Road kill hotspots are frequently used to identify priority locations for mitigation measures. However, understanding the landscape context and temporal dynamics of these hotspots is a challenge. We investigate the factors that drive the spatiotemporal variation of bat mortality hotspots in a Mediterranean landscape. We hypothesise that hotspot locations occur at places where bat activity is higher. Additionally, we hypothesise that this activity is related to vegetation density and productivity because this is related to insect prey abundance. We used a combination of spatiotemporal analysis and generalised mixed models to evaluate the effect of the local spatial variation of vegetation productivity (as measured by the Normalized Vegetation Index -NDVI) on bats space use. Then, we combined this information with bat road kill locations to predict and validate spatial and temporal variation on road kill hotspot locations. During three years (2009, 2010, and 2011) we conducted daily surveys (n=690) along a 51 km long transect that incorporates different types of roads in southern Portugal, searching for bat casualties. Overall, we found 474 bat casualties during the sampling period. Then, to conduct the bat mortality analyses we assigned each road kill to the corresponding 500 m-road segment. We identified mortality hotspots for each year; segments where the number of casualties exceeded the upper 80% confidence limit of the mean, assuming a Poisson distribution of road kill per road segment, and analysed if they changed their location over the years. Overall, only 10% of segments were identified as hotspots during the whole year. 37% of road segments are intermittent road kill hotspots, i.e., they are classified as hotspots only in one or two years. 53% of road segments had very few bat casualties and were not identified as hotspots. For 20% of the road segments, we did not find any bat casualties. Thus, the non-persistent hotspots were the most frequent category. Further analysis of this type of hotspots showed that the spatiotemporal congruence of hotspots locations declined with decreasing vegetation production and the associated reduced bat activity on the proximity of road segments. This supports our hypothesis showing that a decline in overall vegetation productivity, and the presumed lower abundance of prey have a significant effect on the decrease of bat road kill. In this study, we show for the first time that using readily available series remote sensing data, and the indices that can be calculated based on this information, such as the NDVI, can be a powerful tool to predict bat road kill hotspots and their persistence in time. Thus, NDVI can be used in road planning, to prioritise location of mitigation measures or to identify essential road habitats for conservation.
