Almost all of the water of the planet occurs as saltwater in the oceans. Of the 3 percent of the global resource that is freshwater, two-thirds comes as snow and ice in polar and mountainous regions. Hence, liquid freshwater constitutes about 1 percent of the global water resource. At any one time, almost all of this occurs as groundwater, while less than 2 percent of it is to be found in rivers and lakes. In the temperate, humid climates, about 40 percent of the precipitation ends up in the groundwater, and for Mediterranean-type climates the figure is 10-20 percent. For the truly dry climates, the value can be virtually zero (Bouwer, 2002). Not all the water in rivers and lakes and in the groundwater is accessible for use because part of the water flows in remote rivers and during seasonal floods that cannot be captured before the water reaches the ocean. An estimated 9 000 - 14 000 km3 of water is economically available each year for human use. At most, this represents 0.001 percent of the estimated global water. At present, annual withdrawals of water for human use are about 3 600 km3. This may give the impression that there is plenty of water available that could be withdrawn for human use. However, part of the available surface water must remain in the rivers and streams to ensure effluent dilution and safeguard the integrity of the aquatic ecosystem. How large this part should be is little understood. It varies with the time of year, and each river basin has its own specific ecological limit below which the system can be expected to degrade. A global estimate for this demand is 2 350 km3/year. Adding this to the amount withdrawn each year for human use results in nearly 6 000 km3 of economically accessible water that is already committed (FAO, 2002b). This indicates that globally the margin is fairly small. Because water and population are distributed unevenly throughout the world, the water situation is already critical in various countries and regions and likely to become so in several more.
Agriculture is the principal user of all water resources taken together, i.e. rainfall (so-called green water) and water in rivers, lakes and aquifers (so-called blue water). It accounts for about 70 percent of all withdrawals worldwide, with domestic use amounting to about 10 percent and industry using some 21 percent (Figure 2).
Figure 2 Water withdrawals by region and by sector
Source: Crops and drops. FAO, 2002
There is an important distinction between water withdrawn for use and water actually consumed. In irrigated agriculture, about half of the water withdrawn (with considerable variation in this figure) is consumed in evaporation and transpiration from plants and moist soil surfaces. Some of the plants that contribute to this evapotranspiration process are unproductive weeds and plants on wasteland. Water that is abstracted but not consumed infiltrates the soil and is stored as groundwater or flows back through drains into rivers. This drainage water is generally of a lower quality than the water originally withdrawn owing to contamination by agrochemicals and salts leached from the soil profile. Compared with a return flow of 50 percent of the water withdrawn for agriculture, 90 percent of the water for domestic use is returned to rivers and aquifers as wastewater, while industry typically returns up to 95 percent. Poor-quality return flows from urban and industrial areas are sometimes treated before being returned to watercourses, but the non-point character of agricultural pollution makes treatment difficult. In this sense, agricultural water pollution may be better handled by controlling quantitative use and outflow from agricultural land.
Rain is the source of water for crop production in the more humid regions of the world where some 60 percent of the worlds food crops are grown. Rainfed agriculture takes place on some 80 percent of the arable land and irrigated agriculture produces 40 percent of the worlds food crops on the remaining 20 percent. In order to meet future food demands, it is expected that relatively more crops will have to be grown on irrigated than on rainfed land, such that about equal amounts will come from both types of areas. Given the importance of rainfed cereal production, insufficient attention has been paid to potential production growth in rainfed areas. Most attention usually focuses on the possible expansion of irrigated areas. However, increasing cereal yields in rainfed temperate countries, better plant protection and manure techniques, and the use of supplemental irrigation in more arid countries indicate the significant potential for improving rainfed agriculture.
