Bat Species Density Gradients in the New World: A Statistical Assessment

Bat species density gradients in North America, South America, and the continental New World were analysed by a variety of multivariable statistical pro- cedures. Multiple regression analyses were performed to identify which of four possible descriptors (latitude, longi- tude, area, or biome richness) best accounted for variation in bat species density among quadrats. Independent descriptors were considered to be meaningful if they met two criteria: the descriptor must make a significant con- tribution to the multiple regression and it must augment R2 by at least 5%. Because different taxa might respond to independent descriptors in different ways, separate analyses were conducted for molossids, vespertilionids, phyllostomids, non-phyllostomids, and all bats. Latitude was consistently the most important predictor of bat species densities (r2>0.72), except in the case of vespertilionids in the continental New World, in which latitude was unimpor- tant (r2=0.03) and the two most important variables (longi- tude and biome richness) together only accounted for approximately 40% of the variation. Even a control analysis on data for Brazil (for which finer resolution of ecological life zones was available) failed to identify biome richness as an important predictor of species densities but did indicate that latitude was again the most important