The hierarchy of virtue: mutualism, altruism and signaling in Martu women's cooperative hunting

Cooperative hunting is often assumed to be mutualistic, maintained through returns to scale, where, by working together, foragers can gain higher per capita return rates or harvest sizes than they can by hunting alone. We test this hypothesis among Martu hunters and find that cooperation only provides increased returns to poorer hunters while disadvantaging better hunters. Even so, better hunters still cooperate as frequently as poorer hunters. We ask whether better hunters are advantaged in secondary sharing distributions or whether they bias their partner choice to kin or household members. We find that better hunters are not more likely to pair up with kin and they do not gain consumption benefits from acquiring more. They share a greater proportion of their harvest than poorer hunters: no matter how much one produces — better hunter, worse hunter, cooperator, solitary hunter — all eat the same amount in the end. Such a result suggests the hypothesis that cooperation might be a costly signal of commitment to the public interest on the part of better hunters, which generates trust among camp members and facilitates strong social networks, particularly among women, who cooperate more than men. While some foragers may benefit through cooperation from returns to scale or risk reduction, others may benefit more through signaling commitment and generating trust.     

Hunting income patterns among the Hadza: big game, common goods, foraging goals and the evolution of the human diet.

The assumption that large mammal hunting and scavenging are economically advantageous to hominid foragers is examined in the light of data collected among the Hadza of northern Tanzania. Hadza hunters disregard small prey in favour of larger forms (mean adult mass greater than or equal to 40 kg). Here we report experimental data showing that hunters would reduce their mean rates if they included small animals in the array they target. Still, daily variance in large animal hunting returns is high, and the risk of failure correspondingly great, significantly greater than that associated with small game hunting and trapping. Sharing large kills reduces the risk of meatless days for big game hunters, and obviates the problem of storing large amounts of meat. It may be unavoidable if large carcasses cannot be defended economically against the demands of other consumers. If so, then large prey are common goods. A hunter may gain no consumption advantage from his own big game acquisition efforts. We use Hadza data to model this 'collective action' problem, and find that an exclusive focus on large game with extensive sharing is not the optimal strategy for hunters concerned with maximizing their own chances of eating meat. Other explanations for the emergence and persistence of this practice must be considered.     

Hunting, healing, and hxaro exchange: A long-term perspective on !Kung (Ju/''hoansi) large-game hunting

The question of why hunters target large game knowing that most of the meat will end up far beyond their own hearths is an intriguing one. It raises the issue of whether foragers pursue nonnutritional goals in food procurement and which goals they pursue. Here, I will use data from studies that have been carried out over a 34-year period among the Ju/'hoansi (!Kung Bushmen) to evaluate four hypotheses concerning why foragers target large game: reciprocity, costly signaling, nepotism, and long-term political goals aimed at bringing about conditions conducive to cooperative breeding. I will propose on the basis of qualitative and quantitative data that (1) all four hypotheses identify benefits that are gained by Ju/'hoansi through producing and distributing a surplus of meat, (2) different benefits might be important at different stages of a man's life, and (3) the pursuit of broader political goals merits serious consideration in accounting for men's work effort.     

Upper-body strength predicts hunting reputation and reproductive success in Hadza hunter–gatherers

Hunting is a characteristic feature of early human subsistence, and many theories of evolution have emphasized the role of hunting in hominization. Still today hunting ability continues to be selected for in extant foragers with better hunters experiencing greater reproductive success. Yet little is known about the traits that comprise a successful hunter traits that are presupposed to also be under selection. Two complementary empirical analyses were conducted to examine this question using data collected from Hadza hunter–gatherers in Tanzania. First, data on upper-body strength, running speed, target precision and visual and auditory acuity were collected to examine the traits that predict hunting reputation in men. Second, interview data were collected from Hadza informants regarding the traits they deem important for hunting. Results from the first study implicate upper-body strength as the strongest and most consistent predictor of men's hunting reputation. Hadza conventional wisdom also accord with these findings. Although informants stressed the importance of non-physical traits, such as “intelligence” and “heart”, strong arms were cited as the most important physical trait for hunting. Finally, men with stronger upper-bodies experienced greater reproductive success, a result that is largely mediated by hunting reputation. These findings suggest that selection for hunting ability may have acted on men's upper-bodies. Nevertheless, the importance of effort on strength and hunting success cannot be dismissed. This is also discussed.     

