A Multistate Model for Polar Bear Survival and Abundance in Alaska’s Beaufort Sea
Jeff Bromaghin, David Douglas, George Durner, Kristin Simac, Todd Atwood

The Arctic Ocean continues to undergo rapid transformation to a seasonally ice-free ecosystem. As ice-specialized apex predators, polar bears (Ursus maritimus) are being challenged to cope with ongoing habitat degradation and potential changes in their prey base driven by food-web response to climate warming. Knowledge of polar bear response to ecosystem change is necessary to understand ecosystem dynamics and inform conservation decisions. In the southern Beaufort Sea (SBS) of Alaska and northwest Canada, sea ice extent has trended downward since satellite observations began in 1979 and available evidence suggests that polar bear abundance has been declining since the late 1990s. We investigated the population dynamics of polar bears in Alaska’s SBS from 2001 to 2016 using a multistate Cormack-Jolly-Seber mark-recapture model. States were defined as large geographic regions and we hypothesized that state-specific recapture probabilities and bear transitions (movements) between states would explain previously reported heterogeneity in recapture probabilities among individual bears. Estimates of state-transition probabilities were informed by location data from mark-recapture observations and satellite-telemetered bears. Our results corroborate prior findings that SBS polar bears experienced low survival and declining abundance from 2003 to at least 2006. Survival probabilities improved modestly from 2006 to 2008 and afterward rebounded to the comparatively high levels of 2001 to 2003 for the remainder of the study, except in 2012. Abundance continued to fluctuate in concert with survival from 2006 to 2015, but interannual changes were within credible interval widths so no significant trends in abundance were apparent. Abundance from 2006 to 2015 averaged 565 bears, the lowest level ever reported for the Alaska portion of the SBS subpopulation. The multistate model with geographic states provided an effective and plausible mechanism to explain heterogeneity in recapture probabilities.

Assessing Fisher Habitat Associations in a Fragmented and Managed Suburban Landscape in Albany, NY with Occupancy Modeling.
Daniel Winters, Daniel Bogan

Fisher populations are expanding across New York State, rebounding from historical overexploitation and habitat loss. Once believed to primarily inhabit dense, remote coniferous and mixed forests away from human populations, Fisher now inhabit suburban landscapes including in the Albany Pine Bush Preserve (APBP) near Albany, NY. This preserve is fragmented by urbanization, and highly managed to create early successional habitat to benefit threatened and endangered species; however, this management activity may reduce the amount of suitable habitat for fisher and reduce or exclude the species. The objective of this study was to investigate the distribution and habitat use of the fisher population inhabiting the preserve to gain insight into whether they will persist given continued habitat management for early successional, fire dependent vegetative communities. We deployed 50 camera traps, randomly distributed across 4 distinct and dominant habitat cover types in APBP for 8 weeks (January–March 2021). We reviewed photos to construct detection histories and conducted a single-season occupancy analysis with PRESENCE. The 4 selected habitats were the mixed forests, northern hardwood forests, pitch-pine (Pinus rigida) and scrub oak (Quercus ilicifolia and Q. prinoides) barrens, and the northern sandplains grassland. Each habitat was used as categorical site covariates along with a fifth covariate that documented whether the site was in a forest patch or not. From 262,453 photos, we identified 52 fisher detections at 50% of the 50 sampled locations. The top models identified mixed forest habitat being the strongest indicator for habitat use (occupancy) and detection probability. Maintaining a network of forested habitat is likely necessary to sustain fisher within the wildlife community of the preserve. Future research and modeling is necessary to determine a minimum amount of forested habitat patches necessary to sustain fisher.

Distribution and Occupancy of Fisher on Public and Private Lands in Connecticut: Preliminary Results – SRIP
Katerina Gillis, Paul Hapeman
Fishers (Pekania pennanti) are medium-sized furbearers with a wide distribution extending throughout the northeastern United States. In the 19th century, fishers became scarce due to logging, clearing for agriculture, and overexploitation. Through reintroductions, closed trapping seasons, and recolonization, fisher were able to recover throughout their range in the Northeast. In Connecticut, sighting reports and annual harvest data suggest that fishers in Connecticut may have declined in the last 15 years, particularly in the western part of the state. In order to properly manage this harvested species, it is necessary to understand their distribution, habitat use, and abundance. In this study, we conducted winter camera surveys in 41, 5km2 units between 2018 and 2020 located on public and private lands to provide information on their statewide distribution and to identify drivers of habitat use that may be associated with their apparent decline. We detected fisher in 19 out of 41 units. Out of the 26 units surveyed west of the Connecticut River, fisher were only detected in 9 units. Preliminary estimates of occupancy 0.526, 0.090 (SE), 95% CI – (0.354 – 0.693) and detectability 0.543, 0.078 (SE), 95% CI – (0.392 – 0.686) from PRESENCE were much lower than estimates of occupancy and detectability derived from a fisher population in southern Vermont considered to be abundant. These preliminary results support the apparent decline of fisher in the western part of the state. Additionally, the significantly lower estimates of occupancy and detectability in Connecticut compared to Vermont may in part, reflect differences in abundance of fisher in these two populations.  
