How well do species distribution models measure variable importance?

One of the most common applications of SDMs is to identify important variables and measure their relative effect. Despite hundreds of papers assessing the predictive power of SDMs, there are none assessing their inferential power. Maria Santos and Adam recently completed the first such analysis!

Permute-after-calibration test of variable importance. The OMNI model perfectly recreates the species’ range so serves as a benchmark for the other models.

Smith, A.B. and Santos, M.J. Testing the ability of species distribution models to infer variable importance. bioRxiv doi: 10.1101/715904

Tags: ,

Leave a Reply

You must be logged in to post a comment.