Bighorn sheep (Ovis canadensis) experienced widespread population declines and extirpations during the late 19th and early 20th centuries, and recovery has been limited in scope and magnitude. We explored the molecular legacy of this history by comparing genetic diversity and structure in a native Rocky Mountain bighorn sheep (O. c. canadensis) metapopulation in central Idaho to genetic variation in samples collected from the same area up to 138 years earlier and to mitochondrial DNA haplotypes in bighorn sheep populations across western North America. The contemporary Idaho metapopulation was structured into 6 genetic groups with varying levels of admixture that largely corresponded with 5 population management units as defined by the distribution of bighorn sheep and suitable habitat. Nuclear DNA microsatellite diversity (expected heterozygosity He) was higher in historical samples from a private collection dated between 1879 and 1985 from the Lower Salmon (He 0.756) and Middle Fork Salmon populations (He 0.735) than in contemporary samples (Lower Salmon He 0.677, Middle Fork He 0.720). The Lower Salmon historical samples also had higher mitochondrial DNA (mtDNA) haplotype diversity (Hd 0.972), and more private microsatellite alleles (11) than contemporary samples from Lower Salmon (Hd 0.792, 0 private alleles). Thirteen Idaho mtDNA haplotypes were assigned to a haplogroup containing desert (O. c. nelsoni, O. c. mexicana, O. c. weemsi) and Sierra Nevada (O. c. sierrae) bighorn sheep lineages, consistent with gene flow between Rocky Mountain and desert sheep subspecies. This haplogroup contained 35

Wildlife managers have traditionally relied on hunting to manage elk Cervus canadensis population abundance; however, problems with elk over-abundance and/or distributions have arisen across the western US as private landowners restrict public hunting and refuges are created. In the Sapphire Mountains of western Montana, USA, we investigated the effects of an amenity refuge expansion on elk distribution during the hunting season and contrasted this with a common public land management strategy aimed at increasing elk security through road closures. We used GPS collar data collected from male and female elk for two years before and two years after the expansion of an amenity refuge that restricted hunter access to 31 km2 of habitat. We also evaluated elk responses to the closure of 41 km of motorcycle routes and 4 km of vehicle routes on US Forest Service (USFS) lands, implemented concurrently with the refuge expansion. Male elk use of and selection for the expanded refuge increased within the first four years of the expansion. The odds of male elk selection of the refuge in the third and fourth years after expansion were 545% higher (95% CI = 397, 738%) than the years prior to expansion. The odds of male elk selecting public hunting grounds decreased by an additional 48% (95% CI = 37, 57%) after route closures, and both male and female elk predominantly used and increased their selection for private lands with varying degrees of hunter access restrictions. Amenity refuges can rapidly alter elk distribution, while public land route closures may not. Amenity refuges challenge wildlife management. To meet this challenge, new management strategies must be developed in landscapes with increasing amenity refuges, and the effectiveness of public land road closures, as typically applied, needs examination.

Group living is found in only 10-15% of carnivorans and can shape demographic processes. Sociality is associated with benefits including increased ability to acquire resources, decreased risk of mortality, and increased reproductive success. We hypothesized that carnivore group size is influenced by conditions related to competition, prey, and mortality risk, which should affect benefits and costs of sociality and resulting demographic processes. We evaluated our hypotheses with gray wolves (Canis lupus) using a 14-year dataset from a large, heavily managed population in the northern Rocky Mountains, USA. Annual mean group size ranged 4.86-7.03 and averaged 5.92 overall. Most groups were relatively small, with 80% containing <= 8 members. Groups were larger in areas with higher densities of conspecific groups, and smaller where prey availability was low. Group sizes remained largely stable while the population was unharvested or under low-intensity harvest but declined under high-intensity harvest. Results support the hypothesis that as habitat becomes saturated, inclusive fitness may become increasingly important such that subordinates delay dispersal. In addition to direct implications for birth and deaths, conditions related to prey and mortality risk may also influence dispersal decisions. Our work also provided a model to predict group size of wolves in our system, directly fulfilling a management need.

