CIAT Research Online - Accepted Manuscript On-farm diversity offsets environmental pressures in tropical agroecosystems: A synthetic review for cassava-based systems 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: Delaquis, Erik; De Haan, Stefan; Wyckhuys, Kris A.G.. 2017. On-farm diversity offsets environmental pressures in tropical agroecosystems: A synthetic review for cassava-based systems. Agriculture, Ecosystems and Environment. 251: 226-235. Publisher’s DOI: http://dx.doi.org/10.1016/j.agee.2017.09.037 Access through CIAT Research Online: http://hdl.handle.net/10568/88232 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-NoDerivatives 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 change this manuscript in any way or use it commercially. For more information, please contact CIAT Library at CIAT-Library@cgiar.org. 1 For submission to Agriculture, Ecosystems and Environment 1 2 3 4 5 Send correspondence to: 6 7 Kris A. G. Wyckhuys 8 Institute of Plant Protection, Chinese Academy of 9 Agricultural Sciences 10 No. 2 West Yuanmingyuan Rd., Haidian District, 11 Beijing, 100193, P. R. China 12 Tel: 86-10-62813685 13 Contact: kagwyckhuys@gmail.com 14 15 16 17 18 19 On-farm diversity offsets environmental pressures in tropical agro-20 ecosystems: a synthetic review for cassava-based systems 21 22 23 24 25 Erik Delaquis1, Stefan de Haan1 and Kris A.G. Wyckhuys1* 26 27 28 1International Center for Tropical Agriculture (CIAT) Asia regional office, Hanoi, Vietnam 29 30 2 Abstract 31 Ecosystem integrity is at risk across the tropics. In the quest to meet global dietary and market 32 demands, tropical agro-ecosystems face unrelenting agricultural intensification and expansion. 33 Agro-biodiversity can improve ecosystem stability and functioning, but its promotion in 34 smallholder-based systems faces numerous practical hurdles. In the tropics, cassava (Manihot 35 esculenta Crantz) is cultivated on over 25 million hectares and features as the third most 36 important source of calories. Cassava crops are often maintained by resource-poor farmers who 37 operate on marginal lands, at the fringes of sensitive, biodiverse habitats. As traditional 38 intercropping schemes are gradually abandoned, monoculture cassava systems face stagnating 39 yields, resource-use inefficiencies and agro-ecosystem degradation. A global literature search 40 identified 189 cassava intercropping studies, covering 330 separate instances of intercropping 41 systems. We employed a vote-counting approach and simple comparative measure across a 42 subset of 95 studies to document the extent to which intercropping sustains a bundle of 43 ecosystem services. Across geographies and biophysical conditions, a broad range of intercrops 44 provided largely positive effects on five key ecosystem services: pest suppression, disease 45 control, land equivalency ratio (LER), and soil and water-related services. Ecosystem services 46 were augmented through the addition of a diverse range of companion crops. Results indicated 47 25 positive impacts vs. 3 negative impacts with the addition of maize, 5 vs. 1 with gramineous 48 crops, 23 vs. 3 with four species of grain legumes, and 9 vs. 0 with trees. Appropriate 49 intercropping systems can help to strike a balance between farm-level productivity, crop 50 resilience, and environmental health. Our work highlights an urgent need for interdisciplinary 51 research and systems-level approaches to identify intensification scenarios in which crop 52 3 productivity, provision of ecosystem services, biodiversity conservation, and human well-being 53 are all balanced. 54 Keywords: Food security; land-sharing; sustainable intensification; crop diversification; 55 ecosystem services 56 57 1. Agricultural expansion puts tropical ecosystems at risk 58 Rapid population growth, shifting consumption patterns, and resource competition are 59 increasing pressure on the world’s agricultural systems and non-arable land (Godfray et al., 60 2010). Contemporary agricultural trends have dramatically shifted farming practices, promoted 61 rapid expansion of agricultural lands, and triggered global environmental changes that risk 62 destabilizing whole ecosystems (Foley et al., 2011). With agro-ecosystems covering 37.5 % of 63 global national land surfaces in 2014 (FAOSTAT, 2016), environmental impacts linked to farm-64 level management decisions are substantial, and are expected to be exceptionally pronounced in 65 tropical terrestrial ecosystems (Laurance et al., 2014). 66 The pursuit of increased production through both area expansion and farming intensification 67 has resulted in an increase of agricultural areas in the tropics of >100 million ha in the 1980-90s, 68 occurring largely at the expense of intact or disturbed forests (Gibbs et al., 2010). The limits to 69 this expansion have simultaneously driven a need to increase productivity on limited land, 70 sparking research into the causes of sub-optimal yields and the potential for ‘yield gap closure’ 71 (van Ittersum et al., 2016; Sayer and Cassman, 2013). Farmers often respond to the need for 72 increased productivity with intensification measures, many of which have negative 73 4 environmental impacts at field, farm, and agro-landscape levels (e.g., Emmerson et al., 2016). 74 Irrational pesticide and fertilizer use and extractive management are commonplace, leading to 75 soil and water resource degradation in many parts of the tropics (Godfray et al., 2010), while 76 exacerbating biotic and abiotic production constraints in both intensified and low-input farming 77 systems (Poppy et al., 2010). 78 Millions of smallholder farmers eke out a living by continuously cropping in such settings, 79 which are characterized by shrinking natural resource bases and degraded agro-ecosystem 80 functioning (Bai et al., 2008; Barbier, 1997; Bossio et al., 2010). Though they constitute the 81 backbone of global food security, many of the world’s smallholders continue to live in poverty, 82 cultivate marginal lands, and operate on the fringes of sensitive, biodiverse habitats (Tscharntke 83 et al., 2010). 84 In this paper we explore how field-level diversification fosters the provision of multiple key 85 ecosystem services (i.e., soil and water conservation, pest regulation and disease control, and 86 land equivalency ratio) in a major tropical and subtropical food crop. More specifically, we 87 examine the example of intercropping in cassava-based systems through an ecosystem services 88 lens. We provide information on recent trends in cassava cultivation globally, and subsequently 89 discuss associated environmental impacts. Next, we systematically review the literature on 90 intercropping practices in cassava-based systems, present its impacts on multiple ecosystem 91 services, and discuss further implications of these findings for cassava-based farming systems 92 across the tropics. 93 5 2. Cassava: an adaptable ‘survivor’ crop 94 Cassava (Manihot esculenta Crantz) production has increased greatly in the past 50 years (Fig. 95 1b). This starchy, tuberous staple is now cultivated on ~ 25 million ha throughout the global 96 tropics (FAOSTAT, 2016). Originating in the Neotropics (Olsen and Schaal, 1999), cassava is 97 now an important food in sub-Saharan Africa and South America, while in mainland Southeast 98 Asia it is predominantly a cash crop. Cassava is the largest calorie producer among roots and 99 tubers, making it a critical crop in resource-poor farming settings across the global tropics (Fig. 100 1b). Cassava is highly adaptable to variable conditions, being grown in a wide range of agro-101 ecological settings: from Africa’s arid Sahel and the cool highlands of Zambia to Colombia’s 102 Andean lowlands and the limestone uplands of Laos and Vietnam. A perennial woody plant 103 primarily managed as an annual, cassava is cultivated for its starchy roots used as human food, 104 animal feed, a source of industrial starch, and a biomass energy feedstock (Zhou and Thomson, 105 2009; von Maltitz et al., 2009). 106 A hardy ‘survivor’ crop, cassava thrives in degraded settings, under low soil fertility, at high 107 temperatures, and can withstand periodic droughts (El-Sharkawy, 2014). Cassava is highly 108 resilient and adaptable in the face of ongoing climatic changes, providing options for adaptation 109 in challenging environments (Jarvis et al., 2012). Cassava’s ability to grow on poor soils, under 110 sub-optimal climatic conditions, and to provide the advantage of flexible harvest timing, make it 111 the crop of ‘last resort’ across the tropics (Hillocks et al., 2001) and earn it the moniker ‘the 112 drought, war, and famine crop’ (Burns et al., 2010). 