Wolves and Cats

Contributed Paper
ROOM: HCCC, Room 26B

8:10AM Estimating the Abundance of Wolf Kills in Yellowstone National Park Using Multiple Observation Techniques
Matthew Metz; Paul Lukacs; Douglas Smith; Daniel Stahler; Mark Hebblewhite
Accounting for detection probability is a central need for most studies estimating abundance. One common way this is accomplished is through using multiple observers. Here, we present a novel way to integrate different methods of independent observation to estimate abundance. Specifically, we used ground-based observations, aerial observations, and Global Positioning System (GPS) wolf (Canis lupus) location data to detect > 2,300 wolf kills in Yellowstone National Park during two 30-day winter monitoring periods from 1997-2018. We used the information describing detection by each method to estimate wolf kill abundance using the Bayesian technique of data augmentation. Using the spatial location of each wolf kill, we evaluated how the probability of detecting a kill for each method was affected by distance from the road, openness (how forested a location was), whether the location was observable from a viewshed, carcass biomass, and wolf group size. Additionally, we evaluated how the time since a wolf kill ‘appeared’ on the landscape affected the probability of detection. Our analysis indicated that searching wolf GPS location ‘clusters’ most consistently detected wolf kills. Conversely, wolf kills were often missed by aerial and ground-based observations. But detection probability for these methods was affected by some of the covariates that we evaluated. Distance from the road, openness, and whether a spatial location was observable all affected detection through ground-based observations. For aerial-based observations, carcass biomass, distance from the road, and openness influenced detection. Additionally, the probability of detecting a wolf kill decreased the longer it had been on the landscape for ground and aerial-based observations. In our study, using three detection methods and accounting for detection probability provided a more accurate, and precise, estimate of the number of wolf kills and, ultimately, kill rate (i.e., number of prey killed per wolf per unit time).
8:30AM How Are the Wolves Doing in Idaho? an Integrated Population Model to the Rescue
Jon Horne; David Ausband; Mark Hurley; Jennifer Sturthers; Kayte Groth
Monitoring wolf populations through time is important for determining population viability and evaluating the effects of management actions. It is also critical information for understanding the effect that wolf predation has on prey species. As part of a larger study seeking to understand ecological drivers of elk survival in Idaho, we needed a spatially and temporally explicit estimate of wolf abundance in areas where we had monitored elk survival. However, this information proved to be extremely elusive due to sporadic pack counts through time coupled with the challenges of keeping radio-collars on a hunted population of wolves. We developed an integrated population model (IMP) to reconstruct pack abundances through time in Idaho. From 2005 – 2016, we monitored 209 wolves with GPS collars and 121 wolf packs across the state. We combined information from GPS-collared individuals with pack counts to model pack-specific abundance on a monthly basis via the IPM. As part of the IPM we obtained estimates of dispersal, harvest, and mortality other than harvest. By leveraging information from both data sources (GPS collars and pack counts) we were able to estimate changes in pack abundance through time as well as other important parameters for which we have no specific information (e.g., recruitment, observation probability). Our results will be used to provide managers a better understanding of changes in the wolf population through time as well as provide a way to measure predation risk and evaluate its influence on elk survival.
8:50AM Predicting Wolf Recolonization at Regional Scales
Nicolas Jaffe; William F. Porter
In the face of novel natural and anthropogenic conditions, wildlife agencies are under pressure to develop strategies that effectively balance conservation and stakeholder priorities. This is especially true in cases where human-wildlife conflict brings additional social, economic, and even political challenges. Consequently, tools that accurately predict changes in species ranges and population dynamics are invaluable to wildlife management and planning. While wildlife based literature offers a wide range of modeling techniques, agent based models (ABMs) have emerged as a uniquely effective method due to their “bottom-up” approach. ABMs, alternatively termed individual-based models, explore a system’s emergent properties through simulations of interacting, individual agents and their environment. In recent years, ABMs have been used to study species behavior, human-wildlife interactions, and scenario planning, making them increasingly applicable to wildlife management. In the Great Lakes region of the U.S., gray wolves (Canis lupus) have been gradually recolonizing their native range. Already controversial, further expansion of the species is almost certain raise concerns about potential ecological and economic impacts. Agent based