Conservation by native peoples

Native peoples have often been portrayed as natural conservationists, living a “balanced” existence with nature. It is argued that this perspective is a result of an imprecise operational definition of conservation. Conservation is defined here in contrast to the predictions of foraging theory, which assumes that foragers will behave to maximize their short-term harvesting rate. A behavior is deemed conservation when a short-term cost is paid by the resource harvester in exchange for long-term benefits in the form of sustainable harvests. An example of the usefulness of such an operational definition is presented using data on patch and prey choice decisions of a group of subsistence hunters, the Piro of Amazonian Peru. Results indicate that the area around the Piro village is depleted of prey, and that hunters allocate more time to patches where return rates are highest. This response is consistent with both a conservation strategy and foraging theory. Contrary to the expectation of the conservation strategy, however, hunters do not restrain from pursing opportunistically encountered prey in the depleted areas. The implications for conservation policy are briefly discussed. Michael Alvard received his Ph.D. at the University of New Mexico in 1993 and is currently an assistant professor of anthropology at Dickinson College. His interests include the impact of traditional hunting on prey species and the evolution of conservation. He is currently pursuing research opportunities with the Wana, a group of subsistence hunters living in the rain forests of Sulawesi, Indonesia.     

Use of hunters in wildlife surveys: does hunter and forest grouse habitat selection coincide?

Wildlife monitoring performed by volunteer personnel may suffer from bias with regard to their habitat use. Such errors can lead to erroneous population estimates, evidently influencing both management programmes and research that are based on the monitoring. We used a dataset on hunters’ habitat use in forest while searching for black grouse (Tetrao tetrix) and capercaillie (Tetrao urogallus) to test whether hunters’ utilisation of the habitat was independent of that of grouse or if it corresponded to the grouse habitat preference. Twenty volunteer hunters with dogs registered their tracks and all observations of capercaillie and black grouse in Østfold County, Norway, during August 2003 and 2004. We performed an ecological niche factor analysis (ENFA) and a K-select analysis with respect to hunters’ selection for habitat, described with ecogeographical variables related to forest stand characteristics, as well as the habitat preference of the observed grouse, conditional on the habitat utilisation of the hunters. Individual ENFA on the hunter's tracks revealed large variation in the habitat preferences of the hunters. The K-select indicated few overall patterns in the habitat characteristics of grouse observations, conditional on the hunters selected habitat. Accordingly, the results indicate that hunters’ observation of grouse prior to the hunt may give indicators of changes in grouse density unbiased by habitat preference due to the large between-hunter variation in habitat preference, given that a sufficient number of hunters is used. This suggests that such monitoring programmes can provide information about fluctuations in grouse population sizes valuable for both the management and research of forest grouse species.     

The effect of genotype on response thresholds to sucrose and foraging behavior of honey bees (Apis mellifera L.)

Honey bee foragers were tested for their proboscis extension response (PER) to water and varying solutions of sucrose. Returning pollen and nectar foragers were collected at the entrance of a colony and were assayed in the laboratory. Pollen foragers had a significantly higher probability of responding to water and to lower concentrations of sucrose. Bees derived from artificially selected high- and low-pollen-hoarding strains were also tested using the proboscis extension assay. Returning foragers were captured and tested for PERs to 30% sucrose. Results demonstrated a genotypic effect on PERs of returning foragers. The PERs of departing high- and low-strain foragers were consistent with those of returning foragers. The PERs were related to nectar and water reward perception of foragers. High strain bees were more likely to return with loads of water and lower concentrations of sucrose than foragers from the low pollen strain. Low-strain bees were more likely to return empty. We identified a previously mapped genomic region that contains a variable quantitative trait locus that appears to influence sucrose response thresholds. These studies demonstrate a gene-brain-behavior pathway that can be altered as a consequence of colony-level selection for quantities of stored food. Accepted: 3 September 1997     

