A Water Resources Threshold and Its Implications for Food Security

Jun 12, 2003 - However, the next 30 yr may see many poor and populous countries dropping below the threshold in association with their rapid populatio...
0 downloads 8 Views 209KB Size
Environ. Sci. Technol. 2003, 37, 3048-3054

A Water Resources Threshold and Its Implications for Food Security HONG YANG,* PETER REICHERT, KARIM C. ABBASPOUR, AND ALEXANDER J. B. ZEHNDER Swiss Federal Institute for Environmental Science and Technology (EAWAG), Ueberlandstrasse 133, P.O. Box 611, CH-8600 Duebendorf, Switzerland

Cereal import has played a crucial role in compensating local water deficit. A quantitative account of water deficit and cereal import relations therefore is of significance for predicting future food import demand and formulating corresponding national and international policies. On the basis of data for countries in Asia and Africa, we estimated a water resources threshold with respect to cereal import. Below the threshold, the demand for cereal import increases exponentially with decreasing water resources. There appeared to be a declining trend in the threshold, from 2000 m3/(capita year) in the early 1980s to 1500 m3/(capita year) by the end of the 1990s. Until recently, most countries below the threshold were oil-rich and thus were able to afford cereal import. However, the next 30 yr may see many poor and populous countries dropping below the threshold in association with their rapid population growth and the depletion of fossil groundwater. Water deficitinduced food insecurity and starvation could intensify because cereal import may not be affordable for these countries.

Introduction On the world average, agriculture uses about 70% of the total water withdrawals (1), making it by far the largest water user among all sectors. This leads to an intrinsic relationship between a country’s renewable water resources and the capacity for food production. In water-scarce countries, an increasing amount of food has been imported to substitute local water demand for food production. The water that is required for producing the imported food is termed “virtual water” (2). Of the food imported, cereal grains have been the dominant commodities in terms of the quantity and importance for food security to the importing countries. In essence, cereal grains have been major carriers of virtual water in the countries where the resource is scarce (3). Despite the increasing awareness of the constraints of water scarcity on food production and the imperative of food import for water-scarce countries, the relationship between water deficit and food import has mainly been observed empirically and described qualitatively. To the best of our knowledge, no study has so far modeled and rigorously tested the relationship. With the aggravation of water scarcity in many countries and regions in the world and a growing number of the people experiencing water stress (1), a quantitative development of this relationship becomes necessary. The modeling result can help project the scale of the demand for food import that is induced by water deficit * Corresponding author phone: +41-1-8235568; fax: +41-18235375; e-mail: [email protected]. 3048

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 14, 2003

in the coming years. The information can also be useful to the governments in the countries concerned and the international food agencies for developing policies to meet the challenges ahead. Against this background, in this study we took a novel approach to derive quantitatively a threshold with respect to cereal grain import. Below the threshold, lack of water resources is a limiting factor to local food production, and cereal grains have to be imported to compensate for the water deficit. On the basis of this threshold, we project the potential scale of water deficit-induced cereal import in the next 30 yr and address some of the implications for the world food security and international trade.

