Evelyn H. Merrill

Resource Selection Functions Based on Use–Availability Data: Theoretical Motivation and Evaluation Methods

Abstract Applications of logistic regression in a used–unused design in wildlife habitat studies often suffer from asymmetry of errors: used resource units (landscape locations) are known with certainty, whereas unused resource units might be observed to be used with greater sampling intensity. More appropriate might be to use logistic regression to estimate a resource selection function (RSF) tied to a use–availability design based on independent samples drawn from used and available resource units. We review the theoretical motivation for RSFs and show that sample “contamination” and the exponential form commonly assumed for the RSF are not concerns, contrary to recent statements by Keating and Cherry (2004; Use and interpretation of logistic regression in habitat-selection studies. Journal of Wildlife Management 68:774–789). To do this, we re-derive the use–availability likelihood and show that it can be maximized by logistic regression software. We then consider 2 case studies that illustrate our findings. For our first case study, we fit both RSFs and resource selection probability functions (RSPF) to point count data for 4 bird species with varying levels of occurrence among sample blocks. Drawing on our new derivation of the likelihood, we sample available resource units with replacement and assume overlapping distributions of used and available resource units. Irrespective of overlap, we observed approximate proportionality between predictions of a RSF and RSPF. For our second case study, we evaluate the classic use-availability design suggested by Manly et al. (2002), where availability is sampled without replacement, and we systematically introduce contamination to a sample of available units applied to RSFs for woodland caribou (Rangifer tarandus caribou). Although contamination appeared to reduce the magnitude of one RSF beta coefficient, change in magnitude exceeded sampling variation only when >20% of the available units were confirmed caribou use locations (i.e., contaminated). These empirically based simulations suggest that previously recommended sampling designs are robust to contamination. We conclude with a new validation method for evaluating predictive performance of a RSF and for assessing if the model deviates from being proportional to the probability of use of a resource unit.

Scale and heterogeneity in habitat selection by elk in Yellowstone National Park

Abstract Resource selection functions (RSF) can be used to explore the role of scale in determining patterns of habitat use. We estimated RSFs for 93 radiocollared adult female elk (Cervus canadensis) with resource availability defined at four spatial scales and two seasons in Yellowstone National Park. Habitat selection differed markedly among scales and seasonal ranges. During winter elk moved to ranges at lower elevations where snow water equivalents were low and selected landscapes with a mix of forest and open vegetation at all spatial scales. Areas of high vegetation diversity were selected at large spatial scales during summer, whereas elk selected less diverse areas on winter range. During summer elk selected forests that burned 12-14 y earlier, but they used these burns less than expected by chance during winter. Habitat selection by elk occurred at multiple spatial scales; thus, we cannot prescribe a single scale as being best for modelling habitat use by elk. Instead, selection of an appropriate scale will vary depending on the research question or management issue at hand.

Building the bridge between animal movement and population dynamics

While the mechanistic links between animal movement and population dynamics are ecologically obvious, it is much less clear when knowledge of animal movement is a prerequisite for understanding and predicting population dynamics. GPS and other technologies enable detailed tracking of animal location concurrently with acquisition of landscape data and information on individual physiology. These tools can be used to refine our understanding of the mechanistic links between behaviour and individual condition through ‘spatially informed’ movement models where time allocation to different behaviours affects individual survival and reproduction. For some species, socially informed models that address the movements and average fitness of differently sized groups and how they are affected by fission–fusion processes at relevant temporal scales are required. Furthermore, as most animals revisit some places and avoid others based on their previous experiences, we foresee the incorporation of long-term memory and intention in movement models. The way animals move has important consequences for the degree of mixing that we expect to find both within a population and between individuals of different species. The mixing rate dictates the level of detail required by models to capture the influence of heterogeneity and the dynamics of intra- and interspecific interaction.

A MULTI‐SCALE TEST OF THE FORAGE MATURATION HYPOTHESIS IN A PARTIALLY MIGRATORY UNGULATE POPULATION

The forage maturation hypothesis (FMH) proposes that ungulate migration is driven by selection for high forage quality. Because quality declines with plant maturation, but intake declines at low biomass, ungulates are predicted to select for intermediate forage biomass to maximize energy intake by following phenological gradients during the growing season. We tested the FMH in the Canadian Rocky Mountains by comparing forage availability and selection by both migrant and nonmigratory resident elk (Cervus elaphus) during three growing seasons from 2002-2004. First, we confirmed that the expected trade-off between forage quality and quantity occurred across vegetation communities. Next, we modeled forage biomass and phenology during the growing season by combining ground and remote-sensing approaches. The growing season started 2.2 days earlier every 1 km east of the continental divide, was delayed by 50 days for every 1000-m increase in elevation, and occurred 8 days earlier on south aspects. Migrant and resident selection for forage biomass was then compared across three spatial scales (across the study area, within summer home ranges, and along movement paths) using VHF and GPS telemetry locations from 119 female elk. Migrant home ranges occurred closer to the continental divide in areas of higher topographical diversity, resulting in migrants consistently selecting for intermediate biomass at the two largest scales, but not at the finest scale along movement paths. In contrast, residents selected maximum forage biomass across all spatial scales. To evaluate the consequences of selection, we compared exposure at telemetry locations of migrant and resident elk to expected forage biomass and digestibility. The expected digestibility for migrant elk in summer was 6.5% higher than for residents, which was corroborated with higher fecal nitrogen levels for migrants. The observed differences in digestibility should increase migrant elk body mass, pregnancy rates, and adult and calf survival rates. Whether bottom-up effects of improved forage quality are realized will ultimately depend on trade-offs between forage and predation. Nevertheless, this study provides comprehensive evidence that montane ungulate migration leads to greater access to higher-quality forage relative to nonmigratory congeners, as predicted by the forage maturation hypothesis, resulting primarily from large-scale selection patterns.