Ungulate behavior is often characterized as balancing selection for forage and avoidance of predation risk. Within partially migratory ungulate populations, this balancing occurs across multiple spatial scales, potentially resulting in different exposure to costs and benefits between migrants and residents. We assessed how availability and selection of forage and risk from predators varied between summer ranges of migrant and resident mule deer (Odocoileus hemionus; a species in which individual migratory strategies are generally fixed for life) in 3 study areas in western Montana, USA, during summers 2017-2019. We hypothesized that mule deer would face a tradeoff between selecting forage and avoiding predation risk, and that migration and residency would pose contrasting availability of forage and risk at a broad (summer range) spatial scale. We hypothesized deer exposed to lower forage at a given spatial scale would compensate for reduced availability by increasing selection of forage at the cost of reduced avoidance of predators, a mechanism whereby migrants and residents could potentially achieve similar exposure to forage despite disparate availability. We compared the availability of forage (kcal/m(2)) and predation risk from wolves (Canis lupus) and mountain lions (Puma concolor) between summer ranges of each migratory strategy, then assessed how selection for those factors at the home range (second order) and within-home range (third order) scales varied using resource selection functions (RSFs). As forage availability increased among mule deer summer ranges and individual home ranges, selection for forage decreased at the second-order (P = 0.052) and third-order (P = 0.081) scales, respectively, but avoidance of predators varied weakly. In 1 study area, summer range of residents contained lower forage and higher risk than summer range of migrants, but residents compensated for this disadvantage through stronger selection of forage and avoidance of risk at finer spatial scales. In the other 2 study areas, summer range of migrants contained lower forage and higher risk than residents, but migrants did not compensate through stronger selection for beneficial resources. The majority of mule deer in our study system were migratory, though the benefits of migration were unclear, suggesting partial migration may persist in populations even when exposure to forage and predation risk appears unequal between strategies.

Land and wildlife managers use disturbance to reset forests to earlier successional stages and improve the quality and quantity of forage available to ungulates. However, management of vegetation communities to increase forage nutrition has additional implications for other non-forage vegetation. To evaluate responses of vegetation to management of fire and timber harvest, we measured and modeled 3 metrics of vegetative response in 3 forested areas in Montana, USA during 2017-2019: forage nutrition for mule deer (Odocoileus hemionus), invasive species biomass, and floristic quality, a measure of plant communities' tolerance of disturbances and fidelity to particular environments using native species conservatism scores. We found differences in all 3 metrics of plant responses that were specific to disturbance type and study area, and associations between disturbance and vegetation outcomes that were both desirable and undesirable. Generally, deer forage nutrition and invasive species biomass both increased in disturbed areas, whereas floristic quality increased with disturbance in 2 study areas but decreased in the third. Biomass of particular invasive species varied according to both study area and disturbance type. Lastly, we used decision analysis to illustrate trade-offs and overall support for different management actions while also accounting for underlying differences among study areas. For example, management actions with the greatest increase in mule deer forage nutrition tended to also increase invasive species biomass. Compared to other management actions, low-severity harvest scored highly in 2 study areas due to its association with increased forage nutrition while also limiting invasive species biomass and maintaining floristic quality. However, different weighting schemes according to different management priorities among management outcomes alter the relative ranking of actions. Our data show area-specific trade-offs in vegetation outcomes that result from disturbance management, which can be combined with decision analysis to help managers balance objectives and compare potential management actions.

