The outlook for water resources in the year 2020: challenges for research on water management in rice production

R. Barker,^a D. Dawe,^b T.P. Tuong,^b S.I. Bhuiyan^b and L.C. Guerra^c

^a International Water Management Institute (IWMI);
^b International Rice Research Institute (IRRI);
^c University of the Philippines in Los Baños (UPLB)

INTRODUCTION

Over recent years, the scarcity of and competition for water have been increasing worldwide. The opportunities for developing new water resources for irrigation are limited. Irrigated agriculture has traditionally consumed more than two-thirds of the available water supply. As the demand for industrial, municipal and other uses rises, however, less water will be available for agriculture. If food security is to be maintained, ways of increasing the productivity of water must be found.

Within agriculture, rice is the dominant irrigated crop, accounting for approximately 30 percent of the total irrigated area. Over half of the rice area is irrigated. However, the potential for increasing the productivity of water in rice production is as yet unknown.

This article is divided into four sections: first the outlook for global water resources in 2020 is discussed; trends in Asian rice production and consumption and irrigated area are then presented; the concepts of irrigation efficiency and water productivity are discussed briefly, and several alternative practices with potential for increasing water productivity in rice-based systems are presented; and an agenda for research in irrigation and water resource management at the farm, system and basin levels is set forth. This agenda must be followed if potential gains in water productivity are to be realized.

OUTLOOK FOR GLOBAL RESOURCES IN 2020

More than 97 percent of the world's water resources are in the oceans and seas and are too salty for most production uses. Two-thirds of the remainder is locked up in ice caps, glaciers, permafrost, swamps and deep aquifers. Every year, about 108 000 billion m³ precipitates on to the earth's surface. About 60 percent of this (61 000 billion m³) evaporates directly back into the atmosphere, leaving an annual water resource of 47 000 billion m³. If this amount were distributed evenly across the world's population, there would be approximately 9 000 m³ per person per year. However, water availability varies across continents, with North and South America being better watered than Africa, Asia and Europe. Furthermore, much of Africa and Asia's potential supply is lost through runoff caused by heavy seasonal rains. It is estimated that only 9 000 to 14 000 billion m³, or about a quarter of the annual water resource, may ultimately be controllable. At present, an estimated 3 400 to 3 700 billion m³ is utilized (adapted from Seckler, 1993; Postel, Gretchen and Ehrlich, 1996; WRI, 1994).

Agriculture is the largest consumer of water, using 72 percent of the total worldwide and 87 percent in developing countries. With growing demand for water for non-agricultural uses (domestic, municipal, industrial and environmental), the proportion available for agriculture is projected to decline to 62 percent worldwide and 73 percent in developing countries. In developing countries, the growth in water demand for industrial and municipal uses, in absolute terms, is expected to exceed the growth in water demand for agriculture between 1995 and 2020 (Rosegrant, Ringler and Gerpacio, 1997).

In agriculture, other than rainfall, the principle source of water withdrawal is irrigation. In the half century following the Second World War the irrigated area of the world has tripled from 90 million to 270 million ha, an annual compound growth rate of over 2.5 percent. However, there is wide variation among continents in the proportion of cropland irrigated, ranging from one-third in Asia to between 6 and 12 percent in other continents. About 60 percent of irrigated cropland is in Asia, and approximately 50 percent of the irrigated area is devoted to rice production.

A shift in the future allocation of water among competing uses is inevitable. The global trend will reduce the share of water for agricultural use. The growth in irrigated area has slowed in the past decade and is projected to increase at an annual rate of less than 1 percent between 1995 and 2020 (Rosegrant, Ringler and Gerpacio, 1997). Thus, new investments in irrigation and water supply systems will be inadequate for meeting the growing non-agricultural demand for water and for mitigating the impact of water withdrawals from agriculture. Major changes in practices, policies and institutions will be required to ensure that limited water resources are appropriately managed to increase the productivity of water in irrigated agriculture. If these steps are not taken, rice will be the crop most affected, as it depends most heavily on irrigation.

TRENDS IN ASIAN RICE PRODUCTION AND IRRIGATED AREA

World rice production and consumption are dominated by countries situated in a belt from Pakistan in the west to Japan in the east. Historically, these countries have accounted for more than 90 percent of world rice production, and they maintain that position today. This section will focus primarily on trends in these countries. Collectively, the countries will be referred to as Asia, and data from countries in the Near East, such as Turkey, the Islamic Republic of Iran and Iraq, are not included.

Trends in rice production

The most noticeable feature of the trends in Asian rice production over the past 30 years has been a slowdown of growth (Table 1). From 1967 to 1984, while the green revolution was having its maximum impact, Asian rice production grew at an average rate of 3.2 percent per year, and growth was reasonably rapid in all subregions, except Japan and the Republic of Korea. Subsequently however, from 1984 to 1996, growth slowed to an average of only 1.5 percent per year. Trends in area harvested and yield also show a deceleration of growth. Annual average yield growth in Asia was 2.5 percent from 1967 to 1984, but only 1.2 percent from 1984 to 1996. Growth in area harvested was 0.6 percent per year in the first period and only 0.3 percent per year in the second. The slowdown would have been even more pronounced were it not for economic reforms in the emerging market economies of Southeast Asia (Cambodia, the Lao People's Democratic Republic, Myanmar and Viet Nam). These reforms allowed production growth in the region to increase from 2.8 percent per year in the first period to 4.0 percent in the second; but clearly much of this growth is simply recovery from bad policies and is not sustainable indefinitely.

Much of the slower growth in area, yield and production has been caused by lower levels of rice prices in both international and domestic markets. Since the 1960s, world rice prices have declined significantly. Even when the three years of world food crisis in the mid-1970s are excluded, real world rice prices averaged about US$769 per tonne from 1967 to 1981. From 1984 to 1997, however, real world prices averaged just $322 per tonne, a decrease of 58 percent (all prices are in 1997 United States dollars). Real domestic farmgate prices also declined in most countries during this period. Lower prices are the consequence of the introduction of new technologies and the expansion of irrigation in developing countries, and the subsidization of agriculture in developed countries. Sharply lower prices have reduced the profitability of rice farming, especially in the face of higher opportunity costs for labour and land in many rapidly growing Asian economies.