Data Renewable freshwater resources of a country are conventionally defined as the sum of the mean annual surface runoff and groundwater recharge (1, 4, 5). This represents the amount of water that can be withdrawn on an annual basis without violating the concept of sustainability. Water availability can vary yearly and also between different regions within a country. Accounting for this temporal and spatial variability is important for an accurate description of a country’s potential for food production and import needs. In this study, however, we use the yearly average water availability and treat a country as a unit. This limits the degree to which scatter in the residuals of a deterministic model can be reduced. The reason to use country-based data is that most of the available data are annual and country-based, and water transfer is easier to be quantified across political boundaries than within a country. Countries in Asia and Africa are taken for the analysis because water-scarce countries are mostly concentrated in these two continents (6, 7). Many countries in these two continents are net importers of cereal grains. In the late 1990s, the annual net cereal grains imported into the two continents amounted to over 110 million tons, absorbing all the surplus of the rest of the continents (8). Asia and Africa are also home to a majority of the people living in food insecurity and poverty (7, 9). In contrast, North America, Europe, and Oceania are the continents of cereal surplus. South America is more or less self-sufficient. Most of the countries in these continents have a small trade volume of cereal in relation to their domestic production with a few having large surplus and dominating the world export market (8). Except for Denmark, Poland, Bulgaria, and Belgium in the temperate Europe and a number of small island states, all the countries in these continents have water resources over 2200 m3/(capita year), excluding them from the water-scarce country list in any existing classification (9-11). As the aim of this study is to elaborate the relationship between water scarcity and induced cereal import, countries in North and South America, Europe, and Oceania are not considered. Because cereal production and import fluctuated significantly from year to year due to variations in weather and market conditions, 5-yr running averages were calculated for the data from 1980 to 2000. This resulted in 17 partially overlapping data periods. The investigation focuses on countries with more than 1 million inhabitants. This is to reduce the effect of specific local conditions of small countries on the analysis of the general relationship between water deficit and cereal import. Meanwhile, water availability of 5000 m3/(capita year) is used as a cutoff value to guarantee that the water-scarce countries are considered in the analysis while allowing for a comparison with water-abundant countries. All the data used for the simulation are from the 10.1021/es0263689 CCC: $25.00

 2003 American Chemical Society Published on Web 06/12/2003

FIGURE 1. Fits of model 1 for the first (1980-1984; thick dashed blue line and open circles) and the last (1996-2000; thick solid blue line and filled circles with country names) investigation periods. Arrows in the diagram indicate movements of the positions of the countries from the first to the last period. The positions of the parameters a and c of the model are indicated by horizontal and vertical thin lines, respectively (dashed for the first period, solid for the last period). A fit of the nonparametric model (LOESS with span ) 0.75) (13) shows that the parametric model is adequate (thin dashed and solid red lines, respectively, for the first and last periods). A Monte Carlo analysis suggests that the model results are insensitive to moderate variations in the cutoff value of 5000 m3/capita. World Resources Institute (10), the Food and Agricultural Organization of the United Nations (8), and the World Bank (12).

resources” and countries with water resources below the threshold as water-deficient countries.

(1)

The fits of model 1 to the data for the first (1980-1984) and the last (1996-2000) investigation periods are in close agreement with the nonparametric regression (Figure 1). This demonstrates that model 1 is a reasonable parametrization for describing the cereal import as a function of available water. The value of the coefficient of determination (R 2) of around 0.45 suggests that the model explained about half of the variation in net cereal import. The dependence of parameter c on only a small number of the data points, however, leads to a rather high uncertainty of this parameter. Nevertheless, a t-test indicates that parameter c is significantly different from 0 at the 95% confidence level for the last five periods when more countries fall below the threshold.

where the units of net cereal import is in [kg/(capita year)] and water is in [m3/(capita year)], a represents the base net cereal import independent of water resources [kg/(capita year)], b is the theoretical maximum amount of water-induced cereal import (exceeding the base level a) as water approaches 0 [kg/(capita year)], and c is the renewable water resources below which a significant dependence exists between net cereal import and water resources at the country level [m3/ (capita year)]. The postulation of an exponential relationship between cereal import and available water resources allows a certain freedom in defining the value of c. The factor 3 in the exponent of model 1 follows a commonly used practice for obtaining a practical value for c. This defines the parameter c as the amount of available water at which the net cereal import is 5% of b above the asymptote a (i.e., net cereal import ) a + 0.05b). We define c as the “threshold of water

Although the general trend in cereal import is relatively well represented by model 1, there is a considerable amount of variation in the net cereal import per capita for countries with similar water resources. For example, the net annual cereal import in Burundi is almost negligible despite its meager water resources. With the similar level of water resources, the volume of import in Algeria, Egypt, and Morocco all exceeded 120 kg/(capita year). This situation gives rise to a consideration of other factors in the model. To identify such additional factors, correlation coefficients are calculated between the residuals of model 1 and possible influence variables depicting a country’s physical, technological, and socio-economic conditions for the last investigation period (Table 1). The choice of the variables was based on what are commonly used and also on data availability. A scatterplot of the correlation between model residuals and various variables is given in Figure A of the Supporting

