Towards a practice framework for transdisciplinary collaboration in planetary health

Abstract

Non-Technical Summary

Despite growing recognition of the importance of transdisciplinary research in addressing complex sustainability challenges, in practice it has been much hampered by persistent inequities, power disparities, and epistemological disconnect. Planetary health as an emerging field offers a unique lens highlighting the need for knowledge integration across the environment, health, and development (EHD) nexus. Drawing upon extensive analyses, including a meta-analysis of existing transdisciplinary frameworks, a literature review of practices in these fields, and a case study of a planetary health action research project in Indonesia and Fiji, we propose a framework to guide the design and implementation of transdisciplinary research.

Technical Summary

The proposed framework was iteratively designed, starting with existing frameworks, complemented by findings and practice recommendations from a literature review of 36 publications of recent transdisciplinary practices in the EHD fields and an in-depth case study of a planetary health research from Indonesian perspectives. The practice framework focuses on the stakeholder collaboration process, and emphasizes reflexivity and co-learning throughout all research phases: initiation (co-design); implementation (adaptive co-management), and monitoring and refinement (co-monitoring). Foundational considerations for stakeholder engagement could inform process design by reflecting on stakeholder contributions, interactions, integration, and expected outcomes. As suggested by development studies, and implicitly agreed upon but insufficiently elaborated within environment and health, attention to the local context of the research, mapping of power dynamics, and the values of equity and inclusivity are pertinent if research is to produce credible, relevant, and legitimate knowledge and outcomes. A renewed focus on addressing power equities can help ensure stakeholders' perspectives and interests are equally valued and potential solutions are not inadvertently excluded as a legacy of systemic power imbalance. The practice framework is most effectively applied in the initial process co-design, by process initiators and funders assessing proposals for international transdisciplinary research in power-diverse settings or resource-poor contexts.

Social Media Summary

How can researchers across diverse fields collaborate with renewed focus on power inequities to accelerate progress towards the Sustainable Development Goals?

1. Background and context

Transdisciplinary research approaches have been increasingly promoted and practiced in order to co-produce knowledge and urgent action towards the 2030 Sustainable Development Goals (SDGs). In particular, the emerging field of planetary health calls for the expansion of transdisciplinary (TD) knowledge integration and collaboration amongst scientific disciplines, policy, and societal stakeholders for more significant impact and coherence, as systemic disconnect among sectors could potentially delay meaningful and lasting impact (Pongsiri et al., 2017; Zeinali et al., 2020). Planetary health also offers a unique lens highlighting the interconnections amongst the health of natural systems, human health, and socio-economic development, as evidenced by the irreversible harm to human and environmental health caused by socio-economic development systems during the Anthropocene (Whitmee et al., 2015). Previously, such interconnections were conceptualized by human ecologists as biosensitivity (Boyden, 2016, 2004) and echoed by proponents of a broader eco-epidemiological understanding of health (McMichael, 2013). Boyden (2016) also extensively underscores the importance of reducing disparities among all sections of humanity towards intergenerational equity. With the Leave No One Behind (LNOB) imperative at significant risk, collaborative efforts with greater emphasis on addressing inequalities could provide a worthwhile boost towards the SDGs (Browne et al., 2023). Figure 1 shows the intersection of environmental sustainability, public health, and development, previously operationalized as the environment-health-development (EHD) nexus (Wardani et al., 2022).

Figure 1. Biosensitivity and interconnections at the environment-health-development (EHD) nexus.

In recent years, TD research has grown in importance and practice in the respective EHD fields; however, its practice remains undertheorized, underfunded, and underdeveloped (Brown et al., 2015, 2019). Theories, principles, and frameworks for TD research exist, but a number of gaps exist that merit deeper exploration. First, there has been little comparison and synthesis across these diverse yet interrelated fields in search of common ground for collaboration. Meanwhile, increased breadth and scale of collaborations in planetary health could intensify challenges due to deep epistemological, methodological, and cultural differences among distant disciplines, sectors, and development contexts (Ely et al., 2020). Second, much TD research follows an ideal-typical, linear model of ‘linking knowledge to action,’ i.e. producing then applying knowledge, thereby compromising the immediacy and potential for transformative impact of an experimental approach to developing solutions (West et al., 2019). Third, the terms ‘collaboration’ and ‘integration’ have in many cases been used generically, but remain as black boxes without sufficient elaboration of how the process might unfold, especially in power-diverse contexts (Pohl et al., 2021). Finally, in light of widening global inequalities, TD collaboration between High-Income Countries (HIC) and Low- and Middle-Income Countries (LMIC) partners may not sufficiently consider disparities in higher education and research training and historical loss of indigenous knowledge (Jenkins et al., 2018). This is not surprising as most TD theories and frameworks have been conceived through HIC academic perspectives, e.g. Schneider and Buser (2018); Lang et al. (2012); Luederitz et al. (2017).

