Global climate change presents a risk to future livelihoods. The nature and extent of that risk must be located between the poles of global catastrophe and local resiliency. At one pole, the future of the planet is imperiled. This is an extreme view, given the current projections of climate change, but not unthinkable if global greenhouse gas (GHG) emissions are not controlled. The other pole foresees a pace of technological, social, economic, and political progress that enhances resiliency to perturbations in environmental conditions. According to this view, societies have or will develop the means to cope with the expected range of climatic fluctuations.
Three fundamental difficulties arise in assessing the risk of climate change to future livelihoods. First, rapid socioeconomic changes can be expected at the same time as projected global warming. For instance, the probable range for a doubling of greenhouse gas concentrations and the equilibrium climatic response is 30 to 100 years (Houghton et al., 1990, 1992). But, markets fluctuate on time scales of hours to months and few governments last 30 years. To assess realistically effects in the next century, the baseline of environmental, social, and economic conditions must be projected with and without the interactions between the presumed baseline and its perturbation due to climate change.
Second, studies of local effects must be embedded in a global context of environmental, social, and economic processes, impacts, and responses. Local welfare is increasingly part of global processes, for example, of trade, credit, labour, communications, human rights, and aid.
Third, the complexity of linkages among economic sectors, and feedback processes in general, confounds the analysis. The largest agricultural impact of climate change may occur through relatively small changes in comparative advantage, strongly reflecting government policy and indirect market interventions, rather than seemingly large changes in yield potential (Smit 1989). A large range of impact projections are required to capture the uncertainty of method and surprises inherent in futurology.
The Climate Change and International Agriculture project sought to address these challenges to understanding the effects of global climate change. The project, funded by the U.S. Environmental Protection Agency (EPA) and coordinated by Cynthia Rosenzweig (Goddard Institute for Space Studies, GISS), Martin L. Parry and Thomas E. Downing (Environmental Change Unit), involved four components:
Crop Yield Modeling: Agricultural scientists estimated potential changes in yields using the crop models developed by the U.S. Agency for International Development's International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT). Yields were modeled for wheat, maize, rice, and soybeans at 112 sites in 18 countries. The combination of sites and countries represent about three-quarters of world production for wheat, maize and soybeans, and about half of world production of rice.
National Potential Yields: The crop model results and other studies were used to estimate changes in national potential yields. Estimates were made for 34 countries and regions for major commodity groups (e.g., coarse grains).
World Agricultural Trade: The country and regional scenarios of potential yield changes were run through a world food trade model, the Basic Linked System (BLS), at the International Institute for Applied Systems Analysis (IIASA). This general-equilibrium model estimates annual changes in global and regional crop production, prices, trade patterns, and food poverty. The scenarios of world agricultural trade with climate change were compared to baseline projections including the interactions with population growth, economic growth, trade liberalization, and land use controls.
Vulnerable Regions: The implications of climate change were assessed for societies and regions vulnerable to food poverty due to the combination of climate change, resource constraints, population growth and economic development. Methodological development, a global assessment and five country studies (Zimbabwe, Kenya, Senegal, Egypt and Chile) were undertaken.
Results from the first three components of the Climate Change and International Agriculture project are summarized in Rosenzweig et al. (1992). This research report summarizes the results of the vulnerable regions component of the project which drew upon a network of over 25 collaborators (see the list of collaborators in section 12). In order to report the results to the U.N. Conference on Environment and Development (June 1992) and International Geographical Congress (August 1992), this summary report precedes detailed monographs being prepared for each country study (see the acknowledgments in section 11).
The traditional scientific mode of inquiry, building, validating, and linking together formal models based on observations and experiments, provides great insight into the evolving dynamics of socioeconomic systems and their resource utilization. For example, the first three components of the Climate Change and International Agriculture project drew upon field-level, national and world models of yields and trade to reveal regional differences in sensitivity to climate change (Rosenzweig et al., 1992).
However, considering the difficulty of modeling the complex interactions inherent in global change, a complementary approach may be warranted. Such a methodology, as proposed here, derives from a vulnerability/risk framework. It focuses on current vulnerability, risk of present and future climatic variations, and responses to reduce present vulnerability and improve resiliency to future risks. The critical question is what can we do to improve present food security and its prospects in a warmer world?
As developed in this component of the Climate Change and International Agriculture project, the methodology for assessing vulnerable places involves several activities:
Understand Current Vulnerability to Hunger: The multiple dimensions of vulnerability can be used to identify vulnerable socioeconomic groups, map their locations, and assess their degree of vulnerability (see Downing 1991 a, l 991 b for an extensive literature review). The typology presented in section 2.1 forms a basis for gauging world food security in section 2.2.
Assess Trends in Vulnerability: Global trends in food security are briefly reviewed in section 2.2. The Senegal study explicitly projects trends in population, agriculture and climate.
Portray the Risk of Climatic Variations: Climatic variations include episodes that affect present vulnerability and trends that interact with future vulnerability. For example, recent drought episodes have had severe impacts in Chile and Zimbabwe. Trends in climatic resources (e.g., temperature and precipitation) may not be clear at a local or national level, as reviewed in section 2.3. The vulnerability/risk framework looks at a range of possible futures, rather than explicitly forecasting climate change (which requires assigning a probability to future scenarios). Each of the country studies assesses the sensitivity of agricultural systems to climatic fluctuations, using a variety of agroclimatic indices, agroecological data bases, and detailed crop-specific models.
Review Potential Responses: The conclusions address the prospects for adaptive responses to the risk of climate change, while the Zimbabwe, Kenya and Senegal studies discuss specific strategies.
Clearly, implementing a full vulnerability/risk assessment for the world and five country studies is a mammoth task, beyond the resources available in this project. Instead, the project developed an analytical framework and illustrated its application at the world level and in five countries for specific development issues.
2.1. Focus on Vulnerability
Formal studies of vulnerability relative to the risk of climate change have assessed relative sensitivity to sea level rise (Tegart et al., 1990) and identified layers of risk in water resources (Glieck 1988). (For additional studies of the impacts of climatic variations in developing countries, see Kates et al. 1985; Magalhães and Neto l991, and Parry et al., 1988,1992.) Often, however, the term vulnerability is used in a colloquial manner to refer to the adverse consequences of climate change. In such usage, it combines a sense of current (or projected) sensitivity and forecasted risk.
In contrast, the term vulnerability has three connotations in the usage proposed here. First, it refers to a specific consequence (such as food shortage or famine), rather than a cause (e.g., drought or climate change). For example, a household in semi-arid Kenya may be vulnerable to food poverty, that is, their current income and assets may be insufficient to avoid hunger if drought triggers a shortfall in production. (See McKee and Vilhjalmsson 1986 for an example from epidemiology). Second, vulnerability implies an adverse consequence, as opposed to the more neutral term sensitivity. For example, maize yields are sensitive to drought; households are vulnerable to hunger. Societies and regions may be vulnerable to food shortage and hunger, perhaps because their' economic systems are sensitive to climate change. To rhetorically assert that developing countries are vulnerable to climate change assumes climate change will have unspecified, but adverse, consequences. Third, vulnerability is a relative term differentiating between socioeconomic groups or regions, rather than an absolute measurement of deprivation. The analyst or decision maker must assign the thresholds of vulnerability that warrant specific responses (Dever et al., 1988).
A formal definition of vulnerability draws upon concepts in epidemiology, comparative statics in economics and risk mapping of hazards. Vulnerability is an aggregate measure of underlying conditions. Vulnerability to hunger is an aggregate measure of the underlying factors that influence exposure to chronic or episodic hunger and predisposition to their consequences. In a broad sense, vulnerability to hunger is the antithesis of food security, the ability of individuals, households, and regions to meet their nutritional requirements. (For additional reviews of vulnerability, see: Alexandratos 1988; Bohle et al., 1992; Borton and Shoham 1990; Chambers 1989; Downing 1991 a; Drèze and Sen 1989; Higgins et al., 1982; Huss-Ashmore and Katz 1989; Kasperson et al., 1990; Liverman 1990; Maskrey 1989; Messer 1989, 1986, 1984; Millman and Kates 1989; Mortimore 1991; Newman et al., 1989; Ruttenberg 1981; Sahn 1989; Shipton 1990; Swift 1989, World Bank 1986.)
Hunger is not a random experience. It is related to a variety of environmental, social, and economic factors. These factors may be organized conceptually on three levels (with three components to each level) to emphasize the scale of geographical and social organization.
Regional Food Shortage
National Food Balance: A macro-level indication of vulnerability is the ability of national production, storage, and net imports to meet aggregate food consumption requirements of the resident population (measured either by the status quo or nutritional standards). The food balance may be a signal of impending problems, although it is neither a sufficient nor necessary condition of food shortage for specific socioeconomic groups.
Geographical Location: Often geographical location implies the coincidence of a number of factors that could be gauged in more specific analyses of institutions, food poverty, or nutrition. For example, food production on-farm compared to household consumption is a measure of food poverty (see below). But, a simple index of agroclimatic resources may provide an additional indication of the geographical distribution of vulnerability. For example, semi-arid agricultural areas are likely to be more vulnerable than humid zones. In many cases, specific regions have already been identified as being particularly vulnerable to chronic food shortage or famine. Additional geographical factors are civil strife and population density relative to resources.
Institutional Development Using the term institutions in a broad sense, this dimension includes the adequacy of infrastructure to support agricultural production, distribute food to markets, provide health services, and promote income and food entitlement. It may reflect the ability of isolated communities and markets and marginalized ethnic groups to command commercial food or food aid.
Household Food Poverty
Income Components: Characteristics of household livelihood (or food entitlement) from agricultural production on-farm and from communal lands, market exchanges, barter/labour exchanges, transfers, and assets comprise an essential dimension of vulnerability to food poverty. A complete enumeration of household income would reveal different sources of food, shifts between sources during times of stress, and patterns of vulnerability due to, for example, drought or price inflation.
Cultural Preferences: The choice of crops, agricultural practices, diet, income-generating activities, and the utilization of other resources are influenced by cultural patterns that affect household income, expenditure, and consumption.
Demography: The composition of the household (e.g., age-sex distribution, size, life cycle stage) influences consumption requirements, availability of labour, and the intra-household distribution of food.
Individual Food Deprivation
Nutritional Status: Data on malnutrition gauge individual ability to withstand deprivation of food once it occurs. The distribution of malnourishment is often correlated with the risk of exposure to food shortage or famine, which may be related to processes at a higher level of aggregation-- household food poverty, deficient health services, or regional environmental factors. Nutritional status is both an indicator of vulnerability and a measure of the consequences of food shortage.
Health Status: The incidence of diseases, such as cholera, diarrhoea, malaria, and vitamin A deficiency, reflects both individual ability to withstand further food deprivation and the effects of malnutrition and food stress.
Social Status: Although difficult to gauge, the social status of individuals within households affects who suffers first from food poverty and who experiences the greatest deprivation. For instance, women and the elderly may have a lower status than male laborers and sons and therefore receive less food in times of crisis.
This typology provides a conceptual integration of the contributions of natural resource, socioeconomic, and nutritional assessments. It explicitly addresses the dynamics of vulnerability across different spatial and temporal scales--an a priori requirement of studies of global change. The concept of vulnerability has been applied in assessments of vulnerability to famine over the course of the next season (Famine Early Warning System Project 1990, 1991 ) and long-range assessments of food security (reported in section 3).
