Rethinking road mitigation priorities through detection-informed interpretation of roadkill data and road crossability

Roads play a critical role in biodiversity loss by reducing species survival and promoting habitat fragmentation. Most roadkill studies focus on single species or localized areas, using roadkill as a direct measure of risk, often overlooking or misinterpreting inherent biases (i.e. carcass-location, persistence and observation biases). In this study, we apply two concepts to better understand the relationship between wildlife-vehicle collisions and habitat connectivity. First, we interpret roadkill locations modelled using species distribution models as indicators of roadkill detection, rather than actual roadkill risk. Additionally, we introduce road crossability, a measure of habitat connectivity along roads, assessed using circuit theory. We adopt a multispecies approach, analyzing roadkill data from 15 forest mammals with different mobility across different landscapes (natural and anthropized) in northeastern Italy, to investigate the relationship between roadkill detection and road crossability. Our results show positive correlations between roadkill detection and road crossability for medium and high mobility species, particularly in anthropized landscape, while low mobility species exhibit a weaker correlation. Combining these metrics, we classified the regional road network into zones with different conservation and mitigation priorities: Priority Areas for Connectivity Improvement (PACoIs), Potential Areas for Roadkill Mitigation (PARoMs), and Potential Areas for Connectivity Preservation (PACoPs). Our study suggests that multi-specie roadkill detection and road crossability assessment hold promise to facilitate more effective conservation strategies and mitigation measures to maintain habitat connectivity and reduce the impact of roads in different landscapes.