A brief review of transdisciplinarity across the EHD fields reveals some similarities and potential for complementarity, towards finding a common language which can help achieve common understanding and strategic alignment in addressing common risks and opportunities (Demaio & Rockström, 2015). TD scholarship in environment and health fields largely agree on the importance of local contexts and the value of local knowledge to ensure feasibility, relevance, and legitimacy (Luederitz et al., 2017; Peters et al., 2013). Likewise, planetary health scholars have emphasized the importance of local contexts and unique geographies, histories, economies, politics, and cultures (Capon, 2017). However, there has not been a thorough exploration into the perspectives of development studies and decolonizing and indigenous knowledge scholarship and how these may be useful in understanding diverse contexts and knowledge systems (Odora Hoppers, 2011). These fields have deeply established the fallacy of universalizing HIC theories without addressing the specificities and knowledge systems within LMIC contexts (Alsayyad & Roy, 2004; Chakrabarty, 2000; Roy, 2009, 2016). They have also drawn attention to differences in power and resource realities across HIC and LMIC research contexts (Littman et al., 2021). Indigenous scholars have similarly emphasized principles of holism, interconnectedness, self-determination and mutual respect, which are of major importance if indigenous and local knowledge were to contribute globally and locally relevant solutions (Archibald et al., 2019; Kimmerer, 2013; Kovach, 2009; Smith, 2021).

In bringing together diverse knowledge systems, the challenge remains how collaborations can transcend and equally value different disciplines, sectors, and cultures to produce credible, relevant, and legitimate knowledge (Clark et al., 2016b; Pineo et al., 2021). Planetary health scholars have articulated cross-cutting TD research priorities, involving diverse stakeholders in co-design and implementation, and striving for intergenerational equity (Ebi et al., 2020). However, the practice of TD research in planetary health needs more in-depth empirical exploration into recent practices, especially to understand how the process of integration and collaboration among disciplines and sectors may unfold in HIC-LMIC partnerships. As such, the literature on collaborative urban environmental governance may yield important insights, based on four decades of research and observation of collaborative practices (e.g. Innes & Booher, 2018). In addition, while some understanding of cross-disciplinary, team-based research have developed within public health (Hall et al., 2017), a more explicit theoretical and empirical synthesis of existing frameworks and practices among diverse fields at the EHD nexus is needed to understand the collaborative process itself, including the factors that may enable or constrain collaboration (Stokols, 2006), especially in LMIC settings.

In summary, this research seeks to address the abovementioned theoretical and practical gaps in the state of TD research, specifically the need for (1) explicit synthesis of recent TD practices and theoretical frameworks at the EHD nexus; (2) exploration of the collaborative environmental governance literature which may shed light on the collaborative process; and (3) deeper empirical understanding of the implementation of planetary health TD collaboration, especially from LMIC perspectives. The remainder of this section outlines the eventual aim of this research to develop a practice framework, while Section 2 describes the empirical and theoretical analyses conducted to address the abovementioned gaps, and how their findings intertwine and contribute to the framework development process. The practice framework is presented in detail in Section 3, followed by a commentary on its application (Section 4) and some potential limitations of the framework in its infancy (Section 5).

1.1 Aim of the research

Against this background, this research draws upon extensive theoretical and empirical analyses to conceptualize a practice framework guiding the design and implementation of TD collaboration in planetary health conducted in LMIC settings. Typically, such collaboration would be initiated by HIC funders and researchers who then engage with LMIC academic, government, and community stakeholders. Empirical understanding of how such research is operationalized would provide rich insights into the process of collaboration and how diverse stakeholders and knowledges could come together in meaningful, relevant, and legitimate knowledge co-production.

1.1.1 Defining a practice framework

A practice framework, as drawn from the field of social work, commonly combines formal theoretical knowledge, and knowledge accumulated through repeated practice (Healy, 2014, pp. 226–332) In social work, a practice framework ‘integrates empirical research, practice theories, ethical principles, and experiential knowledge in a compact and convenient format that helps practitioners use the knowledge and principles to inform their everyday work.’ A practice framework is a mapping out of the rationale for practice, and is often devised at a scale where values, theories, and practice are clearly linked. Its purpose is to provide schematic guidance for improving, analyzing, and reforming practice (Stanley, 2016). While most existing TD research frameworks describe the ideal-typical research process, a practice framework considers higher-level principles and institutional and cultural contexts driving action-oriented practice; conversely, it creates synergies in individuals' practices to be formalized into knowledge and institutionalize improvements in practice (Connolly, 2007; Healy, 2014). Another point of difference is explicit consideration of values, inclusion, and equity, as social work is a values-driven field serving disadvantaged communities.

