In targeting mature and over-mature forests for harvesting, management in the boreal forest has resulted in a net loss of older forests that often exhibit complex structural variation and multiple cohorts of trees. Multi-cohort forest management has been proposed as a management approach for these older forests that maintains structural wildlife habitat attributes. At the stand level, the approach relies on various partial harvest techniques to emulate the range of structural variation found in natural boreal landscapes. Here, we examine the extent to which boreal bird communities respond to multi-cohort-related structural variation in boreal mixedwood forests. In particular, we test the utility of parameters of Weibull distributions fitted to stand stem diameter distributions, which have figured prominently in methods to characterize multi-cohort structure, to explain variation in the entire bird community and in various species groupings defined by feeding guilds and forest-type associations. We also compare the explanatory power of the two Weibull parameters against 21 forest structure variables and stand age. In general, Weibull parameters outperformed stand age as a correlate of bird community variation and they were significant explanatory variables for the matrix of all species and for four species groupings, whereas age was significant for only one species grouping. When one or the other Weibull parameter was significant, it also tended to be significant even when variation due to the other was partialled out, supporting the importance not only of forest stature, but also of forest heterogeneity in understanding bird community composition. Thus, we found that multi-cohort-associated structural variation was important in explaining variation among boreal bird communities, supporting the idea of silvicultural approaches that aim at diversifying stand structural characteristics.

bird habitat, forest resource inventory, canopy structure, boreal mixedwood forest, multi-cohort management, Weibull, stem diameter distribution