Behavior, nutrition, and environment drive survival of a large herbivore in the face of extreme winter conditions

Abstract

For many species, behavioral modification is an effective strategy to mitigate negative effects of harsh and unpredictable environmental conditions. When behavioral modifications are not sufficient to mitigate extreme environmental conditions, intrinsic factors may be the primary determinant of survival. We investigated how movement behavior, and internal (i.e., nutrition and age) and external (i.e., food availability and snow depth) states affect survival over winter of a long-lived and highly faithful species (mule deer; Odocoileus hemionus). We first tested whether animals changed their behavior during winter based on internal and external states; we subsequently investigated how behavior and state interacted to influence survival in the face of extraordinary winter conditions. Movement behavior changed minimally as a function of age and nutrition; yet, movement behavior affected survival—animals that exhibited more restricted movements were more likely to succumb to mortality overwinter than animals with less restricted movements. Additionally, nutrition and cumulative snow depth had a strong effect on survival: animals that were exposed to deep snow and began winter with low fat were much less likely to survive. Behavior was an effective tool in securing survival during mild or moderate winters, but nutrition ultimately underpinned survival during harsh winters.

INTRODUCTION

Globally, climatic conditions have become increasingly stochastic and may affect species distribution, population dynamics, and ecological processes (Convey & Smith, 2006; Grosbois et al., 2008). Changes in intensity, frequency, duration, and severity of environmental conditions that result from climate change can affect fitness, and thus persistence, of wild animals (Staudinger et al., 2013). For some animals, often the first line of defense to environmental change is altered behaviors, which generally comes at low cost and can be very effective (Muñoz et al., 2015; Verzuh et al., 2021). Yet, to mitigate negative environmental effects through altered behavior, animals must display behavioral flexibility. The degree animals can alter their behavior is contingent on a suite of different circumstances, ranging from reproductive state and environmental conditions to the evolutionary history of an animal (Wolf et al., 2007). Species that exhibit high fidelity may be unable to alter behaviors adequately and require additional strategies to cope with environmental change. The degree to which the evolutionary, life, and natural history of an animal influences adaptive behavioral responses to extreme events, however, remains elusive (Alto et al., 2013; Newman, 1992).

In seasonal environments, animals have adopted a wide breadth of behavioral strategies to allow them to persist during times of resource limitation, from reductions in metabolic processes and body temperatures (i.e., hibernation or torpor) to use of stored energy (e.g., food caches; Smith & Reichman, 1984; somatic reserves: Hjeljord & Histøl, 1999; Parker et al., 2009). Energetic capital (e.g., somatic reserves) is accumulated during periods of resource abundance (e.g., spring or summer, wet season) and then used during periods of resource scarcity (e.g., winter, dry season) to ensure survival and facilitate reproductive effort (Jönsson, 1997). Yet, regardless of the amount of stored energy that an animal enters winter with, energetic capital is not always enough; animals may need to mitigate energy loss over winter by seeking out forage throughout winter to survive.

In temperate environments, winter often represents a nutritional and survival bottleneck for animals (Rogers & Smith, 1993; Unsworth et al., 1999). Stored energy, life history, environmental conditions, and ability to acquire forage all affect survival (Clark & Dukas, 2000). Survival of adult, large herbivores overwinter typically is stable (Unsworth et al., 1999) but can drop substantially when animals are exposed to unpredictable and harsh conditions (e.g., survival has dropped as low as 22% when faced with severe winter conditions; Kautz et al., 2020). Forage quality and quantity are relatively low in winter, and coupled with deep snow, foraging comes at an increased energetic cost (Parker et al., 1984). The balance between energy acquisition and expenditure can create trade-offs for individuals whereby the costs of seeking out forage (e.g., energy expenditure, increased predation risk) must be weighed against the benefits gained from accessing those resources relative to their current energetic state (e.g., state-dependent foraging: Caraco, 1981; risk sensitive-foraging: Blecha et al., 2018). If energy expenditure in seeking out resources is greater than the energy gained by acquiring them, the risk of starvation becomes imminent. Yet, if energetic reserves are inadequate to fulfill daily metabolic needs, not seeking out new resources also can result in starvation. To survive winter, animals must successfully balance the costs and benefits of foraging, which can be difficult, especially since the costs of foraging change under differing environmental and internal conditions—an increasingly tenuous process during severe, stochastic winters.