2.2. Selection and Focus of Country Studies
Section 3 utilizes the above typology to compile an index of food security in developing countries. The project supplemented the global analysis with country studies of vulnerability (reported in sections 4 to 7). The country studies were coordinated by the Environmental Change Unit, but relied on teams of scientists to compile baseline data, validate and run agroclimatic models, and assess the implications for policy. The project resources were not sufficient to launch studies in specific countries or to enforce a unitary methodology ID each country. A tension quickly became apparent in developing the country studies: while a large pool of scientists are interested in or working on global change, global coordination (across five continents) and implementing analytical tools (such as Geographic Information Systems and crop-climate models) are very time-consuming and hampered by distance and technology (i.e., different computer capabilities). Highly qualified collaborators, particularly in developing countries, have numerous responsibilities and little time to devote to fundamental data management and analysis.
The original criteria for selecting country studies included: (I) the degree to which the country represents a dimension of food security; (2) the nature of expected environmental changes; (3) a desire to undertake new work that builds upon existing efforts; (4) a sample of geographical and agroecological regions; and (5) the existence of willing and able collaborators. Preliminary inquiries were made regarding projects in Bangladesh, Brazil, China, Jamaica, Malaysia, Mali, Mauritius, Pakistan, and Thailand. Ultimately, projects were completed in Zimbabwe, Kenya, Senegal, Chile and Egypt. The country studies are a convenience sample as opposed to a systematic representation of the diversity of situations in the developing world.
As the project developed, specific development issues were identified for each country study. The countries, issues and summary results are:
Zimbabwe: A sub-tropical, sub-humid to semi-arid country, Zimbabwe has relatively high food security and a strong agricultural economy. The study focused on the spatial shift in agricultural zones and the distribution of risk for marginal small holders. with a 2deg.C increase in temperature, the core agricultural zone decreases by a third. The semi-extensive farming zone is particularly sensitive to small changes in climate. Farmers in this zone, already vulnerable in terms of self-sufficiency and food security, would be further marginalized due to increased risk of maize production.
Kenya: The agroecological zones of Kenya span the cool, humid highlands to the semi-arid, pastoral lowlands. While the national economy has been relatively healthy, drought and famine have continued to plague vulnerable populations. The project compiled a national assessment of socioeconomic groups subject to food poverty and their vulnerability to changes in agricultural zones and land use. National food production potential in Kenya may well increase with increased temperature and precipitation. However, the impact of climate changes on productivity and vulnerable socioeconomic groups in the semi-arid areas could be devastating. For pastoralists and farmers, reductions in the area suitable for cultivation in the order of 15 to 30 per cent in the sub-humid and semi-arid provinces significantly increases the number of people with inadequate climatic resources for sustainable agriculture. Almost 7 million people already live in food-poor households in these zones.
Senegal: As for much of the Sahel, Senegal has experienced a long-term decline in rainfall while food consumption has been maintained by imports. A critical issue for development in Senegal is the ability of rain-fed agriculture to intensify and expand to keep pace with population growth. The project compiled scenarios of long-term agricultural development and their sensitivity to climate change. If the cultivated area expands and agriculture intensifies at a moderate rate, food production in Senegal can keep pace with projected population growth. However, two-thirds of Senegal's arrondissements presently produce less food than required by their rural populations. with climate change, an additional 1 million people may not be supportable by rain-fed agriculture.
Chile: Economic growth in Chile has benefited from an export orientation and diversification beyond the historic reliance on copper. The expansion of exports of table grapes typifies Chile's development policies. The project assessed the sensitivity of irrigated agriculture and development in one region of Chile to long and short-term climatic fluctuations. The balance of irrigation requirements and water resources in northern Chile is already critical and drought episodes endanger production. A warmer environment entails increased irrigation needs for grapes and possibly dramatic shifts in river basin hydrology. Climate change, particularly if drought risk increases, accelerates the point at which economic expansion becomes constrained by water resources.
Egypt: The agricultural economy, and majority of rural residents, of Egypt are dependent on the Mile. The food-poor in Egypt (over 25 million people in 1990/91, almost half the population) are especially vulnerable to changes in the supply of water from the Nile Basin and market prices of bread (Badr and Darwish 1991). The Egypt country study (coordinated by the University of Colorado) integrates models of the water resources of the Nile, crop yields, sea level rise, the regional economy, and food poverty. Preliminary results are pending and the Egypt study is not included in this research report.
2.3. Potential Climate Change: The Range of Risk
Most scenarios of future climate change are derived from General Circulation Models (GCMs) computer models of complex atmospheric processes (see Carter et al., 1992). GCMs are used to test the climatic sensitivity to specific climatic parameters. The common experiment involves running the GCM for the baseline conditions over several decades to produce an average climatology. A second run includes altered parameters to produce a perturbed climatology. For example, radiative forcing might be increased to correspond to a doubling of greenhouse gas (GHG) concentrations (usually expressed as their carbon dioxide equivalents). The resulting scenario of equilibrium climate change is the difference (or ratio) between the experiment and the control run (often labeled the 2xCO2 - lxCO2 scenario). Transient GCM experiments chart incremental climate changes with increasing GHG concentrations to produce time-dependent, but not equilibrium scenarios. The transient models begin to address the critical issue of how fast climate will change, in addition to the magnitude of the expected changes.
The utility of GCM scenarios for assessing the impact of climate change is a contentious issue. GCM scenarios are internally consistent, based on modeled atmospheric processes and realistic forcing functions. Despite their limitations, GCM based scenarios may be the best choice for studies of the globe or large regions where the focus is specifically on forecasting future conditions. At the global level, GCM scenarios of climate change show a remarkable level of agreement: warming of +1deg. to +5deg.C and 5 to 15 per cent increase in precipitation (Table l and Figure 1). The Climate Change and International Agriculture project utilized three GCM experiments: Goddard Institute of Space Studies (GISS) (Hansen et al., 1983); Geophysical Fluid Dynamics Laboratory (GFDL) (Manabe and Wetherald 1987); and U.K. Meteorological Office (UKMO) (Wilson and Mitchell 1987). These provide a consistent reference for gauging global impacts and assessing policy responses.
However, low confidence in regional changes and the large grid scale of most GCM models limit their direct usefulness in local impact assessments. The Intergovernmental Panel on Climate Change (IPCC) review of climate models suggests that impact researchers use the GCM that most closely matches the current climatology of the region under study (Houghton et al., 1990). Yet, this would severely restrict the range of futures considered.
Since global climate models provide diverging and uncertain regional scenarios of climate change, the study has focused on sensitivity to a range of possible futures, derived from the GISS, GFDL, UKMO general circulation model experiments (Figure 2). This approach places greater emphasis on understanding the sensitivity of impact models to the range of perturbations than on projecting the impacts of specific global scenarios. Artificial increments in key climatic parameters (e.g., +2deg.C) were then selected to represent potential climate changes. The average changes in temperature were added to the baseline temperatures, while the baseline rainfall was multiplied by the percentage change in precipitation. This method of constructing climatic scenarios preserves the baseline climatic variability. Thus, the scenarios used in this project test the impact of mean climate change without any changes in variability, although the frequency of extreme events (e.g., seasonal rainfall less than 200 mm) increases solely as a result of changes in the mean.
3. GLOBAL CHANGE AND VULNERABILITY TO HUNGER
At the global level, current food production is sufficient to feed the world's population if everyone were limited to a basic, principally vegetarian, diet, but would support just over half of world population on a "full-but-healthy diet" with 25 per cent of calories from animal products (Chen 1990, Millman et al., 1991). Global resource capacity is sufficient to feed at least 10 billion people, at the 1980 average level of per capita food consumption for all countries (Murai et al., 1991. cited in Clark University 1992). Regional and local access to food is far more important than global quantities. Climate change does not appear to threaten the global food balance, but is a serious risk for specific regions.
This section presents a global overview of vulnerability to hunger. It establishes the context for detailed country studies of climate change, agriculture, food security, and regional development. The dimensions of vulnerability presented in section 2.1 form a framework for a profile of world hunger. An index of food security classifies countries according to their present vulnerability. The index is used further to assess the extent of food security relative to current resource pressure and future risk of climate change.
3.1. Profile of Current World Hunger
The typology of vulnerability to hunger provides the basis for a multidimensional profile of world hunger based on recent data and drawing upon a variety of sources (Kates et al., 1988, 1989, Chen et al., 1990, Millman et al., 1991). This profile suggests an enduring risk of hunger that affects up to a third of the world's population, but with significant temporal and regional variations. In fact, the actual prevalence of hunger is veiled by conflicting standards and the paucity of data.
National Food Shortage: At the national level, a critical measure of vulnerability is the balance of food supply and consumption. Both supply and consumption can be calculated in several ways, reflecting actual or potential values. The Food and Agriculture Organization (FAO) estimated average (actual) food supply compared to nutritional requirements (potential consumption). By this measure, the population in countries with shortages were:
1,570 million people, 31 per cent of world population, in 1984-86
1,368 million people, 27 per cent of world population, in 1986-88.
However, the definition of total food supply and nutritional requirements is uncertain--modest changes in assumptions (i.e. a 10 per cent allowance for household waste and reduction in nutritional requirements reflecting small body sizes in many countries) could reduce the 1986-88 figures to only 3 per cent of world population (Millman et al., l991).
Household Food Poverty:The calculation of household food poverty is even more uncertain than estimates of the national food balance. In addition to definitions of consumption requirements, supplies from on-farm production, barter, market exchanges and donations must be gauged. Few countries have comprehensive, recent household income and expenditure data to draw upon. The FAO estimated regional prevalence of food poverty according to the assumed distribution of household wealth and a threshold of nutritional requirements that supports healthy growth of children but minimal activity in adults. By this method, food poverty affects:
477 million people, 9 per cent of world population, in 1990.
However, alternative assumptions of the income distribution and nutritional requirements by the World Bank (1989,1990; see Annex) would more than double this estimate:
1,294 million people, 25 per cent of world population, in 1988.
Individual Food Deprivation: The world catalogue of under nutrition, the ultimate measure of hunger, includes:
24 million infants, l 6 per cent of the world's infants, born underweight, in 1990
168 million children,31 per cent of children under 5, underweight for age, in 1988
210 million people, 4 per cent of world population, with iodine deficiencies, in 1990 700 million people, 13 per cent of world population, with iron deficiency, in 1990
700 million children, 15 per cent of children under S, with vitamin A deficiency, in 1988.
To depict the regional patterns of current vulnerability, indicators were chosen for each domain of hunger:
National Food Shortage: food availability in kcal/day per capita for 1986
Household Food Poverty: economic development gross National Product (GNP) per capita in 1987
Individual Food Deprivation: childhood (under 5) mortality per 1000 in 1987.
More systematic data, e.g. on food availability relative to consumption requirements, distribution of wealth (such as the Gini coefficient) and nutritional status (such as the prevalence of malnutrition), were not available for a sufficient number of countries. Data for the three indicators
were drawn from World Bank sources (1989, 1990) for 172 countries (see Annex).
The nominal values of each indicator were normalized, and a composite food security index based on the three indicators was constructed:
Fk is the composite food security index for country k,
Ij,k is the nominal value of indicator j for country k,
Ij,. is the mean value of indicator j for all countries, and
SDi is the standard deviation of indicator j for all countries.
Five classes of vulnerability, based on the distribution of the food security index, are shown in Table 2 and Figure 3. The index illustrates the diversity of countries vulnerable to hunger. The famine-prone countries of Africa (particularly the countries, with a food security index of less than -1.00, comprise 37 countries with a total population of over 800 million in 1988.
