Geographical variation in the influence of habitat and climate on site occupancy turnover in American pika (Ochotona princeps)

Aim: Environmental changes that amplify rates of site or patch occupancy turnover can increase risks of decline in spatially structured populations. We asked whether local habitat and meso-scale climate influenced site occupancy turnover rates in four American pika (Ochotona princeps) metapopulations. We focused on winter cold stress, which is a proposed driver of American pika extinction risk but has been rarely studied. Location: Oregon, Northern California, and Idaho. Methods: We developed Bayesian hierarchical multiseason site occupancy models that accounted for both false-negative and false-positive survey detection errors to explore the winter stress turnover hypothesis. We used remotely sensed meso-scale (1km) snowpack and temperature data and fine-grained local habitat attributes as covariates to model site persistence and colonization rates. Results: Main conclusionsThe estimated magnitude of imperfect detection was greater than previously reported for the species. After accounting for imperfect detection, we found no evidence of declines in site occupancy over the 5-year study period, but our models provided evidence that pika site occupancy turnover can be high (>50% between some years) and apparently exacerbated by winter cold stress, summer heat stress and variation in site habitat quality. However, strength of evidence varied among metapopulations, suggesting influential local contingencies, as reported previously for the species. Main conclusions: Our empirical results suggest that the American pika may be vulnerable to wintertime turnover, negatively exacerbated by climatic events, with implications for future persistence given forecasted snowpack declines across the species' range. Our models suggest a more nuanced dynamism to persistence and extinction risk than the simple scenario of inexorable, monotonic range contraction offered by static range-wide distribution models, and we suggest several ways to strengthen these insights with future studies.