diff --git "a/data/part_5/06fcbddfbbdf3be827da370f4ca24202.json" "b/data/part_5/06fcbddfbbdf3be827da370f4ca24202.json"
new file mode 100644--- /dev/null
+++ "b/data/part_5/06fcbddfbbdf3be827da370f4ca24202.json"
@@ -0,0 +1 @@
+{"metadata":{"id":"06fcbddfbbdf3be827da370f4ca24202","source":"gardian_index","url":"https://cgspace.cgiar.org/rest/bitstreams/1c42a7dd-650e-47c3-ae60-d96a4eb4969b/retrieve"},"pageCount":88,"title":"","keywords":[],"chapters":[{"head":"Executive Summary","index":1,"paragraphs":[{"index":1,"size":2,"text":"The problem"},{"index":2,"size":16,"text":"Livestock systems research (LSR) at the International Livestock Research Institute (ILRI) sought to answer two questions:"},{"index":3,"size":16,"text":"• What are the major livestock systems in the sub-Saharan Africa tropics and subtropics (Ruthenberg, 1980)?"},{"index":4,"size":16,"text":"• What technical and organizational changes can be introduced into these systems to make them productive?"},{"index":5,"size":25,"text":"This chapter reports the answers of decades of research at ILRI, its predecessors and its principal partners to these questions. The chapter continues to ask:"},{"index":6,"size":12,"text":"• What have been the scientific impacts of LSR since the 1970s?"},{"index":7,"size":12,"text":"• What have been the development impacts of LSR since the 1970s?"},{"index":8,"size":16,"text":"• Can the development impacts of LSR be distinguished from long-term trends in African livestock systems?"}]},{"head":"Research spending","index":2,"paragraphs":[{"index":1,"size":110,"text":"ILRI 1 and one of its predecessors, the International Livestock Centre for Africa (ILCA), spent about US$212 million (in 2015 US$) on LSR from 1975 to 2018. This was roughly 22% of ILCA spending before 1995 and about 11% of ILRI spending from 1995 to 2018. The majority of this spending was on ruminants in sub-Saharan Africa, mainly in systems where cattle were the dominant stock. A crude estimate of expenditure by system was 80% on mixed systems and 20% on grazing/rangelands during the 1975-1994 era of ILCA and the International Laboratory for Research on Animal Diseases (ILRAD). No such estimate by system is possible for the ILRI era ."}]},{"head":"Scientific impact","index":3,"paragraphs":[{"index":1,"size":7,"text":"The scientific impact of LSR was substantial:"},{"index":2,"size":27,"text":"• Mapping systems. The greatest scientific impact of LSR was in mapping systems and identifying their technical possibilities, potential for growth and poverty reduction, and research priorities."},{"index":3,"size":51,"text":"• Understanding transversal and system studies. The second most important scientific impact was in understanding the evolution of animal production systems, grazing and mixed alike, to prepare appropriate technical interventions, to avoid repeating past failures in technology generation and transfer, and to identify promising sites for technical transfer from other regions."},{"index":4,"size":38,"text":"• Estimating production parameters. Estimating input and output relationships from field data in the systems studies was feasible for the first time, using the work started in the 1970s by ILCA and continuing to the present by ILRI."},{"index":5,"size":11,"text":"• Creating and defending a new view of African grazing systems."},{"index":6,"size":50,"text":"• Understanding the value of mobility in grazing systems. The various field studies in sub-Saharan Africa showed the importance of mobility in grazing and revealed the costs of new policies and organizations that would compromise mobility. This includes understanding of seasonal and annual variability and the opportunistic character of pastoralism."},{"index":7,"size":17,"text":"• Understanding complementarities in mixed wildlife and domestic livestock systems. This was most important in East Africa."},{"index":8,"size":40,"text":"• Understanding the roles of crop residues and manure in soil nutrient cycling. These included quantifying the benefits and tradeoffs among uses of crop residues and manure in mixed systems as functions of crop and livestock potential in given environments."},{"index":9,"size":52,"text":"• Correcting the neglect of the feed value of crop breeding programmes in Africa by the crop research centres. ILRI work in India and Ethiopia, led by Michael Blümmel, has done much to correct this neglect (see Chapter 14, this volume) and has shown the value of complementary field and station research."},{"index":10,"size":44,"text":"• Quantitative modelling of livestock systems. Models of pastoralism often generated scientifically reliable projections of the outputs of policy experiments. They have succeeded in establishing new research lines, notably 'livestock as a pathway out of poverty' and the study of livestock and climate change."},{"index":11,"size":44,"text":"• Defining how mechanization with animals evolved. This work had an immense scientific impact but little positive development impact in the sense of inducing new investments in mechanization. There was some additional scientific impact in elucidating why some apparently promising technical changes had failed."},{"index":12,"size":13,"text":"• Ley farming in the Mediterranean and in the highlands of East Africa."},{"index":13,"size":27,"text":"• Other efforts to introduce planted forages in smallholder systems, such as pasture improvement in arid areas, or into beef production in the humid and subhumid tropics."},{"index":14,"size":14,"text":"• Smallholder dairying models borrowed from temperate areas 2 and introduced into the tropics."},{"index":15,"size":15,"text":"• Station and farm studies of nutrition and other factors affecting output from working animals."}]},{"head":"Development impact","index":4,"paragraphs":[{"index":1,"size":68,"text":"The development impact of LSR was limited. It consisted of: (i) defining the economic weight and rationality of pastoralism 3 as a means of defending pastoralists' livelihoods against the incursion of crops and wildlife, corruption and bad policy; (ii) valuing land rights of pastoralists; (iii) defining the conditions in which external technologies could be introduced into mixed systems; and (iv) testing and validating improved dairying models for smallholders."}]},{"head":"Defence of pastoralists' interests","index":5,"paragraphs":[{"index":1,"size":123,"text":"The systems studies of ILCA, ILRI and their many collaborators demonstrated the economic rationality of extensive grazing systems in sub-Saharan Africa. Homewood's contemporary book (Homewood, 2008) followed by that of Catley et al. (2013) are thorough analyses of how the new view has contributed to the defence of pastoralists' interests and avoided economic losses to the vulnerable groups. At the same time, such work called into question the 'inevitable overgrazing' critique of pastoralism. In so doing, this research thoroughly discredited, even if it did not eliminate, the 'mainstream view' of pastoralism. The new view of pastoralism, supported by research throughout sub-Saharan Africa, reduced some of the policy threats to those systems, including forced settlement, confinement into inviable grazing schemes and dispossession by elites."}]},{"head":"Valuing land rights","index":6,"paragraphs":[{"index":1,"size":112,"text":"Related to the 'defence of pastoralists' interests' was the impact of work on valuing land rights and thereby permitting better land policies. All of the systems studied -Maasailand, Borana, Kaduna, Niono, the Niger Delta and Fakara, plus the extensive work of independent researchers in East and West Africa -showed the economic and environmental rationality of pastoralism; these demonstrations strengthened the case of the pastoralists against forced sedentarization and other policies that threatened their livelihoods. The review of land rights research in Chapter 17 (this volume) on policy shows how a better understanding of land markets and land rights in Niger and Ethiopia supported policies that could lead to greater efficiency and equity."}]},{"head":"Identifying conditions for successful technical change","index":7,"paragraphs":[{"index":1,"size":68,"text":"The systems studies of ILRI and partners, complemented with findings from the transversal studies, made it possible to identify conditions for success or failure of proposed technical change. For pastoral systems, conditions of success included: (i) components that used local knowledge of production; and (ii) components that did not restrict animal mobility in search of water or pasture. For mixed systems with dominant cropping, conditions of success included:"},{"index":2,"size":136,"text":"(i) a degree of intensification that was compatible with local population density; (ii) a recognition that the introduction of animal power was not sufficient and was in many instances not necessary; and (iii) a recognition that flexibility in managing crops and livestock in the same farm should not be sacrificed to rigid external ideas of an optimal enterprise mix. The single most important development impact was the design and extension of the broad bed maker (BBM) tool for cultivation on Vertisols in highland Ethiopia. This is one of three ILRI research programmes that has documented the costs and benefits of research and extension in relation to a specific product; the other two are ILRI's contribution to Kenya dairy policy (see Chapter 17,this volume) and the vaccine against East Coast fever (ECF) (see Chapter 6, this volume)."}]},{"head":"The problem of translating scientific impact into development impact","index":8,"paragraphs":[{"index":1,"size":17,"text":"It was difficult or impossible to translate scientific impact into development impact for a number of reasons:"},{"index":2,"size":45,"text":"• Improving primary productivity in mobile grazing areas was generally unprofitable because of: (i) competition from natural pastures with introduced pastures; and (ii) the high costs of water, fertilizer, rotational grazing, fencing and other investments in relation to the weak productivity effects of introduced pastures."},{"index":3,"size":24,"text":"• Introducing planted forages into grazing or mixed systems generally failed because of low incremental yields and low adoption rates outside dairy production areas."},{"index":4,"size":33,"text":"• Introducing mixed farming only took place over many years and its utility as an integrated model for smallholders was limited where the conditions -population density, market access and seasonality -were not favourable."},{"index":5,"size":62,"text":"• Introducing mechanization with animals as a project component. International agricultural research centre (IARC) research had no evident impact on technical change in animal production or on efficiency and equity related to the ownership and use of livestock as draught animals. The exceptions to this generalization were found only where profitable cash crops, such as cotton and groundnut, received significant extension support."},{"index":6,"size":29,"text":"• Application of randomized control trial methods for technology evaluation and for making policy recommendations to governments is generally too costly because of sample size problems in livestock research."}]},{"head":"Introduction","index":9,"paragraphs":[{"index":1,"size":81,"text":"This chapter covers the scientific and development impact of international LSR in animal health and management, grazing management and plant production, economics and policy, soil fertility management, farm mechanization and crop-livestock interactions. It reviews both extensive systems, in which mobile herding is the principal activity with little or no arable agriculture, and mixed systems, in which arable farming is the dominant enterprise, animals and crops are managed jointly, and where animals are much less mobile. The chapter answers the following questions:"},{"index":2,"size":12,"text":"• What are the livestock systems in the tropics of sub-Saharan Africa?"},{"index":3,"size":17,"text":"• How have these systems evolved since around 1970 just before the founding of ILCA and ILRAD?"},{"index":4,"size":19,"text":"• What are the major findings of LSR in sub-Saharan Africa beginning around 1970 and continuing to the present?"},{"index":5,"size":13,"text":"• What are the scientific and development impacts of LSR since the 1970s?"},{"index":6,"size":54,"text":"The focus of the chapter is on ILRI, with reference to partners (e.g. Centro Internacional de Agricultura Tropical (CIAT), International Center for Agricultural Research in the Dry Areas (ICARDA), regional and national programmes, and others), given their important contributions in many areas. The chapter first defines the objectives, stages and potential impacts of LSR."},{"index":7,"size":91,"text":"It then summarizes tropical systems with animals as treated by Robinson et al. (2011) and earlier by Seré and Steinfeld (1996), Thornton et al. (2002), Steinfeld et al. (2006)  4 , Kruska et al. (2003), Reid et al. (2008a) and Robinson et al. (2011). Subsequent sections review the scientific and development impacts of ILRI research on pastoral and mixed systems in sub-Saharan Africa, with some reference to South Asia and Latin America, and explains why the scientific impact has often been strong while the development impact has been weak 5 ."},{"index":8,"size":91,"text":"The concluding part of the chapter summarizes the scientific and development impacts and the problems in translating scientific impacts into development impacts and indicates some future priorities. The chapter does not cover mapping methods as discussed in Robinson et al. (2011) and in the agricultural land cover literature, such as Global Land Cover (GLC) 2000 (Mayaux et al., 2004;Bartholomé and Belward, 2005), GlobCover (Bontemps et al., 2011) and MODIS products (Morisette et al., 2002), nor does it treat more recent efforts on composite products (Fritz et al., 2015;See et al., 2015)."}]},{"head":"Objectives of LSR","index":10,"paragraphs":[{"index":1,"size":74,"text":"The broadest objective of system characterization is that of Robinson et al. (2011, p. 17): '…to predict how the production systems may change in the future…and to assess the potential impact of changes in crop-livestock systems on agroecosystem services'. While Robinson's is a recent statement, it is representative of the purposes of system characterization 6 , as they were expressed when international centre livestock research began in the 1970s. The objectives were as follows:"},{"index":2,"size":24,"text":"• To define land and climate units by temperature, rainfall, altitude and soils to express output potential in systems with a major livestock component."},{"index":3,"size":22,"text":"• To describe and map crop and livestock patterns from field observations of land use and production within land and climate units."},{"index":4,"size":33,"text":"• To describe animal health risks by system '…to develop a good understanding of the differences among production systems to…minimize the risk of disease emergence and spread (Robinson et al., 2011, p. ix).'"},{"index":5,"size":31,"text":"• To apply maps of livestock densities and output (Robinson et al., 2014) to illustrate disease risks and to project global and local environmental impacts of livestock (Gerber et al., 2013)."},{"index":6,"size":35,"text":"• To estimate factor and input productivities by system and then to define constraints to development interventions that could be released by research and extension in the 'diagnostic stage' of the farming systems research sequence."},{"index":7,"size":31,"text":"• To apply systems maps -comprising climate, soils, altitude, water, animal diseases and vectors, and productivity -to plan development interventions and to better target public investment in support of sectoral goals."},{"index":8,"size":23,"text":"• To propose technical and managerial changes to relieve productivity constraints in the 'design and testing' stages of the farming systems research sequence."},{"index":9,"size":13,"text":"• To provide information for process and agent models of technologies and policies."},{"index":10,"size":118,"text":"• To improve risk management and assist recovery from shocks 7 . Norman and Collinson (1985) defined four stages in farming systems research: (i) a 'descriptive and diagnostic stage' to analysis of constraints to productivity; (ii) the 'design or planning stage' in which strategies are identified for resolving constraints; (iii) a stage in which strategies (e.g. new varieties) are tested on-farm, with varying degrees of researcher management; and (iv) an extension stage in which recommendations are applied. The comparative advantage of the IARCs, such as ILCA, CIAT and the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) in the 1970s to 1990s was chiefly in the first three stages, with national programmes leading in the fourth stage."}]},{"head":"Stages of systems research","index":11,"paragraphs":[{"index":1,"size":103,"text":"A major part of the descriptive and diagnostic stage in international livestock research was agroecological zoning, which classified agricultural systems by 'climate, soil and terrain' (Collinson, 2000, p. 51;Seré and Steinfeld, 1996) and later work in this century such as that of Kruska et al. (2003) and Robinson et al. (2011), which used data and tools that had not existed when the IARCs were expanding quickly in the 1970s and 1980s. Its purpose was to project carrying capacities, to identify domains for genetic and management improvement, to describe field conditions for experiments 8 and to show the scope for technology transfer among zones."},{"index":2,"size":172,"text":"The basis of the design and planning stage was to create average farm typologies within agro-ecologies. These typologies would be used to make recommendations applicable to dairy farms using cut forages, to smallholders practising rainfed cropping without animal traction, or to systems such as watering practices for transhumant herders. The ILCA systems studies in grazing and mixed systems are a mix of the descriptive/diagnostic and design/planning stages 9 . The most similar CIAT study (Rivas Rios, 1974) is the second stage, in which beef ranches in the same agroecology of Colombia were grouped by farm size. The ICRISAT village-level studies in the semi-arid tropics of India drew samples of farms based, inter alia, on agroecological zoning and ultimately constructed Stage 2 farm typologies using irrigation, mechanization and the principal crop as variables (Walker and Ryan, 1990). The construction of farm typologies was a principal goal of the French tradition of agronomy in West Africa and this had some influence on the work of ILCA and ICRI-SAT in that region from the 1970s."},{"index":3,"size":129,"text":"The third stage was to estimate productivity determinants and to test technologies using experiments in environments and farm types as defined in the first two stages. This stage is to calculate the levels, rates of change and determinants of enterprise and whole-farm productivity to be used in identifying constraints and in targeting extension. This estimation was done using farm data gathered in surveys, sometimes combined with farmer-or researcher-managed experiments (an early summary of mainly crop work from sub-Saharan Africa and Latin America is given by Matlon et al., 1984). The third stage was prominent in the early systems research portfolios; crop research tended to be more prominent, with the early ILCA work in Ethiopia, Mali and Nigeria and some CIAT work in Latin America being the main livestock exceptions."},{"index":4,"size":190,"text":"The fourth stage was 'recommendation and diffusion' of new technologies. This was to be achieved by models of 'entry points' (as described much later by van Wijk et al., 2009) at which technical changes could be introduced based on an understanding of the complex interactions among system components. This stage was broader in LSR than in cropping, which focused on crop cultivars and associated input packages, because of the need to integrate more components -animal species and breed, animal health, animal management (fencing, housing and grazing rotations) and feed production. Livestock-related examples include mechanization: animal traction, the use of cows for draught power and the BBM in Ethiopia; the introduction of trypanotolerant animals in combination with drug treatments and vector control (see Chapters 2 and 3, this volume); a vaccine against ECF, using the infection-and-treatment method (see Chapter 6, this volume); the use of trypanocidal drugs (see Chapter 3, this volume); feed improvements with grasses and legumes (see Chapters 11 and 12,this volume); various forms of land management, such as fodder banks with Stylosanthes spp. in central Nigeria; and alley farming in Nigeria and other humid countries in West Africa."}]},{"head":"Potential impacts of systems research","index":12,"paragraphs":[{"index":1,"size":26,"text":"We set the potential scientific and development impacts of LSR into the three classes of policy studies as defined by Zilberman and Heiman (2004, pp. 278-279)."},{"index":2,"size":41,"text":"Class I is scientific understanding, advanced by estimating system scale, productivity and sectoral accounts. Class I includes advances in research methods, such as using remote sensing to map production systems and carrying out long-term field studies on production and environmental relations."},{"index":3,"size":52,"text":"Class II is contributing to technical change, which lifts productivity. This can involve estimating optimal levels of water, veterinary drug and feed use under controlled situations and testing these levels on farms or among herds. Other examples would be testing innovations, such as vaccines, tropical pastures and multidimensional crops, or improved processing."},{"index":4,"size":54,"text":"A subset of Class II is to estimate the negative externalities (costs) of farming, such as greenhouse gas emissions, land degradation and water pollution from livestock effluents. Examples are the book Livestock's Long Shadow (Steinfeld et al., 2006), which calculated the local and global environmental effects of animal production, and research on antimicrobial resistance."},{"index":5,"size":40,"text":"Class III is advising on policies to raise productivity or to improve income distribution. This can involve adjusting terms of trade, devising better institutions, eliminating distortions in incentives, identifying costs of bad policies and proposing measures to achieve higher growth."},{"index":6,"size":13,"text":"The potential scientific and development impacts of LSR would occur through the following:"},{"index":7,"size":15,"text":"• Defining and me asuring land and climate units with similar production potential (Class 1)."},{"index":8,"size":19,"text":"• Characterizing and mapping production systems -combinations of resources by season -across the land and climate units (Class 1)."},{"index":9,"size":27,"text":"• Estimating productivity relationships to identify priority technologies in animal health, breed, feed and management, or to propose changes in organization or policy (Classes 1 and 2)."},{"index":10,"size":21,"text":"• Targeting these technologies (new methods of production or organization, or policy shifts) for controlled testing in relevant systems (Class 2)."},{"index":11,"size":22,"text":"• Simulating potential growth as functions of hypothesized changes in technologies, policy, organization or species/breed between domesticated animals and wildlife (Class 2)."},{"index":12,"size":13,"text":"• Providing information to guide extension of proposed technologies to farmers (Class 3)."},{"index":13,"size":22,"text":"• Writing ex post evaluations of the development and environmental impact of changes in technologies, policies and organizations (Classes 2 and 3)."},{"index":14,"size":17,"text":"• Proposing new research hypotheses after the ex post evaluations of new methods (Classes 2 and 3)."},{"index":15,"size":16,"text":"• Revising methods of experimentation and modelling to test the new hypotheses (Classes 2 and 3)."}]},{"head":"Classifying Livestock Systems","index":13,"paragraphs":[{"index":1,"size":66,"text":"Classifying livestock systems has three steps: (i) describing the biophysical conditions of production in terms of climate and length of growing period, as determined by rainfall, temperature and altitude; (ii) describing the modes of livestock and crop production with respect to animal mobility, principal product and enterprise scale; and (iii) mapping the modes of crop and livestock production by biophysical conditions to define systems (see ."}]},{"head":"Climate and growing period","index":14,"paragraphs":[{"index":1,"size":33,"text":"The typologies of the Food and Agriculture Organization of the United Nations (FAO, 1975) classified the physical conditions of animal production by climate and length of growing period (LGP) 10 as the following:"},{"index":2,"size":53,"text":"• Arid climates, of the Köppen types BWh (arid, desert, hot arid) and BSh (arid, savannah, hot arid), with an LGP of less than 75 days 11  • Semi-arid to subhumid climates, of the Köppen types Am (equatorial monsoon), Aw (equatorial winter dry), BSh and BWh, and having an LGP of 75-150 days."},{"index":3,"size":18,"text":"• Subhumid to humid climates, of the Köppen types Aw and Am, with an LGP of 181-270 days."},{"index":4,"size":88,"text":"• Humid climates, of the Köppen types Aw and Am, with rare areas in Af ('fully humid'), with an LGP of more than 270 days. Ruthenberg (1980) proposed four modes of tropical livestock production -ranching, commercial dairying, nomadic and semi-nomadic subsistence dairying, and mixed crop-livestock farming with irrigation or under rain-fed conditions. Ruthenberg used two variables -animal mobility (1980, p. 18) andrainfall (1980, pp. 322-323) -in his classification. He grouped animal mobility into permanent nomadism, seasonal (proximate or distant transhumance) or none, as with village grazing or stabulation."}]},{"head":"Modes of production","index":15,"paragraphs":[]},{"head":"Ranching","index":16,"paragraphs":[{"index":1,"size":117,"text":"Ranching is a commercial system on large, private, fenced holdings, usually for beef production using grade animals, with no seasonal or annual mobility outside the ranch. The amount and seasonality of rainfall would vary from arid Botswana (Köppen climates BWh and BSh), Australia (Köppen climates BSk, BWh, plus the humid subtropical classes Cfa and Cwa) to subhumid and humid (Colombia and Brazil; Köppen climates Af, Am and Cwb). There is no reliable estimate of ranched cattle in the African tropics, but that number would be small as a share of all cattle in sub-Saharan Africa and certainly much less than the corresponding shares in the humid tropics of Latin America or in the arid tropics of Australia."}]},{"head":"Commercial dairying","index":17,"paragraphs":[{"index":1,"size":144,"text":"Commercial dairying is milk production for the market, with some form of sown pastures, little seasonal mobility, no annual mobility, and both large and small holdings. Commercial dairying in the tropics is typically in the cool subhumid highlands (East Africa and parts of Latin America) or in the hotter subhumid lowlands, both of which have high rainfall and year-round pasture production. In sub-Saharan Africa, commercial dairying is restricted to highland countries in East Africa with less risk of trypanosomiasis and ECF, or to parts of southern Africa with cooler climates and where insect-borne diseases can be better managed. No reliable estimate of animals or land in commercial dairying is available for sub-Saharan Africa in the 1970s, but both would have been small shares of their respective continental totals and with respect to the numbers of commercial dairy animals in Latin America and the Caribbean."}]},{"head":"Nomadic and semi-nomadic grazing","index":18,"paragraphs":[{"index":1,"size":134,"text":"Nomadic and semi-nomadic grazing is milk production for subsistence, in which extensive seasonal and annual mobility are the defining features. The climate is arid 12 of the Köppen types Bw and Bs with rainfall usually below 300 mm in a single wet season. Grazing systems in sub-Saharan Africa cross a wide band from Senegal to Eritrea, into the northern parts of the West Africa coast (seasonally), into Sudan, Cameroon and the Central African Republic, and then south from Eritrea and Ethiopia around the Rift Valley into Mozambique. Practically all of the African nomadic and semi-nomadic modes are at altitudes below 1500 m above sea level (masl). These systems, although sparsely populated and understocked relative to their potential, were the most important in grazing areas and in ruminant numbers in sub-Saharan Africa at ILCA's founding."},{"index":2,"size":48,"text":"grazing systems. We use 'grazing systems', 'pastoral systems' and 'pastoralism' interchangeably to refer to African farming systems in which ruminants are the main stock. These are livestock/ grazing/arid (LGA), livestock/grazing/humid (LGH) and livestock/grazing/tropical highlands (LGT) in the terminology of Sere and Stenfeld (1996) and Robinson et al. (2011)."},{"index":3,"size":37,"text":"The LGT system is livestock found only in humid and subhumid zones. These are found in Köppen climate Aw (tropical savannah), which is generally hot and wet with the driest month having less than 60 mm rainfall."},{"index":4,"size":41,"text":"The LGH system is livestock only in (temperate and) tropical highland zones. The LGT and LGH systems constitute perhaps 26% of the sub-Saharan Africa grazing area, a similar share of the ruminant populations and perhaps onesixth of the sub-Saharan African population."},{"index":5,"size":110,"text":"African rangelands fall into two groups. The west and central group extends from Senegal to eastern Chad, nearly all of which is in the LGA and LGH systems 13 . All West African grazing systems have a single rainy season at elevations below 1000 masl. The shorter rainy seasons in West Africa (60-90 days), uniformly lower elevations, generally hotter temperatures and strictly monomodal rainfall imply lower average primary productivity, more variable grazing and therefore stronger reasons for extended transhumance (Pratt and Gwynne, 1977;Thornton et al., 2002). The West African group would tend to have lower average pasture quality, as indicated by crude protein (CP) content per unit of dry matter."},{"index":6,"size":109,"text":"The east and central group extends from central Sudan south and east into Eritrea, Ethiopia, Kenya, Somalia and parts of Uganda, Tanzania, Rwanda and Burundi. East African systems are more diverse, ranging from arid to semi-arid (LGA) to subhumid (LGT), with some bimodal rainfall, and some LGH systems at altitudes above 1000 masl. The east and central groups further differ from the western group in having: (i) longer growing periods (60-90 days and sometimes 90-120 days); (ii) colder minimum temperatures; (iii) higher CP content in native pastures; (iv) more competition with wildlife; and (v) because of (i), more diverse mobility patterns and more diverse human ecology (Homewood, 2008;Blench, 1999)."},{"index":7,"size":255,"text":"Systematic area data exist for 1991-1993 (Fig. 15.1; (Fig. 15.1;Seré and Steinfeld, 1996), for 1990-2000(Kruska et al., 2003) and for 2000-2010 (Robinson et al., 2011). The data in Fig. 15.1 for 1991-1993 are adapted from Seré and Steinfeld (1996) as the nearest global benchmark for the ILCA studies that began in the late 1970s. Ruminant systems on grasslands (LGA, LGH and LGT) covered about 1463 million ha, of which 25% was in Latin America and the Caribbean, 40% in sub-Saharan Africa, 27% in Asia, and 8% in West Asia and North Africa. The largest single grassland was LGA in sub-Saharan Africa, covering some 392 million ha, which comprised 27% of the world's grasslands and 67% of sub-Saharan African grasslands. The grassland systems in Latin America and the Caribbean are in the humid and subhumid climates, while those in Central Asia, China, Mongolia and Russia are in the cold and arid climates. Shares of the global totals of ruminant livestock in agroecologies classified as 'grasslands' are (Table 15.1): Latin America and the Caribbean, 55%; sub-Saharan Africa, 33%; Asia, 12%; and West Asia and North Africa, less than 1%. The respective population shares are Latin America and the Caribbean, 22%; sub-Saharan Africa, 28%; Asia, 24%; and West Asia and North Africa, 6%. Animal density -animals per hectare or animals per human population -is highest for the three grassland systems in Latin America and the Caribbean and next highest in sub-Saharan Africa. Reid et al. (2005) reviewed 40 years of research on extensive systems in East African grasslands."