7. Agriculture 7.1 Overview
7.1.1 Context of National Communications Sustainable development includes social, economic and environmental dimensions. Climate change modifies all these dimensions and therefore alters the potential development pathways. In particular, the effects of climate change on agriculture will determine the future of food security and ultimately influence the inequitable North–South divide.
According to the IPCC TAR (McCarthy et al., 2001), climate change is already happening and will continue to happen even if global GHG emissions are curtailed. Many studies document the implications of climate change for agriculture and pose a reasonable concern that climate change is a threat to sustainable development, especially in countries not included in Annex I to the Convention. Identifying which regions and populations are at greatest risk from climate change (i.e., are most vulnerable) can help in setting priorities for adaptation. This chapter focuses on the methods for making these assessments, and includes examples of applications in developing countries and an overview of existing knowledge on the subject. The merits of each approach vary according to the level of impact being studied, and approaches may frequently be mutually supportive. For example, simple agroclimatic indices often provide the necessary information on how crops respond to varying rainfall and temperature in wide geographical areas; crop-specific models are used to test alternative management that can in turn be used as a component of an economic model that analyses regional vulnerability or national adaptation strategies. Therefore, a mix of approaches is often most productive.
7.1.2 Effects of current climate variability Climate is an essential component of the natural capital. In many regions of the world, such as Africa, Southern and Central America, and South and South-East Asia, climates are extremely variable from year to year, and recurrent drought and flood problems often affect entire countries over multi-year periods. These often result in serious social problems. For example, the persistent drying trend in parts of Africa over the past decades has affected food production, including freshwater fisheries, industrial and domestic water supplies, and hydropower generation (Benson and Clay, 1998, 2000).
Agriculture is strongly dependent on water resources and climatic conditions, particularly in the regions of the world that are particularly sensitive to climatic hazards, such as Africa, South and Central America, and Asia. Some countries in these regions, where economic and social situations are often unstable, are extremely vulnerable to changes in environmental factors. It is especially the case in countries where technological buffering to droughts and floods is less advanced, and where the main physical factors affecting production (soils, terrain and climate) are less suited to farming. Crop production is consequently extremely sensitive to large year-to-year weather fluctuations. Crop diseases or pest infestations are also weather dependent and tend to cause more damage in countries with lower technological levels.
7.1.3 Drivers of agricultural response to climate change
Estimation of future agricultural responses to climate change is usually based on scenarios. It is
crucial to understand that there is large uncertainty in the climate scenarios. The scenarios are
essential for evaluating possible future conditions, but they do not necessarily describe the
conditions that will actually occur. Nevertheless, conditions similar to the scenarios are
possible, and as such they should be used to explore possible adaptive measures.
Agriculture is a complex sector involving different driving parameters (environmental,
economic and social). It is now well recognized that crop production is very sensitive to
climate change (McCarthy et al., 2001), with different effects according to region. The IPCC
analysis of climate change impacts (TAR) estimates a general reduction of potential crop yields
and a decrease in water availability for agriculture and populations in many parts of the
developing world (Table 7.1).
The main drivers of agricultural responses to climate change are biophysical effects (Table 7.2)
and socio-economic factors (Table 7.3). Crop production is affected biophysically by changing
meteorological variables, including rising temperatures, changing precipitation regimes and
increasing levels of atmospheric carbon dioxide. Biophysical effects of climate change on
agricultural production depend on the region and the agricultural system, and the effects vary
Socio-economic factors influence responses to changes in crop productivity, with price changes
and shifts in comparative advantage. The final response depends on the adaptation strategies in
each region and agricultural system. The combination of biophysical and socio-economic
effects can result in:
? Changes in the mix of crops grown, and hence in the type of farming and rural land use
? Changes in production, farm income, and rural employment
? Changes in rural income, contribution to national GDP, and agricultural export earnings.