Shifts in the frequency and severity of winter conditions in temperate environments resulting from climate change can have important implications for population dynamics of large mammals and may pose a broader threat to biodiversity (Saltz et al., 2006). For some species, behavioral plasticity allows animals to alter their exposure to unfavorable conditions and cope with changing environments (Verzuh et al., 2021). It remains unclear, however, how species with extreme behavioral fidelity might shift their decision-making when faced with increasingly stochastic environments. We investigated the role of movement behavior, nutrition, and environment on overwinter survival of a long-lived and highly faithful animal using a long-term, individual-based study of adult, female mule deer (Odocoileus hemionus) in western Wyoming, USA from autumn 2015 to spring 2021. Deer in the Wyoming Range experienced two particularly harsh winters (2016–2017 and 2018–2019) with extreme snowpack. During these winters, survival (~65% and 70%) fell far below what is typical for this population (~90%; LaSharr et al., 2023). These severe winters were extraordinary; animals in this population had never experienced winter conditions that harsh during their lifetime; thus, it presented a novel opportunity to evaluate whether animals can mitigate the effects of harsh environmental conditions by shifting behaviors and how dependent survival is on internal state. We evaluated how movement behavior, internal state (i.e., nutritional condition and age), winter conditions (i.e., snow depth), and forage availability interacted to influence survival for a highly faithful and long-lived species. Animals cannot change their age or the condition in which they entered a season, but shifts in behavior are possible throughout winter; thus, behavior represents one of the few axes an animal can adjust in response to harsh conditions. Given the potential for animals to mitigate negative environmental effects through modification of movement behavior, we hypothesized that movement behavior in winter would change under different internal (i.e., nutritional condition and age) and external (i.e., snow depth and forage availability) states. We expected severity in winter conditions to affect survival negatively, whereas survival should be enhanced for animals in good nutritional condition at the beginning of winter and of prime age. Moreover, we expected state-dependent movement behaviors to further buffer survival—animals that shifted their movement based on their internal and external state would increase their probability of surviving.

METHODS Study area

We studied a population of mule deer that winters on two distinct winter ranges in the Wyoming and Salt River mountain ranges in western Wyoming, USA (42°25° N, 110°42° W) from December 2015 to April 2021. The primary vegetation types in the study area are sagebrush species (Artemisia spp.) with additional mountain-shrub communities (Juniperus spp., Cercocarpus spp., Amelanchier spp., and Symphoricarpos spp.). Primary predators on winter ranges for the population included mountain lions (Puma concolor) and coyotes (Canis latrans). Elevations range from 2000 to 2300 m and mean, annual precipitation (30-year average; Applied Climate Information System [ACIS], National Oceanic and Atmospheric Administration Regional Climate Centers) was 26.3 cm in South Wyoming Range (Kemmerer, WY; ACIS Station 5105; elevation 2112 m) and 16.5 cm in North Wyoming Range (Big Piney, WY; ACIS Station 0695; elevation 2079 m; data available online). Mule deer in the Wyoming Range experienced two harsh winters during our study period. From 1991 to 2021, maximum snow depth was highest during the winter of 2016–2017 at both the Fossil Butte, WY and Woodruff, UT weather stations; during the winter of 2018–2019, maximum snow depth was the fourth highest in Woodruff, UT and third highest in Evanston, WY during that time (National Oceanic and Atmospheric Administration Regional Climate Centers). These winters were considered some of the harshest conditions for large herbivores seen in nearly three decades by local biologists and wildlife managers (G. Fralick and N. Hymas, personal communication). Adult, female mule deer were fitted with GPS collars that recorded fixes every 1–2 h. We recaptured all animals each autumn and spring and assessed nutritional condition (i.e., body fat). See Appendix S1 for a detailed methodology on animal capture and handling.