Using the World Bank's country and regional rates of food poverty, almost 400 million people in these countries would be food-poor. World food poverty affects 1,290 million people,26 per cent of world population. This profile of world food poverty, based on World Bank data, is the upper limit of the range of food poverty estimates. As noted above, recent FAO estimates are much lower and China has lowered food poverty rates to less than 10 per cent in 1989 (Burkhi 1992). Regardless of the actual level of poverty, the relative distribution between countries is apparent; as is the diversity of food poverty and the paucity of systematic data.
Long-term vulnerability will be affected by changes in resources. One important dimension is the balance of population and agricultural resources. The FAO calculated potential population supporting capacity in developing countries (Higgins et al., 1982). Carrying capacity was derived from agroecological zones, crop yields for different levels of technological input, and national data on area cultivated. As an indicator of current resource pressure, population in 1988 was compared to resource capacity (with low level of inputs as gauged for 1975 by the FAO. Population/resource capacity was classified as abundant (population less than 85 per cent of resource capacity), balanced (population greater than 85 per cent but less than 115 per cent of resource capacity), or critical (population exceeded resource capacity by at least 15 per cent). Although agriculture has intensified since 1975, this indicator provides a relative ordering of the pressures on agricultural resources presented by the growing demand for food. Over 50 countries have a critical balance of population and resource capacity (Table 3A). One-fifth of the countries, with 1.3 billion people (and half of the food-poor), have both critical resource pressures and low food security.
3.2. Interactions with Climate Change
The implications of climate change for world food poverty may be assessed in two ways: (I) to utilize a model that integrates the dynamics of resources, population and economy; and (2) to adopt a risk framework based on assessments of vulnerability to food poverty and the nature and magnitude of specific impacts of climate change.
The world trade model tested in the Climate Change and International Agriculture project follows the first approach (Rosenzweig et al. 1992). The Basic Linked System simulates world food production, trade and consumption for annual time steps. In projections to the next century, risk of food poverty increases with the magnitude of changes in yield potential. For example, three GCM scenarios (GISS, GFDL and UKMO, each with the direct effects of CO2 enrichment), result in 10, 17 and 58 per cent increases in food poverty in 2060 (respectively) compared to a reference scenario without climate change. However, the Basic Linked System includes individual country models for only 10 developing countries (the rest are represented in regional aggregate models).
The second approach facilitates interpretation of disparate data to chart the interactions of climate change, as a future risk, and vulnerability. Over the next century, the time scale of climate change, current vulnerability will change. The most significant trends include: development of infrastructure (e.g., health, education and transport); technological innovation (e.g., biotechnology); population growth (and urbanization); economic growth and integration (e.g., reduced dependence on agriculture); and global economic and political negotiation (e.g., on trade, development assistance and environment). The project did not attempt to project vulnerability in the 21st century. Surprise (such as the breakdown of civil authority and episodic crises), rather than trends in driving forces, may well dominate patterns of vulnerability. While the prevalence of food poverty should decline, there is little reason to believe that the relative distribution of vulnerability between countries will be substantially altered.
The risk of climate change for agriculture and food security encompasses four categories:
Direct Effects of Cdeg.2 Enrichment: Increased carbon dioxide concentrations enhance plant growth and water-use efficiency. The changes are more pronounced in C3-pathway crops (e.g.,
wheat, barley, rice and potatoes) than in C4 pathway crops (maize, millet, sorghum, and sugar cane). The direct effects of CO2 enrichment have been tested in numerous greenhouse and field trials and a growing number of crop models. For example, in the International Agriculture and Climate Change project, the CO2 effect increased yields for all crops, with a global average of 25 per cent for C3 crops and 7 per cent for C4 crops (Rosenzweig et al., 1992).
Effects of Altered Weather: The relationships between crops and climate have been widely tested for most staple food crops and the literature on the effects of altered climates on crops is rapidly expanding. While most of the research has focused on incremental changes in mean values, the effects of extremes and altered variability warrants further work.
Effects Related to Altered Resources: One of the most significant effects may well be the concerted effect of altered resources on agriculture. Changes in water resources for irrigation, soil characteristics, sea level rise, ecosystems, pest dynamics and health (and labour) may well be more significant in some locations than the effects of changed weather patterns on plant growth.
Secondary Effects: Agricultural systems overlap with household, regional and global economies that provide and allocate resources. The secondary effects of climate change include altered farm profitability, changed regional production costs and comparative advantage, and increased world prices.
Sufficient data are not available at present to undertake a thorough analysis of the multiple threats to food security. Instead, the index of food security portrayed in Figure 3 is classified against the risk of changes in two resources (yield and sea level rise). As above, 172 countries are grouped into five classes of food security (Table 3B). These are relative classes; some countries with moderate food security present, such as Egypt, may suffer long-term deterioration in their resources and economies.
The Climate Change and International Agriculture project estimated changes in national productivity (the weighted aggregate of cereal crops) for three GCM scenarios with the direct effects of CO2. For almost all of the developing countries, potential yield decreased in each scenario. While the range of yield changes was similar, especially compared with the northern mid-latitudes, two classes of changes were distinguished: moderate (maximum yield decreases of 20 per cent) and severe (maximum yield decreases of 30 to 50 per cent).
For countries with a coastline, three classes of the risk of sea level were identified: slight (little coast with low investment relative to GNP); moderate (substantial coastal investment at risk, e.g., Sudan, Togo, Peru); and high (coastal
investment at risk threatens national economy, e.g., Bangladesh. Madagascar, Philippines). In addition, many of the small island states, not included in the countries assessed here, are highly vulnerable to sea level rise (see Pernetta 1992, Han et al., 1990).
Of the 5 billion people in the 172 countries: 1.6 billion reside in countries that have low food security; 2.4 billion reside in countries that are highly vulnerable to sea level rise; and over 2 billion live in countries that may be subject to high decreases in potential yield (Table 3B). The highest vulnerability and risk are in the 5 countries--Bangladesh, Comoros, Ethiopia, Haiti, and India--that have low food security, critical resource pressures, high risk of sea level rise, and severe risk of declines in yield potential (Table 3C). Almost 1 billion people reside in these countries. Thus, almost one-fifth of the world's population live in countries that can be expected to suffer from the extraordinary effects of climate change. For the proportion of these populations already suffering hunger (perhaps one-fifth to one quarter), climate change may threaten their survival.
4. ZIMBABWE: SHIFTS IN AGRICULTURAL RISK
In Zimbabwe, the study focuses on agricultural risk. Agricultural land use in Zimbabwe spans a range of natural farming regions from the productive specialized, intensive farming regions, of central to eastern Zimbabwe, to the lowland, extensive farming zones of western and southern Zimbabwe (Figure 4 (a)). The altitudinal gradient between farming regions is modest. Small changes in agroclimatic resources entail significant spatial shifts and reveal land use zones that are most sensitive to change. The shift in the distribution of household yields highlights critical thresholds of production. In a country where drought is recurrent (Iliffe 1990), with quite serious impacts in 1991/92, the effects of climate change may be acute.
4.1. Spatial Shifts in Agroclimatic Potential
A simple index of the atmospheric water balance, precipitation minus potential evaporation (Eo, based on Pan A measurements), gauges the relative amount of water available from the atmosphere. Changes in either precipitation or Eo (related to temperature) alter the water balance. The spatial shifts in this index illustrate the extent to which agricultural land use may be subject to changes in water resources often the limiting factor in semi-arid regions of Zimbabwe.
A gridded data base, provided by CSIRO (Booth et al. 1990), was used to calculate the atmospheric water balance for the growing season, November to March. The data comprise 4999 0.5deg. latitude by 0.5deg. longitude cells (or pixels), derived from 107 temperature and 484 precipitation recording stations. The present balance (based on the 1941-70 climatology) shows the wetter highlands (with surplus water) and the extreme water-deficits of southern Zimbabwe (Figure 4).
Pan A Eo, over the entire surface of Zimbabwe, was related to temperature (statistically significant with r=0.62, p<0.01):
Eo= 3754+57.9*Tm (2)
Eo is average seasonal (November-March) Pan A evaporation Tm is average seasonal mean temperature.
Thus, a 1deg.C change in temperature is equivalent to an increase in seasonal Eo of 58 mm. This enables scenarios of changes in both temperature
and precipitation to be compared using the same atmospheric water balance. In the following scenarios, this relationship was used for the entire grid, although the relationship between temperature and evaporation would also be related to other variables such as aspect, altitude, relative humidity, and wind. (The results of a monthly water balance model, with potential evapotranspiration and crop specific water requirements, Will be reported in the Zimbabwe country monograph.)
The baseline atmospheric water balance is compared with a range of potential changes in temperature (+1deg., +2deg., and +4deg.C) and precipitation , (20 per cent). These increments are well within the range of GCM scenarios (Figure 2). For .example: the UKMO scenario indicates warming of 3-7deg.C with _ 50 per cent in precipitation; the GFDL model shows 3.5-5deg.C warming and 0-+20 per cent change in precipitation; and the GISS scenario is 3-5deg.C and little change in precipitation.
Three scenarios are mapped: an increase of 2deg.C, n increase of 2deg.C with a 20 per cent increase n precipitation. and an increase of 2deg.C with a '0 per cent decrease in precipitation (Figure 4). Figure 5 summarizes the suite of sensitivity tests. With a temperature increase of 2deg.C, the wet zones (with a water surplus) decrease by a third, from per cent of Zimbabwe to about 2.5 per cent. The driest two zones double in area. A further increase in temperature, to +4deg.C, reduces the summer water-surplus zones to less than 2 per cent of Zimbabwe. An increase in precipitation of 20 per cent would approximately compensate or a 2deg.C increase in temperature: at +4deg.C, a 20 per cent increase in precipitation fails to compensate for the increased evaporative demand.
4.2. Changing Risk in Maize Productivity
The agroclimatic index shows the broad-scale shifts in the atmospheric water balance. However, : does not assess other aspects of climate change: potential effects of increased CO2 concentrations n the atmosphere; the plant-level soil-water balance; distribution of precipitation within the growing season; fertilizer effects; and the probability of drought and yield failure. For these purposes Dr. P. Muchena applied a more detailed crop -specific water- balance model.
The IBSNAT version of the CERES maize model .was validated for current conditions and tested with scenarios of climate change (see Muchena 1991 and Rosenzweig et al., 1992 for a fuller description of the IBSNAT models and results).
Three representative sites were chosen:
Banket in the commercial highlands northwest of Harare (over 800 mm of rainfall per year).
Gweru in the semi-intensive farming region in central Zimbabwe (over 600 mm per year).
Chisumbanje in the semi-extensive, semi-arid farming region in southeastern Zimbabwe (over 600 mm per year).
The model runs on a daily time step and includes multiple soil horizons, nutrient partitioning between the plant's roots, stem, leaves, and grain. and the physiological effects of elevated CO, levels. The model runs were made for the November to February growing season for individual years between 1960 and 1988. Different soils and maize varieties were modeled for each site. The initial soil water content was about 70 per cent of field capacity. The model yields indicate the magnitude of changes with different levels of thermal time and water stress.