113 Because intermediate yields are often attainable even in poor conditions, for example ~ 14T/ha 114 on East and Southern African smallholder farms (Tittonell and Giller, 2013), cassava is often 115 cultivated in monocultures, without proper addition of fertilizer or organic amendments, with 116 6 complete abandonment of rotation schemes, and using low quality planting material. Cassava 117 enjoys a theoretical yield potential (defined as the yield of a crop grown in the absence of biotic 118 constraints, and with non-limiting water and nutrients) approaching 90 T/ha (Cock et al., 1979; 119 van Ittersum et al., 2013). Despite this, average yields in the tropics remain low, and are 120 increasing only slowly (El-Sharkawy, 2012; Tittonell and Giller, 2013). Farm yields throughout 121 Africa as a whole average 10 T/ha, far below both the 15-40 T/ha obtained in local on-farm trials 122 in the same agro-ecozones (Fermont et al., 2009), or the average yield of 20.7 T/ha in Southeast 123 Asia (FAOSTAT, 2016). Although substantial yield gains are predicted when farmers adopt the 124 use of pre-emergence herbicides or appropriate soil amendments (Howeler, 2015; Fermont et al., 125 2009), it is likely that strategic use of low-input technologies can equally achieve significant 126 progress toward these goals while safeguarding food security, improving farmer livelihoods, and 127 off-setting a range of environmental impacts (Pypers et al., 2011; Fermont et al., 2009). Given 128 the large scale of cassava’s cropping, its ubiquity in sensitive tropical environments, and its 129 importance to impoverished smallholders with limited options for investment in agricultural 130 inputs and technologies, the environmental impacts of its cultivation is a topic that demands 131 serious consideration. 132 3. Environmental impacts 133 Long thought to be largely environmentally benign due to the ability to produce acceptable 134 yields on marginal soils with minimal external inputs, studies now demonstrate that improperly 135 managed cassava can contribute to cycles of environmental degradation that ultimately threaten 136 the sustainability of crop production (Reynolds et al., 2015). Crop area expansion and 137 7 unsustainable farming practices influence these outcomes, though patterns of variability with 138 socio-economic factors, geography, and ecosystem composition are poorly understood. 139 Cassava has gained the reputation of promoting soil erosion due to its erect architecture, poor 140 canopy cover in the early growing season coinciding with heaviest rains, soil-disturbing harvest 141 scheme, and ability to continue producing despite mismanagement of degraded soils (Moench, 142 1991; Valentin et al., 2008). However, those factors do not appear to be the intrinsic cause of 143 cassava’s environmental impacts. Cassava is commonly grown on erodible hillsides, drought-144 prone areas or acidic soils, and recently deforested land, generating negative impacts directly 145 related to overall land use and improper management (Reynolds et al., 2015). Soil exhaustion, 146 fertility depletion, and topsoil loss challenge cassava production (and indeed that of all crops) 147 across the tropics (Fermont et al., 2009; De Vries et al., 2010; Waddington et al., 2010; Clement 148 and Amezaga, 2008; Valentin et al., 2008). When inadequately managed, soil biological function 149 and plant immune responses decline, and crops become increasingly prone to arthropod pests and 150 plant diseases (Graziosi et al., 2016; Vurro et al., 2010). These problems are rarely evaluated 151 comprehensively, and symptomatic treatments are often pursued without tackling the underlying 152 drivers or exploring the complex interplay between contributing factors. 153 In the period following the Columbian exchange (see Crosby, 1972), cassava spread far beyond 154 its native range in the Americas to become an important component of agroecosystems across the 155 global tropics. Since the 1980s, notable increases in cassava area have occurred in key 156 production zones including Nigeria, Cambodia, and Vietnam. In West Africa, continuing 157 agricultural expansion is leading to forest loss and degradation, with cassava cropping identified 158 as one of several drivers (Norris et al., 2010). In Cambodia, booming demand for export cassava 159 has contributed to deforestation in upland areas, alongside other cash crops, such as rubber 160 8 (Hought et al., 2012; NEPCon, 2014). Land clearance for cassava cultivation has also occurred 161 in Latin America, but effects on local biodiversity remain poorly documented (Howeler et al., 162 2000). 163 As it is grown with few interventions during a long growing cycle (8-12 months), cassava may 164 provide stable habitat conditions for diverse biota. However, cassava monocultures sustain lower 165 biodiversity than certain agricultural and agro-forestry habitats, or natural areas (Francesconi et 166 al., 2013). As in examples from across multiple production systems and settings (Tews et al., 167 2004), the ability of cassava-based systems to sustain biodiversity may improve with 168 management regimes that enhance habitat heterogeneity, which can improve and sustain 169 biodiversity at a field, farm, and landscape level (Benton et al., 2003). One of the most well-170 known ways to increase agrobiodiversity is through the practice of intercropping. 171 4. Intercropping as a traditional solution to restore ecosystem degradation 172 In the Amazonian center of origin, traditional societies grow dozens of cassava varieties on 173 small plots, interspersed with other food, fiber, or cash crops (McKey et al., 2001). This 174 approach contrasts with the market-driven, large-scale, and genetically uniform systems 175 promoted by industrial agriculture models. Highly diverse farming, including intercropping, 176 prevailed in the tropics prior to the introduction of modern, globalized markets and high intensity 177 production schemes (Hulugalle and Ezumah, 1991; Wargiono et al., 2000), but currently risk 178 being discarded in favor of increasingly uniform cropping systems (Gianessi, 2013). Despite the 179 appeal of monoculture production systems, pockets of smallholder farmers in various parts of the 180 world still maintain diversified cassava systems (de Carvalho et al., 2009). 181 9 As cassava cropping systems shift towards simplified management regimes, they increasingly 182 require interventions to build in resilience and safeguard ecosystem functioning. Intercropping is 183 the production of two or more crops in the same field at the same time, augmenting structural 184 complexity and diversity (Andrews and Kassam, 1976). The introduction of a second crop can 185 take many forms, with spatial designs that are additive, substitutive, or a combination of both. 186 Additive designs maintain the same spatial arrangement as in monoculture, but add an intercrop 187 species for all or part of the production cycle. Substitutive designs entail the removal of 188 individual plants or rows of plants and replacement with the intercrop species. While 189 intercropping persists in subsistence or low-input, resource-limited farming systems, it is 190 commonly under-valued (Altieri, 2004). Similarly, there exists as yet unexploited potential to 191 pursue the development of intercropping strategies tailored specifically to highly-intensified 192 cropping systems (Andrade et al., 2012). 193 Diversification tactics in general, and inter-cropping specifically, are known to enhance overall 194 system productivity, while augmenting stability, resilience, and ecological sustainability 195 (Vandermeer, 1989; Nicholls and Altieri, 2004; Letourneau et al., 2011; Lin, 2011). 196 Intercropping may also be effective in improving water infiltration and storage, increasing 197 carbon sequestration, reducing soil erosion, and contributing to ecological pest, weed, and 198 disease management (Brooker et al., 2015; Bedoussac et al., 2015). Recent research suggests that 199 on-farm diversification supports an array of provisioning and regulating ecosystem services, 200 especially within tropical terrestrial systems (Kremen and Miles, 2012; Oliver et al., 2015; 201 Lundgren and Fausti, 2015). Although intercropping may contribute to solutions for some of the 202 most pressing issues in global agriculture and biodiversity conservation, quantitative syntheses of 203 the existing research are lacking (Kremen and Miles, 2012). 