Katherine A. Zeller

Using simulations to evaluate Mantel-based methods for assessing landscape resistance to gene flow

Abstract Mantel‐based tests have been the primary analytical methods for understanding how landscape features influence observed spatial genetic structure. Simulation studies examining Mantel‐based approaches have highlighted major challenges associated with the use of such tests and fueled debate on when the Mantel test is appropriate for landscape genetics studies. We aim to provide some clarity in this debate using spatially explicit, individual‐based, genetic simulations to examine the effects of the following on the performance of Mantel‐based methods: (1) landscape configuration, (2) spatial genetic nonequilibrium, (3) nonlinear relationships between genetic and cost distances, and (4) correlation among cost distances derived from competing resistance models. Under most conditions, Mantel‐based methods performed poorly. Causal modeling identified the true model only 22% of the time. Using relative support and simple Mantel r values boosted performance to approximately 50%. Across all methods, performance increased when landscapes were more fragmented, spatial genetic equilibrium was reached, and the relationship between cost distance and genetic distance was linearized. Performance depended on cost distance correlations among resistance models rather than cell‐wise resistance correlations. Given these results, we suggest that the use of Mantel tests with linearized relationships is appropriate for discriminating among resistance models that have cost distance correlations <0.85 with each other for causal modeling, or <0.95 for relative support or simple Mantel r. Because most alternative parameterizations of resistance for the same landscape variable will result in highly correlated cost distances, the use of Mantel test‐based methods to fine‐tune resistance values will often not be effective.

Multi-level, multi-scale resource selection functions and resistance surfaces for conservation planning: Pumas as a case study

The importance of examining multiple hierarchical levels when modeling resource use for wildlife has been acknowledged for decades. Multi-level resource selection functions have recently been promoted as a method to synthesize resource use across nested organizational levels into a single predictive surface. Analyzing multiple scales of selection within each hierarchical level further strengthens multi-level resource selection functions. We extend this multi-level, multi-scale framework to modeling resistance for wildlife by combining multi-scale resistance surfaces from two data types, genetic and movement. Resistance estimation has typically been conducted with one of these data types, or compared between the two. However, we contend it is not an either/or issue and that resistance may be better-modeled using a combination of resistance surfaces that represent processes at different hierarchical levels. Resistance surfaces estimated from genetic data characterize temporally broad-scale dispersal and successful breeding over generations, whereas resistance surfaces estimated from movement data represent fine-scale travel and contextualized movement decisions. We used telemetry and genetic data from a long-term study on pumas (Puma concolor) in a highly developed landscape in southern California to develop a multi-level, multi-scale resource selection function and a multi-level, multi-scale resistance surface. We used these multi-level, multi-scale surfaces to identify resource use patches and resistant kernel corridors. Across levels, we found puma avoided urban, agricultural areas, and roads and preferred riparian areas and more rugged terrain. For other landscape features, selection differed among levels, as did the scales of selection for each feature. With these results, we developed a conservation plan for one of the most isolated puma populations in the U.S. Our approach captured a wide spectrum of ecological relationships for a population, resulted in effective conservation planning, and can be readily applied to other wildlife species.

Sensitivity of resource selection and connectivity models to landscape definition

ContextThe definition of the geospatial landscape is the underlying basis for species-habitat models, yet sensitivity of habitat use inference, predicted probability surfaces, and connectivity models to landscape definition has received little attention.ObjectivesWe evaluated the sensitivity of resource selection and connectivity models to four landscape definition choices including (1) the type of geospatial layers used, (2) layer source, (3) thematic resolution, and (4) spatial grain.MethodsWe used GPS telemetry data from pumas (Puma concolor) in southern California to create multi-scale path selection function models (PathSFs) across landscapes with 2500 unique landscape definitions. To create the landscape definitions, we identified seven geospatial layers that have been shown to influence puma habitat use. We then varied the number, sources, spatial grain, and thematic resolutions of these layers to create our suite of plausible landscape definitions. We assessed how PathSF model performance (based on AIC) was affected by landscape definition and examined variability among the predicted probability of movement surfaces, connectivity models, and road crossing locations.ResultsWe found model performance was extremely sensitive to landscape definition and identified only seven top models out of our suite of definitions (<1%). Spatial grain and the number of geospatial layers selected for a landscape definition significantly affected model performance measures, with finer grains and greater numbers of layers increasing model performance.ConclusionsGiven the sensitivity of habitat use inference, predicted probability surfaces, and connectivity models to landscape definition, out results indicate the need for increased attention to landscape definition in future studies.

Estimating large carnivore populations at global scale based on spatial predictions of density and distribution – Application to the jaguar (Panthera onca)

Broad scale population estimates of declining species are desired for conservation efforts. However, for many secretive species including large carnivores, such estimates are often difficult. Based on published density estimates obtained through camera trapping, presence/absence data, and globally available predictive variables derived from satellite imagery, we modelled density and occurrence of a large carnivore, the jaguar, across the species’ entire range. We then combined these models in a hierarchical framework to estimate the total population. Our models indicate that potential jaguar density is best predicted by measures of primary productivity, with the highest densities in the most productive tropical habitats and a clear declining gradient with distance from the equator. Jaguar distribution, in contrast, is determined by the combined effects of human impacts and environmental factors: probability of jaguar occurrence increased with forest cover, mean temperature, and annual precipitation and declined with increases in human foot print index and human density. Probability of occurrence was also significantly higher for protected areas than outside of them. We estimated the world’s jaguar population at 173,000 (95% CI: 138,000–208,000) individuals, mostly concentrated in the Amazon Basin; elsewhere, populations tend to be small and fragmented. The high number of jaguars results from the large total area still occupied (almost 9 million km2) and low human densities (< 1 person/km2) coinciding with high primary productivity in the core area of jaguar range. Our results show the importance of protected areas for jaguar persistence. We conclude that combining modelling of density and distribution can reveal ecological patterns and processes at global scales, can provide robust estimates for use in species assessments, and can guide broad-scale conservation actions.

Incorporating Road Crossing Data into Vehicle Collision Risk Models for Moose (Alces americanus) in Massachusetts, USA

Wildlife–vehicle collisions are a human safety issue and may negatively impact wildlife populations. Most wildlife–vehicle collision studies predict high-risk road segments using only collision data. However, these data lack biologically relevant information such as wildlife population densities and successful road-crossing locations. We overcome this shortcoming with a new method that combines successful road crossings with vehicle collision data, to identify road segments that have both high biological relevance and high risk. We used moose (Alces americanus) road-crossing locations from 20 moose collared with Global Positioning Systems as well as moose–vehicle collision (MVC) data in the state of Massachusetts, USA, to create multi-scale resource selection functions. We predicted the probability of moose road crossings and MVCs across the road network and combined these surfaces to identify road segments that met the dual criteria of having high biological relevance and high risk for MVCs. These road segments occurred mostly on larger roadways in natural areas and were surrounded by forests, wetlands, and a heterogenous mix of land cover types. We found MVCs resulted in the mortality of 3% of the moose population in Massachusetts annually. Although there have been only three human fatalities related to MVCs in Massachusetts since 2003, the human fatality rate was one of the highest reported in the literature. The rate of MVCs relative to the size of the moose population and the risk to human safety suggest a need for road mitigation measures, such as fencing, animal detection systems, and large mammal-crossing structures on roadways in Massachusetts.