The primary application of gradient models has been in analyzing the relationship between the distribution of species and their environmental setting to create habitat models. Habitat models aim to predict where organisms, populations, or communities occur based on the distribution of appropriate biophysical conditions (habitat selection and habitat suitability preferences). Like gradient models of environmental patterns, predictive habitat models are typically static in nature and rely heavily on the statistical relationships between variables of interest that are hard to measure (here the geographical distribution of species) and more easily measured environmental proxy variables. A variety of statistical methods have been developed for evaluating these relationships, including regression models in various forms, classification trees (Classification tree), ordination techniques (Ordination), and Bayesian models (Bayesian network).
Many practical applications of habitat models involve their use to generate predictive maps. An example of this approach and its implications is found in the arena of global change science, where several studies have predicted the spatial displacement of species along elevation gradients in response to a warming climate (global change - impact on the biosphere). In light of these predictions, some have argued for specific management responses, such as the design of networks of nature reserves that incorporate elevation and latitudinal gradients to allow for future species migration. The use of these models to inform management is complicated by (1) their reliance upon indirect explanatory variables such as elevation rather than factors that are more directly responsible for species distributions and (2) their focus on static relationships rather than the dynamic responses of organisms to environmental change.
Was this article helpful?