1.1.2 Focus on stakeholder collaboration

The proposed framework aims to bridge theory and practice by focusing on the praxis of collaboration among stakeholders, providing guidance for practice and a tool for restructuring current institutional contexts of knowledge production (Giddens, 1984). As mentioned earlier, existing frameworks do not specifically elaborate on the stakeholder collaboration process e.g. Lang et al. (2012); Luederitz et al. (2017); Newell and Proust (2012); hence our framework aims to address this gap. The framework is intended to be inclusive, bringing together diverse stakeholders, serving as a ‘boundary object’ (Star & Griesemer, 1989) to enable collaboration. To encourage mutual exploration and understanding, the framework is developed through a bricolage of concepts broadly accepted in the EHD fields, an emergent construction fitted and combined with findings from the literature review of recent practice and planetary health priorities. In qualitative research, bricolage is an interpretive piecing together of concepts, methods, and representation fitted to the specifics of a complex situation in an emergent fashion (Denzin & Lincoln, 2011). While some may argue the fallacy of achieving consensus amidst stakeholder diversity (Nogueira et al., 2021), consensus is not expected in all situations but the process would be actively facilitated to orchestrate and navigate diverse contributions, interests, and power dynamics towards mutual respect and benefit (Touati et al., 2019).

1.1.3 Audience

The intended audience of the practice framework would be the stakeholders of research as a process of knowledge production. A broad definition of the term ‘stakeholders’ is used to signify actors with an interest in and are affected by the governance decisions and sustainability challenge being addressed, including those with relevant knowledge and other resources to contribute, and those who benefit from or are adversely impacted (Deverka et al., 2012). Such a diverse stakeholders setting is expected to be of considerable contestation and unequal power dynamics, requiring a high degree of stakeholder engagement (Schneider & Buser, 2018). Nevertheless, such diversity is crucial for the knowledge produced to pass the credibility, relevance, and legitimacy (CRL) criteria (Clark et al., 2016b), and reflects the breadth of stakeholders in planetary health or sustainable development. Ideally, the audience of the framework would be all research stakeholders, i.e. funders in the public and private sectors, academic institutions and researchers, government, civil society organizations, and communities. This inclusive definition adds a transparency and legitimacy towards a common vision by forming a picture of the whole process for all stakeholders to perceive their potential roles.

1.1.4 When and how to use the framework

We define transdisciplinarity as a research approach involving academic and non-academic stakeholders with a view towards societal application, including action research that simultaneously integrate policy, physical or health innovations. Application to solving societal challenges is needed to accelerate progress towards the SDGs, especially action research by iteratively adapting the intervention through experimentation and reflexivity (Wiek et al., 2017). Application of the research could involve developing a novel community infrastructure, or a cross-cutting environment, health, and equity policy (Ebi et al., 2020; ISC, 2023; West et al., 2019).

Hence, the practice framework could be used by stakeholders to provide guidance for reflexivity and co-learning: (1) during the design and inception stage, (2) at multiple points during the process as a continuous monitoring tool, and (3) as a post-mortem evaluation tool to identify refinements and lessons for future TD collaborations, especially in LMICs. Lessons learned could include elements that have worked better than others, and potential reasons for unrealized or unintended outcomes. However, as elaborated in Section 4, the framework would be most effectively applied at the pre-development and co-design phase by process initiators and funders assessing TD research proposals for potential funding.

2. Methodology and framework development approach

The practice framework development took place from 2019 to 2023 and incorporated three areas of extensive analyses, namely a literature review, an empirical case study, and a meta-analysis of frameworks. The findings and practice recommendations from these previous analyses are described in detail in Table 1, which also provides examples of how they were translated into the framework. In summary, the literature review of practice (Section 2.1) identified a leverage point for transformative change in knowledge production systems, in which funding institutions play a pivotal role in influencing project design. Funding institutions are hence identified as one of the primary audiences of the framework. The empirical case study (Section 2.2) yielded two publications: the first (Wardani et al., 2023) highlighted the essential elements of collaboration as experienced from LMIC perspectives, which were included as Foundational Considerations (Section 3.2.1) around stakeholder engagement in the practice framework; while the second (Wardani et al., forthcoming) illuminated the factors and key stakeholder processes occurring at different stages of the collaboration outlined in the framework, e.g. Stakeholder Contexts corresponding with Structural Factors, Stakeholder Contributions with Input Factors, etc. The meta-analysis of frameworks (Section 2.3) identified the research phases (Section 3.1), and Foundational Considerations gleaned from development studies around Local Context, Values & Ethics, and Power Dynamics; while TD research frameworks in environment and health enabled the construction of the stages of collaboration (Structure, Input, Process, Output, Outcomes) (Sections 3.2.2 to 3.2.5) and the crucial role of Reflexivity and Co-Learning (Section 3.3).

