Rice-based production systems for food security and poverty alleviation in sub-Saharan Africa

T. Defoer,^[17] M.C.S. Wopereis, M.P. Jones, F. Lançon, O. Erenstein and R.G. Guei¹
WARDA, Abidjan, Côte d'Ivoire

INTRODUCTION

The lack of food security for a large proportion of the African population continues to exacerbate poverty and malnutrition. High population growth, the effects of HIV (human immunodeficiency virus) on the productive labour force, the degradation of the environment, poor agricultural development support services and lack of enabling economic policy environment have all aggravated the situation. Rice has great potential and can play a critical role in contributing to food and nutritional security, income generation, poverty alleviation and socio-economic growth in Africa. It is an important food crop in many African countries and is increasingly preferred over many traditional foods, such as sorghum, millet and most root and tuber crops. It is the staple food crop in Côte d'Ivoire, the Gambia, Guinea, Guinea-Bissau, Liberia, Madagascar, Mauritania, Senegal and Sierra Leone. In most countries, rice supply cannot keep up with demand. Consumption demand has grown rapidly over the last two decades and is now more than 6 percent per annum, amounting to over 10 million tonnes of milled rice per year (Figures 1 and 2). In West Africa alone, FAO projected that rice imports would rise to 4 million tonnes per year by 2000, drawing approximately US$1 billion from foreign exchange earnings. This increase is due to both population growth (2.6 percent per year) and the increasing share of rice in the diet of African populations (1.1 percent per year); rice consumption in 1998 was 30 kg per caput per year (FAO, 1999),mainly as aresultof rapid urbanization (Snrech, 1994). Urban rice consumers, faced with a relative increase in the rice price, prefer to maintain their consumption level (at the expense of other categories of goods), rather than shifting to other cereals. This is most probably due to the ease of preparation and the difference in time perception between urban and rural families. The vast majority of rice in Africa is rainfed and grown by smallholder farmers, a disproportionate number of whom are women. Growth in demand is creating opportunities for small-scale producers.

The production of rice in sub-Saharan Africa has steadily increased since the 1970s, reaching almost 7 million tonnes of milled rice by the end of the last decade. The increase in rice production is due to: expansion in area (70 percent) and yield increase (30 percent) (Fagade, 2000; Falusi, 1997). The gap between rice demand and regional supply is increasing and was about 4 million tonnes of milled rice for sub-Saharan Africa as a whole in 1998 (Figure 2). Nigeria was the major rice importer with almost 1 million tonnes in 1999/2000 (Mbabaali, 2000).

The social interest in developing regional rice production goes further than import substitution. As mainly resource-poor farmers grow rice integrating a wide range of other agricultural activities, rice research and development can be considered an entry point for the development of the agricultural sector as a whole. Thus, rice research and development can be seen as a catalyst, reducing production risks, averting natural resource degradation, enhancing food security and income and contributing to poverty alleviation.

The objective of this paper is to provide an overview of the challenges and technical opportunities in developing rice-based systems for food security and poverty alleviation in sub-Saharan Africa. The data presented relate mainly to West and Central Africa.

RICE ECOLOGIES IN WEST AND CENTRAL AFRICA

The potential for rice development in West and Central Africa is largely determined by the agro-ecological conditions in which rice can be produced. Rice is characterized by its plasticity which allows it to grow in almost any biophysical environment in West and Central Africa. Rice is grown in a whole range of agro-ecological zones from the humid forest to the Sahel (Figure 3). The total rice area in West and Central Africa amounts to more than 4 million ha (FAO, 1999). Within these regional agro-ecological zones, five main rice-based systems can be distinguished with respect to water supply and topography in sub-Saharan Africa (Windmeijer, Duivenbooden and Andriesse, 1994):

• Rainfed upland rice on plateaus and slopes;
• Lowland rainfed rice in valley bottoms and flood-plains with varying degrees of water control;
• Irrigated rice with relatively good water control in deltas and floodplains;
• Deep-water, floating rice along river beds and banks; and
• Mangrove swamp rice in lagoons and deltas in coastal areas.