Table 7.1. Climate change and related factors relevant to agricultural production and food security
Consequences and factors that interact with Climate factor Direction of change agricultural production and food security Sea level rise Increase Sea level intrusion in coastal (agricultural) areas and
salinization of water supply Precipitation Intensified hydrological cycle, so Changed patterns of erosion and accretion; changed
intensity/run-off generally increases, but with storm impacts; changed occurrence of storm flooding
regional variations and storm damage, water logging, increase in pests Heat stress Increases in heat waves Damage to grain formation, increase in some pests Drought Poorly known, but significant Crop failure, yield decrease; competition for water
increased temporal and spatial
Atmospheric CO Increase Increased crop productivity but also increased weed 2productivity and therefore competition with crops Source: Data and information from Parry et al. (1998a, 2004), McCarthy et al. (2001).
Table 7.2. Characterization of agronomic impacts, adaptive capacity and sector outcomes Biophysical Uncertainty Expected intensity Adaptive Socio-economic and other impact level of negative effects capacity secondary impacts Changes in crop Medium High for some Moderate Changes in optimal farming systems;
growth conditions crops and regions to high relocation of farm processing industry;
increased economic risk; loss of rural
income; pollution due to nutrient
leaching; biodiversity decrease Changes in High Medium High for Changes in optimal farming systems;
optimal conditions intensive loss of rural income for livestock production
Changes in Medium High for Moderate Increased demand for irrigation;
precipitation and to low developing decreased yield of crops; increased risk
the availability of countries of soil salinization; increased water
water resources shortage; loss of rural income Changes in High to Medium Moderate Pollution due to increased use of
agricultural pests very high to high pesticides; decreased yield and quality of
crops; increased economic risk; loss of
rural income Changes in soil Medium High for Moderate Pollution by nutrient leaching;
fertility and developing biodiversity decrease; decreased yield of
erosion countries crops; land abandonment; increased risk
of desertification; loss of rural income
Table 7.3. Characterization of aggregated farming system impacts, adaptive capacity and
Socio-economic Uncertainty Expected intensity adaptation (private
impact level of negative effects coping capacity) Other impacts Changes in High High for areas where Moderate Changes in crop and livestock optimal farming current optimal production activities; relocation of systems farming systems are farm processing industry; loss of
extensive rural income; pollution due to
nutrient leaching; biodiversity
decrease Relocation of High High for some food Moderate Loss of rural income; loss of farm processing industries requiring cultural heritage industry large infrastructure
or local labour
Increased Medium High for crops Low Loss of rural income (economic) risk cultivated near their
Loss of rural High (Not characterized) Moderate Land abandonment; increased risk income and of desertification; welfare decrease cultural heritage in rural societies; migration to
urban areas; biodiversity decrease
7.1.4 Previous studies
Several hundred studies on climate change as it relates to agriculture have been completed. They provide a first indication of the impact types to expect, and thus the most effective analysis methods to implement. Potential impacts on world food supply have been estimated for several climate change and socio-economic scenarios (Figure 7.1). Some regions may enjoy improved agricultural production, whereas others will suffer from yield losses, and so a reorganization of agricultural production areas may be required. In any given region, crops are expected to be affected differently, leading to the need for adaptation in related support industries and markets, farm-level strategies and rural development schemes.
Although Figure 7.1 shows that global production appears stable (Parry et al., 2004 provide additional quantitative data), regional differences in crop production are likely to grow through time, leading to a significant polarization of effects with substantial increases in prices and risk of hunger in the poorer nations. The most serious effects are at the margins – in vulnerable
regions and among vulnerable groups. Individuals particularly vulnerable to environmental change are those with relatively high exposures to changes, high sensitivity to changes, low coping and adaptive capacities, and low resilience and recovery potential. Adaptation is necessary, but adaptation has limits (technological, biotechnological, political, cultural).
Figure 7.1. Percentage change in average crop yields according to the HadCM2 climate
Source: Parry et al., 2004.