Table 1. Summary of findings and practice strategies from previous publications

Findings and practice strategies from previous analyses were incorporated into the development of the practice framework, supplementing the theoretical meta-analysis of existing TD frameworks in the EHD fields. In summary, these previous analyses identified the leverage point for transformative change of knowledge production (i.e. funding institutions), revealed severe constraints relating to understanding local contexts and power dynamics, provided strategic guidance for process design and management, and highlighted pertinent foundational themes of interdependence and diversity of stakeholders' interests. These findings inform our practice frameworks, specifically in formulating the Foundational Considerations, Structure, Input, and Process Factors, and in identifying the primary audience of the framework (i.e. funding institutions and initiators of collaborations).

2.1 Literature review of recent TD practice

A literature review of 36 publications was conducted to draw insights on lessons learned from recent transdisciplinary research practice in the respective EHD fields. These lessons learned were qualitatively analyzed to obtain second-order understanding of the factors enabling and constraining collaboration. These factors were then inductively clustered into structural, relational, and individual factors enabling and constraining collaboration. These findings have been published (Wardani et al., 2022), and described the interplay between factors that enhanced understanding of the collaborative process. Recommendations for practice were identified, emphasizing leverage points for change at the structural level through funding requirements and the project design of such research.

2.2 Empirical case study of TD research in LMIC setting

A unique case study of a contemporary large-scale planetary health research collaboration allowed for an empirical deep-dive. The case study site was the Revitalizing Informal Settlements and their Environments (RISE) program, a planetary health collaboration aiming to implement and assess the environmental, health, and socio-economic benefits of decentralized green infrastructure upgrades of integrative water and sanitation services in a total of 26 informal settlements in Makassar, Indonesia and Suva, Fiji (Brown et al., 2018). Data collection in the case study consisted of 47 semi-structured interviews and six individual reflections in English and Indonesian languages, and two focus groups in Indonesian. Indonesian interviews and reflections were translated into English, and thematic analysis was conducted using NVivo software following an inductive, grounded theory approach (Braun & Clarke, 2006; Charmaz, 2015). The case study yielded two publications; the first on the meaning of collaboration from the perspectives and experiences of Indonesian and LMIC stakeholders (Wardani et al., 2023) and the second on the enabling and constraining factors found in the case study based on experiences of all participants (Wardani et al., forthcoming). Findings from both publications yielded practice strategies that intertwined with and informed the framework development process as elaborated below.

2.3 Meta-analysis of existing TD frameworks

A theoretical meta-analysis was conducted of existing TD frameworks commonly used in the EHD fields. Seeking commonalities and complementary perspectives across the EHD nexus was intended to result in a more comprehensive framework that is more readily accepted in these fields. Throughout the framework analysis and development, ongoing feedback was sought from an academic panel representing the EHD fields, which served as validation within an expanded community of academic practice (Cundill et al., 2015). The rationale for selection of included frameworks are detailed in Table 2 and further elaborated below as part of the iterative framework development process. Table 2 also highlights how these were incorporated into the framework.

Table 2. List of existing TD frameworks in planetary health and EHD fields compiled in the meta-analysis

Starting with values and approaches proposed by planetary health scholarship, existing frameworks in EHD fields were selected and analyzed to inform the practice framework development. Based on comparative analysis and a bricolage of useful concepts and frameworks, we compiled common elements and explored differences to seek resolution, aiming for broad applicability and common ground across disparate yet interrelated EHD fields. Our practice framework highlights the importance of local context, values of equity and inclusivity, and power dynamics, and focuses on the process of stakeholder collaboration, against the background of research phases and co-concepts found in existing frameworks (Section 3.1).

2.4 Practice framework development process

The framework development followed an iterative design process adapted from Romme and Meijer (2020), starting with (1) selection and compilation of frameworks, (2) comparative assessment interweaving inductive theorization from previous analyses; and (3) theoretical validation. Findings from previous studies provided empirical validation of the frameworks, as they were compared and triangulated with the meta-analysis to inform the practice framework development. The iterative framework development process, including the interplays among previous analyses, is detailed below.

1. (1) Compilation and assessment: comparative analysis of frameworks

The framework development process started with the selection of existing TD frameworks from the EHD fields for comparative analysis. Three selection criteria were used: (1) the existing framework is broadly used and accepted in their respective fields; (2) the existing framework is relevant to and aligns with planetary health priorities and imperatives (Ebi et al., 2020); and (3) the frameworks show some similarities and differences complementing each other. Exclusion criteria are frameworks outside the EHD fields, as they are beyond the scope of comparison for this research. Table 2 lists the frameworks selected and detailed justification for their inclusion.

