Landscape composition drives the impacts of artificial light at night on insectivorous bats.

Among the most prevalent sources of biodiversity declines, Artificial Light At Night (ALAN) is an emerging threat to global biodiversity. Much knowledge has already been gained to reduce impacts. However, the spatial variation of ALAN effects on biodiversity in interaction with landscape composition remains little studied, though it is of the utmost importance to identify lightscapes most in need of action. Several studies have shown that, at local scale, tree cover can intensify positive or negative effects of ALAN on biodiversity, but none have - at landscape scale - studied a wider range of landscape compositions around lit sites. We hypothesized that the magnitude of ALAN effects will depend on landscape composition and species' tolerance to light. Taking the case of insectivorous bats because of their varying sensitivity to ALAN, we investigated the species-specific activity response to ALAN. Bat activity was recorded along a gradient of light radiance. We ensured a large variability in landscape composition around 253 sampling sites. Among the 13 bat taxa studied, radiance decreased the activity of two groups of the slow-flying gleaner guild (Myotis and Plecotus spp.) and one species of the aerial-hawking guild (Pipistrellus pipistrellus), and increased the activity of two species of the aerial-hawking guild (Pipistrellus kuhlii and Pipistrellus pygmaeus). Among these five effects, the magnitude of four of them was driven by landscape composition. For five other species, ALAN effects were only detectable in particular landscape compositions, making the main effect of radiance undetectable without account for interactions with landscape. Specifically, effects were strongest in non-urban habitats, for both guilds. Results highlight the importance to prioritize ALAN reduction efforts in non-urban habitats, and how important is to account for landscape composition when studying ALAN effects on bats to avoid missing effects.