Grizzly bears (Ursus arctos) are an icon of the North American wilderness and a symbol of both conservation and conflict (Proctor et al., 2018; Hughes and Nielsen, 2019; Hughes et al., 2020b). However, as with other large carnivores, human-caused mortality, including illegal killing and vehicle collisions, as well as implications from habitat alteration are the primary threats to grizzly bears (Alberta Sustainable Resource Development., 2008; Nielsen et al., 2009). While scientific research and applied management is ongoing across much of North America, data deficiencies persist in grizzly bears' more northerly range, including those in the boreal northwest region of Alberta, Canada.

Grizzly bears in Alberta were listed as threatened in 2010 due to their small population size, slow reproductive rate, limited immigration from other populations, and increased habitat alteration (Alberta Sustainable Resource Development., 2008). A recovery plan identified seven demographically-separate bear management areas (BMA) with recovery objectives including the necessity for population estimates to be conducted for each BMA, and addressing human-caused bear mortality through access and attractant management along with educational outreach (Alberta Sustainable Resource Development., 2008). From 2004 through 2010, a series of DNA-based population inventories were completed across Alberta, including testing non-invasive genetic techniques and modeling estimates by Foothills Research Institute in the boreal northwest (Alberta Sustainable Resource Development., 2008; Festa-Bianchet, 2010). Based on these data, alongside habitat modeling and expert opinion, Alberta's estimated grizzly bear population was 691, plus additional bears in portions of Banff and Jasper National Parks (Festa-Bianchet, 2010).

Estimating grizzly bear density in the northwest population unit called Bear Management Area 1 (BMA 1) has remained particularly challenging, given the large and relatively remote geographic area and wetland conditions of this landscape. This is contrasted with other BMAs, which are largely comprised of the Rocky Mountain and Foothills natural regions and increased human density, resulting in more road or trail access into grizzly bear habitat (Alberta Environment Parks, 2020). While problematic for human-caused mortality and habitat fragmentation, this increased linear footprint in other BMAs generally reduces the costs associated with efficiently inventorying bear populations.

Another challenge in BMA 1 are the differing perspectives and experiences people have across the region concerning grizzly bears, which can hinder effective conservation and management efforts. Grizzly bears are a charismatic species valued for their aesthetics as well as ecosystem function, but also a species that poses serious human safety risk and economic costs to peoples' livelihoods (Morehouse and Boyce, 2017; Proctor et al., 2018; Hughes and Nielsen, 2019; Hughes et al., 2020a; Morehouse et al., 2020). People across BMA 1 hold values and cultural identity linked to the concept of “frontiersmen,” with their ancestors being hardy pioneers of this harsh boreal landscape (Hughes and Nielsen, 2019). Human-bear relationships are viewed and experienced from the perspective of subsistence lifestyles, generating income, and ensuring human safety, and today this still resonates with many people who call the northwest home (Hughes and Nielsen, 2019). That said, the cultural identity, values and practices in the northwest has conflicted with the provincial government's grizzly bear recovery policy, including how provincial direction governed industrial-scale petroleum and forestry production (Hughes and Nielsen, 2019; Alberta Environment Parks, 2020). Tension between local government and stakeholders across BMA 1 has persisted since the late 1990's (Fullerton, pers. comms). This may be related to a lack of trust in grizzly bear science and scientists, or inaccessibility of scientific information and lack of layperson understanding, as well as local perspectives that problem bears were simply “dumped” (i.e., re- or translocated) into BMA 1 thus contributing to human-bear conflict (Hughes and Nielsen, 2019). During this time public reporting of human-grizzly bear interactions was limited, and a “shoot, shovel, and shut up” sentiment was commonly expressed to occur across rural communities in the northwest (Hughes et al., 2020a).