Modeling Approach and Results The investigation starts by analyzing the net cereal import as a function of renewable water resources alone. The relationship between cereal import and renewable water resources was first analyzed nonparametrically (13). Based on the shape of the function found in this analysis, the dependence of net cereal import on renewable water was then approximated by the following parametrization:

3 net cereal import ) a + b exp - water c

(

)

VOL. 37, NO. 14, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3049

TABLE 1. Correlation Matrix of Residuals of Model 1 with Other Possible Influence Factors for the Last Investigation Period (1996-2000)a residuals water residuals water land irrigation fertilizer log(GDP)

1 -0.01 -0.40 0.06 0.30 0.60

land

irrigation fertilizer log(GDP)

-0.01 -0.40 1 0.15 0.15 1 0.19 0.01 -0.01 -0.10 -0.30 -0.16

0.06 0.19 0.01 1 0.45 0.24

0.30 -0.01 -0.10 0.45 1 0.56

0.60 -0.30 -0.16 0.24 0.56 1

a water ) renewable annual freshwater per capita [m3/(capita year)]; land ) sum of arable land and permanent cropland per capita [ha/ capita]; irrigation ) irrigated area per capita [ha/capita]; fertilizer ) annual fertilizer application per capita [ton/(capita year)]; GDP ) gross domestic product per capita [US$/capita].

Information. Cereal prices were not considered as an independent variable under the assumption that all countries face the same prices at the international market. In this case, cereal prices should not affect a single model fit for a given time period. They can, however, have an influence on temporal changes in parameter a in the model for different time periods. The small correlation coefficient between the residuals and water indicates that model 1 describes the dependence of net cereal import on freshwater resources adequately. The correlation coefficient of 0.61 identifies log(GDP) as the most important potential influence factor for the residuals. The scatterplot given in the Supporting Information (Figure A) shows that a linear dependence of the residuals on log(GDP) leads to an adequate description. Inclusion of log(GDP) yielded model 2:

3 net cereal import ) a + b exp - water + d log(GDP) c (2)

(

)

The parameter d represents the increase in net cereal import per unit of log(GDP) [kg/(capita year log(US$/capita))]. Figure B in the Supporting Information illustrates the effect of the additional term on model results. With a R 2 value of about 0.65, model 2 is a significant improvement over model 1 in explaining the variation in cereal import. This is confirmed by an F-test for comparing nested models (the P value is higher than 99%). A t-test indicates that parameter d is significantly different from 0 at the 99% confidence level for all investigation periods. Noting that land has the second largest correlation coefficient after log(GDP) (Table 1), we included the variable land as a linear term in the model and obtained model 3:

3 net cereal import ) a + b exp - water + c d log(GDP) + e land (3)

(

)

The parameter e represents the increase in net cereal import per unit of arable land plus permanent cropland (values will be negative: decrease in import due to the availability of land) [kg/(ha yr)]. The improvement of model 3 over model 2 is marginal. R 2 increased only by 0.03. An F-test for nested models revealed only a significant improvement of this model at the 95% confidence level over model 2 for the last four investigation periods. A t-test of parameter e shows a similar result. The scatterplot given in the Supporting Information (Figure A) implies that a nonlinear term similar to the dependence on water would lead to slightly better results. However, the regression analysis suggests that this minor improvement cannot justify an additional parameter in the model equation. This was further tested by calculating the 3050

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 14, 2003

value of the BIC model structure selection criterion (14) to identify the “best” of the three regression models. This calculation resulted in model 2 as being the best model for most of the investigation periods (see the Supporting Information for details). We did not further include fertilizer and irrigation in the model simulation because of their small correlation with the residuals of model 1 and the lacking evidence of dependence in the scatterplot. The discussion of each of the parameters below is therefore based primarily on the results from model 2, while the results from models 1 and 3 are used as reference. To compare the results of models 1-3 and observe the trend of changes over the years, the values of parameters a-e for the 17 investigated periods with their associated uncertainty bands are plotted in Figure 2. The stability of these estimates as a function of the time window considered can be viewed as an indirect validation of the model. As model fits for nonoverlapping periods lead to similar parameter estimates, it is evident that simulations with the parameter estimates from one period would lead to reasonable predictions for the adjacent periods. Trends in parameter estimates over time are discussed in the next section.

