Wildlife road kills in southeastern Brazil: A spatiotemporal analysis towards mitigation

Loss of wildlife due to road kills is a critical problem in South America that may lead to a decline in species populations and ultimately threaten biodiversity. In Brazil alone animal-vehicle collisions kill about 1.3 million vertebrates every day, making it crucial to develop studies that identify determinants of such mortalities. During 2007, we investigated the spatial and temporal factors of road kills on two connecting highways in southeastern Brazil. We conducted weekly surveys (60 km/h) by vehicle along 160 km of two connecting state highways (total n=8320 km), passing through Cerrado and Atlantic Forest biomes, seasonal forest, farms and urban areas. For each roadkill, we recorded coordinates (GPS), highway and roadside attributes (road design, topography, hard shoulder, constructions, vegetation and slope variables) as spatial determinants. To evaluate seasonality we applied analysis of variance combining monthly roadkill data in two treatments (dry cool versus rainy warm season). We performed logistic regression to investigate the relation of road kills to landscape and road attributes presence or absence. For multiple category variables (slope-design and topography), we used ANOVA with months as replicates. Additionally to evaluate spatial aggregation, a Ripley K-statistics simulation with a 95% confidence interval was applied. We recorded 615 road kills (3.84 roadkills.km-1.year-1): 57.8% birds (67 species), 23.7% mammals (23 species), 11.1% amphibians (four species), and 7.5% reptiles (20 species). There were proportionately more reptile road kills during the rainy season. Straight road stretches were associated with a doubling risk of roadkill (p<0.001) and vegetation with 26% increase (p<0.05). Road deaths were more frequent on straight/level stretches (p<0.01), and on straight/downhill stretches (p<0.01). In terms of topography, level (p<0.01) and hillside (p<0.01) road terrain was also associated with increased wildlife deaths. The aggregation size for all carcasses was within a 0-46 km radius (with a peak at 35 km); for amphibians it was 0-39 km and 46-97 km (62 km) related to water sources, 69-89 km for reptiles (70 km), 0-46 km for mammals (15 km) and 0-9 km for birds (3.5 km).  The high number of road kills is closely related to traffic conditions such as speed and traffic intensity, and this is further affected by weather and roadside conditions, that favour animal presence. Results suggest that straight stretches of road that encourage high-speed traffic increase the likelihood of animal-vehicle collisions, especially where dense roadside vegetation may restrict drivers’ vision. For mitigation, clearing roadside vegetation and installing speed reduction devices on straight stretches of road would likely reduce the number of wildlife deaths. Large extension of carcasses aggregation (0-97 Km) in four taxonomic categories turns inviable a specialized mitigation measure (e.g. wildlife fences). Therefore, on wildlife habitat cut by roads on level or sloping vegetated terrain, mitigation might entail the use of fences coupled with under or overpasses for wildlife crossing. The construction of new roads should follow the topography of the landscape and include the safety features mentioned above, to avoid wildlife loss to vehicle collisions.

traffic mortality; determinants; vegetation