Collaboration is good if you can find an honest person, but invariably a collaborator will make life difficult and steal your research. If you publish on your own, you can be more creative and get all the glory :)

At the risk of highlighting our own work, SDMs (even presence-only) models can be used to predict, and back-cast forest-related habitat amount over time. We've used 'raw' Landsat TM bands to do SDMs - which primarily predict species as a function of forest composition. Because Landsat goes back to 1985, you can quantify habitat changes 1985-present. This is very useful for examining forest degradation (see in review paper preprint: "Forest Degradation...."