The most significant planetary health imperatives that were not explicitly mentioned in existing environmental sustainability frameworks is that TD collaborations must be inclusive, intergenerationally equitable across HICs and LMICs, and embedded within local contexts and geographies. These led to the inclusion of public health, development studies and Participatory Action Research (PAR) frameworks which emphasize attending to power dynamics to address values of equity, inclusivity, and local context and knowledge (Corburn & Gottlieb, 2005; Littman et al., 2021). Indigenous knowledge and decolonizing methodologies scholarship were also included as they align with these values. Other frameworks that did not mention these values explicitly were not excluded, as they contribute useful complementary understanding about the research process.

From this compilation of frameworks emerged similarities in components, serving as initial building blocks for our framework. The Structure-Process-Outcome format commonly used in public health service delivery (Donabedian, 2003), in particular, resonates with a number of other frameworks in environmental sustainability (Djenontin & Meadow, 2018; Luederitz et al., 2017) and collaborative governance (Innes & Booher, 2018). Hence, our practice framework retains the Structure-Process-Outcome format to enhance familiarity and acceptability across fields. Input and output components were added drawing from the abovementioned environmental sustainability frameworks.

While similarities provided useful starting points for collaboration, differences among frameworks suggested areas of complementarity and tension for further exploration. For example, the TD research framework widely used in sustainability science (Lang et al., 2012), provides a useful ideal-typical model and design principles for a TD research process. The authors identified challenges around lack of integration across knowledge types, organizational structures, and technical and communicative aspects. These challenges reflect structural factors which were not explicitly addressed in that framework, e.g. disciplinary and institutional contexts that predetermine stakeholders' epistemologies, organizational priorities, and communication styles. Therefore, the Structure component was added to our framework, also depicted as Context factors in the co-production framework in environmental management (Djenontin & Meadow, 2018). Advance consideration and improving understanding of these rigid structural factors and how they may shape the process can help facilitate collaboration and address constraints.

Another common element across frameworks in the EHD fields is the time-sequential phases of the research, with one key difference. While some frameworks depict the research process as more linear (Djenontin & Meadow, 2018), most frameworks agree that different phases are interdependent and iterative, e.g. Lang et al. (2012); Luederitz et al. (2017); Stokols et al. (2013); Cornish et al. (2023). Different terms may be used in different frameworks; however, we focus on the overall intentions of each phase which were more alike than different. For example, TD initiatives in public health are described as occurring in four phases, Development, Conceptualization, Implementation, and Translation (Stokols et al., 2013), and we considered additional phases from a more recent iteration to include Reflection & Refinement, and Co-Learning (Pineo et al., 2021). The iterative PAR phases are Observe reality; Reflect on gaps; Plan improvements; Act to test improvements; and Observe outcomes (Crane & Richardson, 2000). Although these reflect a more integrated action-research approach, the phases are similar to those in the public health TD model, for example in encouraging reflection, refinement, and research and observation. A synthesis of these phases, and the collaborative process alongside them, are described in more detail in Section 3.

1. (2) Assessment and inductive theorization: empirical research

Previous studies provided empirical validation for the importance of attention to structural factors. Wardani et al. (2022) found structural factors, including the disciplinary traditions and structures of academia (Becher, 2001), to be the greatest constraints in a collaboration. Beyond academic structures, in LMIC research settings, a lack of understanding of the local socio-cultural, political, economic, geographic and historical context also caused constraints, challenges, and inequitable division of labor due to nuanced communication and cultural differences (Sillitoe, 2018) and power imbalance (Gunasekara, 2020). The importance of local context and knowledge contributed by LMIC stakeholders were also highlighted in Wardani et al. (2023).

Beyond structural factors, the collaborative governance framework suggests underlying preconditions to be considered prior and throughout the collaboration, specifically, notions of interdependence and complementarity among a diversity of stakeholders' interests (Innes & Booher, 2018). These notions are echoed in co-production models in public administration (Ostrom, 1996) and Science and Technology Studies (STS) (Jasanoff, 2004). In public administration, involving the general public as end-users is deemed necessary as the latter contributes relevant knowledge and skills in co-producing public services (Loeffler & Bovaird, 2021). In STS, knowledge is understood to be interdependent and co-evolving jointly with its social and political context (Jasanoff, 2004). Likewise, planetary health scholarship aims to highlight systemic interlinkages (Whitmee et al., 2015), while transnational studies and globalization scholars emphasize interdependence across global development contexts (Sassen, 2016, 2019). Understanding interdependence among diverse stakeholders goes a long way in enabling collaboration, as validated in the empirical case study (Wardani et al., 2023). Section 3 below details further interdependence and other foundational considerations related to stakeholder engagement.