Spatial data

GPS collars recorded locations every 1–2 h throughout winter. First, we subset all collars that had hourly relocation data to 2 h for consistency. Next, we subset movement data for each winter for each individual from 1 December to 30 April. Because we were interested in movement on winter range, and not during migration, if animals were still migrating into the beginning of December, we began the subset on the day following their last migration date in autumn. If animals began migrating in April, we ended the subset the day before their first migration date in spring. If animals spent more than 50% of December or April migrating, we removed those animal-years from the analyses.

To assess environmental factors, we used remotely sensed data of food availability and snow depth. Sagebrush was the primary forage species for mule deer in this population during winter, and makes up >80% of their diet (Dwinnell et al., 2019). We developed a metric of food availability using a remotely sensed layer of shrub height (in meters; Homer et al., 2015) and cumulative summer precipitation (in millimeters; 1 May–30 August) on winter range (Thornton et al., 1997; DAYMET), which directly influences the amount of leader growth available to animals in winter. We used remotely sensed data of snow depth available through Snow Data Assimilation System (SNODAS). At each GPS location, we extracted values of snow depth and forage height with a 50-m buffer using the “terra package” in R (Hijams et al., 2022).

We quantified movement behavior on winter range using first passage time (FPT; Byrne et al., 2014; Fauchald & Tverra, 2003). The size of the circle and tortuosity and speed of the animal all influences FPT; animals that have very restricted movements (e.g., slow and highly tortuous) will have much higher FPT measures than animals with fast and directed movements. At each GPS location, we calculated FPT with a radius of 50 m using the “adehabitatLT” package in R version 4.1.3 (Calenge, 2006; R Core Team, 2021).

Movement may become restricted when animals succumb to malnutrition—to prevent bias in movement that may result from physiological limitations that accompany starvation, we censored movement data 14 days preceding death (Alston et al., 2020) and evaluated differences in movement of animals that lived and died in early winter, before mortality occured (Appendix S1).

Movement analyses

To identify whether movement behavior was dependent on an animal's state, we first evaluated whether there was a relationship between movement behavior and autumn body fat, age, snow depth, and food availability. We used a linear mixed-effects model with FPT at each GPS location as the response variable and autumn body fat, age, snow depth, and food availability as the predictor variables. We included a random intercept of unique animal identifier to account for repeated measures in the model. Analyses were performed using the “lme4” package in R version 4.1.3 (Bates et al., 2015; R Core Team, 2021).

Survival analyses

We used Cox proportional hazards models to evaluate survival of mule from December 2013 to April 2021 using a 14-day encounter history; each winter included 10 encounter periods. We only used animals that had died of malnutrition (Appendix S1).

Overwinter survival of large mammals is generally influenced by the cumulative effects of environment, state, and behavior, not conditions at a single point in time. To account for lag effects that may occur from previous experiences, we calculated a 30-day moving average of food availability and FPT for each encounter period (i.e., for each encounter period, we calculated the average FPT and food availability for the 30 days preceding). Animals that are approaching mortality through malnutrition may reduce movement in the final days of their lives; using a 30-day moving average reduced potential bias in movement that may have resulted from an animal's physiological state of dying and not from their behavioral decisions. Additionally, exposure to snow throughout winter is very energetically expensive. We estimated the cumulative exposure to snow depth across winter. We first calculated the average snow depth for encounter window, and for each encounter window, we summed all previous snow depth values up to the current window.

We used a combination of hypothesis testing and model selection to assess which variables had a significant effect on survival overwinter and test our hypothesis that movement behavior and state would interact to influence survival of mule deer. We determined significance at α ≤ 0.10 to reduce the probability of a type II error. We included autumn body fat, age, cumulative exposure to snow depth, and a 30-day moving average of food availability and FPT in all models. To test our hypothesis, we included interactions between FPT and all other variables and cumulative exposure to snow depth and all other variables to identify whether state-dependent behaviors and changes in snow depth influenced survival. If an interaction in a model was not significant (α > 0.10), we removed the model from the model selection process. We included a random intercept of unique animal identifier in all models. We performed model selection using Akaike information criterion (AIC; Burnham & Anderson, 2004) to identify the top model. All analyses were performed using the “coxme”  package in R version 4.1.3 (R Core Team, 2021; Therneau, 2012).

