Effects of large-scale gold mining on habitat use and selection by American pronghorn.

Anthropogenic disturbances, including extraction of natural resources and development of alternative energy, are reducing and fragmenting habitat for wildlife across the globe. Effects of those disturbances have been explored by studying populations that migrate through oil and gas fields or alternative energy facilities. Extraction of minerals, including precious metals and lithium, is increasing rapidly in remote areas, which results in dramatically altered landscapes in areas of resident populations of wildlife. Our goal was to examine how a resident population of American pronghorn (Antilocapra americana) in the Great Basin ecosystem selected resources near a large-scale disturbance year around. We investigated how individuals selected resources around a large, open-pit gold mine. We classified levels of disturbance associated with the mine, and used a random forest model to select ecological covariates associated with habitat selection by pronghorn. We used resource selection functions to examine how disturbances affected habitat selection by pronghorn both annually and seasonally. Pronghorn strongly avoided areas of high disturbance, which included open pits, heap leach fields, rock disposal areas, and a tram. Pronghorn selected areas near roads, although selection was strongest about 2 km away. We observed relatively broad variation among individuals in selection of resources, and how they responded to the mine. The Great Basin is a mineral-rich area that continues to be exploited for natural resources, especially minerals. Sagebrush-dependent species, including pronghorn, that rely on this critical habitat were directly affected by that transformation of the landscape, which is likely to increase with expansion of the mine. As extraction of minerals from remote landscapes around the world continues to fragment habitats for wildlife, increasing our understanding of impacts of those changes on behaviors of wildlife before populations decline, may assist in the mitigation and minimization of negative impacts on mineral-rich landscapes and on wildlife populations.

Changes in selection of resources with reproductive state in a montane ungulate.

Animals select habitats based on food, water, space, and cover. Each of those components are essential to the ability of an individual to survive and reproduce in a particular habitat. Selection of resources is linked to reproductive fitness and individuals likely vary in how they select resources relative to their reproductive state: during pregnancy, while provisioning young when nutritional needs of the mother are high, but offspring are vulnerable to predation, or if they lose young to mortality. We investigated the effects of reproductive state on selection of resources by maternal female desert bighorn sheep (Ovis canadensis nelsoni) by comparing selection during the last trimester of gestation, following parturition when females were provisioning dependent young, and if the female lost an offspring. We captured, and recaptured each year, 32 female bighorn sheep at Lone Mountain, Nevada, during 2016-2018. Captured females were fit with GPS collars and those that were pregnant received vaginal implant transmitters. We used a Bayesian approach to estimate differences in selection between females provisioning and not provisioning offspring, as well as the length of time it took for females with offspring to return levels of selection similar to that observed prior to parturition. Females that were not provisioning offspring selected areas with higher risk of predation, but greater nutritional resources than those that were provisioning dependent young. When females were provisioning young immediately following parturition, females selected areas that were safe from predators, but had lower nutritional resources. Females displayed varying rates of return to selection strategies associated with access to nutritional resources as young grew and became more agile and less dependent on mothers. We observed clear and substantial shifts in selection of resources associated with reproductive state, and females exhibited tradeoffs in favor of areas that were safer from predators when provisioning dependent young despite loss of nutritional resources to support lactation. As young grew and became less vulnerable to predators, females returned to levels of selection that provided access to nutritional resources to restore somatic reserves lost during lactation.

Linking population performance to nutritional condition in an alpine ungulate.

Bighorn sheep (Ovis canadensis) can live in extremely harsh environments and subsist on submaintenance diets for much of the year. Under these conditions, energy stored as body fat serves as an essential reserve for supplementing dietary intake to meet metabolic demands of survival and reproduction. We developed equations to predict ingesta-free body fat in bighorn sheep using ultrasonography and condition scores in vivo and carcass measurements postmortem. We then used in vivo equations to investigate the relationships between body fat, pregnancy, overwinter survival, and population growth in free-ranging bighorn sheep in California and Nevada. Among 11 subpopulations that included alpine winter residents and migrants, mean ingesta-free body fat of lactating adult females during autumn ranged between 8.8% and 15.0%; mean body fat for nonlactating females ranged from 16.4% to 20.9%. In adult females, ingesta-free body fat > 7.7% during January (early in the second trimester) corresponded with a > 90% probability of pregnancy and ingesta-free body fat > 13.5% during autumn yielded a probability of overwinter survival > 90%. Mean ingesta-free body fat of lactating females in autumn was positively associated with finite rate of population increase (λ) over the subsequent year in bighorn sheep subpopulations that wintered in alpine landscapes. Bighorn sheep with ingesta-free body fat of 26% in autumn and living in alpine environments possess energy reserves sufficient to meet resting metabolism for 83 days on fat reserves alone. We demonstrated that nutritional condition can be a pervasive mechanism underlying demography in bighorn sheep and characterizes the nutritional value of their occupied ranges. Mountain sheep are capital survivors in addition to being capital breeders, and because they inhabit landscapes with extreme seasonal forage scarcity, they also can be fat reserve obligates. Quantifying nutritional condition is essential for understanding the quality of habitats, how it underpins demography, and the proximity of a population to a nutritional threshold.

A machine-learning approach for extending classical wildlife resource selection analyses.

Resource selection functions (RSFs) are tremendously valuable for ecologists and resource managers because they quantify spatial patterns in resource utilization by wildlife, thereby facilitating identification of critical habitat areas and characterizing specific habitat features that are selected or avoided. RSFs discriminate between known-use resource units (e.g., telemetry locations) and available (or randomly selected) resource units based on an array of environmental features, and in their standard form are performed using logistic regression. As generalized linear models, standard RSFs have some notable limitations, such as difficulties in accommodating nonlinear (e.g., humped or threshold) relationships and complex interactions. Increasingly, ecologists are using flexible machine-learning methods (e.g., random forests, neural networks) to overcome these limitations. Herein, we investigate the seasonal resource selection patterns of mule deer (Odocoileus hemionus) by comparing a logistic regression framework with random forest (RF), a popular machine-learning algorithm. Random forest (RF) models detected nonlinear relationships (e.g., optimal ranges for slope and elevation) and complex interactions which would have been very challenging to discover and characterize using standard model-based approaches. Compared with standard RSF models, RF models exhibited improved predictive skill, provided novel insights about resource selection patterns of mule deer, and, when projected across a relevant geographic space, manifested notable differences in predicted habitat suitability. We recommend that wildlife researchers harness the strengths of machine-learning tools like RF in addition to "classical" tools (e.g., mixed-effects logistic regression) for evaluating resource selection, especially in cases where extensive telemetry data sets are available.