An assessment of expert-based marten habitat models used for forest management in Ontario

Abstract

We used marten snow tracking data and a previously developed empirical habitat model from northeastern Ontario to validate a number of expert-based, non-spatial marten habitat models. In particular, we tested the non-spatial Ontario Wildlife Habitat Analysis Model, the Boreal East Habitat Suitability Matrix (including tests of both standard forest units and development stages), and Allen's (1982) HSI model. Marten habitat use as measured by tracks in the snow was consistent with predictions of all the expert-based models, suggesting that these models correctly characterized the stand-level forest cover selected by marten in winter. Suitability ranks for individual stands derived from standard forest units and development stages also were consistent with their use by marten. The empirical model was consistent with the expert-based models in that it considered suitable forest stands to be those with tall trees dominated by spruce (Picea spp.) and balsam fir (Abies balsamea) trees, with a large amount of coarse woody debris, and high canopy closure. Our findings suggested that the expert-based models were able to characterize stand structure used by marten despite some of the models using only inputs available from stand inventories. This was accomplished because stand structural elements such as coarse woody debris were integrated into OWHAM and HSM indirectly, through relationships with stand age and species composition. Key words: boreal forest, forest inventory, habitat, habitat suitability, guidelines, Forest Ecosystem Classification, landscape, Martes americana, resource selection, snow tracking, spatial autocorrelation, stand structure