Testing the predictive capabilities of ecological niche models: a case study examining Red‐bellied Woodpeckers

Abstract

AbstractAs humans continue to alter natural habitats indirectly and directly, species’ geographic ranges may change as they track shifting climate regimes or changing landscapes. Ecological niche models (ENMs) are frequently used to show modern ranges and forecast future range changes. These models, however, assume that a species will exhibit niche conservatism, though this is rarely tested. Here, we examine a woodpecker species with a well‐documented recent range expansion to assess the effectiveness of predictive models by comparing the pre‐ and post‐expansion climate/habitat niche using ENMs and multivariate principal components analysis. Occurrence data for the Red‐bellied Woodpecker (Melanerpes carolinus) were obtained from the pre‐expansion (1910–1939; n = 299) and post‐expansion (1980–2009; n = 868) time periods. Ecological niche models were constructed using GIS layers describing climate data and crop cover for the pre‐ and post‐expansion time periods. We then used the pre‐expansion ENM to forecast the distribution of Red‐bellied Woodpeckers into the post‐expansion time period, and vice versa, and calculated the niche overlap of the projected distributions and the actual distributions in those time periods to determine whether pre‐ and post‐expansion niches were similar. Predictive ENMs did not closely match the actual distributions in the pre‐ and post‐expansion time periods, respectively, and the pre‐expansion and post‐expansion niches were significantly different from each other (Schoener's D = 0.745, P < 0.001). Multivariate analyses revealed that the present‐day niche encompasses the past niche and that Red‐bellied Woodpeckers today have broader temperature and precipitation tolerances and are found in both more‐ and less‐forested areas than they were in the pre‐expansion era. Our analyses reveal that Red‐bellied Woodpeckers are not exhibiting climate or habitat niche conservatism, explaining why predictive ENMs for this species could not effectively track their range shift. This study emphasizes that predictive models may not be effective for species undergoing niche changes.