RESULTS

We evaluated how movement behavior, food availability, snow depth, and animal state influenced overwinter survival of adult mule deer between December 2015 and April 2021 using 152 adult, female mule deer with 329 unique animal-years. Median cumulative snow depth animals experienced on their winter ranges was 0.78 m, but was as high as 7.2 m in the 2016–2017 winter and 4.5 m during the 2018–2019 winter.

Movement analyses

Increases in snow depth, autumn body fat, and age all influenced movement of mule deer (Appendix S1: Table S2), but the effect size for body fat and age were minimal and likely not biologically meaningful. Increasing age slightly restricted movement; there was a predicted increase in FPT of 3.18 min for every one-year increase in age. Animals in better nutritional condition were slightly less restricted in their movements, with an increase of 1 percentage point (ppt) of body fat in autumn prompting a predicted decrease in FPT by 3.24 min. Increasing snow depth resulted in more restricted movement of deer; there was a predicted increase in FPT of 40.7 min with every 10 cm increase in snow depth.

Survival analyses

Our top model from the Cox proportional hazards analysis included autumn body fat, age, FPT at 50 m, snow depth, food availability, and an interaction between age and snow depth; there were no other models within 2 AIC. Probability of mortality increased with FPT, age, and cumulative snow depth, and decreased with increasing fat (Appendix S1: Table S3). Autumn body fat had a strong effect on mortality (β = −0.311); expected hazard decreased 26.7% with every 1 ppt increase in autumn fat. Cumulative snow depth also had a large effect on probability of mortality (β = 1.32), and the expected hazard was 3.7× higher with every meter increase in cumulative snow depth. Age positively influenced probability of mortality (β = 0.44), and there was a negative interaction between age and snow depth (β = −0.122); age had an important effect on survival when animals were exposed to relatively low snow depths, but at high exposure to snow depth survival decreased for old and young animals comparably.

Movement behavior and fat both influenced overwinter survival of mule deer, but the degree in which each could improve survival was dependent on other variables, including winter conditions. During a relatively average winter (i.e., cumulative snow ≤1 m), an average deer (i.e., 7 years old, 10% body fat) could improve survival substantially through shifts in movement behavior alone; survival would increase from 0.9% to 71.5% when movement became much less restricted (a shift from an FPT of 20 h to 30 min). When winters were incredibly harsh, however, fat played an important additive role—when exposure to cumulative snow was high (i.e., >6 m), an average deer would have a 19.5% probability of surviving a harsh winter, even when moving a lot (i.e., FPT of 30 min). Yet, if that animal had begun that harsh winter with 20% fat, their probability of survival would have increased to 84.3%.

DISCUSSION

Climate change has altered the environments and resources of wild animals around the world, and increasingly stochastic conditions are projected into the next century (Brown & Mote, 2009; Cleland et al., 2007). Behavioral alteration can be an effective strategy to buffer against life-threatening conditions (e.g., severe winters, intense droughts; Beever et al., 2017; Verzuh et al., 2021)—conditions that may become increasingly common with global climate change. Yet, shifts in behavior can only do so much to buffer against undesirable environmental effects, and for species that do not exhibit behavioral plasticity, behavioral choices may not be enough to secure survival—the internal state of an animal (e.g., fat and age) often play important and additive roles. For a highly faithful, long-lived mammal that experienced harsh winter conditions that typically only occur once every 40–50 years, movement behavior, age, nutritional condition, and the environment all influenced overwinter survival. Contrary to our hypothesis, mule deer were unable to mediate survival through state-dependent shifts in behavior. Regardless of the fat, age, food, or snow depth an animal was exposed to, more restricted movements (i.e., higher FPT) resulted in a lower probability of surviving winter. When winter conditions are benign or moderate, increased movement may be an effective tool in buffering against overwinter mortality for animals that entered winter with low capital reserves. As winter severity worsens, behavior may not matter if animals do not have sufficient stores of fat—regardless of movement strategy, age and nutrition had important and overpowering effects on survival of deer, particularly when winter conditions were harsh.