204 10 Syntheses focusing on individual intercrops are relatively uncommon (Malézieux et al., 2009), 205 with examples including legumes as intercrops within cereals (including soft wheat, durum 206 wheat and barley; Bedoussac et al., 2015) and maize (Sileshi et al., 2008). Mutsaers et al. (1993) 207 conducted a comprehensive review of intercropping practices for cassava, focusing largely on 208 productivity measures, but including several observations on the provision of other ecosystem 209 services; in particular reduction of weeds and erosion through increased canopy cover. The 210 authors noted importantly that the bulk of cassava research (including global breeding efforts) 211 focus nearly exclusively on monoculture settings. Building on the observations of authors like 212 Mutsaers et al. (1993), we apply the lens of ecosystems services specifically to cassava 213 intercropping systems. In this study we a) carry out a global literature synthesis, across systems, 214 components, management strategies, agro-ecozones, and field-level bio-physical conditions, b) 215 evaluate a broad range of provisioning and regulating ecosystem services, and c) employ the 216 formative concept of ecosystem bundles (Bennett et al., 2009) to evaluate trade-offs and 217 synergies for specific crop associations. Ecosystem service bundles provide a valuable tool for 218 simultaneously evaluating the interactions of multiple ecosystem services in different settings, 219 allowing for the detection of trends or patterns of interaction between services (Bennett et al., 220 2009; Rausepp-Hearne, 2010). By grouping the effects on ecosystem services of production 221 systems with different intercrop components, we take a broad view of the trends in ecosystem 222 services in cassava intercropping systems. 223 5. Literature review and analysis 224 A literature review was employed to evaluate cassava intercropping, to assess the existing 225 evidence for the impacts of intercropping on ecosystem services in cassava production systems, 226 11 and to extract and compare findings on the functioning of these services in contrasting 227 intercropping and monoculture settings. The cassava intercropping literature covers a wide array 228 of systems and geographies, with experiments at differing spatial scales, temporal durations, and 229 levels of scientific rigor. Literature was obtained in July 2015 by searching Web of Science using 230 the keywords ‘cassava’ OR ‘Manihot esculenta’ AND ‘intercrop’ OR ‘polyculture’. Relevant 231 studies were selected in which a) cassava was a focal crop, b) intercropping occurred with both 232 spatial and temporal overlap, c) publication occurred in a peer-reviewed journal or in reports of 233 established research centers. The resulting literature was augmented by references cited in the 234 primary literature. 235 A total of 189 references were found (complete list available in online supporting material), of 236 which 170 investigated intercropping for one or more ecosystem services response variables; the 237 remainder being reviews not specific to cassava or reports of intercropping with no experimental 238 data. A total of 20 studies were reviews of intercropping theory or mechanisms not attached to 239 any geographic location. Publications covered the 1975-2015 time period and originated from 27 240 countries. Overall 63 % of cases evaluated intercropping systems with only 2 components, a 241 further 26 % evaluated three component systems, and the remainder investigated various more 242 complex arrangements. In 62 % of cases the intercropping system included a legume. Only 17 % 243 of experiments combined cassava with perennial species alone, while a further 21 % of 244 combinations included both a perennial and an annual, and the remainder with only annual 245 species. Of the 330 species combinations across these studies, 122 included maize (Zea mays L.), 246 52 included cowpea (Vigna unguiculata L. Walp), 48 included trees, 38 included peanut (Arachis 247 hypogaea L.), and 36 included grass/forages. Other intercrops, including rice (Oryza sp.), 248 soybean (Glycine max L. Merrill), and variety mixtures, were less frequently mentioned. 249 12 Intercropping schemes with annual crops generally took advantage of the initial space provided 250 by the establishment of the relatively long-duration cassava crop. Studies were classified 251 according to type of ecosystem service (TEEB, 2010) evaluated (provisioning, regulating, 252 cultural) (Fig. 2). 253 For further analysis of the effects on ecosystem services, a subset of papers was selected for 254 the inclusion of an appropriate cassava monoculture control, robust methods and description of 255 data, examination of land productivity expressed as land equivalent ratio (LER) or area-time 256 equivalent ratio (ATER), soil services (soil cover, erosion, changes in content of N,P,K, organic 257 matter, or earthworm activity), water services (infiltration, runoff, soil moisture content), and 258 pest regulation or disease control. Only 95 studies met all of the above criteria (for information 259 about their geographic origin see Fig. 1c). We analyzed these studies using vote-counting and 260 synthesized multiple independent studies by summing the numbers of (statistically significant) 261 positive and negative effects. More statistically powerful syntheses based on weighted 262 combination of effects are recognized, but quality and applicability of meta-analysis in 263 agronomy has often been questioned (Philibert et al., 2012), and its application to intercropping 264 issues to date remains scarce (Brooker et al., 2015). Robust analysis of ecosystem services in 265 intercropping systems (based on historical published results) will require greater understanding 266 of trends in research and findings, in order to guide the formulation of analytical methods and 267 approaches specific to this application. Considering the lack of directly comparable measures for 268 many of the ecosystem services reported and the absence of recent systematic reviews on this 269 topic, the authors selected vote-counting as a first measure for compiling an overview of the 270 existing research (Cooper, 1998). Vote-counting is a coarse method of evaluation that does not 271 attempt to generate composite effect sizes, but does permit making comparisons across a wide 272 13 range of indicators and variables. For studies in which data were solely presented in graphical 273 form, data were extracted using WebPlotDigitizer software (Rohatgi, 2011). Due to the common 274 practice of reporting multiple separate experiments in a single journal article, data was extracted 275 from each ‘experiment’. In cases where the intercropping arrangement was kept constant but 276 another variable varied (for example, fertilizer application rate or management scheme), the 277 range is represented by horizontal bars. For cases in which a single journal article presented 278 results from completely separate experiments, these were represented as separate points. For 279 multi-year studies, values were averaged over the whole study period. Non-significant results 280 appear on the bisector. For vote-counting only the directionality of results was considered, with 281 differentiation between 4 categories: benefit, dis-benefit, mixed, and no effect. Studies in which 282 no statistically-significant effects of intercropping were reported were catalogued under the ‘no 283 effect’ category, while those with significant yet inconsistent effects (e.g., between years, 284 locations, climatic conditions, soil types) were listed as ‘mixed effect’. 285 6. Ecosystem service bundles in diversified systems 286 6.1. Land use efficiency: LER and ATER 287 A large share of intercropping studies covered provisioning services (Fig. 2). As proxies for 288 land productivity we used two well-established measures: LER and ATER. LER is a common 289 yardstick for measuring relative land use and is widely employed in intercropping (Bedoussac 290 and Justes, 2011), calculated as a ratio of the relative land area required when growing sole crops 291 to produce the yield from an intercrop (Willey and Osiru, 1972). Due to the prolonged growth 292 period of cassava, the disparity between cultivation cycles of component crops can lead to an 293 overestimation of intercropping advantage (Hiebsch and McCollum, 1987; Fukai, 1993). Hence, 294 14 an alternative measure, ATER, which calculates the sum of the relative yields of the intercrop 295 components corrected for the differences in duration of growing period, may be more appropriate 296 (Hiebsch and McCollum, 1987; Fukai, 1993; Mutsaers et al., 1993). Despite this, LER remains a 297 more often reported metric in intercropping studies (Bedoussac and Justes, 2011), and for this 298 reason was the focus of our literature review. Overall, 43 and 17 measures of LER and ATER 299 were reported in 30 and 7 studies, respectively. 