Two caveats qualify these sensitivity tests: (I) the importance of CO2 is widely acknowledged in laboratory experiments, but only a few field tests have been carried out and simulation models such as CERES maize have not been validated against increased CO2 levels (even on their own without increased temperature) and (2) these model results are for initial soil water conditions well above the typical starting levels in Zimbabwe-the effects of altered precipitation regimes may well be more important than increased temperature. A complete suite of sensitivity tests will be reported in the Zimbabwe monograph.
The results suggest that maize is sensitive to changes in temperature and precipitation in all three sites (Table 4). The results for Gweru and Chisumbanje are particularly relevant for small holder agriculturalists. In these zones, warming of 2-4deg.C reduces maize yields by 10-30 per cent. Gweru appears less sensitive to climatic variations. The direct effect of CO2, on average, increases yields by 10 per cent. Thus, the effects of altered temperature and increased CO2 can be expected to be a 0-20 per cent decrease in yields. However, reduced precipitation would exacerbate these effects. Commercial maize production in these areas is most successful with irrigation--dryland maize production may be seriously affected by projected levels of climate change, particularly if runoff and water for irrigation are reduced.
Even more important than changes in average yields, may be changes in the distribution of yields--the expectation of achieving a target level. The cumulative distribution of yields for Chisumbanje are shown for the baseline and three scenarios of warming in Figure 6. In the present climate, with optimal management farmers may expect to exceed 2.5 t/ha in over 70 per cent of the seasons. With any of the scenarios of warming, however, this target would only be exceeded in 15-40 per cent of the years.
Increased risk in food production affects household food security. Surveys in Buhera, in the semi-extensive, semi-arid farming region near Chisumbanje, provide a basis for gauging the implications of climate change on income and self-sufficiency. Household food security in the Buhera area is affected by erratic rainfall, sandy and infertile soils and low levels of crop technology (Govereh and Mudimu 1991). Maize yields are low--averaging 648 kg/ha with a range of 98 to 1811 kg/ha. Farm sizes are relatively large, but with households of 10 people. Average gross margins are negative. Farmers with low yields could earn higher wages off-farm than by producing their own maize.
A representative household was defined, based on the low yielders in Govereh and Mudimu's ( 1991 ) survey of food security in Buhera (Table 5). The land holding is relatively large, but only SO per cent is cropped, and the household comprises 6 adults and 4 children. Household food requirements are based on average weights for Zimbabwe and caloric requirements reported by the Food and Agriculture organization (FAO)
(James and Schofield 1990). Nutritional requirements are difficult to gauge--in this analysis they serve as a reference line rather than a measure of absolute poverty. The analysis is based on average 1986-89 values: prices and income other than from maize remain constant. Household income includes livestock sales, off-farm income and remittances, although these are constant. Multi-year on-farm storage is not calculated--with low levels of production it is not likely to be significant at present. All cultivated land is assumed to be devoted to maize (or have the same yield/income relationships as maize). The distribution of yields is based on the CERES maize model results for Chisumbanje (Figure 6), adjusted to reflect average yields in Buhera for low-yield farmers (340 kg/ha). Clearly, these assumptions are simplistic. Certainly, off-farm income is variable and may be increasingly important for the alleviation of rural poverty in the future. But the model is sufficient to illustrate household vulnerability to food poverty, and may be a beginning point for more realistic assessments.
In the current environment (1965-86 climate and 1986-89 economy), gross food production for the representative household averages 8000 kcal/day, less than three-quarters of requirements (Table 6). Net farm income varies from ZW$ 5,000 to ZW$ 12,000. If off-farm income and remittances are included, total household income is sufficient to procure almost 10,000 kcal/day. At present, household food availability is 79 per cent of requirements; less than 5 per cent of income is derived from off-farm activities.
With climate change, household food security deteriorates. Food availability averages 71 per cent of requirements with +2deg.C increase in temperature, and 62 per cent of requirements with +4deg.C warming. The distribution of household income also shifts--towards higher risk (Figure 7).
Small holders in semi-arid Zimbabwe are already marginalized; climate change imperils their survival. Yet, strategies to increase production already exist: switch to millet and sunflower (although maize currently has the comparative economic advantage); plant shorter-season maize varieties; change plant spacing; apply appropriate fertilizers at the right time; use insecticides; and rotate cereal crops with legumes. While farmers
recognize the benefits of these strategies, they cited cash, risk, labour, land, and input supplies as major constraints to their adoption (Govereh and Mudimu 1991). Increased household food storage would also benefit farmers, provided yields increase and storage facilities are adequate. Policy responses to climate change must address these issues to ensure that vulnerable farmers have at least a minimum set of practicable adaptive options.
Broad-scale shifts in the atmospheric water balance would affect national food production and alter land use zones (Magadza 1992). The semi-extensive farming zone, on the margin of more intensive land uses, appears to be particularly sensitive to small changes in climate. For farmers in this zone, a shift in the risk of maize production would dramatically reduce net income. Socioeconomic groups in this area, already vulnerable in terms of self-sufficiency and food security, would be further marginalized. Increased variations in rainfall and yields would alter the mix of appropriate response strategies. Successful farming systems would have to be responsive to good seasons, implying improved use of weather information, flexible markets for inputs and produce, and reliable drought responses.
5. KENYA: VULNERABLE SOCIETIES AND LAND USE CHANGE
The distribution of vulnerable socioeconomic groups and the complex nature of climate change in a diverse land use system are shown in Kenya. The most productive zones of Kenya are the central and western highlands where the growing season is longer than 180 days, while the north and eastern lowlands, with less than 90 days in the growing season, are best suited to pastoralism (Figure 8). At a national level, agricultural production may well increase with modest warming, since food production in many highland areas is limited by temperature. In the semi-arid zones, however, production and food security would be adversely affected unless increased precipitation compensates for additional evapotranspiration.
5.1. Current Vulnerability
An initial step in a vulnerability assessment is to define the socioeconomic groups that have different patterns of food production and procurement, and possibly different rates of food poverty. Vulnerability to hunger for each socioeconomic group differs. Similarly, consequences of climate change will vary, according to the patterns of vulnerability and the nature of the climatic variations. In most circumstances, vulnerable groups can be defined by their principal means of livelihood, with additional discrimination according to wealth (e.g., size of holding, levels of skill), reliability of income (related to tenure for instance), and geographical location (related to natural resources, but also the political economy that commands development assistance).
In Kenya, seven vulnerable socioeconomic groups establish the range of livelihoods that might be affected by climate change (Table 7). For each group, the rate of food poverty--the proportion of people in each class with incomes insufficient to procure the recommended minimum level of food--has been estimated (Hunt 1984).
Pastoralists are ethnic groups in range lands (according to the 1979 population census). Nomadic pastoralists practice no agriculture: 85 per cent may be at risk of food poverty. Pastoralists who also farm are divided into two groups. Agro-pastoralists (33 per cent food-poor) often have catch crops in riverine sediments. Migrant farmers (55 per cent food-poor) have more permanent fields but also herd livestock at a distance from their settlements.
Landless Poor are rural households with little or no land (based on the Integrated Rural Survey IV) that lack skills to secure permanent off-farm employment. Half of the rural landless households are judged to be food-poor.
Squatters are farmers occupying large absentee owned estates, with no legal claim to their fields. They account for 7.5 per cent of the farm population (more in the Coast Province where large farm squatters are particularly prevalent). Food poverty affects 33 per cent of the squatters. Small holders, defined as landholders with less than 20 ha per household, account for the majority of the rural population in each province. Farmers on middle to large-sized farms are not considered vulnerable to food poverty. About 29 per cent of small holders are subject to food poverty.
Urban Poor comprise households with low-paid salaried and casual workers in the major urban centres--3 per cent of urban dwellers were judged to be food-poor in Nairobi; 6 per cent elsewhere. (The recent recession and price increases may warrant an increase in these rates.)
Almost 7 million people were estimated to live in food-poor households in 1990. The largest vulnerable group (over half of the food-poor population) encompasses small holder agriculturalists. Pastoralists have a higher rate of food poverty, but with a lower population, total just over 1.5 million food-poor people in 1990. Urban poverty comprises only 3 per cent of the total, but may rapidly escalate in the future as rural urban migration increases (Downing et al. 1990).
5.2. Effects of Climatic Variations
Based on several GCM scenarios, future climatic change in Kenya indicates an increase in mean annual temperature of 2.5deg. to 5deg.C with a 0 to 25 per cent increase in precipitation (Figure 2). For example: the UKMO scenario is the warmest (up to +6deg.C), but with a wide variation in precipitation change; the GFDL and GISS models show less warming, but with little increases in precipitation. Scenarios of +2.5deg.C and +4.0deg.C, with 0, +10 per cent and +20 per cent in precipitation were used.
Growing seasons in Kenya are defined as the period during which rainfall exceeds 50 per cent of potential evapotranspiration (according to the Penman method) and temperatures are sufficient for crop growth (Kassam et al., 1991). The baseline length of growing periods (LGPs) were estimated from 437 stations in Kenya, containing data on mean annual temperature, average decadal ( I O-day) potential evapotranspiration and historical decadal rainfall for about 30 years. The LGPs follow the altitudinal gradient of rainfall and temperature and delineate the productive central and western agricultural areas (Figure 8a).
Climate scenarios were applied to the spatial climate data to derive new length of growing period zones and to reassess potential productivity.
A 4deg.C increase in temperature in Kenya results in a dramatic shortening of the LGP (Figure 8). The arid region (no growing period) doubles in size, the limit of maize cultivation (>90 days LGP) retreats to higher elevations, and the tea/dairy zone (>330 days) becomes drier or disappears in some places.
These changes in an agroclimatic index were used in a more complex analysis of land use potential undertaken by G. Fischer at IIASA. The effects of climate change on land use were judged based on the agroecological zonation data base and methodology developed by the FAO and IIASA (Kassam et al., 199l). The spatial data base includes thermal and moisture regimes, soils, and aspects of land use (cash-crops, forests, irrigation, game parks, and tsetse infestation).
The land use analysis involves: (1) selection of land utilization types and input levels for 64 crop types covering 19 food crops, 6 cash crops, grass, pasture, and 31 fuelwood species, each assessed for three levels of management; (2) description of crop requirements matched to the agroecological characteristics of each location; (3) specification of crop rotation options and constraints (such as sustainable single- and multi-cropping systems) and livestock systems, requirements and productivity; (4) description of the relationship between the crop, livestock and fuelwood sectors; (5) quantification of potential arable land and crop production of each activity by agroecological cell (around 90,000 land units derived by overlaying resource and land use maps) to produce a data base describing productivity of all feasible crop rotations by agroecological cell; and (6) specification of district agricultural planning scenarios with objectives and constraints--used as input to a linear programming model that optimizes the allocation of land.
Changes in the LGP were related to changes in the potential productivity and cultivated area of individual crops, grassland and fuelwood species. Then the "best" crop rotation or grassland was chosen for each agroecological cell. In the optimization model, all crop combinations that fit the prevailing length of growing period and met certain sustainability and risk-avoidance criteria were considered. The optimization is based on the weighted sum of net calorie and protein production for human consumption available from each activity.
There are several important characteristics of this approach. The agroecological analysis does not include the increase in photosynthesis due to carbon dioxide enrichment (less significant for maize, the staple food crop in Kenya, than for wheat and barley). Other optimizing criteria, e.g., economic returns, could be used, but would require further assumptions (for example, on future agricultural prices). The choice of crop combinations independently at each location provides for optimal adaptation in accordance with the changed climatic situation, but does not reflect temporal processes of land use conversion, costs of adaptation, and national market constraints. In all scenarios, it is assumed that soil quality, apart from the erosion hazard, is not affected by climate change (perhaps an unrealistic assumption). Also, present forest reserves, game parks, and irrigation schemes are preserved. (An additional simulation with no reserved land constraints shows similar but more optimistic results.)
