Using a bayesian multistate occupancy model to assess seabird and shorebird status in Glacier Bay, Alaska

The U.S. Department of Interior National Park Service is charged with both monitoring avian communities and evaluating the influence of visitors to National Parks on sensitive species; however, this task is challenging considering that sampling programs often involve multiple species, each with differing behavior, habitat requirements, and detectability. Our objectives were to build a model to describe the status of waterbirds in Glacier Bay National Park, Alaska, USA, and assess effects of area closures on these species. We used a Bayesian multistate occupancy model to describe the status of multiple species and make the best possible use of existing survey data. We modeled up to 5 states per species and evaluated predictors of occupancy, nesting, and abundance, as well as survey-related predictors of state-dependent detection probability. We found that occupancy probability varied across species and habitats (islands vs. glacial outwashes). For most species, occupancy probability was substantially greater at sites occupied in the year previous (site persistence). We found weak evidence that area closures affected the occurrence of species in the study, but this was largely because most sites were closed for the entirety of the study period. The probability of detecting occurrence, nesting, and abundance varied across species and survey methods (ground vs. vessel). Detection parameters provided valuable information for enhancing the efficiency of future surveys, by identifying preferred survey methods and sampling periods for specific waterbird species. © 2020 The Wildlife Society.