Characteristics of the boreal mixedwood forest associated with subnivean access by the American Marten (Martes americana)

Habitat selection by the American marten (Martes americana ), in studies throughout its range, has been associated with forest types that offer higher structural complexity. Such structure has been proposed to facilitate predator avoidance and access to the subnivean environment for thermoregulation and food procurement. The purpose of my study was to assess fine-scale habitat characteristics at points of subnivean access, and to use these characteristics to evaluate stands designated as reserved marten habitat according to the Forest Management Guidelines for the Provision of Marten Habitat in Ontario . In two study areas near Kapuskasing and near Thunder Bay, Ontario, I used point sampling to measure overhead canopy cover and various attributes of downed wood and dead trees, and plot sampling to describe understory woody vegetation, where 31 subnivean access points had been determined by winter tracking of marten. A case-control design and stepwise logistic regression were used to compare habitat at marten access points to habitat available in adjacent areas, using two or more reference points each 50-100 m from an access point. Overhead canopy cover ( P = 0.003), abundance of coarse woody debris (P = 0.020), and deciduous understory stem density (P =0.030) were positively associated with subnivean access. Total volume of standing dead trees (snags) and coarse woody debris in intermediate stages of decay, identified by loose bark and little to no intact fine branch structure, were negatively associated with subnivean access when estimated as volume within a plot ( P = 0.047). In habitat reserves in the Lakehead Forest, near Thunder Bay, the same characteristics were used in a forward stepwise discriminant function analysis comparing sites of used subnivean access and proximally located control points to stands designated as "good" and "fair" suitable marten habitat within marten cores. The discriminant function analysis was able to distinguish the "good" suitable habitat from the case - control model to a greater degree than the "fair" suitable habitat