300 A positive relationship (represented by a ratio above 1) was found between intercropping and 301 overall system productivity, with an overall range from 0.79 – 1.84. LER measures were above 302 parity in nearly all cases (37/43), with consistent over-yielding observed in a number of species 303 combinations (Fig. 3). One notable exception is pigeonpea (Cajanus cajan), for which average 304 LER values below 1 were recorded. Maize and bean-based systems performed particularly well 305 due to their relatively short growing seasons, while peanut and rice-based systems give mixed 306 results and are therefore inconclusive. Within a given intercrop system, substantial variation was 307 observed due to environment, varieties, treatments and management practices. ATER measures 308 were also at or above parity in 13/17 reported cases (not shown). 309 LER and ATER do not reflect economic yield. If couched in a system in which cassava is 310 much more valuable on the market than its intercrop, even a small yield penalty may reduce 311 overall profits. In the majority of cases cassava root yield was depressed by intercropping, but 312 intercrop production was able to compensate for these losses. Similar findings have been 313 previously reported for intercropping in a set of different systems and geographies (e.g., Ngwira 314 et al., 2012). Some systems (particularly those not bound by the onset of a rainy season, which 315 can encourage root rot) may also present opportunities for cassava to remain in the field after 316 intercrop harvest, making up for yield losses incurred by early intercrop competition (Tsay et al., 317 15 1988) and possibly benefiting from higher off-peak root prices. This may be increasingly feasible 318 for smallholders gaining early income from harvest of an intercrop. 319 6.2. Pest and disease suppression 320 Plants grown in association regularly benefit from reduced arthropod pest pressure and 321 stronger immune responses through so-called associational resistance mechanisms (Barbosa et 322 al., 2009; Letourneau et al., 2011). Particular plant associations experience a reduced likelihood 323 of detection or vulnerability to herbivores, as affected by a plethora of biotic and abiotic factors. 324 Not only can plant associations enhance abundance or activity patterns of natural enemies, thus 325 benefiting biological control (e.g., Khan et al., 1997), but they can also directly regulate pest 326 densities (e.g., Ben Issa et al., 2016). These benefits are further amplified at larger scales, in 327 which inter- and intraspecific diversity at field or farm level can contribute to a substantial 328 lowering of pest populations and disease incidence (Boudreau, 2013; Lundgren and Fausti, 2015; 329 Gurr et al., 2016). 330 In our global review, 63 % (n=15/24) of experiments reported a decrease in pest indicators 331 within intercropped systems (Fig. 4). Across whitefly species the average population change with 332 intercropping was a reduction of 27 %, while in mealybugs the average was a reduction of 37 %. 333 We combined metrics that reflect pest pressure, including abundance ratios of various 334 developmental stages or feeding damage ratings. Pests of global relevance, such as mealybugs, 335 mites, and whitefly, were affected to varying extents by intercropping. Whitefly and mealybugs 336 experienced population reductions in 73 % and 60 % of cases (n= 11/15 and 3/5), respectively. 337 The effect of intercrops on herbivorous mites was solely studied for the invasive green mite, 338 Mononychellus tanajoa in Africa, reporting slightly lowered (average -5 %) pest populations; 339 however in 3/5 cases no effect was found. Whether the observed population reductions translate 340 16 into economic gains is under-investigated. Few authors have hypothesized about the mechanistic 341 basis for this reduced vulnerability to pests, and the relative contribution of abiotic or biotic 342 factors (including natural enemies) has not been thoroughly assessed (but see Gold et al., 1989, 343 1990). In this review we focused on pest populations, for which the available data are more 344 robust; in the few studies which attempted to evaluate higher-order interactions with predators or 345 parasitoids (Gold et al., 1989; Toko et al., 1996; Schulthess et al., 2004; Onzo et al., 2014), 346 mixed and inconsistent results were reported (not shown). Although past work has failed to adopt 347 holistic, community-level perspectives (e.g., Memmott, 2009; Wood et al. 2015; Wyckhuys et 348 al., 2017a), our findings suggest that further research is required into the myriad ways in which 349 diversification can enhance crop resilience to pest attack. 350 The impact of intercropping on arthropod pest suppression has direct implications for 351 incidence, virulence and spread of insect-vectored diseases, such as cassava mosaic disease 352 (CMD). Transmitted by different species of whitefly, cassava geminiviruses are debilitating 353 pathogens of cassava worldwide and cause important productivity losses in Africa and South 354 Asia. We noted a consistently beneficial effect of intercropping on CMD, with all five cases 355 reporting a 10-40 % reduction in disease incidence in diversified plots (data not shown) 356 (Agbobli, 1987; Fondong et al., 1982; Night et al., 2011). Addition of an intercrop affects 357 pathogen-host-vector interactions through changes in plant morphology and system complexity, 358 resulting in behavioral modification of the insect vector, and subsequent changes in temporal and 359 spatial aspects of disease spread (Fondong et al., 2002; Night et al., 2011). Similar trends were 360 observed for non-viral diseases, such as cassava bacterial blight (n= 2/3 studies; data not shown). 361 As plant pathogens are propagated by wind, rainfall, or soil, an added intercrop can alter 362 bacterial disease dynamics by impeding infection, disease development or dispersal (Gurr et al., 363 17 2016). Analysis of these encouraging results should be tempered by the limited attention the 364 subject has received in cassava, and the likely effects of variable management, genetic, and 365 abiotic factors or the interplay with resident pest populations (e.g., Wyckhuys et al., 2017b). 366 Despite those interfering factors, intercropping may bring about significant farm-level savings as 367 multiple cassava biotic stressors inflict tangible yield losses (e.g., Nwanze, 1982; Legg & 368 Fauquet, 2004). 369 6.3. Soil- and water-regulating services 370 Our review included 21 studies examining soil variables covering a range of edaphic 371 parameters including measures of soil fertility, erosion, ground cover, moisture content, water 372 infiltration, soil macro-fauna, and organic matter levels (Fig. 5). Excessive erosion has cascading 373 negative effects on carbon cycling, results in substantial nutrient effluxes, and has compounding 374 impacts on a host of soil properties (Quinton et al., 2010; Powlson et al., 2011). Intercropping 375 brought about sharp reductions in erosion levels in a wide range of biophysical settings (n=8/9), 376 with levels regularly halved and beneficial impacts strongly modulated by management tactics 377 (e.g., cultivation, sowing or harvesting timing). This is of primary importance as cassava fields 378 on steep slopes can lose topsoil at a staggering rate of 221 tons per annum (Pimentel et al., 379 1995); soils that are effectively ‘non-renewable’ over human timescales. 380 Little effect was observed in studies investigating N, P, K, pH, or soil organic matter. While 381 legume intercrops increase overall biomass production, contribute to C sequestration, and help 382 meet the nitrogen needs of the standing crop (Bedoussac et al., 2015; Sileshi et al., 2008), this 383 was not reflected in the results of the present review. In short-term experiments, intercropping 384 did not contribute to changes in nutrient storage or soil carbon stocks. Many of the included 385 studies had conspicuously short durations by the standards of soil science (3 years or less). Due 386 18 to the often multigenerational scale of processes involved with soil fertility regulation (Powlson 387 et al., 2011), closer scrutiny should be paid to long-term records of nutrient balances, analysis 388 methodology, and incorporation of surface crop residues. When residues are removed soil 389 benefits are expected to be minimal (Lal, 2010; Makinde et al., 2006). Over the long term, 390 gradual accumulation of organic matter is expected if sufficient crop biomass is reincorporated 391 into the field, mitigating one of the key constraints to cassava crop productivity. 