The bobcat (Lynx rufus) is the most widely distributed and abundant felid in North America, whose status is primarily monitored via harvest and associated data. I used a combination of harvest and field data to investigate factors affecting spatiotemporal dynamics of bobcat harvest and spatial ecology in Minnesota.
In chapter one, I investigate the socioeconomic and ecological factors that affected the number of bobcats harvested in Minnesota. Management of game animals requires understanding of factors that affect harvest levels. Although influenced by international law, bobcat management is the responsibility of state or provincial agencies, and jurisdictional environmental, ecological, and regulatory differences may alter which variables influence harvest. Consequently, our understanding of the factors driving bobcat harvest should be at a scale similar to that at which they are managed. I associated 32 years of bobcat harvest data from Minnesota with socioeconomic (e.g., pelt prices, license sales) and ecological variables (e.g., prey abundance, bobcat-specific index of winter severity) to determine what variables most strongly influenced annual bobcat harvest. I constructed candidate negative binomial generalized linear models based on an information-theoretic approach and used quasi-likelihood Akaike's Information Criterion adjusted for small sample size to assess the relative performance of each model. My best model suggested that annual bobcat harvest in Minnesota was positively related to the proportion of scent stations visited by bobcats and season length, and negatively related to the proportion of days when the maximum temperature remained below the bobcat's lower critical temperature. My results differ from those of other studies examining factors influencing furbearer harvest that have suggested furbearer harvest is driven primarily by pelt price, and suggest that managers can influence the annual harvest of bobcats by changing season length.
In chapter 2, I examine the factors affecting the spatial distribution of bobcat harvest in northeastern Minnesota. Understanding which factors limit species distributions is fundamental to predicting their response to and impact under environmental change. For species that are difficult to monitor, the spatial distribution of their harvest represents a common tool used for monitoring populations and can, with caution, be used to infer ecological relationships. Despite a nearly three-fold increase in abundance over the last 15 years and coincident increase in harvest, the spatial distribution of bobcat harvest in Minnesota has remained relatively static. Of particular interest is the near total absence of bobcat harvest in the northeastern portion of the state because it represents one of five regions designated critical habitat for the federally-threatened Canada lynx (Lynx canadensis), and anecdotal accounts suggest bobcats may threaten the persistence of Canada lynx populations. To explore potential explanations for the lack of bobcat harvest in this region I developed candidate binomial generalized linear models comparing townships where male and female bobcats were and were not harvested to determine whether hunter access and effort, climate, competition, prey abundance, or some combination thereof, accounted for the absence of bobcat harvest in this region. As judged by Akaike's Information Criterion corrected for small samples sizes, top-ranked models for males and females suggest that the distribution of bobcat harvest in northeastern Minnesota is primarily determined by bobcat ecology rather than hunter effort and access. The probability that a male or female bobcat was harvested in a township increased with white-tailed deer (Odocoileus virginianus), density and decreased with coyote (Canis latrans) density; harvest of females was also positively related to the proportion of a township composed of regenerating forest, an index of snowshoe hare (Lepus americanus) abundance. My results correspond with those of previous studies suggesting that bobcat populations can be suppressed by coyotes, that females are more reliant on snowshoe hare than males, and that white-tailed deer form an important component of the diet of bobcats at northern latitudes. Furthermore, my results suggest that reductions in winter-related mortality of white-tailed deer as predicted by climate change and consequent increases in deer density may remove one of the barriers to further colonization of the Arrowhead by bobcats, potentially increasing Canada lynx exposure to competition and genetic introgression.
In chapter 3 I use data from two GPS radio-collared sibling adult female bobcats and compare estimated home range and core area size to previously published studies of bobcat space use in Minnesota and Wisconsin and provide the first published estimate of space use overlap among sibling adult females. Social organization influences carnivore demography, space use, density and abundance. In bobcats, social organization is thought to be affected by multiple interacting factors including relatedness, sex, and prey and conspecific density. To provide baseline data on the effect of relatedness on bobcat social organization, I examined space use and overlap among two sibling, adult female bobcats in east-central Minnesota, and compared these results to previously published research. Estimated bobcat home range size was similar to that of previous studies, suggesting stability in home range size across several decades and reliability in our estimates. Home range and core area overlap was within the range of previous studies. Importantly, the use of two different methods for estimating core area suggested that the subjective use of the 50% utilization distribution would have underestimated core area size and overlap.
In the 4th and final chapter, I estimate how environmental features affect the suitability of habitat for bobcat reproduction and kitten survival and estimate the extent and distribution of bobcat breeding habitat in Minnesota. Distribution models have seen widespread adoption for a diversity of conservation applications because they require minimal data yet have proven highly predictive of the environmental features tolerated by animals. However, there has been limited integration of demography and distribution modeling despite empirical evidence suggesting that the environmental conditions supportive of reproduction are a subset of those supporting survival. I developed a maximum entropy distribution model of habitat suitable for bobcat reproduction and kitten survival using locations where kittens were harvested. My distribution model had good predictive ability and results suggest that the distribution of riparian forest is the preeminent environmental feature explaining the distribution of bobcat reproduction in Minnesota. Coyote abundance, row-crop agriculture and prairie were negatively associated with habitat suitability, mirroring the results of previous studies. Percentage of the study area providing suitable habitat for bobcat reproduction ranged from 23-76%, depending on the threshold used to discriminate between suitable and unsuitable habitat. Notably, I used data that agencies charged with managing bobcat populations largely already gather to develop a highly predictive model of the suitability of habitat for bobcat reproduction. Conservationists without access to harvest data can gather similar data from incidental observations of reproduction to provide better insight into the relative importance of environmental features for conservation planning and prioritization.