},{"index":8,"size":149,"text":"LGA is defined as livestock only. The grazing areas and animal numbers of the most arid system are by far the highest in sub-Saharan Africa. These facts indicate that sub-Saharan Africa's most important growth potential in terms of animal numbers is also its highest cost area in terms of grazing land used per animal unit. LGA types are typically in the arid and semi-arid (Köppen climate BSh). They are generally hot with a single rainy season and with no month having a minimum temperature below 0°C. The LGA system (390 million ha in sub-Saharan Africa) covered some 27% of global grazing and some 53% of sub-Saharan Africa grazing. The great majority of African production in LGA systems is nomadic herding with milk as the subsistence good. Population density is low, typically less than 25 persons/km 2 ; ruminant stocking rates were generally from 5 to 20 ha per animal."}]},{"head":"Mixed farming by smallholders","index":19,"paragraphs":[{"index":1,"size":166,"text":"Ruthenberg's fourth mode of tropical livestock production -mixed farming by smallholdershas always been the focus of international agricultural research in sub-Saharan Africa. Such modes blend crop and livestock activities. Animals eat crop residues and provide draught power, dairy products, meat and manure. Animal mobility is restricted to seasonal grazing with some transhumance as part of animal tenure relationships between crop farmers and pastoralists.   Robinson et al., 2011, p. 48.) mixed systems. We use 'mixed systems' from the classification of Seré and Steinfeld, 1996)  The mixed-farming types 15 -MRA, MRT, MRH and some MIA-formed the majority of rain-fed cropped area and income in sub-Saharan Africa in the mid-1970s. These shares grew after the mid-1970s with the expansion of animal traction in much of West Africa and in parts of East Africa. MRA and MRH systems represented some 121.9 million poor rural livestock keepers in sub-Saharan Africa in 2000 or threequarters of the total of all poor rural livestock keepers (Robinson et al., 2011, pp. 145-152)  16 ."},{"index":2,"size":18,"text":"The mixed systems are characteristically more diverse in enterprise and spatial patterns than the pastoral types 17 ."},{"index":3,"size":11,"text":"African mixed-farming systems 18 differed from pastoralism in the following aspects:"},{"index":4,"size":10,"text":"• Most land use, income and employment were from cropping."},{"index":5,"size":9,"text":"• Crop residues displaced pastures as the main feed."},{"index":6,"size":14,"text":"• The shares of animal products in food consumption were less than among pastoralists."},{"index":7,"size":33,"text":"• Animal mobility was limited to seasonal transhumance or to daily movements near permanent villages. In the Latin American systems, with private and fenced grazing, there would be no seasonal or annual mobility."},{"index":8,"size":8,"text":"• The functions of livestock were more diverse."},{"index":9,"size":35,"text":"They included recycling soil nutrients and using draught animals for cultivation. These functions of animals were less important or even absent in Latin America where mineral fertilizers replaced manure and tractor mechanization replaced animal power."},{"index":10,"size":44,"text":"• There were important shares in income and employment of annual and semi-perennial cash crop -cotton, groundnut, rice and other cereals, roots and tubers -that did not exist in rangelands. In the Latin American and Middle Eastern mixed systems, animals were the cash commodity."},{"index":11,"size":35,"text":"• Important shares in income and employment of perennial cash crops -coffee, tea, cocoa, rubber and oil palm -were found in mixed systems with poultry, swine and small ruminants, although much less often with cattle."},{"index":12,"size":24,"text":"• There was smallholder commercial dairying with zero grazing, planted forages, purchased feeds, and cultivation of annual and perennial crops on the same farm."},{"index":13,"size":25,"text":"• There was a more unequal distribution of cattle among households, although this was compensated to some degree by more equal distribution of small ruminants."},{"index":14,"size":27,"text":"• There was commercial on-farm fattening of ruminants, swine or poultry, in zero-grazing or transhumant systems, using purchased feeds and crop residues from annual and perennial crops."}]},{"head":"Trends in African Agricultural Systems","index":20,"paragraphs":[{"index":1,"size":31,"text":"In this section, we discuss the trends in African agricultural production systems in a sample of countries where livestock production is prominent and where ILRI has had a significant research presence."}]},{"head":"Human populations","index":21,"paragraphs":[{"index":1,"size":92,"text":"Data on the distribution of African populations across agricultural systems have long been patchy. The information compiled by Jahnke (1982), as presented in the Introduction to this volume, show that perhaps 238 million rural people lived in sub-Saharan Africa in the mid-1970s in all five agroecological zones. Of that mid-1970s total, nearly 40% lived in the arid and semi-arid zones that were the initial focus of ILCA. The focus of ILRAD's work can be mapped largely to the humid and subhumid agroecological zones given that trypanosomiasis and Theileria are more present there."},{"index":2,"size":251,"text":"The more detailed estimates of Seré and Steinfeld (1996) for nine livestock production systems gave a 1991-1993 average of some 519 million people who depended on those systems. The latter estimated human populations as: (i) a total of 173 million people in the grassland systems, constituting about one-third of the sub-Saharan African population and implying a population density of approximately 30 persons/km 2 on grazing plus arable lands; and (ii) a total of 346 million in the mixed rain-fed and irrigated systems, implying a population density of approximately 140 persons/km 2 on grazing plus arable lands. Of this 1991-1993 total, about one-third lived in the LGA/LGT/LGH production systems, which is not greatly different from Jahnke's estimated share of the rural population living in the arid and semi-arid agroecological zones. Robinson et al. (2011, p. 48) described four broad classes of livestock systems from vintage 2000 data: (i) agro-pastoral and pastoral, corresponding to the grassland systems LGA, LGH, LGT; (ii) 'mixed extensive' systems, corresponding to MRA, MRH, MRT, MIA, MIH and MIT; (iii) 'mixed intensifying' systems as those parts of MRA, MRH, MRT, MIA, MIH and MIT with higher production potential and better market access; and (iv) 'others', which were mainly urban and areas having less than '10 percent of the total land area covered by crops' (Robinson et al., 2011, p. 46), as shown globally in Fig. 15.1. Sub-Saharan Africa is dominated by agro-pastoral and pastoral systems, and most of the original work of ILCA and ILRAD focused there (Fig. 15.2). "}]},{"head":"Livestock production","index":22,"paragraphs":[{"index":1,"size":47,"text":"We know generally that the evolution of Africa livestock systems after the colonial era has been determined by: (i) rainfall and its effects on pasture and crop growth, as the latter is a partial determinant of fodder availability for ruminants; (ii) control of human and animal diseases;"},{"index":2,"size":79,"text":"(iii) the growth of human populations and income and the resulting demand for agricultural products; and (iv) the expansion of cropland. These factors have been compounded by exogenous changes in farming methods, the introduction of arable and permanent cash crops, and the development of irrigated farming (e.g. Inner Delta of the Niger River in Mali, the Awash Valley in Ethiopia, the Niger-Benue confluence in Nigeria, the Senegal River Valley and along the other major river basins of sub-Saharan Africa)."},{"index":3,"size":98,"text":"Evidence about stock holdings -numbers, species, breeds, productivity and locations -is important for understanding the development impact of livestock research. Data on African livestock numbers and productivity vary erratically over time, agroecological zone, country, season and mode of production. Estimates of animal numbers are particularly unreliable because of systematic undercounting. It is generally impossible to relate stock numbers or their productivity to inputs of research, infrastructure or policy. With these warnings, the following summarizes information over the period 1970-2016 for aggregate information from sub-Saharan Africa on livestock production, stocks of ruminants and the species composition of ruminant numbers."},{"index":4,"size":129,"text":"We defined a productivity measure as growth of the FAO livestock production index, in rural per capita terms, for the East and southern Africa (Fig. 15.3a) for the period 1970-2016. The total index grew at an annual rate of about 1.9% from 1970 to 2016; a related index, normalized by rural population, grew at an annual rate of 2% in those countries. The corresponding values for ten countries in West and Central Africa (Fig. 15.3b) were 2.8% and 0.7% for the same period. In sum, the great majority of livestock production growth in sub-Saharan Africa for nearly 50 years is closely associated with population growth in 23 countries. The principal direct research/development influence would have been through the eradication of rinderpest, which was achieved early in the 21st century."},{"index":5,"size":87,"text":"A measure of ruminant stock numbers is the tropical livestock unit (TLU) 19 . Figure 15.4 expresses 'TLU density' as the ratio of ruminant TLU to arable land. Growth in TLU numbers accelerated sharply in East and southern Africa and in West and Central Africa around the mid-1990s. This 'area-weighted TLU density' grew at an annual rate of 0.6% in East and southern Africa 20 during the period 1970-1994, at a rate of 2.0% for 1995-2016 and at 1% for the entire period of 1970-2016 (Fig.15.2a) (Fig.15.2a)."},{"index":6,"size":44,"text":"The estimated TLU density grew at an annual rate of 1.2% in West and Central Africa during the period 1970-1994 and at 3.6% for 1995-2016. Over the period of 1970-2016, TLU density in West and Central Africa rose at an annual rate of 2.3%."}]},{"head":"Changes in species mix","index":23,"paragraphs":[{"index":1,"size":188,"text":"One factor affecting the relevance and impact of research is long-term changes in ruminant species mix. This mix appeared to have shifted slightly to cattle from sheep and goats from 1970 to 2016 (Fig. 15.5). This continues the trend from 1950 to 1980 as noted earlier by Le Houérou (1989, p. 125). The ratio of cattle TLUs to sheep/goats TLUs was about 6.7 in 1960, about 4.1 in the mid-1980s and about 3.7 in 2009(FAO/CIRAD, 2013, p. 6). De Haan (2016, p. 25) found similar patterns for stock numbers and species composition in East Africa, excluding Ethiopia, over the period 1960-2011. There was little change in the ratio of sheep and goat numbers to cattle numbers in East and southern Africa (Fig. 15.5a) over the period 1970-2016; this ratio rose modestly in West and Central Africa (Fig. 15.5b), which is possibly related to the rising scarcity of grazing in that subregion. The shift in species from cattle to small ruminants indicates that some opportunity has been lost by not focusing more on diseases specific to small ruminants, such as peste des petits ruminants (see Chapter 7, this volume)."},{"index":2,"size":196,"text":"The ILCA systems studies (Ethiopia, Kenya, Mali, Nigeria and Niger) were not conducted over a long enough period to say anything reliable about species mix over time. One exception was an analysis of a mixed sheep and goat flock at Elangata Wuas Group Ranch in Kenya where goats had become more numerous relative to   1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 1970 1975 1980 1985 1990  sheep between 1978 and 1986 (Wilson and Maki, 1989). Coppock (1994, pp. 169-170) found that short-term droughts in Borana affected cattle numbers more adversely than smallruminant numbers. One aspect of species composition is the use of animals, mainly oxen, for power. Indeed, animal traction research was an important part of ILCA research until the 1990s. Systematic and comprehensive information on draught animals is unavailable in sub-Saharan Africa. We do know that animal traction has expanded widely in this area from a very low base of use, while tractor mechanization has displaced animal and human power in the other dryland and subhumid regions of the world, although lack of data make it impossible to estimate changes in stocks of TLUs according to the use of animals for power."}]},{"head":"Land use, fertilizer use and cereal yields","index":24,"paragraphs":[{"index":1,"size":134,"text":"Evidence about land holdings and rural population shows rising land pressure throughout sub-Saharan Africa. The stock of arable land per rural population fell continuously over the long period of 1970-2016. In East and southern Africa, there was about 0.3 ha of arable land per capita in 1970 and this had fallen to about 0.2 by 2016 (Fig. 15.6a); the annual rate of decline is estimated to have been -1.1%. In West and Central Africa, there was about 0.7 ha of arable land per capita in 1970 and this had fallen to about 0.3 by 2016 (Fig. 15.6b); the annual rate of decline is estimated to have been about -1.4%. The amount of arable land per rural population began to fall less sharply after the mid-1990s, reflecting the broad deceleration of population growth across Africa."},{"index":2,"size":28,"text":"Evidence about fertilizer use and cereal yields suggests a modest intensification of farming systems in response to land pressure. National data on fertilizer use (Fig.  2011-2016 (Fig.15.7b (Fig.15.7b)."},{"index":3,"size":167,"text":"The regional mean cereal yield in East and southern Africa was about 1.1 t/ha in 1970-1975 and had risen to 1.9 t/ha in 2011-2016 (Fig. 15.8a) (Fig. 15.8a); the estimated average rate of growth for East and southern Africa was 0.9% from 1970 to 2016. The West and Central Africa mean was 0.6 t/ha in 1970 and had risen to roughly 1.4 t/ha by 2016 (Fig. 15.8b); the estimated average rate of growth for the ten West and Central Africa nations was 1.3% from 1970 to 2016, so this subregion's levels and growth rates lag behind those of southern Asia and of Latin America. The mean cereal yield in West and Central Africa was about 0.7 t/ha in 1970 and had risen to 1.4 t/ha by 2016 (Fig. 15.8b). The index of labour productivity in cereals therefore fell systematically over the long period of 1970-2016; it had been roughly 0.45 t per capita in the 1970s and fell to an average of 0.38 in the period 2010-2016."},{"index":4,"size":79,"text":"There was some small improvement in an index of food production in East and southern Africa and in West and Central Africa over the period of 1970-2016. Food production per rural inhabitant among the 13 East and southern African nations (Fig. 15.9a) rose at an average annual rate of 1.4% from 1970 to 1994; it rose at a rate of 3.1% from 1995 to 2016. The corresponding values for West and Central Africa (Fig. 15.9b) were 0.5% and 1.7%."},{"index":5,"size":147,"text":"The aggregate evidence, with all the qualifications about potential internal errors, shows the following: (i) 23 countries in sub-Saharan Africa have much less land per capita than they had 40-50 years ago; (ii) these countries have increased cereal yields at a rate close to that of rural population growth, indicating that rural labour productivity in the major crop group has barely grown; (iii) livestock productivity has similarly increased at an annual rate roughly equal to that of population growth; (iv) most of the gains in livestock productivity are due to more animals per unit of land, not to more productivity per animal; (v) there have been modest gains in fertilizer use per unit of land and sub-Saharan Africa still lags regions of the world where agriculture has grown more quickly; and (vi) there have been modest gains in food production per capita in the two principal sub-regions.  "}]},{"head":"Inequality of livestock holdings","index":25,"paragraphs":[{"index":1,"size":179,"text":"Inequality of livestock holdings would determine to some extent the distribution of benefits from research across households within given agroecologies. The African field studies carried out by ILCA and later ILRI all showed substantial inequality among farm units in livestock holdings, indicating that research would benefit fewer households than in a more even distribution of animals across production units. Inequalities in ownership of cattle were high within and between Malian survey villages, as were those in ownership of small ruminants (Wilson, 1986, p. 23). High degrees of inequality in stock holdings were found in the Borana study led by Coppock (1994, Table D1). In the Maasailand study of Solomon Bekure et al. (1991, p. 84), 29% of households in three group ranches owned 57% of cattle in the sample while 24% of sample households owned only 4% of the cattle. Hiernaux and Ayantunde (2004, pp. 35-27) found significant differences in land and animal holdings per adult household member in a sample of 492 households in western Niger studied in the 1990s and in the early part of this century."},{"index":2,"size":148,"text":"A comparison of global data from 12 countries 21 confirmed the general finding of livestock holding inequality across households. With household samples divided into expenditure quintiles, measured by the presence or absence of animals and by TLUs per household, the comparison showed that about 65% of rural households in all expenditure groups had livestock (1.9 TLUs on average across countries and expenditure quintiles); about 68% in the lowest expenditure quintile had livestock (1.4 TLUs on average), while 58% in the highest expenditure quintile had livestock (3.3 TLUs on average). Surveys of pastoralists from 2003 to 2011 in 11 sub-Saharan African nations showed a median Gini coefficient of 0.673 on TLU holdings (de Haan, 2016, p. 238). The older evidence is not directly comparable to the more recent data from the 11 countries and hence we cannot say if concentration of ownership has changed in recent years 22 ."}]},{"head":"Climate to 2000","index":26,"paragraphs":[{"index":1,"size":111,"text":"The ILCA systems studies in West Africa, beginning in the 1970s, took place at a time of rising temperatures and drier growing conditions. Global atmospheric studies (Hulme et al., 2001) showed African temperatures to have risen from the mid-1970s to 2000. The most regular increases occurred in the northern hemisphere rainy (June-August) and post-rainy (September-November) months. There was less pronounced annual warming in East Africa and even a cooler period in the Borana and Turkana ranges. The mid-1970s to 2000 tended to be drier than the previous 75 years in the Sahel (about 25%); there was a mix of wetter and dryer trends in East Africa, with more extremely wet years."},{"index":2,"size":80,"text":"Vegetation trends in the Sahel, despite the apparent warming and drying, tended to show recovery from earlier drought cycles. Herrmann et al. (2005) contended that land degradation in the Sahel, caused by drought, had been partly reversed by conservation efforts. The effects of climate on sub-Saharan agricultural production, on the whole, were mixed in the final 30 years of the previous century and no continental generalization can be made about climate as a determinant of research success/failure in that context."}]},{"head":"Poverty","index":27,"paragraphs":[{"index":1,"size":79,"text":"Poverty was not a theme of ILCA/ILRAD/ILRI research before the mid-1990s and the words 'poor' or 'poverty' as keywords in published work rarely appear before 2000. There was some analysis of wealth disparities in Maasailand in the 1980s by King et al. (1984) and Grandin (1988) but no systematic or even sporadic effort to relate ILRI's work to poverty in Africa, or anywhere else, before the work of Thornton et al. (2002) and that of Perry et al. (2002)."},{"index":2,"size":129,"text":"There were no reliable estimates of rural poverty in sub-Saharan Africa before those of Thornton et al. (2002). Robinson et al. (2011) estimated that LGA, LGH and LGT represented some 29.8 million poor livestock keepers in sub-Saharan Africa, or about 18% of poor rural people. The dry (LGA, MRA and MIA) areas were about 45% of poor livestock keepers in sub-Saharan Africa. The most recent compilation (de Haan, 2016) confirms the historical pattern of higher poverty among animal keepers in the drier climates of sub-Saharan Africa, suggesting that little relative progress has been made in those areas. While there was no poverty focus, as such, of ILCA/ILRAD work until this century, the choice of agroecologies for the original systems studies imposed such a focus on research at those sites."}]},{"head":"Livestock Systems Research","index":28,"paragraphs":[]},{"head":"Research impacts on livestock","index":29,"paragraphs":[{"index":1,"size":230,"text":"Evaluations of the impact of LSR in sub-Saharan Africa confront difficult empirical problems of context, measurement and attribution. The availability of baseline estimates for other regions and production systems is limited. Published estimates of the impact of research on long-term productivity changes in livestock are uncommon globally and rare in Africa. A global review of returns to agricultural research in crops, livestock, fisheries and forestry found a livestock focus in 14.4% of a sample of 292 papers, which included a total of 1886 estimates. The great majority of papers were from the USA, Latin America and the Caribbean (Alston et al., 2000), not from sub-Saharan Africa. The authors found livestock research -including 'beef, swine, poultry, sheep or goats, all livestock, dairy, other livestock, pasture, and \"dairy and beef \"'to have a positive but highly variable, rate of return. A sample of 233 estimates gave a mean internal rate of return (IRR) of 120.7%, with a coefficient of variation of 4.0, a mode of 14.0 and a median of 53.0 (Alston et al., 2000, p. 58). A meta-regression across research programmes (crops, animals and natural resources) with rate of return as the dependent variable had a livestock research coefficient of 12.1% (p < 0.90); field crops had a marginal return of 25.1% (p > 0.99) and natural resources research (fisheries and forestry) had a negative return of 94.5% (p > 0.99)."},{"index":2,"size":67,"text":"Other international comparisons of research effects on livestock productivity do not allow firm generalizations. Evenson and Rosegrant (2003) reviewed published studies of research productivity from US, Brazilian and Indian agriculture. They estimated that annual total factor productivity (TFP) growth in US livestock production was 0.55 from 1950 to 1982, of which 9% was due to public research, 17% to extension and 54% to private research and development."},{"index":3,"size":102,"text":"They estimated that annual TFP growth in Brazilian livestock (1970Brazilian livestock ( -1985) ) was only 0.09, of which 55% was due to public research and 40% to other, unidentified, factors. They made no estimates for livestock in India, although their estimates of TFP growth in crops were from 1. 27 (1956-1965) to 1.14 (1977-1989) and had public research shares of 22% and 45%, respectively. Hazell (2008) found no post-Green Revolution research impact on livestock productivity in South Asia. He mentioned the substantial expenditures of the Indian national programme (Hazell, 2008, pp. 5 and 16) but did not analyse their productivity effects."},{"index":4,"size":125,"text":"Alene ( 2010) calculated an annual rate of growth in TFP, for crops and livestock, of 1.8% 23 across sub-Saharan Africa for 1970-2004 and a slightly faster rate of 2.1% from 1991-2004, possibly reflecting better weather or economic reforms. He estimated an elasticity of TFP with respect to research and development expenditures of 0.10-0.22, depending on the statistical model used. One specific estimate of research impact on African livestock productivity is that of Thirtle et al. (1998) for South Africa, which found rates of return ranging from 11% to 16% for livestock, pasture and range improvements, and returns to animal health research exceeding 36%. The estimates from South Africa were made for that country's national programme and hence include research failures as well as successes."},{"index":5,"size":75,"text":"We will discuss mainly African systems with some reference to other regions whose studies are relevant to sub-Saharan Africa, given that most international livestock research has been in sub-Saharan Africa (Table 15.1). We focus first on grazing systems (LGA, LGT and LGH), of both nomadic and seasonally transhumant types 24 , and later on mixed farming, both dryland (MRA, MRT and MRH) and dryland/irrigated (MIA, MIT and MIH), where crops and livestock interact 25 ."}]},{"head":"The development problem of pastoralism","index":30,"paragraphs":[{"index":1,"size":96,"text":"The initial studies of pastoral and agro-pastoral systems confronted a 'mainstream view' 26 which informed official and scientific opinions about the rangelands (Sandford, 1983a, pp. 11-19). This 'mainstream view' held that tropical pastoralism was unproductive, for two reasons. The first reason was the biophysical determinants of animal productivity 27 and the lack of adapted technology and management to raise productivity. This technical pessimism about potential output was attributed to the arid and variable climates 28 in the tropics, which made crop production infeasible without irrigation and which restricted agriculture to extensive stock raising and oasis farming."},{"index":2,"size":92,"text":"Most of the papers in Monod (1975) expressed the 'mainstream view'. Gourou (1966, p. 192), a leading post-war geographer of the tropics, had earlier argued that tropical cattle productivity could converge to that of the temperate zones only under strict technical conditions involving: '…the importation, full application and adaptation of all the technical progress made in the temperate lands, namely selection of fodder plants, the use of irrigation, abundant and precise application of manures, careful breeding of animals, controlled feeding, and, above all, complete and lasting victory over microbial and parasitic diseases.'"},{"index":3,"size":147,"text":"The biological reasons for pessimism were well founded, as recognized early by the United Nations Educational, Scientific and Cultural Organization (UNESCO 1963, pp. 168-169), ILCA (1975), Pratt and Gwynne (1977), Le Houérou (1989) and in the Dutch-Malian study of Sahelian rangelands (Penning de Vries and Djiteye, 1991). The economic reasons for pessimism were given by Pratt and Gwynne (1977, pp. 100-128). They showed that the economics of range improvement -water development, fencing, overseeding of the range, managed grazing and seeding with new species of grasses -were unfavourable in drier conditions and risky in wetter ones. These biological and financial findings were generally confirmed by Le Houérou (1980) for browse and in the global review by Sandford (1983a) of pastoral development. A review nearly 40 years later of rigorous impact studies of CGIAR research in grazing systems found few productivity effects of that research (Jutzi and Rich, 2016)."},{"index":4,"size":15,"text":"Le Houérou (1989, pp. 149-155) synthesized Sahelian investigations over many years and concluded the following:"},{"index":5,"size":22,"text":"• Irregular rainfall, a long dry season, and competition from native grasses and forbs made the introduction of higher-yielding plant species unsuccessful."},{"index":6,"size":27,"text":"• There had been no commercial success in reseeding pastures at rainfall around the 550 mm isohyet, with local or introduced grasses or with legumes 29 ."},{"index":7,"size":23,"text":"• There had been only one success 30 in establishing browse species at rainfall less than the 400 mm isohyet in West Africa."},{"index":8,"size":42,"text":"• The high cost of fencing, firebreaks and water made it nearly impossible to achieve higher plant yields even if browse and pasture production could be improved under experimental conditions (Le Houérou, 1989, p. 150; see also Montgolfier-Kouèvi and Le Houérou, 1980)."},{"index":9,"size":93,"text":"The second reason for 'livestock pessimism' of the 'mainstream view' was the persistent belief among external observers that African pastoralists were inefficient managers for 'cultural' reasons. The cultural reason, common around the time of the founding of the African livestock centres, was that herders kept too many animals relative to range capacity because they derived utility from keeping stock rather than selling them. This interpretation impelled the view that rangelands management practices, especially open-access land tenure, would lead to overgrazing 31 and should therefore be modified to avoid the destruction of the range."},{"index":10,"size":132,"text":"That overgrazing would eventually destroy the range was a common argument in East and West Africa alike. The work of colonial scientists in Niger is representative of francophone scientific and administrative opinion (Peyre de Fabrègues, 1984). The same perceived risk of overgrazing was the basis of the colonists' agricultural policy in Kenya 32 . Pratt and Gwynne (1977, p. 38), working mainly in East Africa, argued that: 'Overgrazing may often be the direct result of human biological needs… It is, however, also a common situation that the pastoralist maintains a herd far larger than needed for his own subsistence. This is usually attributed to sheer greed, prestige, or the concept of livestock as movable wealth. ' Dahl and Hjort (1976, p. 17) were unusually prescient in rejecting the prestige argument about overgrazing."},{"index":11,"size":189,"text":"A leading observer of tropical agriculture, Ruthenberg (1980, p. 332) concluded, at the beginning of the modern livestock research in Africa, that it was '…usually advisable to leave totally nomadic systems undisturbed' because they were unsuited for ranching for both biological and managerial/cultural reasons. Ruthenberg (1980, p. 343) was pessimistic about semi-nomadic systems, contending that regulation of stock numbers in semi-nomadic systems occurred through 'disaster' and not through rational management. Ruthenberg, whose books are deeply insightful nearly 40 years after their final editions, was dissatisfied enough with semi-nomadic herding to refer to its 'malpractices' (Ruthenberg, 1980, p. 342): 'uncontrolled animal densities, over-grazing'; 'inadequate fodder distribution over the year'; inefficient placement of camps to water, resulting in long daily treks; low calf productivity owing to competition with people for milk; and inefficient grazing practices (Ruthenberg, 1980, p. 342). He concluded that neither greater efficiency nor better environmental stability could be achieved without human population control through emigration. He continued to say that replacement of collective land tenure was necessary by allocation to groups or individuals 'whereby grazing rights were both allotted to the herdsmen and enforced' (Ruthenberg, 1980, p. 343)."