Integrated distribution modeling resolves asynchrony between bat population impacts and occupancy trends through latent abundance

Introduction

Understanding where organisms occur and in what local densities, as well as how and why species distributions change in space and time, are central themes in ecology^{1, 2–3}, biogeography^2,3, and conservation biology⁴. Especially of interest is how species’ occupancy and abundance distributions relate to environmental conditions^{1, 2–3}, population processes^{1, 2–3}, stressors⁴, management actions⁴, or with one-another^{1, 2–3}. Conservation monitoring programs collect and analyze ecological data to address these questions and to inform where and in what intensities limited management actions might be targeted to achieve conservation goals^{4, 5, 6, 7–8}. However, despite advancements in collaborative monitoring programs at macro scales^9,10 and in statistical methods to estimate occupancy and abundance distributions^11,12, extracting simple and comprehensive answers to such questions remains challenging¹².

Large-scale monitoring often requires that several field techniques be deployed across different spatial and temporal resolutions, especially for species with cryptic behaviors, seasonal life cycles, migratory behavior, and continental distributions^6,8,13. Data collected from different field techniques and seasons may be prone to different sources and magnitudes of observation bias (false-negatives, false-positives, differential effort) and be best suited to inform different population metrics (occupancy vs abundance). For example, bats in North America exemplify these challenges^6,8. The North American Bat Monitoring Program^8,10 (NABat) was established in 2015 to address a historical lack of coordinated population monitoring efforts for most species of North American bats⁶ in the wake of white-nose syndrome (WNS), which has caused devastating population declines of hibernating bats^14,15. NABat uses information from multiple monitoring methodologies to gain a richer understanding of population status and trends^6,8, including acoustics (deployed at stationary point locations and along mobile transects) and live-capture data in the summer maternity season. In the winter, counts of individuals within hibernating bat colonies are also obtained. From these streams of monitoring data, population status and trend inferences are based on (1) species occupancy in the summer^16,17, (2) relative abundance in the summer¹⁸, or (3) relative abundance in the winter^14,19. Thus, population inferences to date have been mostly confined to ‘multiple lines of evidence’ which...