We tested the prediction that at coarse spatial scales, variables associated with climate, energy, and productivity hy- potheses should be better predictor（s） of bat species richness than those associated with envir...We tested the prediction that at coarse spatial scales, variables associated with climate, energy, and productivity hy- potheses should be better predictor（s） of bat species richness than those associated with environmental heterogeneity. Distribution ranges of 64 bat species were estimated with niche-based models informed by 3629 verified museum specimens. The influence of environmental correlates on bat richness was assessed using ordinary least squares regression （OLS）, simultaneous autoregressive models （SAR）, conditional autoregressive models （CAR）, spatial eigenvector-based filtering models （SEVM）, and Classification and Regression Trees （CART）. To test the assumption of stationarity, Geographically Weighted Regression （GWR） was used. Bat species richness was highest in the eastern parts of southern Africa, particularly in central Zimbabwe and along the western border of Mozambique. We found support for the predictions of both the habitat heterogeneity and climate/productivity/energy hypothe- ses, and as we expected, support varied among bat families and model selection. Richness patterns and predictors of Miniopteridae and Pteropodidae clearly differed from those of other bat families. Altitude range was the only independent variable that was sig- nificant in all models and it was most often the best predictor of bat richness. Standard coefficients of SAR and CAR models were similar to those of OLS models, while those of SEVM models differed. Although GWR indicated that the assumption of stationa- rity was violated, the CART analysis corroborated the findings of the curve-fitting models. Our results identify where additional data on current species ranges, and future conservation action and ecological work are needed.展开更多
文摘We tested the prediction that at coarse spatial scales, variables associated with climate, energy, and productivity hy- potheses should be better predictor（s） of bat species richness than those associated with environmental heterogeneity. Distribution ranges of 64 bat species were estimated with niche-based models informed by 3629 verified museum specimens. The influence of environmental correlates on bat richness was assessed using ordinary least squares regression （OLS）, simultaneous autoregressive models （SAR）, conditional autoregressive models （CAR）, spatial eigenvector-based filtering models （SEVM）, and Classification and Regression Trees （CART）. To test the assumption of stationarity, Geographically Weighted Regression （GWR） was used. Bat species richness was highest in the eastern parts of southern Africa, particularly in central Zimbabwe and along the western border of Mozambique. We found support for the predictions of both the habitat heterogeneity and climate/productivity/energy hypothe- ses, and as we expected, support varied among bat families and model selection. Richness patterns and predictors of Miniopteridae and Pteropodidae clearly differed from those of other bat families. Altitude range was the only independent variable that was sig- nificant in all models and it was most often the best predictor of bat richness. Standard coefficients of SAR and CAR models were similar to those of OLS models, while those of SEVM models differed. Although GWR indicated that the assumption of stationa- rity was violated, the CART analysis corroborated the findings of the curve-fitting models. Our results identify where additional data on current species ranges, and future conservation action and ecological work are needed.