},{"index":12,"size":193,"text":"Ruthenberg's view of the difficulty of realizing the potential of tropical pastoralism was part of a general pessimism about technical and policy measures. Grigg (1974) had earlier concluded that 'the conversion of pastoralism to ranching has proved difficult' 33 . Jahnke's (1982,  p. 149) review of African livestock systems argued that extensive herding in areas under trypanosomiasis challenge should sometimes 'be left to the fly' given that ranching is not economic and that the recurrent costs of long-term tsetse control are unsustainable on small farms and ranches. Coppock's (1994, p. 273) multidisciplinary study of southern Ethiopia found that previous rangelands development projects had not produced a 'documented increase in cattle offtake or a widespread and sustained improvement in human welfare from veterinary campaigns or ponds, roads, and markets'. Coppock further noted that the 'ranch experiment' in southern Ethiopia had 'failed to transform traditional pastoralism' (Coppock, 1994, p. 35). Mortimore's (2000, pp. 104-105) review of northern Nigeria found that efforts to establish ranches and grazing reserves in the 1960s had failed because of high fixed costs per reserve, competition from cropping and the incompatibility of (enforced) limited mobility in the reserves with erratic rainfall."}]},{"head":"Productivity of subsistence and commercial grazing systems","index":31,"paragraphs":[{"index":1,"size":455,"text":"The 'mainstream view' (as described and criticized by Sandford, 1983a, pp. 11-19) was that African rangelands, like commercial ranches in Latin America, the USA and Australia, should be open-air factories to produce meat. Following this view was the contention that traditional grazing, in Asia and sub-Saharan Africa, was less productive in live weight per hectare or per worker than ranching. This contention encouraged measures to raise stocking rates and offtake from grazing systems and to shift herd composition from females to males to make them more productive in terms of meat supply (Sandford, 1983a: pp. 123-126;Wagenaar et al., 1986, pp. 50-51) and hence more efficient in terms of land use. A few early studies tested the proposition that grazing was less productive than ranching in live weight per hectare (Cossins, 1985 for Ethiopia;Sandford, 1983b, for sub-Saharan Africa;Sandford, 1983a, pp. 123-126;Wilson, 1986, for Mali;Behnke, 1985, for sub-Saharan Africa; Wagenaar et al. 1986, p. 47, for five systems in three countries; Table 15.2). These studies tended to show that grazing productivity per unit of land exceeded that of ranches. Extending the comparison, Breman and de Wit (1983) found that the protein production per hectare in Mali was higher than on large ranches in Texas and Australia. Ocaido et al. (2009) found much higher annual returns per hectare for pastoralism than for ranching when the two were compared in the same area of Uganda. Behnke (1985) concluded that the gap between grazing and ranch productivity in the same environment was usually small, based on observations from Botswana, Kenya, Uganda and Zimbabwe. Behnke further argued that the putative gap between ranch and grazing output had been exaggerated in favour of commercial ranches by cultural bias among external observers, including scientists, who were too eager to find overgrazing and low productivity among traditional herders. Homewood (2008, pp. 63-65), using a larger sample of studies, later restated the critiques of the 'mainstream view', notably the lack of evidence for range degradation.A se- cond counterattack against the 'low productivity of subsistence grazing' argument was that the objective of African rangelands systems was not to produce meat for sale but to produce milk for consumption by the herders and to provide security against drought and disease. This alternative view of the objectives of grazing systems -confirmed notably in the Borana, Maasailand, Kaduna and both Mali studies and in the reviews by Dyson-Hudson and Dyson-Hudson (1969), Nicholson (1984) and Behnke (1985) -is consistent with common features of grazing modes. These are: (i) low weaning weight of calves because of competition for milk with people; (ii) more females than males in the herd to sustain milk production with limited feed; and (iii) seasonal grazing rotations between milk and dry herds to provide milk for subsistence."}]},{"head":"The problem of animal mobility","index":32,"paragraphs":[{"index":1,"size":381,"text":"The ILCA systems studies (Ethiopia, Kenya, Mali, Nigeria and Niger) were not designed to estimate long-term changes in the mobility of grazing ruminants, although they are definitive on the role of traditional mobility patterns in exploiting seasonal changes in feed, water and markets. Older studies (ILCA 1975;Pratt and Gwynne, 1977;Le Houérou, 1989) have the same shortcoming as does the most recent major compendium on African pastoralism (Catley et al., 2013). The FAO/ CIRAD (2013) review of Sahel grazing systems, covering 1970-2010, is the only long-term study of the issue. It found that 'movements have become longer and more dispersed, especially southward' in response to climate, markets and cropping density (FAO/CIRAD, 2013, p. 14). Bourn and Wint (1994) reviewed 20 aerial and ground surveys of livestock and land use between 1980 and 1991 in Mali, Niger, Sudan, Chad and Nigeria. The surveys confirmed the greater importance of stock mobility in the West African subhumid climates, related to the stronger seasonal variability of rainfall and hence of plant biomass in the '750-1250 mm rainfall band' (Bourn and Wint, 1994, p. 9). They noted that seasonal mobility was weaker in arid and humid climates than in the subhumid climates, presumably because the latter is a transition zone in West Africa north of the equator. The single-point sampling frame in Bourn and Wint (1994) does not permit conclusions about long-term changes in mobility. Turner et al. (2014) completed an unusually detailed study of livestock mobility in 32 mixed-farming villages in Mali and Niger. They found that mobility was a dominant strategy among all groups (farmers, herders, artisans and fishers) and that mobility became more dominant as stock density rose. Mobility, and the type of mobility (within village, proximate or distant seasonal transhumance), were not strongly affected by species in the rainy season, although it was reduced for all species in the dry season. Apparent longer-term trends affecting mobility in this study were: (i) insecurity in grazing areas outside the survey villages, whether these areas were proximate or distant, reduced mobility; (ii) higher cultivation densities in the survey villages increased the risk of livestock damage to crops and hence the risk of conflicts among farmers and herders; and (iii) changes in livestock corridors related to planning of water points also hampered mobility and changed its paths."},{"index":2,"size":102,"text":"There is a long chain of research confirming the importance of mobility to pastoralists and highlighting the potential costs to their livelihoods if mobility were to be restricted. This chain includes the ILCA systems studies for Mali, Kenya and Ethiopia (Behnke and Scoones, 1993), the studies on Niger (Hiernaux and Ayantunde, 2004;Turner et al., 2014), the FAO/CIRAD compendium on the Sahel (FAO/CIRAD, 2013) and later de Haan (2016). They confirmed the potential damage to productivity from policy restrictions on mobility, and the long-term adverse trends related to insecurity, cultivation density, and the competition for water among herders and between herders and farmers."}]},{"head":"ILCA's programme","index":33,"paragraphs":[{"index":1,"size":104,"text":"ILCA began from the general idea that not enough was known about African grazing systems, or about mixed systems in which the actual productivity of livestock appeared to be less than its potential, to permit the introduction of technical, managerial or policy changes. The specific origin of grazing systems research was to test the founding hypothesis of ILCA: that there had been '…a failure to integrate the biological, economic and sociological components of research and development programmes' (Tribe et al., 1973, p. 1). Multidisciplinary research was therefore proposed to provide the scientific basis for programmes to expand livestock production and to improve herders' welfare."},{"index":2,"size":95,"text":"ILCA accordingly launched a series of livestock systems studies, which continued for 25-30 years in various parts of sub-Saharan Africa, to investigate the economic, biotic, abiotic and organizational constraints to improving livestock systems. ILRAD's involvement with systems analysis was almost nil for its first decade and began only in the 1980s with its role in the African Trypanotolerant Livestock network (ATLN) and its leadership of epidemiology and economics work of ECF. These studies ultimately filled some of the gap between the development importance of livestock in sub-Saharan Africa and the research effort on livestock problems."},{"index":3,"size":94,"text":"ILCA's studies had two generations. The first generation followed extensive communal rangelands in Ethiopia, Kenya and Mali from the early 1970s until the early 1990s. These studies featured little collaboration with the crop centres. The first generation of ILRI research took place mainly in the LGA/LGH and MRH/MRT types. The second followed smallholder mixed systems in Nigeria, Ethiopia and Mali (with enhanced collaboration among ILCA, the International Institute of Tropical Agriculture (IITA) and ICRISAT), in the Middle East (with ICARDA) and in Central America (with CIAT). The second generation concentrated more on mixed systems."},{"index":4,"size":106,"text":"The broad purposes were to describe how the systems functioned, to define types of production units and to estimate productivity parameters and thereby to understand constraints to productivity. A specific goal of the first studies was to collect primary field data on crop and animal performance, which in the late 1970s were unavailable in most of sub-Saharan Africa. A second goal was to estimate input-output relationships and later to test new technologies. Paul Neate's history of ILCA (ILCA, 1994) maps several milestones: • ILCA's pastoral work began with a compendium on mapping sub-Saharan Africa rangelands (ILCA, 1975), with some reference to Tunisia, India, Iran and Australia."},{"index":5,"size":39,"text":"• The ILCA (1975) rangelands book was followed by studies of browse in Africa (Le Houérou, 1980), the rangelands of Kenya ( de Leeuw et al., 1984;Solomon Bekure et al., 1991), Ethiopia (Coppock, 1994) and tropical Africa (Sandford, 1983a)."},{"index":6,"size":45,"text":"• ILCA published extensive literature reviews on trypanotolerance (Trail, 1979a(Trail, ,1979b)), on livestock in the subhumid zone (ILCA, 1979a) and on small ruminants in Africa (ILCA, 1979b;Gatenby and Trail, 1982), which outlined the state of knowledge and suggested priorities for new research in these areas."},{"index":7,"size":29,"text":"• Research on management and productivity of extensive mixed systems in semi-arid central Mali (Wilson et al., 1983;Wilson, 1986;Wagenaar et al., 1986), with data collection from 1978 to 1984."},{"index":8,"size":25,"text":"• Research on management and productivity of mixed systems in subhumid central Nigeria (von Kaufmann et al., 1986), with data collection from 1978 to 1986."},{"index":9,"size":17,"text":"• Mixed systems in highland Ethiopia (Gryseels and Anderson, 1983), with data collection from 1978 through 1985."},{"index":10,"size":38,"text":"• Mixed systems in semi-arid, subhumid and highland areas of sub-Saharan Africa, plus the related work of McCown et al. (1979), McIntire et al. (1992), the collaboration of ILCA with Pingali et al. (1987) and Winrock International (1992)."},{"index":11,"size":32,"text":"• The productivity of pastures, crop residues and other feeds, with a new focus on browse, which had not been well studied in sub-Saharan Africa before the mid-1970s (Le Houérou, 1980;Walker, 1980)."},{"index":12,"size":39,"text":"• Soils, water and vegetation (complemented by the Dutch-Malian work of Penning de Vries and Djiteye, 1991;and Breman and de Wit, 1983) in semi-arid Mali and by the compendia of ILCA (ILCA, 1975) and of Pratt and Gwynne (1977)."},{"index":13,"size":13,"text":"• The volume of Powell et al. (1995) on livestock and nutrient cycling."},{"index":14,"size":22,"text":"The following section sketches the main characteristics of the research sites and the specific experimental and survey treatments done at each site."}]},{"head":"Borana, Ethiopia","index":34,"paragraphs":[{"index":1,"size":86,"text":"ILCA and national partners studied an area of 15,475 km 2 in the Borana Plateau of southern Ethiopia (Coppock, 1994, p. 39)  34 from 1980 to 1991. The site was chosen within the much larger area of a regional livestock development project that provided background information on the environment, social organization and technology of crop and livestock production (Coppock 1994, pp. 38-60). Previous research on the Borana areas of northern Kenya and southern Ethiopia had concentrated on social organization and not on biological constraints to productivity."},{"index":2,"size":8,"text":"Scientists identified four ecologies in the study area:"},{"index":3,"size":28,"text":"• Subhumid, with varying patterns of livestock, forest, cropping and urban, with annual rainfall as high as 750-1000 mm, and some areas of bimodal rainfall (March-May and September-November)."},{"index":4,"size":20,"text":"• 'Upper semi-arid', with about 600 mm of rainfall, with a mixed landscape of savannahs and woodlands, varying in elevation."},{"index":5,"size":23,"text":"• 'Lower semi-arid', with about 400-600 mm of rainfall in a single season, varying soil types and landscapes of wooded areas and grasslands."},{"index":6,"size":19,"text":"• ' Arid', with varying soil types and landforms, and less than 400 mm rainfall in a single season."},{"index":7,"size":107,"text":"Measures of LGP at five sites were between 114 and 151 days, summed between the 'long' and 'short' rains. The systems would be classified as LGA and MRAc. The longer growing seasons, compared with West African areas of similar rainfall, were related to higher elevation and bimodal rainfall not found in West Africa. The extended growing periods typically produce pastures with a higher CP content and with longer periods of CP availability than in West Africa. The longer growing seasons and greater variation in altitude of East Africa produce a much greater diversity of plant species than in the West African Sahel (Le Houérou, 2009, p. 133)."},{"index":8,"size":61,"text":"Household survey and aerial surveys were used to estimate aggregate resource use in the 15,475 km 2 area studied. A combination of the two methods was used in the 1982-1985 period to arrive at estimates of 66,000 people (4.3/km 2 ), 325,000 cattle (21/km 2 ) and TLUs of nearly 350,000. The latter gives an average stocking of about 4.5 ha/TLU."},{"index":9,"size":88,"text":"A rough estimate of land use in the early 1980s was that perhaps 4% was cropped (Coppock, 1994, p. 85). Measurements of cropped area per household were not reported in the grazing territories (madda) studied. The main crops noted were maize, beans, teff (eragrostis tef) and wheat. No strict division of land use between grasslands and wooded areas was possible at the time but the majority would have had 10-40% woody cover (Coppock, 1994, pp. 84-85). Perhaps half of the grazing territories would have been 'unsuitable for cultivation'."},{"index":10,"size":62,"text":"Livestock provided the largest proportion of income for survey households. The average number of livestock stock units per person in a sample of 49 households from 1981 to 1983 (Coppock, 1994, Table  D1) was 13.8 (range 4.6-39.8) averaged across grazing territories. There were major wealth differences across Borana households, reported as livestock holdings per adult equivalent household member (Coppock, 1994, pp. 174-175)."},{"index":11,"size":9,"text":"Experimental and survey treatments in the Borana programme were:"},{"index":12,"size":12,"text":"• Fodder resources (introduction, conservation, feed gardens and rehabilitation of degraded areas)."},{"index":13,"size":9,"text":"• Herd management (offtake, calf feeding and breed improvement)."},{"index":14,"size":8,"text":"• Water surveys and experiments (Nicholson,   1987a,b, 1989)."},{"index":15,"size":10,"text":"• Novel work on the intergroup inequality of herd holdings."},{"index":16,"size":93,"text":"The Borana volume did not clearly present feed composition although it is evident that pasture was the principal feed. The unusual features of the Borana Plateau compared with West African pastoralism -two rainy seasons, hence less intra-annual variability of grazing and better forage quality throughout the year; a scarcity of cropping, hence less crop residue production; and the constant battle against bush encroachment, hence greater incentive to feed browse -impelled feed research into characterization of pasture and browse and away from work on crop residue quality, which was significant in the Kaduna study."},{"index":17,"size":51,"text":"The Borana study outlined a development path for the region. The first step was to raise stocking rates on undergrazed areas through water development. The argument was that higher well density, and lower labour costs of lifting water, would divert energy spent by animals on trekking and watering into weight gain."},{"index":18,"size":124,"text":"The second step proposed for Borana was to manage stocking rates -numbers per unit of land, species, seasonality and stock preferences among forages -to prevent overgrazing, to adapt stocking rates to rangeland potential and to exploit unused feeds, notably browse 36 . This step has failed as it was originally conceived, because it depended on the assumption that cattle would remain the dominant species and hence excluded species shift as an alternative to the management of cattle stocking rates on common land. The explanation for the failure to manage stocking rates was that it was uneconomic. In long wet cycles with good pasture growth, there was no good reason to destock. When good pasture cycles ended, destocking would occur naturally through sales and mortality."},{"index":19,"size":113,"text":"The third step was to augment external inputs, such as fertilizers and new plant species, to raise primary production and therefore carrying capacity. Using external inputs to raise range productivity has typically failed, as one sees little trace of mineral or organic fertilizer use in African rangelands and/or of introduced herbaceous species. There appears to have been progress in leguminous tree hay making and fodder banks (see Chapter 13, this volume, on planted forages including trees), but the benefit and costs of such methods are poorly known and cannot be attributed to research or extension. A variant was to use external inputs (fencing, mineral fertilizers and new forage crops) to rehabilitate degraded sites."}]},{"head":"Maasailand, Kenya","index":35,"paragraphs":[{"index":1,"size":86,"text":"The ILCA studies of Maasailand in Kajiado County of south-eastern Kenya -of water, soils, climate, animal health, herd management, vegetation, pastures, economics and labour usebegan in 1978 and continued into the mid-1980s 37 . The outstanding study is that of Solomon Bekure et al. (1991), covering 1979-1991, followed by the later contributions of Homewood and Rodgers (2004), among others. The general area of Maasailand covering much of southern Kenya and north-central Tanzania has since been studied by scientists of many disciplines over the past 40 years."},{"index":2,"size":108,"text":"The Maasailand studies differed from the others in two respects. One was that they could use historical data on system structure and productivity that was unavailable in Mali, subhumid Nigeria and later Niger 38 . Another was that the Maasailand had been the site of dramatic changes in pastoralism. Land availability and therefore herd mobility had diminished with the appropriation of Maasai lands by the colonists at the beginning of the 20th century (as summarized for an area north of Kajiado by Reid et al., 2008b). The later introduction of group ranches by the independent Kenyan government in the 1960s further limited the land and mobility of Maasai."},{"index":3,"size":81,"text":"The Maasailand system, classed as LGA by Robinson et al. (2011) at an altitude around 1500 masl, was semi-nomadic with a low human population density, poor market access, low stocking rates, milk as the staple and very little cropping. The population density was about eight persons per km 2 in 1979 (Solomon Bekure et al., 1991, p. 16). Stocking rates ranged from 1 TLU/ha to 0.25 TLU/ha across the study sites with significant wet-and dry-season variation. Cropping density was almost nil."},{"index":4,"size":74,"text":"Species and breed composition of herds and flocks changed little in the years before and after the group ranches. Herd structure did not change notably over time from a two-thirds share of females; offtake changed little as a share of herd size. There was limited infrastructure development and new road, water and dipping infrastructure fell into disuse from lack of maintenance. The use of variable inputs was uncommon except for acaricides and veterinary drugs."},{"index":5,"size":98,"text":"Inequality in stock ownership was high. The Gini coefficient of cattle and small stock ownership in three Kenya group ranches (1980)(1981) was about 0.50 (Solomon Bekure et al., 1991, p. 3) but there are no corresponding earlier data. The effect of group ranches on the inequality of stock ownership was unknown. A later summary of the effects of subdividing group ranches found that population pressure in Maasailand in the final quarter of the 20th century would have reduced livestock/person numbers enough to make ranch subdivisions inviable without outside income from wage labour or remittances (Thornton et al., 2006)."},{"index":6,"size":32,"text":"Experimental and quasi-experimental treatments were: (i) higher-yielding fodder resources; (ii) policies to encourage greater offtake from the herd; (iii) grazing and water management; and (iv) management and health of sheep and goats."}]},{"head":"Niono, Mali and the Malian Delta","index":36,"paragraphs":[{"index":1,"size":140,"text":"The Mali livestock systems (LGA) research was led by ILCA and the Malian Institut d'Economie Rurale (IER) around the town of Niono in central Mali near the Niger River. There were two studies, one of the agro-pastoral systems in areas farther from the Niger and the second in and around the Inner Delta of the Niger. The principal themes of both were animal health and productivity, herd management, vegetation and grazing patterns, and crop agronomy. The areas were LGA, with generally low human population density, highly variable seasonal livestock density, monomodal rainfall, good market access in regional towns, and a diverse mix of rain-fed and irrigated crops near the animal production zones. The altitude was between 200 and 300 masl in the Niono and Interior Delta areas. The hot dry conditions (Köppen climate BSh) made trypanosomiasis rare and Theileria unknown."},{"index":2,"size":115,"text":"The main reports were those of Wilson (1986) and Wagenaar et al. (1986). Wilson (1986) was a research project on cattle and small-ruminant productivity among agro-pastoralists 39 near Niono in north-central Mali with field work over the period 1976-1983. The study by Wagenaar et al. (1986) from 1979 to 1983 covered cattle productivity among agro-pastoralists in the Inner Delta of the Niger River in central Mali. Both Mali studies were in areas without major trypanosomiasis challenge and where internal parasites were the principal animal health risk. The Inner Delta study area differed from the Niono one chiefly in that water was available year-round and that the divide between wet-and dry-season pastures was much less pronounced."},{"index":3,"size":174,"text":"There were associated studies of vegetation and biomass carried out by Breman and Cissé, (1977), Hiernaux (1980Hiernaux ( , 1984) ) and Hiernaux et al. (1983), and on primary productivity and soil-water relationships led by the University of Wageningen (Penning de Vries and Djiteye, 1991). A later comprehensive review of Sahel grazing systems (Le Houérou, 1989)  The Mali studies differed from the others in not having formal experiments with productivity treatments. They relied on statistical analysis of implicit 'treatments' arising from surveys of herd and flock effects on reproductive performance, nutritional status, young stock mortality and seasonal respiratory diseases. The Niono study (Wilson, 1986, p. 108) used the statistical results to make general recommendations about feed, management, and prophylactic and curative health measures without benefit-cost analysis. The Inner Delta study did a more thorough analysis of calf weaning policies leading to recommendations that could be applied by herders (Wagenaar et al., 1986, pp. 45-51) and leading to other proposals for pasture management involving existing forage species rather than the introduction of new forage species."}]},{"head":"Kaduna, Nigeria","index":37,"paragraphs":[{"index":1,"size":132,"text":"The Kaduna sites (von Kaufmann et al., 1986)  40  were chosen to contrast livestock systems under different land-use patterns imposed in part by the natural environment and in part by policy. The Kaduna area is in Köppen climate Aw (tropical savannah). The livestock systems are LGT and MRT. Two of the three sites -Kurmin Biri, between a grazing reserve and a national forest, Abet -were around 2500 km 2 . Annual rainfall in Kurmin Biri and Abet was 1200-1300 mm in one season. Ganawuri is a higher-cultivationdensity site, with slightly higher rainfall, of which 800 km 2 was subject to aerial survey and 40 km 2 was subject to ground truthing (von Kaufmann et al., 1986, p. 45). Kurmin Biri and Abet were at about 600 masl and Ganawuri above 1250 masl."},{"index":2,"size":76,"text":"Waters-Bayer and Taylor-Powell (1986a) summarized population and land use in the three sites. They distinguished three groups: (i) pure pastoralists, who do not practise cropping and are nomadic; (ii) agro-pastoralists, whose main activity is livestock but who do have crops and are less nomadic than 'pure pastoralists'; and (iii) mixed farmers, whose main activity is crops but who do keep some ruminants; animal traction for power was rare in this production type, which was the largest."},{"index":3,"size":92,"text":"Cultivation intensity varied from 5-15% of all land in Kurmin Biri to 25% in Abet to as much as 40% in Ganawuri. Wet-season livestock density ranged from five heads/km 2 in Kurmin Biri to 25 in Abet. Dry-season stock density ranged from 10-20 head/km 2 in Kurmin Biri to 40 in Abet to 20-25 in Ganawuri. Kurmin Biri had at one time been declared tsetse free, although the vector was apparently still found at Abet during the study period. Ganawuri, at some 1000 masl, had a cooler climate and lower vector incidence."},{"index":4,"size":35,"text":"The Kaduna studies carried out careful experiments of practices to improve productivity. These tests focused on supplementary feeding of cattle, planted forages (mainly Stylosanthes spp.) and on manure for crop production (Powell, 1986a,b;Mohamed-Saleem, 1986a,b;Bayer, 1986)."}]},{"head":"Fakara, Niger","index":38,"paragraphs":[{"index":1,"size":162,"text":"The ILRI studies in the Fakara subregion of western Niger -of soils, vegetation, grazing practices, soil nutrient recycling, herd management, vegetation, household economics and labour use -followed earlier work in the same zone by ICRISAT, ILCA and the International Fertilizer Development Center (IFDC), beginning in the early 1980s. The Fakara studies (Williams et al., 1999, and Hiernaux and Ayantunde, 2004, reporting on work from 1994to 2005) were in Köppen climate BWh at altitudes of less than 300 masl. Annual rainfall in the Fakara is reported to have been 560 mm from 1905 to 1989. There is a monsoon seasonal pattern with rainfall from June to September, a transition period in October and November, a relatively cool season from December to February and a very hot dry season from March until the onset of the rains. Surveys in the 1990s showed the population density to be low, in the order of 10-50 rural persons per km 2 depending on the proximity to water."},{"index":2,"size":108,"text":"The livestock system is a mix of seasonal and annual transhumance. Animal density per unit of land was low, ranging from 5 to 12 TLUs/km 2 . Animal density in TLUs per person was comparatively high, ranging from 0.15-0.20 among poorer families to as high as 2.5 among richer ones. The principal feeds were natural grazing, mainly of browse and annual grasses, and crop residues. Animal disease prevalence is low enough that Hiernaux and Ayantunde (2004) did not mention it at all; the principal disease problem appears to be internal parasites. The hot, dry conditions made tsetse and therefore trypanosomiasis very rare. ECF is unknown in West Africa."},{"index":3,"size":58,"text":"The Fakara field studies were complemented by station investigations of dry-season supplementation, sheep fattening, alternative grazing practices such as rotational grazing and seasonal grazing, and soil fertility management including manuring and mineral fertilization of arable crops. The field and station studies were subsequently used in bioeconomic models of soil fertility management and household behaviour (Barbier and Hazell, 1999)."}]},{"head":"What did the pastoral systems find?","index":39,"paragraphs":[]},{"head":"Sedentarization","index":40,"paragraphs":[{"index":1,"size":116,"text":"Three tenets of the 'mainstream view' of pastoralism were that it was unproductive; that it inevitably involved overstocking, which led to environmental damage and to conflicts with farmers; and that reducing nomadism was a path to higher productivity and less pasture damage due to lower grazing pressure. One result of these beliefs was that sedentarization of herders was sometimes advanced as an appropriate, even urgent, policy. In addition to raising productivity, sedentarization would, moreover, allow provision of housing, education and health services to herders. It followed from this view, which was widely held among administrators and scientists, that policy should seek to sedentarize completely, settle partially by season or otherwise restrict the mobility of pastoral groups."},{"index":2,"size":113,"text":"A prediction from the agricultural evolution literature is that sedentarization of nomads responds to two forces. The first is a push effectpopulation pressure and the expansion of land for crops or wildlife cause herders to lose grazing access. This loss of grazing limits feasible herd numbers, restricts mobility and productivity of the remaining animals, and eventually obliges pastoralists to abandon nomadism for cropping in whole or in part. A related prediction was that disasters such as drought or animal disease would aggravate herd loss and therefore accelerate sedentarization. The second force is a pull effect -by sedentarizing, herders can benefit from access to land, cereal production, crop residues, water and even political representation."},{"index":3,"size":152,"text":"The strength of these predictions has varied greatly by agroclimate, population density and the emergence of markets for milk, commercial meat and cash crops. Both 'push' and 'pull' predictions have been confirmed in the densely populated areas of northern Nigeria by Waters-Bayer and Taylor- Powell (1986b), and also by Blench (1994) who gives many examples of successful spontaneous sedentarization of herders throughout Nigeria, and in at least three pastoral systems of southern Ethiopia and northern Kenya (Fratkin, 2001). In more recent research in semi-arid western Niger, Hiernaux et al. (2009) showed a progressive sedentarization of herding groups into rain-fed farming, caused by drought, with a concurrent integration of crop and animal activities on previously settled farms, caused by demand for land and access to fodder. Northern Nigeria, and parts of semi-arid India, illustrate the impact of population density and service availability on sedentarization, while East Africa illustrates the impact of recurrent droughtsc."