In the coming decades, global agriculture faces the prospect of a changing climate (International Panel on Climate Change (IPCC), 1990a, 1992), as well as the known challenge of continuing to feed the world's population, projected to double its present level of five billion by about the year 2060 (International Bank for Reconstruction and Development/World Bank, 1990). The prospective climate change is global warming (with associated changes in hydrological regimes and other climatic variables) induced by the increasing concentration of radiatively active greenhouse gases (IPCC, 1990a, 1992). Climate change could have far-reaching effects on patterns of trade among nations, development, and food security. To help prepare for this uncertain but challenging future, this study examined the potential effects of climate change on crop yields, world food supply, and regions vulnerable to food deficits.
Despite technological advances such as improved crop varieties and irrigation systems, weather and climate are still key factors in agricultural productivity. For example, weak monsoon rains in 1987 caused large shortfalls in crop production in India, Bangladesh, and Pakistan, contributing to reversion to wheat importation by India and Pakistan (World Food Institute, 1988). The 1980s also saw the continuing deterioration of food production in Africa, caused in part by persistent drought and low production potential, and international relief efforts to prevent widespread famine. The effects of climate on agriculture in individual countries cannot be considered in isolation. Agricultural trade has grown dramatically in recent decades and now provides significant increments of national food supplies to major importing nations and substantial income for major exporting nations (Table 1). These examples emphasize the close links between agriculture and climate, the international nature of food trade and food security, and the need to consider the impacts of climate change in a global context.
Recent research has focused on regional and national assessments of the potential effects of climate change on agriculture. These efforts have, for the most part, treated each region or nation in isolation, without relation to changes in production in other places. At the same time, there has been a growing emphasis on understanding the interactions of climatic, environmental, and social factors in a wider context (Parry, 1990), leading to more integrated assessments of potential impacts in national impact studies completed in the United States (Adams et al., 1990; Smith and Tirpak, 1989), Canada (Smit, 1989), Brazil (Magalhaes (1992) and Indonesia, Malaysia, and Thailand (Parry et al., 1992). Regional studies have been conducted in high latitude and semi-arid agricultural areas (Parry et al., 1988a, 1988b), and the US Midwest (Rosenberg and Crosson, 1991). The results of these and other agricultural impact studies have been summarized in the IPCC Working Group II Report (IPCC, l990b). Sensitivity studies of world agriculture to potential climate changes have indicated that the effect of moderate climate change on world and domestic economies may be small, as reduced production in some areas is balanced by gains in others (Kane et al., 1991; Tobey et al., 1992). However, there has to date been no integrated (i.e.. combined biophysical and economic) assessment of the potential effects of climate change on world agriculture.
This report describes the study of the potential effects of climate change on world food supply. The research involved estimating the responses of crop yields to greenhouse gas-induced climate change scenarios and then simulating the economic consequences of these potential changes in crop yields. The analysis provides estimates of changes in terms of production and prices of major food crops and the number of people at risk of hunger.
The structure research methods for the world food supply study are illustrated in Figure 1. There were two main components:
Estimation of potential changes in crop yield
Agricultural scientists in 18 countries (see Appendices I and 2) simulated potential changes in grain yields using compatible crop models developed by the US Agency for International Development's International Benchmark Sites Network for Agrotechnology Transfer (lBSNAT 1989). The crops modelled were wheat, rice, maize, and soybeans. Wheat, rice and maize account for approximately 85% of the world cereal exports: soybean accounts for about 67% of trade in protein cake equivalent. The crop models were run for current climate conditions, for arbitrary changes in climate (+2deg. and +4deg.C increase in temperature and +/-20% precipitation), and for climate conditions predicted by general circulation models (GCMs) for doubled atmospheric CO2 levels. The direct effects of increasing levels of CO2 on crop growth and water use were taken into account. For the GCM climate change scenarios, results from crop model sites were aggregated by current regional production to estimate national crop yield changes, at two levels of farmer adaptation. The national crop yield changes were then extrapolated to provide estimates of yield changes (for the three GCM scenarios) for other countries and crops included in the food trade analysis.
Estimation of world food trade responses
The national crop yield changes derived from the first component of the study were used as inputs into a world food trade model., the Basic Linked System (BLS), developed at the International Institute for Applied Systems Analysis (IIASA) (Fischer et al., 1988). The BLS was run first for a reference scenario projecting the agricultural system to the year 2060 assuming no change in climate, and then with the three GCM climate change scenarios. Other BLS simulations included the effects of two levels of farmer adaptation and scenarios of different future trade liberalization policy, and economic and population growth rates. Outputs from the BLS simulations provided information on food production, food prices and on the number of people at risk of hunger (defined as the population with an income insufficient to either produce or procure their food requirements) for these scenarios projected up to the year 2060.
Throughout the climate change study, a distinction is made between farm-level adaptations tested by the crop models which result in yield changes, and economic adjustments to the yield changes tested by the BLS world food trade model which result in national and regional production changes and price responses. Farm-level adaptations tested in the crop models include: plantingdate shifts, more climatically adapted varieties, irrigation, and fertilizer application. Economic adjustments represented by the BLS include: increased agricultural investment, reallocation of agricultural resources according to economic returns (including crop switching), and reclamation of additional arable land as a response to higher cereal prices. These economic adjustments are assumed not to feed back to the yield levels predicted by the crop modeling study.
The crop yield and economic modelling components are described in greater detail below.
Arbitrary climate sensitivity tests were conducted to test crop model responses to a range of temperature (+2deg.C and +4deg.C) and precipitation (+/-20%) changes. While this type of sensitivity analysis is dissociated from the processes that influence climate (temperature and precipitation are physically related in any given region), it does provide better understanding of the factors affecting crop model responses. It can also help to identify climatic thresholds of critical impacts. Sensitivity tests were carried out in 13 of the 18 countries. The sensitivity test results were not utilized in the economic analysis with the BLS world food trade model because of their lack of realism, e.g., temperature change at high latitudes is predicted to be greater than the global mean in winter (IPCC, l990a), rather than increase by 2deg. or 4deg.C in all regions of the world throughout the year. Farm-level adaptations were not tested in the sensitivity studies.
Scenarios based on general circulation model results
Scenarios of climate change were developed in order to estimate their effects on crop yields and food trade. A climate change scenario is defined as a physically consistent set of changes in meteorological variables, based on generally accepted projections of concentrations of carbon dioxide (CO2) and other greenhouse gases in the atmosphere (thought to be the likely cause of future climate change). The set of scenarios used is intended to capture a range of possible effects and set limits on the associated uncertainty.
