Working Group II: Impacts, Adaptation and Vulnerability

Other reports in this collection Historical Analogs of Adaptation

Figure 5-2: Ranges of percentage changes in crop yields (expressed in vertical extent of the vertical bars only) spanning selected climate change scenarios—with and without agronomic adaptation—from paired studies in Table 5-4. Each pair of ranges is differentiated by geographic location and crop. Pairs of vertical bars represent the range of percentage changes with and without adaptation. Endpoints of each range represent collective high and low percentage change values derived from all climate scenarios used in the study. Horizontal extent of the bars is not meaningful. Note that on the x-axis the last name of the lead author is listed as it appears on Table 5-4. See Table 5-4 for details on climate scenarios used and types of adaptation strategies modeled in each study. In the case of Mongolia, adaptation was not modeled when climate change resulted in positive yield change. In Romania, earlier planting of irrigated maize results in slightly lower yields than with current planting dates.

The agriculture sector historically has shown enormous capacity to adjust to social and environmental stimuli that are analogous to climate stimuli. Historical analogs of the adaptability of agriculture to climate change include experience with historical climate fluctuations, deliberate translocation of crops across different agroclimatic zones, rapid substitution of new crops for old ones, and resource substitutions induced by scarcity (Easterling, 1996). In the Argentine Pampas, the proportion of land allocated to crops has increased markedly at the expense of grazing land during historic humid periods, and vice versa during dry periods (Viglizzo et al., 1997). Historical expansion of hard red winter wheat across thermal and moisture gradients of the U.S. Great Plains provides an example of crop translocation (Rosenberg, 1982). At present, the northern boundary of winter wheat in China is just south of the Great Wall and the north edge of China, where large temperature increases are expected under climate change (Lin, 1997). Winter wheat planting has shifted from Dalian (38º54'N) to Shenyang (41º46'N) in Liaoning province. The shift was aided by introduction of freeze-resistant winter wheat varieties from high-latitude countries such as Russia, the United States, and Canada into Liaoning province (Hou, 1994, 1995; Chen and Libai, 1997). Rapid introduction of canola in Canadian agriculture in the 1950s and 1960s shows how rapidly farmers can modify their production systems to accommodate a new crop (National Research Council, 1991). Adaptation to declining groundwater tables by substituting dryland for irrigated crops in regions of the U.S. Great Plains is an example of substitutions to deal with water becoming a scarce production resource (Glantz and Ausubel, 1988). None of these examples, however, deals specifically with an evolving climate change and all are historic—which limits confidence in extending their conclusions to future climate change. Agronomic Adaptation of Yields

Increasing numbers of studies have investigated the effectiveness of agronomic adaptation strategies (e.g., adjustments in planting dates, fertilization rates, irrigation applications, cultivar traits) in coping with climate-induced yield losses and gains since the SAR (see Table 5-4). Considerable costs could be involved in this process, however—for example, in learning about and gaining experience with different crops or if irrigation becomes necessary. In some cases, a lack of water resulting from climate change might mean that increased irrigation demands cannot be met (see Section 4.7.2).

Methodologically, there has been little progress since the SAR in modeling agronomic adaptations. On one hand, the adaptation strategies being modeled are limited to a small subset of a much larger universe of possibilities, which may underestimate adaptive capacity. On the other hand, the adaptations tend to be implemented unrealistically, as though farmers are perfectly clairvoyant about evolving climate changes, which may inflate their effectiveness (Schneider et al., 2000). Some studies find agronomic adaptation to be most effective in mid-latitude developed regions and least effective in low-latitude developing regions (Rosenzweig and Iglesias, 1998; Parry et al., 1999). This finding clearly is supported across the studies summarized in Table 5-4, although the number of studies that include adaptation is not large. A small number of studies in Table 5-4 compare yield changes with and without agronomic adaptation. Percentage changes in yields across a range of climate change scenarios for those studies are shown in Figure 5-2. Each pair of vertical bars represents the range of percentage changes by crop, with and without adaptation, for each study. Clearly, adaptation ameliorates yield loss (and enhances yield gains) in most instances. The median adapted yields (mid-point of the vertical bars) shift upward relative to the median unadapted yields in six of the eight studies. Two studies do not show such an upward shift (Mongolia, Romania) because of peculiarities in the modeling (see figure caption). Adaptation ameliorates the worst yield losses in seven of the eight studies.