To address challenges associated with a population inventory of BMA 1 grizzly bears and improve local relationships to enable progress on grizzly bear management, we implemented a collaborative approach that engaged representative stakeholders from across this multi-use working landscape (Wondolleck and Yaffee, 2000; Wilson et al., 2017; Hughes et al., 2020a). Collaborative approaches have been used around the world, including to address bear hunting in Romania (Hartel et al., 2019), conflicts with gray wolves in Montana (Wilson et al., 2017), lion conservation in Zimbabwe (Sibanda et al., 2020), and human-bear coexistence in southwestern Alberta (Morehouse et al., 2020). Collaborative approaches are considered effective at bringing different people across multiple disciplines, perspectives, and experiences together to identify and achieve defined outcomes (Yang, 2017; Hartel et al., 2019; Hughes et al., 2020a). Additionally, collaborative approaches enable participants to decentralize decision-making and share power, foster fairness, and improve credibility and trust in project or policy processes (Singleton, 1998; Wondolleck and Yaffee, 2000; Carlsson and Berkes, 2005; Mattson et al., 2006; Clement et al., 2020). Moreover, collaboration can help groups access and leverage shared resources and funding opportunities, as well as foster co-learning amongst participants.

Our paper weaves together our collaborative approach with a multi-stakeholder team to help identify and address data gaps for BMA 1 grizzly bears, the results from a qualitative evaluation of these collaborative efforts, and the results of the grizzly bear population inventory. We also provide considerations for implementing collaborative approaches elsewhere, to help address potentially contentious contexts for the conservation and management of wildlife species.

BMA 1 is within a multi-use landscape in the boreal forest of northwest Alberta, adjacent to historical grizzly bear habitat in British Columbia, and covers ~41,000 km² (Poole et al., 2001; Alberta Environment Parks, 2020). It is comprised of boreal and mixed-wood natural regions with extensive wetland complexes, with only 2% of the land base protected under provincial park designation limiting motorized (vehicle) access (Alberta Environment Parks, 2020; Figure 1). Human use across the area includes a history of extensive petroleum developments (i.e., well-sites and pipelines), forestry harvest, electrical transmission, agricultural areas for livestock and crop production, recreational use including hunting, off-highway vehicle enthusiasts, river travel and camping, and small residential communities and farmsteads.

Grizzly bear habitat in BMA 1 is classified either Recovery or Support Zones (Alberta Environment Parks, 2020). The Recovery Zone, an area covering 23,458 km² delineated through habitat modeling and expert opinion on bear occurrences, is the focus of our study area (Nielsen et al., 2009). The Recovery Zone identifies where the Alberta government reasonably expects to manage the presence of grizzly bears and reduce human-caused mortality, which has been associated with open road density (Alberta Environment Parks, 2020). The Support Zone (~18,000 km²) is intended to allow for grizzly bears to disperse, with management focusing on securing food attractants and teaching bear safety, largely with agricultural landowners and recreationalists. The Alberta BearSmart program (www.alberta.ca/alberta-bearsmart-program-overview.aspx), in existence since 2008, is the primary outreach strategy used to educate the public and address human-bear conflicts.

We used a collaborative governance approach, including jointly developing Theory of Change (TOC) to guide and evaluate our efforts, alongside employing a non-invasive genetic sampling project coupled with spatially explicit capture-recapture models to estimate grizzly bear density (0.70/1,000 km²) and abundance (16.3 grizzly bears) in the Recovery Zone of BMA 1 in northwest Alberta.

While collaborative governance is a common term used across public administration literature, how it is defined and applied remains vague (Emerson and Nabatchi, 2015). Our study adopted the principle and spirit of a collaborative governance as suggested by Emerson and Nabatchi (2015), which included decision-making processes that constructively engaged people across their different sectors, perspectives, knowledge, experiences and interests to identify the values, needs and objectives for BMA 1 grizzly bear recovery. Our collaborative approach built and nurtured relationships across otherwise disconnected stakeholders and leveraged human and financial capital to conduct the first-ever population estimate for BMA 1 grizzly bears. A major strength in our approach was collectively and iteratively developing the TOR and the TOC, and learning from our collaborative arrangement by evaluating participants' experiences (Emerson and Nabatchi, 2015; Huge and Mukherjee, 2017). Through this, we learned that collaboration provides for opportunities to develop and improve relationships through active participation in discussions, decision-making and, as in our case, supporting data collection and analysis. This is a cornerstone of collaborative governance literature, where the importance of stakeholder engagement in scientific and decision-making processes is increasingly recognized (Redpath et al., 2017). However, we also learned that greater efforts must be taken in future to engage other stakeholders in grizzly bear science and applied management, and that a BMA 1 management plan utilizing population estimates and linear footprint results is strongly desired.