FIGURE 1
Per caput rice consumption in West and sub-Saharan Africa

Source: FAO, 1999.

FIGURE 2
Milled rice production and imports in West Africa compared to the rest of sub-Saharan Africa (SSA)

Source: FAO, 1999.

The three main rice ecologies across West and Central Africa are the rainfed uplands, the rainfed lowlands and the irrigated systems (Table 1). These ecologies can be found across agro-ecological zones. A distinction is made between irrigated systems in the desert margins of the Sahel and those in the savannah or humid forest zones. The upland rainfed rice-based systems cover the largest area (44 percent), mainly in coastal areas in the humid and subhumid agro-ecological zone. The rainfed lowland systems are the second most important in terms of surface area, accounting for 31 percent of the total rice cultivated area; third are the irrigated rice-based systems (12 percent). Deep-water and mangrove rice systems are relatively unimportant in terms of surface area.

Using average rice yields of 1 tonne/ha in the uplands, 2 tonnes/ha in the lowlands, 3 tonnes/ha in irrigated areas in the savannah/humid zone and 4.5 tonnes/ha in irrigated areas in the Sahel, it is possible to make estimates of production figures (Table 2). Rainfed rice production systems are still predominant, providing more than half the total production. Rainfed lowland systems provide 36 percent of the production, followed by rainfed upland (25 percent). The irrigated systems account for 28 percent of production, with 22 percent in the Sahelian zone. Irrigated systems in the savannah and humid forest zone are rather minimal in terms of rice production.

The major rice ecologies present various communalities, such as weed and pest pressure and soil fertility decline. In addition, interlinkages exist between ecologies (e.g. water or nutrient flow from upland to lowland), influencing the ecological sustainability of farmland. In response to this challenge, The Africa Rice Center (WARDA) has developed the concept of the uplandlowland continuum along a toposequence, based on watertable depth (WARDA, 1989). Furthermore, inter-linkages can blur the borderline between ecologies, like the hydromorphic fringe between the upland and lowland along the continuum. Another fuzzy transition exists between rainfed and irrigated lowland. A water-management continuum, ranging from strict rainfed to fully irrigated lowland, can be distinguished, and it may evolve depending on investments in water control measures (Figure 4).

FIGURE 3
Agro-ecological zones in West and Central Africa

TABLE 1
Share of rice ecologies in rice planted areas by country

Source: WARDA, 1997 and also 1996 data provided by FAO.

TABLE 2
Estimated share (%) of rice production by ecologies and by country

Note: Average rice yields used for % calculation: upland:
1 tonne/ha; lowland: 2 tonnes/ha; irrigated, savannah/humid zone: 3 tonnes/ha; irrigated, Sahel zone: 4.5 tonnes/ha.

Source: WARDA, 1997 and also 1996 data provided by FAO.

CONSTRAINTS, OPPORTUNITIES AND CHALLENGES

Lançon and Erenstein (2002) calculated that with an annual per caput rice consumption growth of 1 percent and population growth of 2.5 percent, total rice consumption in West Africa would reach 10 million tonnes in 2010 and 15 million tonnes in 2020. There are three major options for increasing rice production:

• area expansion;
• increase in cropping intensity (number of crops per year from the same area); and
• yield increase (produce per unit area).

The opportunities for rice production development depend to a large extent on the biophysical and socio-economic environments. The challenges for area expansion, increase in cropping intensity and yield increase vary widely by ecology. The following sections highlight farmers' major biotic and abiotic constraints at plot level, and institutional/organizational constraints at farm, community and regional levels with specific attention to the divergences between the major rice ecologies. Combined with specific ecological opportunities, these constraints determine the major challenges for rice research and development.

Upland ecology

Upland rice is to a large extent produced by subsistence-oriented farm households representing only part of the total cropped area and which do not use external inputs, mainly due to high production risk and poverty. Rice yields in upland systems average about 1 tonne/ha; however, these figures hide large differences between farms. Within a given region or village, differences in cropped area and yields between farms may vary tenfold, partly because of differences in the quality of the land, but they are also as a result of differences in management practices (time of sowing, weed control etc.).

FIGURE 4
The upland-lowland continuum

Weed competition is indeed one of the most important yield-reducing factors (Johnson, 1997), followed by drought, blast, soil acidity and low soil fertility. Farmers traditionally manage these stresses through long fallow periods. Population growth and resulting pressure on the land has led to increasing reduction of fallow periods and extension of cropped areas, often towards the more fragile upper parts of the slopes. In the rainfed uplands, slash-and-burn agriculture and reduced fallow periods have aggravated weed pressure and general decline in land quality through soil erosion (Oldeman and Hakkeling, 1990) and soil nutrient depletion, known as "soil mining" (Van der Pol, 1992). The increase in population pressure aggravates the situation, resulting in low and unstable rice yields. Lack of capital limits the use of external resources and intensification of the system. Weed competition further reduces labour productivity and increases the risk of crop failure. Farmers traditionally use long-duration rice cultivars that further undermine the fragility of the system and limit the cropping intensity. Declining productivity and incomes feed the cycle of poverty and environmental destruction (Cleaver, 1993; Cleaver and Schreiber, 1994).

The major challenges for rice research and development in the upland ecology relate to sustainable stabilization and intensification of upland rice production; there is good opportunity for increasing yields and possibly the cropping intensity. Improved varieties with high-yielding capacity under low-resource input conditions are an underexploited potential. However, low-cost complementary soil fertility management practices will have to be introduced, for example, crop rotation with legumes to maintain or improve soil quality, avoid soil nutrient mining and enhance sustainability. With short-duration varieties, it might be possible to introduce rice-legume cropping systems within the same cropping season if rainfall is adequate. Given the complexity and diversity of the upland rice systems, technologies will have to be adapted and finetuned in situ. This will require new approaches for farmers to play more important roles in technology development and for effective interaction between researchers and farmers to develop appropriate technologies.

FIGURE 5
Major rice production constraints

Technical opportunities

Given the fragility of the upland rice system, technologies are urgently needed to reduce degradation of the fragile resource base and ensure sustainable stabilization of the system.

WARDA's breakthrough in developing New Rice for Africa (NERICA) based on crosses between African rice (O. glaberrima) and Asian rice (O. sativa) provides an exciting opportunity for farmers to stabilize and intensify low-input upland systems. NERICAs tend to have better resistance to most African stresses including weeds and drought. They have high-yielding potential and generally outyield local varieties, under both low- and high-input conditions (Jones and Wopereis-Pura, 2001). Moreover, compared to local varieties, NERICAs generally have a much shorter growing cycle (90 to 110 days), providing the opportunity to produce food during the hungry season. Moreover, short-duration NERICAs allow rice farmers to adjust their agricultural calendar to climatic variation (Jones and Wopereis-Pura, 2001). The early-maturing NERICAs have a comparative advantage over local varieties with respect to demand for labour. Labour-saving technologies can motivate farmers to concentrate agricultural activities in a more limited area, allowing the most fragile areas to revert back to natural vegetation. Resource degradation will be reduced, yields and income will be stabilized and labour productivity and income will be increased. NERICAs may also help stop expansion of cultivated land areas, because double-cropping of NERICAs is possible under sufficient rainfall.

