Abstract
Wind power is a promising clean energy technology that has grown rapidly in recent years (EIA 2013). In spite of its environmentally friendly reputation, industrial wind energy generation can have serious impacts on wildlife. Bat and bird collision fatality rates have been alarmingly high at some wind farms. Proper siting of wind facilities may help minimize collision impacts as the wind energy industry continues to grow. Bat and bird fatality rates vary greatly among sites; however , there is no reliable method for assessing collision risk prior to development. My goal was to develop a method for predicting fatality rates based on nocturnal activity patterns measured by ground-level recording of bat and bird calls. For three years, I monitored bat and bird activity using ultrasonic-acoustic detector s at 160 locations, including eight wind farms and a variety of landscape settings to: 1) examine the capabilities of the detector for use in pre-construction site assessment, 2) evaluate the ability of an automated bat call identification program to identify the species of recorded bat calls, 3) determine how pass rates relate to fatality rates for use in predictive models based on pre-construction recordings, 4 ) examine variation in pass rates with respect to pre-specified landscape and habitat features, 5 ) ex amine how activity patterns might differ before versus after a wind facility is built, and 6 ) investigate whether bat activity levels are elevated near turbines. Ground-based recording was found to be a useful method for studying near-ground bat activity patterns at multiple scales, but patterns of acoustic activity of birds were less clear and apparent only at the most coarse geographic scale. The automated bat c all identification program produced mixed results among species and geographic regions. No relations between bat pass rates and estimated fatality rates among wind farms were found, either for all bats or for migratory tree-roosting species. Large differences in bat and bird activity among geographic regions were found, with highest activity levels near Great Lakes coastlines. Also, bat and bird activity levels near the edge of forested river corridors in agricultural settings in Minnesota were found to be higher than those farther from the edge. Evaluation of a variety of predictive models of pass rates revealed distance to water, distance to trees, and ecoregion as good predictors of bat activity levels. Although some differences in bird activity were evident at the broadest geographic scale, models were of limited usefulness in explaining spatial variation in bird activity. Acoustic activity measured by ground-based recorders was not a good predictor of bat fatalities at wind farms; however , it did reveal local and regional patterns that may be useful for siting wind energy facilities in low-impact areas.