},{"index":4,"size":176,"text":"A related prediction was that pastoralism, because of herders' knowledge of animal production, would evolve more easily into mixed farming than arable farming without livestock. The latter would involve aspects of arable farming, ley farming, animal traction, use of manure for cropping, and commercial livestock, such as dairying or ranching for meat. This set of predictions seems to have succeeded only in some aspects of arable farming and to have generally failed with respect to ley farming, animal traction, planted forages and commercial livestock production. The successful predictions -use of manure in cropping, use of crop residues to feed animals, use of animals for transport and cultivation -were confirmed by McCown et al. (1979), in a study of agro-pastoralists in central Nigeria (Powell and Taylor-Powell, 1984), in the ILCA research on agro-pastoralists in northcentral Mali (Wilson, 1986), more generally across semi-arid and subhumid Africa (McIntire et al., 1992), particularly Chapters 5 and 6) and Hill's (1982, pp. 21-23) book on two dry farming villages in Hausaland and in the international comparisons by Baltenweck et al. (2003)."},{"index":5,"size":223,"text":"Enforced sedentarization policies generally failed. Grigg (1974, p. 122) that '…efforts to establish Maasai families on 'ranches'…foundered on the refusal of the Maasai to cull their herds to a level compatible with the grazing resources available, and their reluctance to sell their cattle' 42 . Sandford's (1983a) review of African and other work (ex-Soviet Union, the Middle East) found that efforts to settle nomadic herds, while integrating their owners into cropping or moving them into ranching, usually failed because of the cost of foregone mobility and the unwillingness of pastoralists to abandon their livelihoods. The sedentarization of herders in northern Nigeria was indeed found to create conflicts with farmers, as indicated by numbers of lawsuits for crop damages caused by animals (Waters- Bayer and Taylor-Powell, 1986b, pp. 213-214). Blench (2001, pp. iv and 61-64) found that policies of enforced sedentarization 'had a very unsatisfactory history', citing the instances of Tanzania, Iran, Kenya and Somalia. Homewood (2008) cites many examples from East and West Africa where explicit or implicit settlement policies have failed, sometimes with disastrous consequences for the rights and livelihoods of herders. Induced sedentarization may, moreover, have produced land degradation by concentrating people and animals in smaller areas of scarce resources, such as dry-season water and pastures, but it is not possible to quantify such adverse effects with existing information 43 ."},{"index":6,"size":131,"text":"After enforced mobility restrictions during the collective era in Mongolia, herders 'very rapidly reverted to traditional mobile transhumance' when the collectives were dismantled (Suttie et al., 2005, p. 294). Arable farming was indeed taken up by pastoralists (Blench, 1994(Blench, , 2001)), with marked differences among subregions in sub-Saharan Africa. The reasons for sedentarization were necessity -herders had to settle after losing most or part of their animals, in part because of long-term conversion of grazing and forest to cropland and in part because of opportunity to exploit interactions between crops and livestock in the same enterprise. Sedentarization could not be sustainably enforced by public policy. It was also reversible -as soon as assets would allow, settled pastoralists would revert, at least for part of their family, back to a mobile system."},{"index":7,"size":129,"text":"The extensive systems most carefully studied -Borana, Maasailand, Kaduna, north-central Mali, south-western Niger, the Kenyan-Tanzanian border -seem to have regressed in area under pressure from arable farming (Kaduna), bush encroachment (Borana) and wildlife reservations (Maasailand). One study of northern Nigeria 44 found that the principal land-use change was loss of woodland. Another factor, which has been more difficult to map, is the taking of land for cropping, whether by governments for sugar estates in the Awash Valley of Ethiopia or by small holders on what was formerly lowland seasonal grazing in central Nigeria (von Kaufmann et al., 1986). The review by Reid et al. (2005, pp. 46-48) noted that the encroachment of crop farming had reduced wildlife (wildebeest) populations but not those of cattle on the Kenyan-Tanzanian border areas."}]},{"head":"Grazing organization","index":41,"paragraphs":[{"index":1,"size":47,"text":"A concomitant policy to sedentarization was to modify the traditional social organization of grazing. This traditional organization was typically labelled as 'collective', 'communal' or 'common property'. It was believed to provide weak incentives to sell animals by lowering the private costs of grazing while simultaneously encouraging overgrazing."},{"index":2,"size":133,"text":"A common measure was to re-allocate pasture by giving selected groups exclusive access to grazing. Such policies were enforced in Maasailand (Solomon Bekure et al., 1991) and in north-central Nigeria (von Kaufmann et al., 1986;Blench, 1994). In Maasailand, this was the introduction of group ranches 45 . In northern Nigeria, it was the definition of grazing reserves with remapping of herding family access to the reserves. In several francophone nations of West Africa, it was the promulgation of a 'code pastoral' that delimited (very large) grazing areas and corridors through arable lands but without reorganization of pastoral groups (FAO/CIRAD, 2013). In Mongolia, it was the collectivization of pasture lands during the later Communist era (1950s-1980s;Mearns, 1996, p. 310) with reorganization of herding districts into brigades under the control of the local collective administration."},{"index":3,"size":111,"text":"There were common problems in each attempt at reorganizing pastoral groups and their grazing rights. First, there was the cost of restricted mobility in terms of increased exposure of herders to drought (or, in Mongolia, to extreme cold) and related pasture loss. Second, there was the devaluation of traditional herding authorities in allocating resources and in managing conflicts about pastures and water. Third, there may have been some productivity loss caused by greater exposure to epizootics as herds became denser in smaller areas. Last, there were conflicts between herding and farming groups, which may again have become more frequent as herds sought access to smaller areas for water and dry-season pasture."},{"index":4,"size":132,"text":"Organizational changes in grazing have often been proposed and have nearly as often failed. An early example of a failure to manage mobility was the attempt to introduce grazing blocks among cattle and camel herding pastoralists in semi-arid north-eastern Kenya (Helland, 1980). Helland found that the complex management required for the grazing blocks was incompatible with the traditional institutions charged with managing the seasonality of water and forage in the area. Galaty (2013a) found that group ranches in Maasailand in both Kenya and Tanzania harmed the interests of the traditional herders through a process of institutionalized land grabs. Similarly, confiscation of grazing lands and associated settlement policies by the Imperial and derg regimes of Ethiopia from the 1950s to the 1980s had adverse effects on pastoralists' interests (Mulatu and Solomon Bekure, 2013)."},{"index":5,"size":242,"text":"One path by which subdivision of grazing would have affected the distribution of animals was through its effect on pasture productivity. Subdivision tended not to affect productivity in ranges of very low or very high productivity but tended to reduce it significantly on ranges of intermediate productivity. The effect of restructuring on the inherent seasonality of pastoralism, especially on dry-season grazing, is not clear at the different study sites. In Maasailand, Solomon Bekure et al. (1991, p. 35) found that 'the degree to which group ranches altered management strategies cannot be determined with available data'. The same study noted impressions of changes in watering frequency, use of salt licks and acaricides, and introduction of the improved Sahiwal breed, but could not quantify these changes or estimate their impact on productivity 46 . Ellis and Swift (1988) had earlier shown that the non-equilibrium rangelands model depended so much on mobility that demarcation of ranches would damage system productivity. The effects of conversion, and related changes, were carefully studied by Angassa and Oba (2008) in Borana. They found that range enclosures led to grazing land fragmentation; associated with new arable farming, the overall effect was a degradation of range productivity. Earlier work related a 'decline in total biomass production and animal performance' to 'increases in human and animal populations' and to 'decreases in grazing land' (Angassa and Beyene, 2003). The adverse effects of restricting pastoral mobility were also observed in Inner Asia (Sneath, 1998)."},{"index":6,"size":93,"text":"A related example is from Borana. The land reform imposed by the derg government (1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991) in Ethiopia was associated with 'semiprivatization' of grazing lands, range enclosures and the suppression of fire as a bush management practice (Angassa and Oba, 2008). These policies had the aggregate effect of '…forage scarcity and greater vulnerability of stock during drought years'. These parallel findings from Maasailand and Borana are good examples of policy ignoring what research had predicted about the value of traditional management in maintaining rangeland productivity and in reducing the environmental costs of invasive species."}]},{"head":"Water","index":42,"paragraphs":[{"index":1,"size":28,"text":"Water was long understood to be an essential component of higher pastoral productivity, of resistance to drought and of recovery after drought. Basic questions about water development were:"},{"index":2,"size":16,"text":"• Could potential water supply support livestock production growth? What are the barriers to water development?"},{"index":3,"size":10,"text":"• Can water consumption by pastoral stock become more efficient?"},{"index":4,"size":11,"text":"• What has been the impact of water development on pastoralism?"},{"index":5,"size":112,"text":"• Have there been adverse effects of water development?   Research on pastoral water was of four types: (i) literature reviews; (ii) experiments on stations and in other controlled conditions; (iii) original field investigations, including the system studies; and (iv) recent surveys and modelling of the 'water footprint' of livestock, which became a way of estimating total water use by animals. literature reviews. ILCA commissioned three literature reviews of livestock water use in Africa. King (1983) studied water intake and metabolism by different species and their effects on productivity, Sandford (1983b) focused on water planning and Classen et al. (1983) characterized water use in tropical Africa 47 . Sandford (1983b) recommended the following:"},{"index":6,"size":17,"text":"• Collecting new data on livestock and water point density, which existed then only in arid zones."},{"index":7,"size":18,"text":"• Gathering new data on watering practices in semi-arid areas, which were (then) scarcer than in arid areas."},{"index":8,"size":18,"text":"• Studying actual water use and comparing it with optimal use to test the hypothesis that producers overwater."},{"index":9,"size":8,"text":"• Studying site design to reduce production costs."},{"index":10,"size":11,"text":"• Recovering water costs from users to pay for system maintenance."},{"index":11,"size":46,"text":"Water for pastoralism has expanded greatly since the 1970s, pushed by voluminous public and private investments. Research has identified defects in the design and management of pastoral water supply (Sandford, 1983b, pp. 63-67;Homewood and Rodgers, 2004, pp. 249-253), but its contribution to better designs is unproven."},{"index":12,"size":90,"text":"experiments. An important point in the studies by King (1983) and Sandford (1983b) is the effects of water availability on feed intake in arid zones and on the energy used in trekking to water. Restricted water availability by season and because of distance incurred costs of foregone milk and live-weight production; estimates of trekking energy and its productivity effects could therefore be used to optimize water investments. The evidence of Sandford (1983a, pp. 73-85) on water-productivity effects was generally positive from arid zones in India, Australia and the Middle East."},{"index":13,"size":158,"text":"Experiments on water and productivity were sparse in Africa, despite the importance of the water-energy-productivity relationship stressed by King (1983). A rare and important contribution to understanding water-productivity functions was Nicholson (1987a;1987b, 1989) in Borana. He established that traditional 2-day and 3-day watering of Zebu cattle would not compromise productivity if the animals were adequately fed 48 . There are few other experiments that studied water-use efficiency among African pastoralists. The reviews of range productivity of ILCA (1975), Pratt and Gwynne (1977, for East Africa), Le Houérou (1989, for West Africa) and de Haan (2016) scarcely mention water except to apply standard consumption coefficients per animal. The limited research on water and livestock productivity in arid Africa can be partly explained by the fact that additional water is seen as unambiguously good. The demand for water among pastoralists is so high that public agencies and private water producers are often willing to invest beyond what is technically optimal."},{"index":14,"size":6,"text":"system studies and other field investigations."},{"index":15,"size":180,"text":"The pastoral system studies (Niono, Mali Delta, Maasailand, Borana and Fakara) identified water as a constraint, yet their contributions to answering the water questions are surprisingly limited. The Maasailand and Borana studies are the only ones to answer two of the basic questions about water adequacy for production growth and the cost of new production, while the Borana study is the only one to conclude about efficiency. The Niono and Mali Interior Delta studies mention water only in passing and have nothing quantitative on consumption, production, cost or efficiency. The subhumid Nigeria study says little about water, except to report daily livestock watering times (von Kaufmann et al., 1986, p. 431) and to make the evident point that seasonal water is less limiting than it would be in a semi-arid zone such as Borana or Niono. The Maasailand study did detailed research on pastoral water infrastructure (type, scale and location), production and consumption. The principal books on African pastoralism, all based at least in part on field investigations since the mid-1970s 49 , say very little about technical water issues."},{"index":16,"size":109,"text":"The pastoral studies usually made general qualitative recommendations to raise water production, to allocate water more equitably among users and to prevent damage around wells and ponds. Water research had little or no impact on productivity except in the sense of understanding system dynamics and explaining geographic patterns (e.g. Mesele et al. 2006 on the Yabelo district in Borana). One of two major recent pieces on African pastoralism analyses many policy/technical options but does not treat water development as one of them (de Haan, 2016); the other (Cervigni and Morris, 2016) makes the rather general recommendation to develop 'water resources to allow better access to underexploited rangelands' (p. 85)."},{"index":17,"size":91,"text":"adverse effects of water development. There are adverse effects of water development. The main negative effect is soil erosion caused by intensive cultivation, or by trampling stock, near water points (Sandford, 1983a, pp. 76-78;Wilson, 2007). A 4-year station study of grazing effects on soil properties in western Niger found variable impacts on soil compaction and water infiltration as functions of grazing pressure (Hiernaux et al., 1999). Pratt et al. (1997) did not quantify the adverse effects of water development although they argued that such effects can be managed with simple technologies."},{"index":18,"size":95,"text":"Another potential adverse effect is that mechanization of boreholes might reduce labour demand. If labour is an important mechanism of patron-client relationships among animal-rich and animal-poor herders, then mechanization can weaken such relationships (Coppock, 1994). Lower labour requirements owing to mechanization would reduce the bargaining power of poor pastoralists and deprive them of employment, income and, potentially, finance. The West African literature is silent on possible labour displacement effects of wells, although it says a great deal about water as a cause of conflict between herders and farmers (Wilson, 1986;Wagenaar et al., 1986;Le Houérou, 1989)."},{"index":19,"size":159,"text":"water footprint and productivity. The water footprint literature can provide information about changes in feed composition and wateruse efficiency per animal (Hoekstra and Mekonnen, 2012; see also Chapter 11,this volume). One application of this literature is to partition change in total livestock water use per unit of output as the sum of direct consumption by animals and indirect consumption by forage plants. One long-term study in arid, semiarid and humid zones of Kenya found that total water consumption by cattle, sheep and goats, and camels in each of the three zones was closely proportional to animal numbers (Bosire et al., 2015(Bosire et al., , p. 36, covering 1977(Bosire et al., -1990(Bosire et al., and 2001(Bosire et al., -2012)). This implies that indirect water consumption has not changed relative to the total because feed composition has not changed, nor has water-use efficiency per animal. Additional inferences are that species and breed composition within zones have not shifted towards water-efficient animals."}]},{"head":"Feed systems","index":43,"paragraphs":[{"index":1,"size":114,"text":"Most African pastoral research focused on its geography, society and anthropology until the volumes of ILCA (1975) and Pratt and Gwynne (1977) and the initial ILCA literature reviews. While it was long evident that feed, after disease, was the principal limitation to livestock production in grazing systems, older research had done little on possible technical changes in feed, such as introducing fencing, legumes, exotic grasses, fertilization or mechanization 50 . The technical aspects of older work emphasized water and veterinary problems almost exclusively, with less attention to other biological issues, to economics or to introduced technical change through research and extension 51 . Feed research in Africa therefore began from a comparatively weak basis."},{"index":2,"size":62,"text":"Feed systems research had four components: (i) field characterization of available and potential feeds and their use, with emphasis on rangelands characterization (many of the papers in ILCA, 1975); (ii) station characterization of the productivity and nutritive value of such feeds as planted forages, crop residues, and browse; (iii) field experiments of these feeds; and (iv) adoption studies of promising new materials."},{"index":3,"size":128,"text":"field characterization. The early pastoral studies found that feed consisted of grass, forbs, crop residues and tree browse, as shown in ILCA (1975), Fricke (1979, for Nigeria) and in the references in Le Houérou (1989), McIntire et al. (1992) and Baltenweck et al. (2003). Purchased feeds and planted forages were practically nonexistent, even in stall-feeding systems where crop residues and cut grasses dominate. Even in South Asia and highland Ethiopia where the use of animals for power is common and hence creates a constant source of demand for feed among immobile animals, the principal forage was residues from multidimensional cereals, such as teff, sorghum, pearl millet, maize or wheat. For nomadic herders without arable crops, pasture grazing and cereals residues, obtained by transactions with farmers 52 , dominated."},{"index":4,"size":65,"text":"Natural pastures, crop residues, browse and planted forages were the major themes of the initial ILCA systems studies 53 . Every pastoral study started from the understanding that average feed availability from all sources -grazing, cut grass, shrubs, forbs, browse and crop residues -limited animal productivity; and that interannual and interseasonal variability in feed supply was a constant threat to herd viability 54 and growth."},{"index":5,"size":75,"text":"Every study of mobile grazing recommended improvements to feed quantity and qualities. Suggested improvements involved planting legumes and more productive grasses (both African and exotic species), control of bush encroachment, fertilization of pastures and grazing management. For example, the study by Penning de Vries and van Heemst (1975) argued that '…it is expected that appropriate legumes and Rhizobium strains can be found to increase the soil nitrogen supply in Sahelian natural grasslands and agricultural fields'."},{"index":6,"size":130,"text":"experiments. Some experiments on feed productivity were done by the colonial and national programmes before the advent of international livestock research. The general conclusion of the many detailed empirical investigations of feed improvements in pastoral zones of sub-Saharan Africa is that it has been practically impossible to make economic improvements in range productivity under African conditions. This holds under such treatments as planted forages, whether legumes or grasses, and with soil fertility amendments, such as mineral fertilizers or manure. Pratt and Gwynne (1977, pp. 110-128) arrived at this conclusion for East Africa. In West Africa, Le Houérou (1989) found the same when referring to more than three decades of research, as did Fricke's (1979) review of field and experimental work on cattle in Nigeria in the 1960s and the 1970s."},{"index":7,"size":46,"text":"A careful long-term experiment was the Projet Productivité Primaire au Sahel (PPPS) study of Mali, (Penning de Vries and Djiteye, 1991), which found that: (i) soils could support higher stocking rates where water is available; and (ii) grazing cover management reduced soil degradation and bush encroachment."},{"index":8,"size":214,"text":"Studies of soil fertility management using mineral fertilizers in African grazing systems were rare because the null hypothesis of no treatment effect was strong, even from qualitative observations. Introduced treatments -fertilizers, recycling manures or residual vegetation, planted forage legumes or grasses -have not generally succeeded. The magisterial work of Penning de Vries and Djiteye (1991, p. xix, passage translated from the French by the editors) covering 5 years of experimentation in semi-arid central Mali noted that, 'The immediate introduction of new forage legume species is likely to fail for the same reasons -competition from cereals, low germination capacity, lack of soil P, and low rainfall'. The review by Thomas and Lascano (1995) found benefits in soil quality, pasture productivity and animal productivity from mixing legumes into acid-soil pastures in Latin America; however, they did not specify the ultimate adoption of the proposed legume innovations and hence it is not possible to estimate the development impact of their research. Macharia et al. (2011) showed that introduced legumes on semi-arid pastures in Kenya could improve soil quality (pH, organic carbon, nitrogen and potassium) on station, but the on-farm phase of the trial only lasted 1.5 years, did not conclusively show a soil-quality effect and apparently did not lead to field adoption of legume technologies 55 ."},{"index":9,"size":295,"text":"Studies at one site in semi-arid central Mali (Penning de Vries and Djiteye, 1991) established the basic abiotic (soil and water) and biotic (competition from grasses) constraints to planting legumes in extensive grazing systems. These conclusions are replicable throughout the semi-arid tropics of Africa and South-east Asia; exceptions in Latin America are discussed in Chapter 12 (this volume). The Mali research concluded that, in arid situations of less than 750 mm annual rainfall, the introduction of legumes into natural pastures was unlikely to be economic (Penning de Vries and Djiteye, 1991, p. 435), in part because of higher temperatures that lowered legume yields and in part because of adverse meat:fertilizer price ratios (Penning de Vries and Djiteye, 1991, p. 470). pasture improvement 56 . Pratt and Gwynne (1977, p. 122) analysed the technical and managerial aspects of sown pastures in East Africa. Despite not considering the costs of sown pastures, Pratt and Gwynne (1977) concluded that research on range reseeding using exotic grasses showed 'little promise' and should be considered 'a last resort'. Ruthenberg (1980, pp. 110-125 and 322-355) was negative on ley farming for pasture improvement for the same reason -in the absence of dairying, expected benefits were negative. The PPPS study in Mali found little average effect of reintroducing perennial grasses (Penning de Vries and Djiteye, 1991, p. 469) but some effect on the 'stability' of soils and pastures. A hypothetical model of ley farming in a cotton-based system of southern Mali gave inconclusive results and does not seem to have stimulated adoption of pasture fertilization (Bosma et al., 1999). Doppler's (1980) book on beef cattle production in subhumid Togo found that mineral fertilizers and pasture improvements were generally too unprofitable for smallholders, although they would be suitable for ranches and larger farms."},{"index":10,"size":217,"text":"ILCA studied the agro-pastoral areas near Niono in central Mali (Hiernaux et al., 1983;Wilson, 1986) over the period 1978-1984. Wilson et al. (1983, Chapter 6) elegantly describes vegetative composition, tree density and cover in the three principal land-use types near Niono. Its analysis of the species composition and yield of natural pastures compared treatments with manure, added phosphorus, grazing management and ploughing. It showed (Wilson et al., 1983, pp. 47-50), as had the PPPS work, that the Sahelian rangelands had a high production potential but that added mineral fertilizers at a rate of 22 kg ha -1 were 'probably not profitable' (Wilson et al., 1983, p. 50, authors' translation of the French original). Farm management research near Niono station, as reported by Toulmin and Fulton (1983, pp. 107-119), found overseeding of introduced forage species to be unsuccessful for both grasses and legumes in agronomic and economic terms over the years 1977-1980. An example of the benefit:cost ratio of planting a tree legume (Leucaena leucocephala), simulated under commercial beef ranch conditions, came from the initial ILCA cattle model in Botswana (Cartwright et al., 1978, pp. 55-58). Depending on projected tree establishment costs and Leucaena yields, the model projected an IRR of between 2.7% and 22.4% at median establishment costs and tree yields (Cartwright et al., 1978, p. 78)."},{"index":11,"size":190,"text":"Recent experience is that the use of planted or protected forages, whether as arable crops or as browse trees and shrubs, has grown very slowly from the 1960s to the present. Such forages are still insignificant in East African and West African pastoralism, even where formerly highly mobile grazing systems have evolved to include more arable farming (e.g. Beyene, 2016, for eastern Ethiopia). bush encroachment. Bush encroachment was carefully studied many years ago. The basic East African text on rangeland management (Pratt and Gwynne, 1977, pp. 128-138) gave an overview of 'troublesome' bush species and their potential control, including by fire, machines and grazing (e.g. using goats to graze species unpalatable to cattle and sheep). The Borana study (Coppock, 1994) reviewed bush encroachment and control in the context of grazing dynamics, vegetation succession and soil. Its conclusion (updated in Angassa and Oba, 2008) is that there is no general prediction about the effects of bush encroachment, which might even be positive in some stages of land-use evolution. Prediction of crop encroachment reducing pasture area and pasture productivity was later confirmed by satellite image studies in Borana (Mesele et al., 2006)."},{"index":12,"size":141,"text":"The economic analysis of bush control is not well established. Pratt and Gwynne (1997) apparently concluded that fire was the most cost-effective technique for controlling bush encroachment in East Africa but did not systematically compare fire with other methods. Sandford (1983a) concluded that chemical, mechanical and biological methods of bush control are unlikely to be effective in the African tropics, as did Homewood and Rodgers (2004, p. 252) for the rangelands of the Ngorongoro Conservation Area of Tanzania. Fricke (1979) did not really discuss the financial aspects of bush management but noted the importance of fire for the sustainability of grazing systems in Nigeria. The central Mali study at Niono ranch examined the effects of fire, species and cutting on vegetation, palatability and soil quality but did not review chemical or mechanical control (Penning de Vries and Djiteye, 1991, pp. 346-352)."},{"index":13,"size":76,"text":"Chief among the reasons for the failure of pasture improvement and related efforts to control bush encroachment through biological means (introduction of competing crops and use of mineral fertilizers) are low soil fertility and lack of water (Penning de Vries and Djiteye, 1991). There is some evidence of benefits from overgrazing in terms of preservation of perennials via the effect of lesser competition from heavily grazed annuals and protection against invasive species, again via lessened competition."},{"index":14,"size":153,"text":"impact. The impact of feed improvements in grazing systems has been very weak in sub-Saharan Africa (Table 15.3). Over the 40 years of research on grazing systems, novel plant materials remained a small fraction of feed intake, although total animal production grew substantially. There is little evidence of planted forages in sub-Saharan Africa in extensive grazing systems, even on limited areas such as feed gardens for segments of mobile herds or as calf supplements. There have been successes in humid areas of Latin America (see Chapter 13, this volume), but these have most often been on large ranches or on commercial dairies. From the Mediterranean mandate area of ICARDA, we know that long-term research on Medicago sativa was unsuccessful in achieving agronomic or development impact on a large scale 57 although there have been some benefits from feeding spineless cactus (Chapter 13, this volume;Jutzi and Rich, 2016) in Brazil and in North Africa."},{"index":15,"size":328,"text":"The negligible impact of African agricultural research on forage productivity in grazing systems has some precedent. Two studies of China mention no significant impact from research on sown pastures or planted forages. Perkins' (2013) study of Chinese farming over six centuries makes no reference to pastures or planted forages in any subsystem -not in grazing or mixed systems, in temperate or tropical areas, or with reference to milk, meat or draught power. Even discussions of cereal crop residues and multidimensional crops, such as sugar beets and soybeans, do not mention their use by animals. A contemporary comparison of the lessons of Chinese agriculture for Africa does not mention livestock research -breeding, management, pastures, sown forages, genetics or nutrition -as a factor in the more rapid agricultural growth of China compared with sub-Saharan Africa (Li et al., 2012).Some of the failure to intensify pasture production stems from the failure to understand that conversion of grazing land to cropland is not necessarily a net loss of feed. The net change is positive at lower levels of cultivation intensity and may not become negative until quite high levels. This apparent paradox is due to the fact that crop residues from newly planted areas are valuable feeds and can be stored more easily into the dry season(s) than pastures. Cropping would be accompanied, to some degree, by destruction of invasive browse species during the fallow parts of the cropping cycle, thus improving the fodder value of the fallow. In addition, such pasture substitutes as processed feeds or cut fodder are sold near expanding towns (e.g. for the groundnut basin of central Senegal (Mortimore, 2000); the Kano close-settled zone and other parts of northern Nigeria; and the areas around Niamey, Ouagadougou, Bamako and Addis Ababa). Moreover, secondary feed effects occur, as noted by Mortimore and Turner (2005), including planting trees on cropland to replace woodlands lost to cropland expansion with variable effects on feed balances depending on climate, crop density and tree species."