CIAT Research Online - Accepted Manuscript Decision support tools for site-specific fertilizer recommendations and agricultural planning in selected countries in sub-Sahara Africa The International Center for Tropical Agriculture (CIAT) believes that open access contributes to its mission of reducing hunger and poverty, and improving human nutrition in the tropics through research aimed at increasing the eco-efficiency of agriculture. CIAT is committed to creating and sharing knowledge and information openly and globally. We do this through collaborative research as well as through the open sharing of our data, tools, and publications. Citation: MacCarthy, Dilys S.; Kihara, Job; Masikati, Patricia; Adiku, Samuel G. K.. 2017. Decision support tools for site-specific fertilizer recommendations and agricultural planning in selected countries in sub-Sahara Africa. Nutrient Cycling in Agroecosystems 1-17 p. Publisher’s DOI: https://doi.org/10.1007/s10705-017-9877-3 Access through CIAT Research Online: http://hdl.handle.net/10568/88013 Terms: © 2017. CIAT has provided you with this accepted manuscript in line with CIAT’s open access policy and in accordance with the Publisher’s policy on self-archiving. This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. You may re-use or share this manuscript as long as you acknowledge the authors by citing the version of the record listed above. You may not use this manuscript for commercial purposes. For more information, please contact CIAT Library at CIAT-Library@cgiar.org. Nutrient Cycling in Agroecosystems Decision support tools for site-specific fertilizer recommendations and agricultural planning in selected countries in Sub-Sahara Africa --Manuscript Draft-- Manuscript Number: FRES-D-16-00348R2 Full Title: Decision support tools for site-specific fertilizer recommendations and agricultural planning in selected countries in Sub-Sahara Africa Article Type: S.I. : Bationo Fertilizer Recommendation West Africa Keywords: Risk management; Resource use efficiency; Sub Sahara Africa; Soil productivity Corresponding Author: Dilys Sefakor MacCarthy, Ph. D. University of Ghana Accra, GHANA Corresponding Author Secondary Information: Corresponding Author's Institution: University of Ghana Corresponding Author's Secondary Institution: First Author: Dilys Sefakor MacCarthy, Ph. D. First Author Secondary Information: Order of Authors: Dilys Sefakor MacCarthy, Ph. D. Job Kihara Patricia Masikati Samuel Adiku Order of Authors Secondary Information: Funding Information: Abstract: Recommendations and decisions of crop management in sub-Saharan Africa (SSA) are often based on traditional field experimentation. This usually ignores the variability of production factors in space and time, variability that itself invalidates such decisions and recommendations outside of the experimental sites. Yet, the use of alternative or complementary decision support approaches such as crop modelling is limited. In this paper, we reviewed the state of the use of crop modelling in informing site specific fertilizer recommendations in some countries in SSA. Even though nitrogen fertilizer recommendations in most countries across Africa are blanket, the limited employment of models show that optimum nitrogen application should be differentiated according to soil types, management and climate. A number of studies reported on increased fertilizer use efficiency and reduced crop production risks with the use of Decision Support Tools (DST). The review also showed that the gross limitation of the use of models as agricultural decision-making tools in SSA could be attributed to factors such as low capacity due to limited training opportunities, and the general lack of support from national governments for model development and application for policy formulation. Proposals identified to overcome these limitations include (i) introduction of the science of DST in the curricula at the tertiary level, (ii) encouragement and support for the adoption of model use by Governmental and Non-Governmental Organizations as additional tools for decision making and (iii) simplifying DSTs to facilitate their use by non-scientific audience to scale uptake and use for farm management. Response to Reviewers: The comments of reviewer 1 and Editor Figures have been re-drawn and contents modified to include the equations, the p and r2 values as requested. Also, all figures are now redone in black and white Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation Comments of reviewer 2 Original line 409 “Special collaboration” has been taken care of in lines 398 to 401 Also the comment on phosphorous deficiencies, weed management e.t.c. have now been addressed in lines 404 to 406 Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation 1 Decision support tools for site-specific fertilizer recommendations and agricultural planning in selected 1 countries in Sub-Sahara Africa. 2 3 Dilys S. MacCarthy1, Job Kihara2, Patricia Masikati3, Samuel G. K. Adiku4 4 1Soil and Irrigation Research Centre, University of Ghana, Kpong, Ghana 5 2International Center for Tropical Agriculture (CIAT), Box 823-00621 Nairobi, Kenya 6 3World Agroforestry Centre, (ICRAF) Lusaka, Zambia 7 4Department of Soil Science, University of Ghana, Legon, Accra, Ghana 8 9 *Corresponding Author: Dr. Dilys S MacCarthy: dsmaccarthy@gmail.com 10 11 Abstract 12 Recommendations and decisions of crop management in sub-Saharan Africa (SSA) are often based 13 on traditional field experimentation. This usually ignores the variability of production factors in 14 space and time, variability that itself invalidates such decisions and recommendations outside of 15 the experimental sites. Yet, the use of alternative or complementary decision support approaches 16 such as crop modelling is limited. In this paper, we reviewed the state of the use of crop modelling 17 in informing site specific fertilizer recommendations in some countries in SSA. Even though 18 nitrogen fertilizer recommendations in most countries across Africa are blanket, the limited 19 employment of models show that optimum nitrogen application should be differentiated according 20 Manuscript Click here to download Manuscript Manuscript.docx Click here to view linked References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 2 to soil types, management and climate. A number of studies reported on increased fertilizer use 21 efficiency and reduced crop production risks with the use of Decision Support Tools (DST). The 22 review also showed that the gross limitation of the use of models as agricultural decision-making 23 tools in SSA could be attributed to factors such as low capacity due to limited training 24 opportunities, and the general lack of support from national governments for model development 25 and application for policy formulation. Proposals identified to overcome these limitations include 26 (i) introduction of the science of DST in the curricula at the tertiary level, (ii) encouragement and 27 support for the adoption of model use by Governmental and Non-Governmental Organizations as 28 additional tools for decision making and (iii) simplifying DSTs to facilitate their use by non-29 scientific audience to scale uptake and use for farm management. 30 31 Key words: Risk management; Resource use efficiency; Sub Sahara Africa; Soil productivity 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 3 Introduction 33 Agriculture, the mainstay of the economies in sub-Saharan Africa (SSA), is dominated by 34 smallholder farmers, holding often between 0.5-2 ha and relying mainly on rainfall (Adiku et al., 35 2015). The soils in the region are generally highly weathered (Sanchez, 2002), comprising of Low 36 Activity Clays (LAC) with low inherent fertility (cation exchange capacity CEC between 3 and 15 37 cmolc/kg soil). In some regions such as the West African Sudano-Sahel, the CEC can be as low as 38 1 cmolc/kg soil and hence a great portion of the inherent fertility is derived from the soil organic 39 carbon, which itself is low, often, < 10 g/kg (Bationo and Buekert, 2001). These, in conjunction 40 with poor management practices such as bush burning, residue removal from fields, very low 41 fertilizer application, mono cropping systems and erratic but intense rainfall lead to accelerated 42 soil degradation and fertility decline. Even then, the use of inorganic fertilizer in SSA is low, being 43 only about 10 kg/ha fertilizer a decade ago (Sanchez et al., 2009) although current evidence suggest 44 that several countries have now increased use. For example, current fertilizer use by farmers in 45 Ghana is about 30 kg N/ha (MacCarthy et al., 2017). 