Article prepared for FAO by Jacob Kijne, Consultant.

[K-1] In 2000, 21 countries are expected to have scarce water resources, defined as 1000 m3 of available water per capita per year or less. Twelve of these countries are in Africa and nine in Sub-Saharan Africa (World Bank, 1997). Although food production world-wide is expected to grow fast enough to ensure that world prices continue to fall, predictions are for a worsening in food security of Sub-Saharan Africa (SSA). Food production has been lagging behind population growth, and the self-sufficiency ratio of cereals has declined since 1970 by about one percent per year. Cereal imports are projected to triple, from 9 million metric tons in 1990 to 29 million tons in 2020 (Rosegrant and Perez, 1995). Countries in SSA will not be able to pay for these growing imports. Of even greater concern is the projected increase of some 13 million in the number of malnourished children in SSA by 2020.

[K-2] Total irrigated land in Africa is estimated to be about 12.2 million ha (FAO, 1996). This figure includes all land where water is supplied for the purpose of crop production, excluding only areas where water harvesting and spate irrigation are practiced. Irrigated land represents, on average, less than 8 percent of the arable land, with large differences between countries. Irrigated land percentages are highest in the Northern region (99% in Egypt) and lowest in the Central region (0.2% in the Democratic Republic of the Congo). The average for SSA is less than 4%. These figures cover a wide range of water management situations, including the productive wetlands of the Gulf of Guinea and the highlands of Central Africa, and modern full-control irrigation systems in Zimbabwe. Informal irrigation is estimated to cover between 35% and 50% of the total irrigated land in Africa. Yields of irrigated land are about 2.2 times higher than from rainfed land (FAO, 1996).

[K-3] Vulnerability to drought varies from country to country, depending - among others - on the stage of development. Particularly vulnerable are economies in the early stages of transition from subsistence farming to a more modern and productive farm economy with a strong urban consumer base. The African continent has a centuries-long history of rainfall fluctuations with droughts of varying lengths and intensities, which continues to the present. The Sahel, the Horn of Africa and the countries around the Kalahari Desert are characterized by high inter-annual and intra-seasonal rainfall variability. Good and bad years do not occur singly or at random but tend to be grouped. This has important implications for food security as food and water may need to be stored over a period of several poor years.

[K-4] For Africa prosperous agriculture is the engine without which poverty cannot be reduced, resources cannot be managed sustainably, and food security cannot be assured (World Bank, 1997). Very few low-income countries have achieved rapid non-agricultural growth without corresponding rapid agricultural growth; in countries with large, poor rural populations, agriculture is the only source of the increase in rural farm and non-farm incomes needed to enable the rural poor to afford more and better food. The key role of agriculture in Africa's economic life is apparent -- agriculture accounts for 35% of the continent's GNP, 40% of its exports and 70% of its employment. Today about 70% of Africa's poor live in rural areas and depend for their livelihoods on agriculture and agribusiness.

[K-5] Rural people play a critical role in protecting the environment. Agriculture is the world's biggest user of land and water resources and agriculture both contributes to environmental degradation and suffers as a result of it. With ever-decreasing potential to expand the agricultural area in many countries of SSA, increases in agricultural production will have to come from more intensive use of land, water and other agricultural resources. However, as land use intensifies, fallow periods decline, and cultivation spreads into marginal and ecologically fragile lands. In the absence of appropriate resource management technologies, intensification inevitably leads to degradation of the resource base with important implications for soil productivity, household food security and rural poverty. There are two types of intensification; that which results from population pressure having a very different effect on farmer livelihoods, cropping patterns and resource productivity than intensification that results from market demand. Farmers intensifying because of population pressure become poorer, affecting the potential for technological change (Freeman, 1994).

[K-6] Africa's ability to earn from exports to buy food on the world market has not improved in recent years. Negative effects of macro-economic policies are often compounded by sector-specific pricing policies, such as border taxes on agricultural exports, price controls, and the gap between border prices and those at the farm gate, created by monopolistic government parastatals and state marketing boards. Removal of input subsidies and exchange rate liberalization in Ghana had a negative impact on the competitiveness of cereal production. To again increase cereal productivity in Ghana would require re-introduction of some type of selective subsidy that is both input- and crop-specific (Asuming-Brempong, 1994). National macro-economic reform can only succeed if there is a response from the world market. How can a country be asked to move from food self-sufficiency to food security if access to the world market is not assured? Computer models estimating the income effects of the Uruguay round of trade negotiations predicted positive effects for most regions. Only for Africa some of the computer models predicted negative effects (FAO,1996). This appears to be born out by recent evidence.