To improve lifetime reproductive success, maternal ungulates should pursue behavioral strategies that reduce risk of offspring mortality. Predation is a leading cause of neonatal mortality; thus, maternal ungulates should select habitat that reduces risk of predation on vulnerable neonates. We examined selection of habitat across subpopulations of federally endangered Sierra Nevada bighorn sheep (Ovis canadensis sierrae) within Sierra Nevada, California, USA, 2008-2018. Our objectives were to understand how maternal Sierra Nevada bighorn minimize risk of predation through selection of habitat and to quantify current and future potential neonatal habitat across the Sierra Nevada. Mountain lions (Puma concolor) are the predominant predators of adult Sierra Nevada bighorn; thus, we hypothesized females would select habitat that minimized risk of predation from mountain lions. We developed a used-available resource selection function to quantify patterns of habitat selection and produced predictive maps identifying highly selected lambing habitat across the Sierra Nevada. Our top model demonstrated females selected habitat where the relative probability of encountering mountain lions was low and near escape terrain. Selection for vegetation type was dependent on risk of encounter with mountain lions; when risk of encounter was low, females selected shrub cover. This suggests that when risk of encountering mountain lions was low, females may have selected areas with dense shrubs to reduce risk of being detected by other predators such as coyotes (Canis latrans) and golden eagles (Aquila chrysaetos). As risk of encountering mountain lions increased, females avoided all vegetation types other than barren, suggesting that despite increased risk of encounter, females may have been able to mitigate overall mortality risk by selecting habitat where approaching mountain lions could be detected and avoided. Our predictive models and resulting maps demonstrate differences in prevalence and connectivity of highly selected lambing habitat across subpopulations, which may explain differences in lamb recruitment and adequacy of lambing habitat observed among subpopulations. Recolonization into historical ranges and increasing connectivity between Sierra Nevada bighorn subpopulations is an important conservation need for species recovery and long-term viability of fragmented subpopulations.

A clear connection between basic research and applied management is often missing or difficult to discern. We present a case study of integration of basic research with applied management for estimating abundance of gray wolves (Canis lupus) in Montana, USA. Estimating wolf abundance is a key component of wolf management but is costly and time intensive as wolf populations continue to grow. We developed a multimodel approach using an occupancy model, mechanistic territory model, and empirical group size model to improve abundance estimates while reducing monitoring effort. Whereas field-based wolf counts generally rely on costly, difficult-to-collect monitoring data, especially for larger areas or population sizes, our approach efficiently uses readily available wolf observation data and introduces models focused on biological mechanisms underlying territorial and social behavior. In a three-part process, the occupancy model first estimates the extent of wolf distribution in Montana, based on environmental covariates and wolf observations. The spatially explicit mechanistic territory model predicts territory sizes using simple behavioral rules and data on prey resources, terrain ruggedness, and human density. Together, these models predict the number of packs. An empirical pack size model based on 14 years of data demonstrates that pack sizes are positively related to local densities of packs, and negatively related to terrain ruggedness, local mortalities, and intensity of harvest management. Total abundance estimates for given areas are derived by combining estimated numbers of packs and pack sizes. We estimated the Montana wolf population to be smallest in the first year of our study, with 91 packs and 654 wolves in 2007, followed by a population peak in 2011 with 1252 wolves. The population declined ~6% thereafter, coincident with implementation of legal harvest in Montana. Recent numbers have largely stabilized at an average of 191 packs and 1141 wolves from 2016 to 2020. This new approach accounts for biologically based, spatially explicit predictions of behavior to provide more accurate estimates of carnivore abundance at finer spatial scales. By integrating basic and applied research, our approach can therefore better inform decision-making and meet management needs.

