Linking co-monitoring to co-management: bringing together local, traditional, and scientific knowledge in a wildlife status assessment framework

Here, we explore how different monitoring approaches and knowledge types can feed into the assessment framework we have outlined. We draw on our experience with a caribou and muskox health monitoring initiative that began in the Sahtu Settlement Area, Northwest Territories in 2003 as a partnership among Indigenous communities, wildlife managers, and academic researchers and has since expanded to communities in the Kitikmeot, Nunavut and Inuvialuit, Northwest Territories regions (Brook et al. 2009; Carlsson et al. 2016; Di Francesco et al. 2017; Tomaselli et al. 2018b). This program has evolved over time to include data from conventional scientific approaches (e.g., live animal sampling, population census, and targeted surveillance; Table 2), HBS (Brook et al. 2009; Kutz et al. 2013), and LEK/TEK documented through structured and semi-structured individual and group interviews, including participatory activities with both Indigenous and non-Indigenous community members (Tomaselli et al. 2018b). The Ekaluktutiak Hunters and Trappers Organization, Kugluktuk Angoniatit Association, and Olokhaktomiut Hunters and Trappers Committee have been integral to the success of these wildlife monitoring programs by providing direction and motivation for monitoring activities and engaging their membership in the HBS program; because of their collective long-term involvement and the fluid nature of both membership and executive of these organizations, we have included the organizations themselves rather than specific individuals as coauthors on this paper.

Conventional scientific approaches to monitoring caribou and muskoxen have been well documented elsewhere (Adams et al. 2008; Kutz et al. 2013). Here, we describe our approach to HBS and documenting LEK/TEK (specific details are provided in associated publications) and then illustrate how these could be used to identify and inform indicators of population health for caribou and muskoxen.

Different approaches to knowledge documentation, including conventional scientific monitoring, HBS, and interviews, may target different knowledge types and can complement each other to generate deeper insights into ecosystems and population status and trends (Table 2). This concept is not novel; government biologists and northern communities have a long-standing history of working together to monitor and manage wildlife (Government of the Northwest Territories 2019). There have also been various community-based wildlife monitoring programs initiated by academic researchers, communities, governments, and co-management organizations (e.g., Ferguson et al. 1998; Pearce et al. 2015; Henri et al. 2018); we do not attempt a comprehensive review of these programs. Rather, we briefly describe our collaborative HBS program for caribou and muskoxen and our approaches to documenting LEK/TEK through interviews and participatory activities (Tomaselli et al. 2018b), with a focus on how this information can feed into the assessment framework.

HBS typically involves hunters in the systematic collection of biological samples and data for scientific analysis. Hunters can provide samples that are otherwise difficult to procure from free-ranging animals through conventional scientific approaches, such as organs that can only be obtained post-mortem. Biological samples collected by hunters provide information on infectious disease and on biological and physiological parameters of sampled animals (Kutz et al. 2013; Patyk et al. 2015; Di Francesco et al. 2017; Wittrock et al. 2019), which are important determinants of population health (Deem et al. 2001; Tompkins et al. 2011). For example, in our HBS program, hunters may submit blood on filter paper that can be used to test for exposure to pathogens (Curry et al. 2011; Carlsson et al. 2019), a piece of hide from a standardized part of the animal that can provide hair cortisol levels (Di Francesco et al. 2017), and the jaw bone with teeth that can yield insight into age, morphometrics, diet, and overall health. Changes in individual health can result in reduced growth, survival, and reproduction, and can therefore be predictive of changes in population size (Couturier 2012). Information on individual health can thus facilitate proactive management to pre-empt (or at least expect) potential declines, whereas a population census would not reveal declines until they were well underway. Information on individual health can also reveal mechanisms of population change (e.g., increase incidence of disease or poor nutritional status), thus complementing abundance estimates and informing targeted, management actions (Couturier 2012). Finally, ongoing monitoring of infectious disease and contaminants in harvested animals is important to assess the risk of zoonotic disease transmission or exposure to high contaminant levels (Tomaselli et al. 2016; Cunningham et al. 2017).

