All knowledge is inevitably situated and partial, and management strategies that are limited to a few forms of knowledge have limited capacity to generate innovative solutions to complex problems (Haraway, 1988; Matos, 2015; Rosendahl et al., 2015). Transdisciplinary knowledge co-production that embeds scientific and non-scientific knowledge into research and decision-making processes has become a popular, yet difficult, objective within water governance (Brugnach and Özerol, 2019). Knowledge co-production is a pluralistic approach that appreciates the validity and relevance of multiple ways of knowing. It brings together diverse sources of knowledge and perspectives to generate “context-specific knowledge” and explore potential strategies for tackling complex problems (Miller et al., 2008; Reyers et al., 2015; Norström et al., 2020). Engaging with different participants in knowledge co-production also improves social networks, cultivates communal ownership, and builds capacity for future engagement (Armitage et al., 2011). Co-production can take place in a multitude of ways, depending on how the participation is structured and the overarching goals of the program. Within adaptive management, co-production can occur from the point of project initiation, when problems are identified, framed, and placed in context. Conceptual and quantitative models can be produced using multiple sources of knowledge and are based on the values and inquiries of the participatory group. Adaptive management learning processes that integrate specialist, local, traditional, and social knowledge bases widen the scope of learning that can be pursued and open up new avenues of exploration in the management of the system (Fernández-Giménez et al., 2019).

The National Freshwater Ecosystem Priority Areas (NFEPA) project in South Africa is a useful example of a national-scale freshwater conservation project that targeted knowledge co-production. Over the course of the 4-year process, over 450 participants were involved in the co-production of conservation goals and spatial data that ultimately resulted in an atlas and implementation manual identifying 37 areas for conservation. These results are available freely on a supporting website and uptake within the identified areas has been promising (Nel et al., 2016). The NFEPA process was designed to create space for dialogue among participants from diverse knowledge backgrounds and they successfully engaged in the co-production from developing objectives to the final production of maps.

Pairing social and technical learning through iterative processes enables wider stakeholder ownership of knowledge and can encourage collaborative environmental governance (Wyborn, 2015; McLoughlin et al., 2016; van der Molen, 2018). Adaptive management traditionally uses a single-loop learning approach, in which management is conceptualized as an experiment within the system and the results of the experiment are used to update knowledge of the system and design new management strategies (Stem et al., 2005). While this type of learning is critical to the adaptive management process, this style of deductive hypothesis testing assumes the ongoing validity of underlying values and goals. Double-loop learning describes institutional learning, in which the decision making processes themselves are updated through iterations (Williams, 2011; Petersen et al., 2014; Williams and Brown, 2018). Beyond this, triple-loop learning encourages stakeholders to revisit and modify the underlying beliefs and perceptions that drive management (Pahl-Wostl, 2006). Social learning to support institutional reframing occurs through collective processes in which learners’ beliefs are updated through successive interactions with one another and the management environment (Pahl-Wostl, 2006; Fernández-Giménez et al., 2019).

Embedding social learning into the adaptive management cycle will allow us to critically examine the institutional frameworks and processes that govern e-flows management Assessing these structures and making incremental changes will allow a shift from rigid river regulation to a “multifunctional dynamic landscape” (Pahl-Wostl, 2006). Social learning, whether structured within management or emerging organically, is highly dependent on “learning spaces” where stakeholders can share knowledge, develop common understandings, and work cooperatively (Lumosi et al., 2019). Moellenkamp et al. (2010) explored how intentional, informal participatory processes run in parallel to the formal water management process facilitated social learning and provided avenues for experimentation. They found that the participatory process facilitated institutional changes and shifts in applied methodology for management (Moellenkamp et al., 2010). Restructuring the adaptive management cycle to create learning spaces through the integration of participatory processes allows participants to engage in critical social learning about the relationships between stakeholders, management frameworks, and the river.