7.2 Methods and Tools
7.2.1 General considerations
The methods for assessing climate impacts in crop production and evaluation of adaptation
strategies are extensively developed and used widely by scientists, extension services,
commercial farmers and resource managers. A major challenge facing all agriculture–climate
evaluations is the analysis of important biophysical and socio-economic impacts, because these
must be derived from complex interactions among biophysical and socio-economic systems that
are difficult to model. The tools presented in this chapter are adequate to be used with modified
mean climate conditions. To evaluate changes in the frequency and intensity of extreme events, Page 7-5 such as droughts or floods, it is important to include a combination of empirical yield responses
based on statistical data and modelling approaches. In all cases, the challenge in interpreting the results originates in the use of uncertain climate change scenarios.
A number of approaches to the assessment of the impacts of climate change on agriculture have been developed from the many studies conducted to date. Approaches used to assess biophysical impacts include:
? Agroclimatic indices and GIS ? Statistical models and yield functions ? Process-based models. In addition, different tools can be used to examine the socio-economic impacts of climate change. A relatively simple economic forecasting tool, such as that developed by the United States Country Studies Program (Benioff et al., 1996), is often useful. More complex approaches, such as economic regression models, microeconomic and macroeconomic models, farm models, and household and village models, can also be used.
Each of these methods yields information on different types of impacts (Table 7.4). For example, simple agroclimatic indices can be used to analyse large-area shifts of cropping zones, whereas process-based crop growth models should be used to analyse changes in crop yields. Effects on income, livelihoods and employment are assessed using economic and social forms of analysis.
In addition, studies can be undertaken using a regional approach or a site-specific approach. In a regional approach, several existing simple tools can be applied and tested under a range of
conditions in a given region and the results visualized on maps. This simple regional approach is essential for integrating climate change, crop production, water demand indices and socio-economic indices on a regional scale, thus providing a first-order evaluating tool to analyse possible adaptation strategies.
Table 7.4. Summary of the characteristics of the main agricultural models Type of model Description and use Strengths Weaknesses Agroclimatic Based on combinations of Simple calculation. Climate based only, lack indices and GIS climate factors important for Effective for comparing management responses or
crops. across regions or crops. consideration of carbon
Used in many agricultural fertilization.
Useful for general audiences.
Statistical Based on the empirical Present-day crop and Do not explain causal models and relationship between climatic variations are well mechanisms. yield functions observed climate and crop described. May not capture future
responses. climate–crop relationships or
Used in yield prediction for CO fertilization. 2famine early warning and
Process-based Used to calculate crop Process based, widely Require detailed weather and
crop models responses to factors that affect calibrated, and validated. management data for best
growth and yield (i.e., climate, Useful for testing a broad results.
soils, and management). range of adaptations.
Used by many agricultural Test mitigation and
scientists for research and adaptation strategies
Available for most major
Economic tools Used to calculate land Useful for incorporating Not all social systems,
values, commodity prices, financial considerations and households and individuals
and economic outcomes for market-based adaptations. appropriately represented.
farmers and consumers based Climate-induced alterations in
on crop production data. availability of land and water
not always taken into account.
Focus on profit and utility-
Models are complex and
require much data.
Household and Description of coping Useful in semi commercial Not generalizable; do not
village models strategies for current economies. capture future climate
conditions by household and stresses, if different from
village as the unit of current.
A site-specific approach involves local studies that analyse the sensitivity of crop yield, farm
management and water use to climate at the local scale and the implications for policy decisions
that affect water management. Crop models typically focus on optimizing timing of production
and efficiency of nutrient use (primarily nitrogen) and irrigation water.
Because economic sectors vary greatly among different countries and physical environments,
different methods of impact assessment will be appropriate. It is likely that a mix of approaches will lead to the most robust set of results for a given area.