The idea is, however, not to promote the replacement of local varieties by NERICAs. Rather, WARDA's strategy is to integrate NERICAs into the existing varietal portfolio of farmers. Indeed, in 1997 WARDAsetup a participatory variety selection (PVS) approach, with farmers selecting varieties from rice gardens with large numbers of local and modern O. sativa and NERICA varieties (Figure 5). The PVS-approach is a 3-year programme allowing farmers to test and evaluate their own selected varieties under their site-specific conditions and according to their needs. The most promising varieties are then proposed to the national release committee (Table 3). This approach is an example of how farmers can be effectively involved early in the research and development process of new technologies. A study conducted by Diagne et al. (2001) shows that the PVS approach leads to increased biodiversity, as the number of varieties grown per farmer can be increased.

Irrigated ecology

Irrigation systems include dam-based irrigation, water diversion from rivers and pump irrigation from surface water or tubewells. Major differences in constraints, opportunities and challenges exist between irrigated rice ecologies in the Sahel and in the humid forest and savannah zones.

TABLE 3
Non-NERICA WARDA-bred upland rice varieties released in West and Central Africa

FIGURE 6
NERICA varieties released or in pipeline for release in Africa

Average farmers' yields in the Sahel are around 4 to 5 tonnes/ha per season, with potential yields varying from 6 to 11 tonnes/ha per season, limited by solar radiation and temperature only. The potential yield gains from improved crop and resource management are, therefore, tremendous. In the Office du Niger in Mali, average yields have increased over the last 15 years from 2 to almost 6 tonnes/ha. Only 10 percent of the area under irrigation is double-cropped, because of extreme temperatures in both hot and wet seasons. The relatively demanding cropping calendar leaves little room for delays in activities, and mechanization is relatively widespread. However, labour remains a limiting factor and about half of the rice grown in the Sahel is direct-seeded. African rice gall midge (AfrRGM), rice yellow mottle virus (RYMV) and blast are the major pests found in the irrigated rice ecosystems.

Moving south into the savannah and humid forest zones, schemes become smaller, they are mainly transplanted and located along major roads and near urban centres. Compared to the Sahel, rice productivity in the savannah and humid forest systems is less constrained by temperature extremes, but potential yields are lower, ranging from 5 to 8 tonnes/ha per season, as a result of lower solar radiation levels. Actual yields are around 3 tonnes/ha per season, indicating considerable scope for yield gains.

Still, substantial investments have already been made in irrigation infrastructure, especially in the Sahel. The development, maintenance and rehabilitation of the existing infrastructure provide concrete opportunities to capitalize on these earlier investments. This is adequately illustrated by the development of rice production in Mali over the last few years and the substantial yield increases achieved in the Sahel (Mali, Senegal) - demonstrating that irrigated rice is a feasible option in the subregion.

In the Sahel, soil alkalinization followed by sodication may affect soil quality, particularly if groundwater tables rise too close to the surface due to lack of drainage combined with irrigation water rich in dissolved inorganic ions. Soil alkalinization and sodication problems are found in the Office du Niger, Mali, in Foum Gleita, Mauritania, and in irrigated systems in Niger.

Until now, rice-rice or other double-cropping schemes are found only in a few zones (e.g. in the Kou Valley in Burkina Faso or in Niger). There is a clear potential for the introduction of short-duration cultivars that may allow to grow two rice crops per year on the same land, or an additional non-rice crop following rice, profiting from residual moisture in the soil or supplementary irrigation.

Technical opportunities

Compared to the upland rice ecosystems, the irrigated rice systems are quite robust and homogeneous in terms of both biophysical and socio-economic characteristics. Innovation and change should concentrate on improving resource-use efficiency and factor productivity.

WARDA and partners have developed improved integrated rice management (IRM) options for irrigated rice cropping in the Sahel that are within farmers' means, based on farm surveys and farmer participatory on- and off-station research (Wopereis et al., 2001; WARDA-SAED, 2000). IRM focuses on land preparation, crop establishment technique, sowing dates and rates and cultivar choice, and provides a farming calendar for best-bet management for a given site x sowing date x cultivar choice x crop establishment technique combination; relevant information can be adapted to a specific site and to farmers' means. IRM options further include: fertilizer rates for specific target yields and taking into account farmers' financial means; weed and water management practices; and harvest and post-harvest techniques.