E-flows management is complicated by the many uncertainties associated with a complex social-ecological system, including unknown flow-ecology relationships (Bunn and Arthington, 2002; Lynch et al., 2018; Watts et al., 2020), measurement uncertainty (Stewardson and Rutherfurd, 2006; Goguen et al., 2020), the dynamics of a non-stationary climate and environment (Milly et al., 2008; ThompsonLaizé et al., 2014; Poff, 2018), and ambiguous or shifting social perceptions of the water resource systems (Gleick, 2000; Hogl et al., 2012). Adaptive management seeks to reduce the uncertainties associated with ecological systems through a structured learning process (Williams, 2011; Webb et al., 2017). Recognition and transparency regarding uncertainty fosters trust between stakeholder groups and increases acceptance of management programs despite these uncertainties (Acreman et al., 2014; Conallin et al., 2018b). Previous literature on e-flows and water resources has focused on ecological and environmental uncertainty; social-ecological relationships and perceptions of management represent another axis of uncertainty. This will be increasingly the case as we adjust to managing under a changing climate where different sets of trade-offs and decisions will be needed (Horne et al., in review). Societal and personal values regarding the environment also shift over time (Kendal and Raymond, 2019) and in response to engagement with environmental issues and political dialogues (Hards, 2011; Corner et al., 2014; Wolsko, 2017). Maintaining legitimacy through time will require transparency about uncertainty and non-stationarity in both physical and social dimensions of environmental water management. We suggest that social values regarding flows and management are also non-stationary and that diverse stakeholder engagement throughout the adaptive management cycle will allow us to capture these changing social-ecological contexts and embed them into management.

Legitimacy is crucial to the success of adaptive e-flows management, as public trust and confidence in management agencies is what allows them to function (Horne et al., 2017; O’Donnell and Garrick, 2017). The concept of legitimacy can be constrained by a focus on the centrality of government institutions and agencies that are presumed to be acting in the public interest and supported by sound technical guidance (Gearey and Jeffrey, 2006; Godden and Ison, 2019). In countries where federal and state institutions are responsible for e-flows governance, community participation is often mandated through legal instruments that may have a narrow, inflexible definition of engagement (Godden and Ison, 2019). Fostering legitimacy for e-flows programs and associated management is two pronged, requiring focus on both input and output legitimacy (Hogl et al., 2012; O’Donnell et al., 2019). E-flows programs have previously depended largely on building output-based legitimacy, defining their credibility based on the success and efficacy of their management programs as shown through scientific indicators. While this is a necessary component of overall legitimacy, building input legitimacy in parallel through process-focused, stakeholder driven initiatives builds public trust and confidence for program success, and in turn helps to bolster trust in program outputs. Input legitimacy focuses on transparency, access, representation, and accountability throughout decision making and management. These values encourage stakeholders to create a shared understanding of the problem and develop a shared vision for success (Cullen, 1990; Webb et al., 2010).

In an analysis of water management projects in the Netherlands, van Buuren et al. (2012) describe throughput legitimacy as the carry through of democratic principles and deliberative opportunities throughout the management process. In one of their case studies, there were protests following a dike improvement proposal that had no community input. In response to the protests, the original proposal was scrapped, and a collaborative process was developed with the agreement that any new proposal must have public support. The new process emphasized communication and transparency by building in key opportunities for citizens to contribute their voice. This new process complemented existing institutions and frameworks, leading to a hybridized strategy with greater throughput legitimacy (van Buuren et al., 2012).

Stakeholders are broadly defined as the individuals, organizations, and institutions that have an interest in the outcome of an e-flows program. This definition casts a wide net, particularly given that stakeholders may define themselves as such and seek engagement while other stakeholders may actively choose not to be involved. The narrative around stakeholder engagement has often focused on the aspirational inclusion of communities, framing the community participants as stakeholders while other participants, such as bureaucrats, managers, and experts, are responsible for facilitating the engagement process. In practice, it is rare that e-flows projects frame these other participants as stakeholders with distinct values and perspectives. However, previous literature in the field of e-flows identified three primary categories of stakeholders, distinguishing researchers, and water managers as discrete stakeholder groups along with local communities (Edelenbos et al., 2011; Webb et al., 2018). Webb et al. (2018) conceptualized these groups within a Venn diagram, where management strategies and projects involve different combinations of groups. In addition to these three groups, we have added Indigenous peoples as a discrete group, given their unique relationship with and knowledge of catchments, their recognition by existing governance frameworks, and their traditional and ongoing role as custodians (Nania and Guarino, 2014; Wilson, 2014; Jackson, 2017; Moggridge et al., 2019). E-flows management, like all natural resource management, is underpinned by the values, data, knowledge, and people involved in the process (Kennedy and Koch, 2004). Management strategies involve different contributions from these stakeholder groups and an ideal adaptive management approach would lie at the intersection of the four groups.

