Implications of rescaling rules for multi-scaled habitat models

Abstract

Recent work in landscape ecology suggests that organisms use environmental cues at a variety of scales to select habitat. As a result, habitat studies that evaluate environmental conditions at multiple spatial scales have become increasingly common. We examined whether the way in which data are rescaled influences inferred relationships between organisms and habitat features. Using a habitat model developed at fine scales, we systematically rescaled habitat (canopy density, slope, and cover type) and distribution maps according to a variety of different rescaling rules, including spatial averaging, thresholding, presence/absence, and majority. We found that the spatial autocorrelation of habitat data interacts with rescaling rules to alter the correspondence between species presence and habitat across scale. Different rules lead to substantially divergent and sometimes opposite correlations among the species and habitat features on the landscape. Such differences in interpretation due to variation in methodology can lead to very different interpretations of a species’ habitat requirements and thus have important implications for both ecology and conservation.