Exploring the relative importance of satellite-derived descriptors of production, topography and land cover for predicting breeding bird species richness over Ontario, Canada

In this paper we investigate the relative predictive power of a number of remote sensing-derived environmental descriptors of land cover and productivity to predict species richness of breeding birds in Ontario, Canada. Specifically, we first developed a suite of environmental descriptors (productivity, land cover, and elevation). These descriptors were based on readily available data, including the MODerate-resolution Imaging Spectroradiometer (MODIS) onboard the Terra and Aqua satellites and terrain data from the Shuttle Radar Topography Mission (SRTM). We then assessed the capacity of the environmental descriptors, using a decision tree approach, to estimate species richness of all breeding birds, and of groups of bird species based on habitat and nesting groupings, using data summarized from the Ontario Breeding Bird Atlas. Results indicated that the variance in the distributions of total bird species richness, as well as richness of habitat and nesting groups, were well predicted by the environmental descriptors (with variance explained ranging between 47 and 75%) with the predictions clearly related to both habitat (as modeled by land cover and land cover diversity) and vegetation productivity. Modeling demonstrates that initial partitioning is most often based on land cover class, indicating that it may be the driving variable of bird species richness; however, information on vegetation productivity and energy were then critical in defining how many species occur in each habitat type. The results indicate that remotely sensed environmental descriptors can provide an effective tool for predicting breeding bird species richness at regional scales. (c) 2008 Elsevier Inc. All rights reserved.