Vertebrate road mortality estimates: Effects of sampling methods and carcass removal

The study of road mortality rates is important to identify species in need of detailed population evaluation, to determine the effectiveness of mitigation measures or to identify road network sections or road stretches where to concentrate mitigation actions. However, the impacts of roads on vertebrate mortality are usually underestimated, since detection probability and removal rates are not considered when road-kill magnitude is evaluated. In this paper, we present differential equations to estimate road-kill mortality rates taking into account carcass detectability and removal rates. The equations presented consider the periodicity and sampling methods normally used in road mortality studies and we discuss the recommended formulae to each sampling context. To exemplify outcomes of each equation we used a data set obtained during a one-year monitoring of vertebrate road-kills in southern Brazil. Our results show that there are differences in carcass detectability and removal rates among different taxonomic and body size groups. For reptiles and birds, respectively, we estimated mortality rates two and 39 times larger when considering carcass removal and detectability. Although our mortality estimates may be affected by some possible biases in sampling detection probability and removal rates, our main goal in this study was to provide a mathematical alternative to incorporate those error sources in road-kill estimates. We believe that the proposed mathematical models, by properly estimating road-kill magnitude, represent a substantial step towards an adequate evaluation of road impacts on wildlife.

► Road mortality magnitude is important to evaluate road impacts on wildlife.
► Carcass detection and removal rates are different among taxonomic and size groups.
► High carcass removal and low searchers’ detection underestimate mortality rates.
► Mathematical models can properly correct estimates of road-kill mortality rates.