Prices are falling even while supply seems to be expanding less rapidly, which points to a slowdown in demand growth. This slowdown has been caused by secular trends in population growth and per caput consumption of rice. Population growth in Asia has been decelerating for the past three decades. In the 1960s, it averaged 2.3 percent per year but slowed dramatically, to just 1.4 percent, in the 1990s. Furthermore, as incomes rise, per caput consumption continues to increase, but at a decreasing rate (at least beyond a very low level of income), until incomes have reached a level at which per caput consumption begins to decline in absolute terms. This has already happened in Japan, Taiwan Province of China, the Republic of Korea and, more recently, Malaysia and Thailand. Thus, the rapid economic growth experienced in much of Asia has been translated into a slowdown in growth of demand for rice. This tendency has been exacerbated by rapid urbanization throughout the region, since per caput rice consumption is lower for urban residents than it is for rural dwellers. The trend of slow growth of demand is likely to continue unless the financial crisis, particularly in Indonesia, leads to increased rice demand caused by a slowdown in income growth, substitution of imported wheat and migration back to rural areas.

Trends in irrigated area

Approximately 56 percent of the world's total irrigated area is in Asia (as defined earlier), a figure roughly the same as the region's share of world population. A large part of Asia's irrigated area, perhaps half, is used for growing rice.

Since 1960, there have been two distinct phases of irrigation growth in Asia. For the period from 1961 to 1980, annual average growth in irrigated area was 2.1 percent (these data refer to total irrigated area, not irrigated rice area). This growth rate was relatively stable across five-year subperiods within that interval. From 1980 to 1995, however, growth slowed significantly, falling to an annual average of 1.3 percent. The decline in growth rates was concentrated in East and South Asia (Table 2). In Southeast Asia, average annual growth from 1980 to 1995 was 1.9 percent, the same as it was between 1960 and 1980. Compared with South and East Asia, however, Southeast Asia is relatively unimportant; it accounted for only 11 percent of Asia's total irrigated area in 1995.

There are at least four interesting characteristics of post-1980 growth in irrigation in Asia: first, growth since 1980 has been much slower than it was in the 1960s and 1970s; second, much of the growth that has occurred has resulted from the use of tubewells, as opposed to canal systems; third, large parts of the new irrigated area are not being cropped with rice; and fourth, although much of the new irrigated area is not being cropped with rice, the share of rice area that is irrigated has increased since the late 1970s.

A slowdown in irrigation growth. For Asia as a whole, total irrigated area grew at an annual average rate of 1.3 percent from 1980 to 1995. Growth was predominantly concentrated in five countries: India, China, Pakistan, Thailand and Bangladesh (Table 3). Collectively, these five countries accounted for 91 percent of the growth from 1980 to 1995 (they also accounted for 87 percent of total irrigated area in Asia in 1980). Indonesia and Japan each had more irrigated area in 1980 than either Thailand or Bangladesh, but neither contributed much to growth after 1980.

Which factors are most responsible for the slowdown in growth of irrigated area in the 1980s and 1990s? Because rice is the major user of irrigation water in Asia, lower rice prices on both international and domestic markets have reduced the incentives for governments and donors to sponsor large-scale irrigation projects. Lower prices have also reduced incentives for small organizations and individual farmers to invest in irrigation technology. At the same time, the land most naturally suited to irrigation has already been developed, with the result that the costs of constructing new irrigation systems are now higher than in earlier years. Rosegrant and Svendsen (1993) state that the real costs of new irrigation more than doubled in India and Indonesia between the early 1970s and the late 1980s, and they cite large increases in several other countries as well. It is likely that this combination of lower rice prices and higher construction costs has been a major factor behind the decline in the growth rates of irrigated area in Asia in the 1980s and 1990s. There are other factors, including the poor performance of existing systems and the increasingly vocal opposition of environmentalist groups to the construction of new large dams.

The increasing importance of groundwater irrigation. Much of the growth in irrigated area in Asia since 1980 has come from tubewells. This is particularly the case in South Asia, which has accounted for nearly two-thirds of the growth in Asia's irrigated area since 1980. In India, tubewells and other wells accounted for nearly 80 percent of the growth in net irrigated area from 1980 to 1992, with canals accounting for only 16 percent of growth. In Pakistan, the area irrigated strictly by canals declined in absolute terms from 1982 to 1995, while tubewell irrigation, either by itself or in conjunction with canals, increased by more than enough to allow an increase in total irrigated area (thus, tubewells accounted for more than 100 percent of the growth in total irrigated area). A similar trend occurred in Bangladesh from 1981 to 1992, with sources of irrigation other than tubewells declining in absolute terms, and tubewells accounting for more than 100 percent of the growth in total irrigated area.

The increasingly important role of tubewell irrigation in the 1980s and 1990s continued a trend that began in the 1960s. In India, tube- and other wells accounted for 77 percent of growth in total irrigated area between 1963 and 1980, nearly the same share as for the 1980 to 1992 period. In Pakistan, tube- and other wells accounted for more than half the growth in total irrigated area from 1961 to 1981. However, watertables are falling in many parts of the region, raising the question of sustainability.

Large sections of new irrigated area are not being cropped with rice. Although rice is the most important irrigated crop in Asia, it appears that large parts of the increase in irrigated area since 1980 are not being planted to rice (Table 4). First, consider India, which was responsible for nearly half of Asia's growth in net irrigated area during this period. While gross irrigated area for all crops increased by 16.6 million ha from 1980 to 1992, gross rice irrigated area increased by only 3.3 million ha during the same period, accounting for only one-fifth of the total expansion in irrigated area. This is a continuation of a long-term trend: in 1960, 45 percent of gross irrigated area in India was cropped to rice, but by 1992 this share had fallen to 30 percent. A similar pattern prevails in Pakistan, where the increase in irrigated rice area was less than one-tenth of the increase in total irrigated area from 1980 to 1995. The situation is even starker in China and Thailand, where irrigated rice area is declining in absolute terms at the same time as total irrigated area is expanding. In China, most of the increase in irrigated area is in the north, where little rice is grown and the area planted to rice is declining, as it is also in the south.

