Ordonez gave an overview of species distribution modeling and had a nice slide on all the different programs that help model such things: Bioclim, Domain, Biomapper, Maxent, Garp, Grasp, Species, Biomod, and others. He also referred to a paper (Elith et al., Ecography 2006) that tested the various programs. My own work these days is focusing somewhat on species distribution modeling, but I have to admit the whole practice makes me nervous. We already have a word for this science: it’s called “ecology,” the study of the distribution and abundance of species. The fact that we smaller set of ecologists have developed a new word for it suggests that we’ve decided to ignore much of ecology — and evolution — in order to simplify the process. And that’s exactly what we’ve done.

The basic approach in species distribution modeling / habitat suitability modeling is you collect occurence data for the target species, collect spatially-explicit environmental variables that might determine the limits of the species’ range, and then model where you would expect to find the species in your study area. In general, it works okay (no models are real, some are good, &c &c). It can be useful, especially in conservation, when rigorous surveys aren’t feasible. But people have been using these methods to model current distribution, and then predict range shifts based on climate change. As with yesterday’s biophysics approach to modeling flyways, such a study ignores things like behavior and adaptation. What limits the current generation might not do so in other conditions.