The bison (Bison bison) of Yellowstone National Park (YNP) and Grand Teton National Park (GTNP) represent two of only three remaining populations of plains bison that have no evidence of hybridization with cattle. Therefore, these bison are an important source for ecological and genetic restoration of wild bison. Little is known regarding genetic population structure and gene flow among the Greater Yellowstone Area (GYA) bison herds. I evaluated the feasibility of fecal DNA sampling for genetic analyses of wild bison populations. I used matched blood and fecal samples from eight radio-collared bison from Hayden Valley breeding group (YNP), and multiplex polymerase chain reaction (PCR) of four microsatellite loci to assess amplification success and genotyping error rates. The amplification success rate was 92% and the genotyping error rate was 12% on average across all individuals, and loci. Exclusion of two poor quality samples from data analyses increased amplification success to 97%, and reduced the genotyping error rate to 4%. I PCR amplified a 470 bp mitochondrial DNA (mtDNA) fragment for sequencing, and successfully identified haplotypes for120 individuals. The error rate for mtDNA sequencing was 0.0005 nucleotide mis-incorporations across all samples. Sequencing and RFLP analysis of mtDNA control region from 179 fecal samples collected over two consecutive seasons was conducted to evaluate population structure among YNP breeding groups, and between GTNP and YNP bison populations. I found significant genetic distinction between YNP and GTNP bison populations (FST = 0.191, p < 0.001). The differences in haplotype frequencies between Hayden Valley and Lamar Valley breeding groups were highly significant (FST = 0.367, p < 0.001), and nearly two times greater than between GTNP and YNP thus providing evidence for at least two genetically distinct breeding groups within YNP. Differences between breeding groups remained significant even though haplotype frequencies were different between years within Hayden Valley (FST = 0.054, p < 0.05). The techniques and protocols developed have allowed high amplification success, low genotyping and sequencing error rates. This study demonstrated that non-invasive fecal DNA sampling is feasible for bison, and detected fine-scale population genetic structure in among GYA bison, suggesting female philopatry.

We studied changes in vegetation and habitat selection by endangered Sierra Nevada bighorn sheep (Ovis candensis sierrae; hereafter Sierra bighorn) for 2 years following wildfire on winter ranges in eastern California. We hypothesized that wildfire would change both forage availability and predation risk. Green forage biomass on Sierra bighorn winter ranges rebounded quickly from wildfire. Within 2 years green forage biomass was equal in burned and unburned areas, although total forage biomass was greater in unburned areas. Plants in the burn had 3% greater crude protein but equivalent digestibility and phenology. Forage composition in burned areas was forb dominated compared with unburned areas that were shrub dominated. Visibility, a measure of predation risk, was 9% greater in burned areas at a 5 m radii compared with unburned areas. We found no change in fecal nitrogen between Sierra bighorn in burned and unburned areas but there was a shift to higher diet composition of forbs in the burn. We evaluated Sierra bighorn resource selection using seasonal resource selection functions that included spatiotemporal models of forage biomass and spatial models of predation risk by cougars (Puma concolor), the main predator of Sierra bighorn. In the first year post-wildfire, Sierra bighorn increased selection for new growth herbaceous biomass in response to the reduced biomass caused by wildfire. While wildfire initially reduced total forage biomass it also created pockets of the highest new forb biomass in areas of high cougar use. These pockets attracted Sierra bighorn causing an increase in overlap with cougars in winter 2008. Sierra bighorn showed consistent selection to be near escape terrain and remained closer to escape terrain in areas of high cougar use compared to areas with low cougar use. By spring 2008 and winter and spring of 2009 Sierra bighorn strongly selected total forage biomass where cougar use was low and in areas of high cougar use, Sierra bighorn avoided total forage biomass. As a result Sierra bighorn overlap with cougar use was reduced. We advise management to consider the effects of fire on both forage availability and predation when implementing prescribed burns to benefit ungulates.