},{"index":16,"size":127,"text":"animal health. Animal health was not the main point of pastoral characterization research, except for field studies of epizootics, because so much was already known and because ILRAD's focus was on parasitology and immunology. The main diseases of grazing ruminants were: (i) trypanosomiasis, caused by protozoan parasites transmitted by the blood-feeding tsetse flies of the genus Glossina, which tend to be more abundant in the humid and subhumid zones; (ii) Theileriosis, most often known as ECF, another protozoan parasite transmitted by a tick of the genus Rhipicephalus (Boophilus), which is limited to 12 East African countries; and (iii) rinderpest. The principal contributions of animal health work, in terms of systems research, were those on trypanotolerance and ECF 58 as shown in Chapters 2-6 and 10 (this volume)."},{"index":17,"size":152,"text":"Animal health was the domain in which ILCA systems research had the weakest impact precisely because the principal animal health work was laboratory based or was already the object of international control campaigns. For Borana, Coppock (1994, p. 140) reported that 'animal health was never a significant focus of research'. The principal afflictions of cattle were pasteurellosis and external parasites, affecting mainly young stock and having a strong interaction with animal nutrition. There was a low incidence of trypanosomiasis in cattle (Coppock, 1994, Table  E2) because herders made careful choices of grazing territories. Despite detailed animal health status measures taken in the Mali Inner Delta study on brucellosis, internal parasites, contagious bovine pleuropneumonia and trypanosomiasis (Wagenaar et al., 1986), this work did not lead to animal health improvements or to associated productivity gains. The same lack of productivity effect of the field health research was seen in the Niono and Kaduna studies."}]},{"head":"Modelling pastoralism","index":44,"paragraphs":[{"index":1,"size":119,"text":"Numerical models, based on coefficients derived from experiments and surveys, are useful for simulating the impact of technology and policy on livestock and crop production. Many models were developed to simulate livestock demography and technology in the expectation that new methods could be adopted by herders or that such methods could be promoted by appropriately targeted policies. Despite the effort invested in models, they have generally failed, in African situations, to promote new technologies or policies. The reasons have been: (i) complex data requirements; (i) failure to measure important parameters in many studies; (iii) sampling variation in model parameters estimated from surveys and experiments; and (iv) inability to use natural parameters, especially rainfall, as control variables for policy experiments."},{"index":2,"size":110,"text":"Significant research in rangeland modelling -soils, water, pastures, live-weight response to feed intake, management, species and breed effects -was made outside ILCA's mandate area before the mid-1970s, as presented at the 1975 rangelands symposium (ILCA, 1975)  59 . Much of this work was not directly applicable to sub-Saharan African conditions because model processes were derived from range situations -climate, breed and animal disease -in the USA, Australia or Latin America 60 . Hence, field modelling work began at ILCA in the 1970s from a narrow database and confronted difficult problems of using data and model structures from situations in the USA, Australia and Latin America that were rarely closely comparable."},{"index":3,"size":68,"text":"primary productivity. Animal production in extensive grazing systems depended more directly on grazing resources than it did on mixed-farming systems because the latter can more economically exploit cut fodder and purchased concentrates. Despite their importance, studies of primary productivity in grazing systems were rare until comparatively recently because of the long-term data requirements necessary for integrated soil-water-plant-animal models and the resulting high cost of such studies 61 ."},{"index":4,"size":229,"text":"What did the models show about the potential primary productivity gains as functions of climate, soils and technology? 62 The literature review by Le Houérou and Hoste (1977) found close relationships between rainfall, as a good proxy for water availability, and pasture production in Mediterranean areas (sites ranging from 70 to 900 mm of rainfall) and in the Sudano-Sahelian zone of sub-Saharan Africa (200-600 mm). The most comprehensive early work on potential primary production around the time of the founding of ILCA and ILRAD was that at the Niono ranch (Penning de Vries and Djiteye, 1991, is a summary of many papers). A model of primary productivity, derived with annual rainfall levels below 600 mm in a hot Sahelian climate, was reported by Penning de Vries and Heemst (1975, pp. 323-328). They concluded that a basic plant production function of water availability was applicable to annual grasses on sandy soils under arid and semi-arid conditions in Mali. Later work (van Keulen et al., 1981) on semi-arid pastures concluded that the model's predictions were not 'very sensitive' to most weather parameters, except rainfall, and could be 'applied with reasonable confidence' to sparse data situations in hot climates. It was noted that this primary model was only applicable to annual grasses and not to legumes or to woody species, even in the same climate and soil conditions (ILCA, 1975, pp. 336-337)."},{"index":5,"size":107,"text":"The principal findings of the Niono ranch study were that day length limits biomass production, and this can only be managed within narrow boundaries set by other constraints, such as water and soil nutrients. Water was identified as the second limiting factor for biomass production; the growth response to water is linear until it reaches a boundary set by soil fertility. Phosphorus was the next limiting factor, followed by nitrogen once the ratio of phosphorus:nitrogen reached a given level as determined by soil type. Subsequent station and field research tested how two major constraints -water and soil fertility -could be managed biologically while producing adequate financial returns."},{"index":6,"size":65,"text":"These soil-water-plant models established the determinants of plant production in the grazing areas. However, it has been impossible to apply their findings to technology and policy in sub-Saharan Africa given that rainfall is so variable and cannot be managed in arid climate without irrigation. The inability to control the main exogenous variable -water -becomes more binding in arid sites where most of the animals graze."},{"index":7,"size":453,"text":"herd demography and stocking rate. Models having stocking rate as a control variable were typically only possible under ranching conditions. For this reason, several early simulations (Cartwright et al., 1978(Cartwright et al., , 1982;;Konandreas and Anderson, 1982;Konandreas et al., 1983) adapted a US model for ranch and rangeland conditions in arid Botswana. These efforts specified a detailed model of cattle production in which pasture productivity was exogenous and in which the live-weight response to feed intake was the main function to be estimated. This model's major result was the projected relationship between feed resources and a 'permissible' stocking rate for different classes of herders over a 30-year period, which included a sample range of very good to very bad years. It evaluated four production alternatives -calf weaning age under ranch conditions (Cartwright et al., 1978, pp. 46-49) and three innovations on traditional rangeland cattle posts: a lower weaning age, a shorter breeding period and reserve pastures for calves (Cartwright et al., 1978, pp. 49-55). None of the rangeland innovations appeared profitable in the simulations and, applied jointly, the three achieved only 'modest' productivity gains (Cartwright et al., 1978, p. 55). Konandreas and Anderson (1982) defined policy experiments for supplementary feeding, calf offtake, weaning, sales/purchases and a drought response consisting of sales and supplementation for stock remaining in the herd. Konandreas and Anderson (1982) did not report policy simulations as functions of price, weather or exogenous technology. calf nutrition. Lambourne and Butterworth (1983;in IDRC/ILCA, 1983) stressed the importance of calf nutrition for herd growth and argued that the conflict between calves and people for milk incurred an opportunity cost in herd productivity. The initial comprehensive cattle model of ILCA (Cartwright et al., 1982, pp. 51, 58-65) examined this argument by simulating alternative milking strategies, with and without supplementary feeding, on traditional rangeland cattle production in Botswana. The economic return to milking, without supplementary feeding, was positive in the first 3 years and fell in the following two, final, model years. A mean IRR of -14.0% was estimated for the no-milking strategy without supplementation, indicating that traditional herders in Botswana had foregone significant cash income by not milking 63 . Wagenaar et al. (1986, p. 48, Table 38) subsequently measured calf live weight by milk offtake for traditional Fulani cattle in semi-arid central Mali. Their major result was an elasticity of calf live weight with respect to milk offtake between -0.27 and -0.38; a doubling of milk offtake would reduce calf live weight by about one-third. The Wagenaar et al. (1986) model was deterministic and did not generate a frequency distribution of productivity or economic benefits with respect to the milk offtake treatment; hence, it could not conclude anything about the riskiness of offtake strategies."},{"index":8,"size":153,"text":"The Borana, Maasailand and both Mali studies all identified calf survival and weight as a major constraint to productivity; accordingly, they proposed supplementary feeding of calves as a solution. The Mali studies reported nutritional stress causing poor reproductive performance in both cattle and small ruminants as the chief cause of productivity below potential (Wilson, 1986;Wagenaar et al., 1986). Wagenaar et al. (1986, p. 51) recommended calf supplementation but only as 'a practice worth exploring', although they concluded, given the prices of meat and milk and expected live weight as a function of milk offtake, that the herders' observed average milking period of 6-13 months was roughly optimal given local conditions 64 . Solomon Bekure  et al. (1991, p. 145) found calf supplementation with purchased feed to be profitable but only if cow and calf mortalities were reduced 'drastically' and only if workers were available to meet the additional labour demands resulting from supplementation."},{"index":9,"size":75,"text":"The work of Coppock (1994) and Solomon Bekure et al. (1991) generated long-term predictions for the viability of the Borana and Maasailand systems. Coppock (1994) predicted that the Borana would be 'increasing dependent on grain purchases' 65 , more cattle marketing and 'less sustainability in terms of per capita milk production and asset accumulation'; an additional implication of these trends would be greater cultivation of crops by the pastoralists and greater recourse to wage labour."},{"index":10,"size":190,"text":"vector control. Biting insects and ticks and other vectors of disease can be controlled by spraying and trapping, by killing wildlife that are reservoirs of the pathogens and by destroying vector habitats (Jahnke, 1982, pp. 142-149). Vector modelling to make control more effective has taken many forms. Estimates of the coverage of spraying campaigns are quite old 66 ; recent detailed estimates are not available, but it is likely that vector control through spraying and trapping has been effective, although sporadic and on relatively small areas (Grace, 2003, p. 2, estimated that vector control was effective on 2% of the tsetse-infested area of sub-Saharan Africa). The killing of wildlife reservoirs has probably adversely affected the vector and hence limited the disease. Destruction of tsetse habitats caused by the expansion of human population and crop cultivation has been successful but is not a manageable control policy. A long-term study by Reid et al. (2000b) for 1960-2040 projected that many of the 23 African tsetse species would 'begin to disappear' but that 'for the foreseeable future' a large area in sub-Saharan Africa would remain 'infested by tsetse and under threat of trypanosomiasis'."},{"index":11,"size":145,"text":"a new grazing model. The major change in grazing systems theory, associated with the field observations derived from farming systems research, was the shift from equilibrium to non-equilibrium models (Ellis and Swift, 1988;Behnke et al., 1993;Vetter, 2004). In the former, systems are density dependent on such biotic factors as grazing intensity and population density; these factors determined the average system performance, its variability and its resilience. Non-equilibrium systems are density independent, and abiotic factors, such as the frequency and duration of drought, are determinant. Non-equilibrium rangeland systems would not revert to a steady state, or to a long-term trend, as a function of grazing pressure or population density 67 . The shift in the basic grazing model from 'equilibrium' to 'non-equilibrium' affected the proposed technical and managerial changes in the domains of feed, species choice, the transition to settled cropping, water use and animal health."},{"index":12,"size":136,"text":"Observations over long periods are too infrequent in semi-arid conditions to test equilibrium and non-equilibrium models. One comparison included a study over 27 years in semi-arid Senegal, which found that dry periods, with higher rainfall variability, tended to be 'non-equilibrium', while wetter periods, with lower rainfall variability, tended to be 'equilibrium' and to show density dependence (e.g. overgrazing) (Miehe et al., 2010). A subsequent paper, using data from the same sites, concluded that vegetation dynamics are 'driven by precipitation not by grazing' and hence are explicable by the non-equilibrium model (Retzer, 2006). Coppock's (1994, p. 193) study of southern Ethiopia argued that density dependence is stronger in higher rainfall areas and weaker in drought-prone areas, and that long-term loss of grazing reserves has tended to increase density dependence, but these arguments have not been rigorously tested."},{"index":13,"size":244,"text":"The policy implications of a non-equilibrium model would be that, because destocking occurs with variations in rainfall, especially in droughts longer than 1 year, efforts to impose generalized destocking independent of weather to preserve range quality fail because they impose the cost of underutilization of available grazing on herders. Blench (1994Blench ( , 2001) ) reviewed global rangelands and argued that destocking has failed and, moreover, that restocking after drought or epizootics has been ineffective because it cannot mobilize the feed and water resources needed to revert to the previous stocking rates. Public restocking programmes have sometimes failed because of inequitable distribution of new stock because of corruption. Homewood (2008 p. 70) asserts that an 'expanded view [one that incorporates both equilibrium and disequilibria reasoning into specific situations of rangeland ecology] has fundamental policy implications for sub-Saharan Africa'. True as this generalization may be, it has been very difficult to find examples of technology or policy research to which economic or environmental benefits may be attributed using either the equilibrium or non-equilibirum model. Attempts to apply non-equilibrium grazing practices -such as rotational grazing from more temperate climates (Vetter, 2004, p. 11) -have often failed in Africa and in the tropics more generally because rainfall is low and variable; low and variable rainfall makes rotational grazing riskier, because of its effects on the level, spatial and temporal distribution of pasture productivity, and enforces mobility across a much larger area than is possible for most pastoralists."}]},{"head":"What did the mixed-systems studies find?","index":45,"paragraphs":[{"index":1,"size":29,"text":"The second focus of international livestock research was on smallholder mixed farming where crops were the dominant source of income and employment and where livestock had a secondary role."}]},{"head":"Characteristics of mixed systems in sub-Saharan Africa","index":46,"paragraphs":[{"index":1,"size":126,"text":"Seré and Steinfeld (1996, p. 11) defined a mixed system as one 'in which more than 10% of the dry matter fed to animals comes from crop by-products or more than 10% of the total value of production comes from non-livestock farming activities. In these systems, more than 90% of the value of non-livestock farm produce comes from rainfed land use. ' Ruthenberg's (1980, pp. 1-18) classification for cropping systems applied the keys of: (i) population density; (i) degree of cultivation in the system (arable land as a share of arable plus pasture); (iii) type and duration of fallow; (iv) share of irrigated cropping in total of irrigated plus rain-fed; and (v) agro-climate, typically measured by LGP and cross-tabulated against temperature, altitude and bimodal/monomodal rainfall regimes."},{"index":2,"size":39,"text":"A more recent classification of mixed cropping systems describes them by matching details of crops and livestock enterprises against the resource base (Dixon et al., 2001). This classification 68 compared with Seré and Steinfeld (1996) has the following groups:"},{"index":3,"size":21,"text":"• Irrigated farming ('MIT' in the model of Seré   and Steinfeld, 1996), included by Ruthenberg (1980, pp. 178-250) under 'arable irrigation'."},{"index":4,"size":13,"text":"• Wetland rice ('MRT'), included by Ruthenberg (1980, pp. 178-250) under 'arable irrigation'."},{"index":5,"size":17,"text":"• Rain-fed farming in high-potential areas, such as the West African savannahs and much of Brazil ('MRT')."},{"index":6,"size":13,"text":"• Mixed crop-livestock rain-fed farming in highland areas, such as central Ethiopia ('MRH')."},{"index":7,"size":23,"text":"• Mixed crop-livestock rain-fed farming in semi-arid areas, such as the West African Sahel north of the 500 mm annual rainfall isohyet ('MRA')."},{"index":8,"size":48,"text":"• Dualistic -a mix of large commercial and smallholder in the same general environment; these were restricted to former areas of European colonization in Africa but were common in much of Central America and Latin America. Dixon et al. (2001, p. 13) defined three specific mixed smallholder systems:"},{"index":9,"size":50,"text":"• Mixed crop-livestock rain-fed farming in the highlands (corresponding to MRT), such as central Ethiopia and parts of Kenya. Such farms were typically less than 3 ha in operated area, with animal traction featuring in Ethiopia although not in Kenya, and feature mixes of annual crops, permanent crops and ruminants."},{"index":10,"size":47,"text":"• Mixed crop-livestock rain-fed farming in low potential areas (corresponding to MRA), overlapping with grazing (LGA), such as the West African Sahel north of the 600 mm annual rainfall isohyet; farms would be much larger in such areas, with low cereal and grain legume yields per hectare."},{"index":11,"size":62,"text":"• Rain-fed farming in high-potential areas (MRT for rain-fed and MIT for irrigated), such as the West African savannahs and in much of Brazil and Colombia. Farms in Latin America were typically much larger than those found in the Sahel or in the highlands of East Africa. Mixed farming in these three systems differed from pastoralism by the presence of the following:"},{"index":12,"size":10,"text":"• Majority shares of cropping in total income and employment."},{"index":13,"size":18,"text":"• Varying shares of livestock in income and employment, with lower shares in food consumption compared with pastoralism."},{"index":14,"size":22,"text":"• More unequal distribution of cattle between households, although this was compensated to some degree by more equal distribution of small ruminants."},{"index":15,"size":38,"text":"• Less animal mobility or none, which was limited to seasonal transhumance or to daily movements near permanent villages. In the Latin American systems, with private and fenced grazing lands, there would be no seasonal or annual mobility."},{"index":16,"size":28,"text":"• New livestock functions, notably nutrient recycling and the use of draught animals for cultivation. These functions would be much less important or even absent in Latin America."},{"index":17,"size":39,"text":"• Important shares in income and employment of annual and semi-perennial cash crops -cotton, groundnut, rice, roots and tubers -that did not exist in rangelands. In the Latin American and Middle Eastern production types, animals are the cash commodity."},{"index":18,"size":36,"text":"• Important shares in income and employment of perennial cash crops -coffee, tea, cocoa, rubber and oil palm -that are found in mixed systems with poultry, swine and small ruminants, although much less often with cattle."},{"index":19,"size":21,"text":"• Commercial dairying with zero grazing, planted forages, purchased feeds, and cultivation of annual and perennial crops on the same farm."},{"index":20,"size":25,"text":"• Commercial on-farm fattening of ruminants, swine or poultry, in zero-grazing or transhumant systems, using purchased feeds and crop residues from annual and perennial crops."},{"index":21,"size":311,"text":"Although it is evident that people and animals are more concentrated in the mixed systems, estimates of the distributions of tropical smallholders, and their livestock, are difficult because of a lack of accurate data linking production and input use to farm size and structure. It is even more difficult to derive productivity estimates for individual farming systems, for example by the share of livestock in income or assets. levels and rates of change in farm sizes. Levels and rates of change in farm sizes in the 1960s-1980s are impossible to estimate by agroclimate in sub-Saharan Africa because of a lack of comprehensive data. However, published metaanalyses make the trends clear for farm size, as measured in area per worker, and in type of farm. Masters et al. (2013) found that ' Africa and Asia experienced a gradual decline in total land available per rural worker' over the period 1960-1970 to 1999-2000. The same decline can be seen over the period 1990-2015 (Masters et al., 2013, p. 2). Associated with this decline would have been some consolidation of subsistence farms into commercial farms in densely populated areas. An earlier paper of Hazell et al. (2010, p. 11) found a decline in median farm size in 16 countries (all regions) over periods ranging from 1970from to 1990from to 1998from to 2001from . Hazell (2013) ) compared farm size trends in Asia and Africa from 1970 to 2011 and further projected trends to 2030 and to 2050. He noted that rural population growth would fall (Hazell, 2013, p. 1) substantially from 2011 to 2030, compared with the earlier period, and would fall again from 2030 to 2050. While the expected fall in rural population growth reduces pressure on farm sizes, it is associated with other problems of farm consolidation, the transition from subsistence to commercial farming and the potential for extreme inequality among smallholders."},{"index":22,"size":317,"text":"A global summary (Lowder et al., 2016) found the usual problem with data coverage in tropical countries. It confirmed other findings about the fall in mean and median holdings in sub-Saharan Africa, while stressing the fact that land abundance in sub-Saharan Africa was limited to a few countries. The great majority of sub-Saharan African smallholdings would be less than 5 ha (Lowder et al., 2016, Fig. 5) and such holdings would be 50-70% of agricultural area. improved plant cultivars. Improved cultivars of cereals have become nearly universal on farms of all types in India, Indonesia, China and Brazil. Sub-Saharan Africa has lagged East and South Asia in adopting modern cultivars but has recently begun to converge. Walker and Alwang (2015, pp. 265-293) studied diffusion of modern plant varieties in sub-Saharan Africa from the 1960s to 2011. Technical progress in crop management, as indicated by uptake of modern varieties in sub-Saharan Africa, was significant during that period. At the beginning of CGIAR research on mixed livestock systems (1976)(1977)(1978)(1979)(1980), the share of cropped areas in seven major cereals -sorghum, maize, pearl millet, rice, teff, wheat and barley -was low, ranging from roughly zero in teff and barley, to 6% in sorghum, to 37% in wheat, with an areaweighted average of 3.0%. The area sown to modern varieties grew at an average annual rate of 1.4% (Walker and Alwang 2015, p. 389) from 1998 to 2011. The average for the seven cereals had grown to 31% in 2006-2010, with values of 52% in maize, 62% in wheat and 36% in rice; most of the area growth was in maize (52% of the total growth from 1976-1980 to 2008-2010) and in sorghum (20% of the total growth from 1976-1980 to 2008-2010). There were important increases in the areas planted to improved cultivars of other crops such as cassava, yam, soybean, cowpea, common beans and groundnut (Walker and Alwang, 2015, p. 344)."},{"index":23,"size":118,"text":"There are no comprehensive data on modern variety adoption by agroclimate in sub-Saharan Africa from Walker and Alwang (2015). The national averages (Walker and Alwang, 2015, p. 346) show rapid growth of area under modern varieties in countries and crops -maize and sorghum in Nigeria; maize, sorghum and teff in Ethiopia; and pearl millet in Mali, Senegal and Niger -where mixed crop-livestock are the principal farm types in humid (Köppen climate A) and semi-arid (Köppen climate B). These patterns imply that important benefits accrued in mixed livestock systems from additional crop residue output produced by modern varieties, although we cannot quantify these benefits or attribute them to (crop) research fields other than plant breeding and pathology 69 ."},{"index":24,"size":165,"text":"Data on the yields of modern cultivars in sub-Saharan Africa are less comprehensive than data on area planted. Walker and Alwang (2015, p. 354) estimated that a 1% increase in the share of area under improved cultivars was associated with an increase in TFP of about 0.47 in a sample of 30 countries covering most of sub-Saharan Africa. Older data reported by Walker and Alwang (2015, p. 355)  The delay in developing and extending productive seed/fertilizer packages for crop farming in sub-Saharan Africa, compared with South Asia and East Asia, had a negative indirect effect on livestock production. Seed and fertilizer packages were not always highly profitable under experimental conditions, especially in more arid sites, indirectly reducing potential feed production for ruminants. New seeds and fertilizer were less profitable under farmers' conditions in sub-Saharan Africa; strong evidence for this fact is the 20-40-year lag between the rapid growth of high-yielding variety/fertilizer packages in South Asia and the slower growth of these packages in sub-Saharan Africa."}]},{"head":"Crop-livestock interactions in mixed systems","index":47,"paragraphs":[{"index":1,"size":69,"text":"Most of the work on mixed systems was done in the semi-arid tropics of central India, the Sahel of West Africa, the subhumid savannahs of sub-Saharan Africa and the highlands of East Africa, mainly in Ethiopia and Kenya. Examples of such studies in the IARCs were chiefly, but not only, from ILRI, ICRISAT and the Centro Internacional de Mejoramiento de Maíz y Trigo (CIM-MYT) as shown in Table 15.4."},{"index":2,"size":197,"text":"Grazing and mixed systems present many complex interactions. Bourn and Wint (1994, pp. 4-5) reviewed 20 aerial and ground surveys of livestock and land use carried out between 1980 and 1991 in Mali, Niger, Sudan, Chad and Nigeria, covering a broad range of grazing and mixed smallholder systems. The surveys found many common elements: (i) positive relationships between plant and animal biomass; (ii) positive relationships between livestock distribution and land-use intensity, rural settlement density and mean annual rainfall (Bourn and Wint 1994, p. 6); (iii) weak seasonal effects on animal biomass in arid and humid climates, implying less seasonal animal mobility in both; (iv) stronger seasonal effects on animal biomass in the '750-1250 mm rainfall band' (Bourn and Wint 1994, p. 9), implying more animal mobility; (v) a declining threat from tsetse and trypanosomiasis because of 'agricultural expansion, deforestation and the removal of wildlife' (Bourn and Wint 1994, p. 15); and (vi) in Nigeria, probable undercounts of 25% of domesticated livestock populations compared with official figures, indicating that the hypothesized conflict between crop and livestock production had been overstated.The principal research goals on mixed systems with comparatively weak integration between crop and animal production were as follows:"},{"index":3,"size":21,"text":"• To define constraints to higher productivity in the power, soil nutrient, feed production and animal management components of mixed farms."},{"index":4,"size":21,"text":"• To introduce or improve animal draught power with a research focus on higher work output through health and feed interventions."},{"index":5,"size":10,"text":"• To improve nutrient cycling by using manure on crops."},{"index":6,"size":11,"text":"• To feed crop residues and higher-quality planted forages to ruminants."},{"index":7,"size":10,"text":"• To introduce dairying and small-ruminant fattening into mixed farms."},{"index":8,"size":29,"text":"• Ultimately to raise livestock and crop productivity jointly by exploiting positive interactions between the two components. Some examples of feasible technical changes in mixed enterprises included the following:"},{"index":9,"size":4,"text":"• Applying seed/fertilizer packages."},{"index":10,"size":17,"text":"• Shifting the livestock enterprise mix to dairying and intensive stock raising (poultry, pigs, small ruminants) components."},{"index":11,"size":65,"text":"• Improving feed production by planting forages or by treating crop residues in densely populated areas (semi-arid tropics of India, Ethiopia, northern Nigeria, and cotton-producing areas of West Africa), whether for animal traction, dairying or on-farm fattening. • Diversifying crop production and processing by adding annual cash crops, such as oilseeds, cotton and grain legumes, which would have a secondary effect of providing more feed."},{"index":12,"size":20,"text":"• Cycling nutrients 70 in mixed grain-livestock farms by feeding crop residues to stock and by restoring manure to fields."},{"index":13,"size":49,"text":"Predictions about the evolution of mixed systems strongly influenced the initial international research in the 1970s. New theories of farming systems (e.g. Boserup, 1965;Ruthenberg, 1980;Binswanger and Rosenzweig, 1986) provided a deeper understanding of how tropical agriculture evolved under the influence of population density, market access, information costs and incentives."},{"index":14,"size":62,"text":"The first prediction was that nomads would settle to become crop farmers. This prediction depended on the view that sedentarization was desirable because it was more productive and would allow better provision of social and infrastructure services to the former nomads. Some mixed-systems research therefore sought to identify the policies and technologies that would reduce the transition costs of the inevitable sedentarization."