In the Northern Rockies of the United States, predators like wolves (Canis lupus) and mountain lions (Puma concolor) have been implicated in fluctuations or declines in populations of game species like elk (Cervus canadensis) and mule deer (Odocoileus hemionus). In particular, local distributions of these predators may affect ungulate behavior, use of space, and dynamics. Our goal was to develop generalizable predictions of habitat selection by wolves and mountain lions across western Montana. We hypothesized both predator species would select habitat that maximized their chances of encountering and killing ungulates and that minimized their chances of encountering humans. We assessed habitat selection by these predators during summer using within-home range (3rd order) resource selection functions (RSFs) in multiple study areas throughout western Montana, and tested how generalizable RSF predictions were by applying them to out-of-sample telemetry data from separate study areas. Selection for vegetation cover-types varied substantially among wolves in different study areas. Nonetheless, our predictions of 3rd order selection by wolves were highly generalizable across different study areas. Wolves consistently selected simple topography where ungulate prey may be more susceptible to their cursorial hunting mode. Topographic features may serve as better proxies of predation risk by wolves than vegetation cover-types. Predictions of mountain lion distribution were less generalizable. Use of rugged terrain by mountain lions varied across ecosystem-types, likely because mountain lions targeted the habitats of different prey species in each study area. Our findings suggest that features that facilitate the hunting mode of a predator (i.e. simple topography for cursorial predators and hiding cover for stalking predators) may be more generalizable predictors of their habitat selection than features associated with local prey densities.

As an outcome of natural selection, animals are probably adapted to select territories economically by maximizing benefits and minimizing costs of territory ownership. Theory and empirical precedent indicate that a primary benefit of many territories is exclusive access to food resources, and primary costs of defending and using space are associated with competition, travel and mortality risk. A recently developed mechanistic model for economical territory selection provided numerous empirically testable predictions. We tested these predictions using location data from grey wolves (Canis lupus) in Montana, USA. As predicted, territories were smaller in areas with greater densities of prey, competitors and low-use roads, and for groups of greater size. Territory size increased before decreasing curvilinearly with greater terrain ruggedness and harvest mortalities. Our study provides evidence for the economical selection of territories as a causal mechanism underlying ecological patterns observed in a cooperative carnivore. Results demonstrate how a wide range of environmental and social conditions will influence economical behaviour and resulting space use. We expect similar responses would be observed in numerous territorial species. A mechanistic approach enables understanding how and why animals select particular territories. This knowledge can be used to enhance conservation efforts and more successfully predict effects of conservation actions.

In cooperatively breeding species, large group size is often positively related to reproductive success and group persistence. We have a poor understanding, however, of how group sizes within a population affect reproduction particularly as density varies. We hypothesized that at low densities, wolves in both small and large groups would have similar reproductive rates. At high densities, however, wolves in small groups would have lower reproductive rates compared to those in large groups. Using empirical data from radio‐collared wolves in Idaho and Yellowstone National Park, WY, USA (1996–2012), we compared reproductive rates (i.e. proportion reproducing, litter size, pup survival) among small and large groups of wolves as density fluctuated within the populations. Reproductive rates were generally lower for individuals in small groups compared to those in large groups, particularly as density increased. Pup survival, however, was slightly higher for wolves in small groups compared to large groups except at very high densities. Polygamy increased with density regardless of group size, suggesting a polygamy threshold for wolves. Large group size resulted in less parturition failure, more breeding females per group, larger litter sizes, and ultimately more pups recruited per group. Large group size appears advantageous for several, but not all, aspects of reproduction particularly when population density is high.

Recently developed methods, including time-to-event and space-to-event models, estimate the abundance of unmarked populations from encounter rates with camera trap arrays, addressing a gap in noninvasive wildlife monitoring. However, estimating abundance from encounter rates relies on assumptions that can be difficult to meet in the field, including random movement, population closure, and an accurate estimate of movement speed. Understanding how these models respond to violation of these assumptions will assist in making them more applicable in real-world settings. We used simulated walk models to test the effects of violating the assumptions of the time-to-event model under four scenarios: (1) incorrectly estimating movement speed, (2) violating closure, (3) individuals moving within simplified territories (i.e., movement restricted to partially overlapping circles), (4) and individuals clustering in preferred habitat. The time-to-event model was robust to closure violations, territoriality, and clustering when cameras were placed randomly. However, the model failed to estimate abundance accurately when movement speed was incorrectly estimated or cameras were placed nonrandomly with respect to habitat. We show that the time-to-event model can provide unbiased estimates of abundance when some assumptions that are commonly violated in wildlife studies are not met. Having a robust method for estimating the abundance of unmarked populations with remote cameras will allow practitioners to monitor a more diverse array of populations noninvasively. With the time-to-event model, placing cameras randomly with respect to animal movement and accurately estimating movement speed allows unbiased estimation of abundance. The model is robust to violating the other assumptions we tested.