7.2.2 Limitations and sources of uncertainty Climate change scenarios. Climate change scenarios are derived from GCMs driven by
changes in the atmospheric composition of GHG, derived from different storylines in socio-economic scenarios (SRES, see below). A main challenge is, how to interpret the results derived from the climate scenarios. In all regions, uncertainties with respect to the magnitude of the expected changes result in uncertainties in the agricultural evaluations. For example, in some regions projections of rainfall, a key variable for crop production, may be positive or negative depending on the climate scenario used. The uncertainty derived from the climate model is related to the limitation of current models to represent all atmospheric processes and interactions of the climate system. The limitations associated with projecting socio-economic development pathways represent an additional source of uncertainty.
Climate variability. Regional climates naturally fluctuate about the long-term mean. For
example, rainfall variability occurs with regard to timing and quantity, affecting agriculture each year. It is clear that changes have occurred in the past and will continue to occur, and climate change modifies these variability patterns, for example, resulting in more droughts and floods. Nevertheless, there are many uncertainties, especially about rainfall scenarios for the future.
Agricultural models. Agricultural models contain many simple, empirically derived
relationships that do not completely represent actual plant processes. When models are adequately tested against observed data (calibration and validation process), the results represent agricultural output under current climate conditions. Nevertheless, the simplifications of the crop models are a source of uncertainty of the results. For example, agricultural models in general assume that weeds, diseases and insect pests are controlled; that there are no problem soil conditions, such as high salinity or acidity; and that there are no catastrophic weather events, such as heavy storms. The agricultural models simulate the current range of agricultural technologies available around the world; they do not include potential improvements in such technology, but can be used to test the effects of some potential improvements, such as improved varieties and irrigation schedules. A range of agricultural models is used widely by scientists, technical extension services, commercial farmers and resource managers to evaluate agricultural alternatives in a given location under different conditions (i.e., drought years, changes in policy regarding application of agrochemicals, changes in water input, among other conditions).
Effects of CO on crops. CO is a component of plant photosynthesis and therefore influences 22
biomass production. It also regulates the opening of plant stomata and therefore affects plant transpiration. As a result, in theory, plants growing in increased CO conditions will produce 2
more biomass and will consume less water. Experiments in greenhouses confirm this; nevertheless because of the multiple interactions of physiological processes, actual changes are smaller than the theoretical ones. In field conditions, the changes are even smaller. Most of the crop models used for climate change evaluations include an option to simulate the effects of COincrease on crop yield and water use (see Rosenzweig and Iglesias, 1998). It is difficult to 2
validate the crop model results because there are only a very limited number of these experiments worldwide, raising uncertainty about the simulated results.
Issues of scale. Scaling up the vulnerability and adaptation results to a regional level is, as in Page 7-8 most scaling exercises, not an easy task. Ideally, one might use information from farms that are
representative of agriculture in the region, and the degree of their representativeness would
need to be established. More frequently, regional assessments have relied on the input provided
by regional planners and economists as to regional-scale effects, based on local data supplied to
them and discussed by a full range of stakeholders.
Socio-economic projections. The limitations of projecting socio-economic changes affect not
only the SRES scenarios but also the potential adaptive capacity of the system. For example,
uncertainty about the population changes (density, distribution, migration), GDP, and
technology determines and limits the potential adaptation strategies that can be employed.
7.2.3 Combining climate change scenarios with agricultural tools and models Given the uncertainties of the scenarios (magnitude of change and sometimes direction of
change), a good approach is to use several possible scenarios as inputs for the agricultural
models. In addition, using sensitivity scenarios combined with agricultural models (for o example, changes in temperature up to +3C and changes in precipitation from –30% to +30%)
provides an idea of the tolerable thresholds of change for a particular system.
