Ray, Justina

Understanding the spatial ecology of large carnivores in increasingly complex, multi-use landscapes is critical for effective conservation and management. Complementary to this need are robust monitoring and statistical techniques to understand the effect of bottom-up and top-down processes on wildlife population densities. However, for wide-ranging species, such knowledge is often hindered by difficulties in conducting studies over large spatial extents to fully capture the range of processes influencing populations. This thesis addresses research gaps in the above themes in the context of the American black bear (Ursus americanus) in the multi-use landscape of Ontario, Canada. First, I assess the performance of a widely adopted statistical modelling technique – spatially explicit capture-recapture (SECR) – for estimating densities of large carnivores (Chapter 2). Using simulations, I demonstrate that while SECR models are generally robust to unmodeled spatial and sex-based variation in populations, ignoring high levels of this variation can lead to bias with consequences for management and conservation. In Chapter 3, I investigate fine-scale drivers of black bear population density within study areas and forest regions by applying SECR models to a large-scale, multi-year black bear spatial capture-recapture dataset. To identify more generalizable patterns, in Chapter 4 I then assess patterns of black bear density across the province and within forest regions as a function of coarse landscape-level factors using the same datasets and assess the trade-offs between three different modeling techniques. Environmental variables were important drivers of black bear density across the province, while anthropogenic variables were more important in structuring finer-scale space use within study areas. Within forest regions these variables acted as both bottom-up and top-down processes that were consistent with ecological influences on black bear foods and intensity of human influences on the species' avoidance of developed habitats. Collectively, this thesis highlights the opportunities and challenges of working across multiple scales and over expansive landscapes within a SECR framework. Specifically, the multi-scale approach of this thesis allows for robust inference of the mechanisms structuring fine and broad scale patterns in black bear densities and offers insight to the relative influence of top-down and bottom-up forces in driving these patterns. Taken together, this thesis provides an approach for monitoring large carnivore population dynamics that can be leveraged for the species conservation and management in increasingly human-modified landscapes.

Author Keywords: animal abundance, black bear, capture-recapture, density estimation, statistical ecology, wildlife management

Central to wildlife conservation and management is the need for refined, spatially explicit knowledge on the diversity and distribution of species and the factors that drive those patterns. This is especially vital as anthropogenic disturbance threatens rapid large-scale change, even in the most remote areas of the planet. My dissertation examines theinfluence of land- and sea-scape heterogeneity on patterns of genetic differentiation, diversity, and broad-scale distributions of island-dwelling ungulates in the Arctic Archipelago. First, I investigated genetic differentiation among island populations of Peary caribou (Rangifer tarandus pearyi) in contrast to continental migratory caribou (Rangifer tarandus) and evaluated whether genetic exchange among Peary caribou island populations was limited by the availability of sea ice – both now and in the future. Differentiation among both groups was best explained by geodesic distance, revealing sea ice as an effective platform for Peary caribou movement and gene flow. With future climate warming, substantial reductions in sea ice extent were forecast which significantly increased resistance to caribou movement, particularly in summer and fall. Second, I assessed genetic population structure and diversity of northern caribou and deciphered how Island Biogeography Theory (IBT) and Central Marginal Hypothesis (CMH) could act in an archipelago where isolation is highly variable due to the dynamics of sea ice. Genetic differentiation among continental and island populations was low to moderate. In keeping with IBT and CMH, island-dwelling caribou displayed lower genetic diversity compared to mainland and mainland migratory herds; the size of islands (or population range) positively influenced genetic diversity, while distance-to-mainland and fall ice-free coastlines negatively influenced genetic diversity. Hierarchical structure analysis revealed multiple units of caribou diversity below the species level. Third, I shifted my focus to the terrestrial landscape and explored the elements governing species-environment relationships. Using species distribution models, I tested the response of caribou and muskoxen to abiotic versus abiotic + biotic predictors, and included distance to heterospecifics as a proxy for competitive interactions. Models that included biotic predictors outperformed models with abiotic predictors alone, and biotic predictors were most important when identifying habitat suitability for both ungulates. Further, areas of high habitat suitability for caribou and muskoxen were largely disjunct, limited in extent, and mainly outside protected areas. Finally, I modelled functional connectivity for two genetically and spatially disjunct groups of island-dwelling caribou. For High Arctic caribou, natural and anthropogenic features impeded gene flow (isolation-by-resistance); for Baffin Island caribou we found panmixia with absence of isolation-by-distance. Overall, my dissertation demonstrates the varying influences of contemporary land- and sea-scape heterogeneity on the distribution, diversity and differentiation of Arctic ungulates and it highlights the vulnerability of island-dwelling caribou to a rapidly changing Arctic environment.

Author Keywords: Circuitscape, connectivity, Island Biogeography, landscape genetics, population structure, species distribution models

Current major issues in conservation biology include habitat loss, fragmentation and population over-exploitation. Animals can respond to landscape change through behavioural flexibility, allowing individuals to persist in disturbed landscapes. Individual behaviour has only recently been explicitly included in population models. Carnivores may be sensitive to changing landscapes due to their wide-ranging behaviour, low densities and reproductive rates. Canada lynx (Lynx canadensis) is a primary predator of snowshoe hares (Lepus americanus). Both species range throughout the boreal forests of North America, however lynx are declining in the southern range periphery. In this dissertation, I developed new insights into the effects of habitat loss and fragmentation on lynx. In Chapter 2, I created a habitat suitability model for lynx in Ontario and examined occurrence patterns across 2 regions to determine if habitat selection is flexible when different amounts of habitat are available. Although lynx avoided areas with <30% suitable habitat where suitable land cover is abundant, I found that they have flexible habitat selection patterns where suitable land cover is rare and occurred in low habitat areas. In Chapter 3, I investigated the effects of dispersal plasticity on occupancy patterns using a spatially explicit individual-based model. I showed that flexible dispersers, capable of crossing inhospitable matrix, had higher densities and a lower risk of patch extinction. In contrast, inflexible dispersers (unable to cross inhospitable matrix), were most limited by landscape connectivity, resulting in a high extinction risk in isolated patches. I developed three predictions to be explored with empirical data; (1) dispersal plasticity affects estimates of functional connectivity; (2) variation in dispersal behaviour increases the resilience of patchy populations; and (3) dispersal behaviour promotes non-random distribution of phenotypes. Finally, in Chapter 4, I examined the consequences of anthropogenic harvest on naturally cycling populations. I found that harvest mortality can exacerbate the effects of habitat fragmentation, especially when lynx densities are low. Dynamic harvest regimes maintained lynx densities and cycle dynamics while reducing the risk of population extinction. These results suggest that lynx display some flexibility to changing landscapes and that the metapopulation structure is more resilient to increasing habitat loss and fragmentation than previously understood. Future studies should focus on determining a threshold of connectivity necessary for population persistence and examining the effects of habitat loss on the fecundity of lynx.

Author Keywords: Fluctuating Populations, Habitat Fragmentation, Landscape Ecology, Occupancy Dynamics, Population Ecology, Spatially Explicit Population Models