In arid regions, water scarcity is the result of insufficient rainfall. Semi-arid regions may receive enough annual rainfall to support crops but the rainfall is distributed so unevenly in space and time that rainfed agriculture is barely possible. Rainfall variability generally increases with a decrease in annual amounts and it is particularly high in the Sahelian countries. These regions are known for their periodic droughts that may last several years. Rain in semi-arid regions also tends to fall in a few hard showers. Such rainfall is difficult to capture for agricultural use. This leads to large amounts of runoff going into drains and eventually seeping down to the groundwater or to rivers. Where river discharge is large and difficult to manage, one way to capture the flow is through spate irrigation in which part or all of the river flow is diverted into fields surrounded by high bunds. In this way, one irrigation of up to 50 cm can provide enough soil moisture for a wheat crop even in the skeletal soils found in Yemen. Floodwater harvesting takes place within a streambed and entails blocking the water flow. This causes water to concentrate in the streambed. After the flood season is over, the streambed area where the water collects is cultivated. A terraced wadi (ephemeral stream) system is one type of floodwater harvesting. Here, a series of low check dams are constructed across a wadi and the wadi area is cultivated. Too much flow will breach the check dams or the diversion structures in spate irrigation. The suitability of these methods also depends on the soil conditions and depth of the wadi bed. Water harvesting, which is the collection and storage of surface runoff, has also proved useful in semi-arid regions with infrequent rains (Chapter 3).
Plate 2 Canal intake and bank-protection gabions built under irrigation repair programme (Afghanistan)
Although there is a great variety of such rainwater technologies, it is not clear whether their widespread use is always feasible, especially for poor farmers. The costs involved in the construction and maintenance of the water harvesting system play a major role in farmers decisions on whether to adopt the technique or not. In the past, water harvesting systems were often installed with financial support from outside agencies, such as NGOs and international funding agencies. Many of those systems failed because of lack of involvement of the beneficiaries and their inability to organize and pay for maintenance. Rosegrant et al. (2001) report construction costs for water harvesting systems in Turkana, Kenya, of US$625-1 015/ha. Labour and construction constitute the bulk of the water harvesting costs as the opportunity cost for using the land is essentially zero. The initial high labour costs of building the water harvesting system often provide a disincentive for adoption of the technique. Moreover, many farmers in arid or semi-arid areas do not have the human resources available to move the large amounts of earth necessary in the larger systems. Therefore, small-scale water and soil conservation techniques that are applicable at field level are often adopted more easily. Investments on a larger scale require the existence or creation of community organizations both to pay for the necessary investment and maintenance, and to manage the benefits of the water harvesting infrastructure. Maintenance of the system is sometimes required in the rainy season when labour is relatively scarce and therefore expensive because of competition with conventional agriculture (Tabor, 1995). Notwithstanding these reservations about the widespread applicability of extensive water harvesting systems, model studies indicate that there is significant scope for increasing rainfed production provided that appropriate investments and policy changes are made (Rosegrant et al., 2002). Crop breeding specifically for rainfed environments is crucial to future cereal growth. Chapter 3 discusses integrating crop and water resource management.
Groundwater use for irrigation presents a paradox: regions where the groundwater resource has been overdeveloped coexist with regions with considerable potential for development of groundwater for use in irrigated agriculture (Box 1). A corollary is the so-called fallacy of aggregation: in aggregate terms, at the global or even national level, groundwater availability appears far in excess of present use. The annual groundwater use for the world as a whole has been placed at 750 - 800 km3 (Shah et al., 2000). This figure may appear modest compared to the overall groundwater reserves, but only a fraction of the worlds groundwater reserves are economically available for agriculture. It is estimated that about 30 percent of the worlds irrigation supply is made up by groundwater but this input accounts for some of the highest yields and highest value crops (FAO, 2003).
Box 1 Overabstraction and sustainability: complex theory, simple practice
Source: Burke and Moench, 2000
There remains confusion in the usage of the terms overabstraction and groundwater mining. The latter refers solely to the depletion of a stock of non-renewable groundwater, so leaving the aquifer dewatered indefinitely. The planned mining of an aquifer is a strategic water resource management option where the full physical, social and economic implications are understood and accounted for over time. However, the replenishment of aquifers by downward percolation of rainwater shows high interannual variability and is a complex physical process that is difficult to evaluate. A declining water table does not necessarily indicate overabstraction of the groundwater resource. Overabstraction should not be defined in terms of an annual balance of recharge and abstraction. Rather, it needs to be evaluated over many years, as the limit between non-renewable stock and the stock that is replenished by contemporary recharge from surface percolation is usually unknown.