Prey selection and annual harvest of game in a rural Zambian community

The selection and hunting patterns of Valley Bisa hunters were studied as part of a broader ecological investigation of human communities between the two sections of the Luangwa Valley Game Reserves in Zambia. This article provides information on the frequency and duration of hunts recorded for 1 year, yields of individual hunters, prey selection, frequency of kill, community meat supplies and the effect of human and carnivore predation on population levels on a study area of some 155 km². Individuals who hunted were classified into four categories–resident, transient, occasional, and official, dependent upon their role in the community, length of time spent on the study area and their village of primary residence. The most frequent hunters were residents followed in rank order by officials, transients and occasionals. Hunts were of shorter duration during the dry season, when game was close to villages, than during the wet season. Generally hunters who owned or had access to modern weapons were more successful in their pursuits of game than were those who used muzzle-loading guns. A total of 27 451 kg of meat (carcass yield) was made available to 466 local residents during the course of the year; 40% of this was attributed to the activities of the game guards. Valley Bisa hunters killed primarily buffalo, impala, warthogs, and waterbuck; most of these were males and adults. The kills recorded for carnivores suggest that they took primarily female buffalo and zebra. There was no evidence to suggest that either type of predation was damaging to the game herds on the study area.     

Clovis in context: New light on the peopling of the Americas

Until recently, the first Americans were thought to be fluted-point spear-hunters from the Siberian steppes. Near the end of the Ice Age, they followed big-game herds over the Bering land bridge into the open, upland habitats of the interior of North America about 12,000 years ago. Rapidly extinguishing the big game herds with their deadly hunting methods, they pressed southward in search of new herds and reached the tip of South America about a thousand years later. Today, nearly 70 years after the first excavations at Clovis, New Mexico, the type site for this culture, new sites and new dates from both North and South America are forcing a revision of the earlier picture of the migrations and adaptations of the first Americans. But despite recurring claims that human colonization of the Western Hemisphere began as early as 20,000 or more years ago with the arrival of generalized foragers lacking a projectile-point tradition, no definitive data gives empirical support for a human presence before c. 12,000 before the present (B.P.). All supposed pre-Clovis cultures except one in Alaska have failed to withstand careful scrutiny of their data. In addition, despite recent claims for cultural and biological links of the migrants to Europe or the Pacific Islands, the skeletons and cultural assemblages of Paleoindians throughout the hemisphere point consistently to a northeast Asian origin. According to new data, Paleoindian ancestors in Beringia c. 12,000 years ago were not specialized, fluted-point hunters of large game, but broad-spectrum hunter-gatherers using triangular or bipointed, lanceolates. Diverse cultures descended from these ancestors, not only the big-game hunting Clovis culture of the North American high plains. And just as Clovis did not set the cultural pattern for the hemisphere, it was not the earliest culture. Fully contemporary with the earliest possible Clovis dates of c. 11,200, in South America there already were maritime foragers on the Pacific coast, small-game hunters in the southern pampas, and tropical forest riverine foragers in the eastern tropical lowlands. The Clovis culture thus was just one of several regional cultures developed in the millennium after the initial migration. It could not have been the ancestor of the other early Paleoindian cultures. This new picture of Paleoindian cultures changes understanding of initial human adaptive radiation in the Americas and has implications for general theories of human evolution and behavioral ecology.     

Altruistic self‐removal of health‐compromised honey bee workers from their hive

Social insect colonies represent distinct units of selection. Most individuals evolve by kin selection and forgo individual reproduction. Instead, they display altruistic food sharing, nest maintenance and self‐sacrificial colony defence. Recently, altruistic self‐removal of diseased worker ants from their colony was described as another important kin‐selected behaviour. Here, we report corroborating experimental evidence from honey bee foragers and theoretical analyses. We challenged honey bee foragers with prolonged CO₂ narcosis or by feeding with the cytostatic drug hydroxyurea. Both treatments resulted in increased mortality but also caused the surviving foragers to abandon their social function and remove themselves from their colony, resulting in altruistic suicide. A simple model suggests that altruistic self‐removal by sick social insect workers to prevent disease transmission is expected under most biologically plausible conditions. The combined theoretical and empirical support for altruistic self‐removal suggests that it may be another important kin‐selected behaviour and a potentially widespread mechanism of social immunity.     

