The level of threat to restricted-range bird species can be predicted from mapped data on land use and human population

A key element in the efficient allocation of scarce resources for conservation is the identification of areas of high biological value and high threat. Habitat loss and human population density have proved useful predictors of spatial variation in the current threat status of species albeit at coarse spatial resolution. We present a global analysis, intersecting Endemic Bird Areas (EBAs), to which restricted-range bird species are endemic, with fine-scale data of agricultural extent and human population density and test: (a) how well variation in land use mapped at 0.5degrees x 0.5degrees resolution predicts spatial variation in threat status of species and (b) how the predictive power compares with that of human population density mapped at the same resolution. Variation among EBAs in the proportion of restricted-range species that are threatened can be predicted by both the proportion of land used for agriculture and human population density. Agricultural land use was a better predictor of threat status than human population. Further, the average levels of threat attributable to agriculture were better predicted by land use than human population density, whereas threats due to causes other than agriculture were equally well predicted by land use and human population density. We fitted quantitative empirical models to describe these relationships. These results could be used, together with spatially explicit future scenarios of land use change, to project the geographical distribution and magnitude of future threats to birds at global and regional scales. (C) 2004 Elsevier Ltd. All rights reserved.