As expected given the large range of agroecological environments in Kenya, the results show a wide spectrum of impacts (Table 8). Currently, land in Kenya suitable for permanent cultivation exceeds 6,000 square/km; most of the high potential land is concentrated in the highlands of the Rift Valley, Central and Eastern Provinces. The low altitude areas in Kenya are generally constrained by water availability while the highlands are constrained by temperature and perhaps moisture--parts of central and western Kenya presently have more than optimal rainfall. Hence, in Eastern and southern Kenya, increases in temperature, without corresponding increases in precipitation to balance increased potential evapotranspiration, would result in dramatic reductions in potential agricultural production. For example, in Eastern Province the potentially arable land decreases by 28 per cent with a 2.5deg.C increase in temperature, but increases 18 to 66 per cent if precipitation increases 10 to 20 per cent. In central and western Kenya, gains from increased arable land (due to higher altitude areas becoming more suitable) and increased intensity (due to temperature increases and multiple crops per year) usually more than outweigh negative effects of reduced soil moisture even in scenarios assuming no change in precipitation. In the presently humid areas (>250 days LGP), this can even reduce the impact of the pest and disease constraints in the model.
National food production potential in Kenya may well increase with increases in temperature, provided there is some increase in precipitation. Contraction of agriculture in the lowlands and increased potential in the highlands would exacerbate current population pressures and the demand for conversion of reserved areas to agriculture.
If precipitation changes do not compensate for the warming, the impact on productivity and vulnerable socioeconomic groups in the semi-arid areas could be devastating. The effects of climate change will be felt most directly by those vulnerable groups that rely on their own agricultural production for a major share of their food consumption: pastoralists and small holder agriculturalists (see Downing et al., 1989 for an assessment of drought vulnerability and responses, a possible analogue for climate change). For these two groups, reductions in the area suitable for maize cultivation in the order of 15 to 30 per cent in the sub-humid and semi-arid provinces would significantly increase the number of people with inadequate climatic resources for sustainable agriculture. Decreases in the growing season would also increase vulnerability as the probability of achieving adequate yields is reduced, as illustrated in the Zimbabwe study.
Other vulnerable groups may be affected through changes in the agricultural economy. For example, small holders in Central Province may find tea and coffee become less reliable crops. This would greatly reduce their income, but may be compensated by increased suitability in the highlands for maize. Other urban and rural groups will be affected by changing hydrological resources and food prices. It is not possible to forecast the balance of such secondary effects.
6. SENEGAL: POPULATION GROWTH AND DEVELOPMENT
Like many sub-Saharan African countries, Senegal has experienced a long-term decline in per capita food production. The population growth rate of 2.7 per cent fuels an increasing demand for food. About 70 per cent of the labour force is engaged in agriculture, but the agricultural sector has been plagued by poor policies, meagre natural resources, drought, high energy prices, and falling terms of trade. This country study illustrates the potential impact of adverse climatic change on trends in rural population carrying capacity.
6.1. Current Population and Agricultural Resources
The methodology and data for assessing aggregate human carrying capacity were developed by the U.S. Geological Survey (USGS, see Moore et al., l991 for additional details and references). The carrying capacity model is limited to rainfed production of millet, sorghum, maize, rice and cowpeas, which comprise 80 per cent of national caloric consumption. But, irrigation, flood recessional cropping, market gardens, livestock, fishing, and food imports are important at present and for future development. Thus, the carrying capacity calculations present only one dimension of the population/agriculture crisis and cannot be used to directly estimate starvation.
The model compares consumption requirements with food production. The minimum daily consumption requirement is estimated as 2300 kcal per person, of which 80 per cent ( 1840 kcal) is met by the five staple food crops in the model. Rural population and growth rates were derived for each arrondissement based on the 1976 and 1988 censuses.
Estimates of rainfed caloric production drew upon a comprehensive resource data base (geology, hydrology, soils, vegetation, and land use) completed in 1986, primarily at a scale of 1:500,000. Crop statistics (yield, area cultivated, and production) for 1986-89 were collected for each arrondissement. Post-harvest losses, milling rates, fallow areas, and current prices were estimated from available surveys. The location of present agricultural areas was taken from the land use survey and the present cultivated area from Government of Senegal statistics. Present crop yields were related to average annual rainfall (e.g., a minimum of 400 mm for short-season millet and groundnuts) and soil constraints. The total caloric value of rainfed agricultural production was taken from the net caloric value of millet, sorghum, maize, cowpeas, and rainfed paddy rice. (The caloric equivalent at current market prices of groundnuts and cotton could be included in the analysis.)
The current (1990) human carrying capacity, rural population, and population/resource balance are shown in Figure 9. Of the 93 arrondissements, two-thirds have rural populations exceeding their rainfed carrying capacity. At the national level, rural population exceeds production by a relatively small margin, 15 per cent. The rural deficit and urban demand are supplied by food imports.
6.2. Future Agricultural Development
The USGS made detailed estimates of options for agricultural development. Here, a summary of the database was used to calculate long-term prospects and their sensitivity to variations in yield potential. The rural population was projected to grow at 2.7 per cent per year initially, but dampening each year to reach an equilibrium in 2050. A series of development scenarios were defined, based on the rate of agricultural change, intensification of yields, and expansion of the area planted:
Rate of Agricultural Change: Baseline: 1990.
Slow growth: potential increases in yield achieved in 60 years, at the end of the projection; expansion of agricultural area at 2.7 per cent per year, the current population growth rate. Moderate growth: yield increases achieved in 30 years, and continued at the same rate until 2050; area expanded at 5 per cent per year. Rapid growth: yield increases achieved every 15 years.
Baseline: 1986-89 average, reflecting a medium level of technology. High yield: presently obtainable yields with increased inputs and management (an average increase of 38 per cent for all food crops and arrondissements), based on government estimates.
Baseline: 1986-89 average area planted, based on current government statistics and a 50 per cent fallow level (including habitation and infrastructure). Expanded: without permitting rainfed agriculture in reserved lands, expansion onto moderate potential soils within the climatically suitable zones, with 15 per cent of the area left in natural vegetation, and 50 per cent of the remainder cultivated.
Altered Crop Mix:
Baseline: present crop mix, based on government statistics. Altered crop mixes: optimize caloric production, but with constraints on the area devoted to rice and cash crops (these scenarios were compiled by the USGS, but not projected for this assessment of climate change).
It is important to note that these scenarios of agricultural development are based solely on the resource potential. While they are feasible and appear modest, without recourse to irrigation or biotechnology, agricultural growth in Senegal has stagnated over the past two decades. Since the 1960s, the area planted to food crops has not expanded and yields show little (if any) upward trend. Drought, declining soil fertility and agricultural policies have hindered development. Realization of these scenarios will require substantial changes in economic policy, enhanced research, and vigorous dissemination of agricultural technology.
Without climate change, the results indicate that rapid intensification (growth in yields) of the agricultural sector would keep pace with the projected rural population, but moderate intensification continues the current trend of declining food production per capita (Table 9 and Figure 10). The difference in human supporting capacity between the two intensification scenarios is 5 million people in 2050. With moderate growth, the "excess" rural population is 4 million people.
Expansion of the agricultural area is also a viable option to maintain the balance of rural population and carrying capacity. Even with slow growth, at the rate of population growth, I million ha could be added to the area planted in food crops in 2050. With moderate growth, at 5 per cent per year, the maximum available area of medium potential soils, over 2.5 million ha, is reached by 2050. With moderate expansion, agriculture would keep pace with population growth, but only until 2020 (Table 9). By 2050, the discrepancy between rural population and carrying capacity is almost 2 million people (or almost 3 million people with slow expansion of the agricultural area).
An agricultural strategy that combines moderate intensification of yields and slow expansion of the cultivated area would match projected population growth for the next 30 years and, by 2050, could produce a substantial surplus of food. Clearly, Senegal has the resource capacity, even without farming natural reserves, to feed itself.
6.3. Effects of Climate Change
Climate change in Senegal may increase temperatures 2deg. to 5deg.C, with a range of possible precipitation changes, from -25 per cent to +50 per cent, based on the three GCM scenarios reported in Figure 2. A 4deg.C increase in temperature and a 20 per cent decline in precipitation matches the UKMO GCM scenario. This scenario presents the extreme case, to test the upper range of potential agricultural impacts of climate change in Senegal.
Climate change will directly alter potential yields in Senegal. Based on a daily water balance model (without the direct effects of CO2 enrichment) for the principal cereal, millet. warming of 4deg.C and a 20 per cent decrease in precipitation reduces regional yields by 33-45 per cent, except in the far south (Figure 11) (Diagne 1992). In comparison, the 1972 and 1983 droughts reduced national millet yields by 63 per cent and 79 per cent, respectively (Diagne 1992). A more comprehensive set of climatic scenarios, including experiments with a monthly water balance model, will be reported in the Senegal monograph.
To test the sensitivity of the carrying capacity balance to variations in agricultural potential, a 30 per cent reduction in projected yields in 2050 for all arrondissements was used (Table 9 and Figure 10). As expected, the reduction in agricultural potential substantially reduces carrying capacity, to some 62 to 89 per cent of that achievable without climate change. With climate change, an additional I to 3 million rural people in Senegal would not be supported by rainfed food production.
The Senegal study illustrates the importance at the national level of considering climatic variations in agricultural planning. If the agricultural area expands, higher yields are achieved, and the crop mix changes, agricultural production may well exceed projected demand from a growing population. However, climate change would quite possibly alter the desirable crop mix and regional focus for development. At present, two-thirds of the arrondissements produce less food than their rural populations require. With the moderate growth and climate change scenario, three quarters of the arrondissements would be food-deficit regions in 2050. The associated implications for economic development and migration, especially to Dakar, are serious.
7. CHILE: CLIMATIC CONSTRAINTS ON REGIONAL DEVELOPMENT
In Chile, the study focuses on economic development in the Norte Chico region, a semi-arid zone on the southern boundary of the Atacama Desert. The physical environment is one of contrast: a narrow coastal strip giving way to rugged terrain and a system of east-west valleys reaching the Andean cordillera; low rainfall (60-200 mm/yr) but high snowfall and accumulation in the Andes; high insolation but sea fogs reaching 50 km inland; and fertile irrigated river valleys punctuating semi-arid slopes.
7.1. Context of Regional Development
Regional development in the Norte Chico is a tension between the use of renewable resources and expansion of economic activity (Gwynne 1992, Meneses 1992, Romero 1992). The potential conflicts between irrigated table grape production (the main export crop) and use of water are sketched in Figure 12. Table grape production for delivery in November-December has grown exponentially. With no indication of market saturation and without environmental constraints, grape exports would continue to expand. Increased grape production requires significant increases in irrigation. The curve of water use reflects increased efficiency in the early stages but higher water requirements later to flush deposited salts. The intersection of the production and water-use curves defines the critical zone of maximum capacity. However, this zone is dynamic in response to declining rainfall and extended periods of drought (as occurred in 1988-1990). Expansion beyond the critical zone moves the system into a zone of semi-reversible crisis where salinization and drought would cause entire valley systems to go out of production (at least temporarily).
