JOHN REILLY
Natural Resources and Environment Division, Economic Research Service, USDA,
Washington DC, USA
Impact assessment methodologies
Crops response estimates for different regions of the world
Global studies and their implications for regional effects
Regional vulnerability
Adaptation potential and policies
The technological potential to adapt
The socio-economic capability to adapt
References
The potential impacts of climate change on agriculture are highly uncertain. The large number of studies conducted over the past few years for many different sites across the world show few, if any, robust conclusions of either the magnitude or direction of impact for individual countries or regions. Where apparent consensus exists it frequently appears to occur because only one or two studies have been conducted using a single climate scenario. Many such studies have focused on doubled (2xCO2,) General Circulation Model (GCM) equilibrium scenarios. These do not begin to describe the variety of climatic conditions any particular region is likely to experience as the actual climate changes over time.
Potential future climate changes are also made more uncertain because of the recently recognized role of sulphate aerosols which may partly offset the warming expected from increased concentrations of CO2, methane, nitrous oxide and other radiatively active trace gases. The significant spatial variation in sulphate aerosol concentrations means that the regional pattern of climate change may be quite different from that simulated on the basis of CO2 increase alone. The short lifetime of aerosols in the atmosphere (a few days) means that if the use of high sulphur coal in India or China increases or efforts to control sulphur emission in the United States or Europe are intensified, the spatial pattern of climate change could change significantly within a relatively short period of time due to changes in the aerosol cooling effect.
Different impact methodologies also yield widely varying results of the direct impacts of climate change on crop yields and agricultural production even when examining the same region and the same climate scenarios. The socio-economic environment, agricultural technology and natural resource base will also necessarily undergo profound changes over the next 100 years whether agriculture meets the many challenges of feeding the world's growing population or fails to do so.
The robust conclusion that does emerge from impact studies is that climate change has the potential to change significantly the productivity of agriculture at most locations. Some currently highly productive areas may become much less productive. Some currently marginal areas may benefit substantially while others may become unproductive. Crop yield studies show regional variations of +20, 30 or more per cent in some areas and equal size losses in other areas. Most areas can expect change and will need to adapt, but the direction of change, particularly of precipitation, and required adaptations cannot now be predicted. Nor may it ever be possible to predict them with confidence. Current evidence suggests that poleward regions where agriculture is limited by short growing seasons are more likely to gain while subtropical and tropical regions may be more likely to suffer drought and losses in productivity. However, these broad conclusions hardly provide the basis for mapping out a long-term strategy for agricultural adaptation. Thus, policy must retain flexibility to respond as conditions change.
A further issue is how do climate change impacts on agricultural production fit within the other pressing challenges facing agriculture in different regions of the world. Is climate change a minor threat, likely to be undetected among the many changes that will reshape the agricultural sectors of the world's economies? Or is it another critical challenge to an agricultural sector straining to cope with a growing population, resource degradation, tighter constraints on available resources, and exhaustion of technological capabilities to expand production using existing land and water resources?
It is useful to place some of the 2xCO2 agricultural projections in the context of other future projections. If we accept long-term demographic trends, the largest absolute addition to the world's population will occur during the decade of the 1990s, the growth rate having already slowed from that of the 1950s and 1960s. By the time 2xCO2 climate scenarios are expected to be realized (some time around 2100 or later), the world population will have stabilized and agricultural research will no longer face the challenge of increasing productivity to keep up with a growing population.
Therefore there is a need for more specific analysis about how climate will change over the next 10, 20 or 30 years rather than over the next 100. It also provides a caution not to consider our response to climate change apart from our response to the immediate needs of agriculture: feeding a growing population where presently an estimated 740 million people still suffer from hunger and malnutrition while maintaining the productivity of basic agricultural resources and meeting the demands placed on agriculture to minimize damage to the environment.
This paper will: (1) briefly discuss the major methodologies used to estimate impacts of climate change as different models lead to substantially different estimates of climate change impacts; (2) review the broad literature reporting results of crop yield studies of climate change conducted for many different areas (how much (or little) do we know?); (3) review the set of estimates that has been made for global agricultural production and what it means for regional agricultural impacts; (4) discuss the issue of vulnerability, adding a precise definition, while reviewing some of the vulnerability concepts that have been used in the literature; and (5) review specific issues of adaption - how can the world's agricultural system, or more to the point, those populations highly dependent on agriculture, make themselves less likely to suffer loss from climate change.
Climate change presents a challenge for researchers attempting to quantify its impact due to the global scale of likely impacts, the diversity of agriculture systems, and the decades' long time scale. Current climatic, soil and socio-economic conditions vary widely across the world. Each crop and crop variety has specific climatic tolerances and optima. It is not possible to model world agriculture in a way that captures the details of plant response in every location. The availability of data with the necessary geographic detail is currently the major limitation rather than computational capability or basic understanding of crop responses to climate. A specific problem has been how to take the detailed knowledge of plant response into aggregate assessments of regional assessments. In general, compromises are necessary in developing quantitative analyses at regional scales.
There are two basic approaches for evaluating crop and farmer response to changing climate: (1) structural modelling of the agronomic response of plants and the economic/management decisions of farmers based on theoretical specifications and controlled experimental evidence; and (2) reliance on the observed response of crops and farmers to varying climate.
For the first approach, sufficient structure and detail are needed to represent specific crops and crop varieties for which responses to different conditions are known through detailed experiments. Similar detail on farm management allows direct modelling of the timing of field operations, crop choices, and how these decisions affect costs and revenues. These approaches typically model a representative crop or farm. Both in the case of economic models of farm decisions and in the case of crop response models, the original purpose was to improve understanding of how the crop grows or how a farmer manages. In the case of models of a representative farm, one might hope to offer prescriptive advice for the farmer - where farm operations differ from the profit maximizing (or cost minimizing) model results, it provides guidance for how farmers might improve farm performance. In both cases, the idealized representation of the crop and farm operation tends to give results that differ markedly from the actual experience on farms operating under real world conditions. This may reflect the fact that farmers do not operate as profit maximizers (they could improve their performance) or that the models fail to consider some of the factors that the farmer takes into account such as risk or lack of immediate employment alternatives. Because of the idealized nature of these models, many analysts consider that these provide evidence of the potential production or potential profitability. Imposing climate change on these models gives estimates of how potential production may change due to climate change. Using these results as indicative of how climate will actually affect agriculture thus rests on the assumption that the change in the potential represents the change likely to be actually experienced. Many approaches of this type have used detailed crop response models requiring daily weather records. For aggregate analyses inferences must be made from relatively few sites and crops to large areas and diverse production systems because of the complexity of the models and the need for detailed data on weather over a decade or more. This is the basic approach of Fischer et al. reported elsewhere in this volume.
The work of Leemans and Solomon (1993) is in a similar vein, choosing much simpler representations of crop/climate interactions but is still related to basic agronomic representation of crop growth in response to temperature and precipitation. The advantage of their approach is that, because of the minimal amounts of climatic data required (mean monthly data on temperature and precipitation), the crop models can be applied at a resolution of 0.5° x 0.5° latitude-longitude grids.
The second approach, relying on observed response of crops and farmers, provided some of the earliest estimates of the potential effects. The simplest example of this approach is to observe the current climatic boundaries of crops and to redraw these boundaries for a predicted changed climate (e.g., Rosenzweig, 1985). In a similar vein, researchers have applied statistical analysis of data across geographic areas to separate climate from other factors (e.g., different soil quality, varying economic conditions) that explain production differences across regions and have used these to estimate the potential agricultural impacts of climate change (e.g., Mendelsohn et al., 1994). An advantage of using direct evidence from observed production is that the data reflect how farmers operating under commercial conditions and crops growing under such conditions actually respond to geographically varying climatic conditions. Here, the most recent work uses extremely reduced form models (e.g., Mendelsohn et al., 1994) although estimation of more detailed structural models is possible. Darwin et al. (1995) use revealed evidence from geographic variation in climate in a global model, allocating production and input use to climatically determined land classes based on current production patterns. Climate change impacts are then simulated by altering the distribution of land classes and assuming that when an area's land class changes, its underlying production level changes to that of the new land class.
The Darwin et al. (1995) approach links the basic agricultural productivity of land classes, described by a production function, with a computable general equilibrium model of the world economy. Thus, actual production in a region or land class depends on the final market clearing prices. The model also treats interactions with other sectors of the economy, most importantly sectors that compete for land and water. My interest in this section is in contrasting approaches used to estimate the initial impact of climate on agricultural production. As demonstrated by Fischer et al. (1994), Reilly et al. (1994) and Adams et al. (1988), given an initial climate shock on productivity, there are a number of ways to introduce this shock into a variety of different types of economic models to generate estimates of the market impact and realize production under new equilibrium prices.
