Tuesday, August 3, 2010 - 8:00 AM

COS 23-1: The influence of landscape factors on long-term beaver site occupancy

Anna M. Harrison, State University of New York College of Environmental Science and Forestry, John C. Stella, State University of New York College of Environmental Science and Forestry, and Stacy McNulty, State University of New York College of Environmental Science and Forestry.


The recovery of beaver (Castor canadensis) populations since the cessation of widespread trapping in the early 20th century represents an important non-equilibrium disturbance process in northern forests. Beaver not only alter ecosystems by impounding water and creating ponds, they also remove woody vegetation from the surrounding terrestrial ecosystems, which changes forest community structure.  Duration and frequency of beaver occupancy is determined by landscape and forest suitability and it is enhanced by the effort and changes made by beaver.  The magnitude of beaver impacts on the landscape, and conversely, the landscape and habitat factors that sustain their long-term populations cannot be fully understood with short-term (<5 yrs) records of beaver occupancy in forests communities that change over decadal scales.  Using a multi-decadal dataset (30 yrs) we established predictors of beaver occupancy, theorizing that long-term occupancy at a site is a function of forest resources (food quality and quantity), the landscape’s capacity to support suitable beaver habitat (e.g., topography and hydrology), and the site-specific costs required for dam construction and maintenance. We used linear modeling to evaluate these influences on the duration of beaver occupancy at 14 pond and wetland sites in the central Adirondack Mountains, New York.  


In single-factor models, habitat extent (pond area and forage area) and forest resources (hardwood stand basal area) were important predictors of occupancy duration, whereas site maintenance cost (number and cumulative volume of dams) was not.  We included all non-correlated variables into alternative, plausible regression models to predict the duration of site occupancy and selected the optimal model using AIC.  The best model for predicting beaver occupancy included forage area, dam volume and hardwood stand basal area within the forage area. Models with total forest basal area (hardwoods and softwoods) were inferior to ones that included hardwoods only, thus reinforcing the role of food quality in factors supporting long-term site colonization. Site maintenance costs, as expressed by total dam volume are an important negative influence on occupancy duration when included in multi-factor models. Forage area associated with dam sites,  which we delineated in field surveys, has a strong linear relationship with pond area (R2=0.90), particularly along small stream reaches. This relationship is relevant to management, because it suggests that the substantial, but difficult to measure impacts to forest composition and structure can be estimated and modeled at a landscape scale using remotely-sensed measurements of pond area.