These IRM options have been evaluated with farmers in Senegal and Mauritania (Häfele et al., 2000), resulting in a mean yield increase of 1.7 tonnes/ha (from 3.8 to 5.5 tonnes/ha). Partial budgeting showed that average net benefit increased from US$215 to US$525 per ha, i.e. an 85 percent increase in net benefits. IRM production costs were slightly higher than farmer production costs, mainly due to basal fertilizer application. Even with this assumption, IRM was economically superior to farmer practices across sites. IRM options most attractive to farmers included improved fertilizer and weed management, use of improved varieties and harvesting and post-harvest technologies. WARDA, NARES (National Agricultural Research and Extension Systems) and NGOs (non-governmental organizations) from Mauritania and Senegal, are now exploring ways to scale up results from these studies to a much larger number of farmers.

Moreover, only a few introduced varieties were grown by farmers in irrigated ecologies of West and Central Africa. WARDA, through inter- and intraspecific crosses, has increased the genetic diversity, and a number of high-yielding, good quality, shorter-duration and salinity-tolerant varieties have been developed and grown by farmers (Table 4).

Rainfed lowland ecology

In West and Central Africa alone, an estimated 20 to 40 million ha of inland valley swamps are found, of which only 10 to 25 percent are currently used (Windmeijer, Duivenbooden andAndriesse, 1994). Although WARDA research has shown that no causal linkages exist between water-borne diseases, such as malaria, and the expansion of irrigation in West and Central Africa (WARDA, 1999), farmers still perceive human health problems as a major constraint for the development of lowland areas.

TABLE 4
New irrigated rice varieties released in West and Central Africa

Rainfed lowland systems are more robust than upland systems and have good potential for intensification, but are largely unexploited. Improved water control is definitely a first step towards improving the productivity of the lowlands and controlling iron toxicity. With improved water control, the use of external inputs may become attractive and rice yields may be increased rapidly. As full water control in large schemes is too costly, improvement of water control will have to be limited to smaller schemes.

The rice yields in rainfed lowlands are substantially higher than those in the rainfed uplands, but nevertheless low, averaging 2 tonnes/ha. The potential yield is 3 tonnes/ha at current input levels and 5 to 6 tonnes/ha at increased input levels and with improved water control (i.e. yield levels that are comparable to irrigated systems with full water control in the savannah and humid forest zones). These yield gaps indicate considerable potential for improvement. Poor water control is a major system constraint, prohibiting more intensive use of these systems. Indeed, with improved water control, use of external inputs may become attractive, potentially resulting in higher yields. Although water management is akey factor for intensification, complementary technologies will be needed to provide sufficient opportunities for increased productivity and profitability. Indeed, major biological constraints for rice production in inland valleys, such as iron toxicity, pests and weeds, require major attention to realize those potentials.

Technical opportunities

Important spillover effects from technologies developed for irrigated and upland systems are expected. The NERICAs currently available have been developed for upland systems, but are now being screened under lowland conditions. Interspecific progenies of O. glaberrima and O. sativa subsp. indica that were bred for irrigated conditions are also evaluated under lowland rainfed conditions. However, in the near future, water control will not reach the level of the irrigated systems and breeding for lowland conditions will have to take into account multiple stresses such as drought, low N and P conditions and major pests such as RYMV, AfRGM and blast.

In addition to varietal improvement, the sustainable development of rainfed lowlands will require complementary technologies, covering all aspects of lowland rice management, from land preparation to harvest and post-harvest technologies. The basis will be formed by IRM practices developed for irrigated rice systems, but fine-tuning and adaptation according to the site-specific conditions will definitely be required. Besides intensification, the potential for diversification will be explored. Particular emphasis will be given to the integration of vegetables in rice production systems, and possibilities to integrate aquaculture and livestock in inland-valley lowlands.

