Model-based assessments of climate change effects on forests: a critical review

While current projections of future climate change associated with increases in atmospheric greenhouse gases have a high degree of uncertainty, the potential effects of climate change on forests are of increasing concern. A number of studies based on forest simulation models predict substantial alteration of forest composition, forest dieback, or even loss of forest cover in response to increased temperatures associated with increasing atmospheric carbon dioxide concentrations. However, the structure of these computer models may cause them to overemphasize the role of climate in controlling tree growth and mortality. Model functions that represent the influence of climate on tree growth are based on the geographic range limits of a species, predicting maximal growth in the center of the range and zero growth (100% mortality) at the range limits and beyond. This modeling approach ignores the fact that the geographic range of a species reflects the influence of both climate and other environmental factors, including competition with other tree species, soil characteristics, barriers to dispersal, and distributions of pests and pathogens. These climate-response functions in forest simulation models implicitly assume that tree species occur in all environments where it !s possible for them to survive (their fundamental niche or potential habitat) and that these potential habitats are entirely defined by climate. Hence, any alteration of climate must result in a fairly rapid decline of species near their range limits and rapid alteration of forest composition and structure. The climate-response functions that lead to these unrealistic conclusions have no basis in plant physiology or actual measurements of tree responses to climate stressors. Rather, these functions were chosen as a necessary expedient for modeling the climatic responses of many tree species for which there were limited or no ecophysiological data. There is substantial evidence, however, that some tree species can survive, and even thrive, in climatic conditions outside their present range limits. This evidence suggests that nonclimatic factors exclude some species from natural forests beyond their present range limits and that climate may not be the only determinant of these limits. Hence, there is reason to suspect that published projections of forest responses to climate change based on forest simulation models may exaggerate the direct impact of climate on tree growth and mortality. We propose that forest simulation models be reformulated with more realistic representations of growth responses to temperature, moisture, mortality, and dispersal. We believe that only when these models more accurately reflect the physiological bases of the responses of tree species to climate variables can they be used to simulate responses of forests to rapid changes in climate. We argue that direct forest responses to climate change projected by such a reformulated model may be less traumatic and more gradual than those projected by current models. However, the indirect effects of climate * Corresponding author. Fax: (+ 1-708) 252-5498. 0304-3800/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved SSDI 0304-3 800(95)00138-7 2 C. Loehle, D. LeBlanc/Ecological Modelling 90 (1996) 1-31 change on forests, mediated by alterations of disturbance regimes or the actions of pests and pathogens, may accelerate climate-induced change in forests, and they deserve further study and inclusion within forest simulation models.