What is important to decision-makers and well users is the overall reliability and productivity of a well (in terms of water levels, volumes and water quality) during a given time period. Therefore, if a well taps a particular aquifer, what is its sustainable rate of exploitation, given variable periods of recharge and drought? The answer to this question is not trivial, and requires a certain level of precision in understanding the system dynamics. Where the system dynamics are understood, the maximum available drawdown can be calculated from a nonlinear equation. This equation can be resolved by an analytical approach or through the application of numerical models. Where the aquifer system is sufficiently well known, the assigned value of the maximum available drawdown may also include the exploitation of a portion of the non-renewable groundwater resource. Such methods can provide a basis for pre-empting aquifer degradation before physical and socio-economic damage is done, by giving indications to users of sustainable abstraction rates.
The number of tubewells providing water to irrigated land in India, China, Pakistan, Mexico and many other countries has grown rapidly in the past 40 years. For example, some 60 percent of irrigated cereal production in India depends on irrigation from groundwater wells. This has led to widespread and uncontrolled overabstraction of the resource and the creation of a bubble groundwater economy (Roy and Shah, 2002). In Yemen, abstraction is estimated to exceed recharge by 400 percent (Box 2). Groundwater abstraction and recharge have rarely been quantified accurately. This should be a first step in assessing the potential for further development of the resource and designing management approaches (Box 2). Where irrigated agriculture depends in part on pumped groundwater, many of the command areas present a mosaic of irrigation methods. These range from totally irrigated by canal water to entirely fed by pumped groundwater, with most of the fields having some combination of the two. Hence, irrigation is by definition conjunctive, but there are few examples of conjunctive management.
Box 2 Participatory modernization of water management to reduce overabstraction of groundwater in Yemen
Source: Dixon et al., 2001
The immediate consequence of the continuous decline in water resources in Yemen has been household food insecurity, especially for poor families in vulnerable rural areas. The only viable option is to improve the management of the available resources through the introduction of appropriate technologies and management tools.
In 1995, conscious of these issues, the Government of Yemen launched a programme to improve the general efficiency of irrigation with groundwater. It included the Land and Water Conservation Project (financed by the World Bank). This project is based on cost sharing, farmer participation and modern irrigation technologies.
Water savings achieved at the farm level have ranged from 10 to 50 percent. At the regional level, the savings in water use have averaged at least 20 percent and have reached as high as 35 percent, particularly in northwest Yemen where most of the farms are equipped with bubbler irrigation systems. Considering the current operational costs that farmers pay for pumping water (even with relatively low energy costs), the cost of investing in modern equipment is recovered in 2-4 years through water savings alone. In addition, the new technology offers benefits beyond water savings. These include significant improvements in yield and product quality, resulting from changes in cropping patterns and increases in the irrigated area.
In China, 52 percent of the irrigated lands are (at least in part) served by tubewells. As a result of overabstraction of groundwater, water tables have fallen by up to 50 m over the last 30 years. For example, in the Fuyung Basin in north China, surface water has been curtailed drastically in order to meet industrial demand, and farmers have responded by resorting to groundwater irrigation. The root of the Asian groundwater crisis alluded to by Shah et al. (2000) that threatens millions of poor rural communities lies in the open access nature of the resource. Paradoxically, it is precisely this feature of groundwater in shallow aquifers that has made it a powerful tool in the fight against poverty (Moench, 2002), i.e. that everyone who can afford to install a pump has free access to water. Irrigation with groundwater is generally more productive than canal irrigation because groundwater is produced close to where it is used with hardly any losses during transport. In addition, farmers are in control of the timing and amount of the water extracted. Evidence in India suggests that crop yield per cubic metre of water on groundwater-irrigated farms tends to be 1.2 - 3 times higher than on farms irrigated with surface water (Shah et al., 2000).
Throughout the world, most groundwater development has proceeded primarily on the basis of individual initiatives. Unlike surface irrigation or drinking-water supply projects where government agencies are generally involved in many aspects of design, financing and implementation, most groundwater development is driven by the decision of individual farmers to drill wells and buy pumps. While governments often facilitate this process through subsidies and rural electrification, large implementation departments are rare. In consequence, there are very few government agencies that have frequent and direct contact with groundwater users. Furthermore, surface water development generally involves the diversion of flows or the construction of storage on a clearly defined stream or water body. The impact of such actions on downstream users is generally clear, at least in a conceptual sense. As a result, large bodies of customary and formal law along with the resource monitoring and enforcement systems required to implement them have developed over the long history of surface water development. This is not generally the case with large-scale groundwater development. It is a recent phenomenon and diversions have a far less directly observable impact on other users. As a result, groundwater extraction remains highly individualistic and tends to occur outside the framework of established institutions for allocating, monitoring or managing the resource base. In locations such as India, tens of millions of individuals own and operate wells. Most of these wells are on private lands. The location, use and even existence of such wells is often unknown to any individual aside from the owners and their immediate community. As a result, no established institutional basis for management exists.
Global indicators of water scarcity tend to ignore variations in the importance of irrigated agriculture for food security among countries. They also fail to account for seasonal differences in supply. For example, more than 70 percent of the total supply in India occurs in the three monsoon months of June, July and August, when most of it floods out to the sea. Moreover, countrywide data ignore regional differences in water supply and withdrawal within the country, an example of the aggregation fallacy. Regardless of these caveats, most observers conclude that many countries do not have a surplus of water available for irrigation. In fact, many countries do not have sufficient annual water withdrawal to irrigate their potential gross irrigated area even at high basin-irrigation efficiencies. Basin-irrigation efficiency includes all reuse of drainage water and is considerably higher than system-irrigation efficiency where drainage flow from one system is used for irrigation again downstream in another system. Most analyses indicate that, whatever the water scarcity indicator used, more than half of the worlds population lives in countries with varying degrees of water scarcity. This scarcity can be physical (there is no more water), economic (the country cannot afford to develop additional water resources) or caused by a lack of social adaptive capacity. Examples of adaptive capacity are the ability to produce more value per unit water consumed, and imports virtual water, which is the water used to produce the crops obtained on the world market (Allan, 1995) (Chapter 3, Box 5).
Plate 3 Farmlands under water as a result of flash floods (Bangladesh)
There is concern that more people may be affected by food insecurity as a result of water scarcity. Competition for the same resources combined with the increasing trend of water pollution exacerbates this problem. Moreover, the largely unknown impact of climate change may make water scarcity in some countries more severe. Several studies suggest that rice yields are likely to increase in the higher latitudes and decrease in the lower latitudes under future climates. It is likely that the poorest countries (and the poorest people within them) will suffer disproportionately as they are less able to adapt to the changing conditions. Future projections by the IFPRI (based on their model studies) indicate that water withdrawals will rise by some 22 percent between 1995 and 2025. Projected withdrawals in developing countries will increase by 27 percent in the 30-year period, compared with 11 percent in developed countries (Rosegrant et al., 2002). Only a very small increase in irrigated area is expected, which will be more than offset by increases in river basin efficiency.
However, FAO expects that the overall water withdrawal for irrigation in all developing countries will increase from 2 128 km3 in the benchmark period of 1997/99 to 2 420 km3 in 2030, an increase of nearly 14 percent. FAO further expects the irrigated area in developing countries as a whole to increase from 202 million ha in 1997/99 to 242 million ha by 2030, an increase of nearly 20 percent. The largest increase is expected in sub-Saharan Africa with 44 percent, and the lowest in East Asia with 6 percent. The expected increase is 32 percent for Latin America, about 10 percent for the Near East/North Africa and 14 percent for South Asia (FAO, 2002c; Faurès et al., 2002). The effectively cultivated irrigated area is expected to grow by 34 percent during the period under consideration, because of higher cropping intensities. Much of the difference in the rates of increase of water withdrawal and irrigated area relates to the higher water productivity in irrigated agriculture that is expected to have occurred by 2030, with some effect also from a change from water-intensive rice to wheat production, especially in China.
For the 93 countries taken together, the irrigation water withdrawals would increase from 8 percent in 1998 to 9 percent in 2030 if expressed as a percentage of annually renewable water resources. This statistic is of limited practical value where rainfall and river flows are highly variable. It is also another example of the fallacy of aggregation. In the Near East/North Africa region, irrigation withdrawal would increase from 40 to 53 percent of the renewable resource and in South Asia from 44 to 49 percent, compared with an increase of 1-2 percent in Latin America. The regions with withdrawals of more than 40 percent of their renewable water for irrigation present the greatest challenge, especially as the differences are greater at the country level. Of the 93 countries, 10 (including Egypt and Pakistan) now withdraw more than 40 percent of their renewable water resources for irrigation, while another 8 (including India and China) abstract more than 20 percent of the renewable water for irrigation.
