Evidence for a metabolic trap and habituation limiting post-release survival of orphaned grizzly bears

The rehabilitation of orphaned grizzly bear (Ursus arctos) cubs of the year is often driven by public and animal welfare considerations, yet its efficacy as a functional conservation tool remains poorly understood. This study evaluates a five-year (2019–2023) captive-rear and release initiative in British Columbia (BC), Canada, representing the only systematically monitored program of its kind in North America. I monitored 14 orphaned cubs to examine how age of entry, captive-rearing duration, social environment (number reared together), and body mass at release influenced post-release survival. Of these, 13 bears were subsequently monitored post-release. These releases also provided a unique opportunity to examine whether greater release mass confers a survival advantage, a relationship previously suggested to reduce predation risk in captive-reared ursids. Results were benchmarked against morphometric data for wild yearlings in western North America. Captive-reared yearlings (n = 14) exhibited significant physiological divergence; mean spring release mass (127.4 kg) was 3.5-fold greater than wild-reared counterparts (36.1 kg; p < 0.0001). Despite this size advantage, the confirmed mortality rate was 83% among bears with known outcomes (n = 10/12). Mortality was heavily concentrated in the short-term, with 75% of the bears failing to survive from release to first denning. Contrary to rehabilitation best practices, minimizing handlers did not prevent habituation; several bears quickly generalized human presence with food security post-release. Survival was inversely related to captivity duration; increased time in captivity, driven by earlier age of intake, was associated with reduced survival, likely reflecting the combined loss of maternal learning and prolonged habituation. Furthermore, the relationship between release mass and survival was non-linear; individuals exceeding 135 kg consistently failed to recruit, suggesting a ‘metabolic trap’ where oversized frames create energetic demands difficult to meet for naïve yearlings foraging without maternal guidance in resource-limited spring landscapes. If rearing protocols diverge too far from natural conditions, facilitating habituation through routine husbandry and producing developmental outliers, rehabilitation remains a form of sympathetic management rather than a conservation tool. I advocate for species-tailored, “hands-off” protocols that prioritize natural growth trajectories and behavioural wariness during captive-rearing.

Empirical evidence on the post-release survival and causes of mortality of captive-reared orphaned grizzly/brown bears (Ursus arctos) is extremely limited worldwide. Furthermore, systematic monitoring to determine end fate is particularly uncommon. Although the captive rearing and release of orphaned or injured bears has increasingly been applied across seven of the eight species within the family Ursidae (Beecham, 2006), including European brown bears (Huber et al., 1994; Pop et al., 2012), Asiatic black bears (U. thibetanus) (Han and Jung, 2006), Andean bears (Tremarctos ornatus) (Castellanos, 2005), and sun bears (Helarctos malayanus) (Fredriksson, 2001), published information for most remains restricted to descriptive accounts, conference proceedings, or short-term observations. In contrast, the most extensive peer-reviewed rehabilitation literature exists for the American black bear (U. americanus), where captive-rearing and release programs have been implemented for more than three decades (Clarke et al., 1980; Alt and Beecham, 1984; Carney and Vaughan, 1987; Blair et al., 2020). As a result, rehabilitation protocols and assumptions regarding post-release success are often broadly inferred from American black bear studies despite ecological and behavioural differences among ursid species. This limitation is particularly relevant for grizzly bears, where concerns persist that increased human exposure during captive rearing may promote habituation to humans and subsequent conflict behaviour (Herrero et al., 2005; Huber et al., 2010), rendering robust, long-term empirical evaluations of their survival, mortality, and behavioural integration critical yet exceptionally rare.

Globally, only a handful of captive-rearing programs focus on orphaned grizzly bears. Among these, only three programs, located in Canada, Romania, and Greece, have implemented systematic post-release monitoring to evaluate outcomes and ultimately their effectiveness. In western Russia, the Pazhetnov facility has released more than 200 grizzly bears, but little monitoring data are available (Beecham, 2006). France and Italy have also released cubs, although without consistent or long-term monitoring programs. Spain is currently establishing a facility for captive-rearing and release but data are limited to short-term tracking devices attached to the fur. Consequently, robust data on the success or failure of released grizzly bears remain limited, and this gap becomes a greater concern where rehabilitation is being used as a conservation tool to support small or declining populations (Komnenou et al., 2018).

Traditional wildlife rehabilitation typically focuses on the medical treatment of ill or injured animals. Animals that are undergoing medical rehabilitation or those that require bottle feeding are unavoidably exposed to repeated human presence and handling, which can reduce wariness toward anthropogenic environments (Blair et al., 2020). This habituation can increase post-release use of human-associated resources and the risk of human–wildlife conflict (Blair et al., 2020). In contrast, the bears examined in this study were predominantly healthy wild-born grizzly bear cubs of the year (COY) that were orphaned. The primary objective of their captive rearing was not medical recovery, but the development and maintenance of natural behaviours necessary for independent survival and long-term recruitment when returned to the wild as yearlings.

Since 2007, the Province of British Columbia (BC), Canada, has authorized the rehabilitation of orphaned grizzly bear COY at the Northern Lights Wildlife Society (NLWS) in Smithers. To my knowledge, NLWS is the only facility in North America currently permitted to captive-rear and release orphaned grizzly bear cubs back into the wild. Although the NLWS was provincially permitted to house orphaned bear cubs, no formal government-regulated protocols exist governing how cubs were to be reared while in captivity. The provincial permit conditions of the facility defined general operational and release requirements but did not prescribe detailed husbandry practices, such as diet composition, feeding regimes, or behavioural conditioning protocols.

In BC, wildlife rehabilitation practices are primarily informed by the Minimum Standards for Wildlife Rehabilitation developed by the Wildlife Rehabilitators’ Network of BC (WRNBC, 2012), which establish professional standards intended to guide rehabilitators on acceptable practices, animal care, and release considerations in the province. These guidelines draw on broader Minimum Standards for Wildlife Rehabilitation developed through international collaboration (Miller, 2012) and provide a framework for the care and eventual release of orphaned wildlife. While they articulate general principles of animal welfare, behavioural conditioning, and criteria for release suitability, this framework functions as best-practice guidance and is broadly applicable across a wide range of wildlife rather than prescribing detailed species-specific rearing protocols or enforceable regulatory requirements.

Of note is the lack of prescriptive, operational protocols for grizzly bear captive rearing, especially within a conservation context. The protocols employed at NLWS align with the internationally recognised guidelines for orphan bear rehabilitation (Beecham, 2006; Beecham et al., 2016), which also serves as the methodological foundation for several European rehabilitation initiatives, including programs in Greece (Komnenou et al., 2018) and Italy (Morini et al., 2017). Best captive-rearing practices emphasize large enclosures, reduced human contact, and conspecific socialization to promote bonding among bears rather than habituation to people (Beecham, 2006; Beecham et al., 2016). The guidelines also recommend nutritionally dense supplemental foods, including dry dog food, to support growth and ensure dietary adequacy during captivity. Beecham (2006) further supports spring release timing to coincide with natural family break-up and suggests that retaining cubs until yearling age may allow increased body mass that could reduce vulnerability to predation. However, while these guidelines provide important best-practice principles (Beecham, 2006; Miller, 2012; Beecham et al., 2016), they remain somewhat general recommendations rather than species-specific operational criteria. As a result, key husbandry decisions, including feeding strategies and behavioural conditioning, were primarily left to the discretion of the facility, highlighting an important gap in standardized, policy-driven captive rearing protocols for grizzly bears. In particular, while increased release weight was found to reduce the post release predation risk for black bears (Beecham et al., 2015) criteria for body mass targets at release are not explicitly defined. Similarly, while minimal human contact is widely recommended (Beecham, 2006; Beecham et al., 2016), operational standards provide limited direction on how this should be consistently achieved during routine husbandry (e.g., feeding and enclosure maintenance), creating potential variability in human–bear exposure during captivity.