Since the mid-1970s, Chile has pursued an outward-oriented economic policy, with diversification beyond the traditional reliance on copper exports. In the Norte Chico, this policy has led to the rapid expansion of table grape cultivation and export. Grape exports from Chile in 1977 were almost 40,000 t, rising to almost 500,000 t in 1990 (4.4 per cent of total exports). Table grapes plantations in the Norte Chico (some 11,000 ha in 1988) account for approximately 25 per cent of the total area devoted to grapes in Chile. The dominant market is the U.S., due to its large size, affluence, health-conscious consumers, and retailers that stress year-round supply with seasonal price fluctuations rather than seasonal supplies (as is more common in France, for example).
Commercial, irrigated agriculture occupies the five east-west river valleys of the Norte Chico (in north to south order, the Copiapó, Huasco, Elqui, Limarí and Choapa), each fed by snowmelt in the Andes. Only the Limarí has large reservoirs (Recoleta, Paloma, and Cogotí), which have substantially extended the irrigated area. The climatic and water resources of the Norte Chico provide a comparative advantage for export grapes. Grapes can be harvested as early as November and December, in time for the premium winter markets in North America.
The high profitability of table grape production in the Norte Chico valleys has created a dynamic land market. However, in the Limarí valley, this has become linked to an important dichotomy. On the one hand, there are large and medium-sized commercial farms that have installed storage and pump irrigation systems with water supplied from a variety of sources. The commercial farmers have greater flexibility after consecutive years of drought. It is these farmers that are purchasing and developing more marginal land for table grape production. On the other hand, small-sized farms or subdivided communal land holdings rely entirely on water from gradient irrigation systems. The small-scale farmers and communal landholders will suffer first during drought, along with the casual labour force as production declines.
7.2. Effects of Climate Change
Chile's geography--spanning a large range of climates from north to south and steep gradients from east to west-hinders compilation of realistic scenarios of climate change. Temperatures may increase 0.5deg.C to perhaps 10deg.C, with a probable increase in precipitation, perhaps up to +25 per cent. Snowfall and snow accumulation in the Andes, however, may well decrease. The reference sensitivity test for the Norte Chico, +3deg.C and -25 per cent precipitation, corresponds to the GFDL GCM scenario.
Potential climate change in the Norte Chico will alter the agricultural environment and the supply of water for irrigation. A dynamic crop simulation model (without the direct effects of CO2 enrichment) was used to evaluate the implications of climate change on grapes, grass, wheat, maize and potatoes in the Norte Chico (Santibanez 1992). In general, the effects would be mixed (Table 10). Increased temperatures will lengthen the frost-free period and expand the area of potential cultivation. Conversely, winter chilling would drastically reduce, requiring additional costs for temperate fruit cultivation. Yields would decrease for wheat but not maize and potatoes; water-use efficiency might increase; and the growing season of heat sensitive crops could be changed, particularly with irrigation and if they are not sensitive to day length. Warming would also increase heat stress.
The adverse effects of warming appear to dominate for viticulture in the Norte Chico (Table 11). Annual chilling hours could decrease, with 4deg. of warming, from 800 (near the minimum required) to less than 100. Warm winters result in less rest for the vines, reduced flowering quality and fertility, and increased irrigation requirements. Sprays and alternative varieties may compensate for this impact, but with additional costs. Accelerated growth (about one week for each degree of temperature) reduces fruit diameter and weight. Lower yields might be compensated by higher market prices. Based on the average relationship of evapotranspiration to temperature, about 7 per cent more irrigation water would be required for each degree of temperature increase. In the case of fruit trees, decreasing yield and increasing water requirements dramatically reduce water-use efficiency.
For one commune at the wetter margin of the Norte Chico, gross per capita income from agriculture might be halved with a climate change of 3deg.C and -25 per cent precipitation (Santibanez 1992). Without access to irrigation, this would intensify the socioeconomic crisis among communal farmers in this marginal area.
Continued agricultural production in this zone requires adequate supplies of water for irrigation. The project has not been able to construct a water resource model for the Norte Chico, due largely to the scarcity of data on snow accumulation, melting and runoff in a complex river basin. However, the sensitivity of water resources to climatic variations can be judged from the current hydrology and the recent drought (Peña and Rivera 1992). Although records are sparse, snowfall in the Andes correlates with rainfall. Anomalies are associated with the ENSO index: El Niño years (positive index anomalies) are correlated with wet years and negative anomalies are correlated with dry years. In each basin, mean annual flow is highly variable and the distributions of flows are highly skewed (Table 12). The historical variability is well illustrated by flows at Rio Grande en Cuyano (in the upper Limarí basin). The 80 per cent discharge (80 per cent of the years have greater flows), as occurred in 1981/82, averaged 2.4 m3/s, while the median was 4.3 m3/s (in 1949/50), and the wet year of 1983/84 (20 per cent probability of being exceeded) averaged over 14 m3/s. With a highly variable supply, reservoirs or subsurface aquifers are required to maintain high seasonal demand. A small change in the water budget could seriously threaten the viability of irrigated agriculture; and snowfall and accumulation may be particularly sensitive to warming.
7.3. Social and Economic Implications
The potential effects of climate change may be similar to the consequences of the recent drought. In fact, climate change may well result in more frequent and prolonged droughts in this marginal environment. Consecutive years of drought were recorded in 1988,1989 and 1990. In August 1987, the Cogotí reservoir was full. By the end of the 1990/ 1991 summer, extreme problems were evident in the Guatulame section of the Limarí valley system, supplied from the Cogotí reservoir. After three years of water demand and little replenishment, the reservoir's water resources had been exhausted. Similarly, the Paloma, a larger reservoir further downstream, had enough water for only one more year of average demand. In the autumn of 1991, the Dirección de Aguas (controlling the gradient irrigation system of the region) decided that another year of below-average winter rainfall would require severe restrictions in the coming irrigation season for the Guatulame section of the Limarí (downstream of the empty Cogotí but above the Paloma). The basic policy was to provide sufficient water for pruned vines to stay alive but without producing grapes. This clearly demonstrates that for one important area, the critical zone of water resource use had been reached. Fortunately, the winter 1991 rainfall was well above average and all reservoirs were replenished.
The rapid growth of the table grape industry has acted to substantially reduce rural depopulation and migration to large towns (Ortiz 1992). For example, in the Guatulame valley rural populations have increased from 5,272 in 1970 to 10,060 in 1988 in the San Marcos/Monte Patria section. Retention of the local population is linked to the communal land holding systems of many rural settlements. Communal landholders normally have relatively modest amounts of land to cultivate (0.5-5 ha), a survival strategy for poor rural families. With the growth of table grape production, poor rural families also have access to wage labour without migration. The rural male population can find up to four months paid work (September-December), while females work up to two months in packaging plants. This has meant that in the Limarí valley, the period of maximum demand for labour (November-December) is met by rural workers from the Limarí valley system; during the 1988/89 season, only 9.4 per cent of the rural workforce came from outside the region (Gwynne et al., 1992). Thus, in a prolonged drought or a semi-reversible crisis in production, the affected population will be large in the Limarí. In the Copiapo valley, however, about half of the labour at peak demand comes from outside the valley system. The impact of a regional agricultural crisis will be distributed as migration is altered and regional urban centres suffer.
Regional development in the Limarí valley of northern Chile typifies a balance of resources and land use that may be threatened by climate change. The balance of irrigation requirements and water resources is already critical and drought episodes endanger production. Climate change, particularly if it includes increased drought risk, significantly accelerates the point at which economic expansion in the Norte Chico becomes constrained by water resources. A warmer environment entails increased irrigation needs for grapes and possibly dramatic shifts in river basin hydrology. The socioeconomic effects of climate change will likely reflect the existing land tenure system: large landholders with access to supplementary water resources, land and capital will be better able to respond to climate change and drought than communal farmers.
The impact of climate change on food security in developing countries is potentially serious. The current vulnerability of the world's food-poor, as many as 1 billion people, remains an enduring human concern. For these people, it is not feasible to balance the costs of limiting greenhouse gases with the costs of adapting to the potential impacts of climate change. The world must do both. In this section, the nature of response strategies to mitigate the effects of climate change on food security are explored.
8.1. Typology of Response Strategies
Considering the long time scale of climate change, a key question is how will resource, economic and social systems respond? While the literature on agricultural impacts has expanded dramatically, little work has systematically assessed the potential to ameliorate the predicted impacts. Two dimensions of a typology can be envisioned, relating the timing of responses to specific strategies (Table 13). Responses can be scheduled according to the timing and magnitude of climate change. Four broad categories are accommodation, planned resiliency, purposeful adjustment, and crisis response:
Accommodation: Socioeconomic systems can gradually adapt to small changes in climate. This level of adaptation is at no cost, since the changes are below a threshold of noticeable economic impact and occur on a time scale that coincides with changes in socioeconomic systems. For example, slow climate change up to perhaps l deg.C will allow plant breeders time to develop new varieties virtually in the altered environment in which they will be grown by farmers. Crop breeding will be coincidental with the higher temperatures and CO2 concentrations. Switches between land uses and altered management practices will occur with improved agricultural technology and changes in world markets. The additional impetus of gradual and low magnitude climate change may not be noticeable.
Planned Resiliency: A set of adjustments can be envisioned for any economic sector that provides for greater resiliency to climatic variations, regardless of the eventual climate change. In many cases, these adjustments will be beneficial for other reasons and have a net benefit given the current range of climatic variations. For example, improved seasonal forecasting and response would reduce the cost of agricultural production and supporting infrastructure. Other specific practices include research into new varieties, more effective mechanisms to diversify farm income and risk, integrated pest management, and soil and water conservation.
Purposeful Adjustment. Specific practices that are designed primarily to cope with expected climate change, that are adopted as forecasts of climate change become more certain, and are not justified by other social and economic benefits can be termed purposeful adjustments. They entail a higher cost than those in the previous level of response and a higher risk of failure if the climatic changes are different from expected. Targeted agricultural adjustments may include development of crops that respond to CO2 enrichment and shorter growing seasons and inter-regional water transfers for irrigation.
Crisis Response: After significant impacts have occurred, societies may respond to crises with a further set of adjustments. These crisis responses entail bearing the full cost of the impact and the additional cost of emergency actions that may or may not reduce vulnerability to additional impacts. The domain of climate changes marked by crisis response indicates the failure of adjustments adopted earlier to mitigate the impacts of climate change.
8.2. Response Strategies
Farmers may well be able to cope with gradual changes of +1deg.C or perhaps +2deg.C. But, their capacity to adopt a suitable set of responses depends on economic, social and political inducements or constraints to agricultural innovation, such as: agricultural research and extension services, credit and marketing arrangements, and availability of labour. Specific responses for agriculture include: soil and water management; crop choice, husbandry, and land use; and economic adjustments:
Soil and Water Management Regulation of the soil water budget through moisture conservation, irrigation, soil drainage, mulching, fallowing, tillage, crop rotation, etc., is already effective in semi-arid areas. with increased temperature stress and potential evapotranspiration these methods will be essential. Much can be accomplished with little technical skill or investment, but substantial benefits depend on capital (for example to terrace a hill farm) and availability of water for marginal irrigation. This domain of adjustments is only effective in a fairly narrow band of rainfall: in very dry or wet seasons there is no rainfall to conserve or no need for additional soil moisture.