The scenarios for this study were created by changing observed daily data from the current climate (1951-80) according to doubled CO2 simulations of three general circulation models available at the initiation of the study in 1989. The GCMs used are those from the Goddard Institute for Space Studies (GISS), Geophysical Fluid Dynamics Laboratory (GFDL), and United Kingdom Meteorological Office (UKMO) (Table 2). The temperature changes of these GCM scenarios (4.0-5.2deg.C) are at or near the upper end of the range (1.5-4.5deg.C) projected for doubled CO2 warming by the IPCC (IPCC, l990a, 1992). The GISS and GFDL scenarios, however, are near the mean temperature change (3.8deg.C) of recent doubled CO2 experiments documented for atmospheric GCMs with a seasonal cycle and a mixed-layer ocean (IPCC, 1992). Mean monthly changes in temperature, precipitation, and solar radiation from the appropriate GCM gridbox were applied to observed daily climate records to create climate change scenarios for each site.
Scientific uncertainty of GCM predictions
While GCMs currently provide the most advanced means of predicting the potential future climatic consequences of increasing amounts of radiatively active trace gases, their ability to simulate current climate varies considerably from region to region. They have been shown to simulate current surface air temperatures reasonably well, but do not reproduce current precipitation as accurately (IPCC, l990a). Of special importance for agricultural climate change impacts, there is a notable lack of consensus among GCMs in prediction of regional soil moisture changes (Kellogg and Zhao, 1988). Furthermore, GCMs are not yet able to produce reliable projections of changes in climate variability, such as alterations in the frequencies of drought and storms, even though these could affect crop yields significantly.
CO2 level and timing
For the crop modeling part of this study, climate changes from doubled CO2 GCM simulations are utilized with an associated level of 555 ppm CO2; the assumed timing for the BLS world food trade projections is that these conditions will occur in 2060. This approach, which mixes equilibrium climate and transient CO2 projections, is the best that can be done given the current lack of availability of GCM transient climate change simulations for impact studies and the need for a dynamic time course for the economic model. It is not known what rates of future emissions of trace gases will be and when the full magnitude of their effects will be realized.
Because atmospheric concentrations of other greenhouse gases besides CO2 (e.g.. methane (CH4), nitrous oxide (N2O)), and the chlorofluorocarbons (CFCs)) are also increasing, an 'effective CO2 doubling' has been defined as the combined radiative forcing  of all greenhouse gases having the same forcing as doubled CO2 (usually defined as ~600 ppm; see Table 2 for doubling levels of the three GCMs used in this study). Level of CO2 is important when estimating potential impacts on crops, because crop growth and water use have been shown to benefit from increased levels of CO2 (Cure and Acock, 1986). A CO2 level of 555 ppm was associated with the effective doubled CO2 climate projections for use in the crop modeling simulations. This was based on the GISS GCM transient trace gas scenario A described in Hansen et al. (1988), in which the simulated climate has warmed to the effective doubled CO2 level of about 4degC by 2060. This level assumes that non-CO2 trace gases contribute ~15% of the change in radiative forcing from 300 to ~600 ppm.
The IBSNAT crop models were utilized by the participating agricultural scientist to estimate how climate change and increasing levels of carbon dioxide may alter yields of world crops at 112 sites in 18 countries. The sites represent both major and minor production areas at low, mid, and high latitudes (Figure 2). The crop models used were CERES-Wheat (Ritchie and Otter, 1985; Godwin et al., 1989), CERES-Maize (Jones and Kiniry, 1986; Ritchie et al., 1989), CERES-Rice (paddy and upland) (Godwin et al., 1993), and SOYGRO (Jones et al., 1989).
The IBSNAT models are comprised of parameterizations of important physiological processes responsible for plant growth and development, evapotranspiration, and partitioning of photosynthate to produce economic yield. The simplified functions enable prediction of the growth of crops as influenced by the major factors that affect yields, ie. genetics, climate (daily solar radiation, maximum and minimum temperatures, and precipitation), soils, and management practices. This type of dynamic process crop growth model is considered to be a significant advance over traditional regression-based methods (eg. Thompson, 1969) which were used to estimate crop yields from simple climate and management inputs with geographical and temporal specificity. The models include a soil moisture balance submodel so that they can be used to predict both rainfed and irrigated crop yields. The cereal models simulate the effects of nitrogen fertilizer on crop growth, and these were studied in several countries in the context of climatic change. For the most part, however, the results of this study assume optimum nutrient levels.
The IBSNAT models were selected for use in this study because they have been validated over a wide range of environments (e.g.. Otter-Nacke et al., 1986) and are not specific to any particular location or soil type. They are better suited for large-area studies in which cropgrowing and soil conditions differ greatly, than more detailed physiological models that have not been as widely tested. The validation of the crop models over different environments also improves the ability to estimate effects of changes in climate. Because the crop models have been tested over essentially the full range of temperature and precipitation regimes where crops are grown in today's climate, and to the extent that future climate change brings temperature and precipitation regimes within these ranges, the models may be considered useful tools for assessment of potential climate change impacts. Furthermore, because management practices, such as the choice of varieties, the planting date, fertilizer application, and irrigation, may be varied in the models, they permit experiments that simulate adaptation by farmers to climatic change.
Physiological effects of C02
Most plants growing in experimental environments with increased levels of atmospheric CO2, exhibit increased rates of net photosynthesis (i.e.. total photosynthesis minus respiration) and reduced stomatal openings. (Experimental effects of CO2 on crops have been reviewed by Acock and Allen (1985) and Cure (1985).) Partial stomatal closure leads to reduced transpiration per unit leaf area and, combined with enhanced photosynthesis, often improves water-use efficiency (the ratio of crop biomass accumulation or yield to the amount of water used in evapotranspiration). Thus, by itself, increased CO2, can increase yield and reduce water use (per unit biomass).
The crop models used in this study account for the beneficial physiological effects of increased atmospheric CO2 concentrations on crop growth and water use (Peart et al., 1989). Ratios were calculated between measured daily photosynthesis and evapotranspiration rates for a canopy exposed to high CO2 values, based on published experimental results (Allen et al., 1987; Cure and Acock, 1986; and Kimball, 1983), and the ratios were applied to the appropriate variables in the crop models on a daily basis. The photosynthesis ratios ppm (555 ppm CO2/330 ppm CO2) for soybean, wheat and rice. and maize were 1.21, 1.17, and 1.06. respectively. Changes in stomatal resistance were set at 49.7/34.4 s/m for C3 crops and 87.4/55.8 s/m for C4 crops, based on experimental results by Rogers et al. (1983). As simulated in this study, the direct effects of CO2 may bias yield changes in a positive direction, since there is uncertainty regarding whether experimental results will be observed in the open field under conditions likely to be operative when farmers are managing crops. Plants growing in experimental settings are often subject to fewer environmental stresses and less competition from weeds and pests than are likely to be encountered in farmers' fields. Recent field free-air release studies have found overall positive CO2 effects under current climate conditions (Hendry, 1993).
Limitations of crop growth models
The crop models embody a number of simplifications. For example, weeds, diseases, and insect pests are assumed to be controlled; there are no problem soil conditions (e.g.. salinity or acidity); and there are no extreme weather events such as tornadoes. The models are calibrated to experimental field data which often have yields higher than those currently typical under farming conditions. Thus, the absolute effects of climatic change on yields in farmers' fields may be different from those simulated by the crop models.
