Estimating occupancy from autonomous recording unit data in the presence of misclassifications and detection heterogeneity

Autonomous recording units (ARUs) are now widely used to survey communities of species. These surveys generate spatially and temporally replicated counts unmarked animals, but such data typically include false negatives misclassified detections, both which may vary across sites in proportion abundance. challenges can bias estimates occupancy, the typical approach verifying individual detections is expensive. We developed a Bayesian implementation two-species, false-positive N-mixture model for estimating occupancy from ARU or other animals that does not require manual verification. The accounts species misclassification abundance-induced detection heterogeneity, as well negatives. To evaluate this model, we simulated 200 datasets each 29 scenarios, including scenarios misclassifications outnumbered correct classifications rare also applied acoustic bats conducted on Fort Carson Army Post Piñon Canyon Maneuver Site, Colorado, USA. In simulation study, bias, coverage root mean square error obtained two-species were superior metrics two competing models. Across absolute was consistently low (range: −0.03 0.07), while averaged 93% 74%–98%). For alternative models, often high −0.36 0.39), 47% 65%. Although our included an abundance parameter, reliable. Myotis bats, estimated 1%–5% field-recorded misclassified. Estimated (0.91 0.76) lower than naïve (1.00 0.94). Competing models implausibly local 0.00 at with numerous detections. Our significant because it heterogeneity improves without expensive verification field application indicated common, yet affected inferences. Given ARUs increasingly broad range taxa, could be useful.