Constraints on Interpretation of Ecological Niche Models by Limited Environmental Ranges on Calibration Areas: Software Script Appendix

Abstract

A common application of correlational models of ecological niches of species is to transfer model rules onto other sets of conditions to evaluate distributional potential under those conditions. As with any model fitting exercise, however, interpretation of model predictions outside of the range of the independent variables on which models were calibrated is perilous. We use novel visualization techniques to characterize model response surfaces for a variety of niche modeling algorithms for a virtual species (wherein the truth is known) and for two transfer-based studies published by one of our group. All modeling algorithms, for each species, showed some sort of extrapolation, such that biologically unrealistic responses were reconstructed. We present MOP metrics, a correction and simplification of MESS metrics currently in use, as a means of identifying areas where such extrapolation is likely. We also discuss the implications of these results for studies of ecological niche evolution. Note: This item consists of R code for implementing two of the analyses developed in this paper.