Combining roadkill hotspots and landscape features to guide mitigation measures on highways

Identifying highway sections with high vertebrate roadkill and associated landscape features is crucial for proposing mitigation measures to reduce the negative impacts of wildlife-vehicle collisions on biodiversity, human safety, and the economy. Focusing on highways in the State of Mato Grosso do Sul, Brazil, this study proposes a prioritization of road segments for implementing mitigation measures based on landscape characteristics and roadkill hotspots recorded between 2016 and 2018. Roadkills distribution was assessed using Hotspot Identification, detecting sections with the highest collision intensity. The effect of landscape variables, such as the area and distance of natural vegetation and anthropic land use within 1 km of the highways, on roadkill was modelled using General Linear Mixed Models, which were evaluated based on two response variables: (1) roadkill presence/absence and (2) roadkill abundance, with model selection by the Akaike information criterion. Based on roadkill hotspot identification analysis and predictive models of the influence of landscape elements on roadkill occurrence and abundance, an evaluation criterion was proposed to determine the priority sites for installing wildlife crossings, speed bumps, and signposts. Wildlife-vehicle collisions were positively associated with riparian vegetation areas and urban areas, while negatively associated with silviculture and distance to “cerradão” (forest Cerrado) patches. Eleven road sections were identified as top priority, two as high priority and 78 as medium priority for the installation of mitigation measures with varying cost-benefit ratios. Our results can potentially guide decision-making at broader scales and provide mitigating alternatives applicable to dry ecosystems such as the Cerrado (Brazilian Savanna).