Print

Long, Robert Armand

Natural Resources

2006

PhD

Carnivores typically require large areas of habitat, exist at low natural densities, and exhibit elusive behavior--rendering their study difficult. Relatively new, noninvasive survey methods, such as remote cameras and hair snares, now provide the means to collect robust data that can be used to estimate carnivore occupancy, distribution, and abundance. More recently, scat detector dogs--dogs trained to locate feces (scat) from target species--have been used to survey carnivores and other elusive species. During the summers of 2003-04, specially trained detector dogs were used to locate scat from black bears (Ursus americanus), fishers (Mattes pennanti), and bobcats (Lynx rufus) on 168 sites throughout Vermont. Five detection teams located a total of 1565 scats (747 putative black bear, 665 putative fisher, and 153 putative bobcat). Of 347 scats genetically analyzed for species identification, 179 (51.6%) yielded a positive identification, 131 (37.8%) failed to yield DNA or the DNA failed to amplify, and 37 (10.7%) yielded DNA but provided no species confirmation. For 70 survey sites where DNA confirmation of a putative fisher or bobcat scat was not possible. I assessed the probability that at least one of the scats collected at a given site was deposited by the putative target species (probability of correct ID; PID). I then assigned a detection to any site where PID for the target species was >̲ 0.90. Based on both DNA confirmations and PID values, dog surveys detected bears at 57.1% (96) of sites surveyed, fishers at 61.3% (103) of sites, and bobcats at 12.5% (21) of sites. I estimated that the mean probability of detecting the target species (when present) during a single visit to a site with a detector dog team was 0.86 for black bears, 0.95 for fishers, and 0.40 for bobcats. These detection probabilities were high when compared with single visit estimates reported for other noninvasive detection methods (e.g., remote cameras, hair snares). The probability of detecting black bears was largely unaffected by site- or visit-specific covariates. Fisher detection, however, varied by team. I found little or no effect of topographic ruggedness, vegetation density, or local weather (e.g., temperature, humidity) on detection probability for either fishers or black bears. Data sparseness precluded the modeling of bobcat detectability with site- or visit-specific covariates.
I also compared the relative efficacy of detector dogs, remote cameras, and hair snares for detecting the 3 target species at a subset of the sites visited by detector dogs. All 3 methods detected black bears; fishers and bobcats, however, were not detected by hair snares. Raw detection rates, unique detection rates, and the probability of detecting the target species (given presence) were all substantially higher for detector dogs than for cameras and hair snares. Using occupancy modeling, multi-model inference, and the detection/non-detection data collected at all sites, I modeled occurrence of each target species as a function of sitespecific variables. Black bear occurrence was related to the amount of forested land (+) and human development (-) within a 5-km radius of the survey site. After model-averaging, neither the fisher nor bobcat occurrence models contained significant covariates. However, variables contained in high-ranking fisher models included amount of coniferous forest (+), wetland cover (-), and human development (+). High-ranking bobcat models consistently contained amount of mixed forest (+), and forested wetland cover (+).