One method shown to be effective for generating climate change scenarios is to study the
changes in the last few decades and then project those changes into the near future. For
example, divide the long-term climate database of one region (or sites) and divide them into
two periods: for example 1930–1960 and 1970–2000. Then study the statistical properties of
each one of those two datasets (means, but also frequency, of dry spells, of storms, probability
of subsequent days with rainfall, etc.). This can be done with “weather generators”. The last
step is to continue (project) the trend observed in all these statistical parameters and create a
synthetic scenario for the near future (e.g., 10–20 years). This method has the advantage that it
is based in observed changes. Of course the projections may be as bad as (or worse than) the
results using the conventional method of GCMs.
Finally, an interesting approach is to use a scenario that occurs within the natural climate
variability of the region, such as a drought scenario. It is essential that the agricultural
evaluations include and test more than one possible scenario and analyse the sensitivity of the
response in the context of the current trends of climate. The use of more than one scenario and
approach results in a span of outcomes, which gives a pertinent notion of uncertainty.
7.2.4 Agroclimatic indices and GIS
Simple agroclimatic indices combined with GIS have been used to provide an initial evaluation
of both global agricultural climate change impacts and shifts in agriculturally suitable areas in
particular regions. The agroclimatic indices are based on simple relationships of crop
suitability or potential to climate (e.g., identifying the temperature thresholds of a given crop or
using accumulated temperature over the growing season to predict crop yields; e.g., Holden,
2001). This type of empirically derived coefficient is especially useful for broad-scale mapping
of areas of potential impact.
When combined with a spatially comprehensive database of climate, crops and GIS, simple
agroclimatic indices are an inexpensive and rapid way of mapping altered crop potential for Page 7-9
quite large areas. Applying agroclimatic indices in Africa (Badini et al., 1997) has provided an
understanding of the relationships between the weather, soils and agricultural production systems and the complexities associated with their variability. Carter and Saarikko (1996) describe basic methods for agroclimatic spatial analysis.
7.2.5 Statistical models and yield functions Complex multivariate models attempt to provide a statistical explanation of observed phenomena by accounting for the most important factors (e.g., predicting crop yields on the basis of temperature, rainfall, sowing date and fertilizer application). A possible weakness in their use for examining the impacts of future climate change, however, is their limited ability to predict effects of climatic events that lie outside the range of present-day variability. Their use has also been criticized because they are based on statistical relationships between factors rather than on an understanding of the important causal mechanisms.
Multiple regression models have been developed to represent process-based yield responses to these environmental and management variables. Yield functions have been used to evaluate the sensitivity and adaptation to climate, e.g., in China (Rosenzweig et al., 1999) and globally (Parry et al., 2004).
7.2.6 Process-based crop models Process-based models use simplified functions to express the interactions between crop growth and the major environmental factors that affect crops (i.e., climate, soils and management), and many have been used in climate impact assessments. Most were developed as tools in agricultural management, particularly for providing information on the optimal amounts of input (such as fertilizers, pesticides and irrigation) and their optimal timing. Dynamic crop models are now available for most of the major crops. In each case, the aim is to predict the response of a given crop to specific climate, soil and management factors governing production.
The ICASA/IBSNAT dynamic crop growth models (International Consortium for Application of Systems Approaches to Agriculture – International Benchmark Sites Network for Agrotechnology
Transfer) are structured as a decision support system to facilitate simulations of crop responses to management (DSSAT). The ICASA/IBSNAT models have been used widely for evaluating climate impacts in agriculture at different levels ranging from individual sites to wide geographic areas (see Rosenzweig and Iglesias, 1994, 1998, for a full description of the method). This type of model structure is particularly useful in evaluating the adaptation of agricultural management to climate change. The DSSAT software includes all ICASA/IBSNAT models with an interface that allows output analysis.
The WOFOST model suite is generic and includes model parameters for certain crops (Supit et al., 1994; Boogaard et al., 1998). There are several versions of the models, which are under continuous development at the University of Wageningen.
The EPIC model (Erosion Productivity Impact Calculator; Sharpley and Williams, 1990) incorporates simplified crop growth functions that respond to climate, environment and management; it has been used in some climate impact assessments.