There are exceptions to this general trend, the most important being Bangladesh, Myanmar and Viet Nam, where it appears that most new irrigated area is being planted to rice. These countries account for a much smaller share of total irrigated area than do India, China and Pakistan, however.

The proportion of rice area that is irrigated is increasing. Although large parts of new irrigated area are not being cropped with rice, the proportion of rice area that is irrigated is increasing. This is primarily because large areas of upland (3 million ha) and deep water rice (1.2 million ha) were lost between the late 1970s and the early to mid-1990s, constituting a 25 percent decline in the rice area in these ecosystems (Huke and Huke, 1997). In addition, irrigated rice area increased during these years at an average annual rate of about 0.9 percent, or about 600 000 ha. Thus, while irrigated area accounted for about 51 percent of total rice area in the late 1970s, this figure increased to 55 percent in the early to mid-1990s. Excluding China, where large areas of irrigated rice were lost during the period, the relative increase in irrigated area is more significant, rising from just 35 percent of total area in the late 1970s to 44 percent in the early to mid-1990s. This increase in the proportion of irrigated area means that it is very important that yields on irrigated land continue to rise in order to meet future demand without a substantial increase in prices.

POTENTIAL IMPROVED PRODUCTIVITY OF WATER AND RICE-BASED SYSTEMS

In this section of the article, some of the issues and problems associated with the measurement of efficiency and productivity at the farm, system or basin level are discussed; specific practices and more general improvements that have the potential to raise water productivity at farm or system level are described; and issues related to basin-level efficiency are examined.

Measuring productivity and efficiency

Irrigation efficiency is defined in terms of the amount of water required for evapotranspiration (ET) divided by the amount of irrigation water diverted into the system. Although the definition of efficiency is not consistent among studies (e.g. it often includes seepage and percolation [S&P] as well as runoff and transpiration [R&T] as requirements), the efficiency of rice-based systems is less than 50 percent and lower in the wet than in the dry season (Table 5). It would seem that there is plenty of scope for improving efficiency.

However, in discussions of potential improvements to the efficiency of rice-based systems, the level or boundary of the target area, whether it be farm, irrigation system or river basin, should be taken into consideration. The strategy for improving efficiency will vary from level to level depending on a number of factors, including cost considerations. Furthermore, increasing efficiency at the farm or system level may not lead to greater efficiency at the next level up (i.e. system or river basin). This is because surface runoff (SRO) and S&P at either the farm or the system level can be used elsewhere in or outside the system (wastewater reuse or recycling). For example, the often recommended practice of canal lining to save water may simply reduce groundwater recharge. Based on this premise, and looking from the basin level perspective, a number of recent reports (Seckler, 1996) argue that improvements in water efficiency, where lost water is recovered downstream, result only in "paper" or "dry" water savings. According to these authors, it is only useful to save water (real water savings) which would otherwise be lost to a sink (saline water body or the ocean). However, wastewater recovery or recycling often involves an additional cost which should influence the water management options selected by farmers, irrigation system managers and regional- or basin-level policy-makers.

Saving water does not necessarily lead to increased water productivity.

The efficiency concept gives little information about the amount of food that can be produced with the amount of water available. In this respect, irrigation water productivity, defined as "the amount of food produced or the gross value of output per unit volume of water used" is a more useful concept (Viets, 1962; Tabbal, Lampayan and Bhuiyan, 1992; Tuong and Bhuiyan, 1997; Molden, 1997). As with efficiency, in productivity assessment the boundary or level (farm, system or basin) must be clearly defined when making measurements.

Interventions at the farm level, such as the introduction of new varieties, which increase the output per unit of ET, will lead to greater farm-, system- and basin-level productivity. However, most interventions are designed to improve water management efficiency by reducing S&P and SRO even though (as noted earlier) unless it is known what happens to water that is lost in this way, i.e. the off-site or downstream impact, it is impossible to judge whether increasing irrigation water efficiency will lead to gains in water productivity. Furthermore, alternative interventions which lead to gains in water productivity should be judged on the grounds of cost-effectiveness.

Moving from the farm to the system and basin levels, accounting for irrigation water use and productivity rapidly becomes more complex. Consideration must be given to the consumption or depletion of irrigation water inflow, not only in crop production but in all other uses, as well as to the committed and non-committed outflow. Process depletion or the amount of water diverted and depleted to produce the rice crop (principally transpiration for crop production) may represent a relatively small portion of inflow (Table 6).

The location of the basin plays a major role in improving water productivity. Benefits have to be looked at from the point of view of hydropower generation, flood protection, instream uses and environmental considerations. Improving water productivity at the basin level therefore assumes a wider significance in managing basin water resources. New tools, such as remote sensing, Geographic Information Systems (GIS) and hydrologic modelling, need to be used for planning and developing appropriate management strategies.

Practices and strategies for increasing irrigation water productivity

As seen in the previous section, the practices and strategies for improving water productivity will vary according to location even within the same basin. For example, if at the head of the system or basin, water lost from S&P or SRO simply flows to downstream users, there will be no benefit in saving water. If, however, at the tail of the system, SRO will be lost to the sea, then reducing runoff or practising recycling will have a high pay-off.

The productivity of irrigation water can be increased by doing one of the following: increasing the value of output per unit of water transpired (T); reducing losses to evaporation (E); reducing losses to S&P; reducing SRO; or reusing or recycling water, either within the system or elsewhere in the basin (RCL). A range of alternative practices and their potential effects on one or more of the above are described in the following subsections and summarized in Table 7.