The captive rearing and release of grizzly bears in BC and elsewhere in Europe has been reported as successful (Beecham et al., 2015); however, the criteria and temporal scale used to define success was limited. In particular, collar retention for grizzly bears was insufficient to generate robust post-release survival estimates, and evaluations of success were largely based on the absence of subsequent human–bear conflicts or records of hunter harvest (Beecham et al., 2015). Collars that went off-line were right censored in which the bear was assumed to still be alive, whereas the actual survival time for these individuals is known only to exceed the duration of their active monitoring period. By 2019, the provincial government determined that the existing data on post-release survival was insufficient to thoroughly assess the success of the BC captive-rearing and release program. The Orphaned Grizzly Cub Rewilding Research Project (hereafter Project Rewild) began in 2019 with the objective of evaluating the facility’s rearing methods and collecting long-term data on the post-release survival of captive-reared grizzly bears. At that time, two grizzly bear COYs were being captive-reared for release in June 2020.

To my knowledge, Project Rewild is the only study that has attempted to follow all released orphaned grizzly bears through to confirmed outcomes. This level of monitoring provides rare insight into survival trajectories, mortality risk, and behavioural outcomes following release. Although the sample size is small, it reflects the biological and logistical reality of studying orphaned, captive-reared grizzly bear cubs rather than a limitation of study design. Importantly, complete post-release monitoring was feasible because the number of eligible bears admitted during the study period remained sufficiently low to allow intensive follow-up of each individual. Grizzly bear COYs are infrequently reported orphaned in BC, and opportunities to rear and release such individuals under controlled conditions are rare. Data collection required five consecutive years (2019–2023), representing the full intake of eligible orphaned cubs in BC during that period.

Rehabilitating wildlife, carries strong social and ethical appeal and resonates with public values of compassion and responsibility, particularly for charismatic species such as large carnivores (Veríssimo et al., 2017). The release of a rehabilitated animal provides a visible and emotionally compelling outcome, reinforcing public engagement with wildlife conservation. Rehabilitation centres can therefore influence societal attitudes even when measurable conservation outcomes remain uncertain, as narratives of rescue and release are widely communicated through media and social platforms (Goldenberg et al., 2022).

The conservation value of rehabilitation is often debated in the context of resource allocation. While individual rescues satisfy immediate animal welfare concerns, they represent a reactive approach to the broader drivers of human–carnivore conflict. For example, if the primary causes of orphaning are anthropogenic, such as lethal management removals due to unsecured attractants, the high cost of rearing a single orphaned cub must be weighed against the potential efficacy of investing those same resources into proactive conflict mitigation. However, rehabilitation has also contributed to population recovery, most notably in the restoration of the Asiatic black bear in South Korea (Jeong et al., 2010). Returning animals to the wild can provide demographic and genetic benefits to small or declining populations of conservation concern (Jeong et al., 2010; Goldenberg et al., 2022), potentially increasing population size, enhancing genetic diversity, and reducing risks associated with inbreeding and demographic stochasticity (Goldenberg et al., 2022). These contributions may be particularly meaningful for species with slow reproductive rates, including bears, where the survival of even a small number of individuals can influence long-term population trajectories (Proctor et al., 2004). Nevertheless, without evaluating whether released individuals survive and contribute to population recruitment, it remains unclear whether rehabilitation functions as a robust conservation tool or primarily as a form of sympathetic management (Jule et al., 2008).

To link individual animal care with population-level conservation outcomes, Goldenberg et al. (2022) proposed the “4 R’s” framework, which integrates rescue, rehabilitation, release, and research within a structured monitoring and management context. This framework aligns rehabilitation practice with conservation objectives and follows established guidance for wildlife reintroductions and translocations (Huber, 2010; IUCN/SSC, 2013). Under the International Union for Conservation of Nature (IUCN) guidelines, the activities described here are categorized as reinforcement when individuals are released into an existing population, or reintroduction when individuals are released into suitable areas of their historical range from which they have been extirpated. Application of such frameworks requires empirical evaluation of post-release outcomes, including survival, reproduction, and causes of mortality. Without these data, rehabilitation remains difficult to assess as a conservation tool.

Most grizzly bear population units in BC are considered secure, with conservation concerns concentrated in a small number of threatened or declining units (Morgan et al., 2020). Developing science-based benchmarks within stable and robust populations provides an important opportunity to evaluate rehabilitation outcomes under conditions where post-release mortality is unlikely to affect population viability. Findings from such systems can inform rehabilitation practices and conservation planning for grizzly bears and more broadly across bear species. At the same time, this evaluation addresses immediate management questions relevant to specific high-priority recovery units where the consequences of post-release failure are substantially greater. For example, in areas requiring active population augmentation, such as BC’s Stein-Nahatlatch and the transboundary North Cascades ecosystems, proposals to translocate healthy wild subadult females from source populations must be weighed against the potential use of available orphaned bears. However, the conservation value of captive-reared orphans as an alternative management tool depends on their ability to survive, function, and integrate successfully following release. Quantifying these outcomes in secure populations therefore provides a necessary foundation for evaluating whether captive-reared bears can contribute meaningfully to recovery efforts in small or threatened populations, both locally and elsewhere.

To determine how specific husbandry practices influence survival this study evaluates the post-release outcomes of orphaned, captive-reared grizzly bears returned to the wild as yearlings. Survival and reproductive success were quantified by monitoring individuals until a definitive fate was established, defined as confirmed survival, confirmed mortality, or the cessation of monitoring due to mechanical collar failure. Specifically, I assess how the duration of captive rearing, cohort socialisation, and specific husbandry practices, such as overwinter feeding, influence subsequent survival probability and the risk of human–bear conflict.

These releases provide a unique opportunity to test the assumption that greater release mass confers a survival advantage, a relationship previously suggested to reduce predation risk in captive-reared ursids (Beecham et al., 2015). I further compare these outcomes to age-specific survival rates reported for wild grizzly bear populations in western North America, including direct comparisons between captive-reared and wild-reared yearlings.

The primary research questions addressed in this study are: (1) how temporal factors, specifically the age of entry and the total duration of rearing in captivity, affect the probability of post-release survival; (2) to what extent the social environment defined as number of cubs reared together influences long-term outcomes; and (3) whether the elevated body mass of captive-reared yearlings translates to a functional survival advantage. I hypothesised that increased time in captivity, particularly for bears entering the facility at a younger age, would be associated with reduced post-release survival due to prolonged exposure to human-mediated environments. Furthermore, I hypothesised that the relationship between release mass and survival would be non-linear, with survival potential reduced at both low and high extremes of body condition. More broadly, this research contributes to the limited global literature evaluating the conservation viability and ecological consequences of releasing captive-reared large carnivores upon which future research may be based.