392 Soils are dynamic systems in which decomposition of organic matter occurs through diverse 393 faunal communities (e.g., Bardgett & van der Putten, 2014). Prolonged vegetative cover 394 maintains structure and function of trophic soil food webs and helps to explain the important 395 increases of earthworm activity in intercropped systems (Curry, 2004; Fig. 5). The contribution 396 of diversification to microbially-mediated nutrient cycling processes is also expected to be 397 positive (Brooker et al., 2015), but requires further investigation in cassava-based systems. 398 Cassava’s particularly low P demand and high use efficiency is a result of an efficient obligate 399 symbiosis with P-scavenging mycorrhizae, making soil health particularly salient to maintaining 400 robust production (Howeler et al.,1982). 401 In six of the seven cases of water-related services reported, the effects of intercropping were 402 considered beneficial, with no effect in the seventh case (Figure 6). Soil moisture content and 403 infiltration were either not affected or increased, while runoff was reduced by 59 % in the single 404 study evaluating this metric (Ghosh et al., 1989). A significant gap in research is evident in that 405 none of the studies evaluated investigated water-related services in cassava - grain legume 406 systems, despite this being one of the most commonly promoted intercrops for cassava (n=40). 407 Lastly, with increased soil moisture and water infiltration rates, judicious intercropping systems 408 19 may be increasingly adaptable to changes in climate as they possess several key attributes to 409 sustain productivity under prolonged drought conditions. 410 6.4. Composite measures of ecosystem function 411 In the above sections, we demonstrate how intercropping helps to sustain specific ecosystem 412 services. We compiled these data to visualize how integration of a specific companion crop (or 413 plant family) contributes to provision of a bundle of ecosystem services in the most commonly 414 reported systems. The concept of ecosystem service bundles takes into account service trade-offs 415 and synergies (Bennett et al., 2009), to provide a balanced picture of system-level benefits and 416 costs. Four common intercropping systems were compared by vote-counting of an ecosystem 417 service bundle in Fig. 6. Vote-counting was undertaken at the global scale. No obvious trends 418 were detected between specific ecosystem services and geographic location. Due to the paucity 419 of studies on certain ecosystem services (e.g., soil microfauna, fertilizer use efficiency) and 420 imbalance of geographic distribution of research (see Figure 1c), the present study cannot draw 421 any conclusions regarding geographic trends. Nevertheless, overall benefits were identified in 422 five ecosystem services, as ranked under supporting, regulating, and provisioning service 423 categories (i.e., pest regulation, disease control, LER, soil- and water-related services) (see 424 Bommarco et al., 2013). Though the small number of studies for particular systems (e.g., water 425 and disease control for grass systems) precluded drawing broader generalizations, the following 426 exceptions were recorded: 1) pest control with the addition of maize (no effect in 8/16 studies), 427 2) LER under legume systems (mixed results in 8/20 studies), and 3) soil-based services for tree 428 intercrops (no effect in 6/12 studies). Despite these anomalies, our work illuminates the under-429 recognized role of intercropping for ecological remediation within degraded settings. Human-430 mediated recovery of agro-ecosystems could concurrently help to restore ecological functioning, 431 20 to rebuild crop yields and to play a role in the on-farm conservation of biodiversity; a strategy 432 which has received no scientific attention in the case of cassava. 433 Benefits of intercropping are widely thought to be highly variable, context-specific, and 434 dependent upon management and crop components (Brooker et al., 2015). Though not explicitly 435 addressed in our study, management factors and genotype x environment interactions do indeed 436 shape the performance of intercropping systems, and in many cases make the difference between 437 relative advantage and disadvantage. Despite certain biases, our study shows that ecosystem 438 service bundles are sustained with a diverse range of companion crops in cassava systems, with 439 25 positive impacts vs. 3 negative ones for maize (total n= 43), 5 vs. 1 for other Poaceae (total 440 n= 10), 23 vs. 3 for four species of grain legumes (total n= 40), and 9 vs. 0 for trees (total n= 24), 441 respectively. Half of the global studies on maize intercrops showed no significant effects for pest 442 suppression, while 6 (out of 16) reported positive impacts. Land productivity ratios of the 443 cassava-grain legume systems include studies with pigeonpea, all of which were conducted at a 444 single location in Australia. While these trials may hint at incompatibility of cassava with 445 perennial legumes, they may not be representative of the potential performance of these systems 446 in other geographical and agro-ecological settings. The comparatively weak impact of trees on 447 soil parameters may be due to variable spatial and temporal coverage, methodological effects of 448 studies focused primarily on designing over-yielding forage or mulch systems, and the inclusion 449 of a wide range of tree types and species. Variability in spatial and temporal coverage may be 450 particularly important in tree-based systems, as studies were commonly done with a range of 451 naturally-occurring trees (with inherent seasonal leaf shedding) at close proximity to cassava 452 plantations. Tree species included leguminous species, such as Flemingia macrophylla (Willd.) 453 Merr., Gliricidia sepium (Jacq.) Steud, and Leucaena leucocephala (Lam.) de Wit, and non-454 21 leguminous species such as Eucalyptus spp., and Cocos nucifera. Tree mixtures are also 455 sometimes employed; one study from Brazil included a mixture of 37 species of indigenous trees 456 intercropped with cassava (Daronco et al., 2012). Perennials have been heavily promoted to 457 build underground carbon storage, soil health and fertility in degraded farming systems in Africa, 458 focusing on the use of N-fixing legumes (Glover et al., 2012). The outspoken variability in the 459 effects of different tree species under particular agro-climatic or biophysical conditions suggests 460 that (trait-based, locality-specific) decision-support systems to choose the right companion crops 461 for complementation of one or more particular ecosystem services likely have considerable 462 merit. 463 Caution needs to be taken when interpreting the results of this study, not solely due to our 464 analytic approach (i.e., vote-counting) but also due to a range of other factors. Many agronomic 465 studies are ‘answer-driven,’ and seek solutions to production problems. Designed to identify 466 ‘improved’ production systems, these risk over-representing positive results, with experiments 467 guided by evidence from systems designed by farmers and practitioners. Of further importance is 468 the risk of publication bias, in which experiments reporting significant results are favored for 469 submission and/or publication (Dickersin, 1990). Cultural ecosystem services, such as traditional 470 uses and networks (Coomes, 2010), food cultures (Lancaster et al., 1982), and local perceptions 471 (Kamau et al., 2011) that could have been captured through participatory approaches, were 472 generally underreported in the literature. Despite these pitfalls our findings echo those of several 473 comprehensive global reviews focused on other crops (Andow, 1991; Malézieux et al., 2009; 474 Kremen and Miles, 2012). 475 22 7. Conclusion 476 Intercropping can add complexity and diversity to the world’s agro-production systems. With 477 roots in traditional systems, intercropping holds considerable potential in cassava production 478 systems if attention is given to key barriers to broad-scale adoption. Intercropping has received a 479 fair amount of research attention, but past work primarily consists of on-station trials with a 480 nearly exclusive emphasis on identifying potential for over-yielding. Our work documents the 481 value of this practice to wider ecosystem functioning in a crop of global significance. 