The crop models simulate the current range of agricultural technologies available around the world, including the use of high-yielding varieties that are responsive to technological inputs, but by the year 2060 agricultural technology is likely to be very different. The models may he used to test the effects of some potential improvements in agricultural production, such as varieties with higher thermal requirements and installation of irrigation systems, but do not include possible future improvements. (The BLS economic model used in the study does include future trends in yield improvement, but not technological developments induced by negative climate change impacts.) Finally, models for crops such as millet and cassava were not yet sufficiently tested for use in this study. Potential yield changes of such crops, which may respond differently to both climate change and increases in CO2, are needed for better assessment of climate change impacts in tropical and sub-tropical regions.
Crop modelling simulation experiments were performed for baseline climate (1951-80), arbitrary sensitivity tests, and GCM doubled CO2 climate change scenarios with and without the physiological effects of C02. This involved the following tasks:
Aggregation of site results
Table 3 shows the percentages of world production of wheat, rice, maize, and soybean for the countries in which simulations were conducted. Simulations were carried out in regions representing 70-75% of the current world production of wheat and maize. Even though model runs for soybean were conducted in only two countries (Brazil and USA), these together account for 76% of world production. Rice production was less well represented in the model simulations than the other crops, because India, Indonesia and Vietnam have significant production areas not included in the study. Further research is needed in these key countries in order to improve the reliability of the projections of climate change impacts on rice production.
Crop model results for wheat, rice, maize and soybean from the 112 sites in the 18 countries were aggregated by weighting regional yield changes (based on current production) to estimate changes in national yields. The aggregations were either calculated by the participating scientists or developed jointly with them (see Rosenzweig and Iglesias, 1992). The scientists in each country selected sites representative of major agricultural regions, described the regional agricultural practices, and provided production data for estimation of regional contributions to the national yield changes. Other production data sources included the United Nations Food and Agriculture Organization (FAO, 1988), the US Department of Agriculture (USDA) Crop Production Statistical Division, and the USDA International Service. The regional yield estimates represent the current mix of rainfed and irrigated production and the current crop varieties, nitrogen management and soils.
Yield change estimates for crops and regions not simulated
Changes in national yields of other crops and commodity groups and for other regions not simulated were estimated based on three criteria:
Limitations of crop yield change estimates
The primary source of uncertainty in the estimates lies in the sparseness of the crop modelling sites and the fact that they may not adequately represent the variability of agricultural regions within countries, the variability of agricultural systems within similar agroecological zones, or dissimilar agricultural regions. However, since the site results relate to regions that account for about 70% of world grain production. the conclusions concerning world totals of cereal production contained in this study are believed to be substantiated adequately. Another source of uncertainty lies in the simulation of grain crops only, leading to estimation of yield changes for other commodities such as root crops and fruit based primarily on previous estimates. The previous estimates tended to be less negative than the crop responses modelled in this study, and this introduced a bias in favour of these other crops in the world food trade model.
In each country, the agricultural scientists used the crop models to test possible responses to the worst climate change scenario (this was usually, but not always, the UKMO scenario). These adaptations included change in planting date, change of cultivar, irrigation, fertilizers and change of crop. Irrigation simulations in the crop models assumed automatic irrigation to field capacity when plant available water dropped to 50% and 100% irrigation efficiency. All adaptation possibilities were not simulated at every site and country: choice of adaptations to be tested was made by the participating scientists, based on their knowledge of current agricultural systems (Table 5).
For the economic analysis in the BLS, the crop model results reported by the participating scientists were then grouped into two levels of adaptation. Adaptation Level 1 implies little change to existing agricultural systems, reflecting relatively easy farmer response to a changing climate. Adaptation Level 2 implies more substantial change to agricultural systems, possibly requiring resources beyond the farmer's means.
Adaptation Level 1 includes:
Limitations of adaptation analysis
The adaptation simulations were not comprehensive because all possible combinations of farmer responses were not tested at every site. Spatial analyses of crop, climatic, and soil resources are needed to test fully the possibilities for crop substitution. Neither the availability of water supplies for irrigation nor the costs of adaptation were considered in this study; these are both critical needs for further research. A related study on the Integrated Impacts of Climate Change on Egypt. which utilized the results of this work, does address future water availability for national agricultural production in that country (Strzepek et al., 1993).
At the local level, there may be social or technical reasons why farmers are reluctant to implement adaptation measures. For example, increased fertilizer application and improved seed stocks may be capital-intensive and not suited to indigenous agricultural strategies. Furthermore, such measures may not necessarily result in sustainable production increases. In the case of irrigation, initial benefits may eventually give way to soil salinization and lower crop yields.
Thus, Adaptation Level 2 represents a fairly optimistic assessment of world agriculture's response to changed climate conditions as characterized by the GCMs tested in this study, possibly requiring substantial changes in current agricultural systems, investment in regional and national agricultural infrastructure, and policy changes. However, changes in regional, national and international agricultural policies relating to farm-level adaptation were beyond the scope of the analysis.
The world food system is a complex dynamic interaction of producers and consumers, interacting through global markets. Related activities include input production and acquisition, transportation, storage and processing. While there is a trend towards internationalization in the world food system, only about 15% of total world production currently crosses national borders (Fischer et al., 1990). National governments shape the system by imposing regulations and by investments in agricultural research, infrastructure improvements, and education. The system functions to meet the demand for food, to produce food in increasingly efficient ways, and to trade food within and across national borders. Although the system does not guarantee stability, it has generated long-term real declines in prices of major food staples (Fischer et al., 1990).
The Basic Linked System consists of linked national agricultural sector models. The BLS was designed at IIASA for food policy studies but it also can be used to evaluate the effect of climate-induced changes in yield on world food supply and agricultural prices. It consists of 16 national (including the European Community (EC)) models with a common structure, four models with country-specific structure, and 14 regional group models (Table 7). The 20 models in the first two groups cover about 80% of attributes of the world food system such as demand, land and agricultural production. The remaining 20% is covered by 14 regional models for countries which have broadly similar attributes (e.g.. African oil-exporting countries, Latin American high-income exporting countries, Asian low-income countries etc.). The grouping is based on country characteristics such as geographical location, income per capita and the country's position with regard to net food trade.
The BLS is a general equilibrium model system, with representation of all economic sectors, empirically estimated parameters and no unaccounted supply sources or demand sinks (see Fischer et al. (1988) for a complete description of the model). In the BLS, countries are linked through trade, world market prices and financial flows (Figure 3). It is a recursively dynamic system: a first round of exports from all countries is calculated for an assumed set of world prices, and international market clearance is checked for each commodity. World prices are then revised, using an optimizing algorithm, and again transmitted to the national model. Next, these generate new domestic equilibria and adjust net exports. This process is repeated until the world markets for all commodities are cleared. At each stage of the iteration, domestic markets are in equilibrium. This process yields international prices as influenced by governmental and inter-governmental agreements. The system is solved in annual increments, simultaneously for all countries. Summary indicators of the sensitivity of the world system include world cereal production. world cereal prices and prevalence of population in developing countries at risk of hunger.
