Modeling the power of acoustic monitoring to predict bat fatalities at wind turbines

Large numbers of bats are killed at wind turbines worldwide. To formulate mitigation measures such as curtailment, recent approaches relate the acoustic activity of bats around reference turbines to casualties to extrapolate fatality rates at turbines where only acoustic surveys are conducted. Here, we modeled how sensitive this approach is when spatial distributions of bats vary within the rotor-swept zone, and when the coverage of acoustic monitoring deteriorates, for example, with increasing turbine size. The predictive power of acoustic surveys was high for uniform or random distributions of bats. A concentration of bat passes around the nacelle or at the lower portion of the risk zone caused an overestimation of bat activity when ultrasonic microphones were pointed downwards at the nacelle. Conversely, a concentration of bat passes at the edge or at the top portion of the risk zone caused an underestimation of bat activity. These effects increased as the coverage of the acoustic monitoring decreased. Extrapolated fatality rates may not necessarily match with real conditions without knowledge of the spatial distribution of bats, particularly when the risk zone is poorly covered by acoustic monitoring, when spatial distributions are skewed and when turbines are large or frequencies of echolocating bats high. We argue that the predictive power of acoustic surveys is sufficiently strong for nonrandom or nonuniform distributions when validated by carcass searches and by complementary studies on the spatial distribution of bats at turbines.