Correlated evolution of herbivory and food chemical discrimination in iguanian and ambush foraging lizards

To efficiently locate and assess foods, animal sensory capacitiesand behavioral discriminations based on them must be appropriatefor the diet and method of hunting. In lizards, actively foraginginsectivores identify animal prey using lingually sampled chemicalcues, but ambush foragers do not. Among plant eaters derivedfrom active foragers, plant chemical discrimination is addedto prey chemical discrimination, resulting in correlated evolutionof plant diet and plant chemical discrimination. Here I presentcomparative evidence on the relationships between plant dietand food chemical discrimination in Iguania, which consistsprimarily of ambush foragers and is one of two major lizardclades, and for ambushing lizards in general. Comparative analysesconducted using phylogenetic methods show that (1) all but onespecies of omnivore studied exhibited both prey and plant chemicaldiscrimination, whereas ambush foragers exhibited neither; (2)significant correlated evolution occurred between plant dietand plant chemical discrimination in Iguania and in omnivoresand herbivores derived from ambush foragers; and (3) correlatedevolution has occurred between prey and plant chemical discriminationin Iguania and, more generally, in taxa derived from ambushforagers. These results are explained by selection on planteaters to assess the nutritional value and possible toxicityof plants and by continued consumption of some animal prey evenin herbivores combined with freedom from factors that selectagainst prey chemical discrimination in ambush foragers.     

Prey chemical discrimination and strike-induced chemosensory searching in lizards: Their absence in a crotaphytid lizard (Crotaphytus collaris) and a proposal for research in zoos

Lizards in seleroglossan families of actively foraging carnivores and the herbivorous iguanids use the tongue to gather chemical samples to detect, identify, and locate food prior to attack, and to relocate lost food. In contrast, previously studied iguanian families other than Iguanidae lack lingually mediated prey chemical discrimination (PCD) and do not exhibit lingually mediated chemosensory searching behavior for bitten prey that has escaped or been lost (SICS = strike-induced chemosensory searching). In the present study, experimental tests showed that PCD and SICS are absent in Crotaphytus collaris, a member of the previously unstudied family of iguanian ambush foragers, Crotaphytidae. Available data suggest that in active foragers natural selection favors use of chemical cues to locate hidden prey, whereas in ambush foragers natural selection favors immobility to avoid detection by predators and SICS precludes simultaneous ambush. In most families the states of PCD and SICS are retained from the ancestors, but when foraging mode shifts, a change in chemosensory behavior appears to be induced. A proposal is made for a research program involving herpetologists at zoos. © 1996 Wiley-Liss, Inc.     

Eland, buffalo, and wild pigs: were Middle Stone Age humans ineffective hunters?

Patterns of faunal exploitation play a central role in debates concerning the behavioral modernity of Middle Stone Age (MSA) peoples. MSA foragers have been portrayed as less effective hunters than their Later Stone Age (LSA) successors on the basis of relative species abundances from ungulate assemblages in southern Africa. Specifically, MSA hunters are said to focus on docile eland while avoiding more aggressive prey, particularly buffalo and wild pigs. To evaluate these arguments and compare subsistence behavior, I present a quantitative examination of 51 MSA and 98 LSA ungulate assemblages from southern Africa to show that: (1) with respect to ungulate exploitation, MSA diet breadth may have exceeded LSA diet breadth, (2) ungulate assemblage evenness is equivalent in the MSA and LSA, (3) eland, buffalo, and wild pig are equally abundant in the MSA and LSA, and (4) large ungulate prey are more common in the MSA than in the LSA. With few exceptions, the broad patterns, which sample a range of geographic and environmental contexts, are supported by an environmentally controlled comparison of Middle and Later Stone Age faunas that accumulated under interglacial conditions along the southern African coastline. When interpreted within a foraging theory framework, these differences suggest that MSA hunters enjoyed increased meat yields due to elevated encounter rates with large prey. These results need not imply cognitive differences, but are consistent with an increase in human populations from the Middle to Later Stone Age, which resulted in diminished abundances of large ungulates.     

Growth-related changes in predation behavior in incipient colonies of the ponerine antEctatomma tuberculatum (Olivier)

Summary We traced the development in the laboratory of 18 young colonies of the arboricolous ponerine antEctatomma tuberculatum. Colony foundation is of the partially-claustral type. During the early stages, when the colony is entirely dependent on the queen's behavior, the growth of the colony in terms of number of workers produced over time was relatively predictable. Afterwards, divergence in colony growth in function of the time increases as fast as the number of workers influences the efficiency of colony provisioning.Comparative analysis indicated clear changes in the predation behavior of foundresses and workers as colonies developed. For any stage of colony growth, all individuals provisioned the nest with dead prey or sugar-rich substances in the same way. However, prey hunting involves two different strategies. Foundresses and nanitic workers (originating from colonies with 9–15 workers) foraged actively, catching prey as the result of random encounters. Post-nanitic foragers (originating from colonies with 20–30 workers) and those from nature colonies developed an ambush strategy. Workers in these colonies gained experience at catching and handling prey during a period when they acted as nest guards, and so tended to be more efficient hunters than poorly experienced foundresses or nanitic foragers. The change in strategy was also positively correlated with an increase in the size of workers as the colony matured. A stable maximum in workers size is apparently reached only after the appearance of efficiently hunting foragers, presumably in numbers sufficient to provide adequate quantity and quality of larval food. Such a correlation between worker size and colony growth, assumed general for all ants, has not been demonstrated for Ponerinae before this work.     