Although involving hunters in the collection of biological samples and data is not unique (e.g., Jessup 2003; Stallknecht 2007; Artois et al. 2009; Bell and Harwood 2012; Ryser-Degiorgis 2013; World Organization for Animal Health 2014), simultaneously gathering broader ecological data from hunters is less common. Combining biological samples and hunters’ observations can provide additional information on individual and population health, as well as changes in these over time. For example, our program systematically maps and documents hunters’ observations of morbidity and mortality, group size and structure, population trends, behaviour, predator occurrence, and more; observations that are grounded in their lifetime of experience. Hunters also report on anything unusual in harvested animals or observed during their hunting trips; when feasible and acceptable, they are asked to sample abnormalities from harvested animals (Kutz et al. 2013; Carlsson et al. 2016; Tomaselli and Curry 2019).

We see an opportunity with our framework to more formally and systematically include information from HBS when identifying indicators, metrics, and benchmarks. Including information from HBS programs may make status assessments more accurate, reliable, and timely by providing current samples and observations on an ongoing basis that complements conventional scientific monitoring (Table 2). For example, an emergent disease or decline in body condition may be detected through HBS long before it may be detected through population surveys, which can both stimulate and guide subsequent monitoring (Tomaselli et al. 2018b). The information provided by hunters, which is grounded in an intimate connection to the environment, can be more complex and holistic than targeted scientific observation and may provide important insights about the processes underlying observed changes in individual and population health (Kutz and Tomaselli 2019). The lived experience of the hunters, documented through surveys and interviews, can also provide historical baselines useful for identifying benchmarks within the framework. Finally, HBS programs can help build capacity in northern communities by strengthening the role of hunters’ and trappers’ organizations, creating employment opportunities such as the program coordinator, and providing training and research opportunities for youth, as well as supporting food safety and food security. As with any monitoring approach, careful consideration of sample sizes and potential biases and limitations is essential to ensure that status outcomes are representative and reliable (Table 2).

Documenting the LEK/TEK held by hunters is an important and novel aspect of our HBS programs, but LEK/TEK of the broader community can also inform status assessments. LEK/TEK include observations derived from direct interaction of local people with the environment (Table 1). TEK also reflects cultural values and norms of Indigenous peoples that have developed and been passed down over time immemorial (Usher 2000; Thorpe et al. 2001b). As such, TEK is both a source of knowledge and a way of knowing or understanding phenomena — a system of knowledge and interpretation like the scientific method (Wilson 2008). Although we use the term “ecological knowledge”, traditional knowledge that is not conventionally “ecological” in nature (e.g., related to cooking, crafting) can still reflect changes in wildlife populations and is not to be excluded from the framework (Kendrick and Lyver 2005; Kermoal and Altamirano-Jiménez 2016).

LEK/TEK can provide historical baselines for determining benchmarks that extend back in time beyond the scientific record (Ferguson et al. 1998; Eckert et al. 2018) and may identify indicators and metrics not described in the scientific record (e.g., Ostertag et al. 2018). Such indicators and metrics may be identified through individual and group interviews with community members and then corroborated through feedback sessions. Further descriptions of qualitative methods that would be appropriate here are described by, for example, Thorpe et al. (2001a), Kendrick and Lyver (2005), Ljubicic et al. (2018a), and Tomaselli et al. (2018b).

LEK/TEK can also inform metrics commonly used in science to estimate mortality and reproduction rates and model population trends (Holmes and York 2003; Bender 2006), such as abundance, sex ratio, calf/cow ratio, age structure, and pregnancy rates (e.g., see Tomaselli et al. 2018b). The current wildlife-management paradigm is still based on conventional science (Löfmarck and Lidskog 2017; Tomasini 2018), and translation of qualitative observations into (semi-)quantitative data that can be more readily included into existing co-management processes can be useful (Berkes 2009; Robinson and Wallington 2012; Kutz and Tomaselli 2019). This has been done using semi-structured individual and group interviews, including participatory activities for knowledge documentation, such as proportional piling, scoring, and mapping (see Tomaselli et al. 2018b for details of these methods). If interviews are facilitated regularly, relative changes over time in health and demographic metrics may become apparent earlier than they would through infrequent or irregular conventional population monitoring (Kutz and Tomaselli 2019). Conversely, conventional scientific approaches such as aerial surveys are designed to provide absolute population sizes; however, this information may come too late to implement proactive management that pre-empts declines.

