Potential indicators of the impacts of forest management on wildlife habitat in northeastern Ontario: A multivariate application of wildlife habitat suitability matrices

Protocols to select ecological indicators of sustainable forest management will benefit from a proactive approach that identifies species likely to be most seriously impacted by management. Here, we use an objective approach that couples forest resource inventory information from logged and unlogged landscapes in northeastern Ontario with information from the provincial vertebrate habitat suitability matrix to assess habitat suitability (the amount of used and preferred habitats) in the landscapes and to identify potential indicators of the impacts of forest management activities. Because they may be most representative of potential future logged landscapes in the region, we contrasted the oldest post-clearcut landscapes in the region (1950s and 1960s era, n = 27) with unlogged landscapes (n = 16). Each landscape was a 2-km radius circle (12.6 km(2)). In light of reported invasions of hardwood species into logged conifer sites, we considered two regeneration scenarios for very young post-harvest stands: full and partial regeneration control. Logged landscapes differed markedly from unlogged landscapes with respect to forest age and forest type. These differences resulted in strong distinctions between logged and unlogged forests for bird and mammal communities, but relatively weak separation for herpetofauna. Redundancy analysis indicated that the percent variation in habitat suitability attributable to logging was 27% for birds, 17-18% for mammals, and 8-12% for herpetofauna. More than 40% of bird and mammal species showed highly significant differences in habitat suitability between the two landscape types (P<0.01), with more species having higher suitability in unlogged than logged forests. This higher habitat suitability in unlogged forests was associated with a greater amount of older forests and a higher richness of forest types in unlogged compared to logged forests. The effect of the two regeneration scenarios was discernible for all communities, but had a relatively weak effect, with the possible exception of higher snowshoe hare (Lepus americana) densities under partial compared to full regeneration control. Various lines of evidence are presented suggesting that these differences between logged and unlogged landscapes may be maintained into the future unless actions are taken to ameliorate these effects of clearcut logging. The technique presented here may prove useful in forest monitoring and adaptive management planning because it is objective, can simultaneously consider a large number of forest taxa, focuses on real rather than projected landscapes, and outlines, in a succinct way, the main habitat-related gradients in habitat suitability matrices.