46 47 It has long been established that increasing the use of inorganic fertilizer on arable land is critical 48 to improving crop productivity and ending hunger in SSA (van Keulen and Breman 1990). But 49 this must go along avoiding the low fertilizer use recoveries under high application rates and high 50 rainfall conditions (Vanlauwe et al. 2011) associated with large losses in runoff or leaching. In 51 other words, efforts towards increasing food production should also include ways to improve 52 efficiency of fertilizer use. In 2003, the heads of states of African countries re-pledged to allocate 53 10% of their annual budget and to attain 6% growth in agriculture by 2015 (CAADP, 2003), with 54 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 4 an enhanced fertilizer use at the core of the strategy. Yet, despite the pockets of increased fertilizer 55 use, the situation has still not changed very much from the observations by Sanchez et al. (2009). 56 57 The low application of fertilizers in agriculture in SSA can be attributed to several challenges. 58 First, there is the socio-economic aspects of low incomes of most farmers, and hence their inability 59 to afford fertilizers. This aspect will not be discussed here. From the biophysical point of view, 60 blanket fertilizer recommendations which have been the general approach in many SSA countries 61 have little scientific rigour. For example in Ghana, the fertilizer recommendations for both 62 sorghum and maize are similar and in Zimbabwe recommendations have been done for most crops 63 grown by both commercial and smallholder farmers across the five agro-ecological zones (FAO, 64 2006). The failure to formulate fertilizer recommendations that are soil- and crop-type specific and 65 that also considers the effect of climate variability results in either wastage or deficiencies in 66 fertilizer use. In sum, current fertilizer recommendation practices in the SSA do not properly 67 address the specific local biophysical agricultural production systems, hence making them 68 unprofitable in several instances (Kihara et al., 2015), and a disincentive for smallholder farmers. 69 70 Improving the formulation of fertilizer recommendations in the SSA is hampered by the expensive 71 and time-consuming field experimentation and soil analysis approaches that are logistically too 72 expensive to conduct at every location of interest. . The results are low adoption rates as the field- 73 and soil analysis-based methods alone do not capture the possible range of yield variabilities that 74 can be associated with a given fertilizer application rate and, in many cases , variable weather. The 75 need for the use of complementary procedures that can more effectively assess the many possible 76 interactive effects of biophysical attributes and management practices including soil and crop 77 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 5 types, varieties, fertilizer types, application rates and timing on crop productivity under varying 78 weather, cannot be overemphasized. Typically, these are known as Decision Support Tools (DST) 79 or crop modelling. The purpose of this paper is to provide a historical review of the use of models 80 as DSTs in SSA, and to understand reasons limiting the wide-scale use of these models for 81 agricultural research and development planning and especially for formulating site-specific 82 fertilizer requirements. 83 84 Globally Available Decision Support Tools (DSTs) 85 Decision support tools range from empirical static models that enable the assessment of soil 86 nutrient concentrations and identify limiting productivity, to dynamic software support that 87 combine soils, crop-specific growth parameters and weather. Empirical and static models date 88 back to 1930s (Akponikpe et al, 2014) when a number of nutrient response functions were derived 89 often for single factors (e.g. rainfall, fertilizer, among others) to predict crop response to nutrient 90 application. Indeed, as early as 1913, Mitscherlich derived simple, easy to follow equations to 91 predict crop response to nitrogen application (Mitscherlich, 1913), the foundations of which 92 continue to play roles in agronomic research and advice. A suite of such empirical response 93 functions led to development of a set of improved response models that consider multiple soil 94 nutrients such as QUEFTS (Jansen et al., 1990), the effects of soil acidity on crop productivity e.g. 95 NuMAS (Maran and Leatherman, 1992) and the effects of soil organic matter management on soil 96 productivity and crop performance, e.g. NUTMON (Stoorvogel and Smaling, 1990). The major 97 limitation of these types of models is the lack of dynamic response to changing management and 98 climate. Their use for future predictions is thus limited. 99 100 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 6 The foundation for the dynamic crop models was laid in the 1950s by de Wit (1958) and van Bavel 101 (1953) (see Jones et al., 2016). These types of models, popularly referred to as “Models of 102 Agricultural Systems” combine physical and biological principles to model agricultural systems. 103 Such models, including APSIM, DSSAT and more recently SEAMLESS, harnessed the strengths 104 of non-system models such as EPIC (Willams, 1983), CENTURY (Parton et al., 1987), NTRM 105 (Shaffer et al., 1983), PARCH model (Hess et al., 1997), STICS (Brisson et al., 1998) and 106 PERFECT (Littleboy et al., 1989) in dealing with soil resources under long-term farming activities, 107 but also recognized their weakness in addressing important systems aspect of cropping such as 108 residue management, crop rotation and dynamic management decisions that are responsive to 109 weather, soil and genotype and hence, affect crop yield (Keating et al., 2003). These model 110 development efforts and applications have occurred in other places such as Australia, America and 111 Europe. Even though model uptake worldwide for agricultural planning beyond the research 112 community has been generally low (Rose et al., 2016), there are indeed efforts and success stories 113 where models have been used in the broader agricultural planning context by farmers, communities 114 and monitors. The FARMSCAPE model (Carberry et al., 2002) provides a proof of one such case 115 in northern Australia. It provides a workable interface between researchers, farmers, communities, 116 among others, enabling model application beyond researchers use. Another DST that is used by 117 farmers and consultants in Australia is the “Yield Prophet” which provides growers with integrated 118 production risk advice and monitoring decision support relevant to farm management. The 119 Monsanto Seed Company employs models to assess the greenhouse gas emission reduction 120 potentials of crops such as maize and soybean under varying soil conditions. Thus, in several 121 respects, some efforts have and continue to be made in modest to popularize the use of models in 122 many ways. In SSA, however, model use is mainly limited to largely donor-funded calibration and 123 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 7 validation studies within the research domain. The more crucial aspect of model development to 124 address the peculiar challenges such as soil acidity, phosphorus fixation, soil salinity, among 125 others, on crop production and the adoption of the models by National Governments to assist policy 126 formulation is almost completely under-funded. 127 128 Though crop modelling in the world spans more than 60 years or more, it was not until the mid-129 1980s that both empirical and functional dynamic models were introduced to SSA. Perhaps the 130 earliest model use in the SSA was in South Africa in the early 1970s (Schultze, 1975), followed 131 by a rather slow spread to the other regions. Empirical and the semi-empirical models such as 132 AQUACROP (Raes et al., 2009), CROPSYST (Stockle et al., 2003), STICS, WOFOST (Van 133 Diepen et al.,, 1989), QUEFT and NUTMON took precedence over the more dynamic ones that 134 simulated the dynamics of the crop growth, development and soil processes. By the mid-1980s, 135 the first application of functional dynamic crop-soil systems model in a developing SSA country 136 was probably in Kenya, within the Australia Dry-land Farming Systems Project (McCown et al., 137 1992; Keating et al., 1991) that spanned 1985 to 1992. This formed the foundation of modeling 138 low input systems with the use of the CERES Maize model and then evolved into the use of the 139 Agricultural Production Systems Simulator (APSIM) (McCown et al., 1992). Other decision 140 support tools in use in SSA include WOFOST (Kassie et al.,. 2015) used to assess the impact of 141 the variability of weather parameter on the yield of maize in Ethiopia and SARA-H, a water 142 balance/stress index based model used mainly in the Sahelian regions of West Africa and that has 143 been used extensively for agrometeorological and food security assessments (Sultana et al., 2005; 144 Akponikpe et al., 2014). 145 146 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 8 Despite efforts by Consortium of International Agriculture Research Centres (CGIAR) (e.g. 147 ICRISAT, CIAT and IITA) and IFDC among others to promote DST using software such as 148 Decision Support System for Technology Transfer (DSSAT; Hoogenboom et al., 2010) and 149 APSIM, most of the users from SSA are from the research domain and not from the policy makers’ 150 domain. In effect, the needs for the types of interface suitable for the non-research community 151 have not been expressed to the model developers. Also, SSA can hardly showcase any model 152 development works except the South African sugar cane model and some limited work to extend 153 some models such as APSIM to include intercropping systems (Adiku, 1995; Adiku et al., 1998). 154 155 Challenges to fertilizer recommendation formulation in the SSA 156 Soil and crop-specific nutrient management recommendations are required to increase farm 157 productivity. The challenge of providing these recommendations to farmers in Africa is huge 158 because soils and climate are highly heterogeneous even over short distances. Local soil variability 159 also results in variability in yields even among replicates of the same treatment (Akponikpe et al., 160 2014). Crop productivity and profitability of fertilizer use vary widely in space and time even on 161 the same soil, particularly under rain-fed agriculture (MacCathy et al., 2015; Naab et al., 2015). 162 Some other studies in the Savannah region of West Africa also point to differences in the use 163 efficiencies of applied N fertilizer as a result of differences in the land use history of the fields 164 (MacCarthy et al., 2010). 165 166 It was noted earlier that several fertilizer recommendations in SSA do not consider variations in 167 local settings but are rather uniform in space and in time. Furthermore, research sites on which the 168 recommendations are based are sometimes higher in fertility due to better management and 169 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 9 residual nutrients from previous trials thus, making them unsuitable as basis for the larger 170 recommendations. Wopereis et al. (2006) observed in the West African Savannah that maize 171 response to fertilizer application was affected by the mineral fertilizer management of maize on 172 farmers’ fields as well as inherent soil organic matter. The crop response to fertilizer is also 173 strongly affected by weather variability. With little or no ability to forecast the weather, investment 174 in fertilizer can lead to farmer indebtedness, a phenomenon that serves as a disincentive for the 175 adoption of innovative practices that enhances intensification (Hansen, 2005). Several other 176 studies have reported the weather dependence of crop response to fertilizer use and the subsequent 177 inter-seasonal yield variations (MacCarthy et al., 2009; MacCarthy et al., 2015; Naab et al., 2015; 178 Akponikpe et al., 2010). 179 180 The response to mineral fertilization is also dependent on the crop and on the variety of crop being 181 used (Haefele et al., 2010). Improved crop varieties which are often used in these fertilizer trials 182 are more responsive than the traditional varieties that most farmers use with the former being less 183 resilient to local weather and disease conditions. Soil physical properties such as texture also 184 influence the response of crops to fertilizer application (Zingore et al., 2007). A large spatial 185 variability in yields can occur on a seemingly uniformly-textured soil over short distances 186 (Voortman et al. 2004), posing a challenge to interpretation and potentially point to other 187 interacting factors. The variation of soil physical, chemical and other properties in space, 188 particularly in smallholder systems, due to previous variations in soil fertility management imply 189 that the responses to mineral fertilization would also vary largely in space. The practice of 190 precision agriculture to address such challenges is yet to get a foothold in the SSA. 191 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 10 Thus, to adequately consider the above-mentioned factors in determining fertilizer 192 recommendations for farmers will require some form of decision support tools that take these 193 factors into account in determining crop yield. Decision support tools provide the opportunity to 194 assess the impact of fluctuations in weather parameters on the inter-annual variability on fertilizer 195 use efficiency of crops. It also allows for the assessment of the impact of different management 196 practices on soil properties and processes as well yield. If the SSA is to meet its aim of increasing 197 its fertilizer use by 2050 (CAADP), then the reliance of field experimental procedures alone cannot 198 provide the necessary policy foundation. 199 Role of decision support in SSA 200 The use of DSTs specifically for fertilizer recommendation formulation in SSA is limited. Several 201 studies, however applied the tools in various ways. Smaling and Fresco (1993) used the NUTMON 202 as a decision support tool to monitor the effects of changing land use, and suggest interventions 203 that improve the nutrient balance in Kisii district of Kenya. They concluded that DST has the 204 potential to inform decision makers in determining the effects of current and alternate land use 205 types on crop productivity and long-term sustainability of cropping systems. De Jager et al. (1998) 206 also used the same model in Kenya and concluded that cash crops such as tea and coffee yielded 207 higher economic benefits to farmers and considerably mined less soil nutrient than food crops such 208 as maize and maize-beans systems. Haefele et al. (2003) applied QUEFTS as a DST to study the 209 internal nutrient efficiencies, fertilizer recovery rates and indigenous nutrient supply of irrigated 210 lowland rice in Sahelian West Africa. Similarly, Wopereis et al. (2003) utilized RIDEV-phenology 211 model in the Sahel to develop a DST for determining appropriate time for cultivating rice to avoid 212 yield lose due to increased temperature. Other studies also calibrated and evaluated DSSAT and 213 APSIM for sorghum, millet and maize-based cropping systems on which fertilizer 214 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 11 recommendations could be made (MacCarthy et al., 2010; Akponikpe et al., 2010; Fosu et al., 215 2012; MacCarthy et al., 2012; Fosu-Mensah et al., 2012). 216 217 In the case of functional dynamic crop models, their use has largely remained on the calibration 218 and validation for specific locations in the SSA. For many years in the past, most publications on 219 crop modelling from SSA focused on model calibration (Mabhaudhi et al., 2014; Fatondji et al., 220 2012; Fosu et al., 2012; MacCarthy et al., 2012; Dzotsi et al., 2010) (Table 1). Zinyengere et al. 221 (2015) tested the usefulness of crop models (DSSAT) under data limited dryland conditions of 222 southern Africa using both experimental trial data and district-wide crop yield estimates. Also, 223 Mabhaudhi et al. (2014) calibrated and evaluated AQUACROP for the taro plant in South Africa. 224 Not all calibration attempts were successful; For example, Fosu et al. (2012) explained the failure 225 to predict appropriately yields at high N level (unlike the good predictions at low N) to water stress 226 in the gravelly and shallow soils at the experimental site. Gungula et al. (2003) reported on the 227 inability of the CERES Maize model to predict maize phenology under nitrogen stress condition. 