The advantage of these approaches is that the response of crops and farmers is based on actual response under current operating conditions rather than an idealized view of how crops and farmers respond. The basic caveat associated with this approach is that one must have faith that land currently producing one set of outputs can change to the new set of outputs once climate changes. Whether these types of approaches accurately capture the productivity impact depends on how well they control for other factors (such as soil quality) and whether farmers can adjust their production as climate changes. This latter consideration leads to the interpretation that these approaches capture the long-run equilibrium response to climate change and may not capture adjustment costs associated with changing to new crops and production practices.
Table 10.1 summarizes the results of the large number of studies of the impact of climate change on potential crop production. While the table does not provide the detail on the range of specific studies, methods and climate scenarios evaluated, it provides an indication of the wide range of estimates. The general conclusion of global studies, that tropical areas may more likely suffer negative consequences, is partly supported by the results in the table. For example, Latin America and Africa show primarily negative impacts. However, very few studies have been conducted in these regions. For Europe, the United States and Canada, and for Asia (including China) and the Pacific Rim, where many more studies have been conducted, the results generally range from severe negative effects (-60, -70%, or complete crop failure) to equally large potential yield increases.
The wide ranges of estimates are due to several, as yet unresolved, factors. Differences among climate scenarios are important and these can generate wide ranges of impacts even when using identical methods for the same regions. For example, a study of the potential impact on rice yields conducted for most of the countries of South and Southeast Asia and for China, Japan, and Korea using the same crop model found yield changes for India to range from -3 to +28%, for Malaysia from +2 to +27%, for the Philippines from -14 to +14%, and for mainland China from -18 to -4% (Matthews et al., 1994a, b) depending on whether the GISS, GFDL or UKMO climate scenario was used.
The impacts across sites can vary widely within a region. Thus, how many and which sites are chosen to represent a region and how the site-specific estimates are aggregated can have important effects on the results. Studies for the United States and Canada demonstrate the wide range of impacts across sites with total or near-total crop failure every year projected for wheat and soybeans at one site in the United States (Rosenzweig et al., 1994) but wheat yield increases of 180 to 230% for other sites in the United States and Canada (Rosenzweig et al., 1994; Brklacich et al., 1994; Brklacich and Smit, 1992).
Whether and how changes in a crop variety are specified in a study can have a large impact. Studies conducted of wheat response in Australia found impacts ranging from -34 to +65% for the same climate scenario and site depending on which known and currently grown wheat cultivar was specified in the crop model (Wang et al., 1992). Similarly, Matthews et al. (1994a,b) concluded that the severe yield losses in South, Southeast and East Asia for rice in many scenarios was due to a threshold temperature effect that caused spikelet sterility but that genetic variation with regard to the threshold likely provided significant opportunity to switch varieties as temperatures rose. Thus, an impact analysis that narrowly specifies a crop variety is likely to generate a much different estimated impact than an analysis that specifies responses on the basis of the genetic variation across existing cultivars. Some studies have attempted to evaluate how future crop breeding may change the range of genetic variability available in future varieties (Easterling et al., 1993).
Table 10.1. Regional crop yield for 2xCO2, GCM equilibrium climates
Region |
Crop |
Yield impact (%) |
Countries studied/comments |
Latin America |
maize |
-61 to increase |
Argentina, Brazil, Chile, Mexico. Range is across GCM scenarios, with and without the CO2 effect. |
wheat |
-50 to -5 |
Argentina, Uruguay, Brazil. Range is across GCM scenarios, with and without the CO2 effect. |
|
soybean |
-10 to +40 |
Brazil. Range is across GCM scenarios, with CO2 effect. |
|
Former Soviet Union |
wheat grain |
-19 to +41 -14 to +13 |
Range is across GCM scenarios and region, with CO2 effect. |
Europe |
maize |
-30 to increase |
France, Spain, N Europe. With adaptation, CO2 effect. Longer growing season; irrigation efficiency loss; northward shift. |
wheat |
increase or decrease |
France, UK, N Europe. With adaptation, CO2 effect. Longer season: northward shift, greater pest damage; |
|
vegetables |
increase |
lower risk of crop failure. |
|
North America |
maize |
-55 to +62 |
USA and Canada. Range across GCM scenarios and |
wheat |
-100 to +234 |
sites with/without CO2 effect. |
|
soybean |
-96 to +58 |
USA. Less severe or increase in yield when CO2 effect and adaptation considered. |
|
Africa |
maize |
-65 to +6 |
Egypt, Kenya, South Africa, Zimbabwe. With CO2 effect, range across sites and climate scenarios. |
millet |
-79 to -63 |
Senegal. Carrying capacity fell 11-38%. |
|
biomass |
decrease |
South Africa; agrozone shifts. |
|
South Asia |
rice |
-22 to +28 |
Bangladesh, India, Philippines, Thailand, Indonesia, Malaysia, Myanmar. Range over GCM scenarios, and sites; with CO2 effect; some studies also consider adaptation. |
maize |
-65 to-10 |
||
wheat |
-61 to +67 |
||
Mainland China and Taiwan |
rice |
-78 to +28 |
Includes rainfed and irrigated rice. Positive effects in NE and NW China, negative in most of the country. Genetic variation provides scope for adaptation. |
Asia (other) and Pacific Rim |
rice |
-45 to +30 |
Japan and South Korea. Range is across GCM scenarios. Generally positive in northern Japan; negative in south. |
pasture |
-1 to +35 |
Australia and New Zealand. Regional variation. |
|
wheat |
-41 to +65 |
Australia and Japan. Wide variation, depending on cultivar. |
Source: Reilly et al. (1996).
Finally, the estimated amount of adaptation likely to be undertaken by farmers varies. Fundamental views about how the farm sector responds to changing conditions (of any kind) shape the choice of methodological approach, and these methodological approaches can give apparently widely different estimates of impact. Specification of the crop variety in a crop response model illustrates this difference. For some analysts, the prospect that farmers will not change the variety of crop grown over the next 100 years as climate, technology, prices and other factors change is so remote that they choose to represent change among varieties as an essentially autonomous response of the farm sector. Other analysts choose more specific crop variety characteristics, viewing even crop variety change as neither automatic nor without cost. For example, different varieties of wheat produce flours with different characteristics and the cultural practices for growing spring and winter wheat differ. Similarly, studies of impacts on Japanese rice production estimate negative impacts for the southern parts of the country because of the climate tolerances of Japonica rice which is preferred over Indica varieties in Japan (Seino, 1993).
The differences from simply whether or not one assumes farmers will adopt the better adapted variety are large but these differences are potentially magnified many fold because the series of potential adaptations are broad with some requiring more specific recognition, action and investment by farmers. How do farmers choose a planting date - by planting at the same time each year regardless of weather conditions or by planting when soil temperatures are sufficient for crop growth, when the rainy season starts, or when the fields can be tilled? If the decision is partly keyed to weather conditions then the farm decision-making process will lead to some amount of autonomous adjustment to climate change. Similarly, will the changes in tillage and irrigation practices, crop rotation schemes, crops, and crop processing and harvesting that are likely to occur over the next 100 years due to many factors also reflect changes in climate that are occurring simultaneously, or will farmers be unable to detect climate change and therefore fail to adapt these systems, becoming and remaining ill-adapted to the climate conditions occurring locally? If they adapt to current conditions (but cannot confidently look ahead) how maladjusted will their long-lived investments be after 3, 5, 10 or 20 years of continuous changes in climate?
Table 10.2 provides the range of estimates for the United States that have been generated based on quite different methodologies and assumptions about the extent to which adaptation will occur. While the table covers only the United States, it is likely that applying this range of approaches in other regions would also generate a similar range of estimates. The Mendelsohn et al. (1994) estimates (columns 1 and 2) are based on an econometric model estimated on cross-section data and reflect, according to the authors, long-run, full adjustment of USA agriculture to a climate change shock. The methodology does not allow consideration of how crop prices may change and thus may be most comparable to the initial crop yield shock used in other methodologies. Except for column 8, none of the reported estimates consider the direct effect of CO2 on plant growth. Unfortunately, the wide ranging methodologies do not or have not generally reported results that are directly comparable, thus some interpretation is necessary.