Montana Fish, Wildlife and Parks (MFWP) is concerned that current management practices are responding inadequately to the increasing frequency and consequences of pneumonia die--offs (epizootics) in wild herds of bighorn sheep. Pneumonia epizootics threaten the per-sis-tence of herds and recovery of the species, causing rapid population declines and prolonged periods of poor recruitment. Generally, wildlife man-ag-ers are poorly prepared to prevent disease outbreaks, relying instead on reactive "crisis management." MFWP sought to develop a more proactive, overarching strategy to assist herd man-ag-ers in determining when it would be beneficial and cost--effective to conduct management to reduce the risk of an epizootic, before it takes place. The system needed to be broadly consistent across the state yet support flexibility for local decisions based on site--specific knowledge about conditions and risks. This case study illustrates how we developed a general framework that integrates a decision tree predicting site--specific probability of a pneumonia epizootic into a multi-criteria decision analy-sis to help guide local decisions about proactive disease management. We formalized this decisionmaking process in a spreadsheet model that has been made available to ma-agers of bighorn herds in Montana. To illustrate an application of this decision tool, we also present the results of an analysis conducted by the man-ag-er of a herd at high risk of an epizootic.

Territorial behavior is a fundamental and conspicuous behavior within numerous species, but the mechanisms driving territory selection remain uncertain. Theory and empirical precedent indicate that many animals select territories economically to satisfy resource requirements for survival and reproduction, based on benefits of food resources and costs of competition and travel. Costs of competition may vary by competitive ability, and costs of predation risk may also drive territory selection. Habitat structure, resource requirements, conspecific density, and predator distribution and abundance are likely to further influence territorial behavior. We developed a mechanistic, spatially-explicit, individual-based model to better understand how animals select particular territories. The model was based on optimal selection of individual patches for inclusion in a territory according to their net value, i.e., benefits (food resources) minus costs (travel, competition, predation risk). Simulations produced predictions for what may be observed empirically if such optimization drives placement and characteristics of territories. Simulations consisted of sequential, iterative selection of territories by simulated animals that interacted to defend and maintain territories. Results explain why certain patterns in space use are commonly observed, and when and why these patterns may differ from the norm. For example, more clumped or abundant food resources are predicted to result, on average, in smaller territories with more overlap. Strongly different resource requirements for individuals or groups in a population will directly affect space use and are predicted to cause different responses under identical conditions. Territories are predicted to decrease in size with increasing population density, which can enable a population's density of territories to change at faster rates than their spatial distribution. Due to competition, less competitive territory-holders are generally predicted to have larger territories in order to accumulate sufficient resources, which could produce an ideal despotic distribution of territories. Interestingly, territory size is predicted to often show a curvilinear response to increases in predator densities, and territories are predicted to be larger where predators are more clumped in distribution. Predictions consistent with empirical observations provide support for optimal patch selection as a mechanism for the economical territories of animals commonly observed in nature.

