Very little work has investigated prospects for natural adaptation of crop
species to climate change, and the results of the few studies that do have been
inconclusive. However, there appears to be a wide range of resistance to high-temperature
stress within and among crop species. For example, moderately large genetic
variation in the tolerance to high-temperature induced spikelet sterility has
been reported among and between indica- and japonica-type rice genotypes (Matsui
et al., 1997). Some rice cultivars have the ability to flower early in the morning,
thereby potentially avoiding the damaging effects of higher temperatures later
in the day (Imaki et al., 1987).
Prospects for managed genetic modification appear to be more optimistic than
for natural adaptation. Intraspecific variation in seed yield of soybean in
response to elevated CO2 was observed by Ziska et al. (1998). Differences
in carbon partitioning among soybean cultivars may influence reproductive capacity
and fecundity as atmospheric CO2 increases, with subsequent consequences
for future agricultural breeding strategies (Ziska et al., 1998). However, no
significant intraspecific variability in responses to elevated CO2
was detected in studies with wheat and temperate forage species (Lüscher
and Nösberger, 1997; Batts et al., 1998). To promote adaptation to an environment
of high CO2 and high temperature, plant breeders have suggested selection
of cultivars that exhibit heat tolerance during reproductive development, high
harvest index, small leaves, and low leaf area per unit ground (to reduce heat
load) (Hall and Allen, 1993). However, prospects to improve adaptation of crop
species to elevated CO2 remain very uncertain, and more research
in this direction is required.
The number of studies that model the yield impacts of climate change (with
and without CO2 direct effects and with and without adaptation) across
individual sites in regions has continued to grow since the SAR. Of particular
note is the expansion of studies that explicitly model the effects of change
in climate variability and means simultaneously versus change in climate means
only (Southworth et al., 1999), use transient climate change scenarios,
and report modeling of agronomic and socioeconomic adaptation. A selection of
major global and regional model-based studies reported since the SAR is summarized
in Table 5-4.
Table 5-4 yields are reported as ranges of percentage change over the climate change scenarios, modeling sites, and crop as noted. Thirteen of the yield rangeswithout adaptationare from studies of tropical crops. Of the 13 ranges, 10 encompass changes that are exclusively lower than current yields. In three ranges, a portion of the range is approximately no different from current yields or slightly above. In the tropics, most crops are at or near theoretical temperature optimums, and any additional warming is deleterious to yields. Thirty ranges of percentage changes in temperate crop yields also appear in Table 5-4. Of these 30 ranges, six encompass changes that are exclusively higher than current yields. In another seven, half or more of the changes were more than current yields. In yet another seven, less than half of the changes extended above current yields. The remaining 10 ranges encompassed changes that were exclusively less than current yields. Hence, in two-thirds of the cases, temperate crop yields benefited at least some of the time from climate change.
New work on climate change scenarios (Mitchell et al., 2000) generated with stabilized radiative forcing at 550 and 750 ppm equivalent-CO2 and unstabilized radiative forcing (i.e., unmitigated emissions) in the HadCM2 model simulated major cereal yield response globally in 2080 (Arnell et al., 2001). The pattern of yield changes with unstabilized forcing duplicates the pattern described above: Generally positive changes at mid- and high latitudes overshadowed by reductions in yields at low latitudes. Stablization at 550 ppm ameliorates yield reductions everywhere, although substantial reductions persist in many low-latitude countries. Stabilization at 750 ppm produces a pattern of yield response that is intermediate relative to the 550 ppm and unstabilized forcing scenarios, with anomalous yield increases in mid-latitudes relative to 550 ppm as a result of interactions between atmospheric CO2, temperature, and moisture. More studies are needed before confidence levels can be assigned to understanding of the agricultural consequences of stabilization, although this work is an important step.
In all agricultural regions, the effects of natural climate variability are
likely to interact with human-induced climate change to determine the magnitude
of impacts on agricultural production. Some analyses postulate an increase in
weather variability (Mearns et al., 1992, 1995; Rosenzweig et al.,
2000); simulations of wheat growth indicate that greater interannual variation
of temperature reduces average grain yield (Semenov and Porter, 1995). Hulme
et al. (1999) simulated natural climate variability in a multi-century
control climate for comparison with changed variability in a set of transient
climate change simulations. Wheat yields were simulated with control and climate
change scenarios. For some regions, the impacts of climate change on wheat yields
were undetectable relative to the yield impacts of the natural variability of
the control climate (see Table 5-4). Greater efforts
to take account of the "noise" of natural climate variability are indicated
(Semenov et al., 1996).
The importance of diurnal climate variability has emerged since the SAR (Reilly et al., 1996). Cold temperatures presently limit the yield of rice in all temperate rice-growing regions. Jacobs and Pearson (1999) provide new field results on irreversible effects and retardation (but recoverable) impacts of cold temperatures on various physiological processes in rice. On the other hand, rice spikelet sterility above 35°C at flowering (usually during daytime) puts rice at risk from increased daily maximum temperatures (Horie et al., 1996). In light of recent observed rises in temperatures that are larger for daily minima than daily maxima (Easterling et al., 1997), Dhakhwa et al. (1997) and Dhakhwa and Campbell (1998) conclude that, compared to equal day-night warming, differential warming leads to less water loss through evapotranspiration and better WUE. This is likely to lead to enhanced photosynthesis, crop growth, and yieldalthough at a possible loss of nutritional quality (Murray, 1997). Possible reduction of frost incidence is not normally considered in these studies. On the negative side, higher nighttime temperatures could extend the overwintering range for some insect pests and broaden the range of other temperature-sensitive pathogens.
Substantial progress has been made in development of transient (time-evolving) scenarios of climate change for use in agricultural impact assessment. An important question arises regarding whether 2 years with exactly the same climate, one produced by a transient scenario and the other by an equilibrium scenario, would give different production system responses. Many crop models contain cumulative functions that retain environmental information over several years (e.g., water balance, soil nutrients). This factor alone could account for substantial yield response differences between transient and equilibrium climate change scenarios. Only a few studies deliberately have compared simulated yields with transient and equilibrium climate change scenarios. Using the UKHIV equilibrium scenario with increased interannual variability at Rothamsted, Semenov et al. (1996) simulate a loss of wheat yield relative to current with two crop models and no change with a third. With the UKTR transient scenario, all three models show yield increases relative to current climate. The U.S. Country Studies Program (Smith et al., 1996a) used the Clouds and Earth's Radiant Energy System (CERES) model to simulate larger average increases in winter wheat across Kazakhstan with the GFDL transient climate change scenario (for the 10th decade) (+21% winter wheat yield) than the GFDL equilibrium scenario (+17% winter wheat yield). Spring wheat yields decreased with both scenarios; again, however, yields simulated with the transient climate change were not as adversely affected as those simulated with the equilibrium climate change. Rosenzweig and Iglesias (1998) also found that wheat, maize, and soybean yields are less adversely affected by transient climate change than equilibrium climate change. Lack of consistency in application of transient climate change scenarios to impact modeling between studies results in competing explanations about differences in impact estimates between the two types of climate change scenarios.
Other reports in this collection