Table 10.2. Estimates of the impact of climate change on United States agriculture, percentage change
Climate scenario |
Mendelsohn et al. (1994); without CO2 effect |
Darwin et al. (1995); without CO2 effect |
Rosenzweig and Parry (1994); USA average yield effects across crops |
|||||
Effects on farm income |
Effects on cereal prod. |
|||||||
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
|
Area wgts. |
Rev, wgts. |
Farm-level adaptation |
Full adjustment |
No adjustment |
Full adjustment |
No adjustment |
Adjustment and CO2 |
|
GISS |
-1.8 |
+2.0 |
+4.1 |
-7.8 |
-24.4 |
-3.0 |
-14 to-21 |
0 to +17 |
GFDL |
-1.2 |
+4.2 |
-16.1 |
+4.3 |
-38.0 |
-2.0 |
-23 to -29 |
+9 to -10 |
UKMO |
-4.5 |
+1.1 |
-4.4 |
-5.4 |
-38.4 |
-5.0 |
-25 to -58 |
+1 to -20 |
OSU |
-3.6 |
-0.7 |
-10.0 |
+5.8 |
-33.3 |
-5.2 |
|
|
Notes: Mendelsohn et al. are armualized impact on land values as a percentage of total value of crop and livestock production. The values of crop and livestock production are for 1990 from Darwin et al. (1995). For a description of the Mendelsohn et al. methodology see Mendelsohn et al. (1994). The simulations of the model for the GISS, GFDL, UKMO and OSU reported above were provided by personal communication with Mendelsohn, 29 March 1995. Darwin et al. results are computed from simulations reported in Darwin et al. (1995). Rosenzweig and Parry summarize the range of crop yield impacts used in their 1994 study for the United States. The US average crop yield shocks estimated by them were reported in Reilly et al. (1993). Specific crop yield studies which were, in part, the basis for these estimates were reported in Rosenzweig et al. (1994).GISS: Goddard Institute for Space Studies.
GFDL: Geophysical Fluid Dynamics Laboratory.
UKMO: United Kingdom Meteorological Office.
OSU: Oregon State University.
The starkest difference in methodology is between Mendelsohn et al. (1994) and Rosenzweig and Parry (1994). Columns 1 and 2 reflect results from models estimated with different weights on the individual observations. Mendelsohn et al. (1994) suggest the column 2 estimates based on revenue weights are more appropriate because they reflect the economic value of crops. They suggest that the more negative estimates based on area weights (column 1) reflect the type of bias that may be introduced by focusing on cereal crops which generally have a lower value per hectare than many other crops such as fruits and vegetables. Contrasting the climate impact shock they estimate (column 2) with the types of yield shocks estimated by Rosenzweig and Parry (1994) (column 7) provides a dramatically different picture of the impact of climate change on US agriculture. Rosenzweig and Parry include some adjustments but, unfortunately, the yield shocks for the United States comparable to the Mendelsohn et al. study (climate change and adaptation with no CO2 effect) have not been reported. However, in their study adaptation did not have a particularly powerful effect on mitigating losses as reported by Reilly and Hohmann (1993). The relatively more benign impacts for the USA in the Rosenzweig and Parry yield estimates (column 8, with the CO2 and adaptation) are, in a large part, less severe because of the CO2. Thus, different methodologies, including adaptation but not the CO2 effect, apparently produce estimates of impact for four major climate scenarios in the order of -1 to +5% using a Mendelsohn et al. methodology but on the order of -10 to -25% using the Rosenzweig and Parry methodology. In deriving the -10 to -25% range, I assume that adaption in their study may have reduced losses by 5 to 10% whereas the CO2 fertilization effect reduced losses by 75 to 100%, which is the relative importance of these two factors on a global basis as in their data as estimated by Reilly and Hohmann (1993).
The Darwin et al. (1995) study used an independently derived set of climate shocks, representing climate change as a change in land class where the productivity of each land class was estimated from current data. Their methodology for estimating the direct effect of climate was more akin to Mendelsohn et al. (1994), using the observed differences in production across geographically varying climate as the basis for the projections. Their results, columns 3-6, help explain and confirm some of the differences between the other two studies. The initial shock on United States cereal production in the Darwin et al. (1995) study (column 5) is similarly (and generally more) severe than the yield shocks estimated by Rosenzweig and Parry (1994) (column 7). However, Darwin et al. (1995) estimate that, by just considering the immediate farm-level adjustment (without price changes and without expansion of agricultural production into new areas), farmers could offset between 70 and 120% of the initial losses (i.e., comparing column 5 and column 3). Note that this comparison is between impacts on cereal production and impacts on farm income, which is comparable (given that the simulation in column 3 does not allow prices to change) except that farm income includes impacts to agriculture for livestock and non-grains production as well. Columns 4 and 6 provide the estimates after full-adjustment including changes in world prices and trade, for cereal production and farm income. Note that the farm income effects with full adjustment (column 4) are sometimes worse than the farm income effects with only farm-level adaptation (column 3) because the Darwin et al. (1995) study considers worldwide effects with international trade. Thus, the impacts that occur in the rest of the world under the GISS and UKMO climate scenarios lead the US to lose international comparative advantage once full adjustment of international markets is considered.
Together these three studies indicate the wide range of estimated impacts for the same region and same climate scenarios depending on the methodology used. Mendelsohn et al. (1994) and Darwin et al. (1995) use methodologies that they argue more completely consider adaptation and they find impacts after adaptation to be generally less than Rosenzweig and Parry (1994), but even between these two approaches there are significant differences in estimated impacts for some climate scenarios in comparable estimates (columns 2 and 3).
The above discussion identified four separate factors that contribute to widely varying estimates of regional impacts of climate change apart from how or whether the CO2 effect on crops is included in the simulation. These factors - varying climate scenarios, wide variation across sites within a region, how genetic variability across known crop varieties is addressed within the crop-response-modelling approach, and differences across impact methodologies particularly in how different methods address the capability of farmers to adapt - appear to be of roughly equal magnitude in explaining the wide range of estimates.
Accurate consideration of national and local food supply and economic effects depends on an appraisal of changes in global food supply and prices. International markets can moderate or reinforce local and national changes. In 1988, for example, drought presented a more severe threat because it occurred coincidentally in several of the major grain-growing regions of the world. Reilly et al. (1994) demonstrate that considering country-level production impacts of climate change in the absence of consideration of the global impacts can generate highly misleading results. Agricultural exporting countries, whose productivity is reduced by climate change, may find themselves with a financial bonanza if world agricultural prices rise because of climate change. These same countries may suffer significant economic loss if climate change turns out to be generally beneficial to world agriculture even if agricultural productivity in their country benefits. This feature of the agricultural economy is well-known and reflects what is, in aggregate, an inelastic demand for food. This point, which is a fundamental observation of agricultural economists, means that absolutely no implications for food availability, price or farm financial success can be drawn from local and country-level estimates of production impacts of climate change unless one assumes that production changes around the world will generally balance to leave little impact on global production and prices. A country may attempt to carry out a set of policies that maintains a neutral effect on the country's agricultural sector vis-a-vis the rest of the world, but maintaining such policies will generally entail significant economic cost through subsidization of domestic agricultural production and/or consumption or through import or export controls. There are many different ways these costs may be borne (higher food prices, government expenditures, lost efficiency in the producing sector, lost export opportunities) depending on how the policies are structured.
There are now a number of different attempts to estimate the impacts of climate change on global agriculture, in part to consider the global impacts but more importantly to consider more accurately what the regional impacts could be recognizing that what happens to global agriculture due to climate change will likely be more important for the viability and economic success of local agriculture than what happens to local production potential itself. Kane et al. (1992) and Tobey et al. (1992) examined the sensitivity of agriculture to potential yield losses in major temperate grain-growing regions based on very stylized climate change impacts. They loosely linked the potential for yield losses in temperate regions to climate projections that showed increasing aridity in the continental mid-latitude areas. They made alternative assumptions about how agriculture might be affected in higher latitudes and in the tropics. They also developed scenarios that reflected the estimated yield impacts for different parts of the world that were summarized in the 1990 Intergovernmental Panel on Climate Change assessment (Parry et al., 1990). The yield response estimates used by Rosenzweig and Parry (1994) also reported in Fischer et al. (this volume) were also the basis of Reilly et al. (1994) and in greater detail Reilly et al. (1993). Many of the general conclusions are similar between the studies indicating that, given a set of yield shocks, economic modelling of international markets in itself is not a major source of difference in results even though there are major differences in the modelling approaches. Rather these different economic modelling approaches focus on different aspects and degrees of detail of agricultural economic interactions among crops, livestock, land use and the rest of the economy.
Among the issues that give rise to uncertainty in these studies are the following factors:
1. The timing of expected climate change. For example, Rosenzweig and Parry (1994) assume that the 4.0 to 5.2°C scenarios occur in 2060, but the most recent IPCC work suggests the mean estimate for 2060 is closer to 1.5°C and that the range of global temperature impacts by 2100 is likely to be between 1.0 and 5°C.2. Aggregation from detailed sites. Detailed plant growth models, the basis for many studies, require daily temperature and precipitation records for a 10- to 30-year historical climate record and detailed soil data, limiting the number of sites for which data are readily available and that can be practicably assessed. An alternative approach (Leemans and Solomon, 1993; Carter et al., 1991) makes use of geographic information system databases that contain more extensive information on current climates across the world. These efforts have not been linked to an economic model. Results confirm the pattern of relative decreased crop potential in the tropical areas and increased potential in the northern areas but are not aggregated to determine the net global effect.
3. Coverage of agricultural activities. Simulation of crop response models has been limited to a few major crops for a region, usually important grain crops, with yield effects extended to other crops. Left out are the indirect impacts of climate change through impacts on insect, disease and weed pests; on soils; and on livestock production. Mendelsohn et al. (1994) argue that their statistical approach accounts for all agricultural activities, implicitly accounting for the full effects of climate.