models have been used effectively to predict gray wolf dynamics, at both the individual pack and population level. However, most wildlife management plans and policy are implemented regionally (county level) and any future actions will require scenario planning at this intermediate scale. In this study, we apply an ABM to simulate wolf population dynamics in the Upper Peninsula of Michigan. Our objectives are to predict regional rates of (1) wolf range expansion, (2) total population size, and (3) population growth rate. The model’s spatial and temporal accuracy is validated using a historical, annual dataset of regional wolf counts. Accurate simulations of this known population will identify implications of recolonization for neighboring regions and help inform wildlife policy planning.
9:10AM Estimating Density of Bobcats in Midwestern Landscapes Using Spatial Capture-Recapture Models
Edward Davis; Tim Swearingen; Robert Klaver; Charles Anderson; Christopher DePerno; Jonathan Jenks; Robert Bluett; Chris Jacques
A variety of increasingly sophisticated methods are available for estimating population density from capture-recapture studies. Among these, spatial capture-recapture (SCR) models provide a rigorous analytical technique for inference that extends standard closed population models to include a spatially explicit model by accounting for the distribution of individuals in space. Spatial capture-recapture models rely on spatial information readily available with camera data and use distance between traps and animal activity centers to model spatially explicit (i.e., camera trap) encounter probabilities and have been used in population density estimation for a range of carnivores. We used Bayesian analyses to evaluate the utility of SCR models for estimating density of bobcats in an agriculturally dominated landscape of west-central Illinois. We defined the continuous state space by overlaying the trap array on a square region extending 5 to 20 km beyond camera traps in each cardinal direction. We deployed 50 camera stations over a 77-day period from 1 February-18 April 2017. We captured 23 uniquely identifiable bobcats 115 times and recaptured these same individuals 92 times. Our analysis revealed a slight effect on the posterior distribution of density for the 5-km continuous state-space model, though posterior summary statistics for the 10-km, 15-km, and 20-km continuous state-space models were similar. Densities ranged from 1.44-1.57 bobcats per 100 km2 with a 95% posterior interval of 1.07 to 1.90. Variation in the state-space extending beyond trap arrays affect bobcat density estimates and should be sufficiently large to minimize encountering individuals with activity centers (i.e., home ranges) beyond the state-space boundary. Increased size of home ranges of bobcats across Midwestern landscapes may necessitate the use of relatively coarser survey grids in SCR models to account for frequent movements to and from the state space or whose core areas are positioned beyond camera survey unit boundaries.
9:30AM Persistence of Red Wolf Ancestry in Southwestern Louisiana Despite Decades of Unmitigated Hybridization
Lisette P. Waits; Jennifer R. Adams; John J. Cox; Sean M. Murphy
Red wolves (Canis rufus) historically occupied the southeastern US, but were restricted to southeastern Texas and southwestern Louisiana by the 1960s. Hybridization with coyotes (Canis latrans) created concerns that red wolves would be driven to extinction. Thus, ~400 canids were captured from this region during the 1970s, and 14 were chosen to found the captive red wolf population. Red wolves were presumed extinct in southwestern Louisiana following the removals, but no major efforts have been made to survey for red wolf ancestry among canids in this region. We investigated the genetic ancestry of large canids in the region of southwestern Louisiana where some of the last remaining wild red wolves were captured. Noninvasive genetic sampling was conducted from 2015–2016, and we obtained 335 fecal samples and 204 hair samples. Mitochondrial DNA (mtDNA) analysis identified 216 samples with coyote or red wolf ancestry, and we identified 32 individual canids using nuclear DNA microsatellites. We also obtained tissue samples for 17 individuals from the region. We attempted to generate genotypes at 18 microsatellite loci and a 320 base pair fragment of the mtDNA control region for these 49 individuals. Red wolf ancestry was estimated from the microsatellite data using Bayesian clustering analysis including genotypes of the red wolf founders, coyotes, gray wolves and dogs. MtDNA sequences were matched to the single haplotype observed in the red wolf founders. Red wolf mtDNA or nDNA ancestry was confirmed in 10 (20%) of the individuals in southwestern Louisiana. Five individuals had >18% red wolf ancestry (19-78%) based on nDNA analyses. Thus, pure red wolves and hybrids with high proportions of red wolf ancestry likely remained in southwestern Louisiana following the 1970s removals. Conservation actions may be warranted for extant canids in southwestern Louisiana, and additional research is needed on canids in this region.


Contributed Paper
Location: Huntington Convention Center of Cleveland Date: October 8, 2018 Time: 8:10 am - 9:50 am