Conceptualizing engagement between stakeholders in complex management scenarios is always difficult. We have defined four broad stakeholder categorizations (Indigenous peoples, water managers, researchers, and local community) to enable a discussion of how stakeholder participants might engage with the adaptive management process. These broad groupings are useful, as this is how stakeholders are often identified for inclusion in the participatory process, regardless of how their role in the process is defined or evolves. Using these categories to identify and recruit stakeholder participants ensures that critical groups of stakeholders are included throughout the process. This is particularly important, given that key stakeholder groups are often left out of the management process (Webb et al., 2018) leading to incomplete learning cycles and poor social legitimacy. We recognize that these categories are broad and include a myriad of possible participants and organizational arrangements. The stakeholder participants for any program will be influenced by the scale of the project and by the governance structures and policies already in place. It is also important to note that this framework, as with any management framework, will not be able to describe all the possible nuance of stakeholder participants and arrangements. Rather, it is a tool for conceptualizing these relationships within the adaptive management context.

Table 1 highlights two e-flows case studies with a range of participants, and groups the participants into the stakeholder categories used for this framework. In the Honduran case study, a large multi-national NGO, The Nature Conservancy, was asked to fulfill the role of facilitator as well as complete much of the technical work associated with the e-flows assessment. Hence, we have defined them as researchers within our stakeholder categories. The only water managers involved in the Honduran case study were technical and engineering representatives from the ENEE, a quasi-governmental hydropower management agency that is responsible for dam operations. The workshop consultation process for their project included representatives from Indigenous communities, local government officials and NGOs, and a range of Honduran and international experts (Esselman and Opperman, 2010). The Australian case study included a range of participants in an e-flows assessment. While the project was organized by the local catchment management authority, a university team acted as the project leads, facilitating workshops and supplying technical modeling expertise. In addition to the university team, a panel of discipline experts were recruited for the project, fulfilling the role of researchers within our framework. Local community participants included local elected council members, landholders, irrigators, and other interested citizens. The water policy officers from two local Aboriginal organizations were also included in workshops and played a role on the project steering committee. Given the complicated governance structure surrounding water in Australia, several different water management agencies were included, representing basin, state, and local scales of management (Horne et al., 2021).

Table 2 outlines some key features of these four stakeholder categories and the importance of their roles throughout the e-flows process. First, we note that here we refer to Indigenous peoples as stakeholders; however, we acknowledge that in some contexts it would be better to refer to them as rights holders (see Jackson, 2018; Latta, 2018; Pomart, 2020). Potential roles of Indigenous groups are highly context specific and depend on multiple factors including but not limited to: the empowerment of groups at both federal and local levels, the degree of colonization, historical disenfranchisement, and the extent of Indigenous diasporas (Woodward and McTaggart, 2016; Stefanelli et al., 2017; Clapcott et al., 2018; Norman, 2018). While we have described Indigenous peoples as a stakeholder group akin to water managers and local communities, it is important to understand the significance of customary management practices and dynamic social-ecological relationships that predate colonial settlement and persist today despite post-colonial institutionalization (United Nations Declaration on the Rights of Indigenous Peoples, 2008; Chen et al., 2018; Magdaleno, 2018). Indigenous groups have varied recognition and legal rights within water governance globally (Macpherson, 2019). Currently, Indigenous communities and organizations assert custodial rights, challenge existing governance frameworks, and form collaborative partnerships with non-Indigenous organizations. The ways in which Indigenous communities engage with water governance varies from country to country and regionally within nations. Leading and participating in various forms of water governance (including e-flows) can play an important role in self-determination (von der Porten and de Loë, 2013; Pirsoul and Armoudian, 2019). Because of the importance of context-based approaches for Indigenous inclusion, we do not make specific recommendations for their role in the e-flows process. However, we recommend that Indigenous groups be included as early as possible and throughout the adaptive management cycle using any participatory guidelines developed by the groups themselves (Jackson et al., 2012; Crow et al., 2018). Community is often the focal point of engagement programs in natural resources management, and it is widely recognized that community support is critical for the sustained success of e-flows programs (Horne et al., 2017; Allan and Watts, 2018; Watts et al., 2020). However, when poorly executed, these attempts can become tokenistic, and shallow engagement can harm the long-term success of flows programs (Conallin et al., 2017; Pirsoul and Armoudian, 2019). The definition of community is often left amorphous but can be broadly defined as individuals and groups who live or work locally and have a stake in the decision-making process. This may include irrigators, recreational users, conservationists, local government and politicians, as well as other concerned citizens who identify themselves as stakeholders. Community participants bring specific local knowledge to the decision-making context and can reflect broader community-based values and perspectives. This localized knowledge is rooted in the day-to-day experiences with the river and reflects social understandings of the system. Because the umbrella of community covers a range of individuals and groups, a broad spectrum of viewpoints will be present and there is a potential for conflict and competing perspectives, both within the group and with other stakeholder groups (Haddaway et al., 2017). While negotiating these dynamics is a challenge, with careful conflict resolution it can also be an opportunity to identify shared values and engage in critical social learning (Carr, 2015; Conallin et al., 2017). Successful engagement with community members will recognize their role in knowledge co-production and increase their capacity to participate in all phases of the adaptive management cycle, including decision making.