4. Other resource changes and competition for resources from other sectors. Land and water resource allocation is a conspicuous limitation in global studies. Water demand for other uses will grow, water use may have reached or passed sustainable levels of use in some areas, irrigation is responsible for salinization and land degradation, and water pricing and water system management are far from efficient under current conditions (e.g., Umali, 1993; Moore, 1991). Climate change also will affect demand for resources from other sectors.
The Darwin et al. (1995) study addresses many of these considerations in a global model including eight world regions. An Applied Computable General Equilibrium Model (ACGE), land and climatic resource changes are based on a geographic information system; changing climate shifts the distribution of land across several agroclimatic land classes. Other resource-using sectors are included and are also affected by climate change. The model is a static one, imposing climate change on current economic and agricultural markets and thus does not directly address the issue of timing.
The global results (Table 10.3) are comparable to Rosenzweig and Parry in terms of direct supply impacts for the world in the 'no adaptation' case, but the study finds that adaptation is able to turn global losses to small global benefits (unrestricted case). Even when the model is constrained to continue to produce on existing amounts of land within each region and prices are not allowed to respond, adaptation mitigates a significant share of the losses. These results contrast with those of Rosenzweig and Parry (1994), giving generally smaller impacts and possible benefits even without the CO2 effect and in that they show adaptation to be quite important.
Table 10.3. Percentage changes in the supply and production of cereals for the world by climate change scenario
|
Supply |
Production |
||
No adaptation |
Land use fixed |
Land use fixed |
No restrictions |
|
World |
||||
GISS |
-22.6 |
-2.4 |
0.2 |
0.9 |
GFDL |
-23.5 |
-4.4 |
-0.6 |
0.3 |
UKMO |
-29.3 |
-6.4 |
-0.2 |
1.2 |
OSU |
-18.6 |
-3.9 |
-0.5 |
0.2 |
Note: Changes in supply represent the additional quantities firms would be willing to sell at 1990 prices under the alternative climate. Changes in production represent changes in equilibrium quantities.Source: Darwin et al. (1995).
Again, the global results are important because they are the first step in considering whether a local economy's consumers will be able to purchase food if it is unavailable domestically, how local producers may be affected by changes in demand for their crops, or how the cost of a country's agricultural policies may change because of changing international market conditions.
The previous sections documented the wide range of uncertainty in the potential direction and magnitude of climate change impact. While many new studies have been conducted, most have focused on specific climate scenarios associated with 2xCO2 GCM scenarios or arbitrary changes in climatic conditions to provide evidence of the general sensitivity of agriculture and crop production to climate change. The wide range of estimates limits the ability to extend, interpolate or extrapolate from the specific climate scenarios used in these studies to 'more' or 'less' climate change or to draw implications for impacts beyond the sites where studies were conducted.
Given these uncertainties in both magnitude and direction of impact, a key issue is vulnerability to possible climate change. Vulnerability is used here to mean the potential for negative consequences that are difficult to ameliorate through adaptive measures given the range of possible climate changes that might reasonably occur. Defining an area or population as vulnerable is, thus, not a prediction of negative consequences of climate change; it is an indication that across the range of possible climate changes, there are some climatic outcomes that would lead to relatively more serious consequences for the region than for other regions.
Vulnerability has been used rather loosely in many discussions. Before discussing some of the research that has examined potential vulnerability, I introduce a more formal definition. For the sake of simplicity, consider that climate can be described as a single variable, C. We are uncertain about what value C will take at some future point but we can describe the probability, p, that C will take on a specific value by the probability density function f(C). Further consider that we are able to describe the sensitivity of agriculture, A, to changes in climate by the function g(C). We can then define the expected loss function, L(C) as f(C) x g(C). A population, region, or crop is relatively more vulnerable under this definition if the area under L(C) where damages occur is larger than for a comparison population, region, or crop. Thus, I use the term vulnerability to describe only that portion of L(C) where damages occur. For other purposes, it useful to consider expected (net) damage (or benefit), that is the mean of the values of loss function which is a probability weighted mean of the damages.
Two, purely illustrative, numerical examples are plotted in Figure 10.1. For these examples I have chosen to represent f(C) as a gamma distribution. In panel A, damages are represented by a quadratic function while in panel B, damages are represented by a logarithmic function. These choices illustrate just two of the ways that our expectations about the degree of future climate change and our understanding of the sensitivity of an agricultural system to climate change may interact. In these numerical illustrations, the system characterized by quadratic losses (Panel A) is more vulnerable to loss than the system described by logarithmic losses. Even though the quadratic sensitivity to climate leads to potentially larger losses at extreme temperature change, the system is less vulnerable because climate change is not likely to be that extreme in this example. In fact, the small region of beneficial warming (negative damages) in Panel A gives rise to a substantial possibility of beneficial effects of warming for the system described in this panel. In Panel B, in contrast, damages initially rise sharply but the rate of increase slows. This characterization of system sensitivity indicates damages across the entire range of expected temperature change. Even though damages do not have the potential to become as severe as in Panel A, the system is more vulnerable to damage because climate is more likely to be in the relatively higher damage range of the sensitivity function.
In practice, multiple dimensions of climate affect any agricultural system. The simple characterization in Figure 10.1 is meant to make the definition of vulnerability mathematically precise even though it is not possible at this time to estimate formally the multidimensional, joint distribution of important climate variables. Nor do we precisely know the damage function that relates changes in these climate variables to agricultural impacts. The advantage is to make explicit that we must consider both our expectations with regard to climate and damage sensitivity. To make the example concrete, a semi-arid area may be extremely sensitive to damage if it becomes more arid. But, if our expectation is that it is highly likely that the region will become wetter, the region is not vulnerable. Another region in a humid agroclimatic zone may be vulnerable if substantial warming and drying are likely for the area.
Up to this point, I have not been explicit with regard to what I propose to measure as a damage. The existing literature suggests several different possible measures and therefore several different dimensions of vulnerability. Many studies focus on crop yields. Evidence suggests that yields of crops grown where temperature could easily exceed threshold values during critical crop growth periods are more vulnerable to warming (e.g., rice sterility: Matthews et al., 1994a, b).
Figure 10.1. Defining vulnerability
Farmer or farm sector vulnerability may be measured in terms of impact on profitability or viability of the farming system. Farmers with limited financial resources and farming systems with few adaptive technological opportunities available to limit or reverse adverse climate change may suffer significant disruption and financial loss for relatively small changes in crop yields and productivity or these farms may be located in areas more likely to suffer yield losses. For example, Parry et al. (1988a, b) focused on semi-arid and cool temperate and cold agricultural areas as those that might be more clearly affected by climate change and climate variability.
Regional economic vulnerability reflects the sensitivity of the regional or national economy to farm sector and climate change impacts. A regional economy that offers only limited employment alternatives for workers dislocated by the changing profitability of farming and other climatically sensitive sectors may be relatively more vulnerable than those that are economically diverse. For example, Rosenberg (1993) examined the Great Plains area of the United States because of its heavy dependence on agriculture. Increasing aridity is expected in this region under climate change and thus it was considered to be potentially more economically vulnerable than other regions in the USA.
Hunger vulnerability has been used to mean the 'aggregate measure of the factors that influence exposure to hunger and predisposition to its consequences' involving 'interactions of climate change, resource constraints, population growth, and economic development' (Downing, 1992; Bohl et al., 1994). Downing (1992) concluded that the semi-extensive farming zone, on the margin of more intensive land uses, appears to be particularly sensitive to small changes in climate. Socio-economic groups in such areas, already vulnerable in terms of self-sufficiency and food security, could be further marginalized. We likely ought not to look only at agriculturally dependent people. The means people have within society and the family to obtain food and how their allocation will change if production potential changes must be considered. A poor urban household may suffer due to production losses elsewhere in the region while the rural farmer may continue to eat. Or, women and children of rural peasant farms may go hungry, while 'excess' production from the region is sold. Assessing who has the means and rights to food during shortfalls is thus likely more critical in a climate vulnerability study than assessing how production may change. For hunger and famine in general, the relative importance of acquiring (versus producing) food has been demonstrated by Sen (1981, 1993).
Given the diverse currently existing conditions, the geographical variation likely to exist in any climate change scenario, and the wide uncertainty that must be associated with local prediction of future climates, some vulnerable agricultural areas and populations likely exist for nearly every region even if the expected value for the region is a net benefit. This makes vulnerability a relative concept - while there may be a few areas where even the most extreme climate change we can imagine would not generate losses, in general, the problem is to consider whether a particular region or population is relatively more vulnerable than others.