TABLE 7
Practice for increasing the productivity of water and the effect on reducing transpiration (T), evaporation (E), irrigation inflow requirement (IIR), S&P, SRO and recycled (RCL)
Practice	T	E	S&P	SRO	SCL
Developing improved varieties
Improving agronomic management
Changing schedules to reduce evaporation
Reducing water for land preparation
Changing rice planting practices
Reducing crop growth water
Making more effective use of rainfall
Water distribution strategies
Water recyling and conjunctive use

Developing improved varieties. The adoption of early-maturing, high-yielding varieties of rice over the past three decades has led to a rapid growth in rice output per unit of land and water in many parts of the world (although data on water productivity gains are very scarce). Advances in biotechnology could facilitate further improvement of varieties with tolerance to drought, salinity and cold temperatures, leading to a further increase in output per unit of T.

Improving agronomic management. Introducing optimum combinations of improved technologies or management practices, such as pest control and nutrient management, can raise crop yields and output per unit of T.

Changing the crop planting date. The rate of evaporation varies between wet and dry zones and, within a given location, between wet and dry seasons. In the Punjab in India, the availability of early-maturing varieties of rice has made it possible to shift the planting date from May to July, avoiding the months of high losses to E before canopy closure.

Reducing water use for land preparation. The land preparation period, which currently lasts more than a month in many areas and accounts for as much as one-third of the water diverted, can be reduced to a few days. This might require using more field channels instead of plot-to-plot water delivery and a change in land preparation practices including dry tillage. The result would be a substantial reduction in losses caused by E, S&P and SRO.

Changing rice planting practices. In research conducted in Central Luzon in the Philippines it was found that the wet-seeded rice (WSR) system used 1 747 mm³ of water, compared with 2 195 mm³ for transplanted rice (TPR) (Bhuiyan, Sattar and Khan, 1995). For a similar change to direct seeding (WSR), the reduction in water use was from 1 836 to 1 333 mm³ (Fujii and Cho, 1996). Dry-seeded rice (DSR) appears to offer even greater potential for water saving (Ho Nai Kin et al., 1993). A substantial amount of this saving appears to be caused by the associated change in land preparation practices described earlier. Thus, the shift results in a reduction in losses to E, S&P and SRO. By contrast, in the Republic of Korea it is reported that direct-seeded rice requires 15 percent more water. This is because, under temperate zone conditions, direct seeding extends the growing season.

Reducing water use in the crop growth period. Puddling the soil during land preparation is one of the most common practices for reducing S&P during the crop growth stage. Numerous studies conducted on the manipulation of depth and interval of irrigation have shown that maintaining a saturated soil or alternate wetting and drying after the flowering stage, compared with continuous shallow submergence, could reduce water applications to the field (E, S&P and SRO losses) by 40 to 70 percent without significant loss in yield (Hatta, 1967; Tabbal, Lampayan and Bhuiyan, 1992; Singh et al., 1996). In fact, in Guangxi and Hunan provinces in China, the introduction of alternate wetting and drying is reported not only to have reduced the water requirement but also to have increased the yields (SWIM, 1997). However, because of the rice plant's sensitivity to drought, these water saving practices require a high level of management control and good infrastructure. More needs to be understood about the costs and benefits of such techniques, including the requirement for other inputs such as labour, the effect on crop plant protection and the ways in which the water saved is used.

Making more effective use of rainfall. Traditionally, rice irrigation systems throughout much of Asia, which are called "run-of-the-river" systems, have been designed to divert water to provide supplemental irrigation as insurance for the wet season crop. However, storage systems have expanded rapidly in the last half century, making it possible to increase the area irrigated in the dry season. To make optimum use of storage for dry season irrigation, water releases must be managed carefully in the wet season to take full advantage of the rainfall and to reduce irrigation inflow requirements. This would reduce losses to S&P and SRO. However, considerable management coordination is needed between farmers, who must adjust their planting schedules, and irrigation administrators, who must provide timely release of water. In Sri Lanka the success rate has been mixed. Projects such as the Kadulla Irrigation Scheme (Bird et al., 1991) and the Walagambahuwa Minor-Tank Settlement Scheme (Upasena, Sikurajapathy and Seneviratna, 1980) reported initial success. But, as one colleague studying the success of the latter project stated: "when we withdrew [i.e. the project ended] they [the farmers] withdrew" (Nimal Ranaweera, Department of Agriculture, personal communication). In spite of this, the management and control procedures for successful implementation of this practice would appear to be fairly modest. The failure to convince farmers may be related to their economic situation (lack of money to finance the inputs for early planting) and their aversion to risk, while failure to convince the irrigation administration may reflect a lack of incentives and rewards for implementing management improvements.

Water distribution strategies. The familiar "head-tail" problem plagues many irrigation systems. Farmers at the head of the lateral or turnout receive ample water, while those at the tail receive too little or, in some cases, too much, which leads to waterlogging. Particularly in the dry season, it may be impossible to achieve an even distribution of water over the upper, middle and lower reaches of a system with rotation, which would reduce losses to S&P and SRO and provide water to a larger area. Several forms of irrigation rotation are possible according to the level in the system, the time schedule, etc. The implementation of rotational water distribution in Gal Oya Left Bank in Sri Lanka, Lower Gugera Branch in Pakistan and Tungabhadra Pilot Irrigation Project in India was not successful (Murray-Rust and Snellen, 1993). Failure was attributed to the lack of communication or cooperation between the irrigation agencies and the farmers.

Water recycling and conjunctive use. SRO and S&P from the field and the conveyance network may eventually be reused or recycled (RCL). In some instances, water may flow from one farm field to the next. In others, water may recharge aquifers and be pumped up for reuse outside the system to which it was initially delivered. This is an effective way of increasing the productivity of water in the basin. Water recycling and conjunctive use of groundwater are rarely considered in the original design of irrigation schemes. The shift to tubewells often happens as a desperate response from farmers who are unable to gain access to irrigation water from the canal. In South Asia, as noted earlier, the area irrigated by private tubewell has been increasing rapidly. In India it now exceeds the area irrigated by tanks and canals. In areas as disparate as Punjab, Bengal and Tamil Nadu, wealthier farmers have been willing to pay relatively high prices (even though electricity is free) for an assured water supply. Not all of this reflects conjunctive use or recycling, and in many areas watertables are falling at an alarming rate.