Animals communicate in a variety of ways and calls are used for a number of important behaviors. Temperature, wind, time of day, and human activities can affect animals' use of calls, particularly over long distances. Effects of group size on the use of calls can be particularly influential in territorial social carnivores. Where gray wolves (Canis lupus) are hunted by humans, for example, howling may make it easier for hunters to locate individuals and ultimately increase mortality. We hypothesized that a suite of factors would affect wolves' responses to simulated howling. Specifically, we predicted that howling behavior would increase with (a) group size, (b) pup age, and (c) during crepuscular time periods and howling behavior would decrease (a) where wolves were harvested and (b) when it was hot or windy. Contrary to our prediction, larger groups did not respond as quickly to simulated wolf howls as smaller groups did and minimum and maximum daily temperatures were not good predictors of wolf howling response rates. Individuals in small litters of pups may have responded more quickly to howls than those in large litters because they are eager to seek safety from and have socialization with adults returning from foraging bouts. Although harvest did not appear to affect vocal communication by wolves, group size, pup age, time of day, wind, and number of howls emitted greatly affected wolves' behavior and responses during howling surveys. Howling responses did not change because of harvest; response rates from wolves were nearly identical with (2.2%) and without (2.3%) harvest. The year-round benefits of long-distance vocal communication may outweigh the costs of increased mortality arising from howling during harvest season.

ABSTRACTEstimating abundance of wildlife populations can be challenging and costly, especially for species that are difficult to detect and that live at low densities, such as cougars (Puma concolor). Remote, motion‐sensitive cameras are a relatively efficient monitoring tool, but most abundance estimation techniques using remote cameras rely on some or all of the population being uniquely identifiable. Recently developed methods estimate abundance from encounter rates with remote cameras and do not require identifiable individuals. We used 2 methods, the time‐to‐event and space‐to‐event models, to estimate the density of 2 cougar populations in Idaho, USA, over 3 winters from 2016–2019. We concurrently estimated cougar density using the random encounter model (REM), an existing camera‐based method for unmarked populations, and genetic spatial capture recapture (SCR), an established method for monitoring cougar populations. In surveys for which we successfully estimated density using the SCR model, the time‐to‐event estimates were more precise and showed comparable variation between survey years. The space‐to‐event estimates were less precise than the SCR estimates and were more variable between survey years. Compared to REM, time‐to‐event was more precise and consistent, and space‐to‐event was less precise and consistent. Low sample sizes made the space‐to‐event and SCR models inconsistent from survey to survey, and non‐random camera placement may have biased both of the camera‐based estimators high. We show that camera‐based estimators can perform comparably to existing methods for estimating abundance in unmarked species that live at low densities. With the time‐ and space‐to‐event models, managers could use remote cameras to monitor populations of multiple species at broader spatial and temporal scales than existing methods allow. © 2020 The Wildlife Society.

The effects of harvest on cooperatively breeding species are often more complex than simply subtracting the number of animals that died from the group count. Changes in demographic rates, particularly dispersal, could offset some effects of harvest mortality in groups but this is rarely explored with cooperative breeders. We asked whether a cooperatively breeding species known for long‐distance dispersal could compensate for the effect of harvest mortality on density by adopting immigrants into the group. We used genetic samples to estimate the minimum density of gray wolves ( Canis lupus ) and proportion of immigrants in groups in the northern US Rocky Mountains after an annual harvest regime was initiated and in the Canadian Rocky Mountains where wolves were managed consistently under an annual harvest regime. We tested whether immigration (1) compensated, (2) partially compensated or (3) did not compensate numerically for harvest mortality in groups and hypothesized immigration would increase with increasing harvest intensity. Density of wolves in groups declined after harvest was initiated whereas immigration into groups was consistently low and did not change with harvest in the US study area. Immigration into groups was similarly low and density even lower in the Canadian study area compared to the US study area. Our results indicate immigration did not compensate for harvest mortality in groups in two separate populations of a cooperatively breeding carnivore. We hypothesize the social structure of wolf groups may limit the potentially compensatory response of immigration in some populations.

