The following quotations from D. Seckler, former Director of the International Water Management Institute (IWMI), are characteristic of statements by many experts and of the tenor of the discussions at the World Water Forum held in March 2000:
"For most of modern history, the worlds irrigated area grew faster than population, but since 1980 the irrigated area per person has declined and per capita cereal grain production has stagnated. The debate regarding the worlds capacity to feed a growing population, brought to the fore in the writings of Malthus two centuries ago, continues. But the growing scarcity and competition for water add a new element to this debate over food security. ... In a growing number of countries and regions of the world, water has become the single most important constraint to increased food production" (Seckler et al., 1998).
"Many of the most populous countries of the world - China, India, Pakistan, Mexico, and nearly all of the countries of the Middle East and North Africa - have literally been having a free ride over the past two or three decades by depleting their groundwater resources. The penalty of mismanagement of this valuable resource is now coming due, and it is no exaggeration to say that the results could be catastrophic for these countries, and given their importance, for the world as a whole" (Seckler et al., 1999).
Water scarcity is recognized increasingly as a global concern and, within that broad concern, more attention is focusing on emerging patterns of groundwater overexploitation and their implications for the availability of water to meet human and environmental needs. Food production and food security are among the more important of these needs. Reliable water supplies, particularly those from groundwater, are the lead input for increasing yields, reducing agricultural risk and stabilizing farm incomes. As a result, strong arguments can be made that access to groundwater plays an instrumental role in food security.
Water availability and reliability are linked closely to food security, but the equation linking water to food security is partial and the links are neither linear nor transparent. The full equation is a function of the interaction between water access, production economics and the wider network of entitlements that water users and others have within society. It cannot be assumed that a one-to-one relationship exists between access to reliable water supplies for irrigated agriculture and food security (Plate 2).
According to technical documents prepared for the World Food Summit in 1996, the main generally available indicator of food security is: "per caput food consumption, measured at the national level by the average dietary energy supply (DES) in calories on the basis of national food balance sheets (FBS) and food supplies as national averages" (FAO, 1996). The FAO definition of food security (above) that this paper follows does not focus on food production and physical availability alone; it also includes the critical dimension of access to available food supplies. Under the definition, food security often depends more on the ability of populations to purchase rather than produce food. This is because global and national food distribution systems now frequently negate the impact of local production problems on the availability of food in the market. As a result, the question of whether people have access to sufficient food when groundwater problems disrupt agricultural production depends heavily on whether they have access to a diverse array of alternative income sources or to reserve capital. It also depends on wider factors such as transportation systems and the ability of countries to purchase and distribute food available on global markets. All this implies that analysis of groundwater availability and reliability on a project or regional basis is by itself a poor indicator of the vulnerability of the global population to food insecurity (Figure 1).
Plate 2 - Pumping in Gujurat, India. One farmer + one pump = food security - or does it
FIGURE 1 - Number of undernourished in the developing world: observed and projected ranges compared with the World Food Summit target
Source: The State of Food Insecurity in the World. FAO, 2002
Nonetheless, access to water, and particularly to highly reliable groundwater sources, does play an important role in food security in many cases. Access to reliable sources of water reduces the production risk. Farm incomes at both micro (farm) and aggregate (regional) levels are buffered from the effects of precipitation variability, drought or general water scarcity conditions. As a result, access to reliable groundwater supplies can ensure the income flow needed to purchase food as well as playing a central role in food production. Furthermore, particularly in remote locations within developing countries, irrigated agriculture constitutes the sole source of income that is available to rural populations. As a result, there can be a direct link between water access and household or regional food security. However, this link is highly dependent on the specific situation. There is no inherent direct link between water and food security. While access to water is important in many situations, in other situations irrigated agriculture is only one of many income sources or available livelihood strategies. Consequently, while falling water levels, irrigation system deterioration, droughts and other direct indicators of water scarcity can serve as signals that food security may be threatened, the actual degree of threat will depend on a wide variety of context-specific factors. Water scarcity measures are warning signals, but they do not on their own indicate the emergence of food insecurity.