Equally important as identifying and informing indicators and metrics is establishing the relative importance of different indicators and their sensitivity and specificity in the context of the assessment. Ideally, this is done prior to undertaking a status assessment, based on LEK/TEK and SK of how changes in the metrics and indicators relate to population dynamics. Different metrics will need to be weighted and integrated when an overall status of the wildlife population is needed (e.g., to inform management). Particularly when different metrics or indicators yield different status outcomes, a collective discussion amongst partners is required for a thorough evaluation of data and interpretation of results (Bohensky and Maru 2011; Armitage and Kilburn 2015). As we have not yet fully implemented this assessment framework, the integration of multiple metrics is beyond our experience, to date, but may be the crux of the status assessment process. Although this framework provides a practical approach to implement LEK/TEK in status assessments, it does not replace the need for meaningful and ongoing consultation and engagement.

To illustrate the traffic light approach to assessing indicators under our assessment framework, we show how different sampling approaches and knowledge types can be used assess the indicator “body condition” for muskoxen and caribou. Conventional scientific monitoring of body condition has used various metrics including a standardized body condition score on live animals (i.e., the CircumArctic Rangifer Monitoring and Assessment protocol; Adams et al. 2008) and measurement of back-fat thickness, kidney fat, or bone marrow fat percentage on carcasses (Kutz et al. 2013). Hunter assessments of harvested animals and interviews with hunters can also inform body condition by drawing on the extensive experiential knowledge of hunters and their routine assessment of animals over broad spatial and temporal scales (i.e., LEK/TEK). These different metrics of body condition can be informed by a combination of approaches (Fig. 1). For example, back fat can be measured by ultrasound during live-animal handling by wildlife professionals, yielding precise but sparse data that are limited in number of samples and temporal coverage. In our HBS program, back fat is measured directly by hunters on harvested animals using a ruler included in the sample kits, providing seasonal patterns of condition over time (Fig. 2) (see Carlsson et al. 2016; Tomaselli and Curry 2019). Hunters also record their subjective assessment of body condition of hunted animals by categorizing animals as “skinny”, “not bad”, “fat”, or “really fat” (Fig. 3). The body condition score that a hunter assigns to an animal may depend on their previous experience, and thus the categorical assessment may be relative to past hunts, whereas the measurement of back fat is absolute. At the population level, the application of participatory activities during interviews can be done annually to document hunters’ perspectives on the condition of all observed animals in a single year and gather insights on changes from previous years (e.g., Lyver and Lutsel K’É Dene First Nation 2005; Tomaselli 2018; Tomaselli et al. 2018b). Participatory activities — such as proportional piling (Tomaselli et al. 2018b) — provide a semi-quantitative approach to documenting LEK/TEK, facilitating the comparisons of condition among locations and over time (Fig. 3). For example, to document perceptions of muskox and caribou body condition, Tomaselli et al. (2018b) had participants divide counters (beans) among four body condition categories, and the quantity of beans in each category were then measured or weighed to yield the proportion of the total number of animals observed that were in each body condition category (Fig. 3).

To understand whether observed changes in body condition over time represent a conservation concern, metrics must be compared to benchmarks to designate a population as having green, amber, or red status. Benchmarks for the body condition metrics that we describe above have not yet been determined, but we illustrate the application of the proposed framework using hypothetical benchmarks for the HBS back-fat measurements (Fig. 2) and body condition scores (Fig. 3) from our HBS programs for muskoxen in Ulukhaktok, Northwest Territories. The resulting status outcomes are for illustration only and should not be considered a definitive status assessment. Defining the true benchmarks needs to be done in partnership with the TEK holders, considering both the Indigenous and scientific understanding of the species’ ecology and ensuring that the TEK is interpreted appropriately.