Discussion Between the first and the last investigation periods, there is a significant decrease in per capita renewable water resources for all the countries considered as the result of population growth (all arrows point to the left in Figure 1). In almost all the countries with water resources below the threshold, there is an increase in per capita cereal import (arrows point upward in Figure 1). In contrast, per capita cereal import remained mostly unchanged in the countries with water resources above the threshold, suggesting no significant relationship between changes in their per capita water resources and the volume of cereal import. The situation in Saudi Arabia and Egypt is rather peculiar and thus deserves an elaboration. Cereal production in Saudi Arabia increased from barely 0.3 million ton in the early 1980s to a peak of over 5 million ton in the early 1990s. The expansion of irrigation by extracting nonrenewable fossil groundwater contributed largely to this increase. The production has dropped in the late 1990s and stood at around 2.6 million ton (8). This reverse trend, however, is not reflected in Figure 1 where only the import volumes of the first and the last investigation periods are presented. As for Egypt, all crops are produced under irrigation, and the irrigated areas expanded by 35% during the period observed (8). This has been the reason for a relatively rapid growth in cereal production, resulting in a decrease in cereal import on per capita basis while the total volume of the import continued to increase. With the country’s water withdrawal approaching the limit of its available resources (15, 16), however, it is unlikely that the past growth rate can be maintained. The proportion of imported cereal in the per capita consumption is expected to rise sooner or later. The inclusion of additional variables in models 2 and 3 led to shifts in parameter a in comparison to model 1 (Figure 2). While the shifts have no intrinsic meaning (they correct for the nonzero mean of the additional term), the rising trend of a since the late 1980s is noteworthy. It suggests that the countries included in the analysis overall tended to import a larger amount of cereal on per capita basis in the later years than in the earlier years. This trend may partly be attributable to the decline in cereal prices at the international market. Between 1980 and 2000, average cereal prices dropped by roughly 50% in constant U.S. dollar terms (8). This would have made cereal import both more affordable and economically efficient to all the countries concerned, resulting in a rise in parameter a.

FIGURE 2. Estimated model parameters a-e with uncertainty bands and values of the coefficient of determination R 2 for model 1 (green, plus symbols), model 2 (blue, cross symbols), and model 3 (red, solid circles). The solid lines indicate estimates ( 1 SE. Parameter b shifts downward by the inclusion of log(GDP). This may be because, in model 1, b captured some of the contributions of GDP to the increase in cereal import. The positive value for the coefficient d is an expected result as a high GDP allows a country to purchase the amount of cereal that cannot be met by domestic production. Meanwhile, a higher income also leads to a greater cereal demand in association with a larger portion of meat and other animal products in the food consumption. Conversely, poor countries have a low affordability for cereal import. Taking Burundi as an example, the negligible cereal import is in direct relation to its very low GDP per capita, below 140 US$ (12). The average calorie intake is only 1628 kcal/(capita day) in the country (8), substantially below the internationally accepted minimum dietary energy requirement of 2500 kcal/ (capita day) (7). Thus, the message that could be drawn here is that, while water deficit presents a rigid constraint to food production, food insecurity and starvation are rather a direct result of low incomes. The negative sign of parameter e reflects an inverse relationship between availability of land resources and cereal import. The increase in the absolute value of e is apparently attributable to the improvement in land productivity over the years, represented by rises in crop yields. The water threshold (c), in which we have a special interest, is very stable from one model structure to the other. This increases the confidence in its value. The uncertainty band of the estimate also narrowed in the later years with more countries falling below the threshold. Hence, the model simulations demonstrate that a water resources threshold can be defined. Below the threshold, a country’s cereal import is strongly dependent on the renewable water resources. Above it, no direct relationship is discernible. The result suggests that water is an important limiting factor to domestic food production in water-scarce countries. Cereal grains have to be imported to compensate for the water deficit. In other countries, the situation of domestic food production and cereal import is not directly related to the amount of water resources per se. Typical examples would be many sub-Saharan countries where the low food production capacity and food insecurity are often

