Improved Analysis of Long-Term Monitoring Data Demonstrates Marked Regional Declines of Bat Populations in the Eastern United States

<div><p>Bats are diverse and ecologically important, but are also subject to a suite of severe threats. Evidence for localized bat mortality from these threats is well-documented in some cases, but long-term changes in regional populations of bats remain poorly understood. Bat hibernation surveys provide an opportunity to improve understanding, but analysis is complicated by bats' cryptic nature, non-conformity of count data to assumptions of traditional statistical methods, and observation heterogeneities such as variation in survey timing. We used generalized additive mixed models (GAMMs) to account for these complicating factors and to evaluate long-term, regional population trajectories of bats. We focused on four hibernating bat species – little brown myotis (<i>Myotis lucifugus</i>), tri-colored bat (<i>Perimyotis subflavus</i>), Indiana myotis (<i>M. sodalis</i>), and northern myotis (<i>M. septentrionalis</i>) – in a four-state region of the eastern United States during 1999–2011.</p><p>Our results, from counts of nearly 1.2 million bats, suggest that cumulative declines in regional relative abundance by 2011 from peak levels were 71% (with 95% confidence interval of ±11%) in <i>M. lucifugus</i>, 34% (±38%) in <i>P. subflavus</i>, 30% (±26%) in <i>M. sodalis</i>, and 31% (±18%) in <i>M. septentrionalis</i>. The <i>M. lucifugus</i> population fluctuated until 2004 before persistently declining, and the populations of the other three species declined persistently throughout the study period. Population trajectories suggest declines likely resulted from the combined effect of multiple threats, and indicate a need for enhanced conservation efforts. They provide strong support for a change in the IUCN Red List conservation status in <i>M. lucifugus</i> from Least Concern to Endangered within the study area, and are suggestive of a need to change the conservation status of the other species. Our modeling approach provided estimates of uncertainty, accommodated non-linearities, and controlled for observation heterogeneities, and thus has wide applicability for evaluating population trajectories in other wildlife species.</p>