For back-fat thickness of harvested muskoxen from Ulukhaktok, Northwest Territories, we illustrate two different approaches to defining benchmarks: (1) based solely on the HBS data for back-fat thickness, setting the limit benchmark as the 25th percentile of all available back-fat data (i.e., 2016–2019) for each season and the target benchmark as the 75th percentile of all available back-fat data for each season (dotted lines in Fig. 2), and (2) based on narratives from interviews with hunters, described by Tomaselli et al. (2018b), which indicated that the back fat on muskoxen harvested in the fall around the neighbouring community of Iqaluktutiaq, Nunavut, during population declines was often less than 1 cm (limit benchmark), and pre-decline was often greater than 5 cm (target benchmark; dashed lines in Fig. 2). The data-informed benchmarks account for natural seasonal variation in back-fat thickness but may result in biased status assessments if the time series is short and (or) there have been directional changes in back-fat thickness that result in a shifting baseline (Pauly 1995). Status outcomes may also be biased if changes in hunting preferences or the availability of animals leads to a directional shift in the age or sex of harvested animals, which would make it inappropriate to compare current measurements to historical data. The benchmarks developed from interviews provide a broader context for the metric, both spatially, because they are based on information from another community, and temporally, because LEK/TEK spans a longer time period than the HBS data (Table 2). They are also subject to potential biases if there have been changes in hunting habits over time. In our case, the interviews specifically referred to the fall season, when muskoxen are the fattest, and are not applicable to other seasons. Follow-up interviews may address this by obtaining more detail on seasonal variability in back fat from LEK/TEK. In this case, both sets of hypothetical benchmarks yield an amber status for the most recent back-fat measurements (Fig. 2).

For the body condition scores of harvested muskoxen recorded by hunters, we summarized the raw data (Fig. 3A) by fitting a general additive model for ordered categorical data that we then used to predict the probability of an animal being classified as fat or really fat through time (Fig. 3B). This probability was the quantifiable metric for which we illustrate benchmarks. In this example, we define hypothetical benchmarks from LEK/TEK only, using the results of group proportional piling exercises led by Tomaselli et al. (2018b) in Iqaluktutiaq, Nunavut. On average, participants reported that 74% of observed muskoxen were in good or excellent condition (i.e., fat or really fat) prior to population declines (target benchmark), but only 47% of observed muskoxen were in good or excellent condition in the decline period (limit benchmark; see fig. 4a of Tomaselli et al. 2018b). These interview-based benchmarks do not capture the seasonal variation in body condition present in the data, but nonetheless provide a reference against which the metric can be compared. We note that a shifting baseline may occur in this case because the metric is relative to the hunters’ experience; Tomaselli et al. (2018a) suggested that body condition scores documented though interviews may have underestimated the deterioration of body condition if the subjective scale of measure of the observers was adapting to poorer baselines. Given the interview-based benchmarks, the most recent hypothetical status under this metric would be amber (Fig. 3B), but this would be optimistic if a shifting baseline has occurred.

The potential applications of our proposed framework are broad. We illustrated the application of the framework using two metrics of body condition of muskoxen, but the concepts introduced can be adapted to other indicators, wildlife species, or even other resources (e.g., water; Castleden et al. 2016; Mantyka-Pringle et al. 2017). In any case, the important first step is to identify indicators and select sensitive, specific, and reliable metrics that allow for timely and representative assessments. Ideally, each indicator is informed by multiple metrics that draw on different knowledge types and data collection methods. Combining LEK, TEK, and SK documented or collected through different approaches (e.g., HBS, interviews, conventional scientific monitoring) to inform the same indicator can help compensate for the limitations, biases, and uncertainties inherent to any single method (Table 2), ultimately increasing the depth, accuracy, and reliability of assessments (Plummer and Armitage 2007; Reed 2008; Berkes 2009; Ljubicic et al. 2018b; Tomaselli et al. 2018b, 2019). Also, documenting resource users’ observations through HBS may facilitate early detection of changes in wildlife population health and status, as observations are more frequent and continuous than conventional scientific monitoring, potentially allowing for proactive wildlife management rather than recovery planning (Kutz and Tomaselli 2019). Bridging different knowledge types, and thus, different data sets, allows the framework to connect information from separate monitoring initiatives. Thus, our framework has the potential to maximize the benefit of data that may already exist in fragmented formats, streamlining a process that may otherwise be a burden on conservation partners.

Selection of metrics and respective methods for informing those metrics should take into account the feasibility and logistics of sampling (e.g., cost), ethical considerations (e.g., respecting traditional practices, minimizing animal mortality), sample sizes, and the limitations associated with each approach (Table 2). Community-based monitoring programs can be ideal for informing metrics, but logistical and scientific challenges to such programs include: securing sustainable funding sources (Bell and Harwood 2012); securing the long-term employment of a local coordinator to manage such programs who can communicate in the “languages” of both SK and LEK/TEK and has the skills and resources (e.g., computing) to do so; ensuring sample and data quality; providing appropriate and timely communication of results (Brook et al. 2009); establishing standardized protocols for sample collection and LEK/TEK documentation to ensure comparability across regions (Adams et al. 2008; Kutz et al. 2013); harmonizing data standards and a common database; and determining if adequate sample sizes are obtained. The best approach for informing wildlife conservation status assessments will make the most of existing activities, can be standardized across localities and years, and answers the key questions at hand in a timely manner (Berkes 2009; Tengö et al. 2017).

In this case study, we examined how different metrics could be informed by knowledge documented through a community-based monitoring program of caribou and muskox health. To fully implement the proposed framework, we need to identify a suite of appropriate indicators and metrics and relevant benchmarks for each metric that delineate status zones. Defining benchmarks for true status assessments is a complex process that requires inclusive discussion with all partners to understand their priorities, risk thresholds, and conservation targets.

Multiple metrics and indicators will also need to be weighted and integrated to determine overall status, requiring the consensus of scientists and LEK/TEK holders (Fig. 1). When different metrics yield conflicting results, partners will be forced to consider the context, limitations, and biases of different knowledge sources to understand the discrepancies (Table 2). For example, conflicting outcomes may arise due to differences in the spatial scale or timing of sampling, or reflect changes in the ecology (e.g., migration patterns) of wildlife populations. This deeper understanding may not become clear until the knowledge is processed together within the framework.

Appropriate indicators are those that are sensitive and specific to changes in wildlife conservation status (Rice and Rochet 2005). Although local and targeted management actions may be taken in response to individual indicators, the cumulative and interactive effects of multiple indicators of health is likely more important for understanding population resilience and viability (Macbeth and Kutz 2018). Population-level studies that link indicators to changes in population size drawing on a combination of LEK, TEK, and SK are needed to understand potential interactions and predict the impact of future change.

In this paper, we focused on intrinsic factors that are indicative of individual and population health, but extrinsic factors are also important determinants of population health and trends. Extrinsic factors, both biotic (e.g., predator abundance) and abiotic (e.g., climatic events such as rain-on-snow), may be driving changes in intrinsic health indicators (e.g., nutrition and body condition) and population dynamics (e.g., mortality and reproduction). TEK inherently understands species on an ecosystem level with both extrinsic and intrinsic factors, and is, thus, naturally suited to informing such connections. Incorporating extrinsic indicators into status assessments using a similar approach that combines multiple knowledge types and data collection and documentation methods will further improve proactive wildlife management.

As climate change exerts increasing pressure on Arctic species, examples of shifts in behaviour and phenology are expected to increase (Beever et al. 2017; Pecl et al. 2017). Documenting LEK/TEK of wildlife health indicators and informing metrics with a combination of LEK, TEK, and SK can help ensure that health monitoring and conservation status assessments are timely and relevant in the face of rapid ecosystem change. Approaches such as those presented here can facilitate proactive co-management that helps to maintain healthy wildlife and healthy communities in the Arctic.