[K-7] Reallocation of water from agriculture to urban water supplies will occur in Africa as it has in other continents. The question is whether it can be accomplished in a rational manner which keeps costs to a minimum, or in the ad hoc manner governing most reallocations today (Rosegrant and Perez, 1995). Inter-sectoral reallocation of water can be effected either through government-imposed supply management or by demand management, which uses incentives to induce water to move among competing demands. Since in most African countries diversions for agricultural use account for more than 80% of the total, relatively small transfers of water from agriculture could meet growing urban and industrial demands. For example, in Morocco, a 5% transfer of water from agriculture would almost double the total supplies for the domestic sector (World Bank, 1993). The negative effects of such transfers have not yet been studied and documented. The withdrawal of water for irrigation is causing serious problems for the production of hydropower in Tanzania; the main power stations in the Rufiji and Pangani basins, which supply 80% of the nation's electricity, are downstream of the irrigation water abstraction points. This is just one example of competition for water between industrial and agricultural development. Finally, as incomes grow, the need to allocate water for environmental purposes will also increase accordingly.

[K-8] Against the background briefly sketched in the above paragraphs, this paper will consider in more detail: trends in irrigated agriculture, trends in rainfed agriculture, and water policy and institutional issues. Each discussion ends with an analysis of what is likely to happen in terms of food production in SSA if no changes are made ('business as usual') and what the impact could be if changes are made.

[K-9] Irrigation is key to the national strategy for increased food production in eight Sub-Saharan countries (Botswana, Burkina Faso, Kenya, Mali, Mauritania, Niger, Senegal and Somalia) which have little or no land with a rainfed growing period above 200 days and hence cannot meet food demands from low-input rainfed farming. Together these countries have almost 14% of the population of SSA. Twelve countries (Botswana, Burkina Faso, Chad, Ethiopia, Kenya, Mali, Mauritania, Niger, Somalia, Sudan, Tanzania and Zimbabwe) have a rainfed growing period of less than 120 days on over a quarter of their territory. In the semi-arid regions of these countries, irrigation is usually necessary for any reliable arable agriculture (Rosegrant and Perez, 1995)

[K-10] FAO, by analyzing a wide range of data, has estimated the distribution of crops grown on nearly three fourths of the irrigated area (see Rosegrant and Perez, 1995). In SSA, rice is grown on some 47% of the irrigated land, mainly in the humid zones of Western and Eastern Africa and in Madagascar, other cereals on 24% (mainly in the South), industrial crops, such as sugarcane, on 14%, and vegetables, fodder and arboriculture on the remaining 15% of irrigated land. Crop yields on irrigated land in SSA are generally lower than on irrigated lands in other parts of the world; both rainfed and irrigated yields in Africa are nearly 25% lower than the corresponding yields when all developing countries are taken together. Informal irrigation often springs up spontaneously (sometimes illegally) around formal irrigation schemes, using drainage water from the scheme (e.g. around small reservoir-based systems in Burkina Faso). It is questionable whether production from informal irrigation is counted in the official statistics.

[K-11] The potential for more irrigation in SSA depends not only on the physical characteristics of available soil and water resources, but at least as much on political choices, investment capacity and likely returns on investment, technological improvements, social and environmental requirements. Considering only physical conditions, estimates of the additional areas of SSA that could be irrigated range from 30 to 150 million ha. FAO's own analysis (FAO, 1996) gives the impression that in SSA an additional 29 million ha could be irrigated, implying that only 18% of the region's physical potential in SSA has so far been realized. These figures should be interpreted with caution; in any case, much of the untapped potential is in Central Africa where water resources are relatively abundant. However, it is certain that SSA, with less than 4% of its arable land under irrigation, is the region where irrigation is least developed with regard to the physical potential.

[K-12] Whether in the near future considerably more of this potential will be developed depends to a large extent on the expected returns on investment in irrigation in SSA. Rosegrant and Perez (1995) have used IFPRI's International Model for Policy Analysis of Agricultural Commodities and Trade to explore the impact of increased irrigation investment in Africa on supply, demand, trade, world prices of major food commodities, and food security to the year 2020. Two scenarios were compared: the baseline scenario (based on recent trends in irrigation expansion) giving an increase in irrigated land in SSA of 35%, and a high irrigation investment scenario, which would more than double the irrigated area of SSA by 2020. For SSA, the incremental increase in cereal production of the high investment scenario compared with the baseline scenario is just under 12%. For both rice and maize, the highest incremental increases in production are from the Southern and Western Sub-Saharan regions with 17% in rice production and 13.5% in maize production. The baseline projection depicts a deterioration in food security, with a projected rise in the number of malnourished children from 28.6 million in 1990 to 41 million in 2020. Even with high irrigation investment in Africa, the expected improvements in per capita food availability will not be adequate to substantially reduce malnutrition in the region. Rosegrant and Perez (1995) predict that without additional income growth and without additional increases in social expenditure for health and education, only a slight improvement in nutritional status (around a quarter of a million less malnourished children) will occur. Increased irrigation investment in SSA would have a significant impact on food production growth. But the amount of land under irrigation and the potential additional area with high irrigation investments are not large enough to generate quantum changes in crop production. Estimated savings in expenditure for cereal imports to 2020 are expected to be more than sufficient to pay for the expansion of irrigation under the high irrigation investment scenario.

