Estimating Grizzly Bear Use of Large Ungulate Carcasses With GPS Telemetry Data

Authors

  • Mike R. Ebinger University of Montana, College of Forestry and Conservation, Missoula 59812 and Montana State University, Ecology Department, Bozeman, Montana 59717
  • Mark A. Haroldson U.S. Geological Survey, Interagency Grizzly Bear Study Team, Bozeman, Montana 59717
  • Frank T. van Manen U.S. Geological Survey, Interagency Grizzly Bear Study Team, Bozeman, Montana 59717
  • Jennifer K. Fortin U.S. Geological Survey, Alaska Science Center, Anchorage, Alaska 99508
  • Shannon R. Podruzny U.S. Geological Survey, Northern Rocky Mountain Science Center, Interagency Grizzly Bear Study Team, Bozeman, Montana 59717
  • Justin E. Teisberg Grizzly Bear Recovery Program, USDI Fish and Wildlife Service, Libby, Montana 59923
  • Kerry A. Gunther Bear Management Office, Yellowstone Center for Resources, Yellowstone National Park, Wyoming 82190
  • P.J. White Yellowstone Center for Resources, Yellowstone National Park, Wyoming 82190
  • Steven L. Cain Grand Teton National Park, Moose, Wyoming 83012
  • Paul C. Cross U.S. Geological Survey, Interagency Grizzly Bear Study Team, Bozeman, Montana 59715

Abstract

Ungulate meat is among the most calorie-rich food sources available to grizzly bears  (Ursus arctos) in the greater Yellowstone ecosystem (GYE). However, the ephemeral and unpredictable nature of carcasses makes them difficult to study and their influence on grizzly bear foraging and spatial ecology is poorly understood. We developed a spatial-clustering technique specifically for detecting grizzly bear use of large ungulate carcasses using Global Positioning System (GPS) telemetry locations (n = 54 bear years). We used the DBScan algorithm to identify GPS clusters of individual bears (n = 2,038) and intersected these clusters with an independent dataset of site  visits to recent bear movement paths based from randomly selected days (n = 732 site visits; 2004–2011) resulting in 174 clusters associated with field measured bear behavior. Using a suite of predictor variables derived from GPS telemetry locations, e.g., duration of cluster, area used, activity sensor values, re-visitation rate, we used multinomial logistic regression to predict the probability of belonging to  each of the five response classes (resting, multiple-use, low-biomass carcass, high-biomass carcass, old carcass). Focusing on the high-biomass carcass category, for which our top model correctly classified 88 percent of the carcasses correctly, we applied our approach to a larger dataset of GPS data to examine trends in large-ungulate carcass using of grizzly bears in the GYE from 2002-2011. We found quantitative support for a positive effect of year and mortality adjusted white bark pine cone counts on the carcass-use index during the fall months (Sep and Oct) from 2002-2011.

Downloads

Published

2014-12-31

Issue

Section

Montana Chapter of The Wildlife Society [Abstracts]