The BLS does not incorporate any climate relationships per se. Effects of changes in climate were introduced to the model as changes in the average national or regional yield per commodity. Ten commodities are included in the model: wheat, rice, coarse grains, protein feed. bovine and ovine meat, dairy products, other animal products, other food, non-food agriculture and non-agriculture. Yield change estimates for coarse grains were based on the percentage of maize grown in the country or region; soybean crop model results were used to estimate the protein feed category; the estimates for the non-grain crops were based on the modelling grain crops and previous estimates of climate change impacts as described above. A positive bias toward non-grain crops was introduced by this procedure, since the previous estimates of yield changes of the non-grain crops were less negative than the modelled results from this study.
Economic growth rates
Economic growth rates are a product of several BLS functions. Non-agricultural production utilizes a Cobb-Douglas production function with labour and capital as production factors. Non-agricultural labour input depends primarily on population growth and somewhat on relative prices between agriculture and non-agriculture by means of a sector migration function. Capital accumulation depends on investment and depreciation, which in turn depend on saving and depreciation rates. Depreciation rates and saving rates are estimated from historical data and are kept constant after 1990. There is an exogenous assumption based on historical data for technical progress in the production function. For the lower growth scenario, the savings rate was reduced, resulting in about 10% lower GDP in 2060.
The economic growth rates predicted by the BLS in the reference case follow historical trends as shown in Table 8. For the period 1980 to 2060, the BLS produces a growth of 1.3%, 1.7%, and 2.4% annually for world, developed, and developing countries, respectively, as compared to an average population growth of 1.1 %, 0.3%, and 1.3%.
In general, the rate of exogenous technical progress starts from historical values and for cereal crops approaches 0.5% per annum by 2060. Representing improvement in agriculture productivity due to technological progress, the annual yield trends used in the BLS for the period 1980-2000 are 1.2%, 1.0%, and 1.7% for global, developed country, and developing countries, respectively. According to FAO data, yields have been growing at an average of around 2% annually during the period 1961-90, both for developed and developing (excluding China) countries (FAO, 1991). Recent increase (1965-85) in annual productivity for less-developed countries is about 1.5%/year. In the 1980s, however, yields grew globally at an average yield increase of only 1.3%, implying a falling trend in yield growth rates.
The falling growth rates utilized in the reference case of the BLS may be justified for several reasons. Historical trends suggest decreasing rates of increase, and yield improvements from biotechnology have yet to be realized. Much of the large yield increases in developed countries in the 1950s and 1960s and in developing countries thereafter has been due to intensification of chemical inputs and mechanization. Apart from economic reasons and environmental concerns which suggest that maximum input levels may have been reached in many developed countries, there are likely be diminishing rates of return for further input increases. In some developing countries, especially in Africa, increase in input levels and intensification of production are likely to continue for some time, but may also ultimately level off. Furthermore, since Africa has the lowest average cereal yields of all the regional groups combined with a high population growth rate. it will contribute an increasing share of cereal production. thereby reducing average global yield increases.
Availability of arable land for expansion of crop production is based on FAO data. In the BLS standard national models, a piece-wise linear time-trend function is used to impose upper bounds (inequality constraints) on land use. In addition, this time trend function is modified with an elasticity term (usually 0.05 or less) that reacts to changes in shadow prices of land in comparison to 1980 levels. The upper limits imposed by the time trend function utilize the FAO data on potential arable land. The arable land limits are not adjusted due to climate change, even though they may be affected positively in some locations by extension of season length or drying of wet soils, or negatively by sea-level inundation or desertification.
Risk of hunger indicator
The indicator of number of people at risk from hunger used in the BLS is defined as the population with an income insufficient to either produce or procure their food requirements in developing countries (excluding China). The measure is derived from FAO estimates and methodology for developing market economies (FAO, 1984 and 1987). The FAO estimates were obtained by stipulating that calorie consumption distribution in a country is skewed and can be represented by a beta distribution. The parameters of these distributions were estimated by FAO for each country based on country-specific data and cross-country comparisons. The estimate of the energy requirement of an individual is based on the basal metabolic rate (time in a fasting state and lying at complete rest in a warm environment). Body weight, age and sex have an impact on this requirement. FAO presents two estimates of undernourished people, based on minimum maintenance requirements of 1.2 and 1.4 (the latter judged as more appropriate) basal metabolic rate. The BLS estimate for 1980, based on the 1.4 basal metabolic rate requirements, is 501 million undernourished people in the developing world, excluding China.
Limitations of world food trade model
The economic adjustments simulated by the BLS are assumed not to alter the basic structure of the production functions. These relationships may be altered in a changing climatic regime and under conditions of elevated C02. For example, yield responses to nitrogen fertilization may be altered due to changing nutrient solubilities in warmer soils. Furthermore, in the analysis of BLS results, consideration is limited to the major cereal food crops, even though shifts in the balance of arable and livestock agriculture are also likely under changed climatic regimes. Livestock production is a significant component of the global food system and is also potentially sensitive to climatic change. The nonagriculture sector is poorly modelled in the BLS, leading to simplifications in the simulation of responses to climatic change.
Finally, recent changes in global geopolitics and related changes in agricultural production are not well represented in the BLS. To account for these changes, prices in previously planned economies were made more responsive compared to earlier versions, 'plan targets' for allocation decisions were replaced, and some constraints were relaxed in the agricultural sector model. Better analysis depends on development of new models for these emerging capitalist economies.
The estimates of climate-induced changes in food production potential were used as inputs to the BLS in order to assess possible impacts on future levels of food production, food prices and the number of people al risk from hunger (see Figure 1). Impacts were assessed for the year 2060, with population growth, technology trends and economic growth projected to that year. Assessments were first made for a reference scenario assuming no climate change and subsequently with the GCM 2 x CO2 scenarios of climate change. The difference between the two assessments is the estimated climate-induced effect. A further set of assessments examined the efficacy of two levels of farmer adaptation in mitigating climate change impacts and the effect on future production of different rates of economic and population growth, and of liberalizing the world trade system.
Results are described for the following scenarios:
The reference scenario
The reference scenario projects the agricultural system to the year 2060 with no climate change and with no major changes in the political or economic context of world food trade. It assumes:
These are projections of the world system including effects on agricultural yields under the doubled CO2 scenarios for GISS, GFDL and UKMO GCMs. The food trade simulations for these three scenarios were started in 1990 and assumed a linear change in yields until the doubled CO2 changes are reached in 2060. Simulations were made both with and without the physiological effects of 555 ppm CO2 on crop growth and yield for the equilibrium yield estimates. In these scenarios, internal economic adjustments in the model occur such as increased agricultural investment, reallocation of agricultural resources according to economic returns (including crop switching), and reclamation of additional arable land as a response to higher cereal prices. These are based on shifts in supply and demand factors that alter the comparative advantage among countries and regions in the world food trade system. These economic adjustments are assumed not to feed back to the yield levels predicted by the crop modelling study.
