The promises and perils of using molecular techniques to investigate the effects of roads on gene flow in wildlife populations

Background/Question/Methods

Roads and vehicles can increase mortality rates and/or act as barriers to movement.  These effects can reduce population sizes and isolate individuals, which in turn can cause decreased genetic diversity and increased differentiation.  Molecular methods have been used to assess the reduction of movement across the landscape, usually manifested as increased differentiation and/or the contribution of roads to overall genetic structure at the landscape level.  The majority of studies across a wide range of taxa have found evidence that roads limit gene flow, demonstrating that molecular methods are generally effective at assessing road impacts.  However, here we focus on the studies that did not find impacts.  Reasons for not detecting road impacts include correctly finding that the species are not impacted, generally because sufficient numbers successfully cross roads and maintain genetic connectivity, or because of false negatives, which are of greater conservation concern because the apparent lack of an impact where one is simply undetected can lead to an inappropriate lack of mitigation efforts.  We reviewed the relevant literature to determine if there are general characteristics of studies that found false negatives in hopes of providing greater ability to distinguish them from situations where there truly are no impacts.  

Results/Conclusions

We reviewed 89 studies on 73 species, for 103 unique species-study combinations.   The effects of roads on gene flow were reported in 90 studies (87%), and 21 (23%) of those reported no detectable effects, although some studies found effects only for larger roads or with a particular analysis.  The most common explanation (33%) was that there was sufficient connectivity, either over the road or via crossing structures, to prevent differentiation (true negatives).  A large effective population size (19%) which reduces the rate of genetic drift, or not enough time for detectable differences to accumulate (19%) were given as reasons for not finding impacts when gene flow was likely reduced (false negatives).  Improvements in study design and analysis methods show great promise to reduce these types of false negatives.  For example, recent studies demonstrate that individual-based analyses can detect changes in gene flow in fewer generations (sometimes < 2) than population-based methods, and techniques such as spatial autocorrelation can detect genetic structuring within relatively continuous populations.  Such improvements will provide a clearer understanding of when genetic methods are likely to not reflect the true road effects, helping to focus mitigation efforts to the locations and species most impacted.