Plate 4 Section of the left bank of the principle canal under construction at Bishenyi (Rwanda)
Although the IFPRI and FAO predictions differ in the details, they agree on the general direction of the expected changes. Closer agreement cannot be expected as the model outcomes depend on the underlying assumptions. A most important assumption is the extent to which water productivity in agriculture can be increased between now and 2025 or 2030.
An accurate global view of irrigation investment trends does not exist, but certain proxies can be used to indicate such trends. For example, there has been a sharp decline in World Bank lending for new irrigation schemes (Jones, 1995). Funding for new irrigation construction has largely stopped and the emphasis is on the sustainability and efficiency of existing systems. According to Thompson (2001), irrigation and drainage is still one of the core investment activities of the World Banks rural portfolio, but now mainly in support of rehabilitation and the devolution of responsibilities to water user associations (WUAs). However, the number of irrigation and drainage projects is expected to decrease to well below what it was in the 1980s. Investments in irrigation systems are perceived to have failed to address the changing needs of irrigation services because rehabilitation of existing systems was mostly carried out to restore original project objectives. This type of rehabilitation is often inappropriate as it tends to ignore desirable changes in cropping patterns and irrigation techniques and thus allows low water-productivity practices to continue. Cost and time overruns in irrigation projects as well as public opposition to large dam building projects have further eroded the confidence of funding agents in irrigation investments. Considering the negative aspects of irrigated agriculture (e.g. salinity, waterlogging, health hazards, and groundwater exploitation), it can no longer be taken for granted that irrigation is a protected, preferred practice and that its negative externalities will be accepted unconditionally. Nevertheless, irrigation development and dam building must continue, even if only to update existing facilities and to replace dams and reservoirs that have lost most of their storage capacity as a result of sedimentation. The loss of effective capacity of Mediterranean dams is currently between 0.5 and 1 percent/year with some as high as 3 percent (in Algeria). In Morocco, the reduction in regulating capacity attributable to reservoir silt-up is equivalent to a loss of irrigation potential of 6 000-8 000 ha/year (FAO, 2002d). Improved erosion control in catchment areas may eventually prolong the life span of reservoirs and dams.
Plate 5 Farmer checking rice crop during rice-cum-fish trials (Zambia)
A reduced level of investment in irrigation may not be all bad. In the past, the construction of many irrigation systems was supply driven as part of internationally funded development aid without significant input from the future users of the scheme and sometimes against their expressed wishes. Irrigation potential was and still is seen as an important indicator for evaluating future irrigation development. This parameter expresses how much a countrys irrigated area could be expanded according to land use and water availability criteria. Hence, its value changes over time depending on the countrys economy and competition for water. However, this notion of irrigation potential has often been used as the sole criterion in setting a countrys agricultural and water resource policies, without a parallel analysis of economic, social, institutional and environmental constraints and without a thorough market analysis. The failure of some irrigation schemes can be attributed to a narrow focus on irrigation infrastructure and water distribution combined with an insufficient focus on the productivity of the agricultural systems and their responsiveness to agricultural markets (Burke, 2002).
The public policy of supply-driven irrigation development, adhered to by many governments and donor agencies, may be justified because of the observed importance of the role of irrigated agriculture in food security (below). However, the role of the private sector in irrigation development is often underestimated or ignored. The many investment decisions of smallholders and commercial farmers could exceed public investment, e.g. in Zambia (FAO, 2002e) and India (Moench, 1994). In particular, where there is a comparative advantage in irrigated production to service local and international markets, which may be for vegetables and cut flowers rather than traditional food crops, significant private irrigation investments appear to follow.
Since 1960, growth in average cereal yields has largely kept pace with the increase in world population. It is widely assumed that it will continue to do so until the population begins to stabilize. Most of the increase in grain production has been the result of yield increases rather than expansion in cropped area. Projections by FAO, the IFPRI and the World Bank assume that the further increases in cereal production will come from continuous increases in yield. However, trends in yield data collected by FAO indicate that the average world cereal yield would have to reach at least 4 tonnes/ha for a world population of 8 000 million people from its present level of about 3 tonnes/ha (Evans, 1998). At present, all the developed countries taken together have not achieved an average cereal yield of 4 tonnes/ha. This is the extent of the challenge.
