Inference in road ecology research: what we know versus what we think we know

Roads and traffic impacts on wildlife populations are well documented. Three major mechanisms can cause them: reduced connectivity, increased mortality and reduced habitat quality. Researchers commonly recommend mitigation based on the mechanism they deem responsible. We reviewed the 2012-2016 literature to evaluate authors' inferences, to determine whether they explicitly acknowledge all possible mechanisms that are consistent with their results. We found 327 negative responses of wildlife to roads, from 307 studies. While most (84%) of these responses were consistent with multiple mechanisms, 60% of authors invoked a single mechanism. This indicates that many authors are over-confident in their inferences, and that the literature does not allow estimation of the relative importance of the mechanisms. We found preferences in authors' discussion of mechanisms. When all three mechanisms were consistent with the response measured, authors were 2.4 and 2.9 times as likely to infer reduced habitat quality compared to reduced connectivity or increased mortality, respectively. When both reduced connectivity and increased mortality were consistent with the response measured, authors were 5.2 times as likely to infer reduced connectivity compared to increased mortality. Given these results, road ecologists and managers are likely over-recommending mitigation for improving habitat quality and connectivity, and under-recommending measures to reduce road-kill.