},{"index":15,"size":97,"text":"Related to the settlement prediction was a later view from the characterization literature. Thornton et al. (2002, maps 16a and 16d) contended that there would be major long-term changes in growing periods and in locations of dense livestock and human populations related to climate. The combined effects of these changes would be to shift the West African rangeland units into mixed systems and to eliminate mixed highland systems in East and southern Africa. Jones and Thornton (2009) noted that climate change might push marginal, low-productivity farmers in West Africa into livestock activities, thereby increasing pressure on rangelands."},{"index":16,"size":66,"text":"A second prediction was that closer integration of crops and livestock would be more productive than separate enterprises. Integration would allow more efficient use of resources -animal power, manure and crop residues -that were underused when crops and livestock were managed separately. Many development projects, and their associated research components, acted on this prediction and attempted to accelerate this integration (McIntire et al., 1992, pp. 4-5)."},{"index":17,"size":47,"text":"A third prediction was that smallholdings would gradually consolidate into larger commercial enterprises because such units could exploit economies of scale in technologies and management. This prediction was a stimulus to mechanization research, which varied among regions, being stronger in Asia and Latin America than in Africa."},{"index":18,"size":40,"text":"This evolutionary model influenced the three principal themes of research on mixed systems: (i) mechanization; (ii) soil fertility management; and (iii) feeding systems. A fourth theme -improved animal health -was viewed as necessary to achieving results in the other components."},{"index":19,"size":113,"text":"mechanization. Animal traction had been practically universal among smallholders in the semi-arid tropics of India and in highland Ethiopia for centuries. It had begun to grow rapidly across sub-Saharan Africa in the 1970s. Despite the infrequency of use of animal traction in most of tropical Africa, it was believed that land-abundant areas within sub-Saharan Africa had the potential for mechanization with animals. Mechanization would allow an expansion of cultivated areas and a concomitant increase in crop yields. The efforts to realize these potential gains made farm mechanization with animals an important part of research at ILCA, and to a much lesser extent of ICRISAT, from the 1970s and to the mid-1990s 71 ."},{"index":20,"size":6,"text":"Mechanization research started from three assumptions:"},{"index":21,"size":80,"text":"1. Animal traction was the core component of mixed farming and could create productive interactions among other components that would not occur if those components were separate; Jahnke's (1982, pp. 134-140) summary of livestock development problems in tropical Africa is an example of this view. 2. Animal draught power could increase the cultivated area per worker, could stimulate diversification into more profitable crops and, by improving the quality and timeliness of farm tasks, could raise yields per unit of land."}]},{"head":"3.","index":48,"paragraphs":[{"index":1,"size":21,"text":"Better animal nutrition could allow more power from livestock and hence extend the hypothesized benefits of area, cropping pattern and yield."},{"index":2,"size":202,"text":"The second assumption -that animal traction would produce substantial area, yield and cropping-pattern benefits -led to many studies of constraints to animal power. These studies often failed to find adoption of animal traction, even where livestock disease was manageable and where adequate feed was available. Pingali et al. (1987) explained the mixed adoption of animal traction across sub-Saharan Africa in terms of labour use and fallow type; they concluded that the driving force for mechanization with animals was not output benefits but labour savings in areas of annual cultivation where the transition from forest or bush fallow to grass fallow had taken place and where heavy soils created additional demand for power over what hand hoes could provide. Pingali et al. (1987) complemented a literature review with original farm interviews about the gains from animal traction. They found that area benefits were positive (Pingali et al., 1987, pp. 99-101) and averaged about 25% per person; that yield gains (animal-cultivated fields over hand-cultivated fields for the same crop) were often zero because the quality of tillage with animal traction was not better than that with hand hoes; and that impacts on cropping-pattern diversity were important only where cotton and groundnut had been introduced."},{"index":3,"size":148,"text":"Animal and machine power was introduced into farming systems as a function of population density, given the need to cultivate the same plots repeatedly to suppress weeds, and in response to better market access from proximity to cities, higher population density areas and the introduction of cash row crops, notably cotton and groundnut. The presence of trypanosomiasis, long held to block draught power by restricting cattle production, was shown to be a secondary constraint in the humid and subhumid areas where other climate and population factors weakened the demand for animal power. The core conclusion -the introduction of animal traction into farming systems was a long-term response to growing population and cultivation density, rising wages and the introduction of cash crops (Pingali et al., 1987;McIntire et al., 1992) -greatly weakened the research emphasis on animal draught power in mixed rain-fed systems where animal power was absent or nascent."},{"index":4,"size":309,"text":"The second assumption was that poor animal nutrition constrained draught power. This assumption was tested in many experiments examining animal breed and type (dairy cows or oxen) and feed type (quantity and quality of additions to basal diet of crop residues) at sites in Mali, Ethiopia, Niger and semi-arid India (Renard, 1997). The ILCA research in Mali summarized their findings as: '…dry-season supplementation and weight gain would not improve work output and would be unlikely to increase the amount of land cropped or crop production' (ILCA, 1994, p. 134). The lead scientist for the Malian trials under experimental conditions concluded that work output during the brief 2-week ploughing season '…seem [s] to be unrelated to an animal's body energy reserves at the start of work' ( Bartholomew 1988: p. 59). An experiment using dairy cows for traction in Ethiopia found that supplementation allowed additional work while (almost) maintaining the milk output and reproductive performance of the cows (ILCA, 1994, pp. 134-136). Further work (ILRI, 1998, pp. 178-183) in Ethiopia compared working cross-bred dairy cows with non-working dairy cows and traction with local oxen alone, plus improved management of feed and animal housing. The working cross-bred treatment did not produce higher incomes compared with the non-working cross-bred treatment or the local oxen treatment. An ILRI experiment on a semiarid experiment station with sandy soils in Niger measured energy expenditure of working bulls and oxen (Fall et al., 1997), feed intake and the effects of body condition on work; two important conclusions were that roughage intake did not increase during work and that weight losses during work did not cause reductions in power output (Fall et al., 1997, Chapter 6)  72 . Station and field research on animal power in sub-Saharan Africa did not generally find that nutrition was a serious constraint to adoption or to optimal use of animal power."},{"index":5,"size":130,"text":"animal power and mixed farming in sub-saharan africa. The view that animal traction had a leading role in developing mixed systems in sub-Saharan Africa was examined by McIntire et al. (1992, pp. 47-72), following the study by Pingali et al. (1987). McIntire et al. (1992) found that components of mixed farming -feeding crop residues to stabled or mobile animals, manuring crops, and investing in dairying and onfarm fattening -existed in many sites without animal traction. They concluded that using animals for power was not a necessary condition for mixed farming. An example was found in a field study of mixed smallholders in south-eastern Burkina Faso, where Delgado (1980) found that sample farmers practised some elements of mixed farming (manuring, crop residue grazing, fattening animals on farm) but not animal traction."},{"index":6,"size":96,"text":"The mechanization research at ILRI with the greatest development impact was that leading to the broad-bed and furrow maker (BBM), as recounted by Rutherford (2001Rutherford ( , 2008)). Rutherford's 2008 analysis found a rate of return to ILCA/ILRI research and development of the BBM to be in the order of 0.1%. This was much higher than the negative return found in the 2001 study, the difference being attributed to the greater availability of credit, farmers' adaptation of the BBM tool to their circumstances, and the resulting area and yield effects from greater experience with the tool."},{"index":7,"size":156,"text":"animal power and mixed farming in the semi-arid tropics of india. ICRISAT began work on mixed irrigated and rain-fed cropping systems in the semi-arid tropics of central India (Jodha et al., 1977). These systems were in a hot and usually dry climate, with a single monsoon season, and consisted of small family farms growing ICRISAT mandate crops (sorghum, pearl millet, groundnut, chickpea and pigeon pea) and others (irrigated paddy, cotton, castor and vegetables), usually with bullock power. The chief livestock research component was on the role of animal power for management of Vertisols, cultivation, transport and processing with some work on crop residues. Other national and international research in South Asia concentrated on food crops, on technical changes in irrigation, and on mechanization compatible with smallholdings; livestock research concentrated on animal traction and smallholder dairying (e.g. Singh, 1990, pp. 204-232, on rural poverty in South Asia; Walker and Ryan (1990, on the semi-arid tropics of India)."},{"index":8,"size":133,"text":"The centrepiece of ICRISAT's limited livestock work in India were the village-level studies (Walker and Ryan, 1990). These studies allowed a better understanding of the roles of livestock in the semi-arid tropics of South Asia and led to a reorientation of mixed-system smallholder research. This reorientation conceded that public research on mechanization either added too little to farmers' knowledge of crop management -the case of most animal traction researchor could be done better and faster through adaptations in the private sector -the case of most tractor research. Herbicides were proposed as an alternative to animal or tractor mechanization (Le Moigne, 1979, pp. 219-220), but herbicides have only recently become competitive in countries where rural wages have risen rapidly; if herbicides have replaced hand weeding in sub-Saharan Africa, it would be a recent development."},{"index":9,"size":179,"text":"In 40 years of national and international work on animal traction in West Africa, there have been no significant technical changes induced by research findings that are clearly distinguishable from the changes induced by rising population density, cheaper market access and external economies arising from lower implement production costs. The arguments of Pingali et al. (1987) and McIntire et al. (1992) derived from the evolutionary models of Boserup (1965) and Ruthenberg (1980), and the mixed findings of station and field studies of animal power, effectively ended the era of projects seeking to introduce animal power as a necessary component of mixed farming. soil fertility management. Low soil fertility was long understood to be a major cause of the lagging productivity of tropical agriculture. Significant work was started in the 1960s and 1970s, in Africa, South Asia and Latin America, on soil fertility management with different emphases for the systems found in each region. Soil fertility work had substantial scientific impact on all continents, but its development impact varied greatly by region, farm type and source of water for cropping."},{"index":10,"size":77,"text":"The Africa soil fertility focus was on nutrient cycling in mixed systems because of the prediction that animal manures were 'slack resources' that could be used more efficiently where crops and animals were managed on the same farm 73 . The integration of crop and animal production would occur in part through nutrient cycling through linked mechanisms: the 'exchange' of plant residues from crops to animals, and the 'exchange' of manure and urine from animals to crops."},{"index":11,"size":38,"text":"Nutrient cycling was studied by Powell et al. (1996). The orientation of ILCA/ILRI work on nutrient cycling was on smallholder farms, where the cycle was soil to food crop to animal to soil and crop vegetation 74 ."}]},{"head":"Feed systems","index":49,"paragraphs":[{"index":1,"size":226,"text":"All characterization research found feed scarcity, in quality and in dry-season quantity, to constrain the livestock component whether animals were used for milk, power or meat production. Three interventions, with increasing levels of management costs, were proposed to unbind this constraint: (i) crop residue management; (ii) sown forages, including alley farming; and (iii) ley farming. crop residue management. Crop residue management -harvesting, storing, chopping, making hay, field grazing or the many other possibilitieswas found to be quite common but has been difficult to improve through research. ILCA's compilation of research on 'Plant Breeding and the Nutritive Value of Crop Residues' (Reed et al., 1988) cited no successful examples of improvement of crop residues in the tropics (nor did the book by Renard, 1997), through plant selection or breeding, or by chemical or urea treatment beyond chopping or other practical methods that farmers can already use (see Chapter 14, this volume). The Walker and Ryan (1990) book covering a decade of on-farm research in the semi-arid tropics of India noted that the unit value of fodder in the study villages was sometimes as high as that of grain, but found few examples of crop residue improvement or of planting of specialized fodder crops to replace crop residues in ruminant diets, even where the fodder value of crop residues was high for dairy production or for animal power."},{"index":2,"size":112,"text":"sown forages and ley farming. Sown forages and ley farming have usually failed in the African tropics. Chapter 13 (this volume) reviews attempts to introduce sown forages into mixed systems in Africa, with reference to the success of pasture grasses in Latin America, while Chapter 14 (this volume) covers multidimensional crops. Nordblom et al. (1992Nordblom et al. ( , 1994)), in analysing a Syrian site, identified land cost as the most direct reason for the failures of sown forages in the tropics. The available alternatives -weeds, field boundaries, crop residues and browse -are cheap and of high enough quality that their benefits surpass those of sown forages (Sumberg, 2002;McIntire and Debrah, 1987)."},{"index":3,"size":215,"text":"Where ley farming did emerge, it did so almost exclusively in 'unregulated' form (Ruthenberg, 1980). Where it does exist, it evolved as part of smallholder dairying in temperate highland areas where cross-bred European-African cattle could be raised, rather than from progressive adoption by sedentarized pastoralists. One limited success was achieved around Kaduna, in subhumid central Nigeria, where Stylosanthes hamata was introduced as 'fodder banks' in mixed systems (von Kaufmann et al., 1986). Despite the technical promise of fodder banks, their long-term area and yield effects have not been precisely estimated in Nigeria or elsewhere in West Africa. Other obstacles to ley farming were seed costs and farm size (Christiansen et al., 2000, p. 191, and the papers cited therein for the Mediterranean; Ruthenberg, 1980, for East Africa;Tiffen et al., 1994, pp. 164-166, for Kenya;Powell, 1986a for central Nigeria). Nordblom et al. (1994) reviewed field and station studies of a rotation of wheat and M. sativa with sheep grazing in north-west Syria. They found historical adoption of M. sativa as a pasture crop to be limited in West Asia and North Africa. Their modelling work showed M. sativa to be 'less profitable than traditional rotations' and its adoption to be sensitive to farm size, milk prices and soil nutrient carryover effects to the following wheat crop."},{"index":4,"size":109,"text":"fodder trees. Fodder trees started with the purpose of adding nitrogen to leached or otherwise infertile soils, thereby improving the soil nutrient stock while contributing nitrogen to cereal and tuber intercrops. The field model of this idea is 'alley farming' -typically planting rows of nitrogen-fixing trees between rows of food crops, such as maize or cassava. Several synthetic works established the basic science of the alley farming model (Kang et al., 1990;Sanchez, 1995;Giller, 2001), followed by many subsequent applications and extensions (Sumberg et al. 1987;Jabbar et al., 1996). Promising work was done on economic and environment benefits of silvopastoral systems in Latin America (Ibrahim et al., 2010, p. 189-196)."},{"index":5,"size":228,"text":"One extension of the alley farming model was to use nitrogenous browse as a supplementary feed for small ruminants, given the lack of CP available in grasses or crop residues in the humid tropics. This extension was successful in raising soil nitrogen stocks, in lifting intercrop yields and in providing higher CP content to livestock (Kang et al., 1990, pp. 340-345), while being profitable in an economic sense (Kang et al., 1990). Experiments with livestock were done at two ILCA sites in Nigeria, on alley farming using nitrogenous trees in the humid zone near Ibadan and using leguminous forages (without alleys) and manure recycling at the subhumid site near Kaduna. The latter had greater impact potential because it took place in an agroclimate with lower trypanosomiasis pressure and less heat and humidity, allowing higher ruminant density and productivity. Jabbar et al. (1994) showed that continuous alley farming would be more profitable than alley farming with short fallows or farming with fallow but without the alleys nitrogenfixing trees. Reynolds and Jabbar (1994) showed that the major benefit of supplementing free-roaming small ruminants in West Africa with the foliage of leguminous trees (Leucaena and Gliricidia spp.) was an increase in survival, and that the forage was best directed at latepregnant and lactating females. In East Africa, cross-bred dairy cows showed a significant response in milk production to supplementation with Leucaena."},{"index":6,"size":51,"text":"The International Centre for Research on Agroforestry (ICRAF) estimated that 'fodder shrubs have been widely adopted in East Africa, by an estimated 205,000 smallholder dairy farmers by 2005' (Place et al. 2009). There is no reliable estimate of direct tree yields or of their indirect effects on crop or livestock yields."},{"index":7,"size":129,"text":"animal fattening on farm. On-farm animal fattening was studied as a familiar technology with potential for improvement by using slack resources, especially crop residues and other cut fodder, with local animal breeds. Research themes included: (i) the productivity of supplementary feeds to crop residues and cut grasses; (ii) the introduction of new feeds, notably tree and herbaceous legumes; and (iii) the timing of fattening and sales. Examples of performance trials are given by Bartholomew (1988) and those cited in McIntire et al. (1992, pp. 135-164). The emergence of many small peri-urban fattening units, for ruminants and for poultry, has been rapid in this century but is a function of income growth, urbanization and falling intermediation costs, and does not appear to be related to technical packages developed by research."}]},{"head":"Conclusions","index":50,"paragraphs":[{"index":1,"size":40,"text":"What were the impacts of LSR, done mainly by ILRI and its predecessors, on the scientific and development problems of grazing and mixed livestock systems in Africa? How did new knowledge improve productivity or equity in the principal livestock systems?"}]},{"head":"Scientific impact","index":51,"paragraphs":[{"index":1,"size":91,"text":"LSR has had a strong scientific impact in pastoral and mixed systems. This scientific impact was notable in: (i) dismantling the 'mainstream view' of pastoralism and replacing it with a new model of pastoralism that is biologically and economically more credible than the mainstream view, which has been antiquated for 50 years; (ii) classifying and mapping systems; (iii) estimating productivity parameters to develop bioeconomic models, which have been used particularly in modelling climate change effects; and (iv) applying the network research model to the study of trypanotolerance at disparate field sites."},{"index":2,"size":142,"text":"a new view of pastoralism. The core of the 'mainstream view' was that African herders kept too many animals for cultural reasons and that such overstocking caused overgrazing. The systems studies and related work destroyed this view. A first attack on the 'mainstream view' was the demonstration that animal productivity per hectare in the grazing systems was not uniformly worse, and was sometimes better, than that of ranching systems. Such research further demonstrated that herd structures did not consist excessively of older males kept for sentimental reasons or held against risks beyond their optimal sale age 75 . The age/sex compositions of herds did not consist of 'excess' males, as the 'mainstream view' had contended, given the milk production objectives of herders and the low marginal costs of feed and labour needed to maintain male stock to a roughly optimal sale age."},{"index":3,"size":74,"text":"A second part of the 'mainstream view' was that pastoralists' management practices determined stocking rates. The 'mainstream view' held that maintenance of stocking rates above biologically sustainable values led to overgrazing and eventually to destruction of the range. The rise of 'non-equilibrium' models of pasture dynamics (led by Ellis and Swift, 1988) fortified the new view by stating how pasture dynamics depend on abiotic factors and not primarily on the decisions of pastoralists themselves."},{"index":4,"size":118,"text":"A third tenet of the 'mainstream view' was that crops and livestock would inevitably face a general land conflict as population density and cultivation density rose. This hypothetical conflict would reduce grazing areas and, with fixed pasture yields for given rainfall, would reduce livestock production. The attack on this tenet came from the practical fact that cereal and legume crops in the semi-arid and subhumid zones, which house the majority of African ruminants, produce crop residues that are valuable as feed. The growth of crop production, even in areas marginal for arable farming and even in the absence of feeding grain directly to animals, allowed more livestock production, not less, by the indirect channel of additional crop residues."},{"index":5,"size":131,"text":"It should be acknowledged here that part of the scientific impact achieved through the new view of pastoralism derives from the older work of anthropologists, notably the work of Dupire (1972), Monod (1975), Dyson-Hudson andDyson-Hudson (1969), Oxby (1975) and Stenning (1994), among others. That work, in East and in West Africa, laid an empirical foundation for the new view of pastoralism which recognized the rationality of extensive grazing (as clearly stated by Barbara Grandin in 1987 in an early ILCA paper on Maasailand). Sandford (1983a, pp. 11-18) best summarized the refutation of the 'mainstream view'. He argued that the 'mainstream view' 76 of grazing systems was wrong, anticipated the nonequilibrium critique and asserted that policies to reduce overgrazing were misguided because the evidence for overgrazing was weak. Sandford (1983a) contended that:"},{"index":6,"size":38,"text":"• definitions of overgrazing -changes in vegetation from an undisturbed state ('climax vegetation') to a disturbed one, or loss of productive capacity over time -were not rigorous enough to permit recommendations of alternative management practices to stop overgrazing;"},{"index":7,"size":22,"text":"• estimates of overgrazing were biased upwards because measures of animal numbers and of pasture quantity and quality were sparse or inaccurate;"},{"index":8,"size":39,"text":"• the 'mainstream view' of the dynamics of overgrazing -a movement from undisturbed vegetation subject to grazing to disturbed vegetation of lowered grazing capacity -was contradicted by observations of pastures whose productivity had been improved by heavy grazing; and"},{"index":9,"size":29,"text":"• there was a logical inconsistency in positing overgrazing where stocking rates were highest while simultaneously assuming that such high stocking rates were infeasible on pastures degraded from overgrazing."},{"index":10,"size":40,"text":"The critique of the 'mainstream view' led to a new development path for pastoralism. The new path allowed herd expansion while promoting public investments around grazing areas and regulating conflicts over land and water. This new development path proceeded by:"},{"index":11,"size":11,"text":"• improving animal health by controlling disease, mainly rinderpest and trypanosomiasis;"},{"index":12,"size":28,"text":"• defending grazing lands by laws and regulations; new laws and regulations reduced conflicts between herding and farming groups and in so doing limited the costs of violence;"},{"index":13,"size":13,"text":"• introducing water interventions to raise productivity per animal by reducing trekking times;"},{"index":14,"size":36,"text":"• introducing planted forages, as arable crops or as browse trees and shrubs, and supplementary feeds targeted to a subset of the herd or flock, such as sedentary milk herds or small ruminants for fattening; and"},{"index":15,"size":10,"text":"• accommodating mixed land use of crops, livestock and wildlife."},{"index":16,"size":54,"text":"The studies of Pratt and Gwynne (1977), Ruthenberg (1980), Sandford (1983a), Le Houérou (1989), Solomon Bekure et al. (1991), Coppock (1994), Blench (2001) and Lesorogol (2008) projected variants of that generic path. A detailed development pathway for pastoralism was drawn for the Borana system of southern Ethiopia (Coppock, 1994, pp. 272-295;on water, pp. 202-209)."},{"index":17,"size":138,"text":"mapping systems. The outstanding scientific achievement of the various LSR studies, carried out by ILRI and many partners from 1975 to the present, has been to develop a new view of grazing systems that can generate better policies for rangeland management. This view started from the observation that direct efforts to improve rangelands -pasture and breed improvement and grazing restrictions -had failed. It continued with the contention that rangelands could become more productive if herders benefitted from public investments in human and animal health, water, transport and markets, communications and social protection. The findings of rangeland scientists about the potential of arid areas did not differ greatly from those of earlier scientists in terms of raising biological potential, but they did differ in terms of defining an appropriate sequence of policy and public investment to achieve that potential."},{"index":18,"size":78,"text":"Landmark papers - Coppock (1994) for the Borana system in southern Ethiopia; Solomon Bekure et al. (1991), for Maasailand in Kenya; Wilson (1986) in north-central Mali; Wagenaar et al. (1986) in the Niger Delta of Mali; von Kaufmann et al. (1986) in subhumid central Nigeria; and Hiernaux et al. (2009) in semi-arid western Niger -not only contributed to knowledge about these areas but also established methods of studying them and induced changes in related research and development programmes."},{"index":19,"size":118,"text":"The major system papers over the past 20 years - Seré and Steinfeld (1996); Thornton et al. (2002); Otte and Chilonda (2002); Kruska et al. (2003); Robinson et al. (2011Robinson et al. ( , 2014)), plus the many papers on climate change as discussed in Chapter 16 (this volume) -modernized the methods for defining production systems and reduced the estimation errors for areas, animal numbers and feed balances. This research has clearly had an impact on scientific understanding (Class I) even if it is impossible to estimate the development benefits of this type of research. Work in this century on livestock and the global environment has had a similarly large Class I effect (see Chapter 16, this volume)."},{"index":20,"size":7,"text":"The specific scientific achievements were as follows:"},{"index":21,"size":10,"text":"• Estimating the economic and environmental weight of livestock systems."},{"index":22,"size":16,"text":"• Development of better survey and analytic tools for estimating system scale and potential across continents."},{"index":23,"size":14,"text":"• Creating methods to compare and reconcile national statistical estimates and remote sensing estimates."},{"index":24,"size":21,"text":"• Reducing errors in estimates of woodland, wildlife, crop and grazing areas, permitting closer analysis of actual and potential resource conflicts."},{"index":25,"size":34,"text":"• Better projections of animal numbers across climates and countries, which eventually produced better estimates of global environmental effects 77 and, very recently, refinement of Tier 2 estimates of greenhouse gas emissions from livestock."},{"index":26,"size":19,"text":"• Planning investments in irrigation, grazing, protected areas, disease management, vector behaviour and control, predator control and wildlife interactions."},{"index":27,"size":17,"text":"• Understanding seasonal effects on plant and animal biomass, wildlife and domestic stock biomass, and animal mobility."},{"index":28,"size":150,"text":"development constraints in african livestock systems. An important scientific impact, from work not always led by ILRI but often done with its support and advice, was in the transversal systems studies. These include the publications of Sandford (1983a)  78 and others on pastoralism; Pingali et al. (1987) on animal traction in the farming systems of sub-Saharan Africa, McIntire et al. (1992) on crop-livestock integration in sub-Saharan Africa, Thornton et al. (2002) on livestock and poverty mapping, Baltenweck et al. (2003) for crop-livestock interactions on three continents, Thornton and Herrero (2001) on crop-livestock simulation models, McDermott et al. (2010) on sustainable mixed systems and Herrero et al. (2012). Moreover, they have added to the scientific and development consensus about what to avoid in grazing systemsintroduction of exotic animal breeds, hurried privatization of communal grazing tenure, restriction of mobility and oversowing of pasture in conditions unfavourable to the viability of introduced plants."},{"index":29,"size":177,"text":"productivity parameters. The representativeness and accuracy of estimated productivity parameters -such as fertility, mortality, morbidity, milk production and animal growth -were poor when the older systems studies were launched. The goal of estimating factor and input productivities for the main African livestock systems was completed in the first half of the modern era, as shown in ILCA (1979a) for the subhumid zone of West Africa, Wilson (1986) for Mali, Wagenaar et al. (1986) for the interior Delta of Mali, von Kaufmann et al. (1986) for central Nigeria, Coppock (1994) for Borana, and Solomon Bekure et al. (1991) for Maasailand 79 . Following the work begun in the 1960s and 1970s, the scientific understanding of grazing systems productivity has deepened such that policies and projects can be prepared with comparatively cheap additional background work. One important example is the contemporary use (e.g. de Haan, 2016) of parameters derived from the ILCA systems studies in Mali (Wilson et al., 1983;Wilson, 1986). The policy recommendations of de Haan (2016) depend in part on data and analysis from earlier field studies."