Bat fatalities at different wind facilities in Northwest Germany

To identify driving factors of bat fatalities, we performed a general linear model analysis (GLM). Bat fatalities were specified as dependent variable of the model; bat activity (contacts), wind-turbine dimensions, weather data and location parameters were specified as independent variables of the model. The dependent variable was logtransformed x' = ln(x+1) prior to analysis. Bat fatalities represented 3-day measurements, and therefore weather data and bat contacts were averaged over three days to match the same time-resolution. Non-linear terms of weather data were included to allow for representation of a hump-shaped relationship between weather and bat fatalities in the model. Backward selection of variables was applied to ensure that the final model was exclusively built from significant variables. The analysis was performed using the Statistica for Windows package (StatSoft Inc., Tulsa, OK, USA).