Another significant foundational consideration drawn from the literature review of practice was that of boundary spanning, or communication and relational factors, which were found to be the most enabling (Clark et al., 2016a; Norström et al., 2020; Pohl et al., 2017). Suggested practice strategies include accounting for transaction costs of boundary spanning in project design and management, and researcher training to build individual capacity (Wardani et al., 2022). The empirical case study further cemented that communication and repeated social interactions were required to build trust and relationships, and eventually mutual understanding and acceptance of differences among stakeholders (Wardani et al, forthcoming). Boundaries were found to be the spaces of knowledge integration, therefore creating these spaces were key to bringing together and bridging across different knowledge systems and stakeholders (Wardani et al., 2023). These findings echo the collaborative governance literature, indicating that ‘communicative rationality’ was an ideal condition that could take extensive effort (Innes & Booher, 2018).

1. (3) Theorization & validation

Regardless of the field, existing TD frameworks often mention ‘collaboration’ and ‘integration’ as a generic process, without detailing how the collaborative process might unfold. This practice framework aims to complement this gap using the collaborative governance framework (Innes & Booher, 2018); hence, the cornerstone and focus of this practice framework is on the stakeholder collaboration process. The Structure, Input, Process, Output, and Outcomes Factors relate respectively to stakeholders' contexts, contributions, interactions, integration, and collective benefits. The Foundational Considerations in Section 3.2.1 relate to stakeholder engagement, which include factors that process initiators might reflect on when building the team, designing the collaboration, and during continuous monitoring. Finally, we draw from natural resource management scholarship on an integrative understanding of ‘co-concepts,’ intended to support and enable TD collaboration (Hakkarainen et al., 2022). Each co-concept is aligned with the stakeholder collaboration process and research phases in our practice framework, as elaborated below in Section 3.

3. Towards a practice framework for transdisciplinary collaboration

While our framework selection and assessment drew from the broader EHD fields, the ‘co-concepts’ (co-creation, co-production, co-design, co-learning, and adaptive co-management) were a useful starting point as they represent a contemporary and integrated understanding of ‘collaborative modes of knowledge production and the engagement of non-academic participants’ intended to support TD collaboration. This practice framework refers to collaboration as a transformative co-production process, where ‘a group of actors engage in developing shared understandings and novel ideas of how to intervene in social-ecological systems, requiring deep and protracted stakeholder engagement’ (Galafassi et al., 2018; in Hakkarainen et al., 2022; Shackleton et al., 2019). This is aligned with co-production in highly contested socio-ecological and knowledge systems in sustainability science (Clark et al., 2016a; Norström et al., 2020; Schneider & Buser, 2018). The factors to be considered are provided in the Foundational Considerations, and in the Structure-Input-Process-Output-Outcomes format, which serves as guide posts in developing a TD collaboration.

The practice framework consists of a graphic diagram (Figure 2) and a matrix of reflexive practice questions (Table 3) designed to guide the initial development of and throughout the collaborative process. Figure 2 illustrates how the different elements of the collaboration are integrated, including the research phases, foundational considerations, and collaborative process components and the factors to be considered under each one, as explained in the next section. Table 3 reinforces the framework's focus on stakeholder engagement, with the Foundational Considerations outlined in the left-most column and suggested questions corresponding to each component in the collaborative process. The suggested questions in Table 3 are intended to clarify the points under each component in Figure 2 but relate to broader interpretation centered upon the Foundational Considerations as they correspond to each collaborative component.

Figure 2. A practice framework for transdisciplinary collaboration in planetary health.

Table 3. Matrix of reflexive practice questions

The above matrix of reflexive practice questions is an integral part of the framework and is to be used together with the framework diagram in Figure 2. The questions are centered upon the Foundational Considerations in the left-most column, reinforcing the framework's focus on stakeholder engagement. The header rows here correspond in color with the framework diagram, with reflexivity and co-learning as important throughout in both. The stakeholder collaboration stages (in orange) are aligned with the research phases (in blue). The practice questions are intended to be used for individual and collective reflection both at the beginning and as subsequent stakeholders join in the collaboration.

3.1 Research phases in the practice framework

The practice framework offers a conceptualization of TD research phases aligned with their respective co-concepts (Hakkarainen et al., 2022), namely Phase 1: Predevelopment & Initiation (Co-Design); Phase 2: Implementation (Adaptive Co-Management); and Phase 3: Monitoring & Refinement (Co-Monitoring). These stages of the research process are depicted in blue in Figure 2, arranged in order from the most open-ended (in lighter shades of blue) to more certain (in darker shades of blue). Specifically, the research design should initially be open-ended and adaptable alongside evolving stakeholder priorities, and gain more certainty during implementation and monitoring.

