Present and future taxonomic selectivity in bird and mammal extinctions

We compared the distribution of historical bird and mammal species extinctions across genera and families with the distribution we would expect if these extinctions had occurred at random with respect to taxonomy. We then repeated the comparison for species listed in various categories of threat according to the 1996 Red List of the World Conservation Union. We found the distributions of extinctions and threat classifications to be almost always nonrandom-'selective'-with clustering in certain genera and families. Furthermore, extinctions tended to be clustered in taxa that contain few species; species in smaller genera tended to have higher probabilities of extinction. This tendency was strong for historical extinctions but was reduced or absent for some categories of threat. We attribute this to a change in the causes of extinction whereby predation and introduced species have been joined or superseded by widespread habitat loss. We then assessed the implications of this variable selectivity for the past and likely future losses of genera and families. In most cases, the number of lost taxa rises. Finally, we made predictions about minimum losses of taxa at specific dates in the future and showed that, despite the reduction in some forms of selectivity, we will still lose more taxa than if species extinctions were random.