Abstract
Species distributions in forest-dominated landscapes are closely tied to vegetation structure and heterogeneity, which can vary across spatial scales. As Fennoscandian wolverines recolonize their historical range in boreal forests, specific structural features linked to better resources, such as prey availability, cover, and suitable denning habitats, may promote occupancy in these areas. We studied wolverine (Gulo gulo) occurrence in mainland Finland between 2009-2010 and 2018-2022. We conducted a multi-scale analysis using wildlife and field triangle data and Multi-Source National Forest Inventory (MS-NFI) remote sensing products. We applied generalized linear mixed models (GLMMs) to assess the influence of forest and landscape variables on the probability of occurrence at two spatial scales: local (3.13 km radius) and landscape (20 km radius). Occupied and unoccupied sites were distinguished by landscape fragmentation, tree volume, tree species composition, and distance to clearcuts. Sites were more likely to become occupied when forests were less fragmented and had broadleaved trees, while the probability of occurrence decreased if the total volume of trees was high or fresh clearcuts were in close proximity. Landscape scale seems to be more relevant than local scale when studying the overall forest structure's impact on wolverine occurrence. Our findings provide new insights into the occurrence of wolverines in Finnish boreal forests and could be used to aid species conservation and forest management planning.