Scenarios including the effects of farm-level adaptations
The food trade model was first run with yield changes assuming no external farm-level adaptation to climate change and was then re-run with different climate-induced changes in yield assuming the two levels of adaptation described above. Adaptation Level 1 includes those adaptations at the farm level that would not involve major changes in agricultural practices. It thus included changes in planting date, in amounts of irrigation, and in choice of crop varieties that are currently available. Adaptation Level 2 includes, in addition to the former, major changes in agricultural practices, e.g.. shifts of planting date of greater than one month, the availability of new cultivars, expansion of irrigation systems and increased fertilizer application. This level of adaptation would be likely to involve policy changes at the regional, national and international level and would also be likely to involve significant costs. However, policy, cost, and water resource availability were not studied explicitly and assumed not to be barriers to adaptation. Switching from one enterprise to another based on production and demand factors is included in the BLS.
Scenarios of different future trade, economic and population growth
A final set of scenarios assumed changes to the world tariff structure and different rates of growth of economy and population, yielding insight into alternate futures. As with the previous experiments, these were conducted both with and without climate change. These scenarios included:
Crop yields with arbitrary sensitivity tests
With the direct effects of CO2, and precipitation held at current levels, average crop yields weighted by national production show a positive response to +2deg.C warming and a negative response to +4deg.C (Figure 4). Wheat and soybean yields increase 10-15%, and maize and rice yields increase about 8% with a +2deg.C temperature rise. Yields of all four crops turn negative at +4deg.C, indicating a threshold of the compensation of direct CO2 effects for temperature increases between 2 and 4deg.C, as simulated in the IBSNAT crop models. Rice and soybean are most negatively affected at +4deg.C. These averaged results, however, mask differences among countries. For example, the effects of latitude are such that in Canada, a +2deg.C temperature increase with no precipitation change results in wheat yield increases (with direct effects of CO2 taken into account), while the same changes in Pakistan result in average wheat yield decreases of about 12%. In general, 20% increase in precipitation improved the simulated yields of the crops tested, and 20% decrease lowered yields of all crops.
Crop yields without adaptation
Table 9 shows modelled wheat yield changes for the GCM doubled CO2, climate change scenarios (the yield changes include results from both rainfed and irrigated simulations, weighted by current percentage of the respective practice). Climate changes without the direct physiological effects of CO2 cause decreases in simulated wheat yields in all cases, while the direct effects of CO2 mitigate the negative effects primarily in mid- and high-latitudes.
The magnitudes of the estimated yield changes vary by crop (Table 10). Global wheat yield changes weighted by national production are positive with the direct CO2 effects, while maize yield is most negatively affected, reflecting its greater production in low-latitude areas where simulated yield decreases are greater. Maize production declines most with direct CO2 effects, probably due to its lower response to the physiological effects of CO2 on crop growth. Simulated soybean yields are most reduced without the direct effects of CO2, but are least affected in the less severe GISS and GFDL climate change scenarios when direct CO2 effects are simulated. Soybean responds positively to increased CO2, but is the crop most affected by the high temperatures of the UKMO scenario.
The differences among countries in simulated crop yield responses to climate change without the direct effects of CO2 are primarily related to differences in current growing conditions. Higher temperatures tend to shorten the growing period at all locations tested. At low latitudes, however, crops are currently grown at higher temperatures, produce lower yields, and are nearer the limits of temperature tolerances for heat and water stress. Warming at low latitudes thus results in accelerated growing periods for crops, more severe heat and water stress, and greater yield decreases than at higher latitudes. In many mid- and high-latitude areas, where current temperature regimes are cooler, increased temperatures, while still shortening grain-filling periods, thus exerting a negative influence on yields, do not significantly increase stress levels. At some sites near the high-latitude boundaries of current agricultural production, increased temperatures can benefit crops otherwise limited by cold temperatures and short growing seasons, although the extent of soil suitable for expanded agricultural production in these regions was not studied explicitly. Potential for expansion of cultivated land is embedded in the BLS world food trade model and is reflected in shifts in production calculated by that model.
Simulated yield increases in the mid- and high-latitudes are caused primarily by:
The GISS and GFDL climate change scenarios produced yield changes ranging from +30 to -30%. Effects under the GISS scenario are, in general, more adverse than under the GFDL scenario to crop yields in parts of Asia and South America, while effects under the GFDL scenario result in more negative yields in the USA and Africa and less positive results in the former USSR. The UKMO climate change scenario, which has the greatest warming (5.25deg.C global surface air temperature increase), causes average national crop yields to decline almost everywhere (up to -50% in Pakistan).
The adaptation studies conducted by the scientists participating in the project suggest that ease of adaptation to climate change is likely to vary with crop, site, and adaptation technique (Table 11). For example, at present, many Mexican producers can only afford to use small doses of nitrogen fertilizer at planting; if more fertilizer becomes available to more farmers some of the yield reductions under the climate change scenarios might be offset. However, given the current economic and environmental constraints in countries such as Mexico, a future with unlimited water and nutrients is unlikely (Liverman et al., (1992). In contrast. switching from spring to winter wheat at the modelling sites in the former USSR produces a favorable response (Menzhulin et al., (1992), suggesting that agricultural productivity may be enhanced there with the relatively easy shift to winter wheat varieties.
Yield estimates for the two levels of adaptation developed for the BLS simulations for the UKMO scenario are shown in Figure 6. As in Figure 5, results shown are averages for countries and groups of countries, and regional variations within countries are not reflected. Direct CO2 effects on crop growth and water use are taken into account. Adaptation Level 1, simulating minor changes to existing agricultural systems, compensated for the climate change scenarios incompletely, particularly in the developing countries. For the GISS and GFDL scenarios, adaptation implying major changes to current agricultural systems (Adaptation Level 2) compensated almost fully for the negative climate change impacts. With the high level of global warming as projected by the UKMO climate change scenario, neither Level 1 nor Level 2 Adaptation fully overcomes the negative climate change effects on crop yields in most countries, even when the direct CO2 effects are taken into account.
The reference scenario (the future without climate change)
Assuming no effects of climate change on crop yields and current trends in economic and population growth rates, world cereal production5 is estimated at 3286 million metric tons (mmt) in 2060 (cf. 1795 mmt in 1990). Per capita cereal production in developed countries increases from 690 kg/cap in 1980 to 984 kg/cap in 2060. In developing countries (excluding China) cereal production increases from 179 to 282 kg/cap. Aggregated world per capita cereal production grows from 327 kg/cap in 1980 to 319 kg/cap in 2060. The declining aggregate trend for the future is caused by the relatively large difference in per capita cereal production in the developed and developing countries and the demographic changes assumed by the model.