[K-13] A stylized benefit-cost analysis of a 'generic' irrigation project in Africa (Rosegrant and Perez, 1995), shows that at border prices equivalent to import parity at the coast, it will require either lower than average investment costs or higher than average incremental rice yields to obtain an internal rate of return of 10%. The picture is more favorable if the import parity price reflects conditions in the interior. With the addition of marketing costs, including transport, storage and handling, the interior market price for rice was expected to be about 60% higher. A similar picture emerged for maize, also with an internal market price about 60% higher than at the coast. These results imply that irrigated maize would not be competitive for exports. An important conclusion is that the margin for error in irrigation development in SSA is small, and hence effective project design and implementation are essential to achieve acceptable rates of return.

[K-14] Spectacular failures of some large-scale irrigation systems in SSA have been used to advocate future investments in small-scale irrigation systems. However, careful analysis of large, medium sized and small irrigation systems shows that with full costing, full-control small-scale irrigation systems show the same wide range in investment costs and the same high end as large-scale irrigation (FAO, 1992). A World Bank review of irrigation investment in Africa estimated that average costs were $18,300/ha when indirect costs for social infrastructure, including roads, houses, electric grids, and public service facilities are included (Jones, 1995). Even the direct costs of irrigation development are higher in SSA than in Asia due to physical and external constraints, ranging from the high spatial variability of irrigable soils and difficult terrain, to over-valuation of most African currencies and lack of local manufacture of equipment. The cost of irrigation equipment, either manufactured, assembled locally, or imported, was found to be two to ten times higher than in Asian countries. A locally manufactured treadle pump was twice as expensive in Malawi, and even four times as much in Zambia, than in Asia (FAO unpublished mission report on technology transfer). Failures and successes in both large-scale and small-scale irrigation systems suggest that the size of the system is less important in determining success than the extent to which control is in the hands of the farmers. In systems managed by irrigation bureaucracies, the quality of management and equitable distribution of income to farmers within the system are important determinants of success. Conditions for success of larger systems include efficient management; relatively low cost infrastructure; low operating costs; good technical design which was fully operational at project completion; and availability of agronomically suitable crops, cropping systems and markets.

[K-15] Failures of large-scale government-operated systems draw more attention than failures of farmer-controlled small-scale irrigation. The latter also fail but the failures simply disappear. Some of these failures have been due to the difficult and highly variable climatic conditions, severe weed problems, pests and diseases, lack of suitable crop varieties and low use of complementary inputs, especially fertilizers, labor scarcity, especially during peak periods, and insecure land tenure and water rights. Conditions for success of small-scale farmer-operated irrigation include simple and low cost technology, e.g. cheap small pumps for drawing water from shallow aquifers, rivers, streams and dams; private and individual arrangements for operating the system; adequate supporting infrastructure to permit access to inputs and to markets for the sale of surplus production; high and timely cash returns to the farmers; and active and committed participation of the farmers in project design and implementation. Women are often more than men involved in small-scale irrigation development in SSA and they face a large variety of social, economic and cultural constraints to participating effectively in irrigation development and management (Zwarteveen, 1998). Peri-urban irrigation, which is spontaneously arising around rapidly growing cities and towns, although suffering from insecure land and water rights, is nevertheless considered to contribute significantly to urban nutrition. Much of its production escapes the official records.

[K-16] Appropriate technology is needed to raise yields per unit of water applied in irrigation. Sophisticated methodology for water control and application, which is appropriate for irrigation development in industrialized countries, is probably not suitable for SSA. Low-volume, high-frequency irrigation is an example of a technology that runs counter to the conditions for success listed in the previous paragraph. A distinctive feature of modern irrigation systems is the service-oriented mode of operation. Water delivery responds to farmers' demands for more flexible water supply, enabling increased crop diversification and market-oriented production. Water is provided as a service that should be as convenient and flexible as possible to the users, who in turn pay for the cost of the service (FAO, 1996). This service approach is also appropriate for small-scale farmer-operated systems in SSA. The wide-spread introduction of low-lift pump schemes in SSA illustrates how the availability of cheap, dependable motors and pumps and the increasing availability of fuel or electric power has changed irrigation in recent years. An example is the development of fadama (inland valley swamps) in Nigeria through the installation of shallow tubewells for the irrigation of market crops. Wide-spread introduction of other technologies, such as drip irrigation utilizing low-cost plastic pipes, and sprinklers, probably has to wait until the value of water increases and the demand for these technologies is strong enough for local entrepreneurs to become interested in their manufacture.

