9th International Deer Biology Congress

AUTHORS: Garrett M. Street, Department of Wildlife, Fisheries, and Aquaculture, Mississippi State University; Arthur R. Rodgers, Centre for Northern Forest Ecosystem Research, Ontario Ministry of Natural Resources and Forestry; Tal Avgar, Department of Integrative Biology, University of Guelph; John M. Fryxell, Department of Integrative Biology, University of Guelph

ABSTRACT: Habitat-based prediction of equilibrium density (i.e., carrying capacity) typically requires a complete characterization of the relationship between landscape configuration and animal density; however, under the correct conditions models of habitat selection (e.g., Resource-Selection Probability Functions, RSPFs) may be solved and summed over space to produce a prediction of animal abundance. This suggests that simple models of habitat selection may enable prediction of carrying capacity using commonly collected and widely available datasets of animal presence. We demonstrate this possibility using a time series of moose (Alces alces) abundance based on aerial census and hunter harvest data from which we estimate the carrying capacities of 34 Wildlife Management Units (WMUs) across the province of Ontario, Canada. We then use regression and information theoretic procedures to relate carrying capacity to environmental covariates of biological and management relevance. The best model is then used to generate predictions of carrying capacity for all WMUs. Next, we fit a RSPF to site-specific aerial survey data collected from multiple sites in central Ontario. The RSPF is solved for every 30-m pixel across all WMUs, and the predicted equilibrium density from the RSPF within a WMU is simply the sum of the probabilities across pixels divided by the area of the WMU. We then compare the predictions from the carrying capacity model to those from the RSPF to evaluate the capability of models of habitat selection to accurately predict carrying capacity. We observe that the RSPF fails to accurately predict carrying capacities; however, the direction and magnitude of difference between carrying capacity and the equilibrium density calculated from the RSPF can be predicted based on the difference in landscape characteristics between the site where the RSPF was fitted and the site onto which a prediction is made. Our findings suggest that prediction of carrying capacity from simple models of habitat selection is possible provided characterization of the habitat functional response such that model coefficients may be adjusted based on resource availability. Such models also benefit from enabling estimation of both landscape-level and local densities, allowing examination of “neighborhood carrying capacities” relevant at sub-population scales.