DOI: 10.1111/j.1472-4642.2008.00523.x © 2008 The Authors 222 Journal compilation © 2008 Blackwell Publishing Ltd www.blackwellpublishing.com/ddi Diversity and Distributions, (Diversity Distrib.) (2009) 15, 222–231 BIODIVERSITY RESEARCH ABSTRACT Aim To highlight the benefit of using habitat use to improve the accuracy of predictive road fatality models. Location The Snowy Mountains Highway in southern New South Wales, Australia. Methods A binary logistic regression model was constructed using wombat fatality presences and randomly generated absences. Species-specific habitat variables were included as predictors in the model selection process as well as two spatially explicit measures of wombat habitat use. Generalized additive models (GAMs) were constructed for each possible combination of predictors in R. The final model was selected by comparing all models subsets for the eight predictors and employing the one standard error rule to select the best model set. Results The final predictive model had high discriminatory power and incorporated both measures of species habitat use, greatly exceeding the variation explained by a previously published model for the same species and road. Main Conclusions Our findings highlight the importance of incorporating variables that describe habitat use by fauna for predictive modelling of animal- vehicle crashes. Reliance upon models that ignore landscape patterns are limited in their capacity to identify hotspots and inform managers of locations to engage in mitigation. Keywords Common wombats, Getis–Ord clustering, habitat use, predictive modelling, road- kill, spatial analysis, Vombatus ursinus. INTRODUCTION The adverse impacts of roads on wildlife are well documented (see reviews by Forman et al., 2003; Seiler, 2003; Coffin, 2007). Although population effects on fauna extend well beyond the boundary of the road (Reijnen et al., 1997; Gaines et al., 2005; Jaarsma et al., 2006; Ramp & Ben-Ami, 2006), fatalities of fauna killed in collisions with vehicles on the road itself are of major concern to conservationists and road managers (Forman & Alexander, 1998; Trombulak & Frissell, 2000). Recently, many quantitative models of animal–vehicle collisions have been developed (Malo et al., 2004; Saeki & Macdonald, 2004; Gaines et al., 2005; Jaeger et al., 2005; Ramp et al., 2005; Orlowski & Nowak, 2006), with the goal of providing effective mitigation techniques for management (Jaarsma et al ., 2007). These probabilistic approaches to predicting locations of animal– vehicle collisions are conducted for two primary purposes: (1) to infer those factors contributing to collisions, and (2) to identify hotspots for targeted mitigation. Driven by the need to develop feasible models, modelling approaches for predicting fatality locations have typically relied on variables that characterize the road environment; such as road sinuosity, road-verge attributes and spatial and temporal traffic variation (Finder et al ., 1999; Taylor & Goldingay, 2004; Clevenger & Waltho, 2005). Often missing, or at best generic in nature, are species-specific variables that describe how the animals in question utilize the landscape. When included, species-specific variables are often restricted to vague character- izations of landscape utilization (Jaeger et al., 2005), and often multiple species are modelled simultaneously using the same suite of generic variables (Clevenger et al., 2003; Taylor & Goldingay, 2004; Ramp et al., 2005). The biological link between these habitat variables and the fauna that are involved in collisions is never explicitly described. This oversight has significant ramifications, as the importance of understanding species-specific distributions in ecological studies in road environments has been shown for a wide range of species (Forman et al., 2002; Alexander et al., 2005; Lesbarreres et al., School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, New South Wales 2052, Australia *Correspondence: Daniel Ramp, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, New South Wales 2052, Australia. E-mail: d.ramp@unsw.edu.au Blackwell Publishing Ltd Incorporating habitat use in models of fauna fatalities on roads Erin Roger and Daniel Ramp*