[K-17] The advantages of shallow aquifers for small-scale irrigation include easy access to water for low capital cost, conveyance of irrigation water over only a short distance, reliance on rainfall to renew the source and on soil layers or fractured rock to store the water until needed, and wide-spread availability. The intervention required in the process may consist of artificially increasing the recharge to the aquifer by small structures allowing rain-water to infiltrate rather than run-off. Shallow aquifer development can be associated with inland valley development, since flooding of the plain during and immediately after the rainy season facilitates storage of water underground, which can then be used during the dry season. However, there is usually insufficient information on the extent and yield of these aquifers. Exploitation of shallow aquifers in fractured rocks is more complicated as it requires special techniques such as horizontal wells. Over-exploitation of this type of aquifer may lead to irreversible closure of fractures thus impairing its storage capacity, as has occurred in South Africa. A general problem is how to avoid excessive pumping and to ensure equal access to the aquifer by all interested parties. Individual pump operators have no incentive to optimize long-term extraction rates, since the value of the water left in the ground can be captured by other farmers. Institutional arrangements for managing aquifers are quantity-based controls, prices and charges, and tradable water rights (or exchangeable permits) for stocks or flows of groundwater (Rosegrant and Perez, 1995). The effectiveness of each of these depends on their fair enforcement. Conditions for successful management of (shallow) aquifers include: agreement on the governance structures and their enforcement by the water users; management arrangements which are responsive to the local conditions and make use of available information (rather than considerations which require theoretically better but unavailable information); and the flexibility to change management arrangements when demand increases or new information about the state of the aquifer becomes available.

[K-18] The conclusion of this section must be that irrigation development in SSA presents enormous challenges and pitfalls. Errors in the design and implementation of both large-scale and small-scale irrigation are easily made due to the particular physical, economic and environmental conditions. As a result, investments in the urgently-needed rapid expansion of irrigation in SSA carry considerable risk, higher than competing projects in other developing countries. Research indicates that opportunities to substantially reduce malnutrition in the region by extension of the irrigated area are unfortunately not present. Even with high irrigation investment, i.e. much greater annual investments than have occurred in the recent past, per capita food availability is expected to increase only marginally compared with the baseline scenario, as discussed in paragraph K-12. For the northern region of SSA the model studies predict that food availability under the high irrigation investment scenario will still be under 2000 kcal/capita/day. The number of malnourished children in SSA will have increased from 28.6 million in 1990 to nearly 41 million in 2020, even under the high investment strategy. Since so little land is irrigated today, even a doubling of the total area will not by itself lead to production of enough additional food to make a significant impact on SSA nutrition. It is clear that rainfed agriculture must contribute its share to achieving this much needed nutritional improvement.

[K-19] Crop yields obtained under rainfed conditions in SSA are not accurately known. The national figures reported by FAO combine yields under irrigation with those for the same crop when grown under rainfed conditions. Variability of rainfed yields is bound to be large. In low-input subsistence farming under marginal conditions, cereal yields are often much below 1 T/ha, and it is likely that some of those data are not even included in the national records. Alexandratos (1995), analyzing FAO data, concluded that for SSA as a whole wheat yields had increased from 982 kg/ha for the average of three years 1969/71 to 1563 kg/ha for 1989/91, an annual increase of 2.35%. Similarly rice yields in SSA increased in those twenty years by 0.8% a year, maize by 1.0%, millet by 0.8% and sorghum by 0.5%. For all but rice, these figures mainly represent rainfed conditions. Analyzing national data, as reported by Alexandratos, for countries with hardly any irrigation, yield trends for wheat range from - 0.6% per year for Angola to 2.2% for Ethiopia and Lesotho. The best estimate of the annual yield increase under rainfed conditions that can be made on the basis of these data appears to be between 1.5% and 2%. As these figures show, food production has not been able to keep up with population growth.

[K-20] Low yield growth - and in many areas even a drop in biomass production under rainfed conditions - is largely due to the decline in plant-available soil moisture and nutrients (Rockstrom, 1995). The lack of plant-available soil water (now generally known as green water after Falkenmark (1995)) is the result of either reduced rainfall or a change in the partitioning of rainwater between infiltration and storage in the soil profile on the one hand and runoff on the other. As noted in the introduction, there is no convincing evidence that rainfall patterns have substantially changed; rainfall variability in SSA is historically inherently high. A change in what happens with rainwater once it hits the soil surface is the more likely cause of green water declines. Human-induced soil degradation is the key. Shortened fallow periods, reduced recycling of organic matter to the root zone, and generally low soil fertility all contribute to crust formation, increased runoff and reduced water holding capacity of the soil profile. Analysis of long term records of how rainwater is partitioned, e.g. those referred to by Savenije (1998) for the Mupfure sub-catchment in Zimbabwe for 1969-1989, would help to identify the rate of change in partitioning, and the scale on which it is significant (i.e. field or catchment).