While perhaps most difficult to evaluate, vulnerability in terms of hunger and malnutrition ought to be the first concern. If so, then we can almost certainly eliminate the richer countries of the world as vulnerable. For poorer regions, it is the poorest members of these areas or those that could be made poor by climate change that are most at risk. The wide uncertainty with regard to local and regional climate change means it is difficult to rule out negative possibilities for any area. Thus, without even considering specific climate scenarios, we can assert that those who are currently poor, malnourished and dependent on local production for food are the most vulnerable in terms of hunger and malnutrition to climate change of the world's populations. Similarly, severe economic vulnerability is most likely where a large share of the population depend on agriculture, leaving little alternative employment opportunities. Again, we need not assess climate scenarios or projected yield changes to establish where these vulnerable populations live. Given these considerations, Table 10.4 presents some of the critical dimensions of areas of the world that might be used to assess vulnerability. While the table is too aggregated to identify specifically vulnerable populations, it is indicative of where many of these people are likely to be.
Table 10.4. Basic regional agricultural indicators and vulnerability
|
Sub-Saharan Africa |
Near East/North Africa |
South Asia |
Southeast Asia |
East Asia |
Oceania |
Former USSR |
Europe |
Latin America |
USA, Canada | |
Ag. land (%)* |
41 |
27 |
55 |
36 |
51 |
57 |
27 |
47 |
36 |
27 | |
|
Cropland (%)* |
7 |
7 |
44 |
13 |
11 |
6 |
10 |
29 |
7 |
13 |
|
Irrigated (%)* |
5 |
21 |
31 |
21 |
11 |
4 |
9 |
12 |
10 |
8 |
Land area (106 ha) |
2390 |
1 167 |
478 |
615 |
993 |
845 |
2227 |
473 |
2052 |
1 839 | |
Climate |
tropical; arid, humid |
subtropical, tropical; arid |
tropical, subtropical; humid, arid |
tropical; humid |
subtropical, temperate oceanic, continental; humid |
tropical, temperate, oceanic subtropical; arid, humid |
polar, continental, temperate oceanic; humid, arid |
temperate oceanic, some subtropical, humid, arid |
tropical, subtropical; mostly humid |
continental subtropical, polar, temp. oceanic, mid, arid | |
Pop. (106) |
566 |
287 |
1145 |
451 |
1 333 |
27 |
289 |
510 |
447 |
277 | |
Ag. pop.(%) |
62 |
32 |
63 |
49 |
59 |
17 |
13 |
8 |
27 |
3 | |
Pop./ha cropland |
3.6 |
3.4 |
5.4 |
5.7 |
12.6 |
0.5 |
1.3 |
3.7 |
2.9 |
1.2 | |
Ag. prod. (106t) | |||||||||||
|
Cereals |
57 |
79 |
258 |
130 |
433 |
24 |
180 |
255 |
111 |
388 |
|
Roots and tubers |
111 |
12.5 |
26 |
50 |
159 |
3 |
65 |
79 |
45 |
22 |
|
Pulses |
5.7 |
4.1 |
14.4 |
2.5 |
6.3 |
2 |
6 |
7 |
5.8 |
2.2 |
|
S. cane and beet |
60 |
39 |
297 |
181 |
103 |
32 |
62 |
144 |
494 |
56 |
|
Meat |
6.7 |
5.5 |
5.7 |
6.4 |
39.6 |
4.5 |
17 |
42 |
20.5 |
33.5 |
1991 GNP/cap.** |
350 |
1 940 |
320 |
930 |
590 |
13780 |
2700 |
15300 |
2390 |
22 100 | |
Annual growth** |
-1.2 |
-2.4 |
3.1 |
3.9 |
7.1 |
1.5 |
N.A. |
2.2 |
-0.3 |
1.7 | |
Ag. (% of GDP)** |
>30% |
10-19% |
>30% |
20 to>30% |
20-29% |
<6% |
10-29% |
<6% |
10-19% |
<6% |
* Agricultural land includes grazing and cropland, reported as a percentage of total land area. Cropland is reported as a percentage of agricultural land. Irrigated area is reported as a percentage of cropland.** GNP is in 1991 USA dollars; annual growth, per cent per annum, is for the period 1980-1991. Source: Computed from FAO (1992); GNP per caput, GNP growth rates, and agriculture as a share of the economy are from World Development Indicators in World Bank (1993) and temperature and climate classes from Rötter et al. (1995). Note: East Asia GNP excludes Japan. Also, regional GNP data generally include only those countries for which data are given in Table 1 in World Development Indicators. Countries with more than 4 million population for which GNP data are not available include Vietnam, Democratic Republic of Korea, Afghanistan, Cuba, Iraq, Myanmar, Cambodia, Zaire, Somalia, Libya and Angola.
Because of the wide range of uncertainty in precipitation, the only climatic dimension likely to enter significantly in an assessment of vulnerability is temperature. Cool regions are more likely to be limited by low temperatures and thus warming may prove beneficial - these areas may still suffer if precipitation changes are adverse. But, further warming is unlikely to benefit already warm regions. Thus, global warming appears somewhat stacked against the already warm areas. Coincidentally (or not), these regions tend to also be home to some of the world's poorest.
The focus on hunger and malnutrition as a first priority does not mean that other types of vulnerability are unimportant. Regional economic development, land degradation, or increased environmental stress resulting from agricultural production under a changed climate are important concerns as well.
The hierarchy of damage considerations as above - hunger, regional economic, farmer/farm sector, and yield vulnerability - helps to focus on adaptive strategies that reduce vulnerability. How can we avoid yield failures? If yields fail, what other crops can be grown? If farming becomes uneconomic, what other opportunities for employment exist? If the people of the region can no longer produce food, what other sources of food are available and how will they earn the income necessary to purchase food or what other means does the society in which they live have to provide food assistance?
Historically, farming systems have adapted to changing economic conditions, technology and resource availabilities and have kept pace with a growing population (Rosenberg, 1992; CAST, 1992). Evidence exists that agricultural innovation responds to economic incentives such as factor prices and can relocate geographically (Hayami and Ruttan, 1985; CAST, 1992). A number of studies indicate that adaptation and adjustment will be important to limit losses or to take advantage of improving climatic conditions (e.g., US National Academy of Sciences, 1991; Rosenberg, 1992; Rosenberg and Crosson, 1991; CAST, 1992; Mendelsohn et al., 1994).
Despite the successful historical record, issues of future adaptation to climate change arise with regard to whether the rate of change of climate and required adaptation would add significantly to the disruption likely due to future changes in economic conditions, technology and resource availabilities (Gommes, 1993; Harvey, 1993; Kane and Reilly, 1993; Smit, 1993; Norse, 1994; Pittock, 1994; Reilly, 1994). If climate change is gradual, it may be a small factor that goes unnoticed by most farmers as they adjust to other more profound changes in agriculture stemming from new technology, increasing demand for food, and other environmental concerns such as pesticide use, water quality, and land preservation. However, some researchers see climate change as a significant addition to future stresses; where adapting to yet another stress such as climate change may be beyond the capability of the system. Part of the divergence in views may be due to different interpretations of adaptation which include: the prevention of loss, tolerating loss, or relocating to avoid loss (Smit, 1993). And, while the technological potential to adapt may exist, the socio-economic capability to adapt likely differs for different types of agricultural systems (Reilly and Hohmann, 1993).
Nearly all agricultural impact studies conducted over the past five years have considered some technological options for adapting to climate change. Among those that offer promise are:
· Seasonal changes and sowing dates. For frost-limited growing areas (i.e., temperate and cold areas), warming could extend the season, allowing planting of longer maturity annual varieties that achieve higher yields (e.g., Le Houerou, 1990; Rowntree, 1990a, b). For short-season crops such as wheat, rice, barley, oats, and many vegetable crops, extension of the growing season may allow more crops per year, autumn planting, or, where warming leads to regular summer highs above critical thresholds, a split season with a short summer fallow may be possible. For subtropical and tropical areas where growing season is limited by precipitation or where the cropping already occurs throughout the year, the ability to extend the growing season may be more limited and depends on how precipitation patterns change. A study for Thailand found yield losses in the warmer season partially offset by gains in the cooler season (Parry et al., 1992).· Different crop variety or species. For most major crops, varieties exist with a wide range of maturities and climatic tolerances. For example, Matthews et al. (1994a, b) identified wide genetic variability among rice varieties as a reasonably easy response to spikelet sterility in rice that occurred in simulations for South and Southeast Asia. Studies in Australia showed that responses to climate change are strongly cultivar dependent (Wang et al., 1992). Longer-season cultivars were shown to provide a steadier yield under more variable conditions (Connor and Wang, 1993). In general, such changes may lead to higher yields or may only partly offset losses in yields or profitability. Crop diversification in Canada (Cohen et al., 1992) and in China (Hulme et al., 1992) has been identified as an adaptive response.
· New crop varieties. The genetic base is broad for most crops but limited for some (e.g., kiwi fruit). A study by Easterling et al. (1993) explored how hypothetical new varieties would respond to climate change (also reported in McKenney et al., 1992). Heat, drought and pest resistance; salt tolerance, and general improvements in crop yield and quality would be beneficial (Smit, 1993). Genetic engineering and gene mapping offer the potential for introducing a wider range of traits. Difficulty in assuring traits are efficaciously expressed in the full plant, consumer concerns, profitability and regulatory hurdles have slowed the introduction of genetically engineered varieties compared with early estimates (Reilly, 1989; Caswell et al., 1994).