It is important to note, however, that differences in modeling methodology and aggregation of results often lead to conflicting conclusions in specific regions. For example, in two studies that used the same GCM scenarios, Matthews et al. (1997) simulate large increases, whereas Winters et al. (1999) simulate large decreases in rice yield with adaptation across several countries in Asia (see Table 5-4c). Hence, confidence in these simulations is low.

Important work has investigated the geographic distribution of crop potential under climate change. Carter and Saarikko (1996) used crop models to demonstrate a poleward shift in Finnish potential cereal cultivation by 100-150 km for each 1°C increase in mean annual temperature. Reyenga et al. (1999) indicate that climate change accompanied by a doubling of CO2 is likely to enable expansion of existing Finnish wheat-growing areas into dry margins, considerably extending potential cropping areas. This expansion is moderated by dry conditions but not effectively enhanced by wetter conditions.

Box 5-5. Extending Uncertainty in Crop Models to Uncertainty in Economic Analysis

Several factors contribute uncertainty to modeling of impacts of climate change on agricultural systems (Parry et al., 1999). Crop modeling studies invariably highlight the need to develop confidence that the outputs are not "model-dependent." There is uncertainty because of the fact that yield estimates obtained in climate impact assessments vary from one impact model to another. In some cases, simulations across models may exhibit good agreement. Comparison of rice models showed that their predictions for potential production were quite close to observed values (Peng et al., 1995). The Erosion Productivity Impact Calculator (EPIC), a generalized crop model, predicts observed yields most closely during years with extreme warmth, lending confidence to its ability to predict yields under climate change conditions (Easterling et al., 1996). In other situations, agreement is not as good. In comparisons between wheat simulation models (Goudriaan et al., 1994; Wolf et al., 1996), grain yield predictions were markedly different between models. Such findings have stimulated work to compare the performance of different models and to analyze the underlying reasons for differences (Gregory et al., 1999; Mearns et al., 1999).

Spatial resolution of crop models is another important source of uncertainty in crop models. Crop models simulate processes that regulate growth and development at fine scales (a few kilometers), whereas climate change scenarios that drive crop models typically are produced by climate models operating at coarse scales (1,000 km or more) (Barrow and Semenov, 1995; Easterling et al., 1998). Studies that use statistical downscaling techniques and nested limited area numerical models to increase the resolution of GCM scenarios of climate change have shown large simulated yield discrepancies between coarse-resolution (GCM) and fine-resolution (downscaled) climate change scenarios. Mearns et al. (1999) demonstrate for a site in Iowa that yields can change algebraic sign depending on the resolution of the climate change scenario: Maize yield decreases (-11%) with low-resolution climate change (CSIRO model) and increases (2%) with high-resolution climate change (RegCM limited area model nested within CSIRO model). Causes of variance in model output must be identified. Adaptation of Livestock

Significant costs can be incurred to counter the effects of climate change on animal production; moreover, the impact of a warmer climate in terms of costs is not linear: Larger changes in climate can increase costs exponentially (Hahn and Morgan, 1999). Possible benefits of climate change during cooler seasons are not well documented, but the benefits are likely to be less than the consequential negative hot weather impacts (Hahn et al., 1992). The ability of livestock managers to cope with climate is demonstrated daily in their coping with normally varying conditions. A variety of management adaptations are available for livestock production systems. For example, Hahn and Mader (1997) outline a series of proactive management countermeasures that can be taken during heat waves (e.g., shades and/or sprinklers) to reduce excessive heat loads. Historical success in coping with climate variability suggests that livestock producers are likely to adjust to climate change successfully. Johnson (1965) provides examples from advances in genetics and breeding as related to the environment. These capabilities should allow adaptation to changing, less favorable circumstances associated with projected rates of climate change. However, coping can entail significant dislocation costs for certain producers. For individual producers, uncertainties associated with potential climate change imply additional risks related to how and when to adapt current production practices (Lewandrowski and Schimmelpfennig, 1999). Confidence in the foregoing projections of the ability of livestock producers to adapt their herds to the physiological stresses of climate change is difficult to judge. The general lack of simulations of livestock adaptation to climate change is problematic. This absence of a well-developed livestock counterpart to crop modeling of adaptation assessments suggests a major methodological weakness. Hence, we give only low to moderate confidence in projections of successful livestock adaptability.

height="1" vspace="12">

Other reports in this collection

IPCC Homepage