a result of poor soil fertility and lack of agricultural investment (17). Social and political conflicts, mismanagement, and inappropriate economic development policies only exacerbate the problem (18). An important observation on the value of the threshold c is that it appears to have a declining trend. The calculated water resources threshold goes from a high of some 2000 [m3/(capita year)] in the early 1980s to approximately 1500 [m3/(capita year)] at the turn of the last century (Figure 2). It is interesting that the widely cited water stress threshold of 1700 [m3/(capita year)], suggested by Falkenmark and Widstrand in the early 1990s (11), falls within the calculated range of c. This demonstration of the dynamic nature of the threshold is distinctly different from the previous notion, which has always expressed it as a fixed value. The decline in the water resources threshold is no doubt attributable to the expansion in irrigated areas (increased ability to tap water for agricultural uses) and the improvement in water use efficiency (defined as the dry matter or harvested portion of crop produced per unit of water consumed). During the last two decades, irrigated areas increased by 23% for the world as a whole. In the water-deficient countries defined in this study, the increase was even larger at 31% (8). Alongside the expansion in irrigation, a significant improvement in water use efficiency also took place in many areas. Worldwide, there has been a remarkable increase in crop yields on irrigated land and, to a lesser extent, rain-fed land (15, 19, 20). The average yield of wheat in developing countries, to which most water-deficient countries belong, increased by over 70%. The increments of rice and corn yields were over 50% and 40%, respectively (8). In this process, technological advances and the increased application of chemical fertilizers have played a substantial role (21, 22). More output under a fixed amount of water resources leads to a decrease in the threshold, c. A point that must draw attention here, however, is the exclusion of nonrenewable groundwater from the estimation of the threshold. The extraction of nonrenewable groundwater has become massive in many water-scarce countries and regions during the last two decades, causing a depletion of aquifers at an alarming rate. The situation in Saudi Arabia VOL. 37, NO. 14, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3051

TABLE 2. List of Countries in Africa and Asia Having Renewable Freshwater Resources below the Calculated Threshold of 1500 [m3/(capita year)] by the Year 2030a Afghanistan Algeria Burkina Faso Burundi Cape Verde Comoros Cyprus a

Egypt Eritrea Ethiopia India Iran Israel Jordan

Kenya Korea Republic Lebanon Libya Malawi Maldives Morocco

Niger Nigeria Pakistan Rwanda Saudi Arabia Somalia South Africa

Tanzania Togo Tunisia Uganda Emirates Yemen Zimbabwe

Bold names are the countries entering the water deficit country list after the year 2000.

is a typical case in point. With total renewable water resources of 2.2 km3/year and desalinated water of some 700 million m3, the country’s water withdrawal exceeded 17 km3/year in the early 1990s and over 13.5 km3/year came from nonrenewable deep aquifers (8). Apparently, such a practice cannot be sustained in the long term. Presumably because of the depletion of groundwater resources, cereal production in Saudi Arabia had dropped after the 1993 peak (8). In the model simulations, however, the use of nonrenewable groundwater was not taken into account because of the lack of systematic data. This led to a downward distortion of the constraint of renewable water resources deficit on food production and the demand for cereal import. In other words, the parameter c (Figure 2) underestimated the real threshold, and the values calculated are rather conservative. It should also be noted that the threshold with respect to cereal import defined here is an aggregate proxy that is tied with the economic efficiency and the capacity of a society in utilizing its resources. The underlying implication of the model result is that, when a country’s water resources is below the threshold, it becomes more economical for that country to import cereal than to produce it locally. Therefore, countries that have the financial capacity to purchase food opt for importing at a level defined by the curve shown in Figure 1. For the countries that lack the financial capacity to import, starvation is likely to occur. This is because the lack of the financial capacity of a country not only affects its ability to purchase cereals in the international market but also impedes the water development and the application of modern technologies and inputs that are crucial for increasing domestic production. In essence, the threshold of water resources with respect to cereal import defines a critical point below which a country would choose importing to fill the domestic shortfall if it has the choice or face food shortage when such a choice is not available due to the financial constraint.