},{"index":30,"size":198,"text":"A related use of the data -to apply the estimated input-output parameters in systemic models that can simulate technical, policy and management changes -has scientific validity but has had little or no productivity impact. Reasonably detailed and accurate policy analyses have used the input-output parameters (Thornton et al., 2006;Nelson et al., 2009;Thornton and Herrero, 2010;de Haan, 2016), but resulting policy recommendations to date have not had measurable and attributable welfare impacts. bioeconomic models. A second methodological contribution was the application of bioeconomic models to grazing and mixed systems, using station, field and remote sensing data. Examples are Penning de Vries and Heemst (1975), Cartwright et al. (1978Cartwright et al. ( , 1982)), Konandreas and Anderson (1982), Konandreas et al. (1983), Itty (1992), Itty et al. (1995a,b,c) andSolomon Bekure et al. (1991). The influence of this work on poverty analytics is discussed by Thornton and Herrero (2001); Thornton et al. (2002Thornton et al. ( , 2006) ) and Rich and Perry (2011). This work has achieved much in establishing empirical models and in applying them to targeting and analysing effective treatments. Many applied examples to the field of livestock and climate change are discussed in Chapter 16 (this volume)."},{"index":31,"size":105,"text":"A recent book (de Haan, 2016, pp. 79-122) presents an integrated bioeconomic model that uses and extends some of the research discussed in this chapter. This book made the first effort to compare the benefits and costs of technical interventions to preserve livelihoods in East and West African drylands. While it is premature to estimate the economic effects of the model, which depend on application of the proposed interventions by governments (health, market integration and promoting recovery from drought through small ruminants), this book is a landmark contribution to more scientific policy making, which uses many of the historical efforts of ILRI and its predecessors."},{"index":32,"size":89,"text":"the network research model 80 . A further methodological contribution was the study of trypanotolerance and trypanotolerant animals through a novel international network of scientists (Trail et al., 1979a,b;Murray et al., 1990;ILCA/ILRAD, 1988;Rowlands and Teale, 1994;Itty, 1992). The purpose of ATLN (ILCA/ ILRAD, 1988, p. 32) was to provide 'a better understanding of genetic resistance, acquired resistance, environmental factors which affect susceptibility and the efficacy of present control measures, and second by ensuring optimal application of both existing knowledge and recent research findings' (see also Chapter 2, this volume)"},{"index":33,"size":104,"text":"ATLN allowed a rapid growth of knowledge about trypanotolerance across 18 countries, mainly in village situations, in the humid and subhumid climates of sub-Saharan Africa. The work included status reports beginning in 1979, the establishment of field sites in the early 1980s and ultimately the production of a series of landmark papers. Published work after more than a decade of the ATLN had established the genetic basis of trypanotolerance, clarified the relationship between the animal's ability to control parasitaemia and to control anaemia, developed novel diagnostic tools and laid the basis for selecting for trypanotolerance in young stock (Murray et al., 1990, p. 381)."}]},{"head":"Development impact","index":52,"paragraphs":[{"index":1,"size":92,"text":"the direct development impact of lsr in pastoral areas. The direct development impact of LSR in pastoral areas has generally been weak. It is generally impossible to measure the ex post economic impact of grazing system studies by estimating a function relating output or productivity to research or by using indirect methods, such as analysis of citations. Attempting to map the use and effects of new technologies, as proposed by LSR, is also impossible. For these reasons, we conclude that LSR in sub-Saharan Africa has failed to contribute significantly to technical change."},{"index":2,"size":68,"text":"The tropical examples of how knowledge of pastoralism, as derived from characterization research or from integrated work on station and on farm, have changed technologies, policy and productivity are from ranches in Latin America. Despite more than three generations of research in pastoral systems of sub-Saharan Africa, no productivity effects have been observed corresponding to those achieved in soil fertility, pasture management or beef breeds in Latin America."},{"index":3,"size":102,"text":"The chief development impact of grazing systems studies since the 1960s has been to defend the economies and rights of pastoral peoples. This defence, which is an indirect effect of research and extension, has taken the form of grazing rights legislation and consultation with pastoral peoples on their economic, demographic and political interests. Such legislation may have contributed to higher productivity in some grazing situations, but this impact cannot be measured without detailed biological, economic and cultural studies (e.g. Lesorogol, 2008, on northern Kenya, which is an unusual study because it used two survey rounds and stratified by 'privatized' and 'communal' tenure)."},{"index":4,"size":276,"text":"Part of the defence of pastoralists' rights has been to enforce pastoralists' rights or to stop bad policies (de Haan, 2016, pp. 59-61 and 76-77). An example of this defence is the research finding that old legislation, such as the 1965 Grazing Reserve Law of Nigeria (Waters-Bayer and Taylor-Powell, 1986a) had set aside land for herders use but that the areas actually allocated were much smaller than projected. A recent example of the new rights of pastoralists has been the genesis of new laws and regulations governing pastoralism in West Africa. IEMVT, relying on its long tradition of work in the Inner Delta of the Niger in Mali, proposed a 'Code Pastoral' (Gallais and Boudet, 1980) specifying political, legal and institutional reforms designed to regulate land use and to protect the rights of all users -arable farmers, herders and fishers -in the Delta. In Niger, Peyre de Fabrѐgues (1984) advocated a pastoral code to define the rights and responsibilities of herders and farmers, thereby allowing the development and preservation of pastures. The later study of FAO/CIRAD (2013) delineated years of progress in establishing and protecting the legal and administrative rights of West African pastoralists. The FAO/CIRAD (2013) study showed advances in national laws and regulations in Mauritania, Senegal, Mali, Burkina Faso, Niger and Chad, and in international treaties and local administrative practices, all of which have served to protect pastoralist economies against encroachment by arable farming, commercial ranching, urbanization and land grabs by outsiders. While pastoralists' rights are still precarious in much of Africa, research by national and international programmes has contributed to the understanding of pastoralism and to the defence of pastoralists' rights and welfare."},{"index":5,"size":97,"text":"Another successful example is the long tradition of research on the economic aspects of controlling tsetse and trypanosomiasis, to which ILCA/ILRAD/ILRI work has contributed in part over many years (see Chapter 3, this volume). This began with Hans Jahnke's path-breaking work (1974,1976,1982), continued through the African Trypanotolerance Livestock Network (ILCA/ILRAD, 1988; Itty and Swallow, 1994), the summary of Swallow (2000), and the detailed investigations by Shaw (2004) and Shaw et al. (2015), for example. This work has allowed the definition of control models and has provided valuable advice to extension services on the application of those models."},{"index":6,"size":122,"text":"Other possible exceptions are usually not applicable to livestock whether in grazing or in mixed systems. These include: (i) research in which crops, not livestock, were the principal treatment; (ii) the finding of positive returns to new livestock technologies using ex ante rather than ex post techniques (Bryant and Snow, 2008;Kristjanson et al., 1999a, for a trypanosomiasis vaccine; Kristjanson et al., 1999b, for genetic enhancement of cereal crop residues; Thornton et al., 2003, for dual-purpose crops); (iii) efforts to adapt research in temperate countries, where there is a long history of research, data and controlled conditions for estimation and attribution of treatment effects; and/or (iv) research on tropical ranches under more controlled conditions and in more favourable climatic conditions at larger scale."},{"index":7,"size":151,"text":"the weak impact of lsr on productivity in grazing areas. This weak or nil impact is often attributable to poor experimental design. For example, the treatments studied had no effect, because the treatment effect was not measured correctly or because the original research was not designed with economic impact in mind. In some instances, treatments spread through extension and hence the research effect could not be separated from the extension effect. This conclusion about the weak impact of technology innovations applies more forcefully to grazing systems than to mixed systems, where crop research has had a significant indirect effect on animal production via the pathway of higher grain and crop residue output. One notable failure of grazing systems research was testing treatments under unrepresentative conditions; the prominent example of this failure was grazing reserves or group ranches that were too small or that lacked adequate water or land for stock movements."},{"index":8,"size":61,"text":"An important example of LSR's effect on animal health is the development of the infection-and-treatment method vaccine against ECF, as discussed in Chapter 6 (this volume) 81 . The targeting of this method to specific types of livestock production in East and southern Africa has resulted from the work of Brian Perry, Adrian Mukhebi and colleagues on the epidemiology of Theileria."},{"index":9,"size":67,"text":"LSR has generally failed to contribute to successful project preparation and management in grazing and mixed systems (Wanyoike and Baker, 2013). The contribution of research to the development impact of pastoral projects funded and/or managed by the International Fund for Agricultural Development (IFAD) or FAO, as measured in a sample of 194 projects from Latin America, East Africa, the Middle East and North Africa, was not significant."},{"index":10,"size":36,"text":"Components of higher plant productivity under ranching conditions -fencing, rotational grazing, notably pasture improvements using legumes or exotic grasses, mineral fertilizers and use of exotic animal breeds -and attempts in projects to introduce them usually failed."}]},{"head":"Mixed systems","index":53,"paragraphs":[{"index":1,"size":75,"text":"Proposed development pathways in mixed croplivestock systems had the common objective of integrating crops and livestock with, in some instances, specialization in dairying. The logic behind specific technical interventions -adapting animals for farm mechanization, using crop residues as feed, recycling soil nutrients through application of manure or crop residues and introducing high-yielding dairy cattle -was that on-farm resources were underused and could be made more efficient if managed in an integrated crop and livestock enterprise."}]},{"head":"Farm mechanization","index":54,"paragraphs":[{"index":1,"size":153,"text":"One core assumption of LSR was that a lack of adapted tools and quality feed for draught animals blocked crop-livestock integration (Le Moigne, 1979). This assumption was investigated by ICRISAT in India and in West Africa and by ILCA in Ethiopia and West Africa along two lines -developing new farm tools and improving the condition of work animals. Neither the tool line nor the work conditions line produced widely adopt innovations by farmers beyond the intensification response to the increasing value of draught power at higher cropping intensities. Station work on tools did not produce successful innovations and was ultimately abandoned in the 1980s by ICRISAT and in the 1990s by ILCA/ILRI. Station research on feeding draught animals was more successful in that a positive effect was observed on work capacity, but this result did not translate into adoption because alternatives to crop residues for supplemental feeding of work animals were too costly."},{"index":2,"size":113,"text":"The books by Pingali et al. (1987) and McIntire et al. (1992) prompted a rethinking about agricultural mechanization in Africa, including views on animal nutrition as a factor in stimulating demand for work animals. The authors of these publications found many sites where animals were commonly used for tillage and cultivation, even with feed being scarce in the same sites. They also found many sites with adequate feed, in terms of quality and seasonal availability, with little or no draught animal power in use. They concluded that livestock nutrition was not a significant constraint to the adoption of animal traction in sub-Saharan Africa and that research on the topic should accordingly be limited."},{"index":3,"size":179,"text":"The sustained growth in farm mechanization in sub-Saharan Africa over the past 30 years is strong evidence that mechanization, with tractors or animals, does not face insurmountable biological or cultural obstacles and has not blocked growth. The present consensus is that animal traction is a viable path for higher farm productivity, but that research has done, and can do, little to widen that path. Where conditions for the use of animal draught power do not existbecause of low cropping intensity, a domination of bush fallow, animal disease and/or poor market access -then research on health and feed can do little to expand animal draught power. The same lack of demand would, of course, occur if there is a weak yield, area or cropping-pattern effect attributable to animal power on farm, but research to strengthen these effects has not succeeded where the systems conditions are unfavourable. Even as seed and fertilizer packages became more widely used, beginning around 1990 (Walker and Alwang, 2015), they could be used without animal traction and hence did not require or induce greater crop-livestock integration."},{"index":4,"size":200,"text":"An exception to generalizations about agricultural mechanization is the simultaneous introduction of animal power with a cash row crop, such as cotton in the subhumid zone of West Africa, groundnut in the sandy soils of Senegal, Mali, Niger and Nigeria, or maize in the subhumid savannah. In such areas, simple extension programmes, combined with profitable crop production packages, have promoted mechanization without significant new research. crop residues. There is scant evidence of any research impact on animal nutrition, livestock productivity or on soil quality despite the effort invested in feed in mixed (and pastoral) systems. Proposed improvements in feed resources have tended to fail to be completely adopted and have only been partially adopted (see Chapter 13, this volume, for an explanation of adoption failures of planted forage grasses and legumes in the tropics). Ley farming, in the Mediterranean or in the highlands of East Africa, generally failed because farm size and environmental conditions were unfavourable to this technology for anything other than specialized dairy production. The failure of such proposed improvements in communal systems contrasts with the success in ranching, where private land tenure and access to markets, finance and veterinary care made investment in primary productivity more remunerative."},{"index":5,"size":118,"text":"A major indirect effect on livestock productivity occurred via the adoption of higher-yielding crop cultivars. Although the Walker and Alwang (2015) record of area and yield increases of improved cultivars did not discuss effects on fodder or by-product yields in any of the species studied, Indian experience with major cereals (rice, wheat, maize, sorghum and pearl millet) has shown such effects to be important (Blümmel et al., 2013(Blümmel et al., , 2014; see also Chapter 14, this volume). A related effect, although not well measured or related to research or extension investments), was the expansion of irrigation and the associated growth of double or triple cropping, allowing higher fodder output per season and longer periods of fodder abundance."},{"index":6,"size":104,"text":"the legacy of inadequate crop residue research. The older work found that crop residues were important shares of animal diets in the livestock systems studied by ILRI, ICARDA and ICRISAT, and in South Asia (Reed et al., 1988;Kelley et al., 1993;Renard, 1997)  82 . Recent studies in sub-Saharan Africa (e.g. Hiernaux and Ayantunde, 2004) have confirmed this pattern. Crop residues have always been a significant share of ruminant feed intake on small mixed farms; there is significant growth potential from better use of crop residues, and it is likely that the increase in crop residues associated with higher grain production has stimulated livestock production."},{"index":7,"size":133,"text":"IARC research has done too little on crop residues as feed, despite their importance for smallholder livestock throughout sub-Saharan Africa, West Asia and North Africa, and South Asia. A recent and major example of this shortcoming is the Africa book by Walker and Alwang (2015), which does not report crop residue yields 83 or relate them to grain yields. While the allocation of crop residues between crops and animals was a major research theme at ILCA, in the semi-arid tropics and the Ethiopian highlands, and to a lesser extent at ICARDA in Mediterranean climates, the impact of this work on livestock systems productivity has been weak. The failure to devote adequate research to crop residue quality, and its relation to the harvest index, is the single most important gap in African livestock research."},{"index":8,"size":154,"text":"nutrient cycling. The extensive soil fertility research for mixed farms in sub-Saharan Africa has had an indirect and weak development impact. Nutrient cycling in mixed systems has been studied by Powell et al. (1995) for sub-Saharan Africa, and Boddey et al. (1996) for Brazil and Colombia. This is not because mineral fertilizer use has not grown; there has been significant growth of fertilizers and other modern inputs in sub-Saharan agriculture since the mid-1970s. First, most of the stimulus for mineral fertilizer use has come from economic growth and market development in many African countries, which had both demand and supply effects on fertilizers applications (Townsend, 1999;Morris et al., 2009). A second stimulus has been intensification of farming systems under the pressure of population growth and greater market access. An indirect effect of research has been on the recent growth in areas sown with new plant varieties, especially of maize, which has stimulated fertilizer use."}]},{"head":"What blocked the translation of scientific impact into development impact?","index":55,"paragraphs":[{"index":1,"size":72,"text":"The contradiction of strong scientific impact and weak development impact was seen in both pastoral and mixed systems. This contradiction had several causes: (i) the difficulty of raising primary productivity in arid and semi-arid regions; (ii) the small farm mandate of the international centres, especially in sub-Saharan Africa; (iii) the persistent misapplication of models from other agricultural systems; (iv) policy bias against agriculture; and (v) lack of background data at the outset."},{"index":2,"size":203,"text":"low primary productivity in arid and semi-arid areas. The translation of knowledge into productivity has largely failed in the dry areas. One strong indicator of this failure is that there have been few sustained investments in plant production (e.g. de Haan, 2016). Hypotheses about the share of primary productivity in the yield gap between actual and potential yield, for both grazing and mixed smallholder systems, proved fruitful in terms of scientific impact but less so in terms of economic impact. New planted forages have not been generally successful in the African tropics, with the chief exceptions of cut and carry forages in highland dairying systems and nitrogenous trees in the humid tropics. The straw components of multidimensional crops have been highly successful as by-products of research on cereals, but breeding programmes targeted at changing plant architecture to lower the harvest index have not done well. An indirect effect of the failure to raise the quantity and quality of feed production was the limited profitability of animal fattening because of the high costs of feed. If the profitability of fattening is changing (e.g. de Haan, 2016), it is more the result of higher incomes shifting demand than it is to research reducing production costs."},{"index":3,"size":144,"text":"the iarc mandate. The mandate of the international centres in sub-Saharan Africa and South Asia -to concentrate on small undercapitalized farms that used few purchased inputs and not on large specialized ranches, tree crop estates or well-capitalized arable farms 84 -explains some of the failure to observe a higher development impact of systems research. This is not a criticism of the mandate -it is a statement of one of its inevitable consequences -nor does this mean that the international centres should have concentrated on larger, more capitalized farms; it means that the choice of the small mixed farmer as the principal IARC client necessarily reduced returns to research because the mandate area is more difficult. The long-term success of private animal breeding research in the USA, for example, is a complement to public research in the USA that has no analogue in sub-Saharan Africa."},{"index":4,"size":327,"text":"factor proportions and the misapplication of external models. One obstacle to applying research results in sub-Saharan Africa was the misapplication of principles from ranching systems in the USA, Australia, New Zealand and Latin America. These principles -commercial orientation, private land tenure, pasture improvements and limited animal mobility -were generally not applicable to African pastoralism. A confirmation of this was the finding, long ago, that African rangeland systems were already efficient in terms of live-weight production per hectare, as shown by Cossins (1985) comparing East Africa and Australia, and Breman and de Wit (1983) for the USA, Australia and Mali. Other examples of technology transfer failure include planted forages and fencing. Despite their efficiency at low cash investment and high labour input, the African pastoral economies have not grown as fast as other sectors in the same economies; this is not because of a failure to apply research results but rather from the inability to generate research results that raise productivity in arid and semi-arid climates. A contributor to weak research impact has been the persistence of errors in defining problems. Beginning with the fundamental error of ILCA -that technologies existed in 1975 that could raise productivity of African livestockthe pattern continued for decades and was seen as recently as the recommendation of Little and Dube (2011) for species and breed diversification in pastoralism. This recommendation has failed because breed selection and upgrading for cattle in the African tropics has long been blocked by heat, thirst, disease and insects; none of these adverse factors has been removed by research (with again the notable exception of the eradication of rinderpest in 2011). Beyond breed diversification, cross-breeding in cattle has been unsuccessful in grazing and mixed systems. The only major success is the use of higher-yielding dairy animals in the cooler highlands of East Africa. The reason, as shown in experiments and in the opinions of the herders, was that exotic races were more susceptible to heat, thirst and disease."},{"index":5,"size":118,"text":"Contrasting factor proportions -historically low labour:land ratios in some parts of sub-Saharan Africa and much higher ratios in others -produced contrasting obstacles to research. It may be argued that rural labour scarcity has disappeared in Africa with the doubling of rural population density over the past two generations, but this is not the way to look at labour scarcity. Labour scarcity is a function of relative wages. If urban productivity, and the purchasing power of urban labour, are higher than the corresponding rural values, then labour will be scarce in rural areas even if rural population density has grown. If rural technologies lag behind urban ones, then so will rural productivity, producing the appearance of rural labour scarcity."},{"index":6,"size":45,"text":"policy bias against agriculture. The long history of bias against market agriculture in Africa is well documented. Macro-policies were unfavourable to technologies that were profitable under experimental conditions (e.g. exchange-rate policies would have discouraged the production of tradable beef and encouraged production of non-tradable milk)."},{"index":7,"size":74,"text":"Lack of complementary policies and extension effort reduced the development impact of technologies. An example is trypanosomiasis control. In the absence of a vaccine against this disease, an integrated campaign was needed to apply the knowledge derived from systems research -drug use, management of resistance, vector control and management of trypanotolerant stock -and practically all elements of such an integrated campaign have been difficult to sustain because of fiscal choices made by African countries."},{"index":8,"size":64,"text":"lack of background data at the outset. System information -soils, rainfall, LGP, land-use mapping and productive capacity -only became generally available 10-20 years after independence in sub-Saharan Africa 85 . It took many years for such information to spread and to be usable to policy makers and producers; lack of information about resources would have limited uptake of technologies adapted to specific resource types."},{"index":9,"size":85,"text":"The lack of background data at the outset of the systems studies was aggravated by some features of older research. Many older studies were too qualitative, had data quality problems, notably small sample sizes, and demonstrated failure to conduct or use station research jointly with field studies. In sharp contrast to the African situation 50 years ago, environmental data in 2020 is much more widely and cheaply available, and lack of such data can no longer be adduced as a reason for costly research delays."},{"index":10,"size":400,"text":"This chapter has found few examples of ex post economic analysis of grazing systems research and development in sub-Saharan Africa. Such analysis rarely shows value in LSR because the data were not collected or because returns were negative where the treatment effect was nil or even negative 86 . Where there were data on inputs and outputs needed to estimate productivity, it is difficult or impossible to establish a functional link between research effort and productivity at the level of the decision-making unit of the farm or the policy unit of the sector. Lastly, in most grazing systems, detailed studies are not needed to show that research has failed to raise productivity (whether by increasing output or by lowering input costs) and hence the necessary work to make productivity research estimates has never been done. data problems in livestock studies. Many livestock studies had data problems that vitiated their use in technology or policy analysis. Many studies were done over periods that were too short to capture interseasonal and interannual variability because longer studies were thought to be too costly. Given the interannual variability in rainfall and feed supply, observations of even as long as 5 years produced noisy estimates of productivity and its determinants. Even where the biophysical data were of adequate quality to allow economic analysis (e.g. the Nigeria study by von Kaufmann et al., 1986, or the Mali studies of Wilson, 1986, Wagenaar et al., 1986, and Penning de Vries and Djiteye, 1991), their samples across units and time were usually too small to make reliable parameter estimates for periods longer than 2 years. Even where data were of adequate coverage, problems of selection bias, definition of the control or relevance of the treatment made data analysis inconclusive at best. lsr was sparse and too often qualitative. There are few examples of rigorous productivity estimates in LSR, even of single livestock technologies. One example is A History of Farming Systems Research (Collinson, 2000), which has 12 chapters and more than 50 authors on dozens of topics under the general theme of farming systems research in the tropics, covering work from 1960 to the late 1990s. The book has only three empirical studies of animal production: goat mange in , and alpaca and dual-purpose cattle in . It fails to mention ILRI and its predecessors, while devoting only a few pages (pp. 110-111) to onfarm research with livestock."},{"index":11,"size":75,"text":"The ICRISAT compendium on 'Socioeconomic Constraints to Development of Semi-arid Tropical Agriculture' (Ryan and Thompson, 1980) published some 5 years after ICRISAT's founding has more than 40 papers, which focus almost entirely on crops. There is one section of the paper on animal traction for smallholdings, one chapter referring to milk production, scant mention of forage crops and one paper (Le Moigne, 1979) on specific problems of on-farm research with livestock or with animal-powered implements."},{"index":12,"size":82,"text":"Another prominent example of a lack of quantitative work is Sandford's (1983a) landmark book on pastoralism in the third world, which has no quantitative analysis of technologies, such as grazing management, pasture improvement, animal breeding, calf supplementation, water investment or public infrastructure. Early quantitative work, such as the cattle model of Konandreas and Anderson (1982) or the Thornton (1987) model for Colombia, and others tended to have short lives and apparently were not applied to generate or to induce gains in productivity."},{"index":13,"size":149,"text":"discontinuation of long-term bioeconomic studies. Work after the merger of ILCA and ILRAD has largely dropped integrated long-term studies, with the exception of the work in Fakara and the Kenyan dairying studies of the mid-1990s. Despite the changes in population density, cropping intensity, market access, animal numbers, herd composition and infrastructure (transport, communications, water and energy) and urbanization since the 1980s, the only recent integrated work in African grazing systems is that of Homewood and Rodgers (2004) and Homewood (2008) for East Africa, Lesorogol (2008) for northern Kenya, FAO/CIRAD (2013) for the Sahel and some chapters in Catley et al. (2013). The decline in long-term field research has stopped the generation of adequate data on research and productivity and has therefore prevented comparisons over time. Replacements with ex ante projections over very long periods will continue to give uncertain results and therefore will continue to be unreliable policy guides."}]},{"head":"The Future","index":56,"paragraphs":[{"index":1,"size":131,"text":"The long-term data requirements of livestock systems studies are made more urgent by secular changes in crop and animal productivity imposed by climate shifts. Therefore, for mixed and pastoral systems alike, it will be necessary to conduct new field data collection and analysis while respecting the hypothesis testing requirements of climate change models, especially with respect to parasitology and resistance to abiotic stresses, notably heat and water scarcity. The long-term effort to sustain and use the ICRISAT village-level studies in India should be a model for LSR in sub-Saharan Africa. The fact that we are unable to make reliable estimates of the economic return to agricultural research, and especially to livestock research, in sub-Saharan Africa means that a new dedicated effort is required over many years; this effort will be expensive."