The published literature has tended to view researchers as unbiased experts in their field and has depended heavily on their guidance in creating water management policy (Stewardson and Webb, 2010); thus valuing this technical knowledge above other forms (Edelenbos et al., 2011). However, casting researchers as impartial and neutral observers can be problematic given observed expert bias (de Little et al., 2018) and the intrinsic personal perspectives individual researchers bring to their interactions (Yamamoto, 2012; Mandel and Tetlock, 2016). Similarly, water agency representatives are seldom framed as stakeholders in e-flows management, as they are often the organizers of stakeholder engagement activities. Water agency representatives also bring their own biases and values to the decision making process, particularly when considering risk (Kosovac and Davidson, 2020). Although researchers and agency representatives are typically considered ‘outside’ the formal engagement process, the groups share key aspects with other stakeholders. They have a unique knowledge of the system, often live within or near the catchment of concern, and have a vested interest in catchment management.

Restructuring e-flows management to acknowledge researchers and agency representatives as distinct stakeholders akin to community and Indigenous peoples opens new avenues of collaboration and creates space for dialogue between the groups. Researchers and agency representatives can play a meaningful role in knowledge creation when framed as stakeholders with unique sets of values, data, and knowledge (Rosendahl et al., 2015). Placing researchers and agency reps amongst other stakeholders allows them to make their values and perspectives explicit. These values then become one piece of the larger management puzzle, on a par with those of other groups, and allows decision makers to balance multiple types of knowledge and varied perspectives (Hare and Pahl-Wostl, 2002; Hare et al., 2006; Raymond et al., 2010; Edelenbos et al., 2011).

It is important to recognize that stakeholder groups are neither homogenous nor static; each group’s perspectives will be dynamic, varying both within the group and through time (Steyaert and Jiggins, 2007; Conallin et al., 2017). Therefore, engagement programs should be conceptualized as long-term programs that continuously reengage with participants, recruit new participants, and are self-reflective enough to capture changing perspectives and relationships. Stakeholder recruitment and analysis, detailed in Table 2 under the Planning heading, is a critical step in identifying the stakeholder participants and beginning the engagement program. Stakeholder analysis can be used to systematically identify the individuals or groups who have a long-standing interest in e-flows decisions, are potentially impacted by management actions, or are already in a position of influence (Reed, 2008; Reed et al., 2009; Conallin et al., 2017). A stakeholder analysis may also identify existing social interactions between the various groups and the river, including potential sources of tension and conflict. Building a flexible engagement strategy will help account for these considerations, allowing different stakeholder groups to participate in the ways they deem appropriate and when they have the capacity to do so.

