Detecting extirpation: A localized approach to a global problem

Societal Impact Statement The global biodiversity crisis stems from a cascading series of extirpations driving species toward extinction. Addressing this crisis requires methods for early detection of extinction at local scales, where communities can mobilize conservation efforts. We present a method for the detection of species extirpation, harnessing biological specimens, community science, and targeted surveys. Our framework aligns with the International Union for the Conservation of Nature (IUCN) extinction criteria and provides a practical means of testing the hypothesis of extinction. By streamlining the integration of data into a practical framework for inference, our approach overcomes key challenges in inferring local extinction risk, supporting efforts to detect and mitigate biodiversity loss. Summary Extinction results when all local populations have been extirpated. We therefore require systems for the detection of extirpation to contend with this global problem at its source. Through a baseline analysis incorporating half a century of historical collection data with 5 years of data gathered via a community biodiversity project, we identified 10 plant species from Galiano Island, BC, Canada, that had not been seen recently as suitable subjects for this study. These became the target of systematic surveys over an additional 4 years. Using these data, we applied a Bayesian framework to estimate the probability of species' absences within regions of suitable (historical and potential) habitat, in accordance with the IUCN extinction criteria. Six target species were redetected. For the others, we inferred conditional extirpation probabilities based on relaxed (only historical habitat patches) and strict (all potential habitat) interpretations of the IUCN criteria. Under a relaxed interpretation of IUCN criteria, we found a high probability ~95% of extirpation (local extinction) for two species ( Crassula connata and Primula pauciflora ); under strict criteria, we inferred extirpation probabilities of 89.1% and 83.7%. We demonstrate a framework for the detection of species extirpation, harnessing historical collections, community science, and targeted surveys to meet the IUCN extinction criteria with explicitly defined degrees of confidence.