Resource Partitioning of Sympatric Carnivores in Western North Carolina Through a Spatial and Diet Analysis – SRIP
Maya Feller, Aimee Rockhill, Liz Hillard, Joshua Johnson
Four species of sympatric carnivores occur in western North Carolina, coyote (Canis latrans), bobcat (Lynx rufus), gray fox (Urocyon cinereoargenteus), and red fox (Vulpes vulpes). The coyote has been known to suppress the population of smaller carnivores in the area, which can lead to a shift in the overall ecosystem through interspecific competition. Our objective is to better understand how carnivores are partitioning the available resources by using geospatial and diet analyses. To assess habitat use through space and time, foot traps were set in the summer and fall of 2020 and GPS collars were fitted to individuals of each species. The results will be analyzed in ArcGIS using the resource selection functions with generalized linear mixed models based on distances to habitat and landscape characteristics. To asses diet, we established scat transects within the study area and collected 68 samples in 2019 and early 2020. A subsample was taken from each sample, washed and contents visually identified. Among the four target species little to no overlap of home ranges has been observed among the canids. Many are also staying close to distinct cover types, such as the gray fox staying near dense brushy habitat and the red fox staying within 123 meters of an open field. Preliminary results of the diet analysis have found small mammals as the primary food source with some vegetation in canid samples as well. Insects have been found in 12.5% of samples and human objects such as plastics and string have been found in 8% of samples. Over the summer and fall we plan to expand the project by collecting more scat samples for a seasonal diet analysis and fitting more individuals with collars. This poster will discuss the project objectives, survey methodologies and preliminary findings.
Wildlife Conservation on the Alaskan Peninsula: Harnessing Deep Learning – SRIP
Thomas Chen
Climate change has displaced wildlife and disrupted ecosystems around the world. This effect is especially understood by local communities in Alaska, which is a state that is home to a diverse array of wildlife. Specifically, The Alaskan Peninsula is a unique refuge that provides habitats for species from brown bears, to moose, to caribou, to river otters and fish. This region, which is distinct from any other environment in the world, has been the subject of significant ecological study in recent years, particularly with global climate change threatening some species. We propose the creation of a novel benchmark dataset for large Alaskan Peninsula wildlife for the purpose of deep learning-based computer vision (computers gaining high-level insights from imagery). We take a machine learning approach by first crowdsourcing labels for this imagery, including taxon classification and bounding boxes for semantic segmentation. Then, we train a convolutional neural network, of the AlexNet architecture, on the dataset. The end goal is to enable instantaneous identification and semantic segmentation of wildlife caught in imagery (camera-trap, tourist photos, etc.), which informs conservation strategies, particularly, for endangered species. The assessment of wildlife and consequent tracking of individuals and population trends is enhanced by machine learning, which is a key asset. The Alaska Peninsula’s ecosystems attract visitors from around the world and their preservation is bolstered by novel technologies and applications such as the one we present in this work.
The Foraging Ecology of Coastal Wolves in Katmai National Park & Preserve, Alaska – SRIP
Ellen Dymit
Although wolves throughout most of their range are assumed to be obligate ungulate predators, wolf populations with access to resource subsidies from the marine system may demonstrate incredible dietary flexibility. Wolves are seen frequently on the coast of Katmai National Park & Preserve in Southwest Alaska, but their role in the nearshore system is not well understood. On the Katmai coast, black bears are completely excluded by brown bears, and beavers and moose occur at very low densities. We hypothesize that these circumstances drive coastal wolves to forage extensively in the marine system and exploit seasonal salmon availability. Katmai wolves have been observed catching salmon alongside bears, carrying sea otter carcasses at offshore haul-outs, and in one instance killing an adult harbor seal in an intertidal area. In 2021 we began a 3-year pilot study using noninvasive genetic sampling techniques and automatic digital cameras to study wolf foraging ecology on the Katmai coast. The main objectives of this study are to reconstruct coastal wolf diet through fecal metabarcoding and stable isotope analysis, document intertidal foraging activity and predation/scavenging of marine mammals using automatic digital cameras, and obtain a baseline estimate for wolf abundance in key regions by genotyping scat samples using single-nucleotide polymorphisms (SNPs). Initial metabarcoding results from 174 scats revealed striking differences in wolf diet between Katmai’s isolated coast and the park’s interior. Wolves living on the coast appear to consume ground squirrel, sea otter, seal, salmon, and flounder in abundance, while inland wolves primarily consume snowshoe hare, beaver, small rodents, and birds. Multiple data streams from scat, hair, and saliva samples collected during our 2021 summer field season (May through September) will further elucidate these regional differences in foraging ecology.
