The destruction and fragmentation of habitats due to anthropogenic land use changes have led to the decline of numerous species by reducing the size and the connectivity of the remaining local populations and so contributing to their isolation. The impact of habitat fragmentation can be modeled using landscape graphs, which have become a popular tool. Habitat reduction and fragmentation has been suggested as one hypothesis for the decline of the lesser horseshoe bat Rhinolophus hipposideros in most of western and central Europe. Consequently, we expected habitat connectivity to influence the spatial distribution of the species, particularly through the presence of maternity roosts, which are essential for the persistence of the species. We designed this study to evaluate the impact of landscape connectivity on the distribution of the lesser horseshoe bat by comparing the predictive power of landscape composition alone with a model including both landscape composition and connectivity. We assessed the impact of landscape composition on maternity roost presence for different distances covered daily by bats (600, 2500 and 5000 m). We then applied a landscape graph-based approach to the roosting habitat of the lesser horseshoe bat to extract several patch-level metrics representing the functional connectivity of the landscape at different spatial scales. The results from those approaches show that the bats’ presence in the Franche-Comté region depends on the availability of wooded elements near small built areas and, at a broader scale, on the spatial integration of maternity roosts into a connected network allowing exchanges of individuals among roosts. This approach is a promising way to establish whether the presence probability of a given species depends on the potential connectivity between habitat patches quantified at different spatial scales. We expect that this method can be applied to taxa for which habitat fragmentation is one of the causes of population decline. We anticipate that using a graph-based species distribution model as a tool to predict species presence may focus conservation efforts on areas where habitat potential connectivity and landscape composition together should be taken into account, e.g. where anthropogenic landscape modifications are the main drivers of habitat connectivity.