Cereal prices are estimated at an index of 121 (1970 = 100) for the year 2060, reversing the trend of falling real cereal prices over the last 100 years (Table 12). This occurs because the BLS standard reference scenario has two phases of price development. During 1980 to 2020, while trade barriers and protection are still in place but are being reduced, there are increases in relative prices; price decreases follow when trade barriers are removed. The number of hungry people is estimated at about 640 million or about 6% of total population in 2060 (cf. 530m in 1990, about 10% of total current population).
Effects of climate change with and without adjustments in the economic system
The BLS includes the ability to simulate adjustments that the world food system might make to changes of yield (e.g.. reallocation of agricultural land use, change in fertilizer use, and application of irrigation water). Simulations of the effects of climate change without such internal adjustments are of theoretical interest only as these would unrealistically imply no economic or behavioural response of producers and consumers. However, as a measure of distortion of the economic system these hypothetical impacts help to define the adjustments taking place in the system over time. Under these conditions the effects of climate change and increased atmospheric CO2 on crop yields derived from the GCM scenarios imply a 5% to almost 20% reduction in total cereal production (Table 13). These estimates are changes to production levels projected for 2060 without climate change.
Adjustments within the economic system tend to counteract negative yield impacts as agricultural production shifts to regions of more favourable comparative advantage. The BLS offsets 65-80% of the potential impact on yield in scenarios for impacts below 10% of global cereal production (the GISS and GFDL climate change scenarios). The offset decreases to 60% under a scenario of greater yield reduction (e.g.., UKMO).
Effects of climate change with adjustment in the economic system, but without farm-level adaptation
Changes in cereal production, cereal prices, and people at risk of hunger estimated for the GCM doubled CO2 climate change scenarios (with the direct CO2 effects taken into account) are given in Table 14. These estimations are based upon dynamic simulations by the BLS that allow the world food system to respond to climate-induced supply shortfalls of cereals and consequently higher commodity prices through increases in production factors (cultivated land, labour, and capital) and inputs such as fertilizer. The testing of climate change impacts without farm-level adaptation is unrealistic, but is done for the purpose of establishing a baseline with which to compare the effects of farmer response. We can safely assume that at least some farm-level adaptations will be adopted, especially techniques similar to those tested in Adaptation Level 1 that do not imply major changes to current agricultural systems.
Under the GISS scenario (which provides lower temperature increases) cereal production is estimated to decrease by just over 1%, while under the UKMO scenario (with the highest temperature increases) global production is estimated to decrease by more than 7%. The largest negative changes occur in developing regions, which average -9% to -11%, though the extent of decreased production varies greatly by country depending on the projected climate. By contrast, in developed countries production is estimated to increase under all but the UKMO scenario (+11% to -3%). Thus, disparities in crop production between developed and developing countries are estimated to increase.
Price increases resulting from climate-induced decreases in yield are estimated to range between 25% and 150%. In the case of the GISS scenario, the 5.3% reduction in yields of the unadjusted scenario causes a disequilibrium that is resolved via market mechanisms in the adjusted case. This results in a -1.2% consumer response and about a +4% (relative) producer response and leads to 24% higher relative prices for cereals. While this price response seems to be high, cereal prices only account for a modest fraction, perhaps one third or less, of retail food prices. Hence, a 24% increase in world cereal prices does not imply a 24% increase in food prices.
These increases in price are likely to affect the number of people with insufficient resources to purchase adequate amounts of food. The estimated number of hungry people increases approximately 1% for each 2-2.5% increase in prices (depending on climate change scenario). People at risk of hunger increase by 10% to almost 60% in the climate change scenarios tested, resulting in an estimated increase of between 60 and 350 million people in this condition (above the reference case of 640 million) by 2060.
Effects of climate change under different levels of farmer adaptation
Globally, both minor and major levels of adaptation help restore world production levels, compared to the climate change scenarios with no adaptation (Figure 7). Averaged global cereal production decreases by up to about 160 mmt (0% to -5%) from the reference case of 3286 mmt with Level 1 adaptations. These involve shifts in farm activities that are not very disruptive to regional agricultural systems. With adaptations implying major changes, global cereal production responses range from an additional 30 mmt to slight increase to a slight decrease of about 80 mmt (+1% to -2.5%).
Level 1 adaptation largely offsets the negative climate change yield effects in developed countries, improving their comparative advantage in world markets (Figure 8). In these regions cereal production increases by 4% to 14% over the reference case. However, developing countries are estimated to benefit little from this level of adaptation (-9% to -12% change in cereal production). More extensive adaptation (Level 2) virtually eliminates global negative cereal yield impacts derived under the GISS and GFDL climate scenarios and reduces impacts under the UKMO scenario to one third.
Figure 9 shows the effects of climate change, and climate change with both levels of adaptation, on cereal prices in 2060. As a consequence of climate change, world cereal prices are estimated to increase by about 25% to almost 150%. Under Adaptation Level 1, price increases range from 10% to 100%; under Adaptation Level 2, cereal price responses range from a decline of about 5% to an increase of 35%.
As a consequence of climate change and Adaptation Level 1, the number of people at risk from hunger increases by about 40 million to 300 million (6% to 50%) from the reference case of 641 million (Figure 10). With more significant farmer adaptation (Level 2), the number of people at risk from hunger is altered by between -12 million for the GISS scenario and 120 million for the UKMO scenario (-2% and +20%). These results indicate that, except for the GISS scenario under Adaptation Level 2, the simulated farm-level adaptations did not entirely mitigate the negative effects of climate change on potential risk of hunger, even when economic adjustments, i.e.. the production and price responses of the world food system, are taken into account.
Effects of climate change assuming full trade liberalization, lower economic growth rates, and population growth rates
For each of these alternate future assumptions, a new reference scenario was established with the BLS, and then tested with the GCM climate change scenarios.
Full trade liberalization
Assuming full agricultural trade liberalization and no climate change by 2020 provides for more efficient resource use. This leads to a 3.2% higher value added in agriculture globally and 5.2% higher agricultural GDP in developing countries (excluding China) by 2060 compared to the original reference scenario. This policy change results in almost 20% fewer people at risk from hunger. Global cereal production increases by 70 mmt, with most of the production increases occurring in developing countries (Table 15).
Climate change impacts were then simulated under these new reference conditions. Under the same trade liberalization policies, global impacts due to climate change are slightly reduced, with enhanced gains in production accruing to developed countries. Losses in production are greater in developing countries. Price increases are reduced slightly from what would occur without full trade liberalization, and the number of people at risk from hunger is reduced by about 100 million.
Reduced rate of economic growth
Estimates were also made of impacts under a lower economic growth scenario (10% lower than reference). These are indicated in Table 16. Lower economic growth results in a tighter supply situation, higher prices, and more people below the hunger threshold.
The effect of climate change on these trends is generally to reduce production, increase prices, and increase the number of people at risk from hunger. Developed countries increase cereal production in the GISS and GFDL scenarios even with the projected lower economic growth rates, but developing countries decrease production under all climate change scenarios.
