Importance of causal analysis of threats to oceanic avifaunas is important: reply to Blackburn et al.

We are pleased that Blackburn et al. (2008 [this issue]) have taken an interest in our causal modeling of threatened bird faunas on oceanic islands (Trevino et al. 2007). Our results indicated possible environmental influences on the number of threatened bird species from the physical and biotic characteristics of islands. Blackburn et al. discuss possible strengths and weaknesses of 2 methods of analyzing such influences: generalized?linear mixed modeling and causal modeling with structural equations and standardized partial?regression coefficients (these last used as measures of effect sizes). The former method excludes information because it does not incorporate the logical structure of causes, whereas the latter takes advantage of the fact that physical aspects of islands (island area and isolation from a mainland) might influence biotic factors (number of species in the avifauna, number of introduced mammalian predators, and time since human colonization of the islands) and that physical and biotic factors might influence the dependent variable (number of bird species under threat of extinction) (Fig. 1). Although physical factors are likely to influence biotic factors, we think the reverse is unlikely to be true. The method we used allowed us to test an a priori theoretical model that is a working hypothesis of causal mechanisms (Mitchell 1992) threatening birds on oceanic islands. This method can be used to refine and deepen understanding of the potential collaborative factors that may influence the extinction threat of ocean island avifaunas.