An Examination of Environmental and Temporal Impacts on Bait Site Visitation Frequency and Duration of Ursus Americanus in Wisconsin – SRIP
Arthur Young, Nathan Kluge, Cady Sartini
Multiple studies have found an inverse relationship between the abundance of the hard mast and the success rate of hunting bears with bait. Even though the interaction between hard mast production and hunter success has been examined in other states, a similar study in Wisconsin is useful because of its unique combination of hunting methods and an extended baiting season. The objective of this study is to examine how hard mast production affects the visitation of bears at bait sites in Wisconsin. Ten bait sites with trail cameras were installed in two central Wisconsin counties in 2020 and 2021 from August 3 to October 4. A hard mast index was created for the counties of Northern Wisconsin by sampling acorn abundance at the bait site for the daily movement scale, and at ten additional random locations within each county for the male and female home range scales. The number of minutes a bear was present each day and the hard mast availability were calculated for each site. Using the data from 2020, we found the strongest correlation between mast production and bear attendance at the daily movement scale (R2 = 0.99), with decreasing strength in the correlation as the scale increased (female home range, R2 = 0.51; male home range, R2 = 0.36), showing that local natural food availability may be a predictor of bait site use by bears in Wisconsin. We are currently examining the effect hard mast production has on bear visitation in conjunction with bear abundance and baiting frequency by using a zero-inflated negative binomial regression test.  
Seasonality of Habitat Selection, Dietary Composition, and Behaviors of Brown Bears Within the Nelchina Basin, Alaska – SRIP
Amanda Zuelke, Jeffrey Stetz, Dominic Demma, Jeffrey Welker
Brown bears (Ursus arctos) are a cornerstone of food webs in many terrestrial Alaska ecosystems. Expanding the understanding of these systems’ function and structure is of even more importance today due to increases in land use by humans and amplified concerns as these northern regions undergo rapid changes related to a warming climate. This study aims to analyze the seasonality of brown bear landscape use, diet proportions, and behavioral states and how they are affected by various environmental and anthropogenetic (NDVI, elevation, trail use, etc.) characteristics within the Nelchina Basin of Alaska, USA. We will quantify landscape use and habitat preferences by using high frequency GPS collar location data and will be paired with seasonal diet proportions determined using stable isotope (13C and 15N) analysis of blood, serum, and sectioned hair. In addition, we will use an animal behavioral model to determine the seasonality of foraging, resting, hunting, and kill events and their association with environmental and anthropogenic conditions. Initial landscape use analysis shows bears using a variety of habitats throughout the season with some preference for upland, mountainous areas. Preliminary diet results indicate that bears within this region consume high amounts of caribou throughout the entire non-denning season with smaller amounts of moose, salmon, and vegetation. Determination of the behavioral states and the relation to environmental and anthropogenic variables is currently in progress. As part of a larger research program on this population, we will contrast inferences from these analyses with independent assessments of bear density patterns and resource use.  Collectively, ourresults could prove be beneficial to conservation efforts and appropriate management of brown bears and prey species within the State of Alaska.
Multi-Scale Assessment of Factors Driving Black Bear Density in Human-Modified Landscapes, Ontario, Canada – SRIP
Brynn McLellan
Many large carnivores are vulnerable to changing environmental conditions and anthropogenic influences, with population declines a concern for global biodiversity. Managing large carnivore populations requires assessment of spatial patterns of population density relative to habitat and anthropogenic activities across spatial scales. For wide-ranging species, such as the American black bear (Ursus americanus), such knowledge is often hindered by inherent difficulties in conducting studies across multiple spatial scales and over large spatial extents. The Ontario Ministry of Natural Resources and Forestry (OMNRF) has monitored black bear density since 2004 using genetic spatial capture-recapture surveys across 498022 km2 of the Great Lakes–St. Lawrence (GLSL) and Boreal forest regions. We will apply spatially explicit capture-recapture methods combined with three years of OMNRF capture-recapture data collected across 66 study areas, each containing traplines of 40-44 baited barbed wire hair corrals, to assess the degree to which black bear density is driven by bottom-up and top-down factors at two scales: within black bear home-ranges and across the landscape containing all study areas. Using capture histories from 3642 individuals (1547 females, 2095 males) and scale-dependent anthropogenic, vegetation, land cover, and climate factors, we will assess the relative effect of these covariates on density patterns within and across scales, including differences between sexes. This research will provide insight into factors influencing black bear populations from one of the largest datasets collected on the species and at an extensive spatial extent. Continued human persecution and changes in Ontario black bear habitat necessitates a comprehensive understanding of the spatial structure of populations to guide land-use practices, inform sustainable harvest quotas, and predict future impacts from climate and land-use changes. 
