| Abstract: | Statistical modeling of areas of distribution of species by correlative analysis of the environmental features of known presences has become widespread. However, to a large degree, the logic and the functioning of many of these applications remain obscure, not only due to the fact that some of the modeling methods are intrinsically complex (neural networks, genetic algorithms, generalized additive models, for example), but mainly because the role of other ecological processes affecting the species distributions sometimes is not explicitly stated. Resorting to fundamental principles of population ecology, a scheme of analysis based on separation of three factors affecting species distributions (environment, biotic interactions and movements) is used to clarify some results of niche modeling exercises. The area of distribution of a virtual species which was generated by both environmental and biotic factors serves to illustrate the possibility that, at coarse resolutions, the distribution can be approximately recovered using only information about the environmental factors and ignoring the biotic interactions. Finally, information on the distribution of a butterfly species, Baronia brevicornis , is used to illustrate the importance of interpreting the results of niche models by including hypothesis about one class of movements. The results clarify the roles of the three factors in interpreting the results of using correlative approaches to modeling species distributions or their niches. |