Predictive Model Application Baseline Projection on the Water Deficit-Induced Food Import. Population growth is one of the major driving forces of water demand increase and the decrease in per capita availability of water resources. With the continuous growth of the world population, the number of those living under water stress will increase. This will lead to an expansion of the demand for food import. Using population prediction figures and the associated uncertainties expressed as low and high variants from the United Nations (23), we calculated the per capita renewable water resources up to the year 2030 for the countries considered in this study under the assumption of constant total water availability. Subsequently, the total annual water deficit of the countries with water resources below the threshold of 1500 [m3/(capita year)] was calculated. Such a deficit is defined as the difference between the threshold and the per capita renewable water resources multiplied by the population of the country and summed over all countries below the threshold. The result shows that the total expected 3052

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 14, 2003

water deficit of the water-deficient countries in Table 2 in the year 2030 is around 1150 km3/yr, a volume roughly 13 times the annual discharge of the Nile River (measured at the Aswan Dam station). Accounting for the population uncertainties yields the low variant of 800 km3/yr and the high variant of 1500 km3/yr (Figure 3a). On the basis of parameters b and c obtained for the period (1996-2000), we calculated the potential demand of cereal import induced by the water deficit estimated with the associated uncertainties (Figure 3b) for the countries in Table 2. Although model 2 was the best estimator of net cereal import, we used model 1 for this calculation because a longterm projection of GDP does not exist. The required cereal volume is predicted at a staggering amount of 140 million ton by the year 2030 (low variant ) 120 million ton and high variant ) 160 million ton) as compared to 30 million ton in the year 2000. The total cereal supply at the international market would have to increase by 50% from the current volume of some 200 million ton to merely cope with this additional need from the countries studied. Adding the demand from countries in other continents that may drop below the threshold in the coming years, water deficitinduced cereal import will be even greater. Options for Addressing the Future Water Shortage and Food Security. Fortunately, the above-projected water deficitinduced food import demand may be reduced if the world community can take appropriate actions now. Our model simulation has shown that the water resources threshold is a dynamic proxy rather than a fixed volume. Its decline in the past has avoided many countries from dropping below the threshold. Clearly, in the future, the actual scale of water deficit-induced food import will largely depend on how fast and by how much the water resources threshold can be further lowered. The increase in the capacity to bring water into irrigation had partly contributed to the decline in the water resources threshold in the past. However, as the availability of additional water supply and the number of suitable sites for water project construction decrease and the competition for water from other sectors escalates, the scope of bringing more water into irrigation will be marginal in most of water-scarce countries (24, 25). Large-scale intra- and inter-river basin water transfers may only be opted in few countries and regions because of the high cost of construction as well as the growing local resistance and the increasing awareness of environmental impacts (26). The political sensitivity and conflicts involving international rivers pose a further constraint to such an approach. Desalination of seawater has been used in some oil-rich countries, notably Saudi Arabia, to alleviate water scarcity. But it is unlikely to become a significant source of water for crop production because of the high cost of energy. Currently the best plants need about 28-59 kJ to treat 1 L of seawater. This translates into an energy cost of 1-1.5 US$/m3 of freshwater produced from seawater (27). At such a level of water cost, food would become very expensive. Therefore, desalination will be limited to extremely water-poor and