The most direct and tangible link between groundwater conditions and food security is that of water availability to meet crop requirements. However, water availability in an aggregate sense has little meaning as crop production is heavily dependent on seasonal and interannual fluctuations in availability, including timing in relation to crop growth stages. Many crops are vulnerable to moisture stress at critical points in plant growth, and their yields can be reduced substantially even if adequate water supplies are available following periods of shortage (Perry and Narayanamurthy, 1998). For example, water stress at the flowering stage of maize can reduce yields by 60 percent even where water is adequate throughout the rest of the crop season (Seckler and Amarasinghe, 1999). Similar impacts on onions, tomatoes and rice have also been documented (Meinzen-Dick, 1996). In addition to the direct impact of water availability on crop growth, assured supplies are a major factor in inducing investment in other production inputs such as labour, fertilizers, improved seeds, and pesticides (Seckler and Amarasinghe, 1999; Kahnert and Levine, 1989). As a result, as the reliability of irrigation water supplies increases there is multiplier effect on yields. Taken with the inherent flexibility of groundwater abstraction (on demand, just in time), these characteristics of groundwater were a major contributor to the role of irrigation in the green revolution. Irrigated agriculture now contributes almost 40 percent of world food production from 17 percent of cultivated land (United Nations, 1997).
Expansion of irrigation was the lead input driving yield increases during the green revolution of the 1960s-70s and subsequent decades. As the most reliable source of irrigation water, a source that can generally be tapped when and in the amounts needed, groundwater played a particularly major role. As Repetto (1994) comments: "The Green Revolution has often been called a wheat revolution; it might also be called a tubewell revolution." To this extent, this turnaround hinged upon high-value crops (with high crop-water budgets) and the ability to pay for energy costs (Plate 3).
Yields in groundwater-irrigated areas are higher (often double) compared to those in canal-irrigated areas (Shah, 1993; Meinzen-Dick 1996). In India, the groundwater-irrigated area accounts for about 50 percent of the total irrigated area and up to 80 percent of the countrys total agricultural production may, in one form or another, be dependent on groundwater (Dains and Pawar, 1987). Similar patterns are also present in other countries. In Chinas Henan province, tubewells serve about 2 million ha, or 52 percent of irrigated lands (FAO, 1994). Parts of Mexico, including some of its most productive agricultural areas, are also heavily dependent on groundwater. The role of groundwater is equally important in industrialized countries. For example, Barraque (1998) estimates that: "irrigation uses 80 percent of all water in Spain and 20 percent of that water comes from underground... The 20 percent, however, produces more than 40 percent of the cumulated economic value of Spanish crops." Recent findings from Andalusia, Spain, indicate that groundwater-irrigated agriculture is economically more than five times more productive (in terms of revenue per cubic metre) and generates more than three times the employment in comparison with surface-irrigated agriculture (Hernandez-Mora et al., 1999). The role of groundwater is important not only through higher yields in normal water years. In an analysis of wheat cropping in the Negev Desert, Tsur (1990) estimated the stabilization value (the value associated with the reliability of the water supply as opposed to just the value of the volume available of groundwater development) as being "more than twice the benefit due to the increase in water supply". In southern California, the United States of America, where surface water supplies are less variable than in the Negev Desert, the stabilization value in agriculture is as much as 50 percent of the total value of groundwater in some cases (Tsur, 1993). During the drought in California, the United States of America, in the early 1990s, economic impacts were minimal largely because farmers were able to switch from unreliable surface supplies to groundwater (Gleick and Nash, 1991). The value associated with the flexibility of pumped groundwater supplies has been a further boost to agricultural productivity as it has allowed intensification and diversification of agricultural production in otherwise inflexible surface-irrigation schemes.
However, the presence of groundwater irrigation alone cannot be given full credit for the increased yields documented around the world. It needs to be seen as part of a complementary and mutually reinforcing set of inputs. Groundwater availability has enabled farmers to invest in complementary inputs that, in combination, have increased crop yields substantially. As FAO (2002b) notes: "the response of crop to fertilizer is higher where supply of irrigation water is assured compared to rainfed conditions." It is the reliability and flexibility of groundwater that allows farmers to take the risk of investing in fertilizer, but which also substantially increases their crop productivity. For example, fertilizer use in Pakistan is highest in areas supplied by both canals and tubewells and thus having a highly assured supply of irrigation water. The total nutrient application in these areas is 68.85 kg/ha compared to 29.0 kg/ha in rainfed areas (FAO, 2002b). These observations point to the dependency of crop yields on interactions within a dynamic agricultural system and the difficulty of isolating a single factor as the primary factor contributing to increased production.