Crop Choice, Husbandry, and Land Use: The choice of cultivars, rotations, area planted, site selection, changes in the seasonal calendar (timing of planting and harvest), planting depth, plant density, use of herbicides, pesticides and fertilizer, nitrogen-fixing crops, conversion to/from pasture, livestock types, stock levels, etc. are all possible adjustments to altered climates. Most would occur gradually with little additional investment. However, a major change in the regional crop mix would require new infrastructure for the dissemination, marketing and storage of produce. Faced with uncertain climate changes, perception of its change, proclivity for innovation, and other resource and economic factors, it is difficult to predict when farmers will begin to change their crop mix. As such, it may be important for agricultural planners to be prepared to support a wide range of innovation, rather than promote specific policies tied to current forecasts of the impacts of climate change.
Economic Adjustments: The broadest range of adjustments are economic--diversification of household and regional income, investment in agriculture, food and seed storage, savings and credit, wage employment, market access to food, altered food consumption, research and experimentation, regional development (e.g., rural growth centres), etc. These responses spread the risk of climate change to larger populations and other sectors. On the other hand, greater economic integration creates a dependency on world trade and vulnerability to the effects of climate change in other regions.
This review of the implications of potential climate change for food security in developing countries warrant four enduring conclusions:
Winners and Losers: Climate change will have mixed impacts, between and even within developing countries. Some high potential areas, such as the highlands of Kenya, could benefit from warming. Semi-arid areas at the margins of cultivation are sensitive to climate change: the balance of changes in temperature and precipitation could dramatically increase or decrease agricultural potential. The concerted effects of decreased agricultural potential, sea level rise, and altered water resources may threaten the existence of inhabitants of small island states, deltas, and vulnerable lowlands. While some areas may benefit, all countries would incur costs as agricultural systems adapt to new conditions.
Vulnerability and Risk: As many as a billion people live in poverty. For those also at risk of adverse climate change, perhaps the fundamental question is who has the right to take the risk that the environmental resources of the world's poor will be altered, to the point that their livelihoods are threatened? In each of the country studies reported here, there are areas and populations that are already highly vulnerable: even relatively modest changes in climatic resources imply significant shifts in the risk of crop failure and decreased food security.
Understanding and Coping: The complex dynamics of food security and global change hinder our understanding of present vulnerability and the future risk of climate change. Climate change is only one dimension of global trends that will interact to alter food security in the 21st century. The rate of climate change is critical--food production in Africa is already lagging behind population growth and demand. Many of the strategies cited as feasible mechanisms for coping with climate change are the same improvements already required to meet rapidly rising demand. For the majority of agricultural areas and farmers, the effects of gradual warming of 1-2deg.C particularly with a modest increase of precipitation) may be mitigated over time through purposeful strategies to increase resiliency and accelerate the pace of agricultural development.
Global Causes and Consequences: Climate change is caused by local emissions of greenhouse gases, but embedded in a global economy. The pathways of climate change are clearly global. Equally apparent, unmitigated climate change would have global consequences--adverse impacts on agroecological potential, water resources and health would fuel increased resource conflicts, environmental migration, and international food crises triggered by drought.
Alexandratos N., ed. 1988. World Agriculture: Toward 2000. An FAO Study New York: New York University Press.
Badr, E. and Darwish, S. 1991. Estimation of Egyptian Population Beneath a Poverty Line in 1990191. Oxford: Environmental Change Unit (manuscript).
Bohle, H.G. 1992. Hungerkrisen and ernährungssicherung. Geographisce Reserche 44(2): 78-87
Bohle, H.G. Gertel, J., Krings, T. and Krüger, F. 1991. Drought Hazard Assessment, Famine Disasters and Vulnerable Food Systems. Freiburg: University of Freiburg (manuscript).
Booth, T.H.. Stein, J.A., Hutchinson, M.F., and Nix, H.A. 1990. Identifying areas within a country climatically suitable for particular tree species: an example using Zimbabwe. International Tree Crops Journal 6: 1-16.
Borton, J. and Shoham, I. 1990. Guidelines for WFP Country Offices Preparing Baseline Vulnerability Maps: The Sudan Case Study. London: Relief and Development Institute.
Burkhi, R. 1992. Personal communication, Oxford (based on a forthcoming World Bank review of poverty in China).
Carter, T.R.. Parry, M.L., Nishioka, S., and Harasawa, T. 1992. Guidelines for Assessing Impacts of Climate change. Report of Working Group 11 of the Intergovernmental Panel on Climate Change. Oxford: Environmental Change Unit and Tsukuba: Global Environmental Change Center (forthcoming).
Central Bureau of Statistics. 1983. Population Projections for Kenya, 1980 2000. Nairobi: Central Bureau of Statistics.
Chambers. R. 1989. Editorial introduction: vulnerability, coping and policy. IDS Bulletin 20(2): 1-7.
Chen, R.S., Bender, W.H., Kates, R.W., Messer, E., and Millman, S.R. 1990. The Hunger Report: 1990. Providence RI: World Hunger Program, Brown University.
Chen, R.S. and Pitt, M.M. 1991. Estimating The Prevalence of World Hunger A Review of Methods and Data Providence, RI: World Hunger Program, Brown University
Cubasch, U. and Cess, R.D. 1990. Processes and modeling In Scientific Assessment of Climate Change, Houghton J., Jenkins, G.J., and Ephraums, J.J., eds., pp. 69-91 Cambridge: Cambridge University Press.
Dever, G.E.A., Sciegaj, M., Wade, T.E., and Lofton, T.C. 1988. Creation of a social vulnerability index for justice in health planning. Family and Community Health 10(4): 23-32.
Diagne, M. 1992. Changement du Climat et Production Agricole au Sénégal. Dakar Institut Sénégalais de Recherches Agricoles.
Downing, T.E 1991a. Assessing Socioeconomic Vulnerability to Famine: Frameworks Concepts and Applications. Washington: U.S. Agency for International Development, Famine Early Warning System Project and Providence, RI: World Hunger Program, Brown University.
Downing, T.E. 1991b. Vulnerability to hunger and coping with climate change in Africa. Global Environmental Change 1(5): 365-380.
Downing, T.E. 1988. Vulnerability to Hunger in Kenya: A National Estimate for 1984. Boulder National Center for Atmospheric Research (manuscript).
Downing, T.E., Gitu, K.W. and Kamau, C.M., eds. 1989. Coping with Drought in Kenya: National and Local Strategies. Boulder Lynne Rienner.
Downing, T.E., Lezberg, S., Williams, C., and Berry, L. 1990. Population change and environment in central and Eastern Kenya. Environmental Conservation 17(2): 123-133.
Dréze, J. and Sen, A. 1989. Hunger and Public Action. Oxford: Clarendon Press.
Famine Early Warning System (FEWS) Project. 1991. Vulnerability Assessment June 1991. Washington: Tulane/Pragma Group for the U.S. Agency for International Development.
Famine Early Warning System (FEWS) Project. 1990. Vulnerability Assessment June ]990. Washington: Tulane/Pragma Group for the U.S. Agency for International Development.
Gates, W.L., Mitchell, J.F.B., Boer, GJ., Cubasch, U., and Meleshko, V.P. 1992. Climate modeling, climate prediction and model validation. In Climate Change 1992; The Supplementary Report to the IPCC Scientific Assessment, Houghton, J., Calendar, B.A., and Varney, SK., eds., pp. 97- 134. Cambridge: Cambridge University Press.
Glieck, P.H. 1988. Climatic change and California: past, present and future vulnerabilities. In Societal Response to Regional Climatic Change: Forecasting by Analogy Glantz, M.H., ed., pp. 307-327. Boulder West view.
Govereh, J. and Mudimu, G.D. 1991. Prospects for increasing household food security and income through increased crop productivity and diversification in low rainfall areas of Zimbabwe. In Market Reforms, Research Policies and SADCC Food Security, Rukuni, M. and Wyckoff, J.B., eds. Proceedings of the Sixth Annual Conference of Food Security Research in Southern Africa, November 12-14, 1990. Harare: Department of Agricultural Economics and Extension, University of Zimbabwe.
Gwynne, R.N. 1992. Outward Orientation and Marginal Environments: The Question of Sustainable Development in Chile's Norte Chico. Paper presented at the Annual Conference of the Institute of British Geographers, Swansea. Birmingham: University of Birmingham (manuscript).
Gwynne, R.N., Meneses, C., and Ortiz, 1. 1992. Export Growth, Land Use Change and the Impact of Settlement Patterns: A Comparison of Two "Norte Chico" Valleys. Working Paper No. 62. Birmingham: University of Birmingham .
Han, M., Hou, J. and Wu, L. 1990. Adverse Impact of Projected One Meter Sea Level Rise on China's Coastal Environment and Cities: A National Assessment. College Park: University of Maryland (manuscript).
Hansen, J., Fung, A., Lacis, D., Rind, G., Russell, G., Lebedeff, S., Ruedy, R., and Stone, P. 1988. Global climate changes as forecast by the GISS 3-D model. Journal of Geophysical Research 93(D8): 9341-9$64.
Higgins, G.M., Kassam, A.H., Naiken, L., Fischer. G., and Shah, M.M. 1982. Potential Population Supporting Capacities of Lands in the Developing World. Rome: Food and Agriculture Organization.
Houghton. J., Calendar, B.A., and Varney, S.K., eds. 1992. Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment Cambridge: Cambridge University Press.
Houghton. J.. Jerkins, G.J.. and Ephraums, J.J., eds. 1990. Scientific Assessment of Climate Change. Cambridge: Cambridge University Press.
Hunt, D.M. 1984. The Impending Crisis in Kenya: The Case for Land Reform. Brookfield, VT: Gower.
Huss-Ashmore. R. and Katz, S.H.. eds. 1989. African Food Systems in Crisis: Part One: Micro perspectives. Food and Nutrition in History and Anthropology, Volume 7. New York: Gordon and Breach.
Iliffe, J. 1990. Famine in Zimbabwe: 1890-1960. Gweru: Mambo Press.
James, W.P.T. and Schofield, E.C. 1990. Human Energy Requirements: A Manual for Planners and Nutritionists. Oxford: Oxford University press.
Kasperson, R.E., Dow, K., Golding, D., and Kasperson, J.X., eds. 1990. Understanding Global Environmental Change: The Contributions of Risk Analysis and Management. Worcester, MA: Clark University.
Kassam, A.H., van Velthuizen, H.T., Fischer, G.W., and Shah, M.M. 1991. Agroecological Land Resources Assessment for Agricultural Development Planing: A Case Study of Kenya; Resources Database and Land Productivity. Rome: FAO and Laxenburg: International Institute for Applied Systems Analysis.
Kates, R.W., Ausubel, J.H., and Berberian, M., eds. 1985. Climate Impact Assessment. New York: Wiley.
Kates, R.W., Chen. R.S., Downing, T.E., Kasperson, J.X.. Messer, E., and Millman, S.R. 1988. The Hunger Report: 1988. Providence, RI: World Hunger Program, Brown University.
Kates, R.W., Chen, R.S., Downing, T.E., Kasperson, J.X., Messer, E., and Millman, S.R. 1989. The Hunger Report: Update 1989. Providence, RI: World Hunger Program, Brown University.
Liverman, D.M.1990. Vulnerability to global environmental change. In Understanding Global Environmental Change: The Contributions of Risk Analysis and Management Kasperson, R.E., Dow, K., Golding, D., and Kasperson, J.X., eds., pp. 27-44. Worcester, MA: Clark University.