Altered rates of population growth
Lower population growth has a significant effect on cereal production, food prices, and number of hungry people (Table 17). Simulations based on rates of population growth according to UN Low Estimates result in a world population about 17% lower in year 2060 when compared to UN Mid Estimates as used in the reference run. The corresponding reduction in the developing countries (excluding China) would be about 19.5%, from 7.3 billion to 5.9 billion. The combination of higher GDP/capita (about 10%) and lower world population produces an estimated 40% fewer people in hunger in the year 2060 compared to the reference scenario.
Even under the most adverse of the three climate scenarios (UKMO) the estimated number of hungry is some 10% lower than the estimated reference scenario without any climate change. Increases in world prices in agricultural products, in particular of cereals, under the climate change scenarios employing the low population projection are around 75% of those using the UN Mid Estimate.
Figure 11 summarizes the generalized relative effects of different policies regarding trade liberalization, economic growth and population growth on the production of cereals and people at risk of hunger. Alternative development assumptions make little difference with respect to the geopolitical patterns of the relative effects of climate change. In all cases, cereal production decreases, particularly in the developing world, while prices and population at risk from hunger increase due to climate change. The beneficial effects of trade liberalization and low population growth are of the same or even greater (in the case of population) order of magnitude as the adverse effects of climate change. This suggests that there may be much to be gained from altering the conditions of trade and development as a strategy for addressing the climate change issue. The magnitude of adverse climate impacts are least, however, under the conditions of low population growth. An assumption of low population growth rate minimizes the population at risk from hunger both in the presence and absence of climate change in the BLS simulations.
Climate change induced by increasing greenhouse gases is likely to affect crop yields differently from region to region across the globe. Under the climate change scenarios adopted in this study, the effects on crop yields in mid- and high-latitude regions appear to be less adverse than those in low-latitude regions. However, the more favourable effects on yield in temperate regions depend to a large extent on full realization of the potentially beneficial direct effects of CO2 on crop growth. Decreases in potential crop yields are likely to be caused by shortening of the crop growing period, decrease in water availability due to higher rates of evapotranspiration, and poor vernalization of temperate cereal crops. When adaptations at the farm level were tested (e.g.. change in planting date, switch of crop variety, changes in fertilizer application and irrigation), compensation for the detrimental effects of climate change was more successful in developed countries.
When the economic implications of these changes in crop yields are explored in a world food trade model, the relative ability of the world food system to absorb impacts decreases with the magnitude of the impact. Regional differences in effects remain noticeable: developed countries are expected to be less affected by climate change than developing economies. Dynamic economic adjustments can compensate for lower-impact scenarios such as the GISS and GFDL climate scenarios but not higher-impact ones such as the UKMO scenario. Prices of agricultural products are related to the magnitude of the climate change impact, and incidence of food poverty increases even in the least negative climate change scenario tested.
When the effects of lower future population and economic growth rates and trade liberalization were tested in the food trade model, reduced population growth rates would have the largest effect on minimizing the impact of climate change. Lower economic growth results in tighter food supplies, and consequently would result in higher rates of food poverty. Full trade liberalization in agriculture, on the other hand, provides for more efficient resource use and would reduce the number of people at risk from hunger by about 100 million (from the reference case of c. 640 million in 2060). However, all of the scenarios of future climate adopted in this study increase the estimates of the number of people at risk from hunger.
It should be emphasized that the results reported here are not a forecast of the future. There are very large uncertainties that preclude this: particularly the lack of information on possible climate change at the regional level, the effects of technological change on agricultural productivity, trends in demand (including population growth), and the wide array of possible adaptations. The adoption of efficient adaptation techniques is far from certain. In developing countries there may be social or technical constraints, and adaptive measures may not necessarily result in sustainable production over long timeframes. The availability of water supplies for irrigation and the costs of adaptation are both critical needs for further research.
Future trace gas emission rates, as well as when the full magnitude of their effects will be realized, are not certain, and only a limited range of GCM climate change scenarios, representing the upper end of the projected warming, was tested. However, it can be argued that the use of scenarios from the higher GCM projections provides perspective on the downside risk of global warming projections. Because of these uncertainties, the study should be considered as an exploratory assessment of the sensitivity of the world food system to a limited number of what is, in effect, a much wider array of possible futures.
Determining how countries, particularly developing countries can and will respond to reduced yields and increased costs of food is a critical research need arising from this study. Will such countries he able to import large amounts of food? From a political and social standpoint, these results show a decrease in food security in developing countries. The study suggests that the worst situation arises from a scenario of severe climate change, low economic growth, and little farm-level adaptation. In order to minimize possible adverse consequences - production losses, food price increases, and people at risk of hunger - the way forward is to encourage the agricultural sector to continue to develop crop breeding and management programmes for heat and drought conditions (these will be immediately useful in improving productivity in marginal environments today), in combination with measures taken to slow the growth of the human population of the world. The latter step would also be consistent with efforts to slow emissions of greenhouse gases. the source of the problem, and thus the rate and eventual magnitude of global climate change.
Acock, B. and Allen, L. H., Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In: Strain, B. R. and Cure, J. D. eds., Direct Effects of Increasing Carbon Dioxide on Vegetation. Washington, DC. US Department of Energy. DOE/ER-0238: 33-97.
Adams, R. M., Rosenzweig, C., Peart, R. M., Ritchie, J. T., McCarl, B. A., Glyer, J. D., Curry, R. B., Jones, J. W., Boote, K. J. and Allen, L. H., Jr. 1990. Global climate change and US agriculture. Nature 345(6272): 219-22.
Allen, L. H., Jr., Boote, K. J., Jones, J. W., Jones, P. H., Valle, R. R., Acock, B., Rogers, H. H. and Dahlman, R. C. 1987. Response of vegetation to rising carbon dioxide: Photosynthesis, biomass and seed yield of soybean. Global Biogeochemical Cycles 1: 1-14.
Cure, J. D. 1985. Carbon dioxide doubling responses: A crop survey. In: Strain, B. R. and Cure, J. D. eds., Direct Effects of Increasing Carbon Dioxide on Vegetation. Washington, DC. US Department of Energy. DOE/ER-0238, 33-97.
Cure, J. D. and Acock, B. 1986. Crop responses to carbon dioxide doubling: A literature survey. Agricultural and Forest Meteorology 38: 127-145.
Fischer, G., Frohberg, K., Keyzer, M. A. and Parikh, K. S. 1988. Linked National Models: A Tool for International Food Policy Analysis. Dordrecht, Netherlands: Kluwer.
Fischer, G., Frohberg, K., Keyzer, M. A., Parikh, K. S. and Tims, W. 1990. Hunger--Beyond the Reach of the Invisible Hand. Laxenburg: International Institute for Applied Systems Analysis. Food and Agriculture Project.
FAO, 1984. Fourth World Food Survey. Rome: United Nations Food and Agriculture Organization.
FAO, 1987. Fifth World Food Survey. Rome: United Nations Food and Agriculture Organization.