482 The present study elucidates intercropping’s potential to meet growing food production needs 483 with minimal environmental costs. Over a wide array of companion plants diversified cassava 484 systems can enhance levels of land productivity and sustain key ecosystem functions and 485 services. Benefits are not only concurrent while intercrops are in place (as shown in this study), 486 but importantly also deliver medium to long-term effects on soil fertility, erosion prevention, and 487 both pest and beneficial insect communities. These long-term benefits are particularly important 488 for soil conservation and erosion prevention, as soils are effectively ‘non-renewable’ over human 489 timescales. Historical evidence suggests that intercropping can be adapted to smallholders’ 490 diverse biophysical and socio-economic contexts. However, the allure of the ecosystem service 491 benefits must be counterbalanced against general reductions in focal crop yields, additional labor 492 costs, and economic considerations. Adoption can be hampered by challenges related to e.g., 493 mechanization, labor requirements, incentive systems, and ultimately the overall economic 494 productivity of a crop producing low-value products. Component crops must each perform well 495 within the agro-ecological niche in which the intercropping system is found, and factors such as 496 spacing, arrangement, input types and levels, and relative harvest timing may significantly 497 influence overall productivity. Nevertheless, even with the existing compelling drivers of 498 23 monoculture, a research-informed evaluation of overall system profitability and the development 499 of optimized management regimes will increase the practicability of diversification schemes. 500 Our exercise also suggests a need for methodological, experimental, and conceptual 501 approaches linking productivity, system resilience, and broader environmental preservation 502 within the cassava agro-ecosystem. A weak mechanistic understanding of variable context-503 dependent ecological processes (e.g. plant-plant, and plant-soil interactions) presently constitutes 504 one of several barriers to more widespread promotion and adoption. The path forward for 505 cassava-based farming systems does not solely lie in advancing technical innovations, but in a 506 combination of policies, institutional engagement, markets, and practices. Interdisciplinary, 507 transdisciplinary, and systems-level approaches will be instrumental for identifying 508 intensification scenarios in which cassava productivity, provision of ecosystem services, 509 biodiversity conservation, and human well-being are all balanced, and in providing (smallholder) 510 farmers with selection aides to evaluate appropriate practices to optimize their unique production 511 realities. 512 8. Acknowledgements 513 This initiative was conducted as part of an EC-funded, IFAD-managed, CIAT-executed 514 program (CIAT-EGC-60-1000004285). Additional funding was also provided through the 515 CGIAR-wide Research Program on Roots, Tubers and Banana (CRP-RTB). 516 517 24 9. References 518 Altieri, M.A., 2004. Linking ecologists and traditional farmers in the search for sustainable 519 agriculture. Front Ecol. Environ. 2, 35-42. 520 Andow, D.A., 1991. Yield loss to arthropods in vegetationally diverse agro-ecosystems. Environ. 521 Entomol. 20(5), 1228-1235. 522 Andrade, J. F., Cerrudo, A., Rizzalli, R.H., Monzon, J.P. 2012. Sunflower–soybean intercrop 523 productivity under different water conditions and sowing managements. Agron. J. 104, 1049-524 1055. 525 Andrews, D.J., Kassam, A.H., 1976. The importance of multiple cropping in increasing world 526 food supplies. ASA Special Publication, American Society of Agronomy. 527 Bai, Z.G., Dent, D.L., Olsson, L., Schaepman, M.E. 2008. Proxy global assessment of land 528 degradation. Soil Use and Management 24, 223-234. 529 Barbier, E.B. 1997. The economic determinants of land degradation in developing countries. 530 Phil. Trans. R. Soc. Lond. B. 352, 891-899. 531 Barbosa, P., Hines, J., Kaplan, I., et al., 2009. Associational resistance and associational 532 susceptibility: having right or wrong neighbors. Annu. Rev. Ecol. Evol. S. 40, 1-20. 533 Bardgett, R.D., van der Putten, W.H. 2014. Belowground biodiversity and ecosystem 534 functioning. Nature 515, 505-511. 535 Bedoussac, L., Journet, E.P., Hauggaard-Nielsen, H., et al., 2015. Ecological principles 536 underlying the increase of productivity achieved by cereal-grain legume intercrops in organic 537 farming. A review. Agron. Sustain. Dev. 35(3), 911-935. 538 Bedoussac, L., Justes, E., 2011. A comparison of commonly used indices for evaluating species 539 interactions and intercrop efficiency: Application to durum wheat-winter pea intercrops. F. Crop 540 Res. 124, 25-36. 541 Ben Issa, R., Gautier, H., Gomez, L., 2016. Influence of neighbouring companion plants on the 542 performance of aphid populations on sweet pepper plants under greenhouse conditions. Agric. 543 For. Entomol. 19, 181-191. 544 Bennett, E.M., Peterson, G.D., Gordon, L.J., 2009. Understanding relationships among multiple 545 ecosystem services. Ecol. Lett. 12, 1-11. 546 Benton, T.G., Vickery, J.A., Wilson, J.D., 2003. Farmland biodiversity: is habitat heterogeneity 547 the key. Trends Ecol. Evol. 18, 182-188. 548 25 Bommarco, R., Kleijn, D., Potts, S. G. 2013. Ecological intensification: harnessing ecosystem 549 services for food security. Trends in Ecol. Evol., 28(4), 230-238. 550 Bossio, D., Geheb, K., Critchley, W. 2010. Managing water by managing land: addressing land 551 degradation to improve water productivity and rural livelihoods. Aricultural Water Management 552 97(4), 536-542. 553 Boudreau, M.A., 2013. Diseases in intercropping systems. Annu. Rev. Phytopathol. 51, 499-519. 554 Brooker, R.W., Bennett, A.E., Cong, W.F., et al., 2015. Improving intercropping: a synthesis of 555 research in agronomy, plant physiology and ecology. New Phytol. 206, 107-117. 556 Burns, A., Gleadow, R., Cliff, J., et al., 2010. Cassava: The drought, war and famine crop in a 557 changing world. Sustainability 2, 3572-3607. 558 Clement, F., Amezaga, J.M., 2008. Linking reforestation policies with land use change in 559 Northern Vietnam: why local factors matter. Geoforum 39, 265-277. 560 Cock, J.H., Franklin, D., Sandoval G., Juri, P., 1979. The ideal cassava plant for maximum yield. 561 Crop Science 19(2), 271-279. 562 Coomes, O.T. 2010. Of stakes, stems, and cuttings: the importance of local seed systems in 563 traditional Amazonian societies. Prof. Geogr. 62(3), 323-334. 564 Cooper, H. 1998. Synthesizing Research: A Guide for Literature Reviews, 3rd edn. Sage 565 Publications, Thousand Oaks, CA. 566 Crosby, Alfred W. 1972.The Columbian Exchange: Biological and Cultural Consequences of 567 1492, Westport: Greenwood Press. 568 Curry, J.P. 2004. Factors affecting the abundance of earthworms in soils. Earthworm Ecology, 569 Volume 9, 113 p. 570 de Carvalho, F.M., Silveira Viana, A.E., Leite Cardoso, C.E., et al., 2009. Sistemas de produção 571 de mandioca em treze municípios da região sudoeste da Bahia. Bragantia 68, 699-702. 572 De Vries, S.C., van de Ven, G.W.J., van Ittersum, M.K., Giller, K.E., 2010. Resource use 573 efficiency and environmental performance of nine major biofuel crops, processed by first-574 generation conversion techniques. Biomass and bioenergy 34, 588-601. 575 Dickersin, K., 1990. The existence of publication bias and risk factors for its occurrence. J. 576 Amer. Med. Assoc. 263(10), 1385-1389. 577 El-Sharkawy, M.A., 2012. Stress-Tolerant Cassava: The Role of Integrative Ecophysiology-578 Breeding Research in Crop Improvement. Open J. Soil Sci. 2, 162-86. 579 26 El-Sharkawy, M.A., 2014. Global warming: causes and impacts on agro-ecosystems productivity 580 and food security with emphasis on cassava comparative advantage in the tropics/subtropics. 581 Photosynthetica 52, 161-178. 582 Emmerson, M., Morales, M.B., Oñate, J.J., et al., 2016. Chapter Two-How Agricultural 583 Intensification Affects Biodiversity and Ecosystem Services. Adv. Ecol. Res. 55, 43-97. 584 FAOSTAT, 2016. Food and Agriculture Organization of the United Nations - Statistics 585 Database. http://faostat3.fao.org/ (accessed 15.05.2016). 586 Fermont, A.M., van Asten, P.J.A., Tittonell, P., et al., 2009. Closing the cassava yield gap: An 587 analysis from smallholder farms in East Africa. Field Crop Res. 112, 24-36. 588 Foley, J.A., Ramankutty, N., Brauman, K.A., et al., 2011. Solutions for a cultivated planet. 