Global Warming and Nonlinear Growth: How Important are Changes in Average Temperature?

Abstract

Increasing concentrations of greenhouse gases, particularly CO 2 , in the lower atmosphere have led to concern about global changes in temperature and precipitation patterns. It has been estimated that mean surface air temperature will rise at a rate of ∼0.3-0.4 C per decade because of the increased greenhouse effect (2,10). These projections are based on outputs from General Circulation Models (GCMs). GCMs are coupled ocean-atmosphere models that simulate the transfer of heat, mass, and momentum in the lower atmosphere. They are reliable tools for simulating climate on large temporal and spatial scales, but they have two important drawbacks that may limit their usefulness for estimating the potential impact of global warming on biological processes. First, impact assessments based on GCMs so far have focused on changes in average temperature, although changes in climate variability may be equally or more important (9,12). A theoretical study by Katz and Brown (9) showed that changes in climate are more closely associated with shifts in the frequency of meteorological extremes than with shifts in the mean. Furthermore, global warming may be associated with an asymmetric diurnal temperature increase, by which daily minima increase while daily maxima remain unchanged (7,8,12). Therefore, impact assessments that rely on scenarios involving only shifts in the mean may be misleading. Second, the low temporal and spatial resolution of GCM outputs makes it difficult to link projections of global warming to biological response models, such as crop growth or plant disease models, which require daily or subdaily data as input (3,16,19). Bonan (3) reviewed several studies that contained sensitivity analyses of ecosystem models to climate change and/or global warming. Results from models based on simple, semi-empirical parameterizations of growth at monthly or annual resolutions differed substantially from more detailed models that explicitly accounted for biophysical and physiological processes on daily or subdaily time scales. As recent advancements in computer modeling make it possible to scale down from monthly averaged GCM outputs to daily or hourly data with stochastic weather generators (16,19), the use of more detailed (and presumably more accurate) biological models for impact assessment of global warming will probably increase. The computational demands will be high, however, given the long-term scope of such studies (3). Interactions between plant pathogens and their hosts occur on subdaily time scales. For example, spores of many fungal pathogens can germinate and infect in less than 12 h. Furthermore, growth and developmental responses of plant pathogens (e.g., to temperature) are often nonlinear (1,15). Nonlinearity may lead to biased results if averaged data are used for estimating responses to changes in meteorological conditions (15). This bias may be amplified by the projected asymmetric increase of daily minimum and maximum temperatures. Therefore, to arrive at realistic predictions about the potential impact of global warming on plant diseases, it may be necessary to specify projected changes in climate to a resolution that is compatible with the time scale of the underlying biological processes. Thus, projections with subdaily resolution may be needed to accurately model the potential impact of global warming on plant pathosystems. The objective of this communication is to direct attention to problems resulting from the use of averaged data as input for biological models when response functions are nonlinear