None.

Woodland caribou (Rangifer tarandus caribou) are federally listed and declining across Canada because of the cumulative impacts of human infrastructure development. The Atlin northern mountain herd, in the territory of the Taku River Tlingit First Nation (TRTFN), British Columbia, is less affected by development than southern herds. However, recent low productivity in this herd suggests that the impacts of development (i.e., roads, mines, cabins and towns) may be accumulating. To predict the cumulative impact of human development on the Atlin herd, we developed seasonal resource selection functions (RSF) at 2 spatial scales with data from 10 global positioning system collared caribou. We modeled habitat selection and assessed cumulative effects by estimating the zone of influence (ZOI) around several types of human development. At the landscape and home range scale caribou avoided the ZOI and selected pine-lichen forests in winter and alpine habitats in summer. Approximately 8 and 2% of high quality habitat was lost due to avoidance of current development at the landscape scale in winter and summer, respectively. Future development of access roads to 2 mines would cause a further loss of 1% of high quality habitat. Negotiating the complex political dynamics that surround caribou conservation often requires new approaches to management and recovery planning. The incorporation of traditional ecological knowledge (TEK) with Western science could improve efficiency of management decisions and enhance the validity and robustness of ecological inferences. Therefore, we evaluated how well RSF and TEK habitat models predicted current woodland caribou observations and compared the spatial predictions of both modeling approaches. Habitat suitability index models were generated from TEK interviews with TRTFN members. Though comparison of habitat ranks between the 2 models showed spatial discrepancies in some cases, overall, both approaches had high model performance and successfully predicted caribou occurrence. Our results suggest TEK can be used to identify caribou habitat and is a useful approach in northern ecosystems that frequently lack long-term ecological data that are needed to inform management decisions. Combining TEK-based habitat suitability index models with cumulative effects assessments will facilitate recovery goals for woodland caribou across northern Canada.

I conducted research on the demography of a harvested north Idaho black bear (Ursus americanus) population to determine the underlying dynamics of changes in population abundance, to determine how much these dynamics were driven by variation in food productivity, and to evaluate how these processes could influence inferences based on mark-recapture analysis. In cooperation with Idaho Department of Fish and Game and the USDA Forest Service, I used barb-wire corrals to collect black bear DNA during 2003-2006 in the Purcell Mountains of Idaho. We analyzed these DNA samples to determine the number of uniquely identified individuals in each year, Nu. I used a combination of both genetic and mark-recapture analyses to evaluate the sources of variation in Nu over the four years and to what extent this variation was driven by changes in productivity of foods on the landscape. Specifically, I investigated deviations of Hardy-Weinberg equilibrium and genetic substructure in relation to changes in abundance, and whether variation in vital rates were a function of changing berry productivity in the study area. I found a heterozygote deficiency and detected genetic substructure indicating I sampled ≤ 4 subpopulations within the same area over the four years (a Wahlund Effect). My mark-recapture analyses suggest this pattern was probably in response to landscape changes in summer berry abundance. My results suggest important variation in population dynamics driven by changes in food productivity, which should be considered when using mark-recapture analyses to monitor population trends for black bears.

Wolf (Canis lupus) diet can be estimated from undigested remains of prey in scats or through stable isotope analysis (SIA) of wolf hair when distinct δ13C and δ15N values of potential diet sources are known. Our objectives were to compare diet analysis methods, to estimate intra-population diet variability, and to determine proportions of prey consumed by wolves. We collected scats of 4 wolf packs in northwestern Montana from June to August 2008, and guard hairs of 45 wolves from 12 packs, May to August 2009. We calculated percent biomass consumed of deer (Odocoileus spp.), elk (Cervus canadensis), moose (Alces alces), and other items from scats, and used Pearsons chi-squared tests of proportions to measure differences among packs. We used hierarchical Bayesian stable isotope mixing models to determine diet and scales of diet variation from δ13C and δ15N values of wolves and prey. We used bootstrapped scat data, and Markov Chain Monte Carlo simulation data from stable isotopes to estimate confidence intervals of difference between results from each technique for 4 packs with matched samples. Diet results were not consistent between techniques. Deer was the most common prey item based on scats, and moose the most common based on SIA. Wolf diet was significantly different among packs based on scats, and differences among packs explained most variability in diet based on stable isotopes. We sampled 3 times as many packs for less than half the cost with SIA compared to scat analysis. Experimental data on wolf hair growth period and wolf-specific δ13C and δ15N fractionation values would provide important information for recommending the better technique.