· Water supply and irrigation systems. Across studies, irrigated agriculture is in general less negatively affected than dryland agriculture but adding irrigation is costly and subject to the availability of water supplies. Climate change will also affect future water supplies. There is wide scope for enhancing irrigation efficiency through adoption of drip irrigation systems and other water-conserving technologies (FAO, 1989,1991) but successful adoption will require substantial changes in how irrigation systems are managed and how water resources are priced. Because inadequate water systems are responsible for current problems of land degradation, and because competition for water is likely to increase, there likely will be a need for changes in the management and pricing of water regardless of whether and how climate changes (Vaux, 1990,1991; World Bank, 1994). Tillage method and incorporation of crop residues are other means of increasing the useful water supply for cropping.
· Other inputs and management adjustments. Added nitrogen and other fertilizers would likely be necessary to take full advantage of the CO2 effect. Where high levels of nitrogen are applied, nitrogen not used by the crop may be leached into the groundwater, runoff into surface water, or be released from the soil as nitrous oxide. Additional nitrogen in groundwater and surface water has been linked to health effects in humans and affects aquatic ecosystems. Studies have also considered a wider range of adjustments in tillage, grain drying and other field operations (Kaiser et al., 1993; Smit, 1993).
· Tillage. Minimum and reduced tillage technologies in combination with planting of cover crops and green manure crops offers substantial possibilities to reverse existing soil organic matter, soil erosion, and nutrient loss and to combat potential further losses due to climate change (Rasmussen and Collins, 1991; Logan, 1991; Edwards et al., 1992; Langdale et al., 1992; Peterson et al., 1993; Brinkman and Sombroek, this volume). Reduced and minimum tillage techniques have spread widely in some countries but are more limited in other regions. There is considerable current interest in transferring these techniques to other regions (Cameron and Oram, 1994).
· Improved short-term climate prediction. Linking agricultural management to seasonal climate predictions (currently largely based on the El Niño Southern Oscillation Phenomenon), where such predictions can be made with reliability, can allow management to adapt incrementally to climate change. Management/climate predictor links are an important and growing part of agricultural extension in both developed and developing countries (McKeon et al., 1990, 1993; Nicholls and Wong, 1990).
While identifying many specific technological adaptation options, Smit (1993) concluded that necessary research on their cost and ease of adoption had not yet been conducted.
One measure of the potential for adaptation is to consider the historical record on past speeds of adoption of new technologies (Table 10.5). Adoption of new or different technologies depends on many factors: economic incentives; varying resource and climatic conditions; the existence of other technologies (transportation systems and markets); the availability of information; and the remaining economic life of equipment and structures (e.g., dams and water supply systems).
Specific technologies can only provide a successful adaptive response if they are adopted in appropriate situations. A variety of issues has been considered, including land-use planning, watershed management, disaster vulnerability assessment, consideration of port and rail adequacy, trade policy, and the various programmes countries use to encourage or control production, limit food prices, and manage resource inputs to agriculture (CAST, 1992; US OTA, 1993; Smit, 1993; Reilly et al., 1994; Singh, 1994). For example, studies suggest that current agricultural institutions and policies in the USA may discourage farm management adaptation strategies such as altering crop mix, by supporting prices of crops not well-suited to a changing climate, providing disaster payments when crops fail, or prohibiting imports through import quotas (Lewandrowski and Brazee, 1993).
Table 10.5. Speed of adoption for some major adaptation measures
Adaptation |
Adjustment time (yrs) |
Reference |
Variety adoption |
3-14 |
Dalrymple, 1986; Griliches, 1957; Plucknett et al., 1987; CIMMYT, 1991 |
Dams and irrigation |
50-100 |
James and Lee, 1971; Howe, 1971 |
Variety development |
8-15 |
Plucknett et al., 1987; Knudson, 1988 |
Tillage systems |
10-12 |
Hill et al., 1994; Dickey et al., 1987; Schertz, 1988 |
New crop adoption: soybeans |
15-30 |
FAO, Agrostat - various years |
Opening new lands |
3-10 |
Medvedev, 1987; Plusquellec, 1990 |
Irrigation equipment |
20-25 |
Turner and Anderson, 1980 |
Transportation system |
3-5 |
(A. Talvitie, World Bank, pers. comm., 1994) |
Fertilizer adoption |
10 |
Pieri, 1992; Thompson and Wan, 1992 |
Existing gaps between best yields and the average farm yields remain unexplained but many are due in part to socio-economic considerations (Oram and Hojjati, 1995; Bumb, 1995); this adds considerable uncertainty to estimates of the potential for adaptation particularly in developing countries. For example, Baethgen (1994) found that a better selection of wheat variety combined with improved fertilizer regime could double yields achieved at a site in Uruguay to 6 t/ha under the current climate with current management practices. Under the UKMO climate scenario, yields fell to 5 t/ha, still well above 2.5-3.0 t/ha currently achieved by farmers in the area. On the other hand, Singh (1994) concluded that the normal need to plan for storms and extreme weather events in Pacific island nations creates significant resiliency. Whether technologies meet the self-described needs of peasant farmers is critical in their adaptation (Cáceres, 1993). Other studies document how individuals cope with environmental disasters, identifying how strongly political, economic and ethnic factors interact to facilitate or prevent coping in cases ranging from the Dust Bowl disaster in the USA to floods in Bangladesh to famines in the Sudan, Ethiopia and Mozambique (McGregor, 1994). These considerations indicate the need for local capability to develop and evaluate potential adaptations that fit changing conditions (COSEPUP, 1992). Important strategies for improving the ability of agriculture to respond to diverse demands and pressures, drawn from past efforts to transfer technology and provide assistance for agricultural development, include:
· Improved training and general education of populations dependent on agriculture, particularly in countries where education of rural workers is currently limited. Agronomic experts can provide guidance on possible strategies and technologies that may be effective. Farmers must evaluate and compare these options to find those appropriate to their needs and the circumstances of their farm.· Identification of the present vulnerabilities of agricultural systems, causes of resource degradation, and existing systems that are resilient and sustainable. Strategies that are effective in dealing with current climate variability and resource degradation are also likely to increase resilience and adaptability to future climate change.
· Agricultural research centres and experiment stations can examine the 'robustness' of present farming systems (i.e. their resilience to extremes of heat, cold, frost, water shortage, pest damage and other factors) and also test the robustness of new farming strategies as they are developed to meet changes in climate, technology, prices, costs and other factors.
· Interactive communication that brings research results to farmers and farmers' problems, perspectives and successes to researchers is an essential part of the agricultural research system.
· Agricultural research provides a foundation for adaptation. Genetic variability for most major crops is wide relative to projected climate change. Preservation and effective use of this genetic material would provide the basis for new variety development. Continually changing climate is likely to increase the value of networks of experiment stations that can share genetic material and research results.
· Food programmes and other social security programmes would provide insurance against local supply changes. International famine and hunger programmes need to be considered with respect to their adequacy.
· Transportation, distribution and market integration provide the infrastructure to supply food during crop shortfalls that might be induced in some regions because of climate variability or worsening of agricultural conditions.
· Existing policies may limit efficient response to climate change. Changes in policies such as crop subsidy schemes, land tenure systems, water pricing and allocation, and international trade barriers could increase the adaptive capability of agriculture.
Many of the above strategies will be beneficial regardless of how or whether climate changes. Goals and objectives among countries and farmers vary considerably. Current climate conditions and likely future climates also vary. Building the capability to detect change and evaluate possible responses is fundamental to successful adaptation. Thus, even without having clear predictions of climate change, is it possible to identify some strategies that reduce potential vulnerability.
Adams, R.M., McCarl, B.A., Dudek, D.J. and Glyer, J.D. 1988. Implications of global climate change for western agriculture. Western J. Agric. Economics 13: 348-356.
Baethgen, W.E. 1994. Impact of climate change on barley in Uruguay: yield changes and analysis of nitrogen management systems. In: Implications of Climate Change for International Agriculture: Crop Modelling Study. C. Rosenzweig and A. Iglesias (eds.). US Environmental Protection Agency, Washington DC. pp. 1-13.
Bohl, H.G., Downing, T.E. and Watts, M.J. 1994. Climate change and social vulnerability: toward a sociology and geography of food insecurity. Global Environ. Change 4 (1): 37-48.
Brklacich, M. and Smit, B. 1992. Implications of changes in climatic averages and variability on food production opportunities in Ontario, Canada. Climatic Change 20: 1-21.
