| Sitka black-tailed deer inhabit a narrow strip of coastal temperate rainforest in Southeast Alaska's Tongass National Forest. Together with coastal British Columbia, this area comprises the largest intact coastal temperate rainforest in the world. Clearcut logging has been practiced in the Tongass since the 1950's, transforming large areas of old growth forest into even-aged seral stands. I examine the effects of this disturbance pattern on the deer inhabiting Heceta Island, a 180 km2 island on the southwestern edge of the Tongass. I report on the first randomized vegetation sampling ever carried out on the Tongass, and use the data to produce a habitat classification from the perspective of a foraging deer. This analysis indicates that, instead of the 4–6 silvicultural categories used by managers, there are seven distinct habitat types meaningful to deer. Comparing forage biomass to deer pellet densities, I show that knowledge of forage biomass in these habitat types is insufficient to predict deer habitat use. Using logistic regression, I examine seasonal and daily patterns of habitat use by individual, radio-collared deer. Habitat selection is evident at the scale of home range compositions and use of habitats within home ranges. Based on concurrent survival analysis, I conclude that patterns of habitat selection differ between deer surviving the study and those that died during the study. A Cox proportional hazards regression identifies two types of seral stands (shrub/sapling and pole-young sawtimber) as significant correlates of mortality. I also use Sitka black-tailed deer as an example of the application of dynamic optimization to predicting habitat selection based on knowledge of habitat quality, forager energetics, and predation risk. Predictions from this model indicate that deer should use a variety of habitats depending on their physiological state, and that further study of predation risk and forage quality is needed to accurately predict deer habitat selection on the Tongass. I conclude that predicting responses to habitat disturbance will best be accomplished by landscape-scale behavioral ecology. This consists of using individually based optimality models in concert with spatial models to scale-up individual decisions across populations. |
