Tabitha A. Graves

Current approaches using genetic distances produce poor estimates of landscape resistance to interindividual dispersal

Landscape resistance reflects how difficult it is for genes to move across an area with particular attributes (e.g. land cover, slope). An increasingly popular approach to estimate resistance uses Mantel and partial Mantel tests or causal modelling to relate observed genetic distances to effective distances under alternative sets of resistance parameters. Relatively few alternative sets of resistance parameters are tested, leading to relatively poor coverage of the parameter space. Although this approach does not explicitly model key stochastic processes of gene flow, including mating, dispersal, drift and inheritance, bias and precision of the resulting resistance parameters have not been assessed. We formally describe the most commonly used model as a set of equations and provide a formal approach for estimating resistance parameters. Our optimization finds the maximum Mantel r when an optimum exists and identifies the same resistance values as current approaches when the alternatives evaluated are near the optimum. Unfortunately, even where an optimum existed, estimates from the most commonly used model were imprecise and were typically much smaller than the simulated true resistance to dispersal. Causal modelling using Mantel significance tests also typically failed to support the true resistance to dispersal values. For a large range of scenarios, current approaches using a simple correlational model between genetic and effective distances do not yield accurate estimates of resistance to dispersal. We suggest that analysts consider the processes important to gene flow for their study species, model those processes explicitly and evaluate the quality of estimates resulting from their model.

Biotic impacts of energy development from shale: research priorities and knowledge gaps

11 Although shale drilling operations for oil and natural gas have increased greatly in the past decade, few studies directly quantify the impacts of shale development on plants and wildlife. We evaluate knowledge gaps related to shale development and prioritize research needs using a quantitative framework that includes spatial and tem- poral extent, mitigation difficulty, and current level of understanding. Identified threats to biota from shale development include: surface and groundwater contamination; diminished stream flow; stream siltation; habitat loss and fragmentation; localized air, noise, and light pollution; climate change; and cumulative impacts. We find the highest research priorities to be probabilistic threats (underground chemical migration; contaminant release during storage, during disposal, or from accidents; and cumulative impacts), the study of which will require major scientific coordination among researchers, industry, and government decision makers. Taken together, our research prioritization outlines a way forward to better understand how energy development affects the natural world.

Estimating landscape resistance to dispersal

Dispersal is an inherently spatial process that can be affected by habitat conditions in sites encountered by dispersers. Understanding landscape resistance to dispersal is important in connectivity studies and reserve design, but most existing methods use resistance functions with cost parameters that are subjectively chosen by the investigator. We develop an analytic approach allowing for direct estimation of resistance parameters that folds least cost path methods typically used in simulation approaches into a formal statistical model of dispersal distributions. The core of our model is a frequency distribution of dispersal distances expressed as least cost distance rather than Euclidean distance, and which includes terms for feature-specific costs to dispersal and sex (or other traits) of the disperser. The model requires only origin and settlement locations for multiple individuals, such as might be obtained from mark–recapture studies or parentage analyses, and maps of the relevant habitat features. To evaluate whether the model can estimate parameters correctly, we fit our model to data from simulated dispersers in three kinds of landscapes (in which resistance of environmental variables was categorical, continuous with a patchy configuration, or continuous in a trend pattern). We found maximum likelihood estimators of resistance and individual trait parameters to be approximately unbiased with moderate sample sizes. We applied the model to a small grizzly bear dataset to demonstrate how this approach could be used when the primary interest is in the prediction of costs and found that estimates were consistent with expectations based on bear ecology. Our method has important practical applications for testing hypotheses about dispersal ecology and can be used to inform connectivity planning efforts, via the resistance estimates and confidence intervals, which can be used to create a data-driven resistance surface.

The influence of landscape characteristics and home-range size on the quantification of landscape-genetics relationships

A common approach used to estimate landscape resistance involves comparing correlations of ecological and genetic distances calculated among individuals of a species. However, the location of sampled individuals may contain some degree of spatial uncertainty due to the natural variation of animals moving through their home range or measurement error in plant or animal locations. In this study, we evaluate the ways that spatial uncertainty, landscape characteristics, and genetic stochasticity interact to influence the strength and variability of conclusions about landscape-genetics relationships. We used a neutral landscape model to generate 45 landscapes composed of habitat and non-habitat, varying in percent habitat, aggregation, and structural connectivity (patch cohesion). We created true and alternate locations for 500 individuals, calculated ecological distances (least-cost paths), and simulated genetic distances among individuals. We compared correlations between ecological distances for true and alternate locations. We then simulated genotypes at 15 neutral loci and investigated whether the same influences could be detected in simple Mantel tests and while controlling for the effects of isolation-by-distance using the partial Mantel test. Spatial uncertainty interacted with the percentage of habitat in the landscape, but led to only small reductions in correlations. Furthermore, the strongest correlations occurred with low percent habitat, high aggregation, and low to intermediate levels of cohesion. Overall genetic stochasticity was relatively low and was influenced by landscape characteristics.

Balancing Precision and Risk: Should Multiple Detection Methods Be Analyzed Separately in N-Mixture Models?

Using multiple detection methods can increase the number, kind, and distribution of individuals sampled, which may increase accuracy and precision and reduce cost of population abundance estimates. However, when variables influencing abundance are of interest, if individuals detected via different methods are influenced by the landscape differently, separate analysis of multiple detection methods may be more appropriate. We evaluated the effects of combining two detection methods on the identification of variables important to local abundance using detections of grizzly bears with hair traps (systematic) and bear rubs (opportunistic). We used hierarchical abundance models (N-mixture models) with separate model components for each detection method. If both methods sample the same population, the use of either data set alone should (1) lead to the selection of the same variables as important and (2) provide similar estimates of relative local abundance. We hypothesized that the inclusion of 2 detection methods versus either method alone should (3) yield more support for variables identified in single method analyses (i.e. fewer variables and models with greater weight), and (4) improve precision of covariate estimates for variables selected in both separate and combined analyses because sample size is larger. As expected, joint analysis of both methods increased precision as well as certainty in variable and model selection. However, the single-method analyses identified different variables and the resulting predicted abundances had different spatial distributions. We recommend comparing single-method and jointly modeled results to identify the presence of individual heterogeneity between detection methods in N-mixture models, along with consideration of detection probabilities, correlations among variables, and tolerance to risk of failing to identify variables important to a subset of the population. The benefits of increased precision should be weighed against those risks. The analysis framework presented here will be useful for other species exhibiting heterogeneity by detection method.