Implications for forest management of bird-vegetation relationships in the mixedwood boreal forest of northeastern British Columbia

Species-habitat models have been used to quantify habitat quality and predict species distributions, with the goal to guide forest managers in setting harvest limits and designing silvicultural practices. However, considerable uncertainties exist regarding this ‘fine filter’ approach to forest management. In this thesis, I examine uncertainties concerning how management targets can be reliably derived from species-habitat models, appropriate measurement scales, and the financial costs of generating reliable habitat models. I address these uncertainties within the context of bird conservation in the mixedwood boreal forest, but several of my findings have general implications for the use of species-habitat models to inform forest management. The first concerns a major paradigm in conservation biology – the idea that non-linearity in species-habitat models reflect ‘threshold’ declines of species to habitat loss, thereby allowing targeted amounts of habitat protection to be set above thresholds. In Chapter 1, I reviewed 37 empirical studies attempting to quantify thresholds in habitat loss. I find that so far non-linearity in species-habitat models have not identified habitat amounts needed for species persistence. However, these relationships can still be used to set management targets, as they likely show changes in habitat quality. As an example, in Chapter 2, I related the densities of bird species to the gradient in forest composition. I used non-linearity in the relationship to delineate ‘good’ quality habitat for generalist species as stands comprised of more than 30% deciduous trees, representing a target for regeneration practices following clear-cutting of old mixedwood stands. The third chapter of this thesis addresses uncertainties concerning scale. Specifically, I show that good quality habitat for a keystone woodpecker species, the yellow-bellied sapsucker, can be measured at the stand scale by measurements made at fine scales - the characteristics of aspen trees chosen for nesting. Finally, I address cost uncertainties by showing that reliable habitat models can be generated using a less expensive roadside survey, but only for bird species that are coniferous or deciduous tree specialists. Generalist species will require surveys for birds and habitat in the forest interior, which is more time consuming and thus expensive but yields more reliable models.