An extinction vortex is one of the greatest threats to endangered species; when demographic, environmental, and genetic stochasticity interact with each other and with deterministic factors, such as habitat quality, to reinforce the demise of a small population. To successfully escape an extinction vortex and enable species recovery, all processes that affect endangered populations should be comprehensively assessed and incorporated into conservation plans. For my dissertation, I worked in conjunction with California Department of Fish and Game to develop a comprehensive research program to guide recovery efforts for federally endangered Sierra Nevada bighorn sheep, the rarest subspecies of mountain sheep in North America. I initiated a combination of demographic, habitat and genetic analyses to identify the stochastic and deterministic factors limiting the recovery of this subspecies, examine the relative and synergistic impacts of these factors on the performance of Sierra Nevada bighorn sheep, and the benefits of different management activities for stimulating recovery efforts. Just as the extinction vortex predicts, I found that small populations of Sierra Nevada bighorn sheep were driven by a number of stochastic and deterministic processes. Demographic, habitat, climate, predation, and genetic factors operated singly and in concert to shape the overall viability of this subspecies. The interaction of factors led to atypical demographic patterns that deviated from theoretical expectations and increased extinction risk. To alleviate extinction processes, I found that management strategies must be tailored to population-specific dynamics, targeting those vital rates and ecological drivers which have the greatest power to increase performance. Results from this study have elucidated critical aspects of Sierra Nevada bighorn sheep ecology, provided a recovery strategy for this subspecies, and supplied new quantitative tools for examining the dynamics of small and endangered populations. Ultimately, this work offers an example of assessing population viability, not in terms of probability of extinction, but in terms of quantifying conservation measures that will alleviate extinction dynamics and achieve endangered species recovery goals.

Understanding how population dynamics respond to landscape-scale disturbance and disease are crucial for effective wildlife management and conservation. Two new potential stressors on greater sage-grouse (Centrocercus urophasianus) populations in the Powder River Basin of Montana and Wyoming are coal-bed natural gas (CBNG) development and West Nile virus (WNv). I first examined how CBNG development, habitat, and other landscape features influenced trends in the abundance of displaying males and the status of sage-grouse leks. Second, I used rates of WNv-induced mortality and seroprevalence from radio-marked birds to estimate rates of WNv infection. Third, I studied the influence of female characteristics, season, and environmental variables on nest, brood, and female survival. I then used population models to estimate potential impacts of WNv on population growth. From 2001-2005, numbers of males on leks in CBNG fields declined more rapidly than leks outside CBNG. Of leks active in 1997 or later, only 38% within CBNG remained active by 2004-2005, compared to 84% of leks outside CBNG. By 2005, leks in CBNG had 46% fewer males per active lek than leks outside CBNG. Persistence of 110 leks was positively influenced by proportion sagebrush habitat within 6.4 km of the lek and negatively affected by CBNG development at multiple scales. Prohibiting CBNG development within 0.4 km of sage-grouse leks is inadequate to ensure lek persistence. From 2003-2005, minimum WNv-related mortality rates from 1 July-15 September ranged from 2.4-13.3% and maximum possible rates ranged from 8.2-28.9%. In spring 2005 and 2006, 10.3% and 1.8% respectively, of newly-captured females tested seropositive for neutralizing antibodies to WNv. Annual WNv infection rates were lower in habitats without CBNG development. Summer mortality from WNv occurred every year, decreased annual female survival rates by 0-27% per year, and reduced estimates of population growth by 7-10% per year. Changes in epizootiology of WNv and in distribution and management of surface water from CBNG development will play an important role in long-term impacts of WNv on greater sage-grouse populations in the Powder River Basin. Management should focus on eliminating man-made water sources that support breeding mosquitos known to vector the virus.

