Predicting bat occurrence : evaluation of a connectivity-based habitat index for Swedish bats

Abstract: Ecological impact assessments are required for all large infrastructure or exploitation projects because of their effects on environment and species. Methods that would make such assessments more standardised, efficient and reliable are highly desirable in today’s society, where many species are decreasing due to human-induced habitat modifications. This thesis evaluates the performance of a habitat index for Swedish bat species (bat habitat index, BHI), which has the potential to be used as a tool for impact assessments. Previous studies have shown that bats are negatively affected by large roads because they represent a barrier for movement and because the noise associated with the traffic impedes efficient foraging. The BHI takes the barrier effect of the highway into account by assigning it the lowest value for permeability. Acoustic effects in proximity to the highway are ignored by the BHI. For this study, 50 sites were inventoried for four nights using automatic ultrasound recorders to get an empiric measure of bat occurrence that could be compared to the BHI’s predictions and the distance to the highway E18 which crosses the study area. The correlation between the predicted values from the BHI and the observed bat activity (number of recordings) or species richness was tested using generalised linear mixed models. Among the models with the same response variable, the best-performing model was selected based on a bias-corrected AIC (Akaike Information Criterion). The results show that the BHI is a good predictor for the number of species that regularly occur at a given site (i.e. species that were observed in a minimum of three out of four nights), but not for the total number of species. The BHI performed also much better at predicting the activity of forest-living species than overall bat activity. Instead of the expected positive correlation between the distance to the highway and bat activity, the observed negative correlation suggests that the BHI overestimates the barrier effect of the highway. This could easily be corrected by adjusting the permeability of highways in the construction of the BHI. For further improvement of the BHI, I also suggest that the goals are set more explicitely by deciding which species the BHI should focus on (e.g. all species or only forest-living species) and which measure of bat occurrence it should predict (e.g. regular occurrence of species, total species richness or activity).