General strategies for increasing water productivity

This section discusses two strategies for increasing irrigation efficiency and the productivity of water: rehabilitation and modernization; and irrigation management transfer.

Rehabilitation and modernization. During the 1980s, following the completion of many major irrigation schemes, there was growing concern about the rapid deterioration of many systems. The focus shifted from new construction to rehabilitation. In its narrow interpretation, rehabilitation is defined as the restoration of infrastructure to its original form. When improvements were considered, the initial emphasis was on physical infrastructure such as regulators and canal linings. However, today rehabilitation investments typically take a far broader view and include institutional, organizational and technical changes. Clearly this involves instituting a higher level of management and control with the aim of reducing the irrigation inflow requirement. Modernization involves all of the above elements, but there is no commonly agreed definition of modernization. This is partly because irrigation system improvements should take into account the institutional setting and management capacity in a given location, as well as the differences in requirements for the irrigation of rice compared with other crops. The often overlooked location- and crop-specific character of irrigation water management suggests that a single concept of modernization is inappropriate.

Jones (1995), recounting the World Bank's experience of irrigation system design problems, indicates that the more reticulated systems, which are supposed to be capable of water delivery on demand, are suitable to dryer regions but inappropriate for the humid tropics and for rice systems. In a study of six rice-based irrigation systems financed by the World Bank in Asia (Rice, 1997), it was concluded that the operations and management (O&M) performance of agencies and irrigators was better than expected from a review of the literature, and that poor O&M had a negligible impact on crop production. The study also observed that, despite the low return, paddy farmers had tended to increase the amount of land devoted to paddy. A case study in the Philippines (Kikuchi et al., 1996) shows a high variability in O&M activities and a rate of deterioration among laterals in the same system.

Rehabilitation has focused on improving irrigation efficiency. There are, however, very few studies in which the impact of rehabilitation on water productivity (crop output per unit of water) has been measured. Among these, the Gal Oya Left Bank rehabilitation project is unusual in that it has been able to analyse data over a period of 23 years, before and after rehabilitation (Amarasinghe, Sakthivadivel and Murray-Rust, 1998). The rehabilitation undertaken in 1982/83 involved both physical improvements and the development of farmers' organizations. Table 8, which compares the periods before and after rehabilitation, shows that the area irrigated increased by 25 percent, the crop yield per hectare by 50 percent and the yield per cubic metre of water by 100 percent. Success was attributed to the simultaneous implementation of physical and institutional improvements. Unfortunately, however, the yield or productivity figures have not been adjusted for national trends in yield over the same period.

Irrigation management transfer (IMT). For the past two decades, an ever-increasing number of countries around the world have undertaken irrigation management transfer, or turnover, defined as "the transfer of responsibility and authority for managing irrigation from government agencies to farmers or other local management organizations" (Vermillion, 1997). There have been a number of studies of this process, and the literature shows a mixture of positive and negative results. Government expenditures for irrigation tend to decline, although costs to farmers often rise and the responsibility for rehabilitation or major maintenance is often unclear. The evidence suggests that there has been little change in yields, water productivity or farm incomes, although it may still be too early to judge these outcomes. As in the case of rehabilitation projects, very little monitoring and evaluation work has been carried out to identify the impact of IMT over time on crop yields and water productivity. In Asia, IMT has moved slowly. There have been efforts in some countries to transfer maintenance, and/or irrigation fee collection, to farmers, but the responsibility for systems operation and water allocation has remained in the hands of the Government. There seems to be a lack of pressure for change from both sides. In some situations, farmers have no incentive to take over management of the system, while in the strongly centralized irrigation bureaucracies there is no desire to relinquish control. Particularly in South Asia, however, where groundwater is available, farmers are "privatizing" by developing their own wells.

CHALLENGES FOR RESEARCH

Compared with the huge amount of money invested in irrigation, very little seems to have been invested in research. There are very few available data on the productivity of irrigation water and the cost of various options for increasing productivity. Many countries do not have a research organization that deals with irrigation and water management. For those that do, the focus tends to be on hydrology and/or farm-level water management. There seem to be two explanations for this. First, as long as plenty of water was available, research to improve efficiency in water management had low priority. Second, the tools of analysis at the system and basin levels were poorly developed.

The picture has changed dramatically over the past few years. Growing scarcity of, and competition for, water have increased the demand for research into improving water productivity. The rapid development of information technologies (remote sensing and GIS) has made it possible to determine water resource flows, in both the surface and the subsurface areas, at the river basin level. Information technology is to water management what biotechnology is to crop improvement.

A major challenge for research over the next ten to 15 years is the identification, for specific situations, of the optimum combination of improved technologies and management practices that can raise water productivity at the farm, system and basin levels. New tools (such as remote sensing, hydrologic modelling and GIS) must be applied at the irrigation system and basin levels so that appropriate strategies can be developed for increasing the productivity of water in agriculture and allocating water among competing uses. This constitutes a largely new area for research and a challenge for a multidisciplinary team of researchers. The following sections give a more specific list of some of the research challenges that lie ahead.

Accounting method for water use and productivity

Data on the efficiency of irrigation systems are scarce and, when they are available, the method of derivation is often not described. There are rarely data on the productivity of water at either the farm or the system level. A common water-accounting procedure is needed for analysing the use, depletion and productivity of water at the farm, system and basin levels. Such a procedure is necessary in order to make comparisons of productivity within and among these three levels. Crop yield per unit of water should become a standard calculation, particularly in areas where water is a scarce resource.

Molden (1997) has developed procedures for measuring the productivity of water resources at the farm, system and basin levels, based on water balance and the measurement of inflows, outflows and internal use over a defined area. These procedures are being tested in watersheds in Sri Lanka and India by scientists from the International Water Management Institute (IWMI) in collaboration with colleagues from national organizations. Procedures include the use of remote sensing which now makes it possible to measure basin-level evapotranspiration and estimate crop yields.

