Habitat and climatic associations of climate-sensitive species along a southern range boundary.

Climate change and habitat loss are recognized as important drivers of shifts in wildlife species' geographic distributions. While often considered independently, there is considerable overlap between these drivers, and understanding how they contribute to range shifts can predict future species assemblages and inform effective management. Our objective was to evaluate the impacts of habitat, climatic, and anthropogenic effects on the distributions of climate-sensitive vertebrates along a southern range boundary in Northern Michigan, USA. We combined multiple sources of occurrence data, including harvest and citizen-science data, then used hierarchical Bayesian spatial models to determine habitat and climatic associations for four climate-sensitive vertebrate species (American marten [Martes americana], snowshoe hare [Lepus americanus], ruffed grouse [Bonasa umbellus] and moose [Alces alces]). We used total basal area of at-risk forest types to represent habitat, and temperature and winter habitat indices to represent climate. Marten associated with upland spruce-fir and lowland riparian forest types, hares with lowland conifer and aspen-birch, grouse with lowland riparian hardwoods, and moose with upland spruce-fir. Species differed in climatic drivers with hares positively associated with cooler annual temperatures, moose with cooler summer temperatures and grouse with colder winter temperatures. Contrary to expectations, temperature variables outperformed winter habitat indices. Model performance varied greatly among species, as did predicted distributions along the southern edge of the Northwoods region. As multiple species were associated with lowland riparian and upland spruce-fir habitats, these results provide potential for efficient prioritization of habitat management. Both direct and indirect effects from climate change are likely to impact the distribution of climate-sensitive species in the future and the use of multiple data types and sources in the modelling of species distributions can result in more accurate predictions resulting in improved management at policy-relevant scales.

Performance of spatial capture-recapture models with repurposed data: Assessing estimator robustness for retrospective applications.

Advancements in statistical ecology offer the opportunity to gain further inferences from existing data with minimal financial cost. Spatial capture-recapture (SCR) models extend traditional capture-recapture models to incorporate spatial position of capture and enable direct estimation of animal densities across a region of interest. The additional inferences provided are both ecologically interesting and valuable for decision making, which has resulted in traditional capture-recapture data being repurposed using SCR. Yet, many capture-recapture studies were not designed for SCR and the limitations of repurposing data from such studies are rarely assessed in practice. We used simulation to evaluate the robustness of SCR for retrospectively estimating large mammal densities over a variety of scenarios using repurposed capture-recapture data collected by an asymmetrical sampling grid and covering a broad spatial extent in a heterogenous landscape. We found performance of SCR models fit using repurposed data simulated from the existing grid was not robust, but instead bias and precision of density estimates varied considerably among simulations scenarios. For example, while the smallest relatives bias of density estimates was 3%, it ranged by 14 orders of magnitude among scenarios and was most strongly influenced by detection parameters. Our results caution against the casual repurposing of non-spatial capture-recapture data using SCR and demonstrate the importance of using simulation to assessing model performance during retrospective applications.

Use of environmental DNA (eDNA) in streams to detect feral swine (Sus scrofa).

Invasive feral swine can damage ecosystems, disrupt plant and animal populations, and transmit diseases. Monitoring of feral swine populations requires expensive and labor-intensive techniques such as aerial surveys, field surveys for sign, trail cameras, and verifying landowner reports. Environmental DNA (eDNA) provides an alternative method for locating feral swine. To aid in detection of this harmful invasive species, a novel assay was developed incorporating molecular methods. From August 2017 to April 2018, water samples and stream data were collected along 400 m transects in two different stream types where swine DNA was artificially introduced to investigate potential factors affecting detection. A generalized linear model (family binomial) was used to characterize environmental conditions affecting swine DNA detection; detection was the dependent variable and stream measurements included stream type, distance downstream, water temperature, velocity, turbidity, discharge, and pH as independent variables. Parameters from the generalized linear model were deemed significant if 95% confidence intervals did not overlap 0. Detection probability for swine DNA negatively related to water temperature (ß =  - 0.21, 95% CI [-0.35 to -0.09]), with the highest detection probability (0.80) at 0 °C and lowest detection probability (0.05) at 17.9 °C water temperature. Results indicate that sampling for swine eDNA in free-flowing stream systems should occur at lower water temperatures to maximize detection probability. This study provides a foundation for further development of field and sampling techniques for utilizing eDNA as a viable alternative to monitoring a terrestrial invasive species in northern regions of the United States.

Environmental conditions for Jamestown Canyon virus correlated with population-level resource selection by white-tailed deer in a suburban landscape.