FIGURE 3. (a) Projection of total water deficit of the countries with renewable freshwater resources below the threshold of 1500 [m3/(capita year)] (expected values in black, high and low UN population forecast variants in blue, and expected values for water thresholds of 1300 and 1700 [m3/(capita year)] in red). (b) Projection of the expected “water deficit-induced” cereal import calculated with model 1 [i.e., b exp(-3/c water)] (black) with uncertainty bands in association to the high and low UN population forecasts (blue), water threshold values of 1300 and 1700 [m3/(capita year)] (red), and uncertainty in the model parameters (green). energy-rich countries and mainly for nonagricultural purposes. Wastewater treatment, reuse, recycling, and artificial recharge of groundwater have received a growing attention in recent years. Recycle and reuse of wastewater for irrigation has been seen in some water-scarce countries and regions. However, in developing countries, such a practice has often been carried out using untreated or not properly treated wastewater, causing a serious health concern and soil pollution. In view of the high cost of wastewater treatment and the health risks involved, using wastewater for irrigation at the moment may only be suitable in limited areas. In the future, however, the prominence of wastewater as a source of irrigation may increase with the improvement in wastewater treatment technologies and the affordability of the treatment cost (28). During the past two decades, technology advances and the associated improvement in water-use efficiency have played an important role in bringing down the water resources threshold. With water becoming increasingly scarce, the development and application of locally suitable technologies and the improvement in water management efficiency will have to play a greater role in dealing with water scarcity and enhancing food security. Two points however are worth noting here. One is the pollution caused by agricultural production. The intensive use of chemical fertilizers and pesticides in pursuing higher production has been rapidly becoming environmentally hazardous in many areas of the world (21, 29). Pollution and environmental degradation have further aggravated the water scarcity by reducing the availability of usable freshwater (30). The situation is more severe in developing countries where environmental regulations and pollution control measures are often either not in place or unenforceable. In the coming years, therefore, greater efforts for increasing food production must be integrated with agricultural practices and technologies that are designed to minimize the adverse environmental impacts. The other point of concern is the potential for further lowering of the water resources threshold. There is a

physiological minimum water demand that is linked to the basic needs for food consumption, drinking, hygiene, and other essential uses of an individual human being. Although the volume of this physiological minimum can vary among individuals depending primarily on the living standards, an estimate of 800-1200 m3/capita may be a reasonable range (31-33). This physiological imperative would be the lowest limit that the water resources threshold could go. With the water resources threshold moving closer to the physiological imperative, further lowering the threshold will become more and more difficult and the marginal cost of the efforts required will be higher. Until the end of the 1990s, most of the countries with water resources below the threshold have been oil-rich and/ or had their GDPs above the low-income level in the World Bank income classification. The ability to purchase food from the international market had enabled these countries to compensate for their water deficit and to meet the domestic demand. Looking ahead, however, the situation becomes increasingly uncertain and worrisome. Many countries that are likely to fall below the water resources threshold in the coming years are poor and hence are unable to afford the purchase of cereals (Table 2). Compounding the situation is an expected increase in cereal prices in response to the overall greater demand (9). Climate change may add further threat on food production, particularly in poor countries (34, 35). Given the large population sizes of some of the newly added poor countries, the scale and incidence of food insecurity could become much higher in the coming years than what has been seen in the past. To avoid the occurrence of the situation, greater efforts must be devoted to scientific research and innovation and application of environmentally sustainable agricultural technologies. Equally important is a capable international food trade regime that can provide sufficient food to the market without political strings attached (32, 36).

Supporting Information Available Additional text including two figures and a table. This material is available free of charge via the Internet at http:// pubs.acs.org. VOL. 37, NO. 14, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3053

Literature Cited (1) (2) (3) (4)

(5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)