Irrigation management interventions at the farm and system levels

In the previous section a range of interventions that have the potential to increase the productivity of water were outlined. However, research is needed to determine the optimum combination of improved technologies and management practices that can meet the component water demands (i.e. water for land preparation, stand establishment and crop irrigation) under various conditions, with the least water consumed and with management of S&P and SRO to ensure basin-level efficiency.

Basin-level strategy

As the competition for water among sectors and users grows, the requirement for water for irrigation must be considered in conjunction with the demands for other uses that ensure food security and environmental sustainability. The importance of a basin perspective in measuring irrigation water productivity has already been noted. A holistic systems approach to research and development needs to be adopted at the irrigation system and basin levels. Remote sensing, GIS and modelling techniques can be used in planning the basin-level allocation of water among competing sectors.

Uses of irrigation water for purposes other than crop production

As water scarcity intensifies, there will be increasing pressure to transfer water from agricultural to municipal and industrial uses. Agriculture will receive less water, largely on the basis of the relatively low output and value per unit of water in irrigated agriculture (especially grain production). However, water for irrigation systems is used for more than irrigating field crops and is thus undervalued. Irrigation supplies critical water resources for a variety of productive uses including livestock, fishing, domestic use and industry. Irrigation provides water for trees, both commercial and non-commercial, and has both positive and negative impacts on the environment. Thus, farmers with irrigated fields are not the only stakeholders in irrigation systems. As the scarcity of and competition for water increase, there is a growing need to account for and evaluate the irrigation water used for purposes other than crop production.

Water quality

As the quantity of water for agriculture decreases, the quality is also declining, with increased salinity and pollution caused by municipal and industrial waste and the use of agricultural chemicals and fertilizers. The amount of land lost annually to salinization may well exceed the area of new land currently being developed for irrigation. The loss in crop production to declining yields resulting from salinization is not known. Pollutants contaminate potable water supplies and water for livestock and fishing. More research is needed to identify the magnitude of the problems and to seek cost-effective solutions.

Groundwater

There has been a rapid growth in groundwater and in the conjunctive use of ground- and canal water, particularly in South Asia. Research is needed to compare the productivity of water supplied by tubewells with that from tanks and canals. If this increasingly important source of water for irrigation is to be sustainable, steps must be taken to monitor the draw-down of watertables, to facilitate the recharging of aquifers and to reduce overexploitation.

Policy, institutional and management reforms

Increasing the productivity of water or the amount of food produced per unit of water consumed will require major changes in policies, institutions and management practices. Research is needed to identify policies, support services and institutional improvements that will increase the productivity of the water used in agriculture. However, institutional and management capacities vary widely across locations. The decentralized systems of East Asia, for example, contrast sharply with the highly centralized bureaucracies of South Asia. Irrigation institutions cannot be treated simply as a response to factor scarcities. It will be necessary to identify the set of improvements and reforms appropriate for the specific conditions.

The economics of water productivity

Interventions that lead to higher water productivity almost always require increased inputs of other resources, such as management, labour and capital. Economic analyses of alternative techniques for raising the productivity of water in a given location are almost completely lacking, owing substantially to the lack of adequate data describing the physical relationships. Such analysis will be in greater demand as attempts are made to establish irrigation systems with greater financial autonomy and less reliance on government subsidies.

Impact assessment of alternative interventions

Too few studies assess the impact of interventions on irrigated area, water and land productivity and related factors in the way that the Gal Oya Left Bank study does (Table 8). Such studies require careful monitoring over time in order to capture the before and after and the with and without effects and to separate the changes caused by intervention from other factors, such as a change in weather. Use of the standard accounting procedures and water balance measurements described in this article will facilitate monitoring activities and the assessment of off-site and downstream impacts at the system and basin levels.

Improvement of varieties and agronomic management practices

While the above research agenda is concerned with water management improvements, it must be remembered that much of the increase in water productivity in agriculture over the past few decades has come from varietal improvement and better crop management practices, such as pest and nutrient management. Further development of varieties, not only with higher yield potential but also with tolerance to drought, salinity and cold temperature, should continue to be one of the most important avenues for raising the productivity of irrigation water.

CONCLUDING REMARKS

The world faces growing scarcity of and competition for water. By far the largest consumer of water is irrigated agriculture. Within agriculture, rice is the dominant irrigated crop, accounting for approximately 30 percent of the irrigated area.

A number of factors characterize irrigation in rice-based systems. Although the total rice area is decreasing, the proportion of the rice area under irrigation is increasing. As a result of declining prices and rising costs, the profitability of rice farming has been declining. The supply of water per unit of rice area is dwindling. The contamination of water by agrochemicals, industry and municipal uses is on the rise. There has been a lack of incentive for irrigation system managers to improve performance or devolve responsibility for operation and maintenance to non-governmental entities. Poorly defined land and water rights and inadequate support systems have discouraged farmers' participation in management. Against this background the rapid growth in tubewells in areas of groundwater availability is not surprising.

In spite of these facts, there appears to be a wide range of options for increasing the productivity of water in systems served by canals and tanks. The most appropriate strategy to adopt will vary according to time and place. Substantial investment in research is needed to identify the policies, management practices and technologies that could increase the productivity of water at the farm, system and basin levels and to determine the most cost-effective strategies. Complementing this research effort is the continuing need for research in genetic improvement of rice and agronomic management, leading to higher productivity per unit of water.

REFERENCES

Ali, S.H. 1983. Water scarcity and the role of operation and maintenance: administration and farmers' organizations. In K.K. Singh, ed. Utilization of canal waters: a multi-disciplinary perspective on irrigation. Workshop on Irrigation Systems Management Related to Chak (Outlet) Requirements, Varanasi, India, July 1981. Publication No. 164, p. 40-47. New Delhi, India, Central Board of Irrigation and Power.

Amarasinghe, U.A., Sakthivadivel, R. & Murray-Rust, H. 1998. Impact assessment of rehabilitation intervention in the Gal Oya Left Bank. Research Report No. 18. Colombo, Sri Lanka, International Irrigation Management Institute (IIMI).