Plate 3 - Extensive groundwater irrigation of wheat from large stratiform aquifers. Sadah, Yemen
Nonetheless, the information available indicates the critical role groundwater has played in agricultural production over recent decades. The relationship between assured supplies of irrigation water, increasing yields and food production is now under stress. According to Rosegrant and Ringler (1999): "the growth rate in irrigated area declined from 2.16 percent/year during 1967-82 to 1.46 percent in 1982-93. The decline was slower in developing countries, from 2.04 percent to 1.71 percent annually during the same periods." Yield increase rates are also declining, and projections indicate that this will continue in coming decades (Rosegrant and Ringler, 1999; FAO, 2000). Furthermore, in some local areas such as Sri Lanka and in the rice-wheat systems of India, Nepal, Pakistan and Bangladesh, yields have been stagnant for a number of years (Amarasinghe et al., 1999; Ladha et al., 2000).
Although stresses on water resources are increasing and there is a logical link between water scarcity and yield stagnation, causal relationships between emerging water problems, yields and food production vulnerability are not proven. According to Ladha et al. (2000), where yield stagnation is concerned: "There is some evidence of declining partial or total factor productivity...The causes for the stagnation or decline are not well known, and may include changes in biochemical and physical composition of soil organic matter (SOM), a gradual decline in the supply of soil nutrients causing nutrient (macro and micro) imbalances due to inappropriate fertilizer applications, a scarcity of surface water and groundwater as well as poor water quality (salinity), and the buildup of pests, especially weeds such as Phalaris minor."
Furthermore, as Seckler and Amarasinghe (1999) note: "It is very difficult to project crop yields. ... The international dataset does not distinguish between yields on irrigated and rain-fed area: they are just lumped together in average yields." Water is only one factor affecting crop yields. Data available at the global level do not provide much insight into the relationship between yields on irrigated and rainfed lands, or enable conclusions about yields on areas irrigated by groundwater or on areas where groundwater depletion is occurring. Recent evaluations of the implications of water scarcity for food security range from the optimistic to the pessimistic. For example, Brown (1999) contends that primarily because of impending water shortages in northern China, the country will have to import up to 370 million tonnes of grain per year to feed its population in 2025. This massive increase in imports could cause steep increases in cereal prices and disruption of the world market (Seckler et al., 1999). On the other hand, analyses by FAO and the International Food Policy Research Institute (IFPRI) indicate that yield increases (rather than increases in cultivated area) will be the dominant factor underlying growth in cereal production in the coming decades and that, in aggregate, production increases will be sufficient to meet demand (Rosegrant and Ringler, 1999; FAO, 2002a). The FAO (2002a) report states that: "The overall lesson of the historical experience, which is probably also valid for the future, seems to be that the production system has so far had the capability of responding flexibly to meet increases in demand within reasonable limits." (Plate 4).
The core point in this discussion of the direct links between groundwater availability and food production is the role of interacting dynamic systems and the uncertainty inherent in predicting outcomes based on partial understanding of any one of them. The role that irrigation, particularly groundwater irrigation, has played in increasing yields is relatively clear. Whether or not emerging water problems are a significant factor underlying the declining rate of yield increases or represent a significant threat to overall production levels is less well documented. While the potential nature of such connections is clear in concept, available data and other evidence are insufficient to test the conceptual relationships. While it is essential not to dismiss the implications of groundwater overabstraction and water scarcity for food production simply because data are insufficient to prove them, it is equally essential not to ignore the wide range of other factors that could be playing equal or greater roles. Therefore, the first part of the equation linking groundwater and food production is clouded even before investigation of the larger question of the role groundwater plays in entitlements and food security.
Food security is a function of three factors: availability; stability; and the ability of individuals to obtain access to food. As Sen (1999) and others (Dreze et al., 1995) have argued for famines in India, starvation is frequently due to the inability of individuals to purchase supplies that are readily available on the market and is not a function of food availability per se. The entitlement approach described by Sen "views famines as economic disasters, not just as food crises." Sen indicates that the main interest in the entitlement approach probably lies in "characterizing the nature and causes of the entitlement failures where such failures occur."