[K-21] Biomass production per unit of water has been studied in different climates and for a wide range of crops. In arid and semi-arid regions it is in the range of 0.4 kg/m3 (Rockstrom,1995). Trials in Niger found that low-fertilized pearl millet produced 0.5 kg/m3, but when high-fertilized the production was 1.1 kg/m3. Interaction between fertilizer and organic matter application and the ratio of actual over potential evapotranspiration (as proxy of relative yield) is nicely illustrated by Rockstrom (1995) with data from Institut de Recherche Agronomique Tropicale in Burkina Faso. The effect of soil fertility on yield had been studied for 18 years for a mono-culture of sorghum. A good correlation between crop water deficit and grain yield was expected and found. There were also large permanent differences in yield level, regardless of crop water deficit, for different soil management techniques. Mono-culture of sorghum without fallow and with no external mineral inputs showed a drastic yield decline already after two years. Other treatments, involving various amounts of fertilizer and organic matter inputs, stabilized yields at levels ranging between three and ten times the no-input yield level. These results demonstrate that enhanced soil fertility produces higher yields per unit of water, i.e. a higher water use efficiency (see K-22). Looking at the economics, however, there is no evidence that the application of fertilizers pays off in the drier climatic zones. This is especially true for nitrogen, whose application has been found to be economic only in areas with at least 900 mm annual rainfall. Application of NPK-fertilizers and urea on millet and sorghum is generally only marginally profitable at best. Low initial yield levels imply that a small relative increase in yield does not produce enough grain to pay for the fertilizer. With higher yield levels, the same relative yield increase produces more grain and hence more income.

[K-22] When water and soil resources are limited, increasing biomass production per unit of water is needed to produce more food. The term water use efficiency has been used in different ways, one of which is biomass produced per unit of water. Not all the rain that falls on arable land contributes to food production. Whether rainwater is productive or not depends on its partition between infiltration and storage in the root zone, versus runoff. The scale at which these phenomena are considered is important because runoff from one field may be productive in evapotranspiration on a neighboring field. Yet, part of the water that infiltrates and is stored in the soil profile is also not productive as it is used up by non-productive plants. It is estimated that uncontrolled invasion of 'alien' plants, i.e. primarily trees introduced from other countries, use approximately 70% of South Africa's mean annual runoff. South Africa now has a million dollar program, Working for Water, that employs over 40,000 workers to eradicate the invading unproductive plants (Stockholm Water Front, February 1999). This nuisance vegetation tends to go to places where one would least like to have them, such as the wetlands and along streams and rivers. Similar situations in irrigated agriculture have led to a distinction being made between beneficial, non-beneficial and reasonable water use (e.g. Solomon and Burt, 1999). Non-productive but beneficial use includes deep percolation of water from the root zone to maintain salt balances, evapotranspiration from wind breaks and cover crops, and also the water used in wetting of seedbeds to enhance germination. These distinctions have proved valuable when considering water balances on different scales to assess the effect of water transfers away from (irrigated) agriculture. Likewise, in rainfed agriculture water use efficiency is scale-dependent.

[K-23] Climatological drought is said to occur when rainfall is below normally expected amounts, whereas agricultural drought occurs when the water supply is insufficient to cover the needs of crops and livestock. Droughts are frequent and severe in most African countries and mitigation of their effects is a crucial element in future development. The extent to which irrigation can contribute to management of the effects of droughts in SSA is yet unclear (FAO, 1996). Africa derives most of its irrigation supplies from highly variable local rainfall. The relation between rainfall and availability of irrigation water has not been studied in detail. It depends on local hydrological conditions, including the presence of (natural) storage areas. A small storage reservoir intended for regulation of water supplies between seasons is unlikely to be of much use over an extended multi-annual drought. Large investments in drought mitigation and management, though, appear justified when considering the huge costs associated with droughts (e.g., a drop in Niger's GDP of 16.6% as a result of the 1984 drought and in excess of $2 billion for drought relief operations in Southern African countries during the 1991/92 drought) and unimaginable suffering by large numbers of people and animals (100,000 deaths in Ethiopia as a result of the 1973 drought) (FAO,1996).

[K-24] Water harvesting can be defined as the collection of runoff and its productive use for irrigation of crops, pasture and trees, as well as for domestic and livestock consumption. In other words, it implies a human-induced change in the partition of rainwater in favor of green water. The concept is simple: runoff is collected from a larger catchment area and concentrated on a smaller target area for storage, either in the soil profile or in tanks, ponds and cisterns. The technology is old, but new developments, such as special machinery for sub-soiling and contour bunding, can accelerate the construction while reducing the labor costs. This is not necessarily a good thing as one of the conditions for success has been found to be the involvement of the water users from the early planning stages onwards, including construction to ensure a sense of ownership of the water harvesting facility. Other conditions of success are: the motivation of the water users resulting from a realistic understanding of the benefits of the facility; the strengthening and updating of existing indigenous water conservation measures; and the economic feasibility of the initial construction and subsequent maintenance costs (more likely to be economically feasible if water harvesting is part of a village or regional land use management plan, involving improved agronomic practices and inputs) (Oweis et al.,1999). Many water harvesting systems have failed because one or more of these conditions were not satisfied. Often the introduction of water harvesting systems is financed and assisted by NGO's. Their contribution to food security is especially valuable at village level in subsistence farming areas as successful water harvesting can alleviate short term effects of climatological droughts. Nevertheless, because of their small scale, these techniques are unlikely to have a significant impact in the foreseeable future on incremental food production in SSA (Rosegrant and Perez, 1995).