Each phase will be described further below, however it is worth mentioning that Phase 2: Implementation (Adaptive Co-Management) could simultaneously integrate research and action, following an iterative, emergent, experimental approach as practiced in sustainability transitions (van Breda & Swilling, 2018; Wiek et al., 2017). While some argue that transdisciplinarity aims towards usable or actionable knowledge (Clark et al., 2016b) this implies production of knowledge (research) then implementation of intervention (action). Some suggest integrated action research would achieve more immediate impact (West et al., 2019), trialing at a smaller scale initially to reduce risks of unintended negative impact. Lessons learned can inform subsequent iterations of the intervention (van Breda & Swilling, 2018), consistent with experiential, ‘learning-by-doing’ approaches in built environment (Raymond et al., 2017) and PAR in urban health (Barke et al., 2020). Development studies and implementation science in public health further agree on locally embedding, developing, implementing, and refining interventions iteratively, as feasibility, effectiveness, and adoption may not be as expected when moving across HIC-LMIC or LMIC contexts (Reidpath et al., 2022; Roy, 2009).

However, academic research globally remains bounded in discovery research separate from direct application or translation. Efforts to transcend disciplines and involving non-academic stakeholders would be necessary to address constraints due to academic disciplinary structures (OECD, 2020). Conducting research with a view towards application to a real-world problem, whether in physical, policy, or other forms, would be the transformative shift required if our knowledge systems were to accelerate progress towards the SDGs. Regardless of approach, research implementation would occur in Phase 2: Implementation (Adaptive Co-Management), followed by Phase 3: Monitoring & Refinement (Co-Monitoring).

3.1.1 Phase 1: Pre-development & Initiation (Co-design)

Reflections on the Foundational Considerations (Section 3.2.1), Structural Factors (Section 3.2.2), and Input Factors (Section 3.2.3) should ideally take place during Phase 1, to inform an analysis of stakeholders to be engaged in setting the agenda and priorities for the research, their institutional and other contexts, and potential contributions. Through meaningful stakeholder engagement and analysis using the reflexive practice questions in Table 3, this phase should also result in a shared understanding of the local context within which the research should be firmly embedded, the current system and sustainability challenge to be addressed, and a broad, inclusive vision of the future transformed system (see Figure 2).

3.1.2 Phase 2: Implementation (Adaptive Co-Management)

As previously mentioned, this practice framework encourages integrated action and research to accelerate impact through direct intervention and experiential learning, as commonly practiced in ‘Living Labs’ (Wiek et al., 2017). An adaptive, emergent approach is recommended in LMIC or other highly fluid and complex contexts (van Breda & Swilling, 2018), which requires high degree of flexibility and adaptability while maintaining vision alignment (Bos et al., 2013). Open-endedness is also important for innovation (Norström et al., 2020). An integrated approach can allow iterative and incremental reflection on stakeholders' perspectives, interests and contributions through the foundational considerations, and reveal other structural issues that could present challenges later.

In conditions of contested values, a high degree of stakeholder engagement is necessary (Schneider & Buser, 2018). Intensity of stakeholder engagement was found to be a heavy burden due to steep learning curves in developing new relationships and repeated back-and-forth communication required (Wardani et al., forthcoming). However, creating the conditions, space, and time for authentic dialog and mutual understanding is critical to ensure meaningful engagement and equity in agenda- and priority-setting, especially in a power-diverse collaborative process (Littman et al., 2021; Pratt et al., 2016). Integrating research and action stakeholders from the start can influence Phase 1: Predevelopment & Initiation (Co-Design) and Phase 2: Implementation (Adaptive Co-Management), by creating spaces for all contributions to be equally valued through a shared learning agenda exploring perspectives and mutual interdependence (Bos et al., 2013; Pineo et al., 2021). These would likely have an effect on the Process Factors around Stakeholder Integration, as described in Section 3.2.4 below.

3.1.3 Phase 3: Monitoring & Refinement (Co-Monitoring)

During this phase, a monitoring of outputs and outcomes could take place, along with ongoing reflection on previously identified priorities and objectives, to identify possible reasons for unmet or unintended outcomes. Stakeholder outputs and outcomes (Sections 3.2.5 and 3.2.6) could be distinguished from but are no less important than research outputs, such as a conceptual framework, publications, and the policy or practice solution. Although co-monitoring is not one of the co-concepts identified by Hakkarainen et al. (2022), it is included in the practice framework as an important element identified in PAR (Crane & Richardson, 2000) to ensure outcomes and changes in reality are observed and monitored, and the proposed improvement or solution is continually refined through ongoing translation as suggested TD frameworks in public health (Pineo et al., 2021; Stokols et al., 2013). Such a co-monitoring approach, when carried out involving diverse stakeholders, could be useful in combining different knowledge systems, including local ecological knowledge, indigenous knowledge, and scientific knowledge and lead to a more effective, equitable, and inclusive monitoring (Peacock et al., 2020).