Sens approach may have particular relevance for analysing the impact of emerging groundwater problems on food security. Studies in the late 1980s highlighted the critical role that access to water, particularly groundwater, plays in poverty alleviation (Chambers et al., 1989). Reliable water supplies are a foundation that enables farmers to afford access to a wide range of development benefits (from food to education and health services) and can also enable farmers to diversify into other, often non-agricultural, income sources. These benefits are accessed through the improved yields enabled by the green revolution package of inputs. However, they carry a substantial risk because farmers must make investments in fertilizer, seed and other inputs in order to achieve them. These investments, which are often made on credit, will be lost if water supplies fail. Consequently, any decline in access to groundwater could have a major impact on the economic condition of small rural farmers. As Burke (2000) argues: "the expansion of irrigated agriculture in the 20th century has de-coupled the water user from the inherent risk of exploiting both surface and groundwater resources. The apparent reliability of storage and conveyance infrastructure and the relative cheapness and flexibility of groundwater exploitation offered by mechanical drilling have sheltered the end user from natural hydrological risk." If substantial groundwater-level declines occur, short-term risk exposure may return to levels not encountered since the spread of irrigation. This risk is predominantly economic.
The economic dimension is also central to understanding the meaning of groundwater overextraction. Most discussions of groundwater overabstraction emphasize the distinction between economic depletion (i.e. falling water levels make further extraction uneconomic) and the actual dewatering of an aquifer. Large-scale aquifers are depleted in an economic sense (the physical limits to pumping and associated energy costs) long before there is any real threat of physical depletion. The Gangetic basin may have 6 000 m of saturated sediment, but only the top 100 m or so are economically accessible for irrigation. Furthermore, wells owned by small farmers are generally shallow. In the context of poverty and famine, falling water tables will tend to exclude those farmers who cannot afford the cost of deepening wells long before they affect water availability for wealthy farmers and other affluent users (Moench, 1992). Consequently, substantial declines in water levels are particularly likely to have a major economic impact on farmers with limited land and other resources. This impact will tend to be particularly pronounced during drought periods when large numbers of small farmers could simultaneously lose access to groundwater as their wells dry up. A more creeping problem would occur during non-drought periods as water-level declines undermined the economic position of small marginal farmers, forcing them onto already saturated unskilled agricultural and urban labour markets. The food security crisis in both these situations would be economic rather than related to foodgrain availability per se. Furthermore, whether there actually is a food security problem would depend as much on larger economic conditions (specifically the opportunities available to farmers transferring out of agriculture into other activities) as on groundwater availability and the economics of agriculture (Plate 5).
Plate 4 - The physical and economic limits to pumping, Eritrea
Plate 5 - Tension between farmers and municipalit users. Taire, Yemen
The question of the larger economic situation is particularly relevant in the context of global demographic and economic changes. Although the latest UN assessment indicates a substantial deceleration in world demographic growth rates, the absolute annual increments in the coming decades will continue to be large. According to FAO (2000): "seventy-seven million persons are added to world population every year currently. The number will not have decreased much by 2015. Even by 2030, annual additions will still be 58 million." Ninety-eight percent of the increase between 1995 and 2020 will occur in the developing world with the largest absolute growth concentrated in Asia and the highest relative increases occurring in sub-Saharan Africa (Pinstrup-Anderson et al., 1999). Population growth will be accompanied by significant changes in where people live. Historically, rural populations have dominated those living in urban areas. However, within the next 15 years, the urban population in developing countries is projected to surpass the rural population (Pinstrup-Anderson et al., 1999). Furthermore, as populations urbanize, their aspirations and food-demand characteristics will change. Such changes are reflected in recent food trade and demand projections. Over the next 20 years, according to Rosegrant and Ringler (1999): "Per capita food consumption of maize and coarse grains will decline as consumers shift to wheat and rice, livestock products, fruits and vegetables, and processed foods. The projected strong growth in meat consumption, in turn, will substantially increase cereal consumption as animal feed, particularly maize. Growth in cereal and meat consumption will be much slower in developed countries. These trends will lead to a strong increase in the importance of developing countries in global food markets: 82 percent of the projected increase in global cereal consumption, and nearly 90 percent of the increase in global meat demand between 1993 and 2020 will come from developing countries. Developing Asia will account for 48 percent of the increase in cereal consumption, and 63 percent of the increase in meat consumption."
To meet changing demand patterns, FAO and the IFPRI project substantial increases in world trade for food, particularly cereals and meat products (FAO, 2000; Rosegrant and Ringler, 1999). Cereal trade is projected to almost double and meat trade to triple by 2020. To date, traditional cereal exporters (North America, Australia, Argentina, Thailand, Western Europe and Viet Nam) have been able to meet sudden rises in demand in developing countries. However, Brown (1999) points out that grain exports by the principal exporting countries (accounting for 85 percent of world exports) have levelled off since 1980. There is debate as to whether developing countries will be able to meet food needs through trade. However, it could also be argued that the high population growth in some water-stressed developing countries (e.g. Jordan and Palestinian Authority) shows that food production is not a limit to food security.