Overall, we suggest our efforts to build a strong, collaborative governance structure ultimately helped move grizzly bear conservation and management forward in northwest Alberta (Ansell and Gash, 2008; Stern and Coleman, 2015). By employing a collaborative process that shared the investment of time, money and collective actions, we were able to use the best available scientific techniques to deliver a grizzly bear population estimate for BMA 1. Indeed, the formalization of the NGBT created an open, respectful, and constructive space that made sense for the context in which we were working in and ultimately take steps toward achieving grizzly bear recovery in a multi-use landscape (Redpath et al., 2017).

Specific to grizzly bear population estimation, we found that BMA 1 contains a low density of grizzly bears and is currently the lowest recorded density of an established grizzly bear population in Alberta. Elsewhere in Alberta, grizzly bear density estimates range from 5.25 to 20.4 /1,000 km² (Alberta Environment Parks, 2020). We note, however, that our detection probability (g0) for grizzly bears was remarkably low and was substantially lower than the parameters we included within our simulation exercise (Table 2, Supplementary Material B). Further, our estimates of sigma (i.e., the spatial area over which a grizzly bear can be detected) were very large (Table 2). Given this, it is likely that we would need to substantially increase our sampling effort in order to achieve a more reliable (i.e., coefficient of variation <0.2) density estimate. However, increased sampling would be expensive and logistically challenging given the characteristics of this landscape.

We did attempt to increase grizzly bear detections by adding sampling sites opportunistically along existing travel routes (i.e., roads) and including sampling stations in British Columbia. However, our low number of detections precludes us from further examining the influence of these decisions on our density estimates. Indeed, we assumed a uniform density in our SECR models, which is an oversimplification because it is likely that grizzly bear density varies as a function of habitat and proximity to roads (Boulanger and Stenhouse, 2014).

We detected only two female grizzly bears and note that sex-specific demographic forces may be driving a male-skewed sex ratio. It is unclear if this ratio is a true component of this population or an unexplained sampling artifact. Lured rub objects represented only 13% of the sampling sites but provided 39% of grizzly bear detections, with 8 of 20 male detections and 1 of 3 female detections occurring at rub objects. Generally, male grizzly bears rub more frequently than female grizzly bears, though this difference dissipates as the season progresses (Lamb et al., 2016; Morehouse and Boyce, 2016). One factor that may have influenced bear detections and rubbing behavior is that we used lured rub objects, whereas most previously published studies used natural rub objects. It is also possible that more female bears occur outside of the Recovery Zone and therefore did not encounter our traps. One potential explanation for this is that the Recovery Zone represents higher quality habitat and is thus used by the more dominant male bears, with females being excluded or seeking alterative habitat (i.e., dominance hypothesis, Elfstrom et al., 2012; Stewart et al., 2013). This theory remains to be tested and is considered a low probability scenario given the large spatial scale over which this process would be occurring. Regardless, our detected female scarcity elevates the importance of minimizing human-caused grizzly bear mortality and further emphasizes the importance of maintaining female survival in BMA 1 (Alberta Environment Parks, 2020). Educational outreach and access management strategies should therefore continue to be implemented across BMA 1 to mitigate human-bear conflict and the potential for female mortality.

Our project also attempted to measure the transboundary population between Alberta and British Columbia, but no grizzly bears were detected in this portion of our study grid. However, this lack of detections is difficult to link to differences in bear density and should be interpreted with caution (Environmental Reporting BC, 2020). And finally, while we detected few grizzly bears, a large portion of our collected hair samples were black bears, suggesting that our sampling methods were appropriate for detecting bear species. Grizzly and black bears have different life histories, where northwest Alberta's boreal landscape is likely more suitable black bear habitat than grizzly bear habitat (e.g., Bonin et al., 2020).

