HOW LANDSCAPE PATTERNS INFLUENCE ROAD-KILL OF THREE SPECIES OF MAMMALS IN THE BRAZILIAN SAVANNA

Abstract

Roads are one of the main threats to mammal species conservation. Identifying relationships between landscape and road-kill patterns is necessary to build predictive models and to propose mitigation measures, particularly in heterogeneous landscapes. We choose three species of medium-sized mammals with high dispersal capacity and opportunistic habitat use as representatives to understand which landscape features affect road-kills: the crab-eating fox Cerdocyon thous, the maned-wolf Chrysocyon brachyurus, and the hare Lepus europaeus. The study was done over two consecutive years on the SP-225 highway in the State of São Paulo, southeastern Brazil. Road-kill data were collected on a daily basis, every three hours, by car travelling at 50-60 km/h. We used a logistic regression considering, as dependent variable, the occurrence of road-kills for each species, and as independent variables, the following landscape features: sugarcane fields cover, pasture cover, native vegetation cover, forestry cover, orange groves cover, urban area, and distance to the nearest river. Three buffer sizes were created around each road-kill occurrence to evaluate the effect of spatial scale. Model selection using Akaike Information Criterion with correction for small samples (AICc) showed that the road-kill incidence of Chrysocyon brachyurus was positive associated to urban area at larger scale, and for Cerdocyon thous and Lepus europaeus was found a positive association to forestry cover at different scales. For C. thous and L. europaeus, some models worth interpreting included water bodies area, distance to the nearest river and sugarcane fields cover. Matrix permeability and river proximity could be used to indicate sites to apply mitigation measures to road-kill for these mammal species in the Brazilian Savanna.