Predicting road mortality risk using life traits of birds and mammals

• Autor(a) principal: Ferreira, Flávio Zanchetta
• Data de Publicação: 2017
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/13323
• topic: Aves - Atropelamento; Mamíferos - Atropelamento; Animais - Atropelamento; Random forest regression; Birds - Road kill; Mammals - Road kill; Animals - road kill; Ecologia Aplicada

Roads are a ubiquitous feature in landscape, and wildlife-vehicle collision can affect populations influencing their long-term persistence. However, some species are more likely to be killed than others. Moreover, many species that are still unstudied or are not detected on road surveys might be affected by road mortality. To help understand why species are road-killed at distinct rates, it is useful to examine the relationship between their traits and the road-kill risk. We developed trait-based models using random forest regression to assess the role of a wide range of species’ traits on road-kill rates for bird and mammal species. We then used these models to predict risk for all bird and mammal species in Brazil. Bird road-kill rates were best explained by body mass, habitat breadth, lifespan and maturity age, whereas mammal road-kill rates were best explained by feeding behavior, home range, habitat breadth, body mass, diet breadth and maturity age. Birds with more than 2 kg and habitat generalists were positively related to high road-kill rates. Short maturity age and lifespan were also associated with high vulnerability to traffic. Mammals exhibiting scavenging feeding behavior, small and intermediate home range sizes, being habitat and diet generalists, with body masses between 3 kg and 45 kg, and earlier maturity age were more susceptible to high road-kill rates. We found that 16 bird and 14 mammal species are potentially vulnerable to road mortality. Our model contributes to a better understanding of the biological characteristics that make species particularly vulnerable to road-kill. We argue that road-kill risk assessment should focus not only on road and landscape related features, but also use the available knowledge on species traits to provide more accurate information for environmental impact assessments.