Bhuiyan, S.I., Sattar, M.A. & Khan, M.A.K. 1995. Improving water use efficiency in rice irrigation through wet seeding. Irrigation Sci., 16(1): 1-8.

Bird, J.D., Francis, M.R.H., Makin, I.W. & Weller, J.A. 1991. Monitoring and evaluation of water distribution: an integral part of irrigation management. In Improved irrigation system performance for sustainable agriculture. Proceedings of the Regional Workshop on Improved Irrigation System Performance for Sustainable Agriculture, Bangkok, Thailand, 22-26 Oct. 1990, p. 112-131. Rome, FAO.

Bos, M.G. & Nugteren, J. 1974. On irrigation efficiencies, Publication No. 19, p. 89. Wageningen, the Netherlands, International Institute for Land Reclamation and Improvement.

Bos, M.G. & Wolters, W. 1991. Developments in irrigation performance assessment. In Improved irrigation system performance for sustainable agriculture. Proceedings of the Regional Workshop on Improved Irrigation System Performance for Sustainable Agriculture, Bangkok, 22-26 Oct. 1990, p. 132-140. Rome, FAO.

Fertiliser Statistics of India. Various issues. New Delhi, Fertiliser Association of India.

Fujii, H. & Cho, M.C. 1996. Water management under direct seeding. In Morroka, S. Jegathesen & K. Yasumobu, eds. Recent advances in Malaysian rice production - direct seeding culture in the Muda area, p. 113-129. Ampang Jajar, Alor Setar, Malaysia. Muda Agricultural Development Authority (MADA) and JIRCAS.

Government of Bangladesh. Statistical Yearbook of Bangladesh, various years. Bangladesh Bureau of Statistics.

Government of Pakistan. 1997. Economic Survey 1996-97. Islamabad, Pakistan, Government of Pakistan, Finance Division, Economic Adviser's Wing.

Hatta, S. 1967. Water consumption in paddy field and water saving rice culture in the tropical zone. Japan Trop. Agric., 11(3): 106-112.

Ho Nai Kin, Chang, C.M., Murat, M. & Ismail, M.Z. 1993. MADA's experiences in direct seeding. Paper presented at the Workshop on Water and Direct Seeding for Rice, 14-16 June 1993. Ampang Jajar, Alor Setar, Malaysia, Muda Agricultural Development Authority (MADA).

Huke, R.E. & Huke, E.H. 1997. Rice area by type of culture: South, Southeast and East Asia. A revised and updated data base. Manila, the Philippines, IRRI.

Hutasoit, F. 1991. Monitoring and evaluation for improving irrigation system performance in Indonesia. In Improved irrigation system performance for sustainable agriculture. Proceedings of the Regional Workshop on Improved Irrigation System Performance for Sustainable Agriculture, Bangkok, Thailand, 22-26 Oct. 1990, p. 230-240. Rome, FAO.

IRRI. 1995. World Rice Statistics, 1993-94. Manila, the Philippines.

Israr, M. 1991. Improved irrigation system performance in Pakistan. In Improved irrigation system performance for sustainable agriculture. Proceedings of the Regional Workshop on Improved Irrigation System Performance for Sustainable Agriculture, Bangkok, Thailand, 22-26 Oct. 1990, p. 270-276. Rome, FAO.

Jones, W.I. 1995. The World Bank and irrigation. Washington DC, Operations Evaluation Department, the World Bank.

Keat, T.S. 1996. Country paper of Malaysia. Paper presented at the Expert Consultation on Modernization of Irrigation Schemes: Past Experiences and Future Options, Bangkok, Thailand, 26-29 Nov. 1996.

Khao-Uppatum, V. 1992. Irrigation/water management for sustainable agricultural development in Thailand. In Irrigation/water management for sustainable agricultural development. Report of the Expert Consultation of the Asian Network on Irrigation/Water Management, Bangkok, Thailand, 25-28 Aug. 1992. RAPA Publication 1994/24, p. 173-190. FAO Regional Office for Asia and the Pacific.

Kikuchi, M., Fujita, M., Marciano, E. & Hayami, Y. 1996. Deterioration of a national irrigation system in the Philippines. A preliminary appraisal. Manila, the Philippines, IRRI. (unpublished paper)

Kitamura, Y. 1990. Management of irrigation system for double cropping culture in the tropical monsoon area. Technical Bulletin No. 27. Ibaraki, Japan, Tropical Agriculture Research Centre.

Mishra, H.S., Rathore, T.R. & Pant, R.C. 1990. Effect of intermittent irrigation on groundwater table contribution, irrigation requirement and yield of rice in Mollisols of the Tarai Region. Agric. Water Management, 18: 231-241.

Molden, D. 1997. Accounting for water use and productivity. SWIM Paper No. 1. Colombo, Sri Lanka, IIMI.

Murray-Rust, D.H. & Snellen, W.B. 1993. Irrigation system performance assessment and diagnosis, p. 20, 148. Colombo, Sri Lanka, IIMI.

Postel, S.L., Gretchen, C.D. & Ehrlich, P.R. 1996. Human appropriation of renewable fresh water. Science, 271: 785-788.

Rice, E.B. 1997. Paddy irrigation and water management in Southeast Asia. Washington DC, Operations Evaluation Department, the World Bank.

Rosegrant, M.W., Ringler, C. & Gerpacio, R.V. 1997. Water and land resources and global food supply. Washington DC, International Food Policy Research Institute (IFPRI).

Rosegrant, M.W. & Svendsen, M. 1993. Asian food production in the 1990s: irrigation investment and management policy. Food Policy, 18(2): 13-32.

Sandhu, B.S., Khera, K.L., Prihar, S.S. & Singh, B. 1980. Irrigation need and yield of rice on a sandy-loam soil as effected by continuous and intermittent submergence. Indian J. Agric. Sci., 50(6): 492-496.