The contribution from irrigated agriculture to achieving this goal will be critical as irrigation provides a powerful management tool against the vagaries of rainfall. Irrigation also makes it economically attractive to grow high-yielding crops and to apply the adequate plant nutrition and pest control required in order to obtain the full potential of these modern varieties. According to the IFPRI, while food production will increase much faster in developing countries than in developed countries, it will not keep pace with demand, and food imports will need to increase. In 1999/2000, developing countries produced 1 030 million tonnes of grain, i.e. 55 percent of world production, and accounted for 61 percent of world grain consumption. To bridge the gap between demand and production, developing countries imported 231 million tonnes of grain, equivalent to 72 percent of worldwide imports. These statistics illustrate that developing countries play a major role in the international agricultural trade and that they are highly susceptible to changes in the world agricultural market in terms of food security. For the poorest countries, an increase in domestic agricultural production is key to improving food security. This explains why expectations about the food security role of irrigated agriculture remain high (Box 3).
Box 3 Water for food security in China
Source: Heilig et al., 2000; Smil, 1996
The question whether China can produce food for its growing population is controversial. Brown (1995) has suggested that the answer is no. One of the counterarguments is that China has more farmland than the government acknowledges officially. Another is that the official data underestimate crop yields by up to 50 percent for crops in hilly interior regions. The data are probably quite reliable for rice from the central and eastern provinces.
Water shortage is probably the single most important problem facing Chinas agriculture today. Water usage in China is predicted to rise by 60 percent by 2050, as an increasing proportion of the people live in cities. Water deficits may affect 36 percent of Chinas grain production, which is produced in areas that either depend totally on irrigation or have significantly higher production when irrigated. However, this also means that 64 percent of crop production is not threatened systematically by water shortages, either because it comes from fields in humid regions, or because precipitation is sufficient for some rainfed production. However, because of drought conditions, this production may not be attainable every year. Without irrigation and water management, paddy fields in the humid south can probably not produce their current two to three harvests but only one or two. However, in a large area in Chinas south and southeast, there is no water scarcity problem but a water scarcity challenge.
More efficient use of water and fertilizers combined with lower post-harvest losses would constitute the most important improvements in Chinas irrigated agriculture. The creation of WUAs has already contributed to ensuring a more regular, guaranteed supply of water to farmers, who then allocate water equitably through the associations. Other improvements could include improved production of pork, reliance on broilers to supply most of the additional demand for meat, expanded production of farmed fish, and increased consumption of dairy products. This combination can contribute considerably to meeting the countrys future nutrition needs without requiring enormous imports of foreign grain.
Agricultural development based on water conservation and irrigation is often considered a promising avenue for poverty alleviation in rural areas. For example, the availability of water for a small domestic garden plot, usually managed by women, can make a significant difference to household nutrition and thus contribute to improved livelihoods. Water harvesting may make this possible (FAO, 2002d). However, this effect is small scale and irrigated agriculture with its higher crop yields is expected to have greater impact on the incidence of poverty and malnutrition. This effect is expected regardless of whether the irrigation project is small or large scale. However, recent studies have shown that poverty alleviation as a result of irrigation development requires that the project be geared towards the needs of the poor (van Koppen et al., 2002). This includes access to training in the technical aspects of irrigation but also in community organization and marketing. One of the recurrent problems is the lack of access to credit, capital or land. Even microcredits have no grace period; repayments typically have to start after a few weeks. This makes them of little use for the purchase of cheap technology, such as treadle pumps and microdrip systems. It has been argued that these technologies are profitable within a short period and do not require a subsidized price for poor people or specific poverty alleviation measures (FAO, 2002d). The problem of credit is not specific to irrigation development and needs to be addressed in a more general sense for successful rural development of poor regions.
