Threats to avifauna on oceanic islands.

Results of the study by Blackburn et al. (2004a) of avifauna on oceanic islands suggest that distance from the mainland and time since European colonization have major influences on species extinctions and that island area is a significant but secondary contributing factor. After augmenting the data of the study on geographical properties for some of the islands they examined, we used a causal analysis approach with structural equation modeling to reexamine their conclusions. In our model geographical properties of islands, such as island area and isolation, were considered constraints on biological factors, such as the number of introduced mammalian predators and existing number of avifauna, that can directly or indirectly influence extinction. Of the variables we tested, island area had the greatest total influence on the threat of extinction due to its direct and indirect effects on the size of island avifauna. Larger islands had both a greater number of threatened bird species and more avifauna, increasing the number of species that could become threatened with extinction. Island isolation also had a significant, positive, and direct effect on threats to island avifauna because islands farther from the mainland had fewer current extant avifauna. Time since European colonization had a significant negative, but relatively weaker, influence on threats compared with the traditional biogeographic factors of island area and distance to the mainland. We also tested the hypothesis that the amount of threat is proportionally lower on islands that have had more extinctions (i.e., there is a "filter effect"). Because the proportion of bird extinctions potentially explained only 2.3% of the variation in the proportion of threatened species on islands, our results did not support this hypothesis. Causal modeling provided a powerful tool for examining threat of extinction patterns of known and hypothesized pathways of influence.