Other Environmental Sciences Commons^™

Do Bioclimate Variables Improve Performance Of Climate Envelope Models?, James I. Watling, Stephanie S. Romanach, David N. Bucklin, Carolina Speroterra, Laura A. Brandt, Leonard G. Pearlstine, Frank J. Mazzotti

United States Geological Survey: Staff Publications

Climate envelope models are widely used to forecast potential effects of climate change on species distributions. A key issue in climate envelope modeling is the selection of predictor variables that most directly influence species. To determine whether model performance and spatial predictions were related to the selection of predictor variables, we compared models using bioclimate variables with models constructed from monthly climate data for twelve terrestrial vertebrate species in the southeastern USA using two different algorithms (random forests or generalized linear models), and two model selection techniques (using uncorrelated predictors or a subset of user-defined biologically relevant predictor variables). There …