228 Wafula (1995) applied CERES-Maize model to support farmers’ decision making with respect to 229 farm management options and the inherent economic implications. The Agricultural Production 230 System sIMulator was applied by Masikati et al. (2014) to show the positive effect of maize 231 mucuna rotation on water productivity in smallholder systems in Zimbabwe. A few studies have 232 recently used crop models for yield gap analysis (van Ittersum et al., 2013; Kassie et al., 2014). A 233 study by Diarisso et al. (2015) in Burkina Faso indicated substantial yield gaps in the smallholder 234 systems which they attributed to low soil fertility, sub-optimal fertilizer input and erratic rainfall 235 condition. Kassie et al. (2014) also applied the DSSAT and the WOFOST DSTs to assess climate-236 induced yield variability and yield gap of maize in the Central Rift Valley of Ethiopia. Dzotsi et 237 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 12 al. 2003 also used the DSSAT model to provide a DST that enabled optimum cultivar-sowing date 238 combination of maize in southern Togo. 239 240 Link between DST and site specific fertilizer recommendation 241 Decision support tools integrate a multiple of parameters known to affect response of crops to 242 inorganic N such as rainfall distribution, type of soil, crop type and crop variety in simulating crop 243 yield. As such, DST is an appropriate tool to enhance farmer decision making especially with 244 regards to site specific fertilizer recommendation. With the use of DST, it can be shown that a 245 wide range of yields can occur even at a given N application rate across soil types, under variable 246 management, or even at same location but under different weather conditions. In Ghana for 247 example, a farmer investing in 120 kg N/ha application rate can obtain yields varying from 1900 248 kg/ha to more than 4000 kg/ha (Fig. 1). This variation can be attributed to rainfall variability. 249 Without the use of DSTs, such yield/fertilizer response information would require many years of 250 field experimentation to obtain. DSTs can be used together with weather forecast for instance to 251 select appropriate sowing time (MacCarthy et al., 2017) or advise on range of fertilizer to use based 252 on the forecast in order to maximize fertilizer use. 253 254 Recently, Nureeden (2014) used the DSSAT – CSM to refine fertilizer recommendations in Sudan 255 Savannah agro-ecological zone in Ghana. Atakora et al. (2014) also used the DSSAT – CSM to 256 determine fertilizer recommendations for a site in the Guinea Savannah Zone of Ghana. A 257 comparison of these two studies which were both located in the northern part of Ghana show 258 differences in recommended N rates that should be applied to maize to optimize yield. These were 259 all applied on point scale just like most other model applications in SSA. Using the N response 260 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 13 data (Fig. 1) for Tamale, Ghana, a strategic analysis of the monetary returns of the various N inputs 261 showed 60 kg N ha-1 as most appropriate to be recommended to farmers since the returns from that 262 were similar to those obtained from N application levels beyond 60 kg N ha-1 (Fig. 2). The 263 economic optimum rate was determined using Gini coefficient (Adnan et al., 2017) which 264 determines the best economic strategy. Environmental limitations combined with management and 265 socio-economic conditions also need to be considered when assessing cost benefit for fertilizer 266 recommendations. For example, at optimal simulated fertilizer application of 60 kg/ha in soil with 267 average % SOC 0.6, 0.8 and 0.5 and annual rainfall of 850, 1200 and 650 mm median maize yield 268 was 5200, 3216 and 2780 kg/ha for Malawi, Mozambique and Zimbabwe, respectively (Fig. 3 a-269 c). Risk is higher in Zimbabwe at the recommended application rate as shown by high variability 270 of both maize grain and stover yields. While 60 kg N/ha is recommended for Zimbabwe, 271 production at that fertilizer rate gives yields that are 20% less than area potential, i.e., due to soil 272 quality, optimal benefits of applying recommended rates can be compromised. In Senegal for 273 instance, yield increases of between 1000 – 2300 kg/ha and profitability of USD 216 – 640 per ha 274 were reported as benefit from using Nutrient Manager for Rice (NMR) decision support systems 275 for irrigated rice (Saito et al., 2015). A simple Microsoft excel decision support tool has been 276 developed in Uganda to help optimize fertilizer use by farmers and about 400 extension workers 277 and farmers trained on their use. This was part of the Optimizing Fertilizer Recommendation in 278 Africa (OFRA) which is a project being done in 7 countries in SSA and is expected to optimize 279 fertilizer use efficiency. The FERRIZ model was also calibrated and evaluated by Segda et al. 280 (2005) and used to improve fertilizer recommendations for irrigated rice in Burkina Faso. These 281 alternative fertilizer recommendations increased the gross returns compared to farmers' practices 282 and existing recommendations. 283 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 14 284 The shape of simulated response of maize to different levels of N fertilizer vary with soil’s water 285 holding capacity as observed in Koutiala, Mali (Fig. 4). While grain yield seemed to have peaked 286 at 120 kg N ha-1 on soil water holding capacity (WHC) of 50 mm, the response curve for soil with 287 a higher WHC (55 mm) suggested further grain yield increase beyond 120 kg N ha-1. Similarly, 288 the response of crops to N fertilization is also influenced by time of planting (Fig. 5). While the 289 use of 120 kg N/ha can result in median yield of about 4000kg/ha with early planting, using same 290 amount of fertilizer in the late planting window produced a median yield of less than 3000 kg N/ha. 291 Decision support tools can also be used to explore what management options to use to minimize 292 yield losses to enhance farmer confidence in fertilizer adoption. Thus, the need to promote site 293 specific fertilizer recommendation to optimize returns on input cannot be over- emphasized. 294 295 Models as DST for future climate 296 Climate change is a major threat to agricultural productivity in the SSA, especially because of (i) 297 high dependence of people and their livelihoods on natural resources, (ii) the rapid degradation of 298 these resources and resilience loss, (iii) extreme poverty and (i) lack of interventions such as crop 299 insurance. The lingering question is how SSA agriculture will be impacted by future climate. This 300 question cannot be addressed without the use of models. Several projections have been put forward 301 based on different models. IFPRI, for example, simulated changes in crop productivity relative to 302 current yield over several countries in Africa. Others reporting impacts of climate change on 303 agriculture productivity include Jones and Thornton (2003) and Thornton et al. (2009). The work 304 of Thornton et al. (2009) in East Africa highlighted the spatial variability of crop response to 305 climate change and, hence, discouraged the use of spatially contiguous developmental domains in 306 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 15 the identification and implementation of adaption options. Areas where yield decline is predicted 307 at current practices are also shown to have yield increases when technological changes, including 308 increased use of fertilizer and varietal improvement, are considered. 309 Traditionally, DST for future predictions were applied in a variety of ways. In some studies, point 310 based scenarios with single General Circulation Models (GCM) were used, whereas others used 311 point simulation but with multiple GCM (Tachie-Obeng et al., 2013). The trend is now towards 312 the use of multi-locations as well as multi-GCMs (Adiku et al., 2015; Masikati et al., 2015; Rao et 313 al., 2015; Beleste et al., 2015). Within the Agriculture Model Improvement and Inter-comparison 314 Project (AgMIP) framework (Rosensweig et al., 2013), a combination of biophysical and socio-315 economic models is being used as DST to assess the impact of climate change on agriculture in 316 various zones of the world. For the West African region, the work is summarized in “Climate 317 Change Impact on West Africa Agriculture: A Regional Assessment” (Adiku et al., 2015). The 318 results showed that net farm income would reduce under climate change. In East Africa, the project 319 focuses on the “Impacts of climate variability and change on Agricultural Systems in East Africa”. 320 The results (Rao et al., 2015; 2012, Kaissie et al., 2015) indicated that the impact of climate change 321 is not uniform across locations, and that some areas will actually benefit from climate change 322 impacts. Hence the impact on the livelihoods of farmers will also vary based on their location. In 323 other studies, it was projected that the production of maize under climate change scenarios in the 324 Bethlehem District, South Africa would reduce by between 10 and 16% if no adaptation measures 325 are employed (Beletse et al., 2015). In the case of Nkayi, Zimbabwe, the impact of climate change 326 on the productivity of crops under current farmer practice was reported to be marginal (7%). The 327 level of impact is low because the current production systems are low input characterized by 328 depleted soils (Masikati et al., 2015). 