Koetke, Lisa

Habitat loss and climate change are two of the leading causes of the global decline of biodiversity. Declines in the abundance of moose (Alces americanus) in British Columbia, Canada, in the 2000s were hypothesized to result from an interaction between a severe climate-induced insect outbreak and resulting increases in salvage harvest of affected forests. I investigated the behavioral and distributional responses of moose to forest harvesting disturbance across a range of spatial scales and tested the use of N-mixture models and camera trap data to estimate population abundance. At a fine spatial scale, I used microhistological analysis of moose fecal samples to assess the effects of logging on the diet of moose. In areas with greater intensities of forest harvesting, moose consumed fewer forbs, shrubs, and fir trees, and their diet was more diverse. These dietary responses were consistent with the Niche Expansion Hypothesis, which predicted that a generalist herbivore would eat a greater diversity of plants to compensate for decreased availability or quality of preferred forage. I used LiDAR and GPScollar data to test hypotheses that explained the use of horizontal and vertical cover by moose. Risk of predation and hunting (Direct Mortality Hypothesis) was the primary factor that influenced the use of cover. Moose used different forest structures, ranging from open to closed, depending on the threat (predation or hunting) and their response was modulated by maternal status. At a coarser spatial scale, I assessed a suite of hypotheses concerning the causes of partial migration, effects of migration on habitat used, and fitness effects of migration. Wildfire disturbance in the winter range was the primary driver of migration and most migratory moose experienced less wildfire disturbance after leaving their winter range. Migrants displaying specific movement tactics (e.g., distance and timing of migratory movements) experienced increased probability of parturition and neonate survival. Migration exposed moose to increased risk of predation but residents were more vulnerable to health-related causes of mortality. While migration provided some moose with fitness benefits, it did not fully mitigate the amount of wildfire disturbance in the summer range, particularly after severe wildfire seasons. Finally, I tested the sensitivity of estimates of population abundance produced by N-mixture models parameterized with camera trap data to ecological conditions, spatial scale of covariates, the potential for temporally non-independent detections, and model choice based on parsimony. Nmixture models produced accurate and reasonably precise estimates of abundance of moose and were robust to model formulation, the spatial scale of associated covariates, and the criteria used to define independent detection. However, I recommend avoiding measures of parsimony for selecting the model or models to generate a population estimate. In total, the results of my dissertation suggest that land management and forest harvest should maintain forest communities that vary in structure and composition. In particular, large-scale disturbance can alter the diet of moose, expose maternal moose to increased risk of mortality, and, in the case of wildfire, it could lead to a decline in migratory behaviors and populations.