Using Noninvasive Survey Methods to Evaluate Occupancy, Abundance, and Population Genetics of Bobcats in Western Maryland – SRIP
Kevin Lamp, Angela Holland, Jacob Haus, Kyle McCarthy, W. Gregory Shriver, Harry Spiker, Lisette Waits, Jacob Bowman
Monitoring bobcat (Lynx rufus) distribution, abundance, and population genetics are important to wildlife professionals tasked with bobcat conservation and management. In Maryland, bobcats are expanding their distribution east of historical bobcat range, likely indicating an increase in abundance and prompting interest in bobcats’ status within western Maryland. We estimated bobcat occupancy and abundance using camera trapping and fecal sampling on 3 study areas in western Maryland. On each study area, we overlaid a 5 × 8 grid network of 40 cells, each cell 5.5km2 in size. In January – March 2019 and 2020, we placed 1 camera in each cell and visited cameras weekly to collect photo data and re-bait the sites. We used a single season site occupancy model to estimate bobcat occupancy over both years and all study areas. Additionally, we used the Royle-Nichols model to estimate bobcat abundance from the camera trapping data. To collect fecal samples, we surveyed transects consisting of hiking trails, closed roads, lightly used roads, and off-road vehicle trails from May – August 2019. We used mitochondrial DNA to identify species and 10 microsatellite loci to identify individual bobcats for use in spatially explicit capture recapture models. In 2019 the camera trapping effort resulted in 105 weekly bobcat detections at 51 of 120 sites. During the 2020 camera trapping effort, we collected 78 weekly bobcat detection events at 42 of 119 sites. In 2019 we completed 5 surveys of 488km of transect, collecting 816 fecal samples for species and individual DNA analysis. Species identification yielded 243 bobcat samples from which we identified 23 individuals with ≥1 recapture and 21 individuals captured only once. This research will establish the foundational knowledge necessary to conserve and manage Maryland’s bobcat population while offering state wildlife managers multiple options to monitor bobcat distribution and demographics into the future.
A Twenty-Year Analysis on the Influence of Acorn Abundance on Black Bear Harvest in States with Varying Harvest Framework – SRIP
Nathan Kluge, Cady Sartini, Ben Sedinger, Andrew N. Tri, Colleen Olfenbuttel, CWB, OiTF Organizer
Abstract: Black bear (Ursus americanus) populations are flourishing in many parts of North America and provide a diverse hunting experience for thousands of hunters annually. Hard mast, such as acorns, makes up a large portion of a black bear’s fall diet and the abundance of mast can alter the relative harvest vulnerability of bear from year to year. States across the U.S. have varied harvest frameworks, some of which allow hunting with the aid of dogs or bait. The objectives of this study are to 1) quantify and standardize annual hard mast production for each state over the past 20 years, 2) determine if different harvest methods are affected by annual hard mast production, and 3) determine if harvest rates of different sex and median age bears correlates to mast production. Historical masting and bear harvest data is being used from California, North Carolina, Minnesota, West Virginia, and Pennsylvania. These states were specifically selected due to their long-term annual mast abundance data and differing harvest framework. Preliminary results show a positive relationship (P < 0.001; R2 = 0.35) between acorn production and black bear harvest and a negative relationship (P < 0.001; R2 = 0.38) between acorn production and median age of both sex bears when all state data are combined. The goal of this analysis is to provide information to wildlife managers who may be evaluating the value, effectiveness, and appropriateness of using masting data as an indicator of black bear harvest results. These comparative conclusions will aid states when considering various management strategies or goals and show which harvest frameworks are most impacted by fluctuations in acorn abundance. This could be increasingly important when implementing strategies such as predictive harvest modeling. 

Location: Virtual Date: November 2, 2021 Time: 11:00 am - 12:00 pm