Adaptive management has previously been identified as a useful approach for e-flows based on its ability to deal with complex and uncertain systems (Webb et al., 2018). The adaptive management cycle is an iterative process divided into three primary phases, planning, learning, and doing. Figure 1 illustrates how the common steps in e-flows management align with the phases of adaptive management. Mapping these two processes together allows us to determine where in the cycle various stakeholder groups should be engaged and in what manner. Each phase of the adaptive management cycle presents an opportunity for meaningful stakeholder engagement allowing the phases to be underpinned by the people, knowledge, data, and values of each stakeholder group. Successful engagement will involve all four stakeholder groups to varying degrees in every phase, with contributions from each group changing over time. It is impossible to define the exact level engagement for different stakeholder groups at every step of e-flows management and being too prescriptive in this regard runs the risk of limiting a program’s ability to evolve through time. We believe that the context of each e-flows program and the stakeholders themselves should determine the level of engagement and guide tool selection.

Table 2 provides detail on the e-flows activities that align with the phases of the adaptive management cycle shown in Figure 1. We suggest guiding principles for each of these phases and elaborate on some the important considerations for these activities. In Table 2, we have focused on activities that typically take place during e-flows management, such as environmental flows assessment and recommendations, modeling, and monitoring, etc. These activities can all include engagement with the four stakeholder groups but are often not structured to include participation as a fundamental component of the activity. The participatory methods highlighted in Table 2 are broad frameworks or concepts that will require a suite of tools to execute, including workshops, steering committees, focus groups and other tasks (Hare et al., 2006). These participatory methods have been identified through a survey of the literature regarding stakeholder engagement in natural resources management, though we do not present a comprehensive list of all possible participation frameworks and tools. These methods should be selected based on the context of individual projects and approaches, including the resources and capacity of stakeholder groups involved.

Many of the activities in the planning phase of e-flows management fall under the broad umbrella of the e-flows assessment (printed in bold within the Planning row in Table 3), an activity that occurs in the initial implementation of a program and is periodically updated (Tharme, 2003; Jain, 2012). Much of the planning for an e-flows program takes place on a long-term time scale, as e-flows assessments take a considerable amount of time and resources to conduct. In practice, a broad e-flows assessment is often conducted every 5–10 years for a catchment. Seasonal and yearly e-flows planning takes place on a much shorter time frame and is informed by the e-flows assessment as well as updated learning. Successful stakeholder engagement for the program begins with the early development of an engagement strategy that is supported by the involvement of participant stakeholders from its conception (McLoughlin et al., 2016; Conallin et al., 2017). Stakeholders should be embedded in this planning phase through an inclusive approach that prioritizes stakeholder values in the creation of a shared vision and the development of program objectives.

The doing phase (detail in second row of Table 3) consists primarily of implementation and monitoring. Implementation of flows or supplying specific flows to the river via a weir or dam, is informed by the planning phase and targets the objectives agreed upon by stakeholders. While flow implementation may fall to water agencies, within-year changes to flow management can be done with stakeholder involvement. Well-designed monitoring programs support adaptive management within the specific context of the program (Williams, 2011; Gawne et al., 2020). Monitoring programs present a unique opportunity for all stakeholders and the wider community to engage in data collection and learn about the hydro-social and ecological conditions of the river through hands-on experience, fostering multi-directional learning and a sense of environmental stewardship (Conallin et al., 2018a).

The learning phase (detail in third row of Table 3) is often ambiguously described in the literature on adaptive management, but it is acknowledged to be the crucial step for reengagement with the adaptive management cycle (Horne et al., 2017). We emphasize here that learning extends beyond updating the flow-ecology relationships and includes social and organizational learning as well through a focus on multi-loop learning with the inclusive involvement of all participants (Williams and Brown, 2018; McLoughlin et al., 2020). A key aspect of adaptive management is continued iterative learning through repeating the cycle, and a lack of repetition is often blamed for the failure of adaptive management projects (Biggs et al., 2011; Fernández-Giménez et al., 2019). Proper documentation and oversight can improve the chances of adaptive management success, ensuring that lessons learned are communicated to participants and the wider community and enabling learning within and between programs. This documentation and oversight can be carried out through a designated Reflector role within the program (see the fourth row of Table 3) (Webb et al., 2018).