[K-25] Inland valley swamps are defined as the upper sections of river drainage systems, comprising valley bottoms, their hydromorphic fringes and flood plains. Their soils are submerged or saturated during part of the year. Their use and hydrological features vary depending on the climate and geology. Traditionally they were used for hunting, fuel wood extraction, fishing and dry season grazing, but population pressure has led to their cultivation mainly with low-yielding rice (FAO, 1996). It is estimated that in West Africa between 20 and 50 million ha could be classified as inland valley swamps, which presumably offer considerable potential for food production because of the availability of water. Examples of specific inland valley swamp development are given in Technical Background Documents of the World Food Summit (FAO, 1996; see also example mentioned in K-16). In these areas, it was found that the hydrological function of these wetlands, such as flood control and groundwater recharge could be adversely affected by uncontrolled drainage. Interference in the partitioning of rainwater always carries a risk. Agricultural development also interferes with fishing. Another lesson was that intensification of inland valley-bottom farming must be integrated within a farming system that includes cultivation of the slopes.

[K-26] Assuming (a) that the estimated growth rate for all cereals of at most 2% per annum was correct for the 20 year period of 1970 to 1990 (paragraph K-19), (b) that yields have continued to increase at that same rate, and (c) that there are about 150 million ha of arable land in SSA (rather heroic assumptions), then a ten percent higher annual growth rate of 2.2% would be needed to produce an equal amount of additional cereals from rainfed land as was expected from irrigated land with the high irrigation investment scenario. To keep up with a population growth rate of 3%, cereal yields on rainfed land will have to average 2.4 T/ha in 2020. Can it be done? Could it be done through improved land use management and fertilizer applications? Or does this level of increase require fast investments in protective supplementary irrigation on traditionally rainfed land? The potential of supplementary irrigation in dryland agriculture is probably dependent on socio-economic and cultural considerations. Subsistence farmers in rainfed agriculture are said to aim for a stable, minimum harvest every year, rather than occasionally high, but more risky (drought sensitive) yields that require high inputs in fertilizers, improved varieties, plant protection measures and perhaps also some degree of farm mechanization (Brouwer et al. 1992).

[K-27] The preferred strategy perhaps then is not to try for equal yield increases on all rainfed land, but rather to focus on high-input agriculture on the most suitable rainfed lands - an agriculture that makes more efficient use of green water through protective irrigation combined with improved agronomic practices. The conclusion of this section is that if nothing is done, yield levels in rainfed agriculture are likely to drop and per capita food availability in rural areas to decline correspondingly. Yield levels, however, could be raised, at least in the most productive regions of dryland agriculture, through major efforts to gradually change from low-input to high-input production. This is especially true if fertilizer use is increased. At present higher water use efficiencies are impossible to attain because of the low fertility status of the soil, and scarce water is wasted. As long as higher fertilizer applications are uneconomic, there is a vicious circle which only can be broken by increasing yields. This obviously asks for policy interventions.

[K-28] A recent World Bank publication on rural development (World Bank, 1997) says succinctly: "Many developing countries have institutional frameworks and agricultural policies that discriminate against the rural sector, under-invest in technology development and dissemination, maintain inappropriate agrarian structures, lock up arable land in low productivity ranching, undervalue natural resources and therefore waste them, seriously under-invest in the health and education of the rural population, discriminate against private sector initiatives in food marketing, and fail to maintain existing or invest in new rural infrastructure. Unless these policies, institutions, and public expenditure patterns are corrected, the world will not have abundant food supplies".

[K-29] Specific water-related policy issues include the need to have secure water rights in terms of water quantity and its quality, appropriate water pricing, and laws pertaining to water user associations. A legally secure water right is needed in (potential) conflict situations between water users, and also in relation to market transactions when (part of) a water right is transferred to enhance the efficiency or productivity of the water resource. Security of the water right will cause the user to benefit from investment in water-saving technology. The concept of water rights, however, is considered completely alien in the rural areas of many SSA countries. Water pricing is often equally controversial. Nevertheless, when water is a scarce good, it becomes part of good management to measure flows and price the water. Laws pertaining to water user associations deal with devolution of water rights from centralized bureaucratic agencies to farmers and water user associations. Devolution has a number of advantages, one of which is the empowerment of the user by requiring user consent to any reallocation of water, and compensation for any water transferred. There is little experience in SSA with turnover of (parts of) irrigation systems, and not much evidence that community-based institutions are better instruments than government agencies for correcting gender inequities and gender-based inefficiencies in the distribution of and access to resources.

[K-30] There are examples of successful private sector involvement in the management of irrigation systems, especially in Southern Africa (FAO, 1996). Managing agents have been contracted by water user associations and have provided essential services or managed schemes, unhindered by the stranglehold of normal government procedures. A private sector approach to the provision of services has been said to improve their efficiency, cost effectiveness and timeliness. The role of government agencies is then to control policy issues, and monitor the agents' performance.