Partner choice in human evolution: The role of cooperation,foraging ability,and culture in Hadza campmate preferences

The ability to choose the partners we interact with is thought to have been an important driver in the evolution of human social behavior, and in particular, our propensity to cooperate. Studies showing that humans prefer to interact with cooperative others is often cited as support for partner choice driving the evolution of cooperation. However, these studies are largely drawn from Western samples, where conditions for partner choice to operate may be especially favorable. Here, we investigate qualities associated with being a preferred partner (i.e., campmate) in Hadza hunter-gatherers of Tanzania in 2016 and 2019. A total of 156 Hadza participants from 17 camps ranked their campmates on generosity, foraging ability, and their preference for them as future campmates. In 2016, Hadza preferred more generous people and better hunters as campmates, with evidence suggesting a stronger preference for better hunters; however, the relationship between generosity and being a preferred campmate was greater in 2019 than in 2016, such that the preference for generous people was stronger than the preference for better foragers, suggesting that campmate preferences are changing. These new findings contrast with reports on data from nearly a decade ago, suggesting that the Hadza do not prefer more cooperative campmates. Further, in 2019, there was anecdotal evidence that Hadza with greater exposure to outside cultural institutions (e.g., schooling, having a job, or living in a village) had a stronger preference for generous campmates than those with less exposure. Taken together, the results suggest that preferences for social partners may, in part, be culturally shaped.     

Dynamic matching of forager size to resources in the continuously polymorphic leaf-cutter ant, Atta colombica (Hymenoptera, Formicidae)

Abstract.  1. Ergonomic optimisation theory proposes that by increasing variation in worker morphology, social insect colonies may increase their dietary breadth; however, little is known about how this relationship operates at the colony level. This study examines the colony-level pattern of forager size allocation to resource sites in a natural setting.
2. Using a biologically relevant measure of toughness, it is shown that leaf-cutter ant colonies exploit a variety of plant resources that vary significantly in toughness at any given time.
3. Forager size is shown to be matched to the toughness of plant material, with larger ants harvesting tougher material.
4. Furthermore, outbound foragers travelling to a harvest site are matched in size to the toughness of plant material contained within the site and are not a random selection of available foragers. The match between forager size and plant toughness may reduce the number of wasted trips and ill-matched foragers.
5. The observed colony-level pattern of forager allocation could be the result of learning by individual foragers, or the result of information shared at the colony level.     

Evaluating Indirect Effects of Hunting on Mule Deer Spatial Behavior

Mule deer (Odocoileus hemionus) are widely hunted throughout western North America and are experiencing population declines across much of their range. Consequently, understanding the direct and indirect effects of hunting is important for management of mule deer populations. Managers can influence deer mortality rates through changes in hunting season length or authorized tag numbers. Little is known, however, about how hunting can affect site fidelity patterns and subsequent habitat use and movement patterns of mule deer. Understanding these patterns is especially important for adult females because changes in behavior may influence their ability to acquire resources and ultimately affect their productivity. Between 2008 and 2013, we obtained global positioning system locations for 42 adult female deer at the Starkey Experimental Forest and Range in northeast Oregon, USA, during 5-day control and treatment periods in which hunters were absent (pre-hunt), present but not actively hunting (scout and post-hunt), and actively hunting male mule deer (hunt) on the landscape. We estimated summer home ranges and 5-day use areas during pre-hunt and hunt periods and calculated overlap metrics across home ranges and use areas to assess site fidelity within and across years. We used step selection functions to evaluate whether female mule deer responded to human hunters by adjusting fine-scale habitat selection and movement patterns during the hunting season compared to the pre-hunt period. Mule deer maintained site fidelity despite disturbance by hunters with 72 ± 4% (SE) within-year overlap between summer home ranges and hunt use areas and 54 ± 7% inter-annual overlap among pre-hunt use areas and 56 ± 7% among hunt use areas. Mule deer diurnal movement rates, when hunters are active on the landscape, were higher during the hunting period versus pre-hunt or scout periods. In contrast, nocturnal movement rates, when hunters are inactive on the landscape, were similar between hunting and non-hunting periods. Additionally, during the hunt, female mule deer hourly movements increased in areas with high greenness values, indicating that mule deer spent less time in areas with more vegetative productivity. Female mule deer maintained consistent habitat selection patterns before and during hunts, selecting areas that offered more forest canopy cover and high levels of vegetative productivity. Our results indicate that deer at Starkey are adopting behavioral strategies in response to hunters by increasing their movement rates and selecting habitat in well-established ranges. Therefore, considering site fidelity behavior in management planning could provide important information about the spatial behavior of animals and potential energetic costs incurred, especially by non-target animals during hunting season. © 2020 The Wildlife Society.     