Study area

BC is geographically extensive and ecologically diverse. Ecosystems range from Pacific coastal temperate rainforests to interior grasslands, alpine tundra, and northern boreal forests. Bears admitted to Project Rewild originated from multiple regions across BC (Figure 1). Three bears originated from northern coastal BC, specifically Kitimat and Kitsault. Two bears originated from north central BC near Vanderhoof. One bear originated from northwestern BC near Houston. Five bears originated from Five bears originated from "Bella Coola on the central coast.". Three bears originated from southeast BC near Meadow Creek and Cranbrook.

The coastal and central coast regions are characterized by Pacific temperate rainforests dominated by western red cedar (Thuja plicata), western hemlock (Tsuga heterophylla), and Sitka spruce (Picea sitchensis), with watersheds that may support up to five species of Pacific salmon (Oncorhynchus spp.). Conversely, interior and northern source locations experience a continental climate; north-central ecosystems consist of sub-boreal spruce forests with lodgepole pine (Pinus contorta), trembling aspen (Populus tremuloides), and white spruce (P. glauca). Northern habitats transition from boreal white and black spruce (P. mariana) to alpine tundra. Southeastern source regions encompass rugged, interior mountain ranges transitioning from low-elevation ponderosa pine (Pinus ponderosa), to mid-elevation Douglas fir (Pseudotsuga menziesii) and western larch (Larix occidentalis), to Engelmann spruce (P. engelmannii) and subalpine fir (Abies lasiocarpa) zones.

Across BC, grizzly bear food resources vary in abundance and distribution by region (MacHutchon, 2021). Seasonal dietary patterns, however, are broadly consistent and reflect local availability. In spring, grizzly bears primarily consume graminoids (Cyperaceae and Poaceae), horsetails (Equisetum spp.), and members of the carrot (Apiaceae) family. Bears also excavate roots and tubers. During summer, diets shift to fruit. Important berry producing shrubs include huckleberries and blueberries (Vaccinium spp.). In drier regions, Canadian buffaloberry (Shepherdia canadensis) is an important food source. Some berry species persist into fall, such as highbush cranberry (Viburnum edule). Bears also resume extensive excavation for roots and tubers, such as glacier lily (Erythronium grandiflorum) and rice root (Fritillaria camschatcensis). Most bears that reside in coastal and southeastern BC also have access to anadromous Pacific salmon (Oncorhynchus spp.), which represent an important seasonal source of protein and lipids for some bears.

Bear intake, holding and rearing

Admission of orphaned cubs to the NLWS facility was determined by the BC Conservation Officer Service (COS), and Regional government biologist based on provincial rehabilitation criteria (FLNRORD, 2016). The NLWS operated under provincial permit #SM21-616820, which authorised the capture, captive rearing, and release of eligible orphaned grizzly bear COY. Under permit conditions, only cubs orphaned during their first year of life were eligible for admission. Trained shelter staff or certified volunteers operating under the shelter permit were authorised to capture and transport cubs to the facility, although in some instances (e.g., bears #F_28, F_29, and M_30) capture was conducted with direct assistance from COS personnel. Cubs could be accepted at any time of year; however, during Project Rewild there were no individuals admitted that required bottle feeding.

Project Rewild operated under a separate provincial research permit (#VI22-689029) issued to evaluate the facility’s captive-rearing methods and assess post-release survival, functioning independently of the facility’s operational permit. Between the fall of 2019-2022, 14 COYs were admitted into NLWS and Project Rewild. Upon intake, cubs underwent a health assessment conducted by shelter staff and, when available, the provincial veterinarian. Following confirmation of acceptable health status and absence of transmissible disease, cubs were placed either directly into the primary grizzly bear enclosure or into a smaller adjoining enclosure to allow acclimation prior to integration. All enclosures were secured with electric fencing.

The main grizzly bear enclosure measured approximately 0.095 ha (≈950 m²; 0.25 acre) and was located ~200 m (Euclidean distance) uphill from the primary human residences and activity areas at the facility. The enclosure was largely spatially separated from areas housing most other rehabilitated species, although it remained in close proximity (~52 m) to the main black bear enclosure. Two adjoining enclosures were attached to the main grizzly bear enclosure one of which occasionally housed black bears without a visual barrier between the species. The grizzly bear enclosure was also located approximately 150 m from the upper moose enclosure.

The enclosure contained a small pool, standing tree trunks, a wooden climbing platform used for environmental enrichment, and small wooden box structures that provided shelter. The smaller adjoining enclosure allowed temporary separation or gradual introduction of different family groups when required. Following acclimation and integration, cubs admitted within a given intake year were reared together in the main enclosure regardless of family, origin or sex.

To minimize habituation to humans, rearing of the grizzly bear cubs was restricted to a single primary caretaker, with a designated secondary handler providing coverage during absences (Beecham, 2006). The facility owners maintained oversight, and public access or viewing was said to be prohibited throughout the rearing process.

Caretakers entered the enclosure for feeding and cleaning. Bears were fed twice daily within the enclosure, typically during morning and evening periods. Food was prepared at a centralized preparation facility located within the human use residential area downslope of the enclosure and transported to the enclosure using an all-terrain vehicle. Diet composition consisted primarily of fruit including apples, grapes and melons, in addition to raw vegetables such as carrots and lettuce. Dry dog food was provided as a dietary supplement (Beecham, 2006). Seasonally available wild bear foods were periodically collected to familiarise the bears with wild foods. Animal protein was offered mainly as salmon and wild game. Live fish were introduced into the enclosure pond in 2022 to promote fishing behaviour development. Enrichment items included rotten logs for insects, piles of rocks with food hidden, suspended animal hides or meat. The primary caretaker entered the enclosure to distribute food during feeding events. Enclosures were cleaned regularly by temporarily securing bears within an adjacent enclosure compartment. Bears could maintain visual contact with caretakers during cleaning procedures.

During Project Rewild, with the exception of two cubs already undergoing rearing at project initiation in 2019, all bears underwent body condition assessment twice by the principal investigator biologist. Assessments occurred at intake approximately two to four weeks following arrival at NLWS and again immediately prior to release between late May and early June of the year following orphaning. Prior to Project Rewild, complete morphometric and body condition datasets were not consistently collected at intake.

Bears were chemically immobilized using zolazepam tiletamine at 2 mg per kg (Zoletil 50 ®, Virbac). Data collected at intake and release included body mass and morphometric measurements comprising body length, girth, neck circumference, and foot pad dimensions. Body condition indices included fat pinch measurements obtained with calipers and body mass index derived from bioelectrical impedance analysis following methods described by Hilderbrand et al. (1998). Resistance values in ohms were measured using a Quantum Legacy II device manufactured by RJL Systems. Measurements were entered into predictive equations developed by Farley and Robbins (1994) to estimate total body water and percent body fat.