Desert bighorn sheep (Ovis canadensis mexicana) are an endangered species in New Mexico. Many of the small, isolated populations of desert bighorn are declining, and factors affecting their growth rates include low lamb recruitment and high mortality of adults due to cougar predation. No one has previously reported cause-specific mortality rates for desert bighorn lambs. My objectives were to determine the causes, extent, and timing of lamb mortality in the Fra Cristobal Mountains, New Mexico, USA. I tested 3 capture techniques during 2001 and 2002: approaching lambs on foot and restraining them by hand; jumping from a helicopter and restraining them by hand; and firing a net-gun from a helicopter. I captured 6 lambs by hand on the ground, 4 lambs by hand from the helicopter, and 11 lambs from the helicopter with a shoulder-mounted and skid-mounted net-gun. No injuries occurred to lambs or capture personnel. The hand capture technique allowed me to capture very young lambs. I then monitored lambs for mortality, and examined carcass and site characteristics to determine cause. I found that the primary proximate cause of lamb mortality was cougar predation, followed by golden eagle predation. Coyotes and bobcats did not kill lambs. Although 1 lamb died from pneumonia, disease was not a critical factor affecting lamb recruitment. I measured habitat characteristics at sites where adults and lambs were killed by cougars and paired control sites, and derived habitat characteristics at predation sites, relocation sites representing used areas, and random sites representing available areas. Visibility was lower at predation than control sites, while slope, elevation, and ruggedness were lower at predation than relocation sites, and predation sites were closer to water and roads than random sites. I suggest selective cougar control of habitual sheep killers over the short term may be an appropriate management strategy to enhance the recovery of desert bighorn populations, while recognizing the importance of carnivore populations to ecosystem health. Wildlife managers may consider prescribed burning to reduce vegetation encroachment and increase visibility and forage quantity and quality. Additionally, assessment of desert bighorn and cougar use of artificial water developments would be beneficial.

Since many cavity-excavating birds require trees with heart rot, managers routinely emphasize decayed trees in their snag retention prescriptions. Where too few trees are present with obvious indicators of decay, apparently sound trees are left to meet retention objectives. These trees may also contain decay, and may offer advantages in longevity and protection over more extensively decayed trees. Better information regarding conditions and heartwood decay-causing fungi important to heart rot in apparently sound trees would aid in retention decisions. We combined data from the 2003 Westside Reservoir and Roberts Fire burns with data from the USFS Forest Health Protections 10-year western larch (Larix occidentalis) merchantability study on the 2001 Moose Fire burn, all collected on the Flathead National Forest. We used dissection data from 284 apparently defect-free fire-killed western larch to predict the probability of heart rot related to tree age, diameter at breast height (dbh), elevation, aspect, habitat type, and sapwood-to-heartwood ratio. Fungi were isolated from trees with heart rot and identified through DNA analysis. We isolated Stereum sanguinolentum, Echinodontium tinctorium, Sistotrema brinkmannii, Antrodia serialis, Phellinus chrysoloma, and Fomitopsis cajanderi from trees in the three burn areas. All variables tested except elevation and slope were significantly associated with probability of heart rot (á=0.05), with sapwood-to-heartwood ratio and dbh showing the highest predictive power through CART analysis. These findings support large tree diameter as a critical characteristic in retaining trees most useable to wildlife, as well as retaining heart rot-causing fungi, which have undergone marked declines in other parts of the world.