329 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 16 330 Limitations and challenges to DST application in SSA 331 In spite of the evidence provided on the improvement in fertilizer use efficiency and reduction in 332 production risks with the use of DST and modelling to inform agricultural management and 333 planning, the use of DSTs to inform decision making is generally poor. This phenomenon is not 334 peculiar to SSA alone. A recent study by Rose et al. (2016) reported of low uptake of DSTs for 335 agricultural decision making in the United Kingdom. The lag in model use as tool for agricultural 336 decision making in Africa may be attributed to several reasons. First, capacity for modelling use 337 is and continues to be grossly lacking. A survey by Adiku (unpublished) on modelling-related 338 publications from the SSA showed that by the year 2009, about 25, 15, 18 and 14 papers were 339 published using DSSAT, APSIM, NUTMON and RUSLE/USLE, respectively. These papers, 340 which emanated from collaborative works between advanced country researchers and SSA 341 counterparts, appeared in reputable journals over a period of about 40 years. On the average, about 342 two modelling papers or so are published annually from the region, with respect to these four 343 models. Against the backdrop of the low capacity, the African Network for Soil Biology and 344 Fertility (AfNet) and their collaborators organized a series of training that culminated in the 345 publication of a book (Kihara et al., 2012). 346 Second, except for donor-funded projects, national support for crop modeling research and 347 application for agriculture development is limited. Over the past 20 years of crop modeling 348 activities within Ghana’s Universities and Research Institutes, for example, direct government 349 funding is negligible. The funding support may appear to be somewhat better in Kenya and 350 southern Africa, but generally not comparable to Europe, Australia, USA, among others. 351 Therefore, as noted, the effect of many peculiar soil challenges of the SSA including soil acidity, 352 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 17 phosphorous deficiency, Mn and Al toxicity, soil erosion and degradation, soil crusts that affect 353 germination and emergence, among others, on crop yields cannot be simulated using the popular 354 DSTs because these processes are not well represented in the models. As a result of the current 355 models lack of sensitivity to these issues, their use in such situations would be limited. Apart, not 356 many institutions in the SSA train expertise in crop modelling and DSTs. Researchers interested 357 in crop modeling must seek training in advanced countries. Interest in modelling among the mainly 358 biology-based students in agricultural sciences in SSA is low, especially because of the need for 359 good mathematical background for modelling. As far back as 1997, the Department of Soil Science 360 at the University of Ghana introduced a curriculum in agricultural systems simulation and 361 modelling. To date, not more than 20 students have participated in the course and not more than 5 362 crop-modelling related thesis have been produced. There is no effort by SSA governments to 363 financially support training in crop modelling. As indicated earlier, there is low capacity in the use 364 of DST even among scientists. Skills on the use of decision support tools are still rare in Sub-365 Saharan Africa (Segda et al., 2005) 366 Third, data unavailability at suitable detail for model validation in particular under broader farm 367 conditions continues to be a major handicap to model use. This requires the need for more research 368 for new versions to include functions that can use routinely collected parameters to estimate those 369 currently required. This will enhance their applicability. The emergence of technologies such as 370 soil-scanners based on IR may be a game-changer for providing extra soil data for areas were data 371 are lacking, particularly with large scale applications. Some efforts have been made to establish 372 minimum data sets and also develop protocols to facilitate the use of DST by other potential users 373 (Hoogenboom et al., 2012; Rosenzweig et al., 2013). 374 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 18 Fourth, the lack of knowledge of the usefulness of DST among agricultural stakeholders for policy 375 formulation is a major handicap. Most DSTs require hardware and computational time and these 376 are often not readily available to potential users in SSA. Organizations that introduce the use of 377 DSTs in SSA often promote specifically those of interest to them whiles smallholder farmers 378 challenges are complex hence require a set of DSTs (DST Toolbox) to adequately address their 379 problems. Critical crops that contribute to food security such as cassava and yam in SSA are 380 usually not adequately captured in most decision support tools. There is also the need to improve 381 use of DST for spatial analysis as most of the existing ones are point based. This will require that 382 they are coupled with geo-spatial tools. Such capabilities already exist in models such as APSIM 383 and DSSAT (Huth et al., 2003) but have not yet been widely applied. 384 385 Conclusions and the way forward 386 Sub-Saharan Africa lags in the use of decision support tools for agricultural decision support even 387 though it is increasingly used in developed countries to support agricultural planning. A great deal 388 of modelling work in SSA has been limited to calibration and validation. Where models were 389 applied to support decision making process, they were hardly used to inform site specific fertilizer 390 recommendation. Inability to capture in models the SSA-peculiar yield limiting factors such as 391 aluminum toxicity, phosphorous deficiency, weeds, and deficiencies of micronutrients limits the 392 application of most of the current models both in representing the real situations and also in making 393 recommendations. The application of models as DST for formulating fertilizer recommendations 394 in the SSA requires much more funding and capacity building support, especially from the national 395 governments and regional bodies in SSA. In sum, for DST to become effective tools for 396 agricultural planning, the following must be achieved: 397 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 19 (i) Capacity building: The introduction of the use of DST in tertiary school curriculum, 398 with a focus on the training especially the next generation not only in model use but 399 more importantly model development. In particular, support from the mathematical 400 disciplines to biological sciences will be required. The setting up of special funds to 401 support students willing to engage in modelling work would be important. 402 (ii) Demonstration of the utility of DSTs beyond research to policy formulation domain 403 (iii) Address peculiar tropical soil and cropping system challenges such as phosphorus 404 deficiency, aluminum toxicity, soil acidity, weed competition, mixed cropping among 405 others to enhance their applicability in SSA. 406 (iv) Development of DST for other important food crops such as cassava and yam. 407 408 409 410 411 References 412 Adiku SGK, MacCarthy DS, Haithie I, Diancoumba M, Freduah BS, Amikuzuno J, Traore PCS, 413 Traore S, Koomson E, Agali A, Lizaso JI, Fatondji D, Adams M, Tigana L, Diarra DZ, 414 N’diaye O (2015) Climate Change Impacts on West African Agriculture: An Integrated 415 Regional Assessment. In: Rosenzweig C, Hillel D (eds) Handbook of Climate Change and 416 Agroecosystems: The Agricultural Model Intercomparison and Improvement Project 417 (AgMIP) Integrated Crop and Economic Assessments–Joint Publication with the American 418 Society of Agronomy, Crop Science Society of America, and Soil Science Society of 419 America. Imperial College Press, London, pp 25–73 420 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 20 Adiku SGK, Rose CW, Gabric A, Braddock RD, Carberry PS, McCown RL (1998) An evaluation 421 of the performance of maize and cowpea in sole and intercropping systems at two savannah 422 zones in Ghana: A simulation study. ‘MODEL–IT Applications of Modelling as an Innovative 423 Technology in the Agri–Food Chain’. Acta Hortic 476:251–259 424 Adiku, SGK (1995) A Field Investigation and Modelling the Effects of Soil, Climate and 425 Management Factors on the Growth of Maize-Cowpea Intercrop. PhD Thesis, School of 426 Environmental Sciences, Griffith University, Australia 458 pp 427 Adnan AA, Jibrin MJ, Kamara AY, Abdulrahman BL, Shaibu AWS (2017) Using CERES–Maize 428 model to determine the nitrogen fertilization requirements of early maturing maize in the 429 Sudan Savanna of Nigeria. J Plant Nutr 40 (7):1066–1082 430 