In seasonal environments with extended periods of resource limitation, many animals rely on both income (e.g., forage) and capital (e.g., stored fat) resources to meet the minimum requirements for survival and maintenance (Parker et al., 2009). Indeed, entering winter with sufficient body fat and access to forage throughout winter strongly improved the probability of surviving a winter (Appendix S1: Table S3). For large herbivores, the condition an animal begins winter in, and accordingly, the capital they have to expend as winter progresses, should be dependent on the environment an animal inhabits (Bårdsen et al., 2014; Monteith et al., 2013). Consequently, if animals that typically do not experience severe winters are exposed suddenly to unprecedented conditions, they may risk mortality if the stored fat they began winter with was insufficient to buffer themselves against the energetic costs of a harsh winter. Winter can be energetically expensive for large herbivores to navigate, and employing movement strategies that minimize energetic expenditure and maximize energetic gain may be crucial to survival, especially when animals are exposed to harsh environmental conditions (Parker et al., 1984, 1996). Contrary to evidence in other species of large herbivores where foraging and resting strategies are dependent on both nutritional state and forage availability (Kohli et al., 2014), age, nutrition, or forage did not have a meaningful effect on movement behavior of mule deer during winter (Appendix S1: Table S2). Nevertheless, movement during winter was crucial to survival—regardless of environment, nutrition, or age, animals that exhibited more restricted movement, and thus lost presumed foraging opportunities, had a higher probability of mortality compared with animals that did not stay in the same place for extended periods.

During winter, snow can be an impressive obstacle for large animals to navigate. Snowpack can increase the energetic costs of moving by up to eight times (Hudson & Haigh, 2002; Parker et al., 1984) and affects the establishment of seasonal ranges and migratory routes for large herbivores (Monteith et al., 2018). Mule deer in the Wyoming Range experienced two particularly harsh winters between 2015 and 2021, with snow depths reaching record highs and adult survival dipping far below what is typical for large herbivores. Snow depth was the biggest driver of movement behavior; as snow depth increased, movement and presumed foraging opportunities became more restricted (Appendix S1: Table S2). Snow depth also had a strong influence on survival over winter; mule deer faced an increasing probability of mortality with increasing exposure to snow depth (Figure 1; Appendix S1: Table S3). Moreover, though younger animals were more likely to survive when exposed to low snow depths, high snow depths overwhelmed the age-dependent relationship by reducing survival of young and old animals similarly. When snow accumulations reached >3.5 m, an average animal (i.e., food availability, fat, and movement all held at the median) faced a high probability of mortality (>70%), regardless of its age (Figure 1). For wild animals, behavioral plasticity and changes in decision-making based on individual state may be critical to survival in rapidly changing landscapes (Verzuh et al., 2021). For animals that exhibit extreme behavioral fidelity, shifts in behavior may be less likely to occur (Jakopak et al., 2022) or simply insufficient to cope with shifts in resources and environmental change, and thus animals may be maladapted to changing environments.

FIGURE 1. Probability of survival as a function of autumn fat, first passage time (FPT), age, the interaction between age and cumulative snow depth, and food availability for adult female mule deer in the Wyoming Range, WY, USA, from December 2015 to April 2021. Predictions represent changes across the range of one covariate with all other covariates held at the median (autumn fat = 10.6%, age = 7.5, FPT = 3.7 h, cumulative snow depth = 0.78 m, food = 23.46 precipitation [in millimeters] × shrub height [in meters]). Predictions for cumulative exposure to snow depth represent probability of survival for animals in the 10th (young) and 90th (old) percentile for age.

Mule deer exhibit high fidelity across many different behavioral contexts (Aikens et al., 2021; Merkle et al., 2019; Sawyer et al., 2019) and exhibited minimal changes in behavior based on nutritional condition, food availability, or age during winter (Appendix S1: Table S2). In predictable landscapes, fidelity can provide benefits including reduced predation risk, reduced energetic costs of movement, and increased fitness (Morrison et al., 2021). Yet, animals that exhibit extremely faithful behaviors in stochastic and unpredictable environments may be unable to keep pace with changing landscapes, which can have serious consequences for fitness (Eggeman et al., 2016; Sawyer et al., 2019). During benign or average winters, shifts in movement behavior were enough to substantially increase survival probability (e.g., shifting FPT from 10 h to 30 min increased probability of survival from <20% to >70%). During particularly harsh winters, movement behavior did little to buffer survival if the condition animals entered the season in was subpar. Energetic capital was a much more powerful buffer against mortality when conditions were harsh—animals could not overcome harsh winter conditions through movement alone and even animals that moved a lot (e.g., FPT of 30 min) likely would succumb to mortality (survival <20%) if they entered a bad winter with insufficient fat reserves. Further, while shifts in movement behavior were important for survival, the relationship between survival and movement was not influenced by the internal or external state of an animal—regardless of their fat, age, or winter conditions, animals who moved more were more likely to survive. For highly faithful species, identifying and implementing management strategies that can help buffer animals against harsh environmental conditions may be crucial to population and species persistence, especially when animals cannot or do not exhibit state-dependent movement strategies.

