Landscape and spatial correlates of vehicle collisions involving large and medium-sized mammals: a systematic review and meta-analysis

Wildlife-vehicle collisions (WVCs), particularly those involving large and medium-sized mammals, represent one of the most widespread and consequential forms of human-wildlife interaction worldwide. These collisions result in human injuries and fatalities, substantial economic losses, direct wildlife mortality, and associated biodiversity loss. Understanding the drivers of mammal-vehicle collisions is therefore essential to mitigate their impacts on both human safety and wildlife conservation. Numerous studies have investigated a wide range of potential predictors influencing the frequency and spatial distribution of WVCs. Among these, landscape and spatial factors have frequently been identified as important correlates. However, findings remain fragmented across regions, species, and methodological approaches, limiting the identification of general patterns. Therefore, the aim of this study is to summarize current evidence on how landscape and spatial features influence WVCs. To address this gap, in the context of the TransWILD Biodiversa+ project, we conducted a systematic review and meta-analysis following PRISMA guidelines. We applied a structured and replicable search strategy using Scopus, Web of Science Core Collection, BIOSIS Citation Index and Zoological Record databases, retrieving 2060 records and conducted title and abstract screening based on the following criteria: 1) presence of landscape and spatial predictors and 2) focus on terrestrial mammals. After the first screening, we retained 533 articles for full-text assessment and selected 185 for further evaluation on their suitability for the meta-analysis based on the presence of quantitative data. Preliminary assessment of these studies involved extracting all reported landscape and spatial predictors to investigate their availability and comparability across studies for quantitative synthesis. Results show a high degree of heterogeneity among the predictors considered for the study of WVCs risk. In total, 172 distinct predictors were extracted from the selected studies. Six predictors related to land cover, water features, and traffic volume were considered in 30-40% of the studies, while 58 predictors appeared in only one study. This first classification of predictors provides the basis for determining which variables can be integrated in the meta-analysis. This work aims to support evidence-based landscape planning and road mitigation strategies that enhance both biodiversity conservation and human safety, by providing a quantitative synthesis of landscape and spatial predictors of WVCs risk. We expect to identify which features increase or reduce collision risk across mammal species and contexts.