Animal-vehicle collisions are undesirable to the general public, to drivers, to insurance providers, to biologists, and presumably to the animals themselves. However, traffic-induced mortality (―roadkill‖) is difficult to mitigate in large part because scientists lack the empirical data required to understand the patterns and processes associated with roadkill. Roadkill is not randomly distributed in space or in time, but what are the primary determinants of roadkill? And do they differ across organismal groups? We monitored vertebrate roadkill at 6 wetland and 6 upland sites in Indiana twice a week for a period of 20 months to determine whether roadkill occurs predominantly near one habitat compared to the other. We documented 14,439 vertebrate carcasses that were mostly distributed near wetlands. A significant fraction of the roadkill was not identifiable based on morphology alone, so we used DNA barcoding as a key element of species assignment. A large proportion of the carcasses (88%) were amphibians, a taxonomic group that has declined precipitously in recent years. Overall, these roadkill data were used along with road and habitat characteristics to develop analytical models that, in the absence of field monitoring, should be useful for predicting sites where roadkill is expected to be substantial. In the case of future highway construction, our models have the potential to help reduce the overall levels of roadkill, and that has the added benefit of translating into fewer human injuries and monetary losses caused by human-wildlife collisions.

Report Number



roadkill, amphibians, DNA barcoding, predictive model, human-wildlife collisions, SPR-3215

SPR Number


Performing Organization

Joint Transportation Research Program

Publisher Place

West Lafayette, Indiana

Date of this Version