Returning to the question of the link between groundwater conditions and food security, Sens framework suggests that access to groundwater will continue to play a critical role in the network of entitlements that determine food security for rural agricultural populations. However, as populations migrate from rural to urban areas, direct access to groundwater for individuals will play less of a role. This is also the case where rural economies become less dependent on agriculture. Furthermore, the food security impact of groundwater-level declines on rural agricultural populations will depend as much on their ability to join the stream of permanent or temporary migrants to urban areas as on their ability to maintain economic livelihoods in rural areas. On an anecdotal level, this dynamic is evident in discussions with farmers in diverse conditions. For example, in many interviews with farmers in Gujarat, India, concerning groundwater overabstraction and the possibility of developing management systems, discussions have elicited the following type of response: "Yes, I know falling water levels will drive me out of production in a few years, but why should I care? The income I am generating now is enabling my children to study for an engineering degree; we will not be here in five years time." Farmers in the United States of America often express similar sentiments. A young farmer in the San Luis Valley in Colorado, the United States of America is 65 years old; a wide set of economic and social factors has induced many young people to prefer a livelihood in the urban or non-agricultural economy. From a food security perspective, they have joined the half of the worlds population that depends on global economic, production and distribution systems within which groundwater availability is only one element. In this sense, they no longer depend on direct individual access to local resources such as groundwater.
For urban residents and the increasing population not engaged in agriculture, food security is likely to become a function of distant production and distribution systems combined with the economic context individuals find themselves in. This implies that food security for many will be influenced at least as much by conditions in the wider economy as by factors such as groundwater conditions that affect agricultural economics and local agricultural production per se.
The discussions above point to the complex nature of the interactions between groundwater conditions and a large number of factors in the wider global economic and demographic context that influence food security. Full analysis of these factors is beyond the scope of this paper. However, recognition of the complexity and identification of points of leverage within it is critical to any meaningful analysis of the implications of groundwater overabstraction for food security. The complexity is also central to identifying meaningful responses to emerging water problems. Many compelling analyses of environmental-social relationships have foundered on seemingly minor gaps in data or system descriptions.
The next section of this paper focuses on groundwater: how well the resource base and emerging overabstraction problems are understood and, beyond that, the implications of emerging problems for food production and security. However, before that, Boxes 1 and 2 present two cautionary examples that highlight the fundamental risk inherent in posing major global consequences where there is partial or weak scientific understanding and where the systems involved are complex.
The cautionary examples in Boxes 1 and 2 contain lessons that are central to the question of evaluating the impacts of groundwater overabstraction on food production and food security.
BOX 1: Rain follows the plough
In the last decades of the nineteenth century, throughout the western United States of America, settlers received land grants of 160 acres (about 65 ha) under the Homestead Act. The High Plains, an area of rolling grasslands between the Mississippi River and the Rocky Mountains were a focal point for settlement. Each year, the waves of settlers drove slightly further west, claimed land and broke the sod. In one decade, nearly 2 000 000 people settled on the Great Plains.1
"God speed the plow.... By this wonderful provision, which is only mans mastery over nature, the clouds are dispensing copious rains... [the plow] is the instrument which separates civilization from savagery; and converts a desert into a farm or garden.... To be more concise, rain follows the plow". (Charles Dana Wilber)2
"Rain Follows the Plow" was a common headline on brochures promoting settlement. This statement was based on accepted scientific analysis of the day. Soil beneath the grasslands was rich and often very moist. Ploughing it would release substantial moisture. According to the theory, as more land was brought under cultivation, more moisture would be released. This would, in turn, contribute to cloud formation and ultimately cause rainfall to increase. The High Plains could be converted into a climate resembling the temperate moist areas of the east coast. This was science, an integrated theory grounded on an apparent understanding of the physical processes incorporated in the model. Furthermore, the climate behaved as predicted. Unusually heavy rainfall in the 1870s and early 1880s, made the claims sound plausible. Rainfall in the High Plains appeared to increase as agricultural areas grew. The story declined as rainfall returned to the lower levels common throughout much of recent history. The endnote was the Dust Bowl, that great event that reshaped much of rural America in the 1930s.