In temperate environments, the conditions of one season can have important and lasting effects on animals in the following season. Body fat accrued over summer had a substantial effect on overwinter survival of mule deer—with all other covariates held at the median, an animal with 2% body fat in autumn had a 7.18% probability of surviving the winter, compared with the 95.4% probability of survival for an animal starting winter with 20% body fat (Figure 1; Appendix S1: Table S3). Habitat and population characteristics on summer range, including population density and forage quality and quantity, can directly affect the nutritional condition of animals as they enter winter and determine the proximity of a population to nutritional carrying capacity (i.e., the number of animals a range can support; Monteith et al., 2014). Additionally, access to food during winter increased the probability of survival (Appendix S1: Table S3). Nutrition, whether stored (body fat) or consumed (foraged), had an overwhelming influence on survival in the face of change. Implementing management strategies that maintain populations below carrying capacity and prioritize restoration and maintenance of forage production on seasonal ranges—particularly in tall forb communities where historic overgrazing has reduced species diversity and productivity on summer ranges of ungulate species—may be important tools in buffering animals against high mortality during severe winters.

Survival in the face of life-threatening conditions was driven by movement behavior, environmental conditions (i.e., snow depth), and internal state (i.e., nutritional condition and age) for a highly faithful, long-lived animal. Moreover, shifts in movement behavior did not increase the probability of surviving winter under differing environmental conditions and internal state. Behavioral modification may be enough to buffer against negative effects on survival resulting from unfavorable environmental conditions (Beever et al., 2017; Verzuh et al., 2021), but sometimes shifts in behavior are insufficient to overcome harsh conditions. For highly faithful animals or when environmental conditions become too severe, the degree of behavioral plasticity may be inadequate to mitigate unfavorable effects of environmental change (Jakopak et al., 2022), and the external and internal states of an animal may have additive and powerful effects on survival. When animals are exposed to deep snow and harsh winter conditions, the marginal gains in foraging may become increasingly tenuous and fail to mitigate loss of body condition. Thus, condition of animals at the onset of winter may be critical to their survival, especially as populations around the world face increasingly stochastic and unpredictable conditions (Vázquez et al., 2017).

AUTHOR CONTRIBUTIONS

Conceptualization: Tayler N. LaSharr and Kevin L. Monteith. Data collection and curation: All coauthors. Formal analyses: Tayler N. LaSharr. Writing—original draft preparation: Tayler N. LaSharr. Visualization: Tayler N. LaSharr. Review and editing: All coauthors. Primary funding acquisition: Kevin L. Monteith, Gary Fralick, Tayler N. LaSharr, and Mark Thonhoff.

ACKNOWLEDGMENTS

We thank H. N. Abernathy, L. E. Hall, and R. A. Smiley for their helpful suggestions, comments, and feedback on this manuscript. The Wyoming Range mule deer study was supported by Wyoming Game and Fish Department, Wyoming Game and Fish Commission, Bureau of Land Management, Muley Fanatic Foundation (including Southwest, Kemmerer, Upper Green, and Blue Ridge Chapters), Boone and Crockett Club, Wyoming Wildlife and Natural Resources Trust, Knobloch Family Foundation, Wyoming Animal Damage Management Board, Wyoming Governor's Big Game License Coalition, Bowhunters of Wyoming, Wyoming Outfitters and Guides Association, Pope and Young Club, U.S. Forest Service, and U.S. Fish and Wildlife Service.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT

Data (LaSharr, 2023) are available in Dryad at https://doi.org/10.5061/dryad.v41ns1s28.