan, IWM Afroza Khanam, MoWR Md. Mamun Khondhoker, Coastal Zone Community Md. Humayun Kobir, Coastal Zone Community Mir Kobir, BTV Tim Krupnick, CIMMYT Swapon Bhadra, BRAC A. La f, SB Ed Barre Lennard, University of Western Australia Syed Mahbub, Naya Diganta Barrister Anisul Islam Mahmud, MoWR Paulceip K. Maih, IWM Mohammed Mainuddin Shourav Maitra Dr. Buddheswar Maji, ICAR-CSSRI Dr. Subhasis Mandal, CSSRI Canning Town M. Maniruzzaman, BRRI S.M. Abdul Mannan, FAO Craig Meisner, WorldFish Tofazzal Hossain Miah, BAU 594 Golok Mistri, Coastal Zone Community Dilip Kr Mistri, Coastal Zone Community Archisman Mitra, IWMI Khaled Mohammed, DUET Haridas Mohuta, BARI Manoranjan Mondal, IRRI Monoranjan Mondal, Coastal Zone Community Dipu Moni, Coastal Zone Community Neaz Morshed, Maasranga TV Abdul Wahab Munshi, Coastal Zone Community M.G. Mustafa Golam Mustaffa, WorldFish Nitasha Nair, IWMI M. Niamul Naser, DU Farhat Naz, IWMI Syed Nicolas, IFAD Ainun Nishat, BRAC University Andrew Noble, WLE (IWMI) Kasia Paprocki, Cornell University Dr. Sanjida Parveen, Sylhet Agricultural University Md. Shamim Parvez, WorldFish Lu a Parvin, World Fish Mike Phillips, WorldFish Ilse Pukinskis, WLE (IWMI) Tausif Ahmed Qurashi, BRAC Asiqur Rahaman M. A. Rahim, BAU Ataur Rahman Azbina Rahman, BUET Akhlasur Rahman, IRRI Mahabubur Rahman, WorldFish Md. Hadishur Rahman BAU-IRRI Md Mahfuzur Rahman, BWDB Md. Nazibur Rahman, BWDB Md. Shahidur Rahman, BWDB Mohammad Saidur Rahman, BAU Moshiur Rahman, LGED Musa Nurur Rahman, BWDB Mustafizur Rahman, Coastal Zone Community Nefaur Rahman, BAU-IRRI Narendra Kumar Rai, Ashok Sansthan Sohel Rana, LGED M. Abdur Rahaman, Rana M. Harunur Rashid, BRRI M A Rashid, BRRI Mike Robson, FAO Azit Roy, Coastal Zone Community Ruma Roy, BAU Steven Rubinyi, World Bank Timothy Russell, IRRI Jayon Saddar, Coastal Zone Community Bidhan Sadhak, WorldFish Nathan Sage, Office of Economic Growth, USAID Nibir Saha, PSTU-IRRI Jiban Krishna Saha, GIZ PK Saha, BRRI Panchali Saikia, IWMI Ibrahim Saiyed Md. Abdus Salek, Coastal Zone Community Popy Rani Sannyashi, BAU Sukanta Sarange, ICAR-CSSRI Khokon K Sarker, BARI Khaled Sarker, Prothom Alo Md. Ruhul Amin Sarker Gazi Abdus Sa ar, Coastal Zone Community Urs Schulthess, CIMMYT S.M Shah-Newaz, IWM Md. Bappy Shahriar, WorldFish Dirk Smits, Blue Gold Is ak Sobhan, Environment, Water Resources and Climate Change (SASDI) South Asia Sustainable Development Department Eyerin Sultana, WorldFish Nazneen Sultana, LGED Mehnaz Tabassum, WorldFish T P Tiwari, CIMMYT Mahmud Hasan Tuhin T P Tuong, IRRI Md. Ekram Ullah, Environment, Forests & Fisheries Sec on Md. Hedayet Ullah, WorldFish Samina Yasmin, CIMMYT 595 596 CPWF Ganges Partner Organizations Bangladesh Agriculture University Bangladesh Fisheries Research Ins tute BRAC Bangladesh Rice Research Ins tute Bangladesh University of Engineering and Technology Ins tute of Water and Flood Management Bangladesh Water Development Board Indian Council of Agricultural Research—Central Ins tute for Brackish-Water AquacultureKakdwip Research Centre Indian Council of Agricultural Research—Central Soil Salinity Research Ins tute—Regiona Research Sta on Canning Town Bangladesh Water Development Board 1970 597 Interna onal Rice Research Ins tute Interna onal Water Management Ins tute Local Government Engineering Department Patuakhali Science and Technology University Shushilan Socioconsult Ltd. SR D I RESOURCE DEVELOPMENT Soil Resource Development Ins tute 598 WorldFish Contact WorldFish, Bangladesh Office House 22B, Road 7, Block F, Banani Dhaka 1213, Bangladesh T: (+880-2) 881 3250 (+880-2) 881 4624 F: (+880-2) 881 1151 E: worldfish-bangladesh@cgiar.org May 2015