Akponikpe P, Gérard B, Michels K, Bielders C (2010) Use of the APSIM model in long term 431 simulation to support decision making regarding nitrogen management for pearl millet in 432 the Sahel. Eur J Agron 32:144–154 433 Akponikpe PBI, Gerald B, Bielders CL (2014) Soil water crop modeling for decision support in 434 millet–based systems in the Sahel: a challenge. Afr J Agric Res 9(22):1700–1713 435 Akponikpe PBI, Michels K, Bielders C (2008) Integrated nutrient management of pearl millet in 436 the Sahel combining cattle manure, crop residue and mineral fertilizer. Exp Agric 44:453–437 472 438 Atakora WK, Fosu M, Marthey F (2014) Modeling Maize Production towards Site Specific 439 Fertilizer Recommendation in Ghana. Global J Sci Frontier Re: D Agriculture and 440 Veterinary 14 (6):70–81 441 Bationo A, Buekert A (2001) Soil organic carbon management for sustainable land use in Sudano-442 Sahelian West Africa. Nutr Cycl Agroecosyst 61:131–142 443 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 21 Beletse YG, Durand W, Nhemachena C, Crespo O, Tesfuhuney WA, Jones MR, Teweldemedhin 444 MY, Gamedze SM, Bonolo PM, Jonas S, Walker S, Gwimbi P, Mpuisang TN, Cammarano 445 D, Valdivia RO (2015) Projected impacts of climate change scenarios on the production of 446 maize in Southern Africa: An integrated assessment case study of Bethlehem District, 447 Central Free State, South Africa. In: Rosenzweig C, Hillel D (eds) Handbook of Climate 448 Change and Agroecosystems: The Agricultural Model Intercomparison and Improvement 449 Project (AgMIP) Integrated Crop and Economic Assessments – Joint Publication with the 450 American Society of Agronomy, Crop Science Society of America, and Soil Science 451 Society of America. Imperial College Press, London, pp 125–158 452 Brisson N, Bruno M, Ripoche D, Jeuffroy MH, Ruget F, Nicoullaud B, Gate P, Devienne–Barret 453 F, Antonioletti R, Durr C, Richard G, Beaudoin N, Recous S, Tayot X, Plenet D, Cellier P, 454 Machet J-M, Meynard JM, Delécolle R (1998) STICS: a generic model for the simulation 455 of crops and their water and nitrogen balances. I. Theory and parameterization applied to 456 wheat and corn. Agronomie 18 (5–6):311–346 457 Carberry PS, Hochman Z, McCown RL, Dalgliesh NP, Foale MA, Poulton PL, Hargreaves JNG, 458 Hargreaves DMG, Cawthray S, Hillcoat N, Robertson MJ (2002) The FARMSCAPE 459 approach to decision support: farmers’, advisers’, researchers’ monitoring, simulation, 460 communication and performance evaluation. Agric Syst 74:141–177 461 de Jager A, Kariuku I, Matiri FM, Odendo M, Wanyama JM (1998) Monitoring nutrient flows and 462 economic performance in African farming systems (NUTMON): IV. Linking nutrient 463 balances and economic performance in three districts in Kenya. Agric Ecosyst Environ 71 464 (1–3):81–92 465 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 22 de Wit CT (1958) Transpiration and crop yields. Volume 64 of Agricultural research report/ 466 Netherlands Volume 59 of Mededeling (Instituut voor Biologisch en Scheikundig 467 Onderzoek va Landbouwgewasses) Verslagen van landbouwkundige onderzoekingen. 468 Institute of Biological and Chemical Research on Field Crops and Herbage 469 Diarisso T, Corbeels M, Andrieu N, Djamen P, Douzet J, Tittonell P (2015) Soil variability and 470 crop yield gaps in two village landscapes of Burkina Faso. Nutr Cycl Agroecosyst 105 471 (3):199–216 DOI:10.1007/s10705-015-9705-6 472 Duncan WG, Loomis RS, Williams WA, Hanau R (1967) A model for simulating photosynthesis 473 in plant communities. Hilgardia 38 (4):181–205 474 Dzotsi K, Agboh–Noaméshie A, Struif Bontkes TE, Singh U, Dejean P (2003) Using DSSAT to 475 Derive Optimum Combinations of Cultivar and Sowing Date for Maize in Southern Togo. 476 In: Bontkes TES, Wopereis MCS (eds) Decision support tools for smallholder agriculture 477 in sub-Saharan Africa: a practical guide, pp 100–112 478 Dzotsi KA, Jones JW, Adiku SGK, Naab JB, Singh U, Porter CH, Gijsman AJ (2010) Modelling 479 soil and plant phosphorus within DSSAT. Ecol Modelling 221:2839–2849 480 Farquhar GD (1979) Models describing the kinetics of ribulose biphosphate carboxylase–481 oxygenase. Arch Biochem Biophys 193:456–468 482 Fosu M, Buah SS, Kanton RAL, Agyare WA (2012) Modelling Maize response to mineral 483 fertilizer on silty clay loam in the Northern Savanna of Ghana Using DSSAT model. In: 484 Kihara J, Fatondji D, Jones JW, Hoogenboom G, Tabo R, Bationo A (eds) Improving Soil 485 Fertility Recommendations in Africa using the Decision Support Systems for Agro–486 technology Transfer (DSSAT). Springer Science + Business Media B. V. pp 157–168 487 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 23 Fosu–Mensah BY, MacCarthy DS, Vlek PLG, Safo EY (2012) Simulating impact of seasonal 488 climatic variation on the response of maize (Zea mays L.) to inorganic fertilizer in sub–489 humid Ghana. Nutr Cycl Agroecosyst 94:255–271 490 Gungula DJ, Kling JG, Togun AO (2003) CERES–Maize predictions of Maize phenology under 491 Nitrogen–stressed conditions in Nigeria. Agron J 95:892–899 492 Haefele SM, Sipaseuth N, Phengsouvanna V, Dounphady K, Vongsouthi S (2010) Agro–economic 493 evaluation of fertilizer recommendations for rainfed lowland rice. Field Crop Res 119:215–494 224 495 Haefele SM, Wopereis MCS, Ndiaye MK, Kropff MJ (2003) A Framework to Improve Fertilizer 496 Recommendations for Irrigated Rice in West Africa. Agr Syst 76(1):313–335 497 Hess TM, Stephens W, Crout NMJ, Young SD, Bradley RG (1997) Predicting Arable Resources 498 in Hostile environments (PARCH), User Guide, Natural Resource Institute. Chatham, UK 499 Hoogenboom G, Jones JW, Traore PCS, Boote KJ (2012) Experiments and data for model 500 evaluation and application; Understanding the Processes using a Crop Simulation Model. 501 In: Kihara J, Fatondji D, Jones JW, Hoogenboom G, Tabo R, Bationo A (eds) Improving 502 Soil Fertility Recommendations in Africa using the Decision Support Systems for 503 Agrotechnology Transfer (DSSAT). Springer Science + Business Media B. V. pp 9–18 504 505 Huth NI, Carberry PS, Poulton PL, Brennan LE, Keating BA (2003) A framework for simulating 506 agroforestry options for the low rainfall areas of Australia using APSIM. Eur J Agron 507 18:171–185 508 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 24 Jansen BH, Guiking FCT, van der Eijk D, Smaling EMA, Wolf J, van Reuler H (1990) A system 509 for quantitative evaluation of the fertility of tropical soils (QUEFTS). Geoderma 46:299–510 318 511 Jones JW, Antle JM, Bruno B, Boote KJ, Conant RT, Ian Foster I, Godfray HCJ, Mario Herrero 512 M, Howitt RE, Janssen S, Keating BA, Munoz–Carpena R, Porter CH, Rosenzweig C, 513 Wheeler TR (2016) Brief history of agricultural systems modeling. Agric Syst 514 Jones PG, Thorton PK (2003) The potential impacts of climate change on maize production in 515 Africa and Latin America in 2055. Global Environ Change 13:51–59 516 Kassie BT, Van Ittersum MK, Hengsdijk H, Asseng S, Wolf J, Rötter RP (2014) Climate –induced 517 yield variability and yield gaps of maize (Zea mays L) in Central Rift Valley of Ethiopia. 518 Field Crop Res 160:41–53 519 Keating BA, Carberry PS, Hammer GL, Probert ME, Robertson MJ, Holzworth D, Huth NI, 520 Hargreaves JNG, Meinke H, Hochman Z, McLean G, Verburg K, Snow V, Dimes JP, 521 Silburn M, Wang E, Brown S, Bristow KL, Asseng S, Chapman S, McCown RL, Freebairn 522 DM, Smith CJ (2003) An overview of APSIM, a model designed for farming systems 523 simulation. Eur J Agron 18:267–288 524 Kihara J, Fatondji D, Jones JW, Hoogenboom G, Tabo R, Bationo A (2012) Improving Soil 525 Fertility Recommendations in Africa using the Decision Support Systems for Agro-526 technology Transfer (DSSAT). Springer Science + Business Media B. V.pp 184 527 Kihara J, Huising J, Nziguheba G, Waswa BS, Njoroge S, Kabambe V, Iwuafor E, Kibunja C, 528 Esilaba AO, Coulibaly A (2015) Maize response to macronutrients and potential for 529 profitability in sub-Saharan Africa. Nutr Cycl Agroecosyst 105: 171–181 DOI 530 10.1007/s10705-015-9717-2 531 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 25 Littleboy M, Silburn DM, Freebairn DM, Woodruff DR, Hammer GL (1989) PERFECT: A 532 computer simulation model of Productivity Erosion Runoff Functions to Evaluate 533 Conservation Techniques. Queensland Department of Primary Industries, Brisbane, 534 Australia, pp 119 535 Mabhaudhi T, Modia AT, Beletse YG (2014) Parameterisation and evaluation of the FAO–536 AquaCrop model for a South African taro (Colocasia esculenta L. Schott) landrace. Agric 537 For Meteorol 193:132–139 538 MacCarthy DS, Adiku SGK, Freduah BS, Gbefo F, Kamara AY (2017) Using CERES–Maize and 539 ENSO as Decision Support Tools to evaluate climate–sensitive farm management practices 540 for maize production in the northern regions of Ghana. Front Plant Sci 8: 541 31. doi: 10.3389/fpls.2017.00031. 