HABITAT SELECTION BY ELK BEFORE AND AFTER WOLF REINTRODUCTION IN YELLOWSTONE NATIONAL PARK

Abstract Prey species are thought to select habitats to obtain necessary resources while also avoiding predation. We examined whether habitat selection by elk (Cervus elaphus) changed following the reintroduction of wolves (Canis lupus) into Yellowstone National Park in 1995. Using conditional fixed-effects logistic regression to build habitat-selection models, we compared seasonal habitat selection by elk based on weekly elk radiolocations taken in 1985–1990 (without wolves) and 2000–2002 (with wolves). Fire-related habitat changes and climate likely interacted with wolf avoidance in shaping habitat selection by elk. In summer, when wolf activity was centered around dens and rendezvous sites, elk apparently avoided wolves by selecting higher elevations, less open habitat, more burned forest, and, in areas of high wolf density, steeper slopes than they had before wolf reintroduction. In winter, elk did not spatially separate themselves from wolves. Compared to the pre-wolf period, elk selected more open habitats in winter after wolf reintroduction, but did not change their selection of snow water equivalents (SWE) or slope. Elk appear to select habitats that allow them to avoid wolves during summer, but they may rely on other behavioral antipredator strategies, such as grouping, in winter. This study provides evidence that wolves can alter seasonal elk distribution and habitat selection, and demonstrates how the return of wolves to Yellowstone restores important ecosystem processes.

Foraging theory upscaled: the behavioural ecology of herbivore movement.

We outline how principles of optimal foraging developed for diet and food patch selection might be applied to movement behaviour expressed over larger spatial and temporal scales. Our focus is on large mammalian herbivores, capable of carrying global positioning system (GPS) collars operating through the seasonal cycle and dependent on vegetation resources that are fixed in space but seasonally variable in availability and nutritional value. The concept of intermittent movement leads to the recognition of distinct movement modes over a hierarchy of spatio-temporal scales. Over larger scales, periods with relatively low displacement may indicate settlement within foraging areas, habitat units or seasonal ranges. Directed movements connect these patches or places used for other activities. Selection is expressed by switches in movement mode and the intensity of utilization by the settlement period relative to the area covered. The type of benefit obtained during settlement periods may be inferred from movement patterns, local environmental features, or the diel activity schedule. Rates of movement indicate changing costs in time and energy over the seasonal cycle, between years and among regions. GPS telemetry potentially enables large-scale movement responses to changing environmental conditions to be linked to population performance.

Multiscale wolf predation risk for elk: does migration reduce risk?

While migration is hypothesized to reduce predation risk for ungulates, there have been few direct empirical tests of this hypothesis. Furthermore, few studies examined multiscale predation risk avoidance by migrant ungulates, yet recent research reveals that predator–prey interactions occur at multiple scales. We test the predation risk reduction hypothesis at two spatial scales in a partially migratory elk (Cervus elaphus) population by comparing exposure of migrant and resident elk to wolf (Canis lupus) predation risk. We used GPS and VHF telemetry data collected from 67 migrant and 44 resident elk over the summers of 2002–2004 in and adjacent to Banff National Park (BNP), Canada. We used wolf GPS and VHF telemetry data to estimate predation risk as a function of the relative probability of wolf occurrence weighted by a spatial density model that adjusted for varying pack sizes. We validated the predation risk model using independent data on wolf-killed elk, and showed that combining wolf presence and spatial density best predicted where an elk was likely to be killed. Predation risk on summer ranges of migrant elk was reduced by 70% compared to within resident elk summer ranges. Because wolves avoided areas near high human activity, however, fine-scale selection by resident elk for areas near high human activity reduced their predation risk exposure to only 15% higher than migrants, a difference significant in only one of three summers. Finally, during actual migration, elk were exposed to 1.7 times more predation risk than residents, even though migration was rapid. Our results support the hypothesis that large-scale migrations can reduce predation. However, we also show that where small-scale spatial variation in predation risk exists, nonmigratory elk may equally reduce predation risk as effectively as migrants under some circumstances.

WILLOW ON YELLOWSTONE'S NORTHERN RANGE: EVIDENCE FOR A TROPHIC CASCADE?

