Abstract
The wolverine (Gulo gulo) is a cold-adapted mesocarnivore and furbearer of conservation and management focus across their North American range. Across their northern North American range wolverines are managed for harvest and have limited conservation focus. In contrast, wolverines in the southern periphery of their range where populations are smaller and more fragmented are receiving increasing conservation focus and management plans. However, potential impacts from climate change and human development like road expansion could pose threats to populations range wide. We applied genetic and genomic methods to assess connectivity of wolverine populations across their northwestern range and used genome wide data to assess whether wolverine populations were locally adapted to environmental conditions across their western range. We used our assessment of connectivity based on landscape features to identify potential vulnerabilities in wolverine populations across their northwestern range and recommended regions for future monitoring. We used our evaluation of local adaptation across the wolverine’s western range to assess their vulnerability to future environmental change and recommended source populations for planned reintroductions to part of the wolverine’s historic range in Colorado.Major technological developments in DNA sequencing methods in the past decade have increased the accessibility of studying wildlife populations with large genomic datasets. We compared population genetic analyses of 201 wolverines genotyped at thousands of SNPs and 501 wolverines genotyped at 12 microsatellite loci across Alaska, U.S. and the Yukon, Canada to assess whether genotyping with thousands of single nucleotide polymorphisms (SNPs) discovered with restriction-site associated DNA sequencing (RAD seq) improved our ability to describe population structure and patterns of gene flow. The SNP dataset detected more population structure and had greater precision in estimating isolation by distance compared to the microsatellite dataset, while both datasets produced similar estimates of genetic diversity and differentiation. The population structure identified with the SNP dataset largely aligned with geographic features and ecoregions. To further investigate the patterns of population structure, we utilized genetic distances calculated from the SNP dataset to assess isolation by resistance and isolation by environment hypotheses. We interviewed trappers to gather local knowledge on wolverine associations with different landscape types, tested landscape variables discussed with trappers and previously identified as important in the wolverine’s southern range, and modeled gene flow via a conductance surface. We assessed the relationship between genetic distances and resistance distances derived from the conductance surface and at site environmental characteristics with linear mixed effects models and found greatest support for an isolation by resistance model. Terrain ruggedness, drainages, and roads were negatively associated with gene flow and timberline was positively associated with gene flow.
With RAD sequencing, we captured genetic loci throughout the genome, including loci near genes subject to selective pressure. Thus, we assessed whether wolverine populations are locally adapted to the unique environmental characteristics across their western range, which include gradients in elevation, precipitation, temperature, and vegetation. In addition to the Alaska and Yukon wolverines, we sequenced individuals from Alberta and British Columbia, Canada, and Idaho and Montana, U.S. and assessed local adaptation with genotype-environment associations (GEAs). We defined conservation units based on patterns of connectivity and adaptive genetic variation and found support for three evolutionary significant units, including the southern group (Idaho and Montana), the middle group (Alberta, British Columbia, Southeast Alaska, and the Yukon), and the northwestern group (the rest of Alaska). We defined adaptive units based on all candidate adaptive loci identified with GEAs and found support for adaptive units in northern Alaska, the Kenai Peninsula, Southeast Alaska, the rest of Alaska and the Yukon, and the southern group (Idaho and Montana). From our identified candidate adaptive loci, candidate loci associated with elevation, temperature and precipitation were linked to genes involved in cardiovascular development, heat tolerance, thermal regulation, and fur growth and color.
Based on our assessment of population connectivity in the northwestern range, we recommend monitoring of our two most isolated and lowest genetic diversity regions which include the geographically restricted Kenai Peninsula and Southeast Alaska. Based on our model support for negative association between roads and gene flow, we also recommend implementing monitoring plans in regions were road and natural resource extraction development takes place to assess the impact of roads and human activities on wolverine movement, dispersal, and breeding. Our detection of patterns of local adaptation also highlights the potential vulnerabilities of wolverine populations to climate change and the need to consider adaptive variation in plans to reintroduce wolverines to their historic range in Colorado. We recommend that wildlife managers source from populations inhabiting regions with environmental characteristics most similar to the proposed reintroduction regions, which would include sourcing from Idaho, Montana, and Wyoming for reintroductions in Colorado.