542 MacCarthy DS, Akponikpe PBI, Narh S, Tegbe R (2015) Modelling the effect of seasonal climate 543 variability on the efficiency of mineral fertilization on maize in the coastal savannah of 544 Ghana. Nutr Cycl Agroecosyst 102:45–64 545 MacCarthy DS, Vlek PLG, Bationo A, Tabo R, Fosu M (2010) Modeling nutrient and water 546 productivity of sorghum in smallholder farming systems in a semi–arid region of Ghana. 547 Field Crop Res 118(3):251–258 548 MacCarthy DS, Vlek PLG, Fosu–Mensah BY (2012) The Response of Maize to N Fertilization in 549 a Sub–humid Region of Ghana; Understanding the Processes using a Crop Simulation 550 Model. In: Kihara J, Fatondji D, Jones JW, Hoogenboom G, Tabo R, Bationo A (eds) 551 Improving Soil Fertility Recommendations in Africa using the Decision Support Systems 552 for Agrotechnology Transfer (DSSAT). Springer Science + Business Media B. V. pp 157–553 168 554 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 26 Maran La R, Leatherman DA (1992) NUMAS – A nutrient management advisory system. 555 Technical summary and user manual. The knowledge Based Systems Research Lab, 556 University of Illinois, Urbana 557 Masikati P, Homann-Kee Tui S, Descheemaeker K, Crespo O, Walker S, Lennard CJ, Claessens 558 L, Gama AC, Famba S, van Rooyen AF, Valdivia RO (2015) Crop–Livestock 559 intensification in the face of climate change: Exploring opportunities to reduce risk and 560 increase resilience in Southern Africa by using an integrated multi–modeling approach. In: 561 Rosenzweig C, Hillel D (eds) Handbook of Climate Change and Agroecosystems: The 562 Agricultural Model Intercomparison and Improvement Project (AgMIP) Integrated Crop 563 and Economic Assessments – Joint Publication with the American Society of Agronomy, 564 Crop Science Society of America, and Soil Science Society of America. Imperial College 565 Press, London, pp 159–198 566 Masikati P, Manschadi A, van Rooyen A, Hargreaves J (2014) Maize–mucuna rotation: An 567 alternative technology to improve water productivity in smallholder farming systems. 568 Agric Syst 123:62–70 569 McCown RL (1973) An evaluation of the influence of available soil water storage capacity on 570 growing season length and yield of tropical pastures using simple water balance models. 571 Agric Meteorol 11:53–63 572 McCown RL, Hammer GL, Hargreaves JNG, Holzworth DP, Freebairn DM (1996) APSIM: a 573 novel software system for model development, model testing and simulation in agricultural 574 systems research. Agric Syst 50:255–271 575 Mitscherlich EA (1913) Soil science for agriculture and forestry. 2nd edition, Verlag Paul Parey, 576 Berlin, pp 317 577 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 27 Monteith JL (1977) Climate and the efficiency of crop production in Britain. Phil Trans Roy Soc 578 Lond 281:277–294 579 Naab JB, Mahama GY, Koo J, Jones JW, Boote K (2015) Nitrogen and phosphorus fertilization 580 with crop residue retention enhances crop productivity, soil organic carbon, and total soil 581 nitrogen concentrations in sandy–loam soils in Ghana. Nutr Cycl Agroecosyst 102:33–43 582 Nurudeen AR (2011) Decision Support System for Agro–technology Transfer (DSSAT) model 583 simulation of maize growth and yield response to NPK fertilizer application on a 584 benchmark soil of Sudan Savanna Agro–ecological Zone of Ghana. MSc Thesis. Kwame 585 Nkrumah University of Science and Technology Kumasi 586 Parton WJ, McKeown B, Kirchner V, Ojima DS (1992) CENTURY user’s manual. Colorado State 587 University, NREL Publication, Fort Collins 588 Raes DP, Steduto P, Hsiao TC, Fereres E (2009) AquaCrop—The FAO crop model to predict yield 589 response to water: II Main algorithms and soft ware description. Agron J 101:438–447 590 Rao KPC, Sridhar G, Mulwa RM, Kilavi MN, Esilaba A, Athanasiadis IN, Valdivia RO (2015) 591 Impacts of climate variability and Change on Agricultural Systems in East Africa. In: 592 Rosenzweig C, Hillel D (eds) Handbook of Climate Change and Agroecosystems: The 593 Agricultural Model Intercomparison and Improvement Project (AgMIP) Integrated Crop 594 and Economic Assessments – Joint Publication with the American Society of Agronomy, 595 Crop Science Society of America, and Soil Science Society of America. Imperial College 596 Press, London, pp 75–124 597 Rose DC, Sutherland WJ, Parker C, Lobley M, Winter M, Morris C, Twining S, Foulkes C, Amano 598 T, Dicks LV (2016) Decision support tools for agriculture: Towards effective design and 599 delivery. Agric Syst 149:165–174 600 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 28 Rosenzweig C, Jones JW, Hatfield JL, Ruane AC, Boote KJ, Thorburn P, Antle JM, Nelson GC, 601 Porter C, Janssen S, Asseng S, Basso B, Ewert F, Wallach D, Baigorria G, Winter JM (2013) 602 The Agricultural Model Intercomparison and Improvement Project (AgMIP): Protocols and 603 pilot studies. Agric Forest Meteorol 170:66–182 604 Sanchez P, Denning G, Nziguheba G (2009) The African green revolution moves forward. Food 605 Secur 1:37–44 606 Sanchez PA (2002) Soil fertility and hunger in Africa. Science 295:2019–2020 607 Segda Z, Haefele SM, Wopereis MCS, Sedogo MP, Guinko S (2005) Combining Field and 608 Simulation Studies to Improve Fertilizer Recommendations for Irrigated Rice in Burkina 609 Faso. Agron J 97:1429–1437 610 Shaffer MJ, Gupta SC, Linden DR, Molina JAE, Clapp CE, Larson WE (1983) Simulation of 611 nitrogen, tillage, and residue management effects on soil fertility. In: Lauenroth WK, 612 Skogerboe GV, Flug M (eds) Analysis of Ecological Systems: State-of-the-Art in 613 Ecological Modelling. Developments in Environmental Modelling, 5. Elsevier, 614 Amsterdam, pp 525–544 615 Smaling EMA, Fresco LO (1993) A decision–support model for monitoring nutrient balances 616 under agricultural land use (NUTMON). Geoderma 60(1–4):235–256 617 Stoorvogel JJ, Smailing EMA (1990) Assessment of soil nutrient depletion in Sub-Saharan Africa: 618 1983 – 2000. Vol 2: 28 Nutrient balances per crop and per landuse systems. ISRIC 619 Stöckle CO, Donatelli M, Nelson R (2003) CropSyst, a cropping systems simulation model. Eur J 620 Agron 18(3–4):289–307 621 Tachie-Obeng E, Akponikpe PBI, Adiku S (2013) Considering effective adaptation options to 622 impacts of climate change for maize production in Ghana. Environ Develop 5:131–145 623 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 29 Theophrastus (1989) Enquiry into Plants (Historia Plantarum), Books VI-IX, translated by Arthur 624 F. Hort, Loeb Classical Library, Vol 79, Book VIII, 7.6. Harvard University Press, 625 Cambridge, Massachusetts 626 Thornton PK, Jones PG, Alagarswamy G, Adresen J (2009) Spatial variation of crop yield response 627 to climate change in East Africa. Global Environ Change 19:54–65 628 Van Bavel CHM (1953) A Drought Criterion and its Application in Evaluating Drought incidence 629 and Hazard. Agron J 45:167–172 630 Van Diepen C, Wolf J, Van Keulen H, Rappoldt C (1989) WOFOST: A simulation model of crop 631 production. Soil Use Manag 5:16–24 632 Van Ittersum MK, Cassman KG, Grassini P, Wolf J, Tittonell P, Hochman Z (2013) Yield gap 633 analysis with local to global relevance—A review. Field Crops Res 143:4–17 634 Vanlauwe B, Kihara J, Chivenge P, Pypers P, Coe R, Six J (2011) Agronomic use efficiency of N 635 fertilizer in maize–based systems in sub–Saharan Africa within the context of integrated 636 soil fertility management. Plant Soil 339:35–50 637 Van Keulen H, Breman H (1990) Agricultural development in the West African Sahelian region: 638 a cure against land hunger? Agric Ecosyst Environ 32:177–197 639 Voortman RL, Brouwerd Albersen PJ (2004) Characterization of spatial soil variability and its 640 effects on miller yield on Sudano Sahelian Coversnad in SW Niger. Geoderma 121:65–82 641 Wafula BM (1995) Application of crop simulation in Agricultural extension and Research in 642 Kenya. Agric Syst 49:399–412 643 Whitbread AM, Robertson MJ, Carberry PS, Dimes JP (2010) How farming systems simulation 644 can aid the development of more sustainable smallholder farming systems in southern 645 Africa. Eur J Agron 32:51–58 646 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 30 Wopereis MCS, Haefele SM, Dingkuhn M, Sow A (2003) Decision Support Tools for Irrigated 647 Rice–Based Systems in the Sahel. Decision Support Tools for Rainfed Crops in the Sahel 648 at the Plot and Regional Scales In: Struif Bontkes TE, Wopereis MCS (eds). Decision 649 Support Tools for Smallholder Agriculture in Sub–Saharan Africa: A Practical Guide, pp 650 114–126 651 Wopereis MCS, Tame´lokpo A, Ezui K, Gnakpe´nou D, Fofana B, Breman H (2006) Mineral 652 fertilizer management of maize on farmer fields differing inorganic inputs in the West 653 African savanna. Field Crops Res 96:355–362 654 Zingore S, Murwira HK, Delve RJ, Giller KE (2007) Influence of nutrient management strategies 655 on variability of soil fertility, crop yields and nutrient balances on smallholder farms in 656 Zimbabwe. Agric Ecosyst Environ 119:112–126 657 Zinyengere N, Crespo O, Hachigonta S, Tadross M (2015) Crop model usefulness in drylands of 658 southern Africa: an application of DSSAT. South Afr. J Plant Soil 32 (2):95–104 659 660 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 31 List of figure captions 661 Fig. Response curve of maize yield to different levels of nitrogen application over 30 years (1980-662 2009) simulation period for Tamale, Ghana. 