589 Nature 478(7369), 337-42. 590 Fondong, V.N., Thresh, J.M., Zok, S., 2002. Spatial and temporal spread of cassava mosaic virus 591 disease in cassava grown alone and when intercropped with maize and/or cowpea. J. 592 Phytopathol. 150, 365-374. 593 Francesconi, W., Nair, P.K.R., Levey, D.J., et al., 2013) Butterfly distribution in fragmented 594 landscapes containing agroforestry practices in Southeastern Brazil. Agroforest. Syst. 87, 1321-595 1338. 596 Fukai, S., 1993. Intercropping - bases of productivity. F. Crop Res. 34, 239-245. 597 Gianessi, L.P., 2013. The increasing importance of herbicides in worldwide crop production. 598 Pest Manag. Sci. 69(10), 1099-1105. 599 Gibbs, H., Ruesch, A.S., Achard, F., et al., 2010. Tropical forests were the primary sources of 600 new agricultural land in the 1980s and 1990s. P Natl. Acad. Sci. USA 107, 16732-16737. 601 Glover, J.D., Reganold, J.P., Cox, C.M., 2012. Plant perennials to save Africa’s soils. Nature 602 489, 359-361. 603 Godfray, H.C.J., Beddington, J.R., Crute, I.R., et al., 2010. Food security: the challenge of 604 feeding 9 billion people. Science 327, 812-818. 605 Gold, C.S., Altieri, M.A., Bellotti, A.C., 1989. The effects of intercropping and mixed varieties 606 of predators and parasitoids of cassava whiteflies (Hemiptera: Aleyrodidae) in Colombia. Bull. 607 Entomol. Res. 79, 115-121. 608 Gold, C.S., Altieri, M.A., Bellotti, A.C., 1990. Effects of intercropping and varietal mixtures on 609 cassava hornworm, Erinnyis ello L. (Lepidoptera: Sphingidae), and the stemborer, Chilomima 610 clarkei (Amsel) (Lepidoptera: Pyralidae), in Colombia. Trop. Pest Manage. 34 (4), 362–367. 611 27 Graziosi, I., Minato, N., Alvarez, E., et al., 2016. Emerging pests and diseases of South‐east 612 Asian cassava: a comprehensive evaluation of geographic priorities, management options and 613 research needs. Pest Manag. Sci. 72(6), 1071-1089. 614 Gurr, G.M., Lu, Z.X., Zheng, X.S., et al., 2016. Multi-country evidence that crop diversification 615 promotes ecological intensification of agriculture. Nature Plants 2, 16014. 616 Hiebsch, C.K., McCollum, R.F., 1987. Area×Time Equivalency Ratio: a method for evaluating 617 the productivity of intercrops. Agron. J. 79, 15-22. 618 Hillocks, R.J., Thresh, J.M., Bellotti, A.C., 2001. Cassava biology, production and utilization. 619 CAB International, Wallingford, UK. 620 Hought, J., Birch-Thomsen, T., Petersen, J., de Neergaard, A., Oelofse, M., 2012. Biofuels, land 621 use change and smallholder livelihoods: A case study from Banteay Chhmar, Cambodia. Appl. 622 Geogr. 34, 525-532. 623 Howeler, R.H., 2015. Sustainable cassava production in Asia for multiple uses and for multiple 624 markets: Proceedings of the ninth regional workshop held in Nanning, Guangxi, China PR, 27 625 Nov-3 Dec 2011. 336 p. International Center for Tropical Agriculture, CIAT. 626 Howeler, R.H., Cadavid, L.F., Burckhardt, E., 1982. Response of cassava to VA mycorrhizal 627 inoculation and phosphorus application in greenhouse and field experiments. Plant Soil 69, 327-628 339. 629 Howeler, R.H., Oates, C.G., Allem, A.C., 2000. Strategic environmental assessment. An 630 assessment of the impact of cassava production and processing on the environment and 631 biodiversity, vol. 5. FAO, IFAD. 632 Hulugalle, N.R., Ezumah, H.C., 1991. Effects of cassava-based cropping systems on physico-633 chemical properties of soil and earthworm casts in a tropical Alfisol. Agric. Ecosyst. Environ. 634 35, 55-63. 635 Jarvis, A., Ramirez-Villegas, J., Herrera Campo, B.V., et al., 2012. Is cassava the answer to 636 African climate change adaptation? Trop. Plant Biol. 5, 9-29. 637 Kamau, J., Melis, R., Laing, M., et al., 2011. Farmers’ participatory selection for early bulking 638 cassava genotypes in semi-arid Eastern Kenya. J. Plant Breed. Crop Sci. 3(3), 44-52. 639 Khan, Z.R., Ampong-Nyarko, K., Chiliswa, P., Hassanali, A., Kimani, S., Lwande, W., Overholt, 640 W.A., Pickett, J.A., Smart, L.E., Woodcock, C.M. 1997. Intercropping increases parasitism of 641 pests. Nature 388, 631-632. 642 Kremen, C., Miles, A., 2012. Ecosystem services in biologically diversified versus conventional 643 farming systems: Benefits, externalities, and trade-offs. Ecol. Soc. 17, 40. 644 28 Lal, R. 2010. Managing soils for a warming earth in a food-insecure and energy-starved world. J. 645 Plant Nutr. Soil Sci. 173, 4–15. 646 Lancaster, P.A., Ingram, J.S., Lim, M.Y., et al., 1982. Traditional cassava-based foods: survey of 647 processing techniques. Econ. Bot. 36(1), 12-45. 648 Laurance, W.F., Sayer, J., Cassman, K.G., 2014. Agricultural expansion and its impacts on 649 tropical nature. Trends Ecol. Evol. 29, 107-116. 650 Legg, J.P., Thresh, J.M., 2000. Cassava mosaic virus disease in East Africa: a dynamic disease in 651 a changing environment. Virus Research, 71, 135-149. 652 Laurance, W.F., Sayer, J., Cassman, K.G., 2014. Agricultural expansion and its impacts on 653 tropical nature. Trends Ecol. Evol. 29, 107-116. 654 Letourneau, D.K., Armbrecht, I., Salguero Rivera, B., et al., 2011. Does plant diversity benefit 655 agro-ecosystems? A synthetic review. Ecol. Appl. 21, 9-21. 656 Lin, B.B., 2011. Resilience in agriculture through crop diversification: adaptive management for 657 environmental change. BioScience 61, 183-193. 658 Lundgren, J.G., Fausti, S.W., 2015. Trading biodiversity for pest problems. Sci. Adv. 1(6), 659 e1500558. 660 Makinde, E.A., Oluwatoyinbo, F.I., Ayoola, O.T. 2006. Intercropping and Crop Residue 661 Incorporation: Effects on Soil Nutrient Status. J. Plant Nutr. 29, 235-244. 662 Malézieux, E., Crozat, Y., Dupraz, C., et al., 2009. Mixing plant species in cropping systems: 663 concepts, tools and models. A review. Agron. Sustain. Dev. 29, 43-62. 664 MAPSPAM, 2016. http://mapspam.info/ (accessed 22.06.2016). 665 McKey, D., Emperaire, L., Élias, M., et al., 2001. Gestions locales et dynamiques régionales de 666 la diversité variétale du manioc en Amazonie. Genet. Sel. Evol. 33, S465-S490. 667 Memmott, J., 2009. Food webs: a ladder for picking strawberries or a practical tool for practical 668 problems? Phil. Trans. R. Soc. B 364(1524), 1693-1699. 669 Moench, M., 1991. Soil erosion under a successional agroforestry sequence: a case study from 670 Idukki District, Kerala, India. Agroforest. Syst. 15, 31-50. 671 Mutsaers, H.J.W., Ezumah, H.C., Osiru, D.S.O., 1993. Cassava-based intercropping: a review. 672 Field Crop Res. 34, 431-457. 673 NEPCon, 2014. Forest-land conversion and conversion timber estimates: Cambodia case study. 674 Nature Ecology and People Consult (NEPCon), Copenhagen, Denmark. 675 29 Ngwira, A.R., Aune, J.B., Mkwinda, S. 2012. On-farm evaluation of yield and economic benefit 676 of short term maize legume intercropping systems under conservation agriculture in Malawi. 677 Field Crop Res. 132, 149-157. 678 Nicholls, C.I., Altieri, M.A., 2004. Designing species-rich, pest-suppressive agroecosystems 679 through habitat management. Agronomy 43, 49-62. 680 Night, G., Asiimwe, P., Gashaka, G., et al., 2011. Occurrence and distribution of cassava pests 681 and diseases in Rwanda. Agric. Ecosyst. Environ. 140, 492-497. 682 Norris, K., Asase, A., Collen, B., et al., 2010. Biodiversity in a forest-agriculture mosaic – The 683 changing face of West African rainforests. Biol. Conserv. 143, 2341-2350. 684 Nwanze, K.F. 1982. Relationships between cassava root yields and crop infestations by the 685 mealybug, Phenacoccus manihoti. Int. J. Pest Man., 28(1), 27-32. 686 Oliver, T.H., Heard, M.S., Isaac, N.J., et al., 2015. Biodiversity and resilience of ecosystem 687 functions. Trends Ecol. Evol. 30, 673-684. 688 Olsen, K.M., Schaal, B.A. 1999. Evidence on the origin of cassava: phylogeography of Manihot 689 esculenta. PNAS 96(10), 5586–5591. 690 Onzo, A., Hanna, R., Toko, M., 2014. Mix-planting pubescent and glabrous cassava affects 691 abundance of Typhlodromalus aripo and its prey mite Mononychellus tanajoa. J. Appl. Entomol. 692 138, 297-306. 693 Philibert, A., Loyce, C., Makowski, D., 2012. Assessment of the quality of meta-analysis in 694 agronomy. Agric. Ecosyst. Environ. 148, 72-82. 695 Pimentel, D., Harvey, C., Resosudarmo, P., Sinclair, K., Kurz, D., McNair, M., Crist, S., Shpritz, 696 L., Fitton, L., Saffouri, R., Blair, R., 1995. Environmental and economic costs of soil erosion and 697 conservation benefits. Science 267, 1117-1123. 698 Poppy, G.M., Chiotha, S., Eigenbrod, F., et al., 2014. Food security in a perfect storm: using the 699 ecosystem services framework to increase understanding. Philos. T. R. Soc. B. 369(1639), 1-14. 