Thursday, August 6, 2009 - 2:10 PM

COS 113-3: Optimizing sampling timing and location to capture genetic material from brown bears: Expert opinion is good value

Grant M. Harris1, Sean D. Farley2, Gareth J. Russell3, John Morton4, and Jeff Selinger2. (1) United States Fish and Wildlife Service, (2) Alaska Department of Fish and Game, (3) New Jersey Institute of Technology and Rutgers University, (4) US Fish and Wildlife Service

Background/Question/Methods

Rare and cryptic animals inhabiting densely vegetated or complex terrain challenge efforts to count them.  Sampling methods often rely on broad, systematic schemes to sample individuals (e.g. photography or DNA), but these approaches grow expensive over large areas. We tested if sampling at focal times and in biologically important locations would improve capture rates, and lessen effort, compared to conventional methods. Our unique approach synergized GPS telemetry data from brown bears inhabiting coastal Alaska with virtual hair-snares, to evaluate the disparate methods. Grid scenarios (3) relied on systematic, landscape-wide sampling, to intercept bear movements. Snares were placed in cell centers, randomly, or by expert opinion (5-20km spacing with 43 bear/years).  We also sampled grid systems after moving snare locations between sessions. Stream scenarios (4) had snares systematically placed (5 – 20km spacing), located by expert opinion along anadromous streams, and these stream scenarios sampled in concert with the arrival of salmon to streams (52 bear/years).  Since “captures” rely on “visits”, we defined the latter as 100m or 500m distances between a GPS fix and snare. Models contained GPS data unaltered, or standardized to the largest fix interval of 13 hours (28 total models).  Tests consisted of 5, 10 day sampling sessions.

Results/Conclusions

The sampling periods corresponded to the dates that maximized the distances between GPS fixes (movements; June 1 – July 20), or bears presence at anadromous streams (<500m; July 10 – August 28). Capture rates are compared by snare spacing and across all scenarios sharing equivalent assumptions. Sampling along streams and using expert judgment adjusted for the timing of salmon, generated capture rates 10 – 14 times higher than grid systems with comparable effort (visitation distance 100m). Overall, stream systems required half the effort of grid systems to acquire capture rates of .1 (20-40% the effort for .2). Such capture rates are necessary to get within 25% of the true population size with a 95% confidence interval (assuming 900 or 300 bears, respectively). Capture rates declined with increasing distance between snares, grid faster than stream scenarios. Moving snares within a cell increased grid capture rates, but not above stream systems. Overall, expert opinion describing where and when to sample improved capture rates and reduced efforts to get them. This increases opportunities for overcoming prohibitive cost and logistical concerns for sampling other elusive species elsewhere.