Despite these challenges, our abundance estimate represents the first empirical estimate for BMA 1 and is a baseline against which future BMA 1 monitoring and management actions can be measured. This includes setting objectives to address the limiting factors (i.e., human-caused mortality) for grizzly bears, implementing local management policy co-designed by the NGBT, exploring opportunities for creative and innovative strategies to implement for this landscape, and assisting in evaluating the efficacy of grizzly bear and human behavior management as well as land use and forestry practices.

Importantly, we also provide a starting point from which to evaluate the achievements and impacts of a complex collaborative governance arrangement for an at-risk species and demonstrate success despite the challenges encountered. Going forward, priority actions the NBGT has considered for local planning includes collectively mapping actual linear footprint and taking coordinated action on motorized (i.e., on-highway vehicle) access and associated impacts (i.e., reduction and restoration of existing linear features, evaluating open road densities and access barriers). Additionally, the NGBT agrees that locally relevant educational outreach, using Alberta BearSmart principles and materials, must be delivered across BMA 1. This includes engagement with petroleum and forest industry, landowners, recreationalists, and community residents to teach and encourage proactive bear safety and conflict mitigation strategies (e.g., electric fencing). Lastly, the NGBT members recognize the need for their organizations' staff and the broader public to contribute to grizzly bear observations, and as such, supports the use of the smartphone-based reporting application GrizzTracker (www.grizztracker.ca) as an effective reporting method.

This study represents the first attempt to empirically estimate the grizzly bear density and abundance across Alberta's northern BMA 1, through a multi-stakeholder collaborative arrangement called the Northwest Grizzly Baer Team (NGBT). Collaboration was at the core of our success, which went beyond simply acknowledging the need to collaborate, to enabling proactive participation by NGBT members in project design, implementation and evaluation. In turn, the relationships we fostered enabled us to leverage financial, human, social, and policy resources, and helped to build trust, reciprocity and exchange across stakeholders and the broader public through open and transparent communications (Ostrom, 1990; Pretty and Smith, 2004; Ansell and Gash, 2008; Kallis et al., 2009). Additionally, our collaborative arrangement provided extensive opportunities to co-learn, share, and engage across the broad membership of the Northwest Grizzly Bear Team.

Our summative evaluation was an important step in understanding the efficacy of our collaborative arrangement, helping to provide evidence to inform future decisions on alternative governance structures. Indeed, this team has agreed to continue to build and strengthen relationships, facilitate constructive dialogue, and share data and knowledge going forward, to develop a locally relevant BMA 1 plan. In addition to our collaborative outcomes, our efforts provide a robust grizzly bear population dataset to help fulfill Alberta's recovery policy objectives and can contribute to future performance evaluations of integrated land management. Future collaborative governance arrangements would be well-suited to look to our project as a guideline for establishing multi-sectoral teams that applied various social and natural science methodologies to define and work toward resolving a complex, real-world, species at risk problem. Overall, the relationships, funding, datasets, and commitment to future efforts would not have been possible without coming together in the spirit and practice of collaboration.

The data supporting the conclusions of this article will be made available by the authors upon request.

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study. Ethical review and approval was not required for the animal study because Animals were not handled. Non-invasive genetic sampling (hair samples) was explicitly used.

CH, LF, LV, and NM conceived of the project. CH and LV directed the project. LF led fieldwork logistics, with participation by NM, CH, and LV and project partners. LV completed the secr analysis, with AM and RS reviewing. CH completed the questionnaire analysis with review by project partners and led the manuscript writing process, with all other authors contributing content and revisions. All authors contributed to the article and approved the submitted version.

This work was supported by a Forest Resources Improvement Association of Alberta Grant (FRIAA-FFI-15-004), annual direct funding support from Mercer Peace River Pulp Ltd., and Alberta government in-kind contributions. These funders were not involved in the study design, collection, analysis, interpretation of data, the writing of this article, or the decision to submit it for publication.

JW was employed by Mercer Peace River Pulp Ltd. at the time of this study. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.