Predicting Bat Activity in the Rotor Swept Area by Modeling Bat Activity at High and Low Elevations in Open Habitats

Most wind facilities in the Midwestern United States are in open habitats, typically cropland, where spatial activity patterns of migratory tree bats (hoary, silver-haired, eastern red) may differ from that in woodland. Past... [ view full abstract ]

Most wind facilities in the Midwestern United States are in open habitats, typically cropland, where spatial activity patterns of migratory tree bats (hoary, silver-haired, eastern red) may differ from that in woodland. Past studies investigating the spatial activity patterns of bats focused on woodlands and riparian zones and concluded that bat activity detected at ground level was insufficient to represent bat activity at high elevations in the rotor-swept area (RSA)(Menzel et al. 2005, Collins & Jones 2009). To address this issue and describe bat activity and diversity across the vertical airspace of wind facilities in Midwestern farmed landscapes, project owners often install detectors on towers at high and low elevations. We investigated whether detectors installed at low elevations (~4 m) in open, primarily cropland habitat accurately reflected bat activity at high elevations (~55 m) at 13 towers in Michigan and Ohio in the high bat-fatality migration period (July 15-October 15, 2009-2011). This study’s objective was to investigate correlations between high and low elevation bat activity in order to develop confidence intervals (CIs) of bat activity at high elevations using a regression-based model and low-elevation data. We used R for linear regression analysis, calculated CIs from regression equations, and validated results with actual data. We found that there was significantly more bat activity and higher species diversity at low elevations compared to high elevations (t[n=5-11]<0.05), except for hoary bat, whose activity did not significantly differ across elevations (t[n=11]>0.05). The proportion of migratory tree bats was higher at 55 meters than at 4 meters; in addition, total migratory tree bat activity, as measured by calls/detector-night, was significantly correlated between high and low elevations (adj. r2=0.83, p<0.001). This was largely due to the frequent presence and strong correlation of hoary bats at both elevations (adj. r2=0.86, p<0.001). Eastern red bat activity also had a significant high-low positive correlation, but a lower adjusted r2 (adj. r2=0.38, p<0.03). We concluded that bat activity detected at low elevations in midwestern open habitats may be sufficient to partially model hoary bat activity in the RSA. It does not, however, fully represent the spatial activity pattern of other migratory tree bat species, especially at sites with high bat activity at low elevations, which results in large CIs at high elevations, making prediction difficult. With additional data from other high-low bat detectors deployed simultaneously in the migratory period, the model may be useful for predicting hoary bat activity in the rotor-swept area of open Midwestern landscapes, as well as for predicting total migratory tree bat activity, and perhaps eventually eastern red bat activity. This would be more flexible and cost-efficient than deploying bat detectors on high towers to detect these two common migratory tree bats, or migratory tree bats in general. These findings may best be applied early in the project development stages for evaluating sites before met towers are available, or for the rapid detection of migratory bat concentration areas at multiple locations across a region.