Magadza, C.H.D. 1992. Climate Change: Some Likely Impacts in Zimbabwe. Lake Kariba: University Lake Kariba Research Station, University of Zimbabwe.
Magalhães, A.R. and Neto, E.B. 1991. Impactos Socials e Economics de Variacões Climaticãs e Respostas Governamentais no Brasil. Fortaleza, Brazil: Imprensa Oficial do Ceara.
Manabe, S. and Wetherald, R.T. 1987. Large scaie changes of soil wetness induced by an increase in atmospheric carbon dioxide. Journal of Atmospheric Science 44: 1211-1235.
Manarolla, J.A. 1989. A Methodology for Ranking Countries According to Relative Food Insecurity. Paper presented at the A.I.D. Economists Conference, November 12- 17. Washington: U.S. Agency for International Development.
Maskrey, A. 1989. Disaster Mitigation: A Community Based Approach. Development Guidelines No. 3. Oxford: Oxfam.
McKee, D. and Vilhjalmsson, R. 1986. Life stress, vulnerability, and depression: a methodological critique of Brown et al. Sociology 20(4): 589-599.
Meneses,C. 1992. Caracteristicas del Uso del Suelo Agua en los Valles del Norte Chico, con Expecial Enfasis en el Valle del Rio Guafulame Santiago: Department of Geography, University of Chile.
Messer, E. l 989. Seasonality in food systems: an anthropological perspective on household food security. In Seasonal Variability in Third World Agriculture: The Consequences for Food Security, Sahn, D.E., ed., pp. 151-175. Baltimore, MD: Johns Hopkins University
Messer, E. 1986. The "small but healthy" hypothesis: historical, political, and ecological influences on nutritional standards. Human Ecology 14: 57-75.
Messer, E. 1984. Anthropological perspectives on diet. Annual Review of Anthropology 13: 205-249.
Millman, S.R. and Chen, R.S.1991. Measurement of Hunger: Defining Thresholds. Providence, RI: World Hunger Program, Brown University.
Millman, S. and Kates, R.W. 1989. Toward understanding hunger. In Hunger in History: Food Shortage, Poverty, and Deprivation, Newman, LF., Crossgrove, W., Kates, R.W., Matthews R., and Millman, S., eds., pp. 3-24. New York: Basil Blackball.
Millman, S.R., Chen, R.S., Emlen, J., Haarmann, V., Kasperson, J.X., and Messer, E. 1991. The Hunger Report: Update 1991. Providence RI: Brown University World Hunger Program.
Moore, D.G., Tappan, G.G., Howard, S.M., Lietzow, R.W., Nadeau, C.A., Renison, W., Olsson, J., and Kite, R. 1991. Geographic Modeling of Human Carrying Capacity from Rainfed Agriculture: Senegal Case Study Sioux Falls, SD: EROS Data Center, U.S. Geological Survey.
Mortimore, M., ed. 1991. Environmental Change and Dryland Management in Machakos District, Kenya 1939-90. Working Paper 53. London: Overseas Development Institute.
Muchena, P. 1991. Impact of Potential Climate Change for Agriculture in Zimbabwe: Crop Productivity. Harare: Plant Protection Research Institute (manuscript).
Murai, S., Honda, Y., Asakura, K., and Goto, S. 1991. An Analysis of Global Environment by Satellite Remote Sensing What Population Can the Earth Feed? Tokyo: Institute of Industrial Science, University of Tokyo (manuscript, cited in Clark University, Critical Zones in Global Environmental Change Newsletter #2, March 1992).
Newman, L.F., Crossgrove, W., Kates, R.W., Matthews, R., and Millman, S., eds. 1989. Hunger in History: Food Shortage, Poverty, and Deprivation. New York: Basil Blackwell.
Ortiz, J. 1992. Socio-Spatial Impact of the Agricultural Modernization Process in the Chilean "Norte Chico" and Socio-Geographic Implications of the Drought. Santiago: Department of Geography, University of Chile.
Parry, M.L., Carter, T.R., and Konijn, N.T., eds. 1988. The Impact of Climatic Variations on Agriculture, Volume 2, Assessments in Semi-Arid Areas. Dordrecht: Kluwer.
Parry, M.L., de Rozari, M.B., Chong, A.L., and Panich, S., eds. 1992. The Potential Socio-Economic Effects of Climate Change in South-East Asia. Nairobi: U.N. Environment Programme.
Peña, H. and Rivera, A. 1991. Efectos Potenciales de un Cambio de Clima en la Agricultura y Desarrollo Regional en el Norte Chico: Analisis de la Variabilidad Hidrologica. Santiago: Direction General de Aguas.
Pernetta, J.C. 1992. Impacts of climate change and sea-level rise on small island states: National and international responses. Global Environmental Change 2(1): 19-31.
Romero, H.I. and Ihl, M. 1992. Short-term Climatic Change and Topoclimatology of the Chilean Semi-Arid Norte Chico. Santiago: Department of Geography, University of Chile.
Rosenzweig, C., Parry, M.L., Fischer, G., and Frohberg, K. 1993. Climate Change and World Food Supply. Oxford: Environmental Change Unit.
Ruttenberg, S. 1981. Climate, food and society. In Climate's Impact on Food Supplies, Slater, L.E. and Levin, S.K., eds., pp. 23-38. Boulder, CO: Westview.
Sahn, D.E., ed. 1989. Seasonal Variability in Third World Agriculture: The Consequences for Food Security. Baltimore: Johns Hopkins University Press.
Santibãnez, F. 1992. Impact on Agriculture due to Climatic Change and Variability in South America: A Case Study of the Arid Zone of Chile. Santiago: University of Chile.
Shipton, P. 1990. African famines and food security: anthropological perspectives. Annual Review of Anthropology 19: 353-394.
Smit, B. 1989. Climate warming and Canada's comparative position in agriculture. Climate Change Digest CCD8901. Downsview: Environment Canada.
Swift, J. 1989. Why are rural people vulnerable to famine? IDS Bulletin 20(2): 8-15.
Tegart, W.J.MCG.., Sheldon, G.W., and Griffiths, D.C. 1990. Climate Change: The IPCC Impacts Assessment. Canberra: Australian Government Publishing Service.
Wilson, C.A. and Mitchell, J.F.B. 1987. A doubled CO2 climate sensitivity experiment with a GCM including a simple ocean. Journal of Geophysical Research 92: 13315-13343.
World Bank. 1990. World Development Report 1990. Washington: World Bank.
World Bank. 1989. Social Indicators of Development 1989. Washington: World Bank.
World Bank. 1986. Poverty and Hunger: Issues and Options for Food Security in Developing Countries. Washington: World Bank.
This research report is based on a suite of country studies of climate change in developing countries carried out for the U.S. Environmental Protection Agency's Climate Change and International Agriculture project (directed by Professor M.L. Parry and Dr. C. Rosenzweig). The members of the country study teams are listed below. I am also indebted to colleagues at the World Hunger Program at Brown University (particularly Dr. R.W. Kates and Dr. R.S. Chen) for development of concepts of vulnerability and food security. Specific comments were received from B. Boardman, A.P. Brignal, P.A. Harrison, G.J. Kenny, D. Norse, A. Street-Perrot, F. Wangati, and R. White. The contributions of my colleagues are gratefully acknowledged. This report is my interpretation of their contributions--credit is shared, responsibility for errors and omissions is mine.
Other research reports that expand on the conclusions of the Climate Change and International Agriculture project are in preparation.
Working titles are:
Climate Change and World Food Supply (in cooperation with the Goddard Institute of Space Studies).
Climate Change in the Norte Chico, Chile; Volume I: Climate, Agriculture and Hydrology and Volume 11: Sociogeographic Consequences and Regional Development (in cooperation with the University of Birmingham and University of Chile).
Climate Change in Zimbabwe: Spatial and Household Shifts in Risk (in cooperation with the University of Zimbabwe).
Vulnerable Groups, Land Use and Climate Change in Kenya (in cooperation with the International Institute for Applied Systems Analysis).
Climate Change in Senegal: Implications for Rural Agricultural Development (in cooperation with the Institut Sénégalais de Recherches Agricoles).
LIST OF COLLABORATORS
Professor Alfredo Apey
Comisión del Desarollo Rural
Ministerio de la Agricultura
Teatintos 40, Casilla 31-D
Dr. Emad Badr
Agricultural Economic Research
Nadi El Said Street
Giza, Cairo, Egypt
Dr. Trevor H. Booth
Division of Forestry, CSIRO
Banks Street, Yarralumla
P.O. Box 4008
Queen Victoria Terrace
Canberra ACT 2600, Australia
Dr. Samir Darwish
Agricultural Economic Research
Nadi El Said Street
Giza, Cairo, Egypt
Dr. Madiagne Diagne
Dr. H.M. Eid
Soil and Water Research Institute
Agricultural Research Center
Ministry of Agriculture
Dr. Günther Fischer
A-2361 Laxenburg, Austria
Dr. Robert N. Gwynne School of Geography University of Birmingham Edgbaston Birmingham B15 2TT, U.K.
Dr. Peter Hutchinson
Zimbabwe Meteorological Services
P.O. Box BE 150
Mrs. R.P. Karimanzira
Zimbabwe Meteorological Services
P.O. Box BE 150
Dr. Ann Kelly International Food Policy Research Institute (IFPRI) c/o Institut Senegalais de Recherches Agricoles Centre de Recherches Oceanographiques de Dakar Thiaroye BP 2241, Dakar, Senegal
Dr. Gavin J. Kenny Environmental Change Unit University of Oxford Oxford OXI 3TB, U.K.
Mr. Joseph Kinuthia
Kenya Meteorological Department
P.O. Box 30259
Dr. C.H.D. Magadza
University lake Kariba Research
P.O. Box 48
Mr. Juan Manuel U.
Facultad de Ciencias Agraria y
Universidad de Chile
Professor Claudio Meneses
Departmento de Geografia
Facultad de Arquitectura y
Universidad de Chile
Marcoleta No. 250
Dr. Donald G. Moore and Dr. G. Tappan
U.S. Geological Survey
EROS Data Center
Sioux Falls, SD 57198, USA
Dr. Paul Muchena
Plant Protection Research Institute
Box 8100, Causeway
Dr. Godfrey D. Mudimu
Department of Agricultural
Economics and Extension
University of Zimbabwe
P.O. Box MP 167
Professor Jorge Ortiz
Departmento de Geografia
Facultad de Arquitectura y
Universidad de Chile
Marcoleta No. 250, Casilla 3387
Mr. W. Ottichilo Department of Resource Surveys and Remote Sensing P.O. Box 47146 Nairobi, Kenya
Professor Martin L. Parry Environmental Change Unit la Mansfield Road Oxford OXI 3TB, U.K.
Ing. Humberto Peña T.
Departmento de Estudios
Dirección General de Aguas
Morande 59, 7deg. Piso
Dr. Hugo Romero A.
Facultad de Arquitectura y
Universidad de Chile
Marcoleta No. 250
Dr. Cynthia Rosenzweig
Goddard Institute for Space Studies
New York, NY 10025, USA
Dr. Fernando Santibãnez
Facultad de Ciencias Agraria y
Universidad de Chile
Dr. Mamadou Sidiba Institut Sénégalais de Recherches Agricoles Centre de Recherches Oceanographiques de Dakar Thiaroye BP 2241 Dakar, Senegal
Dr. Benson Wafula
Katumani National Dryland Farming Research Station
P.O. Box 340
Dr. Fred J. Wangati
P.O. Box 29203