Sustainable intensification and diversification of lowlands require important investments. Therefore, particular attention will be given to the identification of bottlenecks preventing access to capital and other necessary resources and their adequate management. Market forces drive both diversification and intensification; therefore, a particular focus on peri-urban environments and the efficiency of market linkages is warranted. The capacity of locally produced rice to compete with imported rice should be explored, with respect to the efficiency (cost-effectiveness and rice quality) of small-scale processing units.

Inland valleys have very complex, dynamic and diverse human, social, natural and physical dimensions and inter-connections that need to be understood in order to determine options for improved and integrated crop and natural resources management. There is a need to:

• elucidate the different functions of inland valleys and their management;

• unlock constraints to the more widespread use of inland valleys for rice-based systems;

• identify, develop and adapt a wide variety of options and methods for sustainable inland valley development and management; and

• scale up results.

It is clear that with such a high degree of complexity and diversity, research and extension will never arrive at tailor-made recommendations to individually suit the numerous rice growers in the inland valleys. Agricultural research and extension has to provide much more specific information and move beyond simple delivery of general messages, recipes or blanket recommendations to be passed to farmers. Sustainable management of inland valleys requires a fundamental change in innovation, development and learning processes. Indeed, what is needed are approaches that strengthen farmers' capacity to make optimal use of the available resources and the best choices of alternative approaches to resource management. This has important implications in terms of intervention methodology and institutions that support change and development, including farmer organizations. Indeed the inherent complexity, diversity and dynamics of inland valley ecosystems call for a bottom-up, social learning process. Only by doing so, can a sustainable and lasting impact on food security be achieved in the region. A participatory learning and action research approach among inland valley development stakeholders (farmers, change agents, extension, research) at grassroots level is required. This will help build bridges between local and indigenous knowledge and scientific expertise. This will ultimately lead to a network of farmer associations in contact with a wide variety of external stakeholders, forming an inland valley platform at regional and national level. It is also important to develop an integrated natural resources management (INRM) framework and curriculum for farmer learning for inland valleys in West Africa.

CONCLUSIONS

Thanks to the development of NERICAs, it has been possible to break the yield barrier in the upland rice ecology. NERICAs are stress resistant, have short growth duration and respond well to both low and high input conditions. Participatory varietal selection (both researcher- and extensionled) and community-based seed supply systems are currently being used to get seeds to farmers. This development will ensure that the contribution of the upland ecology to regional rice production will increase in the years to come. NERICAs are also currently being developed for rainfed and irrigated lowland systems.

Considerable scope for yield improvement also exists in irrigated ecologies. The introduction of integrated rice management (IRM) in irrigated systems in the Sahel has resulted in considerable yield increases (2 tonnes/ha) without major changes in input use, thereby reducing the very large yield gap between actual and potential yields in these systems. IRM is developed with farmers and provides them with options ranging from land preparation to harvest and post-harvest interventions. Existing farmer organizations can often be used as dissemination channels for IRM.

In the long term, the rainfed lowlands also show great promise. Large areas of rainfed lowland have yet to be developed. Considerable scope exists to increase cropping intensity and yields in these systems because of sub-optimal crop management and poor water control. This again calls for an IRM approach. However, these systems are characterized by a much larger complexity and diversity than the irrigated systems. The message that needs conveying is, therefore, also much more complex and will take more time and effort. It is proposed to introduce IRM through a participatory learning and action research (PLAR) approach and the creation of rural knowledge centres. Pilot farmers in rural knowledge centres should ultimately become agents of innovation and change, stimulating the diffusion of knowledge. Research and extension staff should change in the process to assume a much more facilitating role, building bridges between local knowledge and expertise from outside.

REFERENCES

Cleaver, K.M. 1993. A strategy to develop agriculture in sub-Saharan Africa and a focus for the World Bank. WB technical paper No 203. Washington, DC, USA, WB.

