Estimating extinction risk with minimal data

Anthropogenic pressures on the global biome are causing widespread species declines and extinctions. Assessing the extinction risk faced by individual species is a critical first step in combating this trend. However, we lack high quality demographic data to do so for the vast majority of plant and animal species. We present an efficient modeling approach to estimate extinction risk based on a statistical framework from the mark-recapture literature. We assessed the model's performance using a combination of simulated data, results from a protist microcosm experiment, and data from a long-term, large-scale habitat fragmentation experiment in southeastern Australia. Simulation experiments showed the model is robust to missing data as well as biological processes not included explicitly in the model's assumptions. Fitting the model to data from the protist experiment yielded accurate predictions of the regional extinction dynamics observed in the system, even with a relatively low level of replication. Finally, the model was able to accurately predict the observed dynamics in the habitat fragmentation experiment. The model provides a robust and accurate method to evaluate a species' extinction risk. Since it only requires presence/absence data, applies to a wide range of survey designs, and allows for observational uncertainty and missing data, it can be applied to many datasets that existing models cannot accommodate. For these reasons, the model should be useful in conservation settings.