There seem to be more and more habitat models being published. The idea is usually to take a set of geo-referenced records of a particular plant or animal in an area; find a statistical association with a gridded set of environmental variables; then extrapolate the likelihood of that species occurring in all of the grid squares in the study area. The result can be called a habitat suitability index, or a similar term, and can be used to help inform management policies, protected area planning, and forecasting the effects of future change scenarios.
I guess that the increasing number of such modelling efforts is in part due to the very welcome increase in environmental data – factors such as rainfall, elevation, protected areas, towns, rivers, land cover and vegetation indices are readily available for pretty much anywhere in the world. Added to this is the availability of free software, GIS and statistical, as well as far greater access to computers and knowledge of relevant techniques.
So I was interested and pleased to come across this piece from Colin Beale at York University: it is not only a constructive critique of a particular research paper that used such techniques, but it widens the discussion to the nature of explanations in biology: