Front-country recreation statistics

We obtained data on total number of recreational visits (front-country park visits through entrance stations) to YNP for the years 1992–2023 from the National Park Service Visitor Use Statistics web page ( https://irma.nps.gov/Stats/SSRSReports/YELL/Yellowstone). We obtained recreation statistics for the years 1979–1991 from the YNP Visitor Services Office. To calculate the risk of bear attack for front-country recreationists, we divided the total number of park visits recorded at entrance stations during 1979–2023, by the number of bear-inflicted human injuries that occurred within or ≤250 m from developed areas, roadsides, and front-country trails. Incidents that occurred inside roadside campgrounds were analyzed separately.

Vehicle-accessible campground use statistics

Camping overnight in vehicle-accessible campgrounds may potentially incur greater risk of bear attack than staying in hotels, lodges, or cabins; therefore, we analyzed the risk of bear-inflicted human injury in campgrounds separately from other front-country areas. We obtained data on total number of overnight stays in vehicle-accessible campgrounds for the years 1992–2023 from the National Park Service Visitor Use Statistics web page. We obtained campground statistics for the years 1979–1991 from the YNP Visitor Services Office. To calculate the risk of bear attack while camping in vehicle-accessible campgrounds, we divided the number of overnight stays in campgrounds by the number of people injured by bears in campgrounds. We calculated the total number of overnight stays in campgrounds by adding together the number of nightly tent and RV campers in NPS-managed campgrounds, concessioner-operated campgrounds, and NPS- and concessioner-operated campground group campsites.

Backcountry campsite use statistics

We obtained records of the number of people obtaining backcountry permits to camp overnight in backcountry campsites and dispersed camping zones, and the total number of overnight stays in the backcountry, from the YNP Central Backcountry Office. Records of backcountry overnight stays for the years prior to 1992 did not distinguish the mode of travel used by backcountry campers. Therefore, our estimate of the risk of attack while in backcountry campsites combines all the different types of overnight backcountry campers (e.g., those traveling by foot, stock animal [horse, mule, llama], nonmotorized boat [canoe, kayak, sailboat], and motorized boat).

To calculate the risk of attack while in backcountry campsites, we divided the total number of backcountry overnight stays by the number of people injured by bears while at their campsites. Overnight stays are calculated by multiplying the number of people in each group obtaining a backcountry permit, by the number of nights spent camping in backcountry campsites. For example, a group of 3 people spending 2 nights in the backcountry is recorded as 6 overnight stays. The risk of attack was likely different between different modes of travel while people were traveling to their campsites. However, once backcountry travelers arrived at their campsites, the risk of bear attack was likely similar between all types of recreationists regardless of mode of travel.

Overnight backpacker recreation days use statistics

The YNP Central Backcountry Office has maintained records that distinguish the number of people traveling by foot while on overnight backcountry camping trips from 1992 to the present. For backpackers, we calculated the risk of attack while hiking, independently from the risk of attack while they were in their campsites. To calculate the number of recreation days backpackers spent hiking in the backcountry, we added the number of people in each party to the number of overnight stays for each party, to give us the number of backpacker recreation days. This method accounted for the last day of backpacking trips when backpackers were hiking out to the trailhead and would not be spending the night in the backcountry. For example, 3 backpackers spending 2 nights in the backcountry would be spending 3 days on the trail or a total of 9 recreation days (no. of people × no. of nights, + no. of people).

Overnight stock user backcountry recreation days

Records that distinguish the number of people traveling with stock animals and camping in the backcountry have been kept since 1992. For overnight stock riders, we calculated the risk of attack while riding to campsites independently from the risk of attack while they were in their campsites. To calculate the number of overnight stock rider recreation days, we added the number of people in each stock party to the number of overnight stays for each party. This method accounted for the last day of pack-trips when stock riders were riding out to the trailhead but would not be spending the night in the backcountry. For example, 8 stock riders spending 2 nights in the backcountry would be spending 3 days on the trail or a total of 24 stock rider recreation days (no. of people × no. of nights + no. of people).

Bear-inflicted human injury statistics

We collected data on bear-inflicted human injuries and fatalities during interviews with injured recreationists and incident field investigations conducted by YNP law enforcement and bear management office staff. These records included information on the activity of visitors and the location where the injury occurred. We compared the number of people injured or killed by bears when engaged in different recreational activities with the total number of park visitors that engaged in those activities.

Censored data

We censored the data from 7 bear-inflicted injuries to YNP and other agency or contracted employees that occurred while conducting job-related work activities. These included (1) a biological technician that was injured by a grizzly that woke up early from immobilization drugs during a research capture operation in 1981; (2) a park ranger that was bitten while moving an injured black bear off of the road in 1983; the bear had temporarily lost consciousness after being struck by a vehicle but regained consciousness and bit a ranger that was carrying it off of the road; (3) a biological technician that was injured by a grizzly bear while approaching its very high frequency (VHF) telemetry signal during a study of recreation impacts on bear movements in 1984; (4) a wolf project technician that was bitten by a black bear after approaching a wolf-killed elk carcass in 2007; (5) a wildland firefighter that was injured after surprising a grizzly while conducting a back-burn in the potential wildlife escape path from a large forest fire in high-density grizzly bear habitat in 2008; (6) a power company employee that was injured after surprising a grizzly while riding a utility task vehicle on a power-line corridor in 2008; and (7) a U.S. Geological Survey employee that was bitten by a grizzly that woke up unexpectedly from immobilization drugs during a research capture operation in 2019. These bear-inflicted human injuries were omitted from our analysis because YNP does not keep records of the number of staff (NPS, concessions, contractor) field days where employees are potentially exposed to risk of bear attack. Additionally, in some of those incidents the job duties (moving injured bears, trapping and handling bears, approaching ungulate carcasses, simulating hiker disturbance of bears during a research project, starting back-burns in high-density grizzly bear habitat) put agency or partner staff at greater risk of bear attack than the visiting public are generally exposed to.

Data analysis and statistical methods

We aggregated injury data by year and stratified them by like units of visitation (i.e., recreation days, overnight stays, or entrance station visit counts). The effects of bear species and recreation activity type, if applicable, on injury rates for each type of visitation were initially assessed using generalized linear mixed-effects models executed with the glmmTMB package (Brooks et al. 2017) in Program R version 4.3.1 (R Core Team 2023). We specified models with a Poisson error distribution and included the log of total visits, recreation days, or overnight stays for each year as an offset. We initially included year as a random effect because activity type and species were nested within each year. In most cases, the random effect term resulted in overparameterization and model convergence failures. In these instances, we dropped the random effect term and ran models again as generalized linear models (glms) with a Poisson error structure. The exception to this was the combined use model that pooled all visitor recreation types to assess the effects of species on injury rates, which converged and showed improved performance with the random effect term included versus glms. We assessed variable importance using backwards step selection. We used likelihood ratio tests (LRT) to compare models with and without main effect terms and interactions, where applicable, to determine statistical significance of effects. Models produced estimates as log rate ratios. We calculated rate ratios and 95% confidence intervals post hoc by exponentiating model coefficients and confidence intervals. We checked models for overdispersion and zero inflation using the Program R package “performance” (Lüdecke et al. 2021) and assessed fit using standard model checks.

Definitions of terms

Unique terms used in this manuscript are defined as follows:

All other human–bear interaction terms we used are consistent with definitions by Hopkins et al. (2010).

Injuries in front-country areas

Total park-visits ranged from 1,892,908 to 4,860,537 with a mean of 3,097,635 visits/year from 1979 to 2023. There were 2 injuries caused by grizzly bears and 1 injury caused by black bears in front-country areas during this period, yielding an injury rate of 1/ 69,696,787 visits for grizzly bears and 1/139,393,574 visits for black bears (Table 1). The model-estimated mean rate ratio of 2.00 (95% CI [0.18, 22.81]) indicated a trend toward grizzly bears causing injuries at a rate roughly twice that of black bears in front-country areas (Table 2). However, this effect was not statistically significant (LRT, χ²₁ H 0.34, P H 0.56; Fig. 2A).

Table 2.

Model summaries and estimated rate ratios derived from generalized linear (A–C) and mixed-effects (D) models examining the risk of being injured by grizzly bears (Ursus arctos) or American black bears (U. americanus) while engaged in different recreational activities in Yellowstone National Park, Wyoming, Montana, and Idaho, USA, 1979–2023. Black bears serve as the reference group contained in the model intercept values. Rate ratios and 95% confidence intervals were calculated post hoc by exponentiating model coefficient and confidence intervals. Significant effects (P < 0.05) of explanatory variables evaluated using Likelihood Ratio Tests are denoted with an *.

Fig. 2.

Model-estimated effects of bear species (black [Ursus americanus] or grizzly [U. arctos]) on the rate of bear-inflicted injury to park visitors in Yellowstone National Park, USA, from 1979 to 2023. (A) In developed areas, roadsides, or front-country (FC) trails; (B) in vehicle campgrounds (CG) and backcountry (BC) campsites, with an additional effect of campsite type; (C) for overnight backpackers and horseback riders; (D) for all visitor types combined with additional inclusion of day hikers and day stock users. Black bears serve as the reference group contained in the model intercept values. Points are model-estimated mean log rate ratio and error bars represent 95% confidence intervals. Where intervals include zero, effect was considered unimportant and variable significance was formally assessed using likelihood ratio tests for verifiation.

Injuries in vehicle-accessible campgrounds versus backcountry campsites

Annual overnight stays in vehicle-accessible campgrounds from 1979 to 2023 ranged from 448,286 to 791,382 with a mean of 648,015. Overnight stays in backcountry campsites for the same period ranged from 25,188 to 55,060 with a mean of 41,231/year. Grizzly bears caused 1 injury in vehicle-accessible campgrounds and 1 in backcountry campsites, while black bears caused no injuries in campgrounds and 3 injuries in backcountry campsites. This yields an injury rate of 1/29,160,667 overnight stays for grizzly bears and 0/29,160,667 overnight stays for black bears in vehicle-accessible campgrounds, and 1/1,855,377 overnight stays for grizzly bears and 1/618,459 overnight stays for black bears in remote backcountry campsites (Table 1). Model-estimated mean rate ratios indicated that there was a highly significant effect of campground type, with a ∼98% lower rate of bear-inflicted injury (RR H 0.016; 95% CI [0.002, 0.144]; Table 2) in vehicle-accessible campgrounds than in backcountry campsites (LRT, χ²₁ H 17.65, P < 0.0001; Fig. 2B). Grizzly bears caused fewer injuries in backcountry campsites, with a model-estimated mean ∼43% lower rate of injuries (RR H 0.667, 95% CI [0.11, 4.03]) than that of black bears (Table 2); however, this effect was not statistically significant (LRT, χ²₁ H 0.201, P H 0.65; Fig. 2B). Accordingly, there was no significant species–location interaction (LRT, χ²₁ H 2.231, P H 0.14), resulting in only campground type being retained in the best fit model.

Injuries to overnight backcountry users outside of backcountry campsites

Annual backcountry recreation days for overnight backpackers ranged from 36,316 to 50,437 with a mean of 42,960. Annual overnight horseback riding recreation days ranged from 2,779 to 6,714 with a mean of 4,663. Sufficient data to calculate recreation days for these users were only available from 1992 to 2023. There were 5 injuries to backpackers caused by grizzly bears and 1 injury caused by black bears yielding a rate of 1 injury/275,144 recreations days for grizzlies and 1/1,375,721 recreations days for black bears (Table 1). No injuries to horseback riders were recorded for either species. No bear-inflicted human injuries have been recorded to overnight stock users in the 32-year data set, so models were unable to accurately estimate an effect size for overnight horseback riders compared with backpackers (log rate ratio H –1.98, 95% CI [–49,493, 49,454]), with standard errors approaching infinity. Therefore, backcountry overnight horse and backpacker users were modeled using a Poisson generalized linear model examining only the effect of species on the rate of injury in backcountry areas outside of campsites among overnight users. The model-estimated mean rate ratio for grizzly bears versus black bears was 5.00 (95% CI [0.57, 43.70]; Table 2). This suggests that in backcountry areas outside of campsites, the injury rate attributable to grizzly bears is approximately 5 times the rate of injury caused by black bears. This effect was marginally significant (LRT, χ²₁ H 2.911, P H 0.088; Fig. 2C).

Combined injury rate for all park users and locations

Total combined annual park visits ranged from 1,892,908 to 4,860,537 with a mean of 3,097,635 visits. This combined data set includes injuries to day hikers and stock users that were not included in the other data sets analyzed. From 1979 to 2023, there were a total of 39 injuries caused by grizzly bears and 6 caused by black bears yielding a rate of 1 injury/3,574,194 recreational visits for grizzly bears and 1 injury/23,232,262 visits for black bear (Table 1). There was a highly significant effect of species (LRT, χ²₁ H 27.043, P < 0.0001; Fig. 2D) with model-estimated average injury rates attributable to grizzly bears being approximately 6.5 times higher than those of black bears (RR H 6.50, 95% CI [2.72, 15.54]) across all years and visitor use types combined (Table 2).

Grizzly bear

Our calculated grizzly bear injury rate for all types of recreational activities combined in YNP is within the range of rates reported for other Canadian and U.S. national parks. The rate of 1 grizzly bear–inflicted human injury/3.6 million recreational visits to YNP is higher than the rates reported for Kootenay (1 injury/12.6 million visits), Yoho (1 injury/6.7 million visits), and Banff (1 injury/4.1 million visits) national parks, but lower than the rates for Mount Revelstoke and Glacier (Canada; 1 injury/2.6 million visits), Waterton (1 injury/2.0 million visits), Jasper (1 injury/1.2 million visits), Glacier (USA; 1 injury/0.9 million visits), Denali (1 injury/0.3 million visits), and Kluane (1 injury/0.3 million visits) national parks (Herrero 1976, Herrero and Fleck 1990, Herrero and Higgins 1999).

Although the park-wide grizzly bear–inflicted human injury rate makes it clear that grizzlies are not very dangerous to the average YNP visitor, not all visitors had equal risk of exposure to grizzly bear attack. Visitors that remained within front-country areas had a much lower risk of being injured by grizzly bears (1 injury/ 69.7 million visits) than did other user groups. Camping overnight in roadside campgrounds more than doubled the risk of injury compared with recreating in other front-country areas, although risk remained at a very low level (1 injury/29.2 million overnight stays). Backcountry camping entailed much greater, but still very low, risk (1 injury/1.9 million overnight stays); that risk was >7× greater during hiking between campsites (1 injury/ 275,144 recreation days) than while at a campsite.

During our study period, no people on backcountry overnight horse pack-trips were injured by grizzly bears while riding or in their campsites. However, 2 day-riders were injured, and neither were on their horses when attacked. One was walking, leading his mule when he surprised a grizzly in its daybed. The other was sitting under a tree eating lunch with his horse hobbled nearby when a grizzly climbed out of an adjacent steep gulley, was surprised by the man's presence, and reacted with defensive aggression injuring him slightly before fleeing. In general, people riding stock are less likely to be involved in surprise encounters and therefore defensive bear attacks (Herrero 1970, 1976). Horses and mules have more acute senses of sight, hearing, and smell than people do and will usually detect a bear's presence before a person does, alerting the rider and reducing the chances of surprise encounters at close distances where bears are more likely to respond with defensive aggression (Herrero 2002). The large size of horses and mules is also more intimidating to bears, making them less likely to make contact during defensive charges after surprise encounters. Additionally, unlike humans, when charged by bears, horses and mules have enough speed and agility to outrun bears, thus providing an added margin of safety if the rider can stay in the saddle.

The grizzly bear injury rate for backpackers while they were hiking in YNP (1 injury/275,144 backpacker recreation days) was lower than the risk reported for hunters in Scandinavia (1 injury/144,000 hunters; Støen et al. 2018) and for hikers in Banff (1 injury/53,220 recreation days), Jasper (1 injury/38,175 recreation days), Mount Revelstoke and Glacier National Parks, Canada (1 injury/ 2,620 backcountry recreation days; Herrero 1976, 2002), but considerably higher than reported for all recreationists combined in the Bob Marshall Wilderness of Montana, USA (1 injury/4.5 million recreation days; U.S. Fish and Wildlife Service 2000). However, the recreation data from the Bob Marshall Wilderness combined campsite, hiking and stock use data; therefore, these data are not directly comparable to the backpacker recreation days we calculated for YNP.

Yellowstone National Park does not collect records of the number of day-use recreationists backcountry hiking in the park, so the risk of attack for that user groups could not be calculated. However, we believe the risk of grizzly bear attack for day-hikers is likely very similar to that incurred by overnight backpackers injured while they were hiking (1 injury/275,144 backcountry recreation days).

Black bear

Five of the 6 black bear–inflicted injuries involved people in prone or supine positions; 4 of the 5 were sleeping when bitten. The bites to people in these positions may have been curious soft nips to test people as potential prey but did not appear to be full predatory attacks. In most documented predatory diurnal attacks on people in North America, black bears were extremely focused on their intended prey and not easily dissuaded (Herrero 2002, Smith 2006). In all 5 black bear incidents with sleeping or resting people during our study, the bears disengaged with the victims after they yelled, sat up, stood up, or the bear was driven off by a companion.

Our calculated rate for all types of recreational activities in YNP combined, 1 black bear–inflicted human injury/23.2 million visits, is similar to the rate of 1 injury/21.0 million visits reported for Yoho National Park (Herrero and Higgins 1999) but considerably lower than the black bear–inflicted injury rates reported for Jasper (1/2.8 million visits), Great Smoky Mountains (1/2.6 million visits), Waterton (1/2.0 million visits), and Yosemite (1/0.6 million visits) national parks (Herrero and Fleck 1990).

Grizzly bear versus black bear comparison

In most areas of YNP, grizzly bears posed greater risk to visitors than did black bears. From 1979 to 2023, grizzly bears inflicted ∼6.5× more injuries than did black bears parkwide, even though black bears occurred at higher densities. However, black bears injured people in backcountry campsites more often than did grizzlies. Overall, grizzly bears are more aggressive toward people than black bears are (Herrero 1978, 2002; Stirling and Derocher 1990). Our data from YNP support that conclusion. The overall risk (for all recreational activities combined) of being injured by a grizzly bear in YNP was 1 injury/3.6 million visits, compared with 1 injury/23.2 million visits inflicted by black bears (Table 1). The greater aggressiveness of grizzly bears was also readily apparent for attacks on backcountry hikers. In backcountry areas, grizzly bears injured 1 backpacker for every 275,144 backpacker recreation days, whereas black bears injured 1 backpacker for every 1.4 million recreation days.

The greater defensive aggressiveness by grizzly bears as compared with black bears in North America may be a result of their evolutionary history (Herrero 1978, Stirling and Derocher 1990). Grizzly bears evolved to select more open, nonforested habitats where lack of trees dictated that females with cubs must directly confront threats and defend their cubs on the ground, thus favoring defensive aggression as an evolutionary trait (Herrero 1978, Stirling and Derocher 1990). In contrast, black bears evolved inhabiting forested environments where cubs can escape predators by climbing trees (Herrero 1978, Stirling and Derocher 1990). Therefore, female black bears do not need to engage threats directly but can send their cubs up trees to escape harm, flee themselves, then come back for their cubs after the threat has gone (Herrero 1978, Stirling and Derocher 1990).

Trends in bear-inflicted human injuries

Injury rates caused by grizzly bears and black bears in YNP steadily declined during our study. The decline occurred concurrently with park visitation increasing by approximately 2.5× over the same period (Fig. 1). We attribute the decline in bear attacks to YNP's efforts to prevent bears from obtaining human foods, garbage, and other anthropogenic attractants (Leopold et al. 1969, Cole 1971, Meagher and Phillips 1983), combined with educational efforts to teach park visitors how to react to surprise encounters, interactions, and confrontations with bears when they occur (Table 3). Additionally, YNP has designated 16 Bear Management Areas (BMA) encompassing 21% of the highest quality bear habitats within the park. Recreational activity is limited within BMAs through a variety of seasonal trail, campsite, and area closures, requirements of no off-trail travel, and prohibitions of backcountry travel during nocturnal and crepuscular time periods, implemented during seasons when bear activity is concentrated on specific high-quality foods in predictable locations. The BMAs have likely reduced human–bear encounters, interactions, and confrontations within the park (Coleman et al. 2013, Loggers et al. 2023).

Table 3.

Contact point and media used to provide bear safety information to visitors to Yellowstone National Park, Wyoming, USA.

Relevancy of Yellowstone National Park injury rates to other areas

The probabilities of bear attack for different recreational activities in YNP are likely comparable to the probabilities of attack on many other public lands occupied by grizzly bears in North America. However, because big game hunting is not allowed inside YNP, the risks of bear attack in the park are not comparable to areas where hunting occurs, at least not during the hunting season. Most grizzly bear attacks in the Greater Yellowstone Ecosystem (GYE) outside of YNP involve ungulate-hunting-related activities (Gunther et al. 2004). In Scandinavia, hunters' per capita risk of bear attack was ∼12× higher than that of nonhunters (Støen et al. 2018). Although ungulate hunters are often vigilant, to be successful they generally do not hunt in large groups, stay on designated trails, or make noise to forewarn bears of their presence like many other types of recreationists do. Staying on maintained trails, making noise in areas with limited visibility, and traveling in large groups reduce the risk of surprise encounters and defensive reactions by bears (Herrero 2002, Herrero and Higgins 2003, Smith and Herrero 2018, Gunther and Haroldson 2020), but would likely significantly reduce ungulate hunter success rates.

Additionally, coastal salmon (Oncorhynchus spp.) –feeding brown bears generally occur at higher population densities than do interior populations, which can result in bear-to-bear habituation and much shorter overt reaction distances (ORD), making coastal bears less likely to react with defensive aggression during surprise encounters with people (Smith et al. 2005). In contrast, interior brown bear populations such as the Greater Yellowstone Ecosystem population, generally occur at lower densities, leading to longer ORDs and making interior bears more likely to respond to surprise encounters with defensive aggression (Smith et al. 2005). Higher levels of food stress among interior bear populations, which generally live in less productive habitats and do not have access to abundant coastal salmon spawning runs, may also contribute to higher rates of defensive aggression by interior brown bears (Aumiller and Matt 1994; Herrero et al. 2005, 2011; Smith et al. 2005). Therefore, the probabilities of grizzly bear attack we calculated for YNP are not likely comparable to occupied grizzly bear habitats where ungulate hunting occurs nor to coastal lands with abundant salmon populations.

Bear safety messages

From an early age most people are taught behaviors that decrease the risk of injury in an urban setting; however, much less effort is expended to teach people how to safely enjoy outdoor activities in a wilderness environment (Penteriani et al. 2016). Different people learn most efficiently in different ways, so YNP uses a wide variety of media to promote bear safety. Media used include written signs, posters, brochures, and web page material; video links; and social media posts (Table 3). Yellowstone National Park attempts to educate visitors both before they arrive and while they are present in the park (Table 3). Although YNP implements an extensive bear-safety messaging program, many visitors do not follow the park's safety recommendations (Gunther 1990, Gunther et al. 2023b). For example, based on bear encounter–party size research (Herrero 1970; Chester 1976, 1980), YNP recommends recreationists hiking into the backcountry travel in groups of ≥3, yet most visitors hike alone or with just one partner (Oosterhous 2000, Gracia-Longares 2005, Coleman et al. 2013, Gunther et al. 2023b). Additionally, based on bear spray research (Hunt 1984, Rogers 1984, Herrero and Higgins 1998, Smith et al. 2008), YNP also recommends all backcountry hikers carry capsaicin-based bear deterrent spray, but only 57% of overnight backpackers and 18% of day hikers comply with that recommendation (Gunther et al. 2023b). Bear safety messaging is especially challenging because, even though the consequences of bear attack can be quite severe (severe injuries or even death; Herrero 1970, Gunther 2022), the risk of attack for most visitors is extremely small, 1 grizzly bear–inflicted human injury/ 3.6 million park visits and 1 black bear–inflicted injury/ 23.2 million visits. To add perspective to those risk calculations, the injury rate (fatal and nonfatal combined) for passenger vehicle crashes in the USA from 1999 to 2003, was 1 injury/123,092 person trips (calculated from Beck et al. 2007). It is difficult to convince visitors to adhere to bear safety recommendations to reduce the risks of an already rare event.

In YNP, improvements in information and education efforts aimed at recreational safety in bear country are paramount in the face of significant increases in visitation, concurrent with grizzly bear and black bear recovery in YNP and the GYE following the high mortality incurred after the closure of garbage dumps throughout the GYE in the 1960s and 1970s (Craighead et al. 1974, Haroldson and Gunther 2013, Meagher and Phillips 1983, Schullery 1992). As visitation continues to increase, innovative messaging campaigns may be needed to improve visitor compliance with bear safety recommendations. The effectiveness of any new educational campaigns designed to improve visitor safety in bear country should be evaluated. Reducing the frequency of bear-inflicted human injuries is important not only for visitor safety, but also for reducing management removals in threatened bear populations and for maintaining societal support for bear conservation (Røskaft et al. 2007).