Reintroduction of wolves (Canis lupus) to Yellowstone National Park in 1995-1996 has been argued to promote a trophic cascade by altering elk (Cervus elaphus) density, habitat-selection patterns, and behavior that, in turn, could lead to changes within the plant communities used by elk. We sampled two species of willow (Salix boothii and S. geyeriana) on the northern winter range to determine whether (1) there was quantitative evidence of increased willow growth following wolf reintroduction, (2) browsing by elk affected willow growth, and (3) any increase in growth observed was greater than that expected by climatic and hydrological factors alone, thereby indicating a trophic cascade caused by wolves. Using stem sectioning techniques to quantify historical growth patterns we found an approximately twofold increase in stem growth-ring area following wolf reintroduction for both species of willow. This increase could not be explained by climate and hydrological factors alone; the presence of wolves on the landscape was a significant predictor of stem growth above and beyond these abiotic factors. Growth-ring area was positively correlated with the previous years ring area and negatively correlated with the percentage of twigs browsed from the stem during the winter preceding growth, indicating that elk browse impeded stem growth. Our results are consistent with the hypothesis of a behaviorally mediated trophic cascade on Yellowstones northern winter range following wolf reintroduction. We suggest that the community-altering effects of wolf restoration are an endorsement of ecological-process management in Yellowstone National Park.

Statistical Methods for Identifying Wolf Kill Sites Using Global Positioning System Locations

Abstract Accurate estimates of kill rates remain a key limitation to addressing many predator–prey questions. Past approaches for identifying kill sites of large predators, such as wolves (Canis lupus), have been limited primarily to areas with abundant winter snowfall and have required intensive ground-tracking or aerial monitoring. More recently, attempts have been made to identify clusters of locations obtained using Global Positioning System (GPS) collars on predators to identify kill sites. However, because decision rules used in determining clusters have not been consistent across studies, results are not necessarily comparable. We illustrate a space–time clustering approach to statistically define clusters of wolf GPS locations that might be wolf kill sites, and we then use binary and multinomial logistic regression to model the probability of a cluster being a non–kill site, kill site of small-bodied prey species, or kill site of a large-bodied prey species. We evaluated our approach using field visits of kills and assessed the accuracy of the models using an independent dataset. The cluster-scan approach identified 42–100% of wolf-killed prey, and top logistic regression models correctly classified 100% of kills of large-bodied prey species, but 40% of small-bodied prey species were classified as nonkills. Although knowledge of prey distribution and vulnerability may help refine this approach, identifying small-bodied prey species will likely remain problematic without intensive field efforts. We recommend that our approach be utilized with the understanding that variation in prey body size and handling time by wolves will likely have implications for the success of both the cluster scan and logistic regression components of the technique.

Are All Global Positioning System Collars Created Equal? Correcting Habitat-Induced Bias Using Three Brands in the Central Canadian Rockies

Abstract Global positioning system (GPS) collars are changing the face of wildlife research, yet they still possess biases such as habitat-induced fix-rate bias, which is a serious concern for habitat selection studies. We studied GPS bias in the Central Canadian Rockies, a critical area for wildlife conservation, to provide a statistical approach to correct GPS habitat bias for habitat selection studies using GPS collars. To model GPS habitat bias we deployed 11 different collars from 3 brands of GPS collars (Advanced Telemetry Systems [ATS], Asanti, MN; LOTEK Engineering Ltd., Newmarket, ON, Canada; and Televilt, Lindesberg, Sweden) in a random-stratified design at 86 sites across habitat and topographic conditions. We modeled the probability of obtaining a successful location, PFIX, as a function of habitat, topography, and collar brand using mixed-effects logistic regression in an information theoretic approach. For LOTEK collars, we also investigated the effect of 8 and 12 GPS channels on fix rate. The ATS collars had the highest overall fix rates (97.4%), followed by LOTEK 12 channel (94.5%), LOTEK 8 channel (85.6%), and Televilt (82.3%). Sufficient model selection uncertainty existed to warrant model averaging for logistic regression PFIX models. Collar brand influenced fix rate in all PFIX models: fix rates for ATS and LOTEK 12 channel were not statistically different, whereas LOTEK 8 channel receivers had intermediate fix rates, and Televilt had the lowest. Fix rate was reduced in aspen stands, closed coniferous stands, and sites in narrow mountainous valleys but was higher on upper mountain slopes. Slight discrepancies between fix rates from field trials and observed species fix rates (wolf [Canis lupus] and elk [Cervus elaphus]) suggest uncorrected behavioral or movement-induced bias similar to other recent studies. Regardless, the strong habitat-induced bias in GPS fix rates confirms that in our study area habitat effects are critical, especially for poorer performance brands. Based on previous studies of effects of the amount of bias on inferences, our results suggest correction for GPS bias should be mandatory for Televilt collars in the Canadian Rockies, optional for LOTEK (dependent on the no. of channels), and unnecessary for ATS. Thus, our GPS bias model will be useful to researchers using GPS collars on a variety of species throughout the Rocky Mountain cordillera.