Where did the analysis go wrong? First, although the processes were understood, at least at a gross level, the orders of magnitude on the flows involved were not. Ploughing may release moisture from the soil, but the amounts involved were far too small to affect the regional climate. Second, the model neglected interactions with other systems. Regional and hemispheric wind patterns are such that any water released from ploughing soil in the High Plains is often transported huge distances before it reaches the ground as running water again. Rain did not follow the plough, it only appeared to.
First, the database relevant to the question being asked was weak in both cases. This was natural enough in the 1870s-1880s when long-term rainfall records were unavailable and changes were noted primarily on the basis of the personal observations of settlers (a situation with many parallels to the current groundwater debate). However, the database was also weak in the case of Himalayan deforestation. Despite massive increases in information gathering technologies, good historical records of forest cover were generally unavailable and data relevant to critical components of the model, such as suspended sediment and baseload transport in rivers, remain inadequate to this date. In many ways, the challenge was one of recognizing that, despite the large amounts of information available, core data relevant to the questions being asked were absent.
BOX 2: Himalayan deforestation
One of the most compelling recent environmental stories was that surrounding deforestation in the Nepal Himalayas. In a classic analysis, Eckholm (1976) painted a picture of environmental degradation in the hills having major regional consequences. The model was clear. It envisioned a direct cause-and-effect relationship between population growth, mountain deforestation and lowland flooding. "As wood scarcity forces farmers to burn more dung for fuel, and to apply less to their fields, falling food output will necessitate the clearing for ever larger, ever steeper tracts of forest - intensifying the erosion and landslide hazards in the hills, and the siltation and flooding problems downstream in India and Bangladesh." (Eckholm, 1976). The interaction between population, food and fuel lay at the heart of deforestation problems. As forest cover declined, erosion and the speed of runoff increased. These, in turn, increased sediment loads, caused riverbeds in the Indian plains to aggrade, and led to increases in flooding throughout the Gangetic basin and the growth of islands in the Ganges Delta of Bangladesh.
The model was integrated and the database relatively strong. As Ives and Messerli (1989) stated: "the most compelling and trend-setting characterization of the Himalayan region and its anticipated eco-disaster is that published by Erik Eckholm (1975, 1976)...". Forest cover and flooded areas were being monitored through satellite imagery. Stream gauges were, at least in some locations, in place and had long-term monitoring records. Furthermore, as with the conceptual foundations for the rain-follows-the-plough model, the physical processes were relatively well understood. However, as with that model, interactions with other systems, in this case plate tectonics and the expansion of population into marginal agriculture, were ignored or at least underestimated. As Ives and Messerli pointed out: "The large literature that depicts the imminence of environmental catastrophe in the Himalayan region has tended to confuse cause and effect, has largely missed the essential historical depth, and has assumed the existence of dramatic upstream-downstream interrelationships without requiring rigorous factual substantiation." Subsequent research demonstrated that the causal links between many elements in the model were weak and it was far from clear that forest declines were anywhere near as widespread as portrayed. Furthermore, even if deforestation was causing erosion, natural erosion rates related to tectonic uplift were orders of magnitude higher than the human contribution. Ultimately, the whole basis of understanding Himalayan deforestation came under question.
"The wide uncertainties that currently exist at the bio-physical level - uncertainty as to whether the consumption of fuelwood exceeds or is comfortably within the rate of production, uncertainty as to whether deforestation is a widespread or localized phenomenon, uncertainty as to whether it is population pressures or inappropriate institutional arrangements that lie behind instances of mismanagement of renewable resources ... uncertainty as to whether deforestation in the hills (if it indeed exists) has any serious impact on the flooding in the plains - means that a wide range of mutually contradictory problems are credible." (Thompson et al., 1986).
