Modelling to forestall extinction of Australian tropical birds

The tropics support some of the world's richest centres of species diversity and endemism, yet these biomes are now dangerously imperilled by anthropogenic change. If we are to avert, or at least mitigate, catastrophic loss of species in these areas, it is vital to understand the direct and indirect effects of these agents of threat. Moreover, provision of a robust theoretical and empirical underpinning for the relationship between evolved characteristics (life-history traits and ecological preferences) and extinction risk may provide a general theory of extinction process that would be useful for conservation management. Birds provide some of the best quantitative data in tropical regions on the rate and selectivity of extinctions. Also, their autecology is better known than most other tropical taxonomic groups, making them ideal candidates for the application and testing of extinction theory and viability models. Using the relatively well-studied avifauna of the Australian tropics as an example, we apply population, habitat and threat data based on observed trends, in combination with various lines of surrogate information, to develop a predictive framework for extinction vulnerability of the region's birds. Our inferences are based on generalized linear mixed modelling from an a priori set of models that include range, population, life history and threat variables. The best-selected model, with 92% of the AIC weight of evidence, was then used to rank the 387 species in terms of relative threat. We compare these results with predictions based on trends in habitat alone, and consider the likely impact of projected future threats during the twenty-first century.