Cleaver, K.M. & Schreiber, G.A. 1994. Reversing the spiral: the population, agriculture and environment nexus in sub-Saharan Africa. Washington, DC, USA, WB.

Diagne, A, Kouamé, K.A, Abi, M.I. & Guei, R. 2001. Impact of modern varieties on rice biodiversity. Internal report. Bouake, Côte d'Ivoire, WARD A. 31 pp.

Fagade, S.O. 2000. Yield gaps and productivity decline in rice production in Nigeria. Paper presented at the Expert Consultation on Yield Gap and Productivity Decline in Rice, FAO, Rome, Italy, 5-7 Sept. 2000.

Falusi, A.O. 1997. Agricultural development and food production in Nigeria: problems and prospects. In B. Shaib, N.O. Adedipe, M. Aliyu & M. Jir, eds. Integrated agricultural production in Nigeria: Strategies and mechanisms, p. 151-170. NARP Monograph No. 5.

FAO. 1999. FAOSTAT database (available at www.fao.org).

Häfele, S., Johnson, D.E., Diallo, S., Wopereis, M.C.S. & Janin, I. 2000. Improved soil fertility and weed management is profitable for irrigated rice farmers in Sahelian Africa. Field Crops Res., 66: 101-113.

Johnson, D.E. 1997. Weeds of rice in West Africa. Bouaké, Côte d'Ivoire, WARD A.

Jones, M. & Wopereis-Pura, M. 2001. History of Nerica and PVS. Internal report. Bouaké, Côte d'Ivoire, WARD A.

Lançon, F. & Erenstein, O. 2002. Potential and prospects for rice production in West Africa. Paper presented at Sub-Regional Workshop on Harmonization of Policies and Co-ordination of Programmes on Rice in the ECOWAS Sub-Region, Accra, Ghana, 25-28 Feb. 2002.

Mbabaali, S. 2000. International rice trade: a review of 1999 and prospects for 2000. IRCNewsl., 49: 1-5.

Oldeman, L.R. & Hakkeling, R.T.A. 1990. World map of the status of human-induced soil degradation: An explanatory note. Nairobi, Kenya, UNEP.

Snrech, S. 1994. Pour préparer l'avenir de l'Afrique de l 'Ouest: une vision à l 'horizon 2020. Synthèse de l'étude. SAH/(94)439. Cellule CINERGIE. Paris, France, Club de Sahel.

Van der Pol, F. 1992. Soil mining: An unseen contributor to farm income in southern Mali. KIT Bulletin No 325. Amsterdam, Netherlands, Royal Tropical Institute (KIT).

WARDA. 1999.1998 Annual report. Bouaké, Côte d'Ivoire, WARDA.

WARDA. 1997. WARDA Taskforce estimate. Bouaké, Côte d'Ivoire, WARDA.

WARDA. 1989.1988 Annual report. Bouaké, Côte d'Ivoire, WARDA.

WARDA-SAED (Société Nationale d'Aménagement et d'Exploitation des Terres du Delta du Fleuve Sénégal et de la Vallée du Fleuve Sénégal et de Falémé). 2000. Manuel pratique pour la riziculture irriguée dans la Vallée du Fleuve Sénégal. St Louis, Senegal, WARDA. 100 pp + annexes.

Windmeijer, P.N., Duivenbooden, N.V. & Andriesse, W. (eds). 1994. Characterization of rice-growing agro-ecosystems in West Africa: semi-detailed characterization of inland valleys in Côte d'Ivoire. Wageningen, Netherlands, SC-DLO- Wageningen Agricultural University.

Wopereis, M.C.S., Haefele, S.M., Kebbeh, M., Miézan, K. & Diack, B.S. 2001. Improving the productivity and profitability of irrigated rice production in Sahelian West Africa. In Yield gap and productivity decline in rice production, Proc. of the Expert Consultation held in Rome, 5-7 Sept. 2000. Rome, Italy, FAO.