In sum, the system and the interactions between systems were too complex and poorly understood to be captured adequately. Furthermore, the uncertainty created opportunities for groups in society to advance agendas that matched their worldviews. The international community made major investments in reforestation on the basis of the Himalayan deforestation model. It was practical and pointed to things organizations could do or invest in in order to solve problems. It not only created a problem for organizations to remedy but a problem for them to perpetuate as a means of defining themselves. Ives and Messerli pointed this out: "we must emphasize again that this uncertainty is not merely technical; that it is not just the absence of certainty. Rather, it is structural in the sense that, without their realizing it, certain actors in the Himalayan debate have succeeded in imposing their desired uncertainties within it." Perceptions of environmental degradation became so ingrained in the way organizations approached the Himalayan dynamic that the problems and solutions began to feed on each other. Thompson et al. saw it as "generated by institutions for institutions. The survival of an institution rests ultimately upon the credibility it can muster for its idea of how the world is; for its definition of the problem; for its claim that its version of the real is self-evident." The model even became a major factor in arguments by India for the construction of high dams in the Himalayas. These were portrayed as essential to control floods, the fact that the dams would produce large amounts of electricity which India wanted and Nepal could sell was an added benefit.
As the model of Himalayan deforestation has come under question, the interest of international donors in financing reforestation and watershed work has waned. Doubts regarding the Eckholm model led many to question the importance of forestry work in Nepal. However, whether or not they cause flooding in Bangladesh, forestry problems in Nepal are major and have a direct effect on the livelihoods of local populations. Linking these local problems to a regional crisis model may have had short-term benefits where work on forestry was concerned. However, it may have also undermined the long-term focus essential to addressing the real local problems that afflict populations and the forests they depend on in Nepal. This is also a risk in any model posing groundwater overdraft as a major threat to food security.
Second, both the cases involved the use of explanatory models based on a partial understanding of systems. Although accurate, the physical-process elements underlying the models were partial. As a result, their predictive value was weak. This is also the case with most groundwater systems. Models with a strong predictive capacity are unavailable in all except the most rigorously monitored and analysed aquifers. These aquifers tend to be in wealthy locations such as the Central Valley of southern California, the United States of America, and not in developing countries. Furthermore, even rigorous monitoring and analysis may not enable accurate evaluation of aquifer water availability. For example, in the San Luis Valley of southern Colorado, the United States of America, there is a more than 30-percent gap in water balance estimates despite 40 years of monitoring and analysis driven by litigation over water availability. Experts believe this gap may be related to deep inflow from outside the basin or inaccurate estimates of evapotranspiration from native vegetation, but no one knows for sure. Problems of this type would be exacerbated under conditions in developing countries where groundwater monitoring is a relatively new phenomenon.
Third, as with the case of Himalayan deforestation, answers to the question of the impact of groundwater overabstraction on food security hinge on complex interactions between water-resource, economic and social systems (Plate 6). It is unlikely that all three of these systems are understood to a sufficient degree of precision to develop definitive management responses. Furthermore, even if the systems were understood, interactions between non-linear systems often produce unpredictable and counterintuitive results. This is particularly evident in recent debates regarding the effect of vegetation on stream flows. In South Asia, re-vegetation of watersheds is widely advocated as essential for regenerating springs and river flows. However, studies in Australia document rises in groundwater levels (and the destruction of pastureland through waterlogging) due to removal of tree cover (Moench, 1998). The effects of vegetation on water availability depend on the delicate balance between recharge and evapotranspiration. Improvements in soil characteristics and reductions in runoff associated with vegetative cover generally enhance recharge. At the same time, the vegetation requires water to survive, and evapotranspiration increases. Which dominates depends on a wide variety of factors: species, wind speeds, temperature, soil types, etc. Significantly different outcomes commonly emerge from subtle interactions between such factors in local contexts.
All of the above point to the inherent risks in attempts to link real local problems to global consequences of dubious clarity: it is known that groundwater overabstraction is a major problem in specific regions. Such problems have substantial environmental, economic and other consequences whether or not they have direct implications for global food security. The danger in focusing on macro food-security concerns is that, if these concerns prove open to question, attention will be diverted from groundwater problems that are important in their own right. Furthermore, as in the case of Himalayan deforestation, approaches designed to respond to global problems often obscure responses that could be more effective but would only emerge if the problems were defined in a different way. For example, approaching groundwater overabstraction from the perspective of global food security will tend to focus efforts into global and national attempts to manage the resource base and control use in ways that maintain local and, by implication, global production. In contrast, if groundwater overabstraction is viewed as more of a regional concern, then approaches that encourage people to adapt to scarcity by migrating or transferring out of agriculture (rather than attempting to maintain production levels) could prove viable.
Plate 6 - Effluent groundwater seepage maintaining baseflows in Nepal
In order to evaluate whether the above cautions apply to the debate on groundwater overabstraction and food security, Chapter 3 examines the extent to which emerging groundwater problems are understood and the nature of the available data in key regions.