FAO. 1988. 1987 Production Yearbook. Rome: United Nations Food and Agriculture Organization. Statistics Series No. 82.
FAO. 1991. AGROSTAT/PC. Rome: United Nations Food and Agriculture Organization.
Godwin, D., Ritchie, J. T., Singh, U. and Hunt, L. 1989. A User's Guide to CERES-Wheat- V2.10. Muscle Shoals: International Fertilizer Development Center.
Godwin, D., Singh, U., Ritchie, J. T. and Alocilja, E. C. 1993. A User `s Guide to CERES-Rice. Muscle Shoals: International Fertilizer Development Center. (in press).
Hansen, J., Russell, G., Rind, D., Stone, P., Lacis, A., Lebedeff, S., Ruedy, R. and Travis, L. 1983. Efficient three-dimensional global models for climate studies Models I and II. Monthly Weather Review 111(4): 609-662.
Hansen, J., Fung, I.. Lacis, A., Rind, D., 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): 9341-9364.
Hendry, G. R. 1993. FACE: Free-Air CO2 Enrichment for Plant Research in the Field. Boca Raton: Smoley, C. K. CRC Press.
IPCC, 1990a. Houghton, J. T., Jenkins, G. J. and Ephraums, J. J. eds. Climate Change: The IPCC Scientific Assessment. International Panel on Climate Change. Cambridge: Cambridge University Press.
IPCC, 1990b. Tegart, W. J. McG., Sheldon, G. W. and Griffiths, D. C. eds. Climate Change: The IPCC Impacts Assessment. Canberra: Australian Government Publishing Service.
IPCC, 1992. Houghton, J. T., Callander, B. A., and Varney, S. K. eds. Climate Change 1992. The Supplementary Report to the IPCC Scientific Assessment. Cambridge: Cambridge University Press.
International Bank for Reconstruction and Development/World Bank. 1990. World Population Projections. Baltimore: Johns Hopkins University Press.
International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT). 1989. Decision Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2. l). Honolulu: Dept. of Agronomy and Soil Science. College of Tropical Agriculture and Human Resources: University of Hawaii.
Jones, C. A. and Kiniry, J. R. 1986. CERES-Maize. A Simulation Model of Maize Growth and Development. College Station. Texas A&M Press.
Jones, J. W., Boote, K. J., Hoogenboom, G., Jagtap, S. S. and Wilkerson, G. G. 1989. SOYGRO V5.42: Soybean Crop Growth Simulation Model. Users' Guide. Gainesville: Department of Agricultural Engineering and Department of Agronomy, University of Florida.
Kane, S., Reilly, J. and Tobey. J. 1991. Climate Change: Economic Implications for World Agriculture. US Department of Agriculture. Economic Research Service. AER-No. 647.
Kellogg, W. W. and Zhao, Z.-C. 1988. Sensitivity of soil moisture to doubling of carbon dioxide in climate model experiments. Part 1: North America. Journal of Climate 1: 348-366.
Kimball, B. A. 1983. Carbon dioxide and agricultural yield. An assemblage and analysis of 430 prior observations. Agronomy Journal 75: 779-788.
Magalhaes, A. R. 1992. Impacts of Climatic Variations and Sustainable Development in Semi-arid Regions. Proceedings of International Conference. ICID. Fortaleza, Brazil.
Manabe, S. and Wetherald, R. T. 1987. Large-scale changes in soil wetness induced by an increase in CO2. Journal of Atmospheric Science, 44: 1211-1235.
Otter-Nacke, S., Godwin, D. C. and Ritchie, J. T. 1986. Testing and validating the CERES-Wheat model in diverse environments. AgGRISTARS YM-15-00407. Houston: Johnson Space Center No. 20244.
Parry, M. L., Carter, T. R. and Konijn, N. T. eds. 1988a. The impact of climatic variations on agriculture. Vol. 1 Assessments in cool temperate and cold regions. Dordrecht, Netherlands: Kluwer.
Parry, M. L., Carter, T. R., and Konijn, N. T. eds. 1988b. The Impact of Climatic Variations on Agriculture, Volume 2, Assessments in Semi-Arid Areas. Dordrecht, Netherlands: Kluwer.
Parry, M. L. 1990. Climate Change and World Agriculture. London: Earthscan.
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: UN Environmental Programme.
Peart, R. M., Jones, J. W., Curry, R. B., Boote, K. and Allen, L. H ., Jr. 1989. Impact of climate Change on Crop Yield in the Southeastern USA. In: Smith, J. B. and Tirpak, D. A. eds. The Potential Effects of Global Climate Change on the United States. Washington, DC: US Environmental Protection Agency.
Ritchie, J. T. and Otter, S. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield model. In: Willis, W. O., ed. ARS Wheat Yield Project. Washington, DC. Department of Agriculture, Agricultural Research Service. ARS-38.
Ritchie, J. T., Singh, U., Godwin, D. and Hunt, L. 1989. A User's Guide to CERES-Maize V2.10. Muscle Shoals: International Fertilizer Development Center.
Rogers, H. H., Bingham, G. E., Cure, J. D., Smith, J. M. and Surano, K. A. 1983. Responses of selected plant species to elevated carbon dioxide in the field. Journal of Environmental Quality. 12: 569-574.
Rosenberg, N. J. and Crosson, P. R. 1991. Processes for Identifying Regional Influences of and Responses to Increasing Atmospheric CO2 and Climate Change: the MINK Project. An Overview. Washington, DC: Resources for the Future. Department of Energy. DOE/RL/01830T-H5.
Rosenzweig, C. and Iglesias, A. eds. 1993. Implications of Climate Change for International Agriculture: Crop Modeling Study. Washington, DC: US Environmental Protection Agency. (in press).
Smit, B. 1989. Climatic warming and Canada's comparative position in agricultural production and trade. In Climate Change Digest. CCD 89-01. Environment Canada. pp. 1-9.
Smith, J. B. and Tirpak, D. A. eds. 1989. The Potential Effects of Global Climate Change on the United States. Report to Congress. Washington, DC: US Environmental Protection Agency. EPA-230-05-89-050.
Strzepek, K. M., Onyeji, S. C. and Saleh, M. 1993. A SocioEconomic Analysis of Integrated Climate Change Impacts on Egypt. CADSWES Working Paper. Boulder: University of Colorado.
Thompson, L. M. 1969. Weather and technology in the production of corn in the US corn belt. Agronomy Journal 61: 453-456.
Tobey, J., Kane, S. and Reilly, J. 1993. An empirical study of the economic effects of climate change on world agriculture. Climatic Change. (in press).
United Nations. 1989. World Population Prospects 1988. New York: United Nations.
Wilson, C. A. and Mitchell, J. F. B. 1987. A doubled CO2 climate sensitivity experiment with a global climate model including a simple ocean. Journal of Geophysical Research, 92, (13): 315-343.
World Food Institute. 1988. World Food Trade and U.S. Agriculture, 1960-1987. Ames: Iowa State University.