No abstract

Range declines, habitat connectivity, and trapping have created conservation concern for wolverines throughout their range in North America. Previous researchers used population models and observed estimates of survival and reproduction to infer that current trapping rates limit population growth, except perhaps in the far north where trapping rates are lower. Assessing the sustainability of trapping requires demographic and abundance data that are expensive to acquire and are therefore usually only achievable for small populations, which makes generalization risky. We surveyed wolverines over a large area of southern British Columbia and Alberta, Canada, used spatial capture-recapture models to estimate density, and calculated trapping kill rates using provincial fur harvest data. Wolverine density averaged 2 wolverines/1,000 km 2 and was positively related to spring snow cover and negatively related to road density. Observed annual trapping mortality was >8.4%/year. This level of mortality is unlikely to be sustainable except in rare cases where movement rates are high among sub-populations and sizable untrapped refuges exist. Our results suggest wolverine trapping is not sustainable because our study area was fragmented by human and natural barriers and few large refuges existed. We recommend future wolverine trapping mortality be reduced by ≥50% throughout southern British Columbia and Alberta to promote population recovery.

Train-wildlife collisions can impact wildlife populations as well as create human and resource management challenges along railways. We identified locations and railroad design features associated with train-wildlife collisions (strikes) on a 134 km section of the Canadian Pacific Railroad (CPR) that travels through the Banff and Yoho National Parks. A 21-year dataset of train strikes with elk (Cervus elaphus), deer (Odocoileus spp.), American black bears (Ursus americanus) and grizzly bears (U. arctos) were compared to relative abundance estimates, and nine train and railroad variables. Train strikes and relative abundance varied spatially for elk, deer and bears. Hotspots and relative risk estimates were used to identify potential problem locations. Hotspots were defined as segments of the train line where strike counts were above the 95% confidence interval based on a Poisson distribution, and could be identified for elk and deer but not bears. Relative risk was estimated as the ratio of strike counts to that expected based on relative abundance. High relative risk locations, where more strikes occurred than were expected, were identified for elk, deer, and bears. Relative abundance was positively correlated with strikes for elk and deer but not bears. Train speed limit was positively associated with strikes for elk and deer. For bears, the number of structures (e.g., overpasses, tunnels, snow sheds and rock cuts) and bridges were positively correlated to strikes. To reduce the risk of train strikes on wildlife, our management recommendations include train speed reduction, habitat modifications and railroad design alterations.

Wildlife exclusion fencing has become a standard component of highway mitigation systems designed to reduce collisions with large mammals. Past work on the effectiveness of exclusion fencing has relied heavily on control–impact (i.e., space-for-time substitutions) and before–after study designs. These designs limit inference and may confound the effectiveness of mitigation with co-occurring process that also changes the rate of collisions. We used a replicated (n = 2 sites monitored for over 1000 km years combined) before-after-control-impact study design to assess fencing effectiveness along the Trans-Canada Highway in the Rocky Mountains of Canada. We found that collisions declined for common ungulates species (elk, mule deer, and white-tailed deer) by up to 96% but not for large carnivores. The weak response of carnivores is likely due to the combination of fence intrusions and low sample sizes. We calculated realized fencing effectiveness by applying the same change in collision rates observed at control (unfenced) sites as the expected change for adjacent fenced sections. Compared with the apparent fencing effectiveness (i.e., the difference in WVCs rates before and after fencing was installed), the realized estimates of fencing effectiveness declined by 6% at one site and increased by 10% at another site. When factoring in the cost of ungulate collisions to society, fencing provided a net economic gain within 1 year of construction. Over a 10-year period, fencing would provide a net economic gain of > $500,000 per km in reduced collisions. Our study highlights the benefits of long-term monitoring of road mitigation projects and provides evidence of fencing effectiveness for reducing wildlife–vehicle collisions involving large mammals.