[K-31] Private sector investment was thought until recently to be the most promising source of finance and technical expertise likely to be expanded as the business climate in Africa improved, and one from which the water sector would benefit greatly. Per unit profit margins are much higher on irrigated farms, including small-scale pump-irrigated ones. To fully exploit the potential of irrigation, however, farmers need a complete package of production inputs, improved access to land, land consolidation through the formation of producer cooperatives, regular maintenance of irrigation systems to prevent frequent pump breakdowns, and training programs on operation, maintenance and repair. Private sector investment was seen as an important source of just such a package (FAO, 1996). Foreign direct investment was $80 billion in all developing countries in 1993 and was growing until the slump in East Asia. In 1998 private capital flows to developing countries fell by more than $100 billion (The Economist, February 13th, 1999).

[K-32] Environmental issues typically relate to water quality problems resulting from the discharge of untreated waste water (including raw sewerage from villages and towns) into streams, irrigation canals and drainage ditches. A legal basis for controlling water pollution already exists in all of the countries in North Africa, and many of the countries in SSA, but the laws and regulations have been mostly unenforceable (Rosegrant and Perez, 1995). Privatization of urban water services in SSA, e.g. Cote d'Ivoire, has accelerated the rate at which the urban population obtained access to safe water. Under the same private sector arrangement, 80% of the rural population of Cote d'Ivoire was served by water points equipped with hand pumps with an obvious impact on the health of the rural population (Rosegrant and Perez, 1995). Environmental protection depends on a similar set of policies as are needed for other water users (agriculture, industry and domestic water users), i.e. water rights, participatory approaches to involve local communities and indigenous people in the management of wetlands, education, public awareness, and legally enforceable regulations. In other words, conservation and wise use of wetlands in SSA must be an integral part of river basin management.

[K-33] In many of the important aspects of water strategy, water use efficiency and conservation and environmental sustainability are fully complementary. Yet, as populations grow, cities become bigger and the industrial sector develops, environmental demands for water are bound to come into direct competition with the use of water for directly productive purposes in agriculture, household, and industrial sectors. Experience in other continents, especially in the USA and Australia, shows that the effectiveness of local institutions charged with implementing water allocation rules in managing scarce water resources depends critically on institutions at the state and federal levels. Coordination in regulation and enforcement is necessary, involving the lowest level of organized water users, such as farmer organizations, but also central government institutions (see for example Peterson et al. 1993). An adequate monitoring system and the willingness to comply to the rules for the long-term benefit of all are especially needed. If compliance to rules and regulations in society at large is weak, it is unlikely to be any better when it comes to the management of scarce water.

[K-34] The conclusion of the section is - paraphrasing the last sentence of the quotation from World Bank, 1997 - that unless those policies, institutional arrangements and public expenditure patterns which are counter-productive to integrated water resource management are changed, water that could be used successfully for improving agricultural productivity in irrigated and dryland agriculture alike, will be wasted and per capita food availability will continue to fall. Nevertheless, if policies and institutional arrangements are introduced which support increased agricultural productivity a start can be made to reverse these trends in yield per unit water and land, and in per capita food production.

[K-35] Many parts of Africa have been ravaged by persistent civil wars and political unrest. The world has seen many a country progress from relative prosperity to a situation where the countryside is covered with landmines and tens of thousands of people are displaced from the rural areas to live in dismal camps, unable to cultivate their land. These are the stories which draw world-wide attention. Their repeated viewing on television screens does not create a sense in the western world that further investments in sustaining and improving the continent's food production systems are worthwhile. Not when there are so many competing demands, and certainly not at a time of ever-diminishing foreign aid. On average 21 OECD countries devoted 0.22% of their GNP to development assistance in 1997, down from 0.25% a year earlier. Total overseas aid dropped by 12.8% from 1996 to 1997. Aid from the G7 group of big, rich countries has dropped by almost 30% in real terms since 1992 (The Economist, 13th February 1999).

[K-36] Although not sufficient by itself, irrigated agriculture was identified in section K-18 as a necessary element of enhanced food production in SSA. Without proper techniques for monitoring the physical performance of irrigation systems, it is impossible to assess the potential benefits that may accrue from further investments to improve them. Identifying and evaluating indicators for irrigation performance has started only in recent years, and the use of such indicators is not wide-spread in Africa. Although performance accountability in public irrigation systems is rare, it seems unlikely that without incentives to improve performance in a quantifiable manner, irrigation staff will make the necessary management changes. Values of performance indicators, both of physical and institutional performance, are at best known for only a short period of time. This points to a dilemma when looking for trends. Recent trends will be uncertain, while long term measurements, for instance of yields, may have ceased to be relevant as input levels have probably changed over the period needed to reduce the uncertainty generated by random year-to-year fluctuations to an acceptable level.

