SI Acoustic detection range.

SI Statistical analysis—Covariable extraction. SI Statistical analysis—Model weight calculation. S1 Fig. Map showing all 776 wind turbines in the studied counties (within Bretagne and Pays-de-la-Loire regions, western France), the land cover of the area, and the 154 sampled sites. S2 Fig. Boxplot showing that a large gradient of distance from the wind turbine was sampled each night. Horizontal line: median; box: first and third quartiles; whiskers: range; dots: outliers. S3 Fig. Boxplot of the wind incidence angles sampled each night. Various wind incidence angles were sampled each night to obtain a gradient of location around the turbine in relation to wind direction. Horizontal line: median; box: first and third quartiles; whiskers: range; dots: outliers. S4 Fig. Distribution of average wind turbine blade speed rotation per night for the entire dataset (left) and for the three subdatasets (right). S5 Fig. Distribution of the interaction of the tested gradients (wind incidence angle, depending on the distance from the wind turbine) for the entire dataset (left) and for the three subdatasets (right). S6 Fig. Distribution of average wind gusts per night for the entire dataset (left) and for the three subdatasets (right). S7 Fig. Distribution of average wind speed per night for the entire dataset (left) and for the three subdatasets (right). S8 Fig. Number of Pipistrellus pipistrellus each night depending on the average wind speed of the night (i.e. for the three subdatasets). S9 Fig. Candidate models within a ΔAICc < 7 containing (in color) or not (in grey) the variables of interest: distance to wind turbine (blue, on the left), wind incidence angle (orange, in the center), and their interaction (green, on the right). S10 Fig. Estimates, 95% confidence intervals, and p-values for the variables of interest contained in each candidate model within a ΔAICc < 7. S1 Table. Correlation matrix between variables included in the models for all datasets. No variables were correlated (r < |0.7|). Besides this correlation check, we checked for potential collinearity problems in the full models using the Variance Inflation Factor (VIF) before modelling (R package performance). WT = wind turbine. S2 Table. Mean ± standard deviation (min.-max.) of all variables included in the statistical analysis. S3 Table. AICc and R² of null, full and best models for each dataset. S4 Table. Estimates ± standard errors and p-values for the variables of interest in the full and best models (GLMMs). S5 Table. Estimates ± standard errors and p-values for the co variables in the full and best models (GLMMs). S6 Table. Estimates ± standard errors and p-values (in italics) for the predictors of bat activity for the model resulting from a complementary analysis.

(DOCX)