Understanding how environmental factors interact to determine the abundance and distribution of animals is a primary goal of ecology, and fundamental to the conservation of wildlife populations. Studies of these relationships, however, often assume static environmental conditions, and rarely consider effects of competition with ecologically similar species. In many parts of their shared ranges, grizzly bears Ursus arctos and American black bears U. americanus have nearly complete dietary overlap and share similar life history traits. We therefore tested the hypothesis that density patterns of both bear species would reflect seasonal variation in available resources, with areas of higher primary productivity supporting higher densities of both species. We also hypothesized that interspecific competition would influence seasonal density patterns. Specifically, we predicted that grizzly bear density would be locally reduced due to the ability of black bears to more efficiently exploit patchy food resources such as seasonally abundant fruits. To test our hypotheses, we used detections of 309 grizzly and 597 black bears from two independent genetic sampling methods in spatially‐explicit capture–recapture (SECR) models. Our results suggest grizzly bear density was lower in areas of high black bear density during spring and summer, although intraspecific densities were also important, particularly during the breeding season. Black bears had lower densities in areas of high grizzly bear density in spring; however, density of black bears in early and late summer was best explained by primary productivity. Our results are consistent with the hypothesis that smaller‐bodied, more abundant black bears may influence the density patterns of behaviorally‐dominant grizzly bears through exploitative competition. We also suggest that seasonal variation in resource availability be considered in efforts to relate environmental conditions to animal density.

ABSTRACTDuring late summer and fall, elk (Cervus canadensis) need access to adequate nutrition to support physiological requirements for reproduction and overwinter survival. The archery hunting season often occurs during this period and can affect distributions of elk as they seek areas that minimize perceived harvest risk. Areas that confer lower harvest risk may provide relatively low‐value nutrition, resulting in a potential tradeoff between minimizing risk and accessing adequate forage. We used radio‐collar data collected from female elk during late summer and fall (Aug–Oct) and estimated resource selection models to evaluate the extent of this potential risk‐nutrition tradeoff. To evaluate if elk exposed to a greater hunting risk altered selection for forage resources, we assessed the relationship between individuals’ selection coefficients for forage and the proportion of their late‐summer‐fall home range accessible to hunters (our metric of hunting risk). Our results indicate that during the archery season, elk with higher‐risk home ranges selected more strongly for areas farther from motorized routes than elk with lower‐risk home ranges. Regardless of the level of risk, however, elk maintained or increased selection for areas with higher forage quality, suggesting that elk did not compromise access to nutritional resources during the archery season. Elk with higher‐risk home ranges were also exposed to the poorest nutrition and increased their selection for areas with higher forage quality more strongly than elk with lower‐risk home ranges during the hunting season. Elk with lower‐risk home ranges had access to the highest nutrition, which may be due to the availability of concentrated sources of high‐quality forage from irrigated agricultural areas on private lands that restricted hunter access. Resource agencies interested in encouraging elk to remain on public lands during the hunting seasons might consider closing motorized travel during the archery season to increase security on public lands, limiting hunter pressure on public lands, improving forage quality on public lands, and working with private land owners to enhance hunter accessibility and restrict elk access to high‐quality forage resources. © 2019 The Wildlife Society.

Ungulates migrate to maximize nutritional intake when forage varies seasonally. Populations of ungulates often include both migratory and non-migratory individuals, but the mechanisms driving individual differences in migratory behaviour are not well-understood. We quantified associations between hypothesized drivers of partial migration and the likelihood of migration for individual ungulates that experienced a range of environmental conditions and anthropogenic influences. We evaluated the effects of forage variation, conspecific density, and human land uses on migratory behaviour of 308 adult female elk in 16 herds across western Montana. We found irrigated agriculture on an individual's winter range reduced migratory behaviour, but individuals were more likely to migrate away from irrigated agricultural areas if better forage was available elsewhere or if they experienced high conspecific density on their winter range. When the forage available during the summer growing season varied predictably between years, elk were more likely to migrate regardless of whether they had access to irrigated agriculture. Our study shows that predictable availability of beneficial native forage can encourage migration even for ungulates with irrigated agriculture on their winter range. Perturbations that can affect the forage available to ungulates include wildfires, timber harvest, livestock grazing and changing weather patterns. If these or other disturbances negatively affect forage on summer ranges of migrants, or if they cause forage to vary unpredictably across space and time, our results suggest migratory behaviour may decline as a result.