A clear challenge for ecological niche modeling when using commonly collected biodiversity data is determining how to best mitigate the effects of sampling bias. Here, we review some of the recent approaches to the effects of sampling bias on models, focusing on filtering occurrences in overrepresented regions based on geographic and environmental proximity. We tested the efficacy of filtering in geographic and environmental space using occurrence data from a restricted rodent, the Gulf Coast kangaroo rat, and widespread passerine, Audubon's Oriole. Our evaluation strategies examined 14 distance measures in geographic and environmental spaces and eight combinations of environmental variables and their ordinations. This resulted in 78 datasets for each species, which we evaluated using multiple evaluation statistics. The degree of change produced by filtering on predicted suitability and evaluation statistics increased with increasing range size. Environmental filtering resulted in higher model fit at larger extents and retained more occurrences than geographic filtering. Our results indicate that models should be evaluated using multiple evaluation statistics at multiple thresholds. We recommend that ecological niche modeling using natural history collection data should use multiple filtering schemes to see the effects on predictions.