700 Powlson, D.S., Gregory, P.J., Whalley, W.R., Quinton, J.N., Hopkins, D.W., Whitmore, A.P., 701 Hirsch, P.R., Goulding, K.W.T., 2011. Soil management in relation to sustainable agriculture and 702 ecosystem services. Food Policy 36, S72-S87. 703 Pypers, P., Sanginga, J.-M., Kasereka, B., et al. 2011. Increased productivity through integrated 704 soil fertility management in cassava–legume intercropping systems in the highlands of Sud-Kivu, 705 DR Congo. F. Crop Res. 120, 76–85. 706 30 Quinton, J.N., Govers, G., Van Oost, K., Bardgett, R.D., 2010. The impact of agricultural soil 707 erosion on biogeochemical cycling. Nature Geosci. 3, 311-314. 708 Raudsepp-Hearne C., Peterson G.D., Bennett E.M., 2010. Ecosystem service bundles for 709 analyzing tradeoffs in diverse landscapes. Proc. Natl. Acad. Sci. USA 107: 5242–7. 710 Reynolds, T.W., Waddington, S.R., Anderson, C.L., et al., 2015. Environmental impacts and 711 constraints associated with the production of major food crops in Sub-Saharan Africa and South 712 Asia. Food Security 7, 795-822. 713 Rohatgi, A., 2011. WebPlotDigitizer version 3.9. 714 Sayer, J., Cassman, K.G., 2013. Agricultural innovation to protect the environment. Proc. Natl. 715 Acad. Sci. USA 110, 8345-8348. 716 Schulthess, F., Chabi-Olaye, A., Gounou, S., 2004. Multi-trophic level interactions in a cassava-717 maize mixed cropping system in the humid tropics of West Africa. Bull. Entomol. Res. 94, 261-718 272. 719 Sileshi, G., Akinnifesi, F.K., Ajayi, O.C., et al., 2008. Meta-analysis of maize yield response to 720 woody and herbaceous legumes in sub-Saharan Africa. Plant Soil 307, 1-19. 721 TEEB, 2010. The Economics of Ecosystems and Biodiversity: Ecological and Economic 722 Foundations. Taylor & Francis, Oxford, UK. 723 Tews, J., Brose, U., Grimm, V., Tielbörger, K., Wichmann, M.C., Schwager, M., Jeltsch, F., 724 2004. Animal species diversity driven by habitat heterogeneity/diversity: the importance of 725 keystone structures. J. Biogeogr. 31, 79-92. 726 Tittonell, P., Giller, K.E., 2013. When yield gaps are poverty traps: The paradigm of ecological 727 intensification in African smallholder agriculture. Field Crop Res. 143, 76-90. 728 Toko, M., Yaninek, J.S., O’Neil, R.J., 1996. Response of Mononychellus tanajoa (Acari: 729 Tetranychidae) to cropping systems, cultivars, and pest interventions. Environ. Entomol. 25, 730 237-49. 731 Tsay, J. S., Fukai, S., and Wilson, G. L. 1988. Effects of relative sowing time of soybean on 732 growth and yield of cassava/soybean intercropping. F. Crop. Res. 19, 227–239. 733 Tscharntke, T., Clough, Y., Wanger, T.C., et al., 2012. Global food security, biodiversity 734 conservation and the future of agricultural intensification. Biol. Conserv. 151(1), 53-9. 735 Valentin, C., Agus, F., Alamban, R., et al., 2008. Runoff and sediment losses from 27 upland 736 catchments in Southeast Asia: impact of rapid land use changes and conservation practices. Agr. 737 Ecosyst. Environ. 128, 225-238. 738 31 van Ittersum, M.K., van Bussel, L.G.J., Wolf, J., et al., 2016. Can sun-Saharan Africa feed itself? 739 PNAS 113(52), 14964-14969. 740 Vandermeer, J.H., 1989. The ecology of intercropping (Cambridge University Press). 741 von Maltitz, G.P., Haywood, L., Mapako, M., Brent, A., 2009. Analysis of opportunities for 742 biofuel production in sub-Saharan Africa. Environment brief. CIFOR, Bogor, Indonesia. 743 Vurro, M., Bonciani, B., Vannacci, G., 2010. Emerging infectious diseases of crop plants in 744 developing countries: impact on agriculture and socio-economic consequences. Food Security 2, 745 113-132. 746 Waddington, S.R., Li, X., Dixon, J., et al., 2010. Getting the focus right: production constraints 747 for six major food crops in Asian and African farming systems. Food Security 2(1), 27-48. 748 Wargiono, J., Widodo, Y., Hadi Utomo, W., 2000. Cassava agronomy research and adoption of 749 improved practices in Indonesia – major achievements during the past 20 years. Cassava’s 750 Potential in the 21st Century: Present Situation and Future Research and Development Needs. 751 6th Regional Workshop, eds. Howeler, R., Tan, S. International Center for Tropical Agriculture, 752 CIAT. 753 Willey, R.W., Osiru, D.S.O. 1972. Studies on mixtures of maize and beans (Phaseolus vulgaris) 754 with particular reference to plant population. J. Agri. Sci. 79, 517-529. 755 Wood, S.A., Karp, D.S., DeClerck, F., Kremen, C., Naeem, S., Palm, C.A., 2015. Functional 756 traits in agriculture: agrobiodiversity and ecosystem services. Trends Ecol. Evol. 30(9), 531-539. 757 Wyckhuys, K.A.G., Burra, D.D., Tran, D.H. et al. 2017a. Soil fertility regulates invasive 758 herbivore performance and top-down control in tropical agro-ecosystems of Southeast Asia. 759 Agric. Ecosyst. Environ 249, 38–49. 760 Wyckhuys, K.A.G., Graziosi, I., Burra, D.D., Walter, A.J. 2017b. Phytoplasma infection of a 761 tropical root crop triggers bottom-up cascades by favoring generalist over specialist herbivores. 762 PLoS ONE 12(8): e0182766. 763 Zhou, A., Thomson, E., 2009. The development of biofuels in Asia. Appl. Energ. 86, S11-S20. 764 32 Figures 765 766 767 † World Bank, Global Poverty Working Group. Rural poverty headcount at national poverty lines. Accessed June 2015. http://data.worldbank.org/ 768 ‡Calculated from 2014 cassava area data (FAOSTAT, 2016) total agricultural area (FAO, electronic files and web site), online table at 769 http://wdi.worldbank.org/table/3. ‘Arable land’ using FAO definition. 770 771 Figure 1. Trends in cassava production and research: cassava production has intensified over the past 5 772 decades, both in terms of area and yield (a). Countries across the developing-world tropics have a high 773 degree of dependence on cassava in their agricultural systems and high levels of rural poverty as seen in 774 (b) where each bubble represents a single country (FAOSTAT, 2016). Studies on cassava intercropping 775 originate from a wide geographic area (c). The green backdrop indicates harvested cassava area in 2014 776 (MAPSPAM, 2016), while bubble size indicates total number of studies and blue segments indicate the 777 proportion selected for final ecosystem services analysis. 778 779 33 780 Figure 2. Categories of ecosystem services investigated by cassava intercropping literature. Numbers in 781 brackets indicate the overall number of studies identified. Complete list of references available in online 782 supplementary material. 783 784 34 785 786 Figure 3. Land Equivalent Ratios of 2-crop combinations reported in cassava intercropping literature. 787 Diamond-Africa, Square-Asia, Triangle-Americas, Circle-Oceania/Pacific. Horizontal bars indicate ranges 788 for studies reporting multiple values. In cases where the intercropping arrangement was kept constant but 789 another variable varied (for example, fertilizer application rate or management scheme), the range is 790 represented by horizontal bars. Results from separate experiments within a given study are represented 791 as separate points. For multi-year studies, values were averaged over the whole study period. 792 793 794 35 795 796 Figure 4. Means and ranges of effects on pest indicators reported in cassava intercropping literature 797 relative to their respective monoculture controls. Nonsignificant results are located on the median line and 798 horizontal bars indicate ranges found in studies with multiple means reported. Studies that contained 799 separate ‘experiments’ are presented as separate points. Diamond-Africa, Square-Asia, Triangle-800 Americas, Circle-Oceania/Pacific. 801 802 803 36 804 Figure 5. Means and ranges of effects on soil and water indicators reported in cassava intercropping 805 literature relative to their respective monoculture controls. pH is measured as the percentage difference in 806 acidification of soils under monoculture and intercrop, with negative percentages indicating that 807 intercropping leads to less acidic soil for this particular study. SOC= Soil organic carbon, OM=Organic 808 matter, C factor of USLE= crop management factor of universal soil loss equation. Cover % indicates 809 percentage of total soil coverage achieved. Nonsignificant results are located on the median line and 810 horizontal bars indicate ranges found in studies with multiple means reported. Diamond-Africa, Square-811 Asia, Triangle-Americas, Circle-Oceania/Pacific. 812 37 813 814 Figure 6. Vote-count of study findings for key ecosystem services and key intercrop species in cassava. 815 Numbers on bars indicate number of studies, N= total studies evaluated for each trait/crop combination. 816 *Poaceae excepting maize, †Soybean, peanut, cowpea, and pigeonpea. 817 818 819 820