At release processing, all bears were fitted with Vetronic Vertex Lite 2D global positioning system collars operating on the Iridium platform and equipped with continuous very high frequency beacon and a mortality sensor. Individuals were also marked with numbered ear tags and implanted with a passive integrated transponder positioned subcutaneously in the dorsal muzzle region. Collars incorporated safety features including canvas spacer rot-off mechanisms, radio controlled drop-off capability, and a programmed absolute release date. All morphometric, mass, and body condition measurements were repeated during release processing.

All bears were released as yearlings during the spring following their orphaning year, with release timing occurring between mid-May and June. Year pairs are reported using a slash notation in which the first year represents intake and the second year represents release (e.g., 2020/21). To account for seasonal variations in grizzly bear behaviour and availability of forage across the broad latitudinal gradient of BC, the active season was partitioned into three biologically distinct periods: spring, summer, and fall. Spring (15 April – 14 July) encompassed the period from moving away from one’s den site, primarily foraging on herbs and bulbs until the onset of fruit ripening. Summer (15 July – 20 September) was defined by the shift from green vegetation to the consistent use of high-energy berries (Vaccinium spp., Shepherdia canadensis) and insects as primary food sources (Ciarniello et al., 2015; McLellan and McLellan, 2015). Fall (21 September – 15 November) represented the post-berry season until the initiation of denning. These dates provide a phenological midpoint between the earlier berry seasons of southern BC (McLellan and McLellan, 2015) and the later cycles of the central interior (Ciarniello et al., 2015), allowing for a standardized comparison for bears originating from BC’s diverse ecoregions.

Mortality signals generated by collars were investigated as soon as feasible following notification. Mortalities were categorized as natural, human-bear conflict, or human related (e.g., vehicle strike). Natural mortality was assigned when evidence indicated causes such as predation. A mortality associated with release was included in the captive growth and development analyses (rearing) but excluded from post-release survival analyses.

Post release success evaluation incorporated assessment of survival and human bear conflict. Survival was examined across three temporal intervals: (1) short-term survival from release to first denning; (2) mid-term survival from den emergence in the second year to approximate dispersal age within the release region; and, (3) long-term survival defined as attainment of breeding age or confirmed reproduction.

Human-bear conflict was defined according to criteria established by the IUCN Human-Bear Conflict Expert Team (HBCET, 2019). Successful outcomes were characterized by absence of undesirable property use or damage, lack of harm to people, and absence of actual or perceived threats to human safety or property. Conflict management followed the Procedures Preventing and Responding to Conflicts with Large Carnivores decision framework developed by the BC Ministry of Forests Lands Natural Resource Operations and Rural Development (FLNRORD, 2016). Project personnel did not influence management actions undertaken by the COS during human-bear conflict events. Detailed incident documentation and, when feasible, interviews with responding officers were conducted. When study animals were destroyed, body condition at death was evaluated where possible.

Statistical analysis

All analyses were conducted at the individual level, with each bear treated as an independent observational unit (the primary unit of replication). Given the small cohort size of orphaned grizzly bears that were captive-reared (n = 14) and released (n = 13) with known fates (n = 12), the primary findings are reported as single-variable comparisons to preserve statistical power. All statistical tests were conducted using R (version 4.3.2) and Microsoft Excel, with the significance threshold set at α ≤ 0.05.

Captive rearing and growth metrics

Captive rearing duration was defined as the interval, in days, between facility intake and release. To evaluate differences in rearing duration among release years, a one-way analysis of variance (ANOVA) was used. Because release timing was consistently scheduled within a specific late-spring window (mid-May to mid-June) across all study years, variation in total rearing duration was primarily driven by, and interpreted relative to, intake timing.

Absolute weight gain was calculated as the difference between release mass and intake mass in kilograms. Individuals were categorized by intake season based on admission date. Differences in weight gain between intake seasons, as well as sex specific differences in release mass, total weight gain, and morphometric variables (including percent body fat at intake and release), were assessed using two sample t-tests assuming equal variances for within Rewild bear comparisons. Additional exploratory comparisons were conducted to assess weight gain patterns among individuals housed concurrently within release years. These descriptive analyses were intended to identify potential cohort or social context effects on growth. Categorical associations between sex and survival outcomes were assessed using Fisher’s Exact Test. Univariable logistic regression was used to examine the influence of captive-rearing duration and release mass on the probability of human-bear conflict.

Comparative body mass and morphological analysis

Morphometric data for wild yearling grizzly bears (n = 63) were obtained directly from primary investigators and provincial datasets across BC and the transborder regions of the United States. Data for the Flathead Valley and Westslopes of BC (n = 35 of 63) were provided by the BC Provincial Government (G. Mowat), data for the transborder Cabinet-Yaak and Selkirk Mountains of BC, Montana, Idaho, and Washington (n = 16) were contributed by U.S. Fish and Wildlife Service (W. Kasworm and M. Proctor). Additional BC regional datasets included eight bears from the Parsnip River, BC (contributed by the author, L. Ciarniello), and two bears each from the Elk Valley, BC (C. Lamb and L. Smit), and Pemberton Meadows, BC (N. Bickerton). These reference data represent wild, free-ranging yearling grizzly bears captured during the course of long-term population monitoring, demographic research, and university-led field studies conducted by provincial, federal, and academic investigators. To ensure statistical independence, only the initial capture event was retained for individuals captured multiple times within a single season. Filtering was unnecessary for rewilded bears as they were only processed once prior to release.

Mean spring body mass was compared between wild and rewild cohorts using a Welch two-sample t-test to account for unequal variances. Logistic regression was used to examine the relationship between release mass and the probability of one-year post-release survival. The logit function was applied as follows:

Where P is the probability of survival, β 0 represents the intercept, and β 1 is the coefficient associated with release mass.

Morphological differences were assessed by comparing wild yearling body lengths to rewild yearling release lengths again using a Welch’s t-tests. To account for temporal variation, morphological data were examined across spring, summer, and fall seasons, with the majority of sampling occurring within the spring season to match Project Rewild release dates. All comparative analyses were performed with a significance threshold of α ≤ 0.05.

Rewild bear survival analysis

Post-release survival was observed for the 12 rewild bears with known fates. Observed post-release duration among sexes and groups was compared using a Mann-Whitney U test. For survival analysis Mann-Whitney U test ranked survival durations without incorporating censoring structure. To account for censored observations and estimate time dependent survival probabilities, Kaplan-Meier survival analysis was performed, with time measured in days from release to mortality or censoring. Kaplan–Meier methods were applied because the study involved a small (n = 13 including censored bear), intensively monitored cohort with staggered entry and release dates, a design commonly encountered in reintroduction and rehabilitation studies where sample sizes constrain model complexity and fate determination is high (Beringer et al., 2004). Survival estimates were therefore interpreted descriptively and in conjunction with documented end fates. Time to event was defined as the number of days from release to confirmed mortality (n = 10). Individuals that were lost due to presumed collar failure (n = 1) and those that remained alive at the end of the monitoring period (n = 2) were treated as right-censored using the last confirmed date alive.

Daily survival rates, S ^ t, for radiocollared grizzly bears were estimated using a staggered entry Kaplan-Meier design following Pollock et al. (1989):

where n refers to the number of times bears were checked, d i is the number of deaths, and r i is the number of bears at risk recorded at time i. Annual survival was calculated by taking the product of daily survival for all 365 days. The variance for survival rates was estimated using Greenwood (1926). The GPS mortality collar message was used as the date when bears died or for “problem” wildlife kills the known mortality date. Contact was lost with one bear due to what was thought to the be failure of the GPS transmitter. This record was censored at the last known location on the date when the bear went off-line. The calculated mortality rate might be conservative because it is possible that the missing bear was deceased. Sex specific differences in post release survival were evaluated by comparing Kaplan-Meier survival curves using a log rank test. Temporal patterns of mortality were additionally assessed descriptively by examining clustering and dispersion of survival times within sex classes.

Bear intake and holding

From 2019 to 2022, 14 wild-born orphaned grizzly bears entered Project Rewild (7 males, 7 females). One female from a two-cub family unit died during helicopter longline transport to the release site. To facilitate longline transport, both siblings were immobilized and secured back-to-back on a single net-enclosed pallet; however, a mid-flight banking maneuver caused a shift in the longline load. This shift resulted in bear F_27 being wedged between the net, pallet, and her sister, which ultimately resulted in severe, fatal neck compression and asphyxiation. Because this mortality occurred before release into the wild, bear F_27 lacked post-release telemetry data and was excluded from subsequent post-release survival analyses. As a result, 14 bears were captive reared, and 13 were successfully released into the wild.

The 14 captive-reared bears represented five family units and three lone individuals (Table 1). In 2019/20, two male siblings were housed and reared together. In 2020/21, five bears were housed together, representing two sibling family units: one pair of females and one group of two females and one male. In 2021/22, five bears were housed together, representing three family units: two sets of male siblings and one lone female. In 2022/23, two female bears from different family units were housed together.

Season of intake varied among years according to when bear(s) were orphaned. Overall, six bears entered the facility during the spring season and eight bears entered during the fall, with intake patterns varying chronologically based on the orphaning event (Table 1). In 2019/20, the two male siblings were admitted in spring. In 2020/21, both sibling family units were admitted in fall. In 2021/22, one pair of male siblings was admitted in spring, while a second pair of male siblings and one lone female were admitted in fall. In 2022/23, two lone females from different family units were admitted in spring.

Of the eight distinct intake events (representing five family units and three singletons), the mother’s fate was confirmed in four cases: two mothers were shot as “problem” bears by the COS (F_26 & F_27; F_36), one was shot by a landowner on the ranch (M_31 & M_32), and one mother died in an accident during a human-provoked conflict (F_28, F_29 & M_30). In the remaining four cases, the mother’s fate was unknown; these cubs were admitted after being found wandering alone with no maternal female present in the area (M23 & M_24; M_33 & M_34; F_35; F_37). The two individuals that survived to the end of the monitoring period originated from the case where maternal mortality was attributable to a provoked human-bear conflict.

Captive rearing

Bears were captive-reared for an average of 278 days (n = 14, SE = 18, SD = 63, range = 198–373). Duration of captivity differed among release years (ANOVA F 3, 10 = 15.68, p = 0.0004). Although release timing was consistent (late May to mid-June), intake dates varied because bears entered the facility at the time of orphaning. Consequently, interannual variability in days held primarily reflect differences in intake timing.

Release weight and gain

On average, bears gained 81.0 kg (178.3 lbs; SE = 8.7) during captive rearing (n = 12). Bears admitted to the facility in spring (n = 4) gained significantly more weight (115.0 ± 18.6 kg) than those admitted in the fall (n = 8; 64.0 ± 16.9; t 10 = -4.77, p = 0.00075). Mean release weights for yearling females across all years were 116.6 kg (256.5 lbs, n = 7), while males averaged 138.3 kg (304.2 lbs, n = 7), representing a 21.7 kg (~48 lbs) difference between the sexes.

In 2019/20, when the annual cohort consisted of only two bears, two male siblings were housed together and exhibited similar release weights (142.8 kg vs. 143.0 kg). For 2020/21 and 2021/22, representing years when larger cohorts of males and females were housed together concurrently, males gained significantly more weight than females despite free feeding (t 8 = -3.73, p = 0.0058). In 2020/21, the single male had a greater release mass than all four females housed concurrently. A similar pattern was observed in 2021/22, when four males and one female were housed together, with the female again exhibiting the lowest release weight. During this period, males gained an average of 98.7 kg (217.2 lbs), whereas females gained 68.4 kg (150.5 lbs).

In 2022/23, two females were housed together without males, and the difference in weight gain between sexes was no longer significant at release (t 10 = -1.92, p = 0.084). When only females were housed together, release weights diverged substantially, with a difference of 32.4 kg (71.2 lbs) between the two females. Inclusion of the 2023 releases increased mean female weight gain by 14.9 kg (32.8 lbs), driven by the release of the largest yearling female in the study (146 kg, 321.5 lbs). Despite this, overall males (n = 7) remained significantly heavier than females (n = 7) at release (t 12 = 2.74, p = 0.018; Figure 2).

At intake, body length was similar between male and female Project Rewild grizzly bear cubs (males: 106 ± 5.9 cm, range = 90–127 cm, n = 5; females: 114 ± 6.8 cm, range = 89–134 cm, n = 7). This lack of initial sexual dimorphism was driven by a higher proportion of females being admitted during the fall season when cubs were older and larger. Overall, body length at intake differed significantly by season (t 10 = 4.28, p = 0.002); cubs admitted in the fall (n = 8, 5 females:3 males) were longer (119.7 ± 11.1 cm) than cubs admitted in the spring (n = 4, 2 females:2 males; 93.6 ± 6.7 cm). At release, males were significantly longer than females (males: 165 ± 5.5 cm, range = 143–181 cm, n = 7; females: 151 ± 2.6 cm, range = 142–159 cm, n = 7; t 12 = 2.25, p = 0.04).

After feeding throughout the winter and not hibernating, males were heavier, longer, and had greater chest girth than females. Percent body fat at release did not differ between sexes (t 12 = 0.061, p = 0.95, n = 7 per sex). Mean percent body fat was similar for males (26.8 ± 2.1%) and females (27.0 ± 1.8%). Percent body fat ranged from 17.9–32.7% in males and 21.1–32.2% in females.

Post release survival: observed mortality rates and causes

Of the 13 orphaned grizzly bears released as yearlings (2020–2023), 12 had known outcomes and one was censored due to collar failure. Survival time post-release ranged from 7 to >1, 168 days (3.2 years; n = 12, x ¯ = 288.7 ± SD = 413.5 days, median = 107 days). Confirmed mortalities accounted for 83% of the cohort (n = 10), with mortality heavily concentrated in Stage 1 (short-term survival; release to first denning). During this initial stage, nine bears (75% of the known releases) died, representing 90% of all confirmed mortalities recorded during the study being in Stage 1.

Three bears (25%), all of whom were siblings from a social captive-rearing group of five, successfully reached Stage 2 (mid-term survival; den emergence to dispersal age). However, one female (F_28) from this sibling group was killed via intraspecific predation shortly after den emergence in early spring, leaving two siblings (one male, one female) to progress toward Stage 3 (long-term survival; breeding age).

Table 2 summarizes known survival outcomes relative to release mass. Causes of mortality included human-bear conflict (5 of 10, 50%), with four of those bears lethally removed by the BC COS after being classified as “problem” wildlife (F_26, M_33, M_34, F_36). One mortality (F_35) was classified as a human-bear conflict but occurred in a remote alpine zone (1, 730 m) within 24 hours of her encountering a pop-up mining operation with two workers, rather than through attraction to human settlements. Two (20%) mortalities were due to illegal killing and found only due to their tracking devices (M_32, F_37), and three (30%) were consistent with intraspecific predation (M_23, F_28, M_31). Intraspecific predation was the documented cause of death for one of the bears with the lowest release mass and for the bear with the highest release mass. Natural intraspecific mortalities were evenly distributed across spring, summer, and fall (33% each), whereas human-bear conflict mortalities occurred primarily in fall (57%), with 29% in summer and 14% in spring. The two surviving (17%) bears were siblings last monitored three years post-release and confirmed alive and remaining together through camera trap imagery at 4 years old (F_29 and M_30). Survival did not differ significantly by sex (Fisher’s Exact Test, p = 1.0), likely because one male and one female survived.

Duration of post-release survival of known fate individuals did not differ between sexes (n = 12, Mann–Whitney U = 17.5, p = 1.00). However, post-release survival time showed contrasting distributional patterns between sexes (Figure 3A). Male bears exhibited consistently short survival times following release, with low variability and most mortalities occurring within the first year. Female bears showed greater variability in survival time, including both early mortalities and the longest observed post-release survival durations. The two bears still alive at the end of the monitoring period, one male and one female from the same family unit, appear as high-value outliers in Figure 3A.

Bears admitted to the facility in fall (n = 7) survived significantly longer post-release than those admitted in spring (n = 5, Mann–Whitney U = 28.5, p = 0.044; Figure 3B). All spring-admitted bears died during the monitoring period. Survival times for fall-admitted bears were more variable, whereas spring-admitted bears exhibited consistently short post-release survival durations. The two sibling bears surviving to the end of monitoring were held for 226 days, representing one of the shorter holding durations observed. Bears held for the longest durations in captivity did not survive post-release.

Post-release survival time also varied with sibling group size. Bears reared in smaller sibling groups (n = 2 years of 2 individuals/year) exhibited uniformly short survival times following release, and no individuals from this group survived to the end of the monitoring period (Figure 3C). In contrast, bears reared in larger sibling groups (n = 2 years of 5 individuals/year) showed greater variability in survival time and included the only individuals surviving at the end of monitoring.

In relation to human-bear conflict, univariable logistic regression indicated no statistically significant relationship between duration in captivity and the probability of mortality due to human-bear conflict (β = 0.0035, SE = 0.0095, p = 0.714; Odds Ratio = 1.004, 95% CI = 0.985-1.022). Release mass also showed no statistically significant relationship with human-bear conflict mortality (β = −0.0254, SE = 0.0342, p = 0.458; Odds Ratio = 0.975, 95% CI = 0.912-1.042). In an exploratory multivariable model including both duration in captivity and release mass, coefficient estimates were consistent with the univariable results and remained statistically non-significant.

Four bears (2F:2M) had post-mortem weight taken. Two male sibling bears with access to high-quality forage resources, specifically spawning salmon at a weir, experienced an average weight loss of 0.25 kg per day (0.55 lbs/day over 107 days; M_33, M_34). In contrast, two female bears both released in areas without salmon access, lost an average of 0.48 kg per day (1.05 lbs/day, range: 1.0−1.1 lbs/day; F_35, F_36) over 59 and 92 days, respectively. These data indicate that even with abundant food sources, Project Rewild bears may still face significant energetic challenges post-release.

Comparative morphometric analysis of wild and Project Rewild yearlings

Morphometric data were collected for wild yearling grizzly bears (n = 63), including body mass (n = 61), total body length (n = 41), chest girth (n = 42), and neck circumference (n = 36). Spring body mass of captive-reared rewild yearling bears was substantially greater than that of wild yearling bears (t = -16.33, df = 22, p < 0.0001), with Project Rewild bears (x ¯ ± SD = 127.4 ± 18.1 kg, n = 14) weighing 3.5 times as much as wild-reared yearlings (x ¯ ± SD = 36.1 ± 13.5 kg, n = 23). This represents a 91.33 kg difference in spring body mass (SE = 5.59 kg; 95% CI 79.72-102.94 kg) between cohorts.

Sex-specific comparisons showed the same general pattern, with captive-reared bears substantially heavier than wild bears within both sexes (Figure 4). Project Rewild males exhibited the greatest body masses overall, while wild males and females showed broadly overlapping weight distributions with each other but not Project Rewild bears. Despite some variation within groups, the body mass distributions of Project Rewild bears remained well above those of wild yearlings.

Body mass of wild bears exhibited a steady increase from spring (x ¯ = 36.08 kg, n = 23), through summer (x ¯ = 53.71 kg, n = 22), to fall (x ¯ = 74.66 kg, n = 16)(Figure 5). This pattern reflects the expected seasonal weight gain associated with increasing food availability and pre-hibernation fat accumulation. Regardless, wild bears rarely match Project Rewild spring release size, even in fall. Captive-reared bears released during spring and early summer remained significantly heavier than most wild bears at comparable seasonal stages. Only the largest fall wild individuals approach the lower range of Project Rewild spring weights, but the distributions remained largely non-overlapping (Figure 5).

Comparison of yearling body length reveals a consistent developmental disparity between captive-reared and wild bears. Comparing spring measurements, wild yearlings (n = 16, x ¯ = 122.27 cm) were significantly shorter in body length than Project Rewild yearlings at the time of spring release (n = 14, x ¯ = 158.26 cm; t 26 = -8.03, p < 0.0001) regardless of sex. This disparity persisted when comparing all seasonal lengths for wild yearlings (n = 41, x ¯ = 133.80 cm) against Project Rewild spring release lengths (t 29 = -5.65, p < 0.0001).

Chest girth and neck circumference were also compared between wild and Project Rewild yearlings. Comparing spring measurements, wild yearlings were significantly smaller in chest girth (n = 16, x ¯ = 65.0 cm) and neck circumference (n = 13, x ¯ = 39.7 cm) than Project Rewild yearlings at the time of spring release (n = 14, x ¯ chest = 106.5 cm; t 28 = -17.72, p < 0.0001; x ¯ neck = 62.9 cm; t 25 = -12.84, p < 0.0001). This disparity also persisted when comparing chest girth (n = 42, x ¯ = 76.5 cm) and neck circumference (n = 35, x ¯ = 45.7 cm) for all wild yearlings across all seasons against the Project Rewild spring releases (n = 14, x ¯ chest = 107.1 cm; t 51 = -11.25, p < 0.0001; x ¯ neck = 62.9 cm; t 37 = -9.19, p < 0.0001). These results confirm that the excessive growth observed in captive-reared bears is consistent across both mass and skeletal development metrics.

Relationship between release weight and survival outcomes

Stage 1 post-release survival exhibited substantial variability across the range of release weights (Figure 6). Although greater release mass was initially hypothesized to improve post-release success by providing increased energetic reserves, and deter intraspecific predation larger release mass did not improve survival. Survival probability generally declined with increasing release weight across the observed range.

Although mortality occurred across much of the observed release-weight distribution (105.1–167.8 kg), all individuals released above approximately 135 kg died during this initial stage. In contrast, the three individuals that survived beyond one year were released within a lower to intermediate weight range. This included one 105.1 kg individual (F_28) who successfully achieved short-term survival through her first denning period (Stage 1 success) and entered the mid-term survival phase upon den emergence in her second year (Stage 2) before succumbing to intraspecific killing at 331 days post-release.

Kaplan-Meier survival analysis

Kaplan–Meier analysis estimated post-release survival at 0.25 (SE = 0.12; 95% CI = 0.01–0.49) over the monitoring period, indicating low survival with substantial uncertainty due to the small sample size and high mortality. This estimate likely represents a minimum, as the fate of one bear lost to follow-up remains unknown and could not be confirmed (i.e., censored). The observed post-release survival based on known outcomes (17%) is slightly lower than the Kaplan–Meier estimate but within the 95% confidence interval.

Low post-release survival for rewilded grizzly bears was found for both sexes, with most mortality occurring within the first year following release. Survival curves suggested sex-specific patterns in the timing and variability of mortality, with males exhibiting clustered early deaths and females showing greater dispersion in survival times; however, a log-rank test indicated no significant difference in post-release survival between male and female bears (χ² = 0.96, df = 1, p = 0.33), possibly due to the small sample size. Two individuals (one male and one female from the same family unit) remained alive at the end of the monitoring period, contributing to extended survival tails in both groups.

The low post-release survival observed among captive-reared grizzly bears in this study highlights substantial challenges associated with current grizzly bear rehabilitation outcomes in BC. I suggest that post-release failure to survive is not driven by a single factor, but by a causal chain: intake season dictates duration of captivity, which in turn dictates the cumulative volume of human exposure. My results indicate that temporal factors may be important drivers of reduced survival, with younger age of intake and subsequent prolonged rearing duration significantly reducing the probability of survival. Bears admitted in the fall experienced shorter durations in care and benefited from an extended period of wild maternal learning which is central to the development of non-conflict behavioural traits (Morehouse et al., 2016). Conversely spring-admitted cubs lacked this maternal foundation and were subjected to prolonged captive-rearing, which likely facilitated habituation to humans and may have reduced the development of behavioural wariness. While the relationship between days in captivity and conflict mortality was not statistically significant, uncertainty was high due to limited sample size, as is common in in studies of captive-rearing or small, threatened wildlife populations. In such cases, descriptive patterns may still be useful for generating biologically relevant hypotheses and should be interpreted alongside biological understanding rather than significance tests alone (Bissonette, 1999). Biologically, even the shortest captive duration was substantial (~200 days), which may indicate that habituation effects reach a plateau after an initial seasonal threshold rather than progressing linearly, although this cannot be confirmed with the present data. Similarly, release mass was not a statistically significant predictor of conflict mortality, although the observed negative trend was influenced by the two heaviest males in the cohort (M_31 and M_23), both of which died from intraspecific predation rather than human-bear conflict. Furthermore, although the sample size is limited, the social environment appeared to modulate outcomes; bears reared in larger sibling groups exhibited greater survival variability and included the only successful recruits, suggesting that increased intraspecific interaction may partially buffer against the detrimental effects of the human-mediated rearing environment.

In BC, the lack of provincial husbandry standards for grizzly bear captive rearing meant that enclosure design, location, and feeding protocols were left to facility discretion, rather than a species-specific biological framework. In the absence of a specialized provincial mandate, these practices defaulted to conventional rehabilitation models that prioritize operational feasibility over the preservation of behavioural wildness. While these protocols may be sufficient for general wildlife or the care of black bears, my results suggest they are inadequate for the specialized requirements of grizzly bear rehabilitation. This underscores the need for species-specific captive rearing protocols grounded in behavioural ecology and evolution; black bears are highly adaptable and often persist near human development, whereas grizzly bears generally exhibit lower tolerance of human activity and a greater propensity for defensive responses. Consequently, the public tolerance for grizzly bears in human-dominated landscapes is substantially lower. This highlights a significant policy gap: without rigorous, government-led technical specifications, even well-intentioned efforts may unintentionally produce rearing environments that facilitate habituation and food conditioning rather than successful wild transition.

Current best practices for captive rearing bears suggest that habituation can be mitigated by limiting the number of handlers to no more than two people (Beecham et al., 2016) and by releasing bears in remote areas to avoid humans for the first two weeks post-release (Beecham, 2006; WRNBC, 2012). At the time of their inception, these guidelines represented the gold standard for ursid rehabilitation (Beecham, 2006; Beecham et al., 2016), particularly for black bears where survival rates are reported to be high (0.50–0.90; Beecham et al., 2015; Blair et al., 2020). While these foundational protocols have long guided the field, contemporary results suggest that generalized bear rehabilitation guidelines may not adequately account for species-specific differences in behaviour, risk tolerance, and responses to human-associated cues, particularly for grizzly bears. Even with minimal handlers, as stated by the rearing facility, bears in this study remained extremely habituated and food conditioned post-release (as further detailed via post-release behavioural case studies below). This suggests that for grizzly bears, the environmental conditioning of the rearing facility may have a greater impact on habituation than simply limiting the number of handlers. The extended residency for early admitted bears increased their direct and daily contact with the caretaker for their twice daily feeding as well as the frequency of auditory and visual triggers, associated with the husbandry activities of neighboring species (moose and black bears). These environmental cues appeared to condition the bears to associate human presence and mechanical sounds with food security.

The failure of the two-week avoidance of humans’ post-release theory (Beecham, 2006; WRNBC, 2012) is most acutely illustrated by the post-release behaviour of female Bear F_35. Despite an initial 59-day period in a remote alpine zone without human contact, the bear exhibited immediate and persistent food-seeking behaviour upon her first encounter with a temporary mining operation at 1, 730 m. F_35 was noted to persistently approach the two employees, ultimately obtaining their lunches. Critically, F_35’s habituation was not mitigated by the significant auditory and visual disturbance of helicopter activity, nor by the overnight absence of human personnel. The bear remained at the site overnight, approached the helicopter upon arrival, and continued to seek anthropogenic food sources from employees upon their return the following morning. F_35 was killed less than 24 hours after encountering the site for persistent, food-seeking behaviour. This case demonstrates that habituation in captive-reared grizzly bears is not a transient state that diminishes after a few weeks of “wild” living; rather, it represents a fundamental, behavioural shift that persists long after release and cessation of human contact. The physiological toll of this transition is evidenced by F_35’s post-mortem data, which revealed a substantial weight loss of 29.04 kg (64.03 lbs) during her time in the wild. This represents a loss of 0.50 kg per day (1.1 lbs/day), further illustrating the severe energetic deficit faced by over-conditioned yearlings.

Importantly, not all failures of the two-week hypothesis are expressed through immediate, high-intensity habituation and food-conditioned behaviour once encountering a human. In some cases, bears may appear to function normally for extended periods moving through townships, again with conflict behaviour emerging only after an opportunistic encounter with an unsecured human-provided food source (e.g., pet food, livestock feed, or garbage). These delayed conflict trajectories are particularly important from a management perspective because they may be misinterpreted as evidence of successful reintegration, despite the persistence of underlying food-conditioning and tolerance of human-associated environments.

Siblings M_33 and M_34 were destroyed 107 days after release (released 6 June 2022; returned 5 July; destroyed 21 September 2022), further illustrating how this underlying habituation persists as a non-transient state well beyond the two-week post-release risk period. Both bears returned to the community where they were orphaned approximately one-month post-release, they were frequently reported near residences foraging on berries and fishing in a productive spawning channel. During this period, they were described as behaving “respectfully” and typically left the area when directed. However, on September 17 th a bag of dog food was left visible and accessible inside a vehicle and the bears were reported to have damaged the vehicle to access the food reward 103 days post-release. Following this event, the bears returned to the property on 18 th September and were observed in an open barn during daylight hours feeding on unsecured grain, suggesting reinforcement learning and an increasing willingness to exploit accessible human structures. While a single unsecured attractant can draw in any bear, the critical distinction here is the highly accelerated escalation trajectory that followed. When a wild-reared bear encounters an opportunistic food reward, its innate, maternally reinforced wariness likely moderates its subsequent actions, with serious conflict behaviours generally emerging through repeated reinforcement of anthropogenic food rewards and learned reductions in human avoidance rather than immediately following a single feeding event (Herrero 2002; Gunther et al., 2004; Morehouse et al., 2016).This sequence is notable given that high-quality, natural foods were readily available and dog food was fed as a supplementary food source during captive-rearing; prior exposure may have contributed to attraction toward a high-reward anthropogenic item post-release. Further, post-mortem data revealed that these two bears experienced an average weight loss of 0.25 kg per day (0.55 lbs/day) over their 107 days post-release despite being regularly sighted fishing at the spawning channel and feeding on berries. This event indicates that even with abundant food sources, Project Rewild bears may have still faced significant energetic challenges post-release. Taken together, these human-bear conflict events indicate that even months after release, a single unsecured attractant can precipitate repeated site use and rapid escalation to lethal management outcomes.

Physiological barrier: a potential metabolic trap

A central tenet of bear rehabilitation has been that maximizing release weight reduces predation risk. This is supported by guidelines suggesting that a target release weight of approximately 1.5 times the weight of wild-reared cohorts increases post-release survival (Beecham et al., 2016). However, my results suggest that current rearing may produce bears of such large size that survival is compromised rather than improved. Hilderbrand et al. (1999) demonstrated that predictable, high-quality nutrition during early development promotes accelerated skeletal growth. This pattern was reflected in the present study, where Project Rewild yearlings were significantly larger than wild yearlings across every metric, including body length (160 cm vs. 123.8 cm), chest girth, neck circumference (all p < 0.0001), and body mass.

This disparity suggests that captive-reared bears were released as developmental outliers, exhibiting significantly larger skeletal dimensions and approximately 3.5-fold greater body mass than wild yearlings. Regardless of whether this disparity reflects enhanced winter nutrition, prolonged access to supplemental feeding, or the cumulative effects of captive rearing, bears were released as developmental outliers relative to wild-reared yearlings. Although this far exceeds the 1.5x weight factor recommended for ursid rehabilitation (Beecham et al., 2016), such excessive growth may be counter-productive for grizzly bears. In contrast to a wild yearling that enters spring lean and adapted to a low-calorie landscape, a Project Rewild bear may have been released with a massive skeletal frame and elevated energetic requirements. This could potentially create a metabolic mismatch similar to that observed in some of the Yellowstone grizzly bears following the closure of open-pit garbage dumps (Robbins et al., 2004).

Robbins et al. (2004) noted that bears conditioned on predictable, high-calorie garbage dumps grew to excessive sizes that could not be sustained by wild forage alone once those subsidies were removed. In that system, only the smaller bears and subadults, whose caloric demands fell within a more natural energetic range, were able to transition successfully to wild resources. My data suggest a potentially similar mechanism: the Project Rewild yearlings in this study (averaging 116.6 kg for females and 138.3 kg for males) were structurally adapted to a high calorie nutritional environment that does not exist for naïve yearlings released into spring landscapes.

When released into unfamiliar environments without maternal guidance to locate high-density seasonal food resources, these over-conditioned bears may experience immediate energetic deficits. The substantial post-release weight loss documented in the recovered mortalities further supports the possibility that these bears were unable to meet energetic demands under natural foraging conditions. This energetic mismatch may also contribute to increased attraction toward anthropogenic food sources and subsequent human-bear conflict. Although higher mass was not a statistically significant predictor of conflict in this study, which may due to limited sample size, larger bears also failed to recruit into the wild population, and two of the three largest bears released (>140 kg) succumbed to intraspecific killing.

While the present study was not designed to directly test energetic physiology, the observed relationship between excessive release mass and poor post-release outcomes is consistent with the hypothesis that over-conditioned captive-reared bears may experience an ecologically maladaptive energetic burden following release. Additional studies incorporating larger sample sizes, direct measures of energetics, and post-release behavioural assessments are recommended to further evaluate this proposed metabolic trap mechanism more rigorously.

Management implications

If captive rearing produces human-habituated, over-conditioned bears that are socially and/or physically ill-equipped to navigate wild landscapes, then rehabilitation remains a form of sympathetic management rather than a functional conservation tool. To transition toward the latter, captive rearing protocols must prioritize minimal human contact and wild-type metabolic trajectories to ensure both behavioural wariness and social fitness of released bears.

Government oversight

Rehabilitating bears would benefit from a formalised regulatory framework and enhanced government oversight rather than leaving these decisions up to the facility. Implementing standardised, policy-driven protocols for housing and nutrition is essential to mitigate habituation to humans and food conditioning post-release. In BC, strengthening the coordination between the Conservation Officer Service, resource stewardship branches, and regional government biologists would help to ensure that rehabilitation efforts consistently meet scientific and animal husbandry standards. Furthermore, integrating independent scientific oversight to support facility operations would provide the technical expertise necessary for robust data collection and adaptive program evaluation.

Standardized “hands-off” rearing protocols

The foundation for post-release survival begins at the rearing facility, where bears are first exposed to human-associated stimuli during captive-rearing. Enclosures should be geographically and acoustically isolated from other rehabilitating species and facility operations to reduce conditioning to routine husbandry sounds and human activity. Enclosures should be sufficiently large and naturalized to preclude the need for cleaning or other frequent human entry. Manual entry into enclosures for feeding should be strictly avoided; instead, food should be delivered using remote systems (e.g., overhead cables, gravity chutes, sling shots, or timed drop mechanisms) to prevent the development of associative learning between humans and food provisioning. Human presence should be limited to essential husbandry and veterinary interventions, with behavioural observation and facility monitoring conducted primarily through remote surveillance technologies (e.g., CCTV, drones, and motion-activated cameras) to minimize disturbance and reduce the risk of habituation.

Evidence-based growth benchmarks

Facilities should move away from intensive maximum-growth feeding models and instead apply evidence-based developmental targets. Diets should be adjusted to achieve body mass ranges consistent with wild yearlings at pre-hibernation mass, rather than maximizing size. Aligning captive growth trajectories with natural physiological benchmarks is expected to improve post-release foraging performance and survival. Ultimately, the conservation value of rehabilitation will depend on standardized, measurable rearing and release criteria that prioritize the development of the bears’ behavioural competence after release.