[K-37] There is, however, a general, if unquantified, awareness of the variety of good and bad experiences in irrigated and dryland farming alike across the countries of SSA. The conditions for success, listed in paragraphs K-14, K-15, K-17, K-24 and K-29 are based on analyses of best practices. Much more should be done to enumerate and further define these conditions for success. The lessons learned from the most successful production systems can then be applied and refined under different circumstances. Generalizing the potential for yield increases in irrigated and rainfed agriculture as was attempted in paragraphs K-12 and K-26, completely ignores the rich variability in inherent production potential across the African continent. However an acceptance that growth will be much greater in some parts of SSA than in others, leads to the inevitable conclusion that the disparity between rich and poor farmers in SSA is bound to increase when the pressure to produce more per unit land and per unit water becomes even greater than it is now. This has happened already in other parts of the world. Vice President Serageldin of the World Bank (as quoted in World Bank, 1997) asked: "How should resources be distributed between favorable and marginal areas? The first means higher food output, and low food prices for rural and urban consumers, the latter reduces rural poverty." The question takes on special significance in view of the dire predictions about future food production in SSA even with high investments in irrigation and an all-out effort to raise input levels in dryland agriculture.

[K-38] The section on water policies and institutional matters, gives the impression that in the end it all boils down to the need to improve the institutional arrangements governing the management of water. It is probably true to say that such improvements constitute a necessary condition for substantial and sustainable improvements in food production in SSA. What is really needed is an optimal combination of improvements in both technical and institutional arrangements, in which the institutional arrangements - in their broadest sense - provide the enabling environment for technical improvements to bear fruit. Unfortunately, changing institutions and policies for the benefit of the greatest number of people, is one of mankind's most difficult, unfinished businesses.

Alexandratos, N. (Ed.), 1995. World Agriculture: Towards 2010. FAO, Rome, Italy

Asuming-Brengpong, S., 1994. Effects of exchange rate liberalization and input subsidy removal on the competitiveness of cereals in Ghana. pp. 43-60 in Issues in African Rural Development 2. African Rural Social Sciences Research Networks (S.A. Breth, Ed.). Winrock International Institute, Arlington, VA, USA

Brouwer, J., J. Bouma and C. Renard, 1992. Improving water and nutrient use efficiency in subsistence farming in the Sahel: making the most of a poor and very variable environment. Progress Report, ICRISAT, Niamey, Niger and Wageningen Agricultural University, The Netherlands

Freeman, H.A., 1994. Population pressure, land use, and the productivity of agricultural systems in the West African Savanna. pp. 103-114 in Issues in African Rural Development 2. African Rural Social Sciences Research Networks (S.A. Breth, Ed.). Winrock International Institute, Arlington, VA, USA

Falkenmark, M., 1995. Coping with water scarcity under rapid population growth. Conference of SADC ministers, Pretoria 23-24 November 1995

Food and Agricultural Organization, 1992. Water for Sustainable Food Production and Rural Development - UNCED Agenda: Targets and Costs Estimates. FAO, Rome, Italy

Food and Agricultural Organization, 1996. Technical Background Documents 6-11. World Food Summit. FAO, Rome, Italy

Jones, W.I., 1995. The World Bank and Irrigation. The World Bank, Washington, D.C., USA

Oweis, T., A. Hachum and J. W. Kijne, 1999. Water harvesting and supplementary irrigation for improved water use efficiency in dry areas. SWIM Paper 7. International Water Management Institute, Colombo, Sri Lanka

Peterson, E.W.F., J.D. Aiken, and B.B. Johnson, 1993. Property Rights and Groundwater in Nebraska. Agriculture and Human Values 10:41-49

Rockstrom, J., 1995. Biomass production in dry tropical zones: how to increase water productivity. Pp31-47 in Land and Water Integration and River Basin Management. FAO Land and Water Bulletin 1. FAO, Rome, Italy

Rosegrant, M.W. and N.D. Perez, 1995. Water Resources Development in Africa: A Review and Synthesis of Issues, Potentials, and Strategies for the Future. International Food Policy Research Institute, Washington, D.C., USA

Savenije, H.G.H., 1998. How do we feed a growing world population in a situation of water scarcity. Pp.49-58 in Water the Key to Socio-Economic Development and Quality of Life. Proceedings of the 8th Stockholm Water Symposium, SIWI, Stockholm, Sweden.

Solomon, K.H. and C.M. Burt, 1999. Irrigation sagacity: a measure of prudent water use. Irrigation Science 18:135-140

World Bank, 1993. A Strategy for managing Water in the Middle East and North Africa. The World Bank, Washington, DC, USA

World Bank, 1997. Rural Development: From Vision to Action. A Sector Strategy. The World Bank, Washington D.C., USA

Zwarteveen, M., 1998. Identifying gender aspects of new irrigation management policies. Pp. 67-86 in Gender Analysis and Reform of Irrigation Management: Concepts, Cases, and Gaps in Knowledge (D. Merrey and S. Baviskar, Eds.). International Water Management Institute, Colombo, Sri Lanka