Expanding irrigated areas, increasing the control of water and applying high-yield technology in irrigated agriculture have given rise to large increases in farm income, especially in Asia. However, this increase has been disproportionately in the hands of the larger peasant farmers. They are not the poorest of the poor, but their increased expenditure pattern has driven increased employment of those who are the poorest of the poor. The latter have little or no land and they benefit little even from agricultural production programmes directed most closely to them. However, they benefit from lower food prices, increased wages and growth in demand for rural non-farm goods and services (FAO, 2002d; Mellor, 2001; Briscoe, 2001). By contrast, the capital- and import-intensive consumption patterns of large-scale farmers, and especially absentee farmers, contribute much less to poverty reduction. This is more typical for some Latin American countries than for Asia and Africa.
Cost recovery from poor farmers for the operation and maintenance (O&M) of irrigation systems is controversial. Subsidizing these services and providing irrigation water far below cost is financially unsustainable. Stepped tariffs in which the basic need is provided free to poor people may work in the case of drinking-water but is difficult to implement for irrigation water. Monitoring the efficiency of water use in agriculture for many small farmers each using a small amount of water is expensive, but providing irrigation water below cost contributes to wasting of water (FAO, 2002d).
In developing countries, agriculture generally produces many non-tradable goods, such as food crops of lower quality and goods with unusually high transaction costs. This aspect gives agriculture a prominent role in poverty reduction. It also buffers the national economy from shocks to international markets in agricultural commodities. For the rural poor in low-income countries, increased employment opportunities allow them to escape from poverty and hunger. Because they generally have few skills, the poor are more likely to find employment in the production of goods and services that cannot be marketed on the international market. Examples of this type of employment include maintenance of irrigation and drainage structures, watershed management, and afforestation, and where there is a sizeable storage reservoir, employment could be found in fisheries, ecotourism and navigation. Thus, increased employment and, hence, poverty reduction depend on increased domestic demand for these non-tradable, non-farm goods and services. Agriculture is the principle source of such demand and so it is only with rising farm incomes that poverty can be reduced and food security increased (FAO, 2001c).
Plate 6 Women watering cabbages in a vegetable garden with water drawn from a deep well (Mali)
Hence, investments in irrigation development may achieve additional goals such as enhancing economic growth and poverty alleviation in rural areas. Nonetheless, the question may be asked whether investments in other parts of the infrastructure are not more likely to achieve these goals. For example, the steady decline in poverty in India from the mid-1960s to the early 1980s was strongly associated with agricultural growth, particularly the green revolution, which coincided with massive investments in agriculture and rural infrastructure (Fan et al., 1999). According to IFPRI studies in India, the impact of additional irrigation investments on poverty reduction ranked third after rural roads, and agricultural research and extension. Additional government spending on irrigation had a significant impact on productivity growth, but no discernible impact on poverty reduction. While spending on irrigation and power has been essential in the past for sustaining agricultural growth, the levels of irrigation may now be such that it may be more important to maintain rather than increase the systems. The IFPRI studies have also indicated that the marginal returns to several infrastructural investments in India are now higher in many rainfed areas. They also have a potentially greater impact on reducing rural poverty (Bhalla et al., 1999).
A global analysis of the link between farming systems and poverty indicates that the prospects for reducing agricultural poverty in the Near East and North Africa are good (Dixon et al., 2001). However, for the region as a whole, exit from agriculture is the best household strategy for poverty reduction, followed by increased off-farm income. The study indicates that the priority roles of the State are to support the development of vital infrastructure, such as roads, water supplies, services and power supply, and to regulate resource use and pricing for water and power. By comparison, in South Asia, measures that support households in small-farm diversification and also for growth in employment opportunities in the off-farm economy were found to be the most likely to contribute to poverty reduction.
When weighing the pros and cons of new irrigation investments against the benefits of other investments, all additional potential benefits of irrigation, such as health benefits resulting from better nutrition (i.e. more calories and a more balanced diet) and greater rural employment, should be taken into account. Many of the irrigation benefits are site-specific and generalizations cannot be made. Moreover, 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. Notwithstanding these caveats, the fundamental question concerning the economic utility of further investments in irrigation development as a means towards rural development and poverty alleviation is an important one. At least two conclusions can be drawn from the discussion of the role of water in sustainable food production, poverty alleviation and rural development. The first is that donor agencies and governments have a difficult choice to make when investing for poverty reduction and rural development. The choice is not automatically for agriculture or water. The second conclusion is that the right set of government policies can make a large difference in food production, poverty reduction and rural development.