Extinctions of montane mammals reconsidered: putting a global-warming scenario on ice [Extinctions de mammifères de montagne reconsidéré: mettre un scénario de réchauffement global sur la glace].

Résumé

McDonald and Brown developed a model based on island biogeographic principles to predict the magnitude and composition of small mammal extinctions from isolated boreal habitats atop mountains of the Great Basin following global warming. The model predicts that three of 14 boreal mammals will go extinct regionally and that four of 19 mountain ranges will lose upwards of 50% of their present faunas. Here, we re-examine the model on the basis of its underlying assumptions, on the statistical and biogeographic protocols used, and on its predictive power. A key assumption, that populations of these small mammals are isolated by absolute barriers to dispersal, is challenged by published field observations and by extensive trapping records. Statistical procedures used to construct the model are questionable and the model itself yields imprecise estimates. The biogeographic principle used to identify extinction-prone species, nested subsets of species, makes predictions that are at odds with available autecological information. The demonstration of a nested pattern of species occurrences does not provide definitive evidence in resolving SLOSS-or whether a single large island or several small islands of equivalent total area will contain more species. We conclude that the model is not a reliable method for forecasting species extinctions following global warming. The final resolution of the biogeography of montane mammals (and predictive models of extinction) in the Great Basin must await a full and accurate accounting of past and present species distributions.