}]},{"head":"Pastoral systems","index":57,"paragraphs":[{"index":1,"size":27,"text":"Governance research should be extended in pastoral areas with the explicit political goals of defending the land rights of pastoralists and preserving minimum areas for these groups."},{"index":2,"size":90,"text":"Long-term data collection and analysis should be restarted for pastoral systems in northern Kenya, southern Ethiopia and in parts of Mali, Niger and Sudan. Special efforts must be made where the conflict between wildlife and domestic livestock is acute, in both economic and biological terms; this is pointedly true along the entire length of the Rift Valley. Such data collection needs to be linked to verifying predictions of climate change and disease distribution models as a way of contributing to policy debates and of estimating the costs of policy measures."}]},{"head":"Mixed systems","index":58,"paragraphs":[{"index":1,"size":32,"text":"Efforts should be made to strengthen or restart long-term data collection and analysis in densely cultivated areas, such as northern Nigeria, the East Africa highlands and the river basins in the Sahel."},{"index":2,"size":2208,"text":"unproductive notion of pastoralist mismanagement was insightfully criticized by ILCA/ILRI scientist Ralph von Kaufmann (von Kaufmann,1976). 33 Grigg continued to say, correctly, that it would be unwise to enforce sedentarization of pastoralists. 34 Many distinguished scientists were involved in the Borana work including Jean-Claude Bille, Assefa Eshete, Michel Corra, C.S. Kamara, Mark Nicholson, Jess Reed, Solomon Desta and Andrea Woodward. 35 Recent work (de Haan, 2016) introduces subclasses of LGA and livestock grazing (LG) semi-arid and LG sub-humid, which would apply to parts of the Borana study area. 36  Toutain and Boudet (1980, pp. 427-432) mentioned historical restrictions on cutting browse from Acacia albida, imposed either by the traditional authorities in eastern Niger, by the French colonists, or by the government of independent Mali in the 1960s, following the colonial model. 37 ILCA abandoned a study of Afar pastoralists in eastern Ethiopia for security reasons in the 1970s. 38 One example of the information available to the Maasailand researchers is von Kaufmann (1976) covering pastoral problems and proposed solutions in Kenyan rangelands up to the mid-1970s. 39 ' Agro-pastoralists' in West Africa usually means Fulani farmers who practised a mix of settled rain-fed farming and seasonally nomadic grazing. In some instances (Blench, 1999, for northern Nigeria;Waters-Bayer and Taylor-Powell, 1986b) such groups had been settled for many years. In others (Benoit, 1979, for western Burkina Faso), herders had recently settled or were in the process of settling as recently as the 1970s. 40 This book is an outstanding work that is regrettably barely cited in Google Scholar and not cited at all in Scopus. It made a major contribution to the scientific literature, notably through the work of Wolfgang Bayer, David Bourn, M.A. Ibrahim, Salisu Ingawa, J.A. Maina, E O. Otchere, Mark Powell, Mohammed Saleem, Ellen Taylor-Powell, Ralph von Kaufmann, Ann Waters-Bayer and William Wint. 41  Ellis and Swift (1988) stressed that herders could recover well from droughts of one year, but that longer dry periods would cause greater losses of animals, work and income, forcing herders into other jobs. 42 Grigg's observation is correct in that the ranches did fail, but his explanation of the failure is wrong. 43 The studies by Fratkin et al. (1994) and Galaty (1994, p. 189) did not quantify the losses in productivity and equity from enforced sedentarization, but it is clear from these works that such losses were substantial. 44  Mortimore (2000, p. 4), in three states in northern Nigeria, found a decline in agricultural land (cropped) of 8. 4% from 1976-1978 to 1993-1995, of which -12.4% was woodlands, +1.3% was grassland and +3.0% was degraded. 45 This was decades after the British had taken much of the Maasai grazing land, which had narrowed the resource base of the group ranches before they were ever constituted. 46 See Rutten (1992) on the individualization of ownership in Kajiado district of southern Kenya from 1890 to 1990. 47 The technical reviews by King (1983) and Sandford (1983a) concerned cattle for meat and dairy. Sandford (1983a, p. 49) noted a lack of information about water use by draught animals in highland Ethiopia. 48  Homewood and Rodgers (2004, p. 252) reported from Tanzania that energy loss incurred by trekking for water was the 'single biggest constraint to milk production' . 49 The books of Smith (1992), Fratkin et al. (1994); Scoones and Wolmer (2002), Behnke and Scoones (1993), Homewood (2008), Bollig et al. (2013) and Catley et al. (2013) do not mention water experiments, on station or in the field. The range and herd modelling in Behnke and Scoones (1993) did not use water as an objective or a constraint, nor did that of Konandreas and Anderson (1982). Mengistu et al. (2007) observed successful adaptive physiological mechanisms in Ethiopian Somali cattle when subjected to intermittent watering. 50 This statement does not apply to Latin America, where the market potential of beef had stimulated much earlier research on soils, pastures, livestock disease and animal breeds for commercial ranching. 51 Examples are Monod (1975) and papers cited in Sandford (1983a). 52 Chapter 13 discusses the research impact of quality improvements in crop residues for mixed systems, mainly in semi-arid India and in semi-arid and subhumid West Africa. 53 The book by Pratt and Gwynne (1977), while limited geographically to East Africa, covered more than 30 years of work and made careful reference to rangeland studies in the USA and Australia. 54 Feed quality was a major theme in the early CIAT Beef Program (CIAT, 1973). 55 The papers cited by Macharia et al. (2011) confirm their findings: station results are positive, field trials are mixed, adoption data are rare and benefit-cost analysis is absent. The study by Nicholson and Mengistu (2016) confirmed the lack of adoption of forage legumes in grazing systems and on mixed farms in Kenya and Ethiopia using surveys in 2014 and 2015. 56 Chapter 11 (this volume) covers the African range ecology work of ILCA, ILRI and partners. Successes with planted forage grasses in Latin America were mainly on large commercial farms in Brazil and on some smallholder areas in central America, as discussed in Chapter 12 (this volume). 57 Medicago sativa (known as alfalfa in the USA and lucerne in much of Europe) is the most common temperate forage legume but is rarely planted in the semi-arid and subhumid tropics. 58 Detailed papers on these diseases are in the respective chapters of Part I. Screw worm was accidentally introduced into Libya from the new world in the late 1980s and was eradicated with sterile insect techniques and quarantine measures by the mid-1990s.The eradication of screw worm in Libya and the global eradication in 2011 of rinderpest, a viral disease of cattle and some ungulates, are special cases and are not discussed here. 59 Konandreas and Anderson (1982, pp. 3-5) reviewed earlier models, including that of CIAT for ranching in Latin America and that of Texas A&M University (Cartwright et al., 1982). 60  Norton (1975, pp. 313-322) wrote that it would be 'unwise to apply our understanding of American deserts directly to the Sahel' because of differences in stocking rates, climate and flora. 61 The bibliography in Penning de Vries and Djiteye (1991) cites many individual studies of soils, water, plants and animals, but few analysed these elements jointly over time, at the same site, or applied simulation models to validate and project the field results; a significant exception is Hiernaux and Ayantunde (2004). 62 The IFPRI IMPACT model was not detailed enough in 2009 to project outcomes for individual livestock systems of any type (Nelson et al., 2009). 63 It is not possible to calculate an IRR for the 'milking with supplementation' strategy from the original paper, but it appears from Cartwright et al. (1978, p. 60, Table 5.4) that it would be positive. 64 Water was not a major cost of cattle production in the Inner Delta of the Niger River in Mali and hence the studies there had at least one large difference from other work in semi-arid West Africa. Wagenaar et al. (1986, p. 5) found that, 'Throughout the year, the herds are watered at least twice daily.' 65 Coppock (1994, p. 196) referred to 'grain imports' but his context implied not 'imports' in the sense of purchases from foreign suppliers but 'imports into the grazing system' in the sense of purchases from domestic and/or foreign suppliers. 66 Jahnke (1982, p. 143) estimated that roughly 275,000 km 2 had been cleared of tsetse in Nigeria, Zimbabwe, Tanzania and Uganda between 1947 and 1978 in a total endemic area of 10.3 million km 2 in sub-Saharan Africa (see Box I.1 in the Introduction in this volume). 67 Vetter (2004) summarized the origins of the equilibrium model, the critique by advocates of the nonequilibrium model and the situations in which one or both might apply. 68  Dixon et al. (2001) defined eight groups, of which six are relevant here; the others are coastal artisanal fishing and urban agriculture. 69  Walker and Alwang (2015, p. 210) found that fewer than 10% of maize scientists in East and southern Africa were from 'social science' or 'farming systems' . They also found that more than 60% of scientists in 20 crops were in plant breeding, plant pathology, molecular biology and tissue culture (p. 378). 70 The term 'conservation farming' has been introduced more recently as a model of nutrient cycling (e.g. Dixon et al., 2001, pp. 51-52). 71 Farm mechanization with power tillers and tractors was part of the research portfolios of the International Rice Research Institute (IRRI; Chancellor, 1998), and of ICRISAT at times. The goal of this research was to accelerate field tasks by replacing animals to achieve higher cropping intensity. IRRI and other centres also studied mechanization of water supply, threshing and milling, but this work typically involves engines rather than animal power. 72 Some loss of potential impact from the Niger study was due to using a sample size of only seven animals in its trial of energy expenditures, and to sampling animals who weighed roughly one-third more than those animals typically found on farms in western Niger (Fall et al., 1997). 73 There was little soil fertility work 'on farm' in the pastoral areas. 74 CIAT's orientation was on improving soil fertility by adding mineral fertilizers and raising the pH of acid soils without returning manure to the soil in a systematic way. The CIAT work had a major development impact on the acid soils of the Latin American savannahs, estimated to be more than 250 million ha in Brazil, Colombia and Venezuela alone (Lynam and Byerlee, 2017, p. 83) and is one of the principal successes of natural resource management work from the international agricultural research centres. 75 The cattle age/sex pyramid for the Malian Delta in Wagenaar et al. (1986, p. 10) showed the ratio of females to males increased sharply after the age of 3-4 years. The cattle age/sex data for Mali (Wilson,1986, p. 36) found higher female:male ratios in herds not used for animal draught; in herds used for draught power, the female:male ratios were roughly equivalent in one sedentary herd for milk and draught and much less than 1 in another sedentary herd. The female:male ratio in the Maasailand study (Solomon Bekure et al., 1991, p. 83) ranged from 1.86 (rich households) to 2.23-2.30 (poor households) with a mean of 1.97. The ratio of females:males was greater than 3 before the 1983 drought in southern Ethiopia and about 4 in 1985 after the drought (Coppock, 1994, p. 168) 76 The foreword to Pratt and Gwynne (1977, p. vii) illustrated the 'mainstream view': '…in the last few decades, stock numbers have increased so much that extensive areas have been severely overgrazed and now have an extremely low annual productivity, far below that which the land is capable of producing under good management' . 77 Examples are Thornton et al. (2006), Thornton et al. (2009) and Thornton and Herrero (2010). 78 The books of Katherine Homewood (notably, Homewood and Rodgers, 2004;Homewood, 2008), on East African pastoralism, although not ILRI work, are in the same tradition. 79 An exception is parts of East and southern Africa, where ILCA and ILRAD did little apart from early cattle modelling on Botswana (Konandreas and Anderson, 1982) and the epidemiological work of Brian Perry and colleagues in the second half of the 1980s. 80 Network models have also been used in animal traction, forages, animal genetics and economics. 81 The exception to this generalization about the lack of a productivity effect from livestock research in sub-Saharan Africa is the eradication of rinderpest (Roeder and Rich, 2009). Even for rinderpest, the modern research effect was limited to a policy for delivery systems. The vaccine was developed long before eradication was declared in 2011 and the role of systems research in the campaign was very small. 82 There is a long history of crop research in the USA, Canada and Australia on the value of crop residues as a soil amendment. A review by Wilhelm et al. (2004) on removal of stover in the Maize Belt in the USA found studies covering nearly 70 years. 83 Chapter 13 (this volume) reports rough estimates of the contribution of improved cultivars of dryland cereals (maize, pearl millet and sorghum) to animal feed availability for India and some countries in sub-Saharan Africa. Walker and Alwang (2015, pp. 32, 228 and 232) made occasional reference to straw as feed but did not report data on crop residues as feed, changes in harvest index in modern varieties or changes in feed quality of crop residues. 84 There was a focus on large farms at CIAT in research on rice cultivars, beef ranching, forage grasses and acid soil management. 85 This had been available for decades in the livestock systems in South Asia and parts of Latin America. The ILCA (1975) proceedings of 'Inventaire et cartographie des Pâturages Tropicaux Africains: Actes du Colloque Bamako' was a landmark in compiling and publishing this information. 86 The chapter by Evenson (2001) in the Handbook of Agricultural Economics cited one estimate on pasture research among more than 100 estimates of returns to research, and fewer than five papers on livestock in a global total of more than 100 papers on crops and livestock."}]},{"head":"84","index":59,"paragraphs":[{"index":1,"size":43,"text":"This had been available for decades in the livestock systems in South Asia and parts of Latin America. The ILCA (1975) proceedings of 'Inventaire et cartographie des Pâturages Tropicaux Africains: Actes du Colloque Bamako' was a landmark in compiling and publishing this information."}]}],"figures":[{"text":" Fig. 15.1. Characteristics of global livestock systems by region, 1991-1993. CSA, central and southern Africa; EA, East Africa; SA, South Africa; SEA, South-east Asia; SSA, sub-Saharan Africa; WANA, West Asia and North Africa. (Data fromRobinson et al., 2011, p. 48.)     "},{"text":"Fig. 15 . 2 . Fig. 15.2. Characteristics of sub-Saharan Africa livestock systems by agroecology, 1991-1993.(Data fromSeré and Steinfeld, 1996.)  "},{"text":"Fig. 15 . 3 .Fig. 15 . Fig. 15.3. Livestock production index for East and southern Africa (a) and West and Central Africa (b), 1970-2016. Black lines, selected countries; red dashed line, regional trend; green dashed line, regional trend per capita terms estimated by ln(Y) = 0.3599 + 0.0021 × year (a) or by ln(Y) = -9.1269 + 0.0068 (b). (Data from www.faostat.org; accessed 30 April 2020.) "},{"text":"Fig. 15 . 5 . Fig. 15.5. Sheep and goat numbers relative to cattle numbers for East and southern Africa (a) and West and Central Africa (b), 1970-2016. Black lines, selected countries; red dashed line, regional trend estimated by ln(Y) = 0.1 + 0.0001 × year (a) or ln(Y) = -33.2 + 0.0171 × year. (Data from www.faostat.org; accessed 30 April 2020.) "},{"text":"Fig. 15 . 6 .Fig. 15 . 7 .Fig. 15 . 8 . Fig. 15.6. Arable land per capita for East and southern Africa (a) and West and Central Africa (b), 1970-2016. Black lines, selected countries; green dashed line in per capita terms, regional trend estimated by ln(Y) = 21.4 -0.0115 × year (a) or ln(Y) = 27.5 -0.0143 × year. (Data from www.faostat. org; accessed 30 April 2020.) "},{"text":"Fig. 15 . 9 . Fig. 15.9. Food production index for East and southern Africa (a) and West and Central Africa (b), 1970-2016. Black lines, selected countries; red dashed line, regional trend estimated by ln(Y) = -59.3 + 0.0319 × year (a) or ln(Y) = -68.6 + 0.0365 × year (b); green dashed line, regional trend in per capita terms estimated by ln(Y) = -5.6 + 0.0051 × year (a) or ln(Y) = -19.2 + 0.0119 × year (b). (Data from www. faostat.org; accessed 30 April 2020.) "},{"text":"  "},{"text":"  "},{"text":"Table 15 .1. Selected pastoral systems studies, various years.     Population       Population   Country/  density at Farming Scales of  Livestock Cultivation Proposed Country/density atFarmingScales ofLivestockCultivationProposed Study Region/period sub-region/area Climate outset (n/km 2 ) systems research Species density intensity technologies StudyRegion/periodsub-region/area Climateoutset (n/km 2 )systemsresearchSpeciesdensityintensitytechnologies Dyson- East Africa, Uganda/ Aw, BSh <20 Subhumid Regional Cattle Variable, Very low Not mentioned Dyson-East Africa,Uganda/Aw, BSh<20SubhumidRegionalCattleVariable,Very lowNot mentioned Hudson 1966 Karamoja   pastoral territory  typically   Hudson1966Karamojapastoralterritorytypically (1966)        low   (1966)low Monod 1950s-1970s North Africa and Arid Variable Arid grazing Saharan, Camels, Variable Very low Not highlighted in Monod1950s-1970sNorth Africa andAridVariableArid grazing Saharan,Camels,VariableVery lowNot highlighted in (1975)  rangelands of    Sahel cattle,   book (1975)rangelands ofSahelcattle,book   sub-Saharan     goats    sub-Saharangoats   Africa         Africa ILCA (1975) 1960s-1975 Sub-Saharan Arid, semi-arid, Variable Arid grazing, Arid and Cattle, Varied across Not studied Grazing ILCA (1975) 1960s-1975Sub-SaharanArid, semi-arid,VariableArid grazing,Arid andCattle,Varied acrossNot studied Grazing   Africa BWh, BSh across arid semi-arid goats, continent  management AfricaBWh, BShacrossaridsemi-aridgoats,continentmanagement     continent cropping sub-Saharan sheep,    continentcroppingsub-Saharansheep,       Africa camels    Africacamels Dahl and East Africa, East Africa, Arid, semi-arid, <20 Arid grazing, Semi-arid Cattle, Very low Not studied Herd Dahl andEast Africa,East Africa,Arid, semi-arid,<20Arid grazing,Semi-aridCattle,Very lowNot studied Herd Hjort 1976 Somalia BWh, BSh  limited and goats,   management, Hjort1976SomaliaBWh, BShlimitedandgoats,management, (1976)     cropping subhumid sheep,   milk offtake (1976)croppingsubhumidsheep,milk offtake       East Africa camels    East Africacamels Fricke Nigeria, 1960s Nigeria Semi-arid, Variable from LGA, LGT,  Cattle   Herd FrickeNigeria, 1960sNigeriaSemi-arid,Variable fromLGA, LGT,CattleHerd (1979) and 1970s  subhumid 20 to 250 MRA, MRT     management, (1979)and 1970ssubhumid20 to 250MRA, MRTmanagement,           animal health, animal health,           pasture pasture           production production Jahnke Sub-Saharan Sub-Saharan Arid to humid Variable LGA, LGH, Nation, Ruminants Variable Variable  JahnkeSub-SaharanSub-SaharanArid to humidVariableLGA, LGH,Nation,Ruminants VariableVariable (1982) Africa, 1970s Africa   LGT, MRA, territory,     (1982)Africa, 1970sAfricaLGT, MRA,territory,      MRH MRT farm     MRH MRTfarm Sandford 1960s-1970s Global tropics Many, mainly Variable; not LGA, LGH, Territory Cattle, Variable, Zero to low Herd Sandford1960s-1970sGlobal tropicsMany, mainlyVariable; notLGA, LGH,TerritoryCattle,Variable,Zero to low Herd (1983a)   arid and highlighted LGT  camels, typically  management, (1983a)arid andhighlightedLGTcamels,typicallymanagement,    semi-arid in book   goats low  water semi-aridin bookgoatslowwater           development, development,           secure land secure land           rights rights  "},{"text":"Table 15 . 2 . Productivity comparisons of ranching and pastoralism, various years.      Pastoral productivity Ranching Pastoral productivityRanching Study Region/period Subregion/country Climate/system Species estimates (TLUs/ha) (TLUs/ha) StudyRegion/periodSubregion/countryClimate/systemSpeciesestimates (TLUs/ha)(TLUs/ha) Dyson-Hudson East Africa, Karamoja, Uganda Warm semi-arid Cattle Not specific Unknown Dyson-HudsonEast Africa,Karamoja, UgandaWarm semi-aridCattleNot specificUnknown (1966) 1966  (Bsh), tropical    (1966)1966(Bsh), tropical    savannah (Aw),    savannah (Aw),    LGH, LGT    LGH, LGT Dyson-Hudson and Global Global Tropical, temperate, Ruminants, Not specific Not specific Dyson-Hudson andGlobalGlobalTropical, temperate,Ruminants,Not specificNot specific Dyson-Hudson   highland camels   Dyson-Hudsonhighlandcamels (1980)       (1980) ILCA (1975) 1960s and Drylands, sub- Semi-arid (Bsh), Cattle, goats, Variable; 0.05-0.2 in Verify in ILCA (1975)1960s andDrylands, sub-Semi-arid (Bsh),Cattle, goats,Variable; 0.05-0.2 inVerify in  1970s Saharan Africa tropical savannah sheep, camels Sahelian conditions Breman 1970sSaharan Africatropical savannahsheep, camelsSahelian conditionsBreman    (Aw), LGA, LGH,    (Aw), LGA, LGH,    LGT, MRA, MRH    LGT, MRA, MRH Dahl and Hjort East Africa, Western Sudan, Arid, semi-arid, Cattle, camels, Not stated Not stated Dahl and HjortEast Africa,Western Sudan,Arid, semi-arid,Cattle, camels,Not statedNot stated (1976) 1976 Kenya LGA, LGH sheep, goats   (1976)1976KenyaLGA, LGHsheep, goats Fricke (1979) 1970s North-central Semi-arid, Mainly cattle   Fricke (1979)1970sNorth-centralSemi-arid,Mainly cattle   Nigeria subhumid, LGH,    Nigeriasubhumid, LGH,    MRH    MRH Pratt and Gwynne 1960s and East Africa Arid, semi-arid, Mainly cattle   Pratt and Gwynne1960s andEast AfricaArid, semi-arid,Mainly cattle (1977) 1970s  subhumid, LGA,    (1977)1970ssubhumid, LGA,    LGH    LGH Jahnke (1982) Sub-Saharan Sub-Saharan Africa Very arid to humid, Ruminants 0.1-0.5 by climate,  Jahnke (1982)Sub-SaharanSub-Saharan AfricaVery arid to humid,Ruminants0.1-0.5 by climate,  Africa, 1970s  LGA to MIT  ~0.35 considered  Africa, 1970sLGA to MIT~0.35 considered      max subhumid  max subhumid Wilson (1986) West Africa, Niono, Mali Arid to semi-arid, Ruminants 0.02-0.25 in arid and  Wilson (1986)West Africa,Niono, MaliArid to semi-arid,Ruminants0.02-0.25 in arid and  1980s  LGA, LGH, MRA,  semi-arid conditions  1980sLGA, LGH, MRA,semi-arid conditions    MRH    MRH Breman and 1970s and East Africa and Arid, semi-arid, Ruminants In protein kg/ha, 0.4 In protein kg/ Breman and1970s andEast Africa andArid, semi-arid,RuminantsIn protein kg/ha, 0.4In protein kg/ de Wit (1983) early 1980s some temperate temperate, LGA,  for nomads, 0.6-3.2 ha,0.3-0.5 de Wit (1983)early 1980ssome temperatetemperate, LGA,for nomads, 0.6-3.2ha,0.3-0.5   countries LGH  for seasonal for USA, 0.4 countriesLGHfor seasonalfor USA, 0.4      transhumants, 0.3 for Australia transhumants, 0.3for Australia      for sedentary  for sedentary Sandford (1983a) 1960s-1970s Developing Mainly arid to Cattle, camels,   Sandford (1983a)1960s-1970sDevelopingMainly arid toCattle, camels,   countries semi-arid goats   countriessemi-aridgoats  "},{"text":" relied heavily on the ILCA/Malian IER work at Niono and in the Niger Delta of Mali and on earlier research by the L'Institut d'Elevage et de Médecine Vétérinaire des pays Tropicaux (IEMVT) on Sahelian livestock and agrostology in Niger and Senegal.  "},{"text":"Table 15 . 3 . Feed recommendations in systems studies, various years.   Country/     Labour intensity of Country/Labour intensity of Study Region/period subregion/area Climate Farming system Species Feed recommendations recommendations StudyRegion/periodsubregion/areaClimateFarming system SpeciesFeed recommendationsrecommendations Dahl and Hjort East Africa Western Sudan, Arid to LGA Cattle, camels, Not specific Unknown Dahl and HjortEast AfricaWestern Sudan,Arid toLGACattle, camels,Not specificUnknown (1976)  northern Kenya semi-arid  sheep, goats   (1976)northern Kenyasemi-aridsheep, goats Dyson-Hudson Mainly East Several Arid to Extremely Ruminants None Unknown Dyson-HudsonMainly EastSeveralArid toExtremelyRuminantsNoneUnknown and Dyson- Africa  subhumid variable    and Dyson-Africasubhumidvariable Hudson (1980)    mobility types    Hudson (1980)mobility types Jahnke (1982) Sub-Saharan  Very arid to LGA, LGH, Ruminants Grazing, deferred Probably much higher Jahnke (1982)Sub-SaharanVery arid toLGA, LGH,RuminantsGrazing, deferredProbably much higher  Africa, 1970s  humid LGT, MRA,  grazing plus fertilizer  Africa, 1970shumidLGT, MRA,grazing plus fertilizer     MRH MRT    MRH MRT Sandford (1983a) Sub-Saharan More on East Mainly arid to LGA, LGH, LGT Cattle, camels, Variable across Unknown Sandford (1983a) Sub-SaharanMore on EastMainly arid toLGA, LGH, LGT Cattle, camels,Variable acrossUnknown  Africa, Africa semi-arid  goats production  Africa,Africasemi-aridgoatsproduction  1960s-1970s     subsystems  1960s-1970ssubsystems de Leeuw and Sub-Saharan Many Arid, semi-arid LGA, LGT, LGH Ruminants and Not specific Unknown de Leeuw andSub-SaharanManyArid, semi-arid LGA, LGT, LGH Ruminants andNot specificUnknown Milligan (1984) Africa, 1984    wildlife   Milligan (1984)Africa, 1984wildlife von Kaufmann 1978-1986 Kaduna, Nigeria Semi-arid to MRH Cattle, sheep, Crop residues, pasture Mixed (highest for S. von Kaufmann1978-1986Kaduna, NigeriaSemi-arid toMRHCattle, sheep,Crop residues, pastureMixed (highest for S. et al. (1986)   subhumid  goats with S. hamata hamata) et al. (1986)subhumidgoatswith S. hamatahamata) Wilson (1986) West Africa, Mali Arid, semi-arid LGA, LGH, Ruminants Non-specific fodder and Unknown Wilson (1986)West Africa,MaliArid, semi-arid LGA, LGH,RuminantsNon-specific fodder andUnknown  1980s   MRA, MRH,  forage crops,  1980sMRA, MRH,forage crops,     MIA, MIH  including browse  MIA, MIHincluding browse Ellis and Swift East Africa, Northern Kenya Arid, semi-arid, LGA Cattle, camels, Not specific Unknown Ellis and SwiftEast Africa,Northern KenyaArid, semi-arid,LGACattle, camels,Not specificUnknown (1988) 1980s  subhumid  goats, sheep   (1988)1980ssubhumidgoats, sheep Solomon Bekure East Africa, Kajiado County, Semi-arid to LGA, LGH, Cattle Feed gardens with Mixed (highest for Solomon BekureEast Africa,Kajiado County,Semi-arid toLGA, LGH,CattleFeed gardens withMixed (highest for et al. (1991) 1980-1991 Kenya subhumid limited MRA,  Panicum maximum Panicum, less for et al. (1991)1980-1991Kenyasubhumidlimited MRA,Panicum maximumPanicum, less for   (Maasailand)  MRH  or Pennisetum browse) (Maasailand)MRHor Pennisetumbrowse)       purpurem, cowpea,  purpurem, cowpea,       pigeonpea or  pigeonpea or       Leucaena  Leucaena Homewood and East Africa Kenya/ Semi-arid to LGA, LGH, Ruminants and Natural vegetation Unknown Homewood andEast AfricaKenya/Semi-arid toLGA, LGH,Ruminants andNatural vegetationUnknown Rodgers  Maasailand, subhumid MRA wildlife (pasture, browse)  RodgersMaasailand,subhumidMRAwildlife(pasture, browse) (2004)  Tanzania/      (2004)Tanzania/   Serengeti      Serengeti Behnke et al. East Africa, Ethiopia, Kenya, Several LGA, LGH, Ruminants and Natural vegetation Unknown Behnke et al.East Africa,Ethiopia, Kenya,SeveralLGA, LGH,Ruminants andNatural vegetationUnknown (1993) 1970s-1980s Tanzania  MRA wildlife (pasture, browse)  (1993)1970s-1980sTanzaniaMRAwildlife(pasture, browse)        Continued Continued  "},{"text":"Table 15 .   "},{"text":" gave average yield increases of 41% for nine crops and 38% for 21 crops at varying periods from 1970 to 2000. fertilizer use. Fertilizer use in the initial ILCA studies of mixed farming -Debre Zeit and Debre Berhan in highland Ethiopia, Kaduna in subhumid central Nigeria, Ibadan in southwest Nigeria, and Niono and the Inner Delta of the Niger in semi-arid Mali -probably did not exceed 10 kg/ ha at any site before 1990. Input use grew as average cultivated land per person fell. While fertilizer-use data are unreliable before 2000, it was clearly very low across sub-Saharan Africa before this century. Current estimates are less than 15 kg/ha in West Africa and less than 20 kg/ha in East and southern Africa.  "},{"text":"Table 15 . 4. Mixed crop-livestock systems studies, various years.   Country/  Farming Scale of      Country/FarmingScale of Study Region/period subregion Climate systems research Animals Main crops Biotic factors Abiotic factors Technologies StudyRegion/periodsubregionClimatesystemsresearchAnimalsMain cropsBiotic factorsAbiotic factors Technologies Von Kaufmann West Africa, Nigeria, Arid to MRH Household, Cattle Pearl millet, Soil, water, Rainfall, Subsistence Von KaufmannWest Africa,Nigeria,Arid toMRHHousehold,CattlePearl millet,Soil, water,Rainfall,Subsistence et al. 1979-1986 Kaduna semi-arid  territory  sorghum, health, feed temperature dairy et al.1979-1986Kadunasemi-aridterritorysorghum,health, feedtemperaturedairy (1986)       maize, cassava    (1986)maize, cassava Wilson West Africa, Mali Arid to MRA, MIA Household, Cattle, Pearl millet, Soil, water,   WilsonWest Africa,MaliArid toMRA, MIAHousehold,Cattle,Pearl millet,Soil, water, (1986) 1978-1986  semi-arid  territory camels, cowpea health, feed   (1986)1978-1986semi-aridterritorycamels,cowpeahealth, feed       sheep,     sheep,       goats     goats Powell Sub-Saharan West Africa Arid to MRA, MIA Household, Ruminants Pearl millet, Soil, water, Rainfall, Animal traction PowellSub-SaharanWest Africa Arid toMRA, MIAHousehold,Ruminants Pearl millet,Soil, water,Rainfall,Animal traction (1986a,b) Africa,  semi-arid  territory  sorghum, microbes temperature  (1986a,b)Africa,semi-aridterritorysorghum,microbestemperature  1980s-1980s      maize, cassava    1980s-1980smaize, cassava     MRH    (Primary Water, fire (?) Mixed farming MRH(PrimaryWater, fire (?)Mixed farming         studies)  without animal studies)without animal           traction traction         (Primary  Manuring crops (PrimaryManuring crops         studies)   studies)         Nutrient cycle Altitude Animal fattening Nutrient cycle AltitudeAnimal fattening          (Primary Grazing reserves (PrimaryGrazing reserves          studies)  studies) Pingali et al. Sub-Saharan Many Semi-arid MRA, MIA Sub- Ruminants Maize, cotton, pearl Grazing, feed  Animal traction Pingali et al.Sub-SaharanManySemi-aridMRA, MIASub-Ruminants Maize, cotton, pearlGrazing, feedAnimal traction (1987) Africa    Saharan  millet, sorghum,    (1987)AfricaSaharanmillet, sorghum,      Africa  paddy, cassava,    Africapaddy, cassava,        cowpea    cowpea    Subhumid MRH      Tractor SubhumidMRHTractor           mechanization mechanization    Highlands MRT      Mineral fertilizer HighlandsMRTMineral fertilizer    Humid MRH       HumidMRH Gryseels East Africa, Ethiopia Highlands MRH, MRT Territory Draft oxen, Wheat, barley, teff, Grazing, feed Rainfall, Dairy GryseelsEast Africa,EthiopiaHighlandsMRH, MRT TerritoryDraft oxen,Wheat, barley, teff,Grazing, feed Rainfall,Dairy (1988); 1970s early     cattle, faba  temperature,  (1988);1970s earlycattle,fabatemperature, Gryseels 1980s     sheep   frost risk  Gryseels1980ssheepfrost risk and           and Anderson           Anderson (1983);           (1983); Getachew           Getachew Asamenew           Asamenew et al. (1993)           et al. (1993)  "},{"text":"Table 15 .4. Continued   "}],"sieverID":"06f50695-feba-4e28-ab72-a4f53e2089c5","abstract":"Many models were inapplicable for policy advice. Using models, quantitative or not, failed to raise productivity. Many models were developed and later not used at all, even for scientific purposes, or were never used for investment analysis or policy simulations; some of the models were poorly documented and their results could not be replicated."}
\ No newline at end of file