Brklacich, M., Stewart, R., Kirkwood, V. and Muma, R. 1994. Effects of global climate change on wheat yields in the Canadian prairie. In: Implications of Climate Change for International Agriculture: Crop Modelling Study. C. Rosenzweig and A. Iglesias (eds.). US Environmental Protection Agency, Washington DC. pp. 1-23.
Bumb, B. 1995: Growth potential of existing technology is insufficiently tapped. In: Population and Food in the. Early 21st Century: Meeting Future Food Demand of an Increasing World Population. N. Islam (ed.). Occasional Paper. International Food Policy Research Institute (IFPRI), Washington DC. pp. 191-205.
Cáceres, D.M. 1993. Peasant Strategies and Models of Technological Change: A Case Study from Central Argentina. M. Phil. Thesis. University of Manchester, Manchester.
Cameron, D. and Oram, P. 1994. Minimum and Reduced Tillage: Its Use in North America and Western Europe and its Potential Application in Eastern Europe, Russia, and Central Asia. International Food Policy Research Institute, Washington DC. 121 p.
Carter, T.R., Porter, J.R. and Parry, M.L. 1991. Climatic warming and crop potential in Europe: prospects and uncertainties. Global Environ. Change 1: 291-312.
CAST [Council for Agricultural Science and Technology]. 1992. Preparing U.S. Agriculture for Global Climate Change. Task Force Report No. 119, CAST, Ames, Iowa. 96 p.
Caswell, M.F, Fuglie, K.O and Klotz, C.A. 1994. Agricultural Biotechnology: An Economic Perspective. AER No. 687. US Department of Agriculture, Washington DC. 52 p.
CIMMYT. 1991. Annual Report: Improving the Productivity of Maize and Wheat in Developing Countries: An Assessment of Impact. Centro Internacional de Mejoramiento de Maiz y Trigo, Mexico City.
Cohen, S., Wheaton, E. and Masterton, J. 1992. Impacts of climatic change scenarios in the prairie provinces: A case study from Canada. SRC Publication No. E-2900-4-D-92. Saskatchewan Research Council, Saskatoon, Canada.
Connor, D.J. and Wang, Y.P. 1993. Climatic change and the Australian wheat crop. In: Proceedings of the Third Symposium on the Impact of Climatic Change on Agricultural Production in the Pacific Rim. S. Geng (ed.). Central Weather Bureau, Ministry of Transport and Communications, Republic of China.
COSEPUP [Committee on Science, Engineering and Public Policy]. 1992. Policy Implications of Global Warming. National Academy Press, Washington DC.
Dalrymple, D.G. 1986. Development and Spread of High-Yielding Rice Varieties in Developing Countries. 7th edn. US AID, Washington DC.
Darwin, R., Tsigas, M., Lewandrowski, J. and Raneses, A. 1995. World Agriculture and Climate Change: Economic Adaptation. Report No. AER-709. Economic Research Service, Washington DC.
Dickey, E.C., Jasa, P.J., Dolesh, B.J., Brown, L.A. and Rockwell, S.K. 1987. Conservation tillage: perceived and actual use. J. Soil Water Conserv. (Nov.-Dec.): 431-434.
Downing, T.E. 1992. Climate Change and Vulnerable Places: Global Food Security and Country Studies in Zimbabwe, Kenya, Senegal, and Chile. Research Report No. 1, Environmental Change Unit, University of Oxford. 54 p.
Easterling, WE. Ill, Crosson, P.R., Rosenberg, N.J., McKenney, M., Katz, L.A. and Lemon, K. 1993. Agricultural impacts of and responses to climate change in the Missouri-Iowa-Nebraska-Kansas (MINK) region. Climatic Change 24: 23-61.
Edwards, J.H., Wood, C.W, Thurow, D.L. and Ruff, M.E. 1992. Tillage and crop rotation effects on fertility status of a hapludult soil. Soil Science Society of America Journal 56: 1577-1582.
Fischer, G., Frohberg, K., Parry, M.L. and Rosenzweig, C. 1994. Climate change and world food supply, demand and trade. Global Environ. Change. 4 (1): 7-23.
FAO. 1961-1990. Agrostat 1961-1990. FAO, Rome.
FAO. 1989. Guidelines for designing and evaluating surface irrigation systems. Irrigation and Drainage Paper 45. FAO, Rome.
FAO. 1991. Water harvesting. AGL Miscellaneous Paper 17. FAO, Rome.
FAO. 1992. Agrostat. FAO Statistics Division, Rome.
Gommes, R. 1993. Current climate and population constraints on world agriculture. In: Agricultural Dimensions of Global Climate Change. H.M. Kaiser and T.E. Drennen (eds.). St. Lucie Press, Delray Beach, Florida. pp. 67-86.
Griliches, Z. 1957. Hybrid corn: an exploration in the economics of technological change. Econometrica. 25: 501-522.
Harvey, L.D.D. 1993. Comments on 'an empirical study of the economic effects of climate change on world agriculture'. Climatic Change 21: 273-275.
Hayami, Y. and Ruttan, V.W. 1985. Disequilibrium in world agriculture. Agricultural Development: An International Perspective. Johns Hopkins Univ. Press, Baltimore. pp. 367-415.
Hill, PR., Griffith, D.R., Steinhardt, G.C. and Parsons, S.D. 1994. The Evolution and History of No-Till Farming in the Midwest. Purdue University, West Lafayette, IN.
Howe, C. 1971. Benefit-cost analysis for water system planning. Water Resources Monograph 2. American Geophysical Union, Washington DC.
Hulme, M., Wigley, T., Jiang, T., Zhao, Z., Wang, F., Ding, Y., Leemans, R. and Markham, A. 1992. Climate Change due to the Greenhouse Effect and its Implications for China. CRU/WWF/SMA, World Wide Fund for Nature, Gland, Switzerland.
James, L.D. and Lee, R.R. 1971. Economics of Water Resources Planning. McGraw-Hill, New York.
Kaiser, H.M., Riha, S.J., Wilks, D.S., Rossiter, D.G. and Sampath, R. 1993. A farm-level analysis of economic and agronomic impacts of gradual global warming. Amer. J. Agr. Econ. 75: 387-398.
Kane, S. and Reilly, J. 1993. Reply to comment by L.D. Danny Harvey on 'an empirical study of the economic effects of climate change on world agriculture'. Climatic Change 21: 277-279.
Kane, S., Reilly, J. and Tobey, J. 1992. An empirical study of the economic effects of climate change on world agriculture. Clim. Change 21: 17-35.
Knudson, M. 1988. The Research and Development of Competing Biological Innovations: The Case of Semi- and Hybrid Wheats. Ph.D. dissertation, University of Minnesota, St. Paul.
Langdale, G.W., West, L.T. and Bruce, R.R. 1992. Restoration of eroded soil with conservation tillage. Soil Technology 5: 81-90.
Le Houerou, H.N. 1990. Global change: vegetation, ecosystems and land use in the Mediterranean basin by the twenty-first century. Israel J. Botany 39: 481-508.
Leemans, R. and Solomon, A.M. 1993. Modelling the potential in yield and distribution of the earth's crops under a warmed climate. Clim. Res. 3: 79-96.
Lewandrowski, J.K. and Brazee, R.J. 1993. Farm programs and climate change. Climatic Change 23: 1-20.
Logan, T.J. 1991. Tillage systems and soil properties in North America. Tillage Research 20: 241-270.
Matthews, R.B., Kropff, M.J. and Bachelet, D. 1994a. Climate change and rice production in Asia. Entwicklung und Ländlicherraum 1: 16-19.
Matthews, R.B., Kropff, M.J., Bachelet, D. and van Laar, H.H. 1994b. The impact of global climate change on rice production in Asia: A simulation study. Report No. ERL-COR-821. US Environmental Protection Agency, Environmental Research Laboratory, Corvallis.
McGregor, J. 1994. Climate change and involuntary migration. Food Policy, 19 (2): 121-132.
McKenney, M.S., Easterling, W.E. and Rosenberg, N.J. 1992. Simulation of crop productivity and responses to climate change in the year 2030: The role of future technologies, adjustments and adaptations. Agricultural and Forest Meteorology 59: 103-127.
McKeon, G.M., Day, K.A., Howden, S.M., Mott, J.J., Orr, D.M., Scattini, W.J. and Weston, E.J. 1990. Management for pastoral production in northern Australian savannas. Journal of Biogeography 17: 355-372.
McKeon, G.M., Howden, S.M., Abel, N.O.J. and King, J.M. 1993. Climate change: adapting tropical and subtropical grasslands. In: Proceedings XVII International Grassland Congress. pp. 1181-1190.
Medvedev, Z.A. 1987. Soviet Agriculture. W.W. Norton, New York.
Mendelsohn, R., Nordhaus, W. and Shaw, D. 1994. The impact of climate on agriculture: a Ricardian approach. Amer. Econ. Rev. 84: 753-771.
Moore, M.R. 1991. The Bureau of Reclamation's new mandate for irrigation water conservation: purposes and policy alternatives. Water Resources Res. 27: 145-155.
