Statistical power of mobile acoustic monitoring to detect population change in southeastern U.S. bat species, a case study

Abstract

Bat populations in eastern North America have experienced precipitous declines following the spread of white-nose syndrome (WNS) and other population stressors. It is imperative to understand changes in bat populations as WNS spreads to provide appropriate guidance for species management. We developed generalized linear mixed-models of population trend and habitat associations for five indicator bat species on U.S. Fish and Wildlife Service Mobile Acoustic Bat Monitoring program routes across 86 sites in the southeastern United States from 2012 to 2017. We estimated substantial declining annual trends in relative abundance of tricolored bat (Perimyotis subflavus; -15.1% [-20.6 to -9.1% 95% CI]) and little brown bat (Myotis lucifugus; -13.9% [-22.9 to -3.8% 95% CI]). Relative abundance of bat species increased throughout the summer, and associated positively with the amount of woody cover along survey routes in all but P. subflavus. Fewer evening bats (Nycticeius humeralis) and eastern red bats (Lasiurus borealis) were detected along more developed routes. Using these models, we conducted a prospective power analysis to examine sampling effort necessary to detect moderate to catastrophic population changes in bat populations. We estimated that it would require 10–20 years of surveys on 50–100 routes to detect 5% annual declines in all species at 80% power and α = 0.1. Detecting a 2.73% annual decline may require >200 surveys over >20 years; whereas a 1.14% annual decline was nearly impossible to detect via our program. We demonstrate and caution that underpowered monitoring programs may misrepresent the magnitude and/or sign of population trajectories. We recommend project-specific power analysis continue to be emphasized as an important study design component for effective long-term monitoring programs.

Publication
Ecological Indicators