Behavioral responses of male elk to hunting risk

Prey respond to predation risk with a range of behavioral tactics that can vary based on space use and hunting mode of the predator. Unlike other predators, human hunters are often more spatially and temporally restricted, which creates a period of short-duration, high-intensity predation risk for prey. Consequently, identifying the roles different hunting modes (i.e., archery and rifle), hunts for targeted and non-targeted species, and landscape features play in altering spatial and temporal responses of prey to predation risk by humans is important for effective management of harvested populations. From 2009 to 2016, we used a large-scale experiment including 50 animal-years of location data from 38 unique male elk (Cervus canadensis) to quantify changes in movement and resource selection in response to hunters during 3 separate 5-day controlled hunts for antlered males (elk archery, deer [Odocoileus spp.] rifle, and elk rifle) at the Starkey Experimental Forest and Range in northeast Oregon, USA. We evaluated competing hypotheses regarding elk responses to varying levels of prey risk posed by the different hunt types. We predicted that the strength of elk behavioral responses would increase with perceived hunter lethality (i.e., weak response to elk archery but similar response to elk and deer rifle hunts) and that prey response would be closely associated with hunter activity within the diel cycle (greater during diurnal than nocturnal hours) and across hunting seasons. Elk responses were strongest during diurnal hours when hunters were active on the landscape and were generally more pronounced during both rifle hunts than during the archery hunt (supporting our perceived lethality hypothesis). Male elk avoided open roads across all periods except during nocturnal hours of the breeding season and alternated between avoidance of areas with high canopy cover during nocturnal hours and selection during diurnal hours. In combination these patterns led to distinct distributional changes of male elk from pre-hunt to hunt periods. Patterns of male elk selection highlight the importance of managing for heterogeneous landscapes to meet a variety of habitat, harvest, hunter satisfaction, and escapement objectives.     

Temporal variation in habitat selection breaks the catch‐22 of spatially contrasting predation risk from multiple predators

Predator avoidance depends on prey being able to discern how risk varies in space and time, but this is made considerably more complicated if risk is simultaneously present from multiple predators. This is the situation for an increasing number of mammalian prey species, as large carnivores recover or are reintroduced in ecosystems on several continents. Roe deer Capreolus capreolus in southern Norway illustrate a case in which prey face two predators with contrasting patterns of predation risk. They face a catch‐22 situation: spatially avoiding the risk from one predator (lynx Lynx lynx in dense habitat) implies exposure to the other (hunters in open habitat). Using GPS‐data from 29 roe deer, we tested for daily and seasonal variation in roe deer selection for habitat with respect to the habitats’ year‐round average risk level. Generally, roe deer altered their habitat selection between night and day in a pattern consistent with being able to avoid predicted risk from the nocturnal lynx during night and predicted risk from human hunters during day. However, seasonal variation in habitat selection only partially corresponded with the predicted seasonal variation in risk. Whereas roe deer avoided areas with high risk from hunters more strongly during the hunting season than in other seasons, there was a lack of selection towards areas and time periods lowering the risk of lynx predation during winter. It seems likely that the risk of starvation and thermal stress constrain roe deer habitat selection during this energetically challenging season with cold temperatures, snow cover and limited natural forage. The habitat selection pattern of roe deer fits thus only partly with the two contrasting risk gradients they face. Adjusting risk‐avoidance behavior temporally can be an adaptive response in the case of several predators whose predation patterns differ in space and time.