Nicholls, N. and Wong, K.K. 1990. Dependence of rainfall variability on mean rainfall, latitude and the Southern Oscillation. Journal of Climatology 3: 163-170.
Norse, D. 1994. Multiple threats to regional food production: environment, economy, population? Food Policy 19 (2): 133-148.
Oram, P.A. and Hojjati, B. 1995. The growth potential of existing agricultural technology. In: Population and Food in the Early 21st Century: Meeting Future Food Demand of an Increasing World Population. N. Islam (ed.). Occasional Paper. International Food Policy Research Institute (IFPRI), Washington DC. pp. 167-189.
Parry, M.L., Carter, T.R. and Konijn, N.T. (eds.). 1988a. The Impact of Climate Variations on Agriculture: Volume 1: Assessments in Cool Temperate and Cold Regions. Kluwer Academic, Dordrecht. 876 p.
Parry, M.L., Carter, T.R. and Konijn, N.T. (eds.). 1988b. The Impact of Climatic Variations on Agriculture: Vol. 2 Assessment in Semi-arid Regions. Kluwer Academic, Dordrecht.
Parry, M.L., Duinker, P.N., Morison, J.I.L., Porter, J.H., Reilly, J. and Wright, L.J. 1990. Agriculture and forestry. In: Climate Change: The IPCC Impacts Assessment. W.J. McG. Tegart, G.W. Sheldon and D.C. Griffiths (eds.). Australian Government Printing Office, Canberra. pp. 2-1 - 2-45.
Parry, M.L., Blantran de Rozari, M., Chong, A.L. and Panich, S. (eds.). 1992. The Potential Socio-Economic Effects of Climate Change in South-East Asia. United Nations Environment Programme, Nairobi.
Peterson, G.A., Westfall, D.G. and Cole, C.V. 1993. Agroecosystem approach to soil and crop management research. Soil Science Society of America Journal 57: 1354-1360.
Pieri, C. 1992. Fertility of Soils: The Future for Farmers in the West African Savannah. Springer-Verlag, Berlin.
Pittock, A.B. 1994. Climate and food supply. Nature 371: 25.
Plucknett, D.L., Smith, N.J.H., Williams, J.T. and Anishetty, N.M. 1987. Gene Banks and the World's Food. Princeton University Press, Princeton.
Plusquellec, H. 1990. The Gezira Irrigation Scheme in Sudan: objectives, design, and performance. Tech. Paper No. 120. World Bank, Washington DC.
Rasmussen, P.E. and Collins, H.P. 1991. Long-term impacts of tillage, fertiliser and crop residue on soil organic matter in temperate semiarid regions. Advances in Agronomy 45: 93-134.
Reilly, J. 1989. Consumer Effects of Biotechnology. AIB No. 581. US Department of Agriculture, Washington DC. lip.
Reilly, J. 1994. Crops and climate change. Nature 367: 118-119.
Reilly, J. and Hohmann, N. 1993. Climate change and agriculture: the role of international trade. Amer. Econ. Rev. 83: 306-312.
Reilly, J., Hohmann, N. and Kane, S. 1993. Climate Change and Agriculture: Global and Regional Effects Using an Economic Model of International Trade. MIT-CEEPR 93-012WP. Massachusetts Institute for Technology, Center for Energy and Environmental Policy, Boston.
Reilly, J., Hohmann, N. and Kane, S. 1994. Climate change and agricultural trade: who benefits, who loses? Global Envir. Change 4: 24-36.
Reilly, J., Baethgen, W., Chege, RE., van de Geijn, S.C., Erda, L., Iglesias, A., Kenny, G., Patterson, D., Rogasik, J., Rötter, R., Rosenzweig, C., Sombroek, W. and Westbrook, J. 1996. Agriculture in a changing climate: Impacts and adaptation, In: Changing Climate: Impacts and Response Strategies, Report of Working Group II of the Intergovernmental Panel on Climate Change. Chapter 13. Cambridge University Press, Cambridge, UK (in press).
Rosenberg, N.J. 1992. Adaptation of agriculture to climate change. Climatic Change 21: 385-405.
Rosenberg, N.J. 1993. Towards an Integrated Assessment of Climate Change: The MINK Study. Kluwer Academic Publishers, Boston. 173 p.
Rosenberg, N.J. and Crosson, PR. 1991. Processes for Identifying Regional Influences of and Responses to Increasing Atmospheric CO2 and Climate Change: the MINK Project, An Overview. DOE/RL/01830T-H5. Resources for the Future and US Department of Energy, Washington DC. 35 p.
Rosenzweig, C. 1985. Potential CO2-induced climate effects on North American wheat-producing regions. Climatic Change 7: 367-389.
Rosenzweig, C. and Parry, M.L. 1994. Potential impacts of climate change on world food supply. Nature 367: 133-138.
Rosenzweig, C., Curry, B., Richie, J.T., Jones, J.W., Chou, T.Y., Goldberg, R. and Iglesias, A. 1994. The effects of potential climate change on simulated grain crops in the United States. In: Implications of Climate Change for International Agriculture: Crop Modelling Study. C. Rosenzweig and A. Iglesias (eds.). US Environmental Protection Agency, Washington DC. pp. 1-24.
Rötter, R., Stol, W., van de Geijn, S.C. and van Keulen, H. 1995. World Agro-climates. I. Current Temperature Zones and Drylands. AB-DLO, Research Institute for Agrobiology and Soil Fertility, Wageningen. 55 p.
Rowntree, P.R. 1990a. Predicted climate changes under 'greenhouse-gas' warming. In: Climatic Change and Plant Genetic Resources. M. Jackson, B.V. Ford-Lloyd and M.L. Parry (eds.). Belhaven Press, London. pp. 18-33.
Rowntree, PR. 1990b. Estimates of future climatic change over Britain. Part 2. Results. Weather 45: 79-89.
Schertz, D.L. 1988. Conservation tillage: an analysis of acreage projections in the United States. J. Soil Water Conser. May-June: 35-42.
Seino, H. 1993. Impacts of climatic warming on Japanese agriculture. In: The Potential Effects of Climate Change in Japan. N. Shuzo, H. Hideo, H. Hirokazu, O. Toshiichi and M. Tsuneyuki (eds.). Environment Agency of Japan, Tokyo. pp. 15-35.
Sen, A. 1981. Poverty and Famines: An Essay on Entitlement and Deprivation. Oxford University Press, London.
Sen, A. 1993. The economics of life and death. Scientific American (May) 268 (5): 40-47.
Singh, U. 1994. Potential climate change impacts on the agricultural systems of the small island nations of the Pacific. Draft, Los Baños, Phillipines. IFDC-IRRI.
Smit, B. (ed.). 1993. Adaptation to Climatic Variability and Change. Occasional Paper No. 19. University of Guelph, Guelph, Canada. 53 p.
Thompson, T.P. and Wan, X. 1992. The Socioeconomic Dimensions of Agricultural Production in Albania: A National Survey. PN-ABQ-691. Intnl. Fert. Dev. Center, US AID, Washington DC.
Tobey, J., Reilly, J. and Kane, S. 1992. Economic implications of global climate change for world agriculture. Journal of Agricultural and Resource Economics 17: 195-204.
Turner, H.A. and Anderson, C.L. 1980. Planning for an Irrigation System. American Association for Vocational Instructional Materials, Athens, GA.
Umali, D.L. 1993. Irrigation-induced salinity: a growing problem for development and the environment. World Bank Tech. Paper No. 215. World Bank, Washington DC.
US National Academy of Sciences [US NAS]. 1991. Report on Adaptation to Climate Change. Committee on Science, Engineering, and Public Policy, National Academy Press, Washington DC.
US OTA [US Congress, Office of Technology Assessment]. 1993. Preparing for an Uncertain Climate. Vol. 1. OTA-0-567. US Government Printing Office, Washington DC.
Vaux, H.J. 1990. The changing economics of agricultural water use. Visions of the Future: Proceedings of the 3rd National Irrigation Symposium. ASAE, St. Joseph, Michigan. pp. 8-12.
Vaux, H.J. 1991. Global climate change and California's water resources. In: Global Climate Change and California. J.B. Knox and A.F. Scheuring (eds.). University of California Press, Berkeley. pp. 69-96.
Wang, Y.P., Handoko, J.R. and Rimmington, G.M. 1992. Sensitivity of wheat growth to increased air temperature for different scenarios of ambient CO2 concentration and rainfall in Victoria, Australia - a simulation study. Climatic Research 2: 131-149.
Woodward, F.I. 1993. Leaf responses to the environment and extrapolation to larger scales. In: Vegetation Dynamics and Global Change. A. Solomon and H. Shugart (eds.). International Institute for Applied Systems Analysis, Chapman Hall, New York. pp. 71-100.
World Bank. 1993. World Development Report. Washington DC.
World Bank. 1994. A Review of World Bank Experience in Irrigation. Report 13676. World Bank Operations Department, Washington DC.