|Forest Investment Account|
|Abstract of FIA Project 400408|
|" ALT="See below to download" WIDTH="100" HEIGHT="129" BORDER="0" bordercolor="#000000">|
Resource selection analysis of caribouís preference for winter range in the Chase herd study area
|Author(s): Doucette, Andrea; McNay, R. Scott||Imprint: Mackenzie, B.C.: Abitibi-Consolidated Company of Canada, 2003||Subject: Forest Investment Account (FIA)||Series: Forest Investment Account (FIA) - Land Base Investment Program - Innovative|
Current policy for protecting caribou in north-central British Columbia is focused on conservation of winter range. As a preliminary task in implementing that policy, the future, 200-year, spatially-explicit supply of seasonal ranges for caribou were forecasted using a habitat supply model based on: 1) a stochastic disturbance simulator controlled by policy rules, 2) a Bayesian Belief Network (BBN) portraying the ecological factors and relationships used to estimate caribouís seasonal range preferences, 3) a Geographic Information System (GIS) to implement spatial rules, and 4) database tools to handle information transition among model components and to simulate forest succession. Predicting caribou preference for winter range depended on terrain and vegetation values that were modified by potential snow conditions and by risk of predation from wolves. We assess the winter range components of this habitat supply model, pine-lichen winter range (PLWR) and high-elevation winter range (HEWR), using winter-time locations (n = 5207) of 57 radio-collared caribou. We also combined the PLWR and HEWR models to create a Winter Range model. For each winter range model, two nodes were tested. The PREF node models habitat preference without the risk to predation by wolves while the VAL node includes predation risk. Although the combined winter range model worked well for the PREF and VAL nodes (selection rate of 50% and 48% respectively), higher selection rates for the modeled pine-lichen winter range (selection rate of 86% for PREF and 83% for VAL) provided more qualitative confidence in performance. The high-elevation winter range model was judged to be unreliable in its current form because of high error rates and low selection. We expected for the PLWR model that animals sensing predation risk (i.e., VAL node) would improve the error rates found for the PREF node. Instead, the results showed a clear increase in error rates and less predictive selection when predation risk is included in the analysis. This may suggest that caribou are not selective about their habitat with respect to perceived predation risk, but rather are more selective about food habitat quality. We concluded that despite the potential dangers of using complex, forest-level modeling to assess real-world policy options, the results for the pine-lichen winter range model allow us to move forward cautiously within a framework of continuous improvement.
Andrea M. Doucette, R. Scott McNay.
|To view this document you need
Adobe Acrobat Reader,
available free from the Adobe Web Site.
Updated August 02, 2006
Please direct questions or comments regarding publications to For.Prodres@gov.bc.ca