Suburban landscapes can alter spatial patterns by white-tailed deer (Odocoileus virginianus) and increase animal contact with vectors, pathogens, and humans. Close-contact relationships at a landscape level can have broad implications for disease epidemiology. From 1995-1999, we captured and radio-collared 41 deer in two suburban forest preserves in Chicago, Illinois. We collected blood to determine if animals were seronegative or seropositive for Jamestown Canyon virus and tracked deer movements within suburban habitats. We developed utilization distributions at the population-level and evaluated resource selection for seronegative and seropositive deer. We used maximum likelihood estimation for model selection via Akaike information criterion and then restricted maximum likelihood estimation to attain unbiased estimates of the parameters in the top-ranking models. The top-ranking model describing the resource selection of seronegative deer received almost the full weight of evidence (Akaike information criterion ωi = 0.93), and included the proportion of wetlands, precipitation in year t, and an interaction of the proportion of wetlands and precipitation in year t. The top-ranking model describing resource selection of seropositive deer received the full weight of evidence (Akaike information criterion ωi = 1.00). The model included distance to nearest populated place, distance to nearest river, length of road in each grid cell, precipitation in year t, and an interaction of the length of road in each grid cell and precipitation in year t. These results are valuable for mapping the spatial configuration of hotspots for Jamestown Canyon virus and could be used to educate local residents and recreationalists to reduce human exposure.

Accounting for Age Structure and Spatial Structure in Eco-Evolutionary Analyses of a Large, Mobile Vertebrate.

The idealized concept of a population is integral to ecology, evolutionary biology, and natural resource management. To make analyses tractable, most models adopt simplifying assumptions, which almost inevitably are violated by real species in nature. Here, we focus on both demographic and genetic estimates of effective population size per generation (Ne), the effective number of breeders per year (Nb), and Wright's neighborhood size (NS) for black bears (Ursus americanus) that are continuously distributed in the northern lower peninsula of Michigan, United States. We illustrate practical application of recently developed methods to account for violations of 2 common, simplifying assumptions about populations: 1) reproduction occurs in discrete generations and 2) mating occurs randomly among all individuals. We use a 9-year harvest dataset of >3300 individuals, together with genetic determination of 221 parent-offspring pairs, to estimate male and female vital rates, including age-specific survival, age-specific fecundity, and age-specific variance in fecundity (for which empirical data are rare). We find strong evidence for overdispersed variance in reproductive success of same-age individuals in both sexes, and we show that constraints on litter size have a strong influence on results. We also estimate that another life-history trait that is often ignored (skip breeding by females) has a relatively modest influence, reducing Nb by 9% and increasing Ne by 3%. We conclude that isolation by distance depresses genetic estimates of Nb, which implicitly assume a randomly mating population. Estimated demographic NS (100, based on parent-offspring dispersal) was similar to genetic NS (85, based on regression of genetic distance and geographic distance), indicating that the >36000 km2 study area includes about 4-5 black-bear neighborhoods. Results from this expansive data set provide important insight into effects of violating assumptions when estimating evolutionary parameters for long-lived, free-ranging species. In conjunction with recently developed analytical methodology, the ready availability of nonlethal DNA sampling methods and the ability to rapidly and cheaply survey many thousands of molecular markers should facilitate eco-evolutionary studies like this for many more species in nature.

Detecting black bear source-sink dynamics using individual-based genetic graphs.

Source-sink dynamics affects population connectivity, spatial genetic structure and population viability for many species. We introduce a novel approach that uses individual-based genetic graphs to identify source-sink areas within a continuously distributed population of black bears (Ursus americanus) in the northern lower peninsula (NLP) of Michigan, USA. Black bear harvest samples (n = 569, from 2002, 2006 and 2010) were genotyped at 12 microsatellite loci and locations were compared across years to identify areas of consistent occupancy over time. We compared graph metrics estimated for a genetic model with metrics from 10 ecological models to identify ecological factors that were associated with sources and sinks. We identified 62 source nodes, 16 of which represent important source areas (net flux > 0.7) and 79 sink nodes. Source strength was significantly correlated with bear local harvest density (a proxy for bear density) and habitat suitability. Additionally, resampling simulations showed our approach is robust to potential sampling bias from uneven sample dispersion. Findings demonstrate black bears in the NLP exhibit asymmetric gene flow, and individual-based genetic graphs can characterize source-sink dynamics in continuously distributed species in the absence of discrete habitat patches. Our findings warrant consideration of undetected source-sink dynamics and their implications on harvest management of game species.

Application of large-scale parentage analysis for investigating natal dispersal in highly vagile vertebrates: a case study of American black bears (Ursus americanus).

Understanding the factors that affect dispersal is a fundamental question in ecology and conservation biology, particularly as populations are faced with increasing anthropogenic impacts. Here we collected georeferenced genetic samples (n = 2,540) from three generations of black bears (Ursus americanus) harvested in a large (47,739 km2), geographically isolated population and used parentage analysis to identify mother-offspring dyads (n = 337). We quantified the effects of sex, age, habitat type and suitability, and local harvest density at the natal and settlement sites on the probability of natal dispersal, and on dispersal distances. Dispersal was male-biased (76% of males dispersed) but a small proportion (21%) of females also dispersed, and female dispersal distances (mean ± SE  =  48.9±7.7 km) were comparable to male dispersal distances (59.0±3.2 km). Dispersal probabilities and dispersal distances were greatest for bears in areas with high habitat suitability and low harvest density. The inverse relationship between dispersal and harvest density in black bears suggests that 1) intensive harvest promotes restricted dispersal, or 2) high black bear population density decreases the propensity to disperse. Multigenerational genetic data collected over large landscape scales can be a powerful means of characterizing dispersal patterns and causal associations with demographic and landscape features in wild populations of elusive and wide-ranging species.