Johnson, N.; Revenga, C.; Echeverria, J. Science 2001, 292, 1071. Allan, T. Ground Water 1998, 36, 546. Yang, H.; Zehnder, A. J. B. World Dev. 2002, 30, 1413-1430. Alcamo, J.; Henrichs, T.; Rosch, T. World Water in 2025: Global Modelling and Scenario Analysis for the World Commission on Water for the 21st Century; Kassel World Water Series Report 2; University of Kassel: Kassel, Germany, 1999. Vo¨ro¨smarty, C.; Green, P.; Salisbury, J. Science 2000, 289, 284288. Cosgrove, W.; Rijsberman, F. Making Water Everybody’s Business, World Water Vision; The World Water Council, Earthscan Publications Ltd.: London, 2000. Smith, L.; Obeid A.; Jensen, H. Agric. Econ. 2000, 22, 199-215. FAO. FAOSTAT, On-line Database; 2002; www.fao.org. Rosegrant, M.; Cai, X.; Cline, S. World Water and Food to 2025: Dealing with Scarcity; International Food Policy Research Institute: Washington, DC, 2002; www.ifpri.org. World Resources Institute. Environmental Data Tables; Online Database; Washington, DC, 2002; www.wri.org. Falkenmark, M.; Widstrand, C. Population Bull. 1992, 47 (3), 1. World Bank. World Development Indicators (CD-Rom), 2000. Cleveland, W.; Grosse, E.; Shyu, W. In Statistical Models in S; Chambers, J., Hastie, T., Eds.; Wadsworth & Brooks/Cole: Pacific Grove, CA, 1992; Chapter 8. Schwarz, G. Ann. Stat. 1978, 6, 461-464. Elarabawy, M.; Tosswell, P.; Attia, B. J. Water Serv. Res. Technol.Aqua 2000, 4, 111-125. El-Kady, M.; El-Shibini, F. Desalination 2001, 136, 63-72. Sanchez, P. UC Berkeley expert in agroforestry and tropical resources. Campus News 2002; World Food Prize;http:// www.irinnews.org/report. Appelgren, B.; Klohn, W. Phys. Chem. Earth (B) 1999, 24, 361373. Rosegrant, M.; Cai, X. Water Sci. Technol. 2001, 43, 61-70. Pan, J. Z., Zhang, J. Z., Eds. Issues on Rational Allocation of Water Resources in Northern China and South-North Water Transfer; China Water Publishing House: Beijing, 2001.

3054

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 14, 2003

(21) Huang, J.; Pray, C.; Rozelle, S. Nature 2002, 418, 678-684. (22) Khush, G. S. GENOME 1999, 42, 646-655. (23) United Nations. World Population Prospects: The 2000 Revision; 2002; esa.un.org/unpp. (24) International Water Management Institute. PODIUM, The Policy Dialogue Model-A Water and Food Security Planning Tool; 2002; www.cgiar.org/iwmi/tools/podium.htm. (25) Postel, S. Pillar of Sand, Can the Irrigation Miracle Last? W. W. Norton & Company: New York, 1999. (26) World Commission on Dams. Dams and Development. A New Framework for Decision-making; Earthscan Publications Ltd.: London and Sterling, VA, 2000. (27) Postel, S. Dividing the water: food security, ecosystem health, and the new politics of scarcity; World Watch Paper 132; Worldwatch Institute: Washington, DC, 1996. (28) Rahman, A.; Kadi, M.; Rockstrom, J. Water Sci. Technol. 2002, 45, 191-193. (29) Zehnder, A. J. B. EAWAG News 1999, 46, 18-20. (30) Postel, S.; Daily, G.; Ehrlich, P. Science 1996, 271, 785-788. (31) Smith, M. Agric. For. Meteorol. 2000, 103, 99-108. (32) Zehnder, A. J. B.; Yang, H.; Schertenleib, R. Aquat. Sci. 2003, 65, 1-20. (33) Rockstro¨m, J.; Cordon, C.; Folke, M.; Falkenmark, M.; Engwall, M. Linkages among water vapor flows, food production, and terrestrial ecosystem services. Conserv. Ecol. 1999, 3 (2); www.consecol.org/vol3/iss2/art5. (34) Doll, P. Climatic Change. 2002, 54, 269-293. (35) Yates, D.; Strzepek, K. Clim. Change 1998, 38, 261-287. (36) Kilman, S.; Thurow, R. Wall Street J. December 3, 2002, p A1.

Received for review November 27, 2002. Revised manuscript received April 27, 2003. Accepted May 6, 2003. ES0263689