663 Fig. 2: Monetary returns on the use of inorganic fertilizer in maize production at a site in Tamale, 664 Ghana 665 Fig. 3: Simulated maize grain and stover yields in response to mineral N fertilization in 3 countries 666 in southern Africa. 667 Fig. 4: The Simulated effect of soils from Koutiala, Mali with different water holding capacity 668 on the response of maize yield to mineral nitrogen fertilization. WHC is water holding capacity. 669 Fig. 5: The simulated effect of sowing dates on the response of maize yield to mineral fertilizer 670 application in Nioro. Senegal. 671 672 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Fig 1 y = -0.0001x2 + 0.024x + 0.69 R² = 0.99, P=0.0001 y = -0.0002x2 + 0.04x + 1.1 R² = 0.99, P=0.0001 y = -0.0004x2 + 0.07x + 1.5 R² = 0.99, P=0.0001 0 1 2 3 4 5 6 0 50 100 150 G ra in y ie ld ( t/ h a) N applied (kg/ha) Min Median Max Figure 1 Click here to download Figure Fig 1.docx Monetary Returns (Cedis ha -1 ) 0 1000 2000 3000 4000 5000 6000 C u m m u la ti v e p ro b ab il i t y 0.0 0.2 0.4 0.6 0.8 1.0 0 kg N ha -1 15 kg N ha -1 30 kg N ha -1 45 kg N ha -1 60 kg N ha -1 75 kg N ha -1 90 kg N ha -1 120 kg N ha -1 Fig. 2 Figure 2 Click here to download Figure Fig 2.docx c) Zimbabwe, Grain N applied (kg/ha) 0 50 100 0 2 4 6 8 10 12 Min Median Max f) Zimbabwe, Stover 0 50 100 150 Min Median Max b) Mozambique, Grain Y ie ld ( t/ h a ) 0 2 4 6 8 10 12 Min Median Max e) Mozambique, Stover Min Median Max a) Malawi, Grain 0 2 4 6 8 10 12 14 Min Median Max d) Malawi, Stover Min Median Max y = -0.0004x 2 + 0.09x + 2.41 R² = 0.99, P<0.01 y = -0.0005x 2 + 0.09x + 2.01 R² = 0.97, P<0.01 y = -0.0004x 2 + 0.07x + 1.56 R² = 0.93, P<0.01 y = -0.0004x 2 + 0.09x + 0.21 R² = 0.97, P<0.01 y = -0.0003x 2 + 0.07x + 0.28 R² = 0.98, P<0.01 y = -7E-05x 2 + 0.01x + 0.32 R² = 0.70, P<0.01 y = -0.0004x 2 + 0.08x + 0.59 R² = 0.98, P<0.01 y = -0.0003x 2 + 0.06x + 0.5 R² = 0.99, P<0.01 y = -5E-05x 2 + 0.01x + 0.16 R² = 0.95, P<0.01 y = -0.0005x 2 + 0.13x + 4.85 R² = 0.99, P<0.01 y = -0.0005x 2 + 0.08x + 4.44 R² = 0.88, P<0.01 y = -0.0006x 2 + 0.12x + 4.55 R² = 0.98, P<0.01 y = -0.0008x 2 + 0.17x + 2.33 R² = 0.99, P<0.01 y = -0.0008x 2 + 0.16x + 2.09 R² = 0.99, P<0.01 y = -0.0003x 2 + 0.06x + 1.60 R² = 0.90, P<0.01 y = -0.0008x 2 + 0.17x + 3.29 R² = 0.99, P<0.01 y = -0.0007x 2 + 0.12x + 2.961 R² = 0.98, P<0.01 y = -6E-05x 2 + 0.01x + 1.94 R² = 0.61, P=0.09 Fig. 3 Figure 3 Click here to download Figure Fig 3.docx a) WHC 50mm 0 1 2 3 4 5 Min Median Max b) WHC 55mm N applied (kg/ha) 0 50 100 G ra in y ie ld ( k g /h a ) 0 1 2 3 4 Min Median Max c) WHC 32mm Min Median Max d) WHC 40mm 0 50 100 150 Min Median Max y = -8E-05x 2 + 0.04x + 0.66 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.05x + 0.49 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.05x + 0.37 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.04x + 0.19 R² = 0.99, P<0.01 y = -0.0001x 2 + 0.05x + 0.81 R² = 0.99, P<0.01 y = -0.0001x 2 + 0.03x + 0.56 R² = 0.98, P<0.01 y = -0.0002x 2 + 0.06x + 0.73 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.04x + 0.55 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.03x + 0.27 R² = 0.98, P<0.01 y = -0.0002x 2 + 0.06x + 0.93 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.05x + 0.68 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.03x + 0.71 R² = 0.95, P<0.01 Fig. 4 Figure 4 Click here to download Figure Fig 4.docx Late planting N applied (kg/ha) 0 50 100 150 0 1 2 3 4 5 6 Min Median Max Mid planting G ra in y ie ld ( k g /h a ) 0 1 2 3 4 5 6 Min Median Max Early planting 0 1 2 3 4 5 6 7 Min Median Max y = -0.0003x 2 + 0.09x + 0.02 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.06x + 0.01 R² = 0.99, P<0.01 y = -9E-05x 2 + 0.02x + 0.13 R² = 0.98, P<0.01 y = -0.0002x 2 + 0.08x + 0.12 R² = 0.99, P<0.01 y = -0.0002x 2 + 0.06x + 0.04 R² = 0.99, P<0.01 y = -7E-05x 2 + 0.03x + 0.08 R² = 0.98, P<0.01 y = -0.0002x 2 + 0.05x + 0.10 R² = 0.99, P<0.01 y = -5E-05x 2 + 0.01x + 0.03 R² = 0.99, P<0.01 y = -0.0003x 2 + 0.06x + 0.19 R² = 0.99, P<0.01 Fig. 5 Figure 5 Click here to download Figure Fig 5.docx 1 Table 1. Selected publication on the use of Decision support tools in Sub Sahara Africa (SSA). 1 Source Crop Treatment Application Location MacCarthy et al. 2012 Maize N CSM-CERES (DSSAT v 4.0) Ghana Fatondji et al. 2012 Millet Manure CSM-CERES (DSSAT v 4.0) Niger Fosu et al. 2012 Maize N CSM-CERES (DSSAT v 4.0) Ghana Zinyengere et al. (2015) Maize Variable CSM-CERES (DSSAT v 4.0) Malawi Zinyengere et al. (2015) Groundnut None CropGro (DSSAT v 4.0) Malawi MacCarthy et al 2009 Sorghum N & P APSIM v 4.0 Ghana MacCarthy et al 2015 Maize N APSIM v 7.4 Ghana Fosu-Mensah et al. 2013 Maize N & P APSIM v 6.1 Ghana Tetteh and Nurudeen (2015) Maize N & P CSM-CERES (DSSAT v 4.0) Ghana Chisanga 2014 Maize N and planting dates CSM-CERES (DSSAT v 4.0) Zambia Kisaka et al. 2015 Maize N and manure APSIM Kenya Delve et al. 2009 Maize P APSIM Kenya Delve et al. 2009 Maize P APSIM Kenya Delve et al. 2009 Bean P APSIM Kenya Chimonyo et al. 2016 Sorghum Water regime APSIM South Africa Chimonyo et al. 2016 Cowpea Water regime APSIM South Africa Robertson et al. 2005 Velvet bean N and velvet bean as previous crop APSIM Malawi Chikowo et al. 2008 Maize Fertilizer and rainfall APSIM Kenya Katambara et al. 2013 Rice Water productivity and efficiency AQUACROPP Tanzania Ngwira et al. 2014 Maize Climate change, CA, CT CSM-CERES DSSAT Malawi Estes et al. 2013 Maize, Wheat Climate impacts, N CSM-CERES DSSAT v 4.5.0.047 South Africa Estes et al. 2013 Wheat Climate impacts GAM model South Africa Bontkes et al. 2003 Maize N, P, K QUEFTS Togo Micheni et al. 2004 Sorghum, cowpea, pearl millet Manure APSIM Kenya Tsubo et al. 2004 Maize Cereal-legume intercropping APSIM South Africa Tsubo et al. 2004 Beans Cereal-legume intercropping APSIM South Africa Smaling and Janssen, 1993 Maize N, P, K QUEFTS Kenya Okwach and Simiyu 1999 Maize Land management practices APSIM Kenya Gaiser et al. 20010 Maize (West Africa) Improved varieties, soils EPIC West Africa Table 1 Click here to download Table Table 1.docx 2 Folberth et al. 2013 Maize N, P, improved seeds GEMIC Sub-Sahara Africa O’Leary, 2000 Sugarcane N, water, temperature APSIM South Africa O’Leary, 2000 Sugarcane N, water, temperature CANEGRO South Africa O’Leary, 2000 Sugarcane N, water, temperature QCANE South Africa Ncube et al. 2009 Sorghum N uptake APSIM Zimbabwe Srivastava et al. 2012 Yam Fallow EPIC Benin Jansen 2010 Maize SOM, residual P, N NUE Kenya Tittonell et al. 2013 Maize N, P, K manure QUEFTS Kenya Tittonell et al. 2008 Maize Fertilizer, Manure FIELD Kenya Kurwakumire et al. 2014 Maize N, P, K, water use efficiency QUEFTS Zimbabwe Mowo et al. 2006 Maize N, P, K QUEFTS Tanzania Araya et al. 2010 Barley Water regime, planting dates AQUACROP v 3.0 Ethiopia Mabhaudhi et al. 2014a Taro Water regime, Taro landraces AQUACROP South Africa Mabhaudhi et al. 2014b Groundnut Water regime AQUACROP South Africa Karunaratne et al. 2011 Groundnut Soil moisture regime AQUACROP Swaziland & Botswana Beletse et al. 2012 Sweet potato Irrigation treatment AQUACROP South Africa Kipkorir et al. 2010 Maize Water regime AQUACROP Kenya Mugalavai and Kikorir et al. 2015 Maize AQUACROP Kenya Mhizha et al. 2014 Maize Sowing management options AQUACROP Zimbabwe Nyakudya and Stroosnijder, 2014 Maize Rooting depth, planting density, planting date AQUACROP Zimbabwe Masanganise et al. 2013 Maize Cultivars, planting dates, climate AQUACROP Zimbabwe Singels and Bezuidenhout, 2002 Sugarcane Temperature and water stress CANEGRO South Africa Dzotsi et al. 2003 Maize Cultivar, sowing date DSSAT (CERES- Maize) Togo Dzotsi et al. 2010 Maize N, P DSSAT Ghana Jagtap et al. 1999 Maize N, varieties DSSATv2.1 (CERES- Maize) Nigeria Hansen et al. 2009 Maize (Kenya) Precipitation, fertilizer management GCM Kenya Mupangwa and Jewitt, 2011 Maize (South Africa) No-till (NT) and CT systems APSIM South Africa Adnan et al. 2017 Maize N DSSAT v 4.6 (CERES-Maize) Nigeria 2