The predictive accuracy of climate models is quite low at a regional scale—the very scale where climatic factors become ecologically relevant. Until more precise regional estimates of changes in rainfall, temperature, seasonality, and weather disturbance frequencies are available, only general impact assessments are possible. But even when more detailed regional information becomes available, assessments will be hindered by the lack of information on the response of individual species to changes in climate. Because communities are not deterministic equilibrium assemblages of species, any climatic changes will have different consequences for different species. For tree species, which are some of the most structurally significant species in many North American communities, we do have relatively good paleoecological information on their rates of dispersal in response to past climate changes. Dynamic forest models also allow some prediction of how changes in climatic factors will alter forest community composition (Botkin 1990). For most other species, however, changes in response to climate shifts cannot be accurately predicted.

Moreover, the effect of changing distributions of species on various ecosystem services is even less certain. For example, changes in species distribution as a result of climatic changes will alter patterns of disease incidence. Such an effect can be seen in evidence that links the outbreak of cholera in South America in 1991 to El Ni ño (Epstein et al. 1993, Stone 1995). The warming of the waters off the coast of South America may have stimulated growth of a plankton harboring the cholera bacterium. While the increased frequency of El Ni ño in recent years cannot be conclusively tied to human-caused changes in climate, the example demonstrates how ocean current changes that are likely to occur in the event of global warming could have substantial effects on human health. Current models of changes in the distribution of disease vectors under likely future climates suggest that developing countries will see an increase in malaria, schistosomiasis, sleeping sickness, dengue, and yellow fever. The outbreak of the hantavirus in the southwestern U.S. has also been linked to weather conditions, possibly also associated with El Ni ño.

Introductions of new cultivated varieties of crops (developed either through traditional breeding methods or through genetic engineering) into natural or agroecosystems can pose significant risks, such as the introgression of genes into wild populations, weediness, and pathogenicity. Genetic material from introduced plants, animals, and micro-organism can be transferred into wild populations of related species through the formation of fertile hybrids. The new genetic material can then potentially alter the ecological interactions of that wild relative. For example, a disease or frost resistance gene transferred into a wild weedy relative of the crop could extend the range of that wild relative. To date, most gene flow in agricultural systems has taken place between crops and their weedy relatives, but in aquatic systems, substantial gene flow occurs between hatchery-reared fish and wild populations.