How do landscape context and fences influence roadkill locations of small and medium-sized mammals?

J Environ Manage. 2019 Apr 01;235:511-520

Authors: Plante J, Jaeger JAG, Desrochers A

Abstract

Road mortality is the most easily visible effect of roads and traffic on wildlife populations. Mitigation measures such as fences and wildlife passages have been applied to reduce these effects. During the widening of Quebec's Highway 175 from two to four lanes between 2006 and 2012, 33 wildlife passages designed specifically for small and medium-sized mammals were installed under the road in combination with short fences. This study examined the effectiveness of the fences at reducing the number of small and medium-sized mammals killed along a 68?km section of the road while controlling for the potential confounding effects of landscape variables. Repeated daily mortality surveys were conducted by car during the summers of 2012-2015 to measure roadkill occurrence and detection probability. A total of 893 dead animals from 13 taxa were detected. Roadkill occurrence was significantly greater at fence ends than in fenced sections and unfenced sections (fence-end effect), indicating that the fences were not long enough to discourage animals from moving along the fence to the fence ends. Greater length would be required to meet the target of reduced road mortality. Shrubby vegetation in the median strip separating the two directions of the highway was associated with high roadkill occurrence for medium-sized species. Roadkill detection probability for all species combined was 0.72, ranging from 0.17 for small mammals (<1?kg) to 0.82 for medium-sized mammals (>1?kg). To reduce road mortality, when wildlife passages are constructed along with new highways or retrofitted to old highways, fences either should be continuous or sufficiently long to encourage passage use rather than movement around the fence ends. Future road mortality studies should be combined with data about wildlife abundance and detection probability to more accurately estimate the effects at the population level.

PMID: 30711836 [PubMed - indexed for MEDLINE]