Seckler, D. 1993. Designing water resources strategies for the twenty-first century. Water Resources and Irrigation Division Discussion Paper No. 16. Arlington, Virginia, USA, Winrock International.

Seckler, D. 1996. The new era of water resources management: from "dry" to "wet" water savings. Research Report No. 1. Colombo, Sri Lanka, IIMI.

Singh, C.B., Aujla, T.S., Sandhu, B.S. & Khera, K.L. 1996. Effect of transplanting date and irrigation regime on growth, yield and water use in rice (Oryza sativa) in northern India. Indian J. Agric. Sci., 66(3): 137-141.

SWIM. 1997. Water saving techniques in rice irrigation. Report of the SWIM Mission to Guilin Prefecture, Guangxi Region, China, 25-30 May 1997.

Tabbal, D.F., Lampayan, R.M. & Bhuiyan, S.I. 1992. Water-efficient irrigation technique for rice. In V.V.N. Murty & K. Koga, eds. Soil and water engineering for paddy field management. Proceedings of the Inter-national Workshop on Soil and Water Engineering for Paddy Field Management, Asian Institute of Technology, Bangkok, Thailand, 28-30 Jan. 1992, p. 146-159.

Tuong, T.P. & Bhuiyan, S.I. 1997. Increasing water use efficiency in rice production: farm level perspectives. In Proceedings of the International Workshop More from Less; Better Water Management, Department for International Development, Cranfield, UK, 21-23 Sept. 1997, p. 108-111.

Upasena, S.H., Sikurajapathy, M. & Seneviratna, T. 1980. A new cropping system strategy for the poorly irrigated ricelands of the dry zone. Trop. Agric., 136: 51-58.

Vermillion, D.L. 1997. Impacts of irrigation management transfer: a review of the evidence. Research Report No. 11. Colombo, Sri Lanka, IIMI.

Viets, F.G. 1962. Fertiliser and efficient use of water. Adv. Agron., 14: 223-264.

WRI. 1994. World resources 1994-95. New York, Oxford University Press and the World Resources Institute (WRI).

Perspectivas de los recursos hídricos en los años 2020:
la investigación sobre ordenación de aguas en la producción arrocera

L'eau est devenue de plus en plus rare et de nombreux intérêts rivaux entrent en concurrence pour son utilisation. La demande d'eau destinée à l'agriculture, aux usages ménagers, à l'industrie, à l'énergie hydroélectrique et à l'environnement devrait augmenter de 35 pour cent d'ici l'an 2020. Par contre, la part affectée à l'irrigation devrait diminuer et passer de 72 pour cent à 62 pour cent, en faveur d'utilisations plus prisées. L'irrigation, du fait d'un accroissement des superficies irriguées et des rendements accrus obtenus sur les terrains déjà irrigués, a permis d'augmenter considérablement la production vivrière, dans le cadre de la révolution verte. Toutefois, les possibilités d'expansion des zones irriguées et d'exploitation de nouvelles ressources d'eau à des coûts raisonnables sont réduites. Ainsi, faudra-t-il trouver les moyens d'accroître la productivité de l'eau dans l'agriculture irriguée.

Soixante pour cent environ des terres irriguées du globe se situent en Asie, dont 80 pour cent sont consacrées à la production de riz. La mousson semble apporter beaucoup d'eau en Asie, mais les variations saisonnières et annuelles du niveau des précipitations et l'accroissement des demandes non agricoles entraînent des pénuries d'eau dans
des régions apparemment bien arrosées. Qui plus est, les possibilités de réaliser de véritables économies d'eau, grâce à une meilleure gestion des systèmes d'irrigations du riz, ne sont pas connues et sont peut-être plus limitées que les prévisions de nombreux observateurs. Il est nécessaire d'investir des sommes importantes dans le domaine de la recherche pour pouvoir améliorer le rendement des ressources en eau au niveau des bassins, et pour évaluer les coûts des nouvelles stratégies. Cet effort fait pendant à la nécessité de recherches orientées vers l'amélioration génétique du riz, afin d'obtenir des rendements plus élevés par unité d'eau.

Perspectives en ce qui concerne les ressources en eau en 2020:
défis à relever pour la recherche sur la gestion de l'eau aux fins de la riziculture

Ha comenzado una época de aumento de la escasez de agua y de competencia por la misma. La demanda de agua para usos agrícolas, uso doméstico, industria, energía hidroeléctrica y el medio ambiente aumentará, según se prevé, un 35 por ciento desde ahora hasta el año 2020. Las previsiones son que la proporción de agua empleada en riego bajará del 72 al 62 por ciento al desviarse el agua para dedicarla a empleos más valiosos. El riego (tanto debido a la ampliación de los regadíos como al aumento de los rendimientos en la superficie de riego actual) ha contribuido a gran parte del aumento de la producción de alimentos que se ha verificado como consecuencia de la revolución verde. No obstante, las oportunidades para ampliar el riego y desarrollar nuevos recursos hídricos a un costo razonable son de carácter limitado. Por consiguiente, hay que encontrar la forma de aumentar la productividad del agua en la agricultura de regadío.

Aproximadamente el 60 por ciento de la superficie de regadío en el mundo se halla en Asia, dedicándose un 80 por ciento de ella a la producción de arroz. Si bien la región monzónica de Asia parece contar con agua suficiente, las variaciones estacionales y anuales del régimen de lluvias y la creciente demanda de empleos no agrícolas están dando lugar a escaseces de agua en zonas aparentemente bien regadas. Por otra parte, las posibilidades de conseguir ahorros reales de agua mejorando la ordenación de los sistemas de riego basados en el arroz no son conocidas y tal vez sean bastante más limitadas de lo que muchos observadores creían en un principio. Hacen falta sustanciosas inversiones en investigación para mejorar la eficiencia de los recursos hídricos a nivel de cuenca y evaluar el costo de otras posibles estrategias. Como complemento a este esfuerzo está la constante necesidad de investigación sobre mejoramiento genético del arroz, con el fin de obtener un aumento de los rendimientos por unidad de agua.