3.2 Process of stakeholder collaboration

Against the background of these research phases and ‘co-concepts,’ this practice framework focuses and elaborates on stakeholders as active agents, mediating towards shifts in structural power and institutional change as drawn from the scholarship on collaborative praxis and sustainability transitions (Giddens, 1984; Innes & Booher, 2018; Sovacool & Brisbois, 2019). The factors that may influence the collaborative process, are arranged in order in Figure 2 and Table 3, from most foundational and difficult to change (in darker shades of orange), to most open and uncertain in the future (in lighter shades of orange).

3.2.1 Foundational considerations | stakeholder engagement

Following collaborative governance literature, preconditions underlying collaboration require reflection prior to Phase 1: Predevelopment & Initiation (Co-Design), but these Foundational Considerations (Figure 2 in purple, including all elements therein) should be applied continuously throughout the process, e.g. when additional stakeholders are engaged. These Foundational Considerations provide a set of principles that guide process design on a philosophical level, which are necessary to reach in-depth value and power differences among stakeholders. Along with questions in Table 3, these are considerations for stakeholder engagement especially in relation to the Local Context and Vision Alignment, and can influence the framing of the Sustainability Challenge being addressed. Process initiators would need to maintain openness and flexibility in their vision of the transformed system to adapt particular research questions, methodology, and desired outputs to meet all stakeholders' interests. In LMIC and resource-poor contexts, these considerations are especially pertinent to safeguard against unintended consequences of asymmetrical power dynamics. The Foundational Considerations are: Local Context; Values & Ethics; Power Dynamics; Diversity & Interdependence; Boundary Spanning; and Aligned Vision.

• • Local context: Depicted in darker peach in Figure 2, the socio-cultural, political, economic, geographic and historical context of the research setting is of paramount importance to ensure relevance and legitimacy of the knowledge or solution produced (Capon, 2017). Lack of deep and nuanced understanding of the local context, including administrative and legal barriers and operational conditions could present severe structural constraints, such as risks of failure and increased costs due to uncertainties, ambiguities, and constant changes (Cundill et al., 2018; Pineo et al., 2020; White et al., 2018). LMIC stakeholders also provide specific local and indigenous knowledge systems, research and societal priorities, and knowledge on political and power dynamics (Corburn & Gottlieb, 2005). Power and resource differences especially among HIC and LMIC partners need to be acknowledged so as to avoid marginalization of indigenous knowledge and stakeholders (Littman et al., 2021; van Breda & Swilling, 2018). Moreover, nuances in power dynamics and other complexities within and among LMICs which can influence outcomes (Pratt et al., 2016; Reidpath et al., 2022). LMIC stakeholders' priorities and interests need to be central in the co-design for equitable outcomes and to meet relevant needs (Pratt & Hyder, 2017); hence LMIC stakeholders need to be engaged in research agenda setting, leadership, and decision-making (Clark et al., 2016a; Littman et al., 2021; Peters et al., 2013).

• • Values & ethics: Consistent with planetary health priorities for intergenerational justice and equity (Ebi et al., 2020; Zeinali et al., 2020), TD collaborations must equally value and include relevant actors, sectors, and scales (Pongsiri & Bassi, 2021) through open listening, dialog, and respect for different perspectives (Newell & Proust, 2012). These values of equity and inclusivity are compatible with indigenous and decolonizing principles and methodologies (Chilisa, 2011; Smith, 2021). These values are also central in social work and PAR approaches and in health which recognize the co-production of values and relations in collaborations (Corburn & Gottlieb, 2005; Filipe et al., 2017).

  Intergenerational stewardship values are implicit in environmental sustainability frameworks (Lang et al., 2012; Luederitz et al., 2017), and in producing ‘target knowledge’ about the normatively desirable future (Schneider & Buser, 2018). Likewise, environmental sustainability explicitly recognize personal values as intertwined with authentic leadership (McIntosh & Taylor, 2013) and social learning and innovation processes (Bos et al., 2013; Bos & Brown, 2012). These latter processes are inherently value-laden, requiring higher order or ‘double-loop’ and ‘triple-loop’ learning (Tosey et al., 2012) asking the questions ‘are we doing the right things’ and ‘how do we decide what is right?’, in contrast to single-loop learning which focuses on efficiency and maintaining the status quo by asking ‘are we doing things right’. Social learning, through opening up perceptions of diverse possibilities, thus facilitates a fundamental questioning of the status quo with the potential to transform power relations, ways of knowing, and underlying values.

  It is important to note that sustainability challenges would require collaboration among stakeholders with diverse values; however, as suggested by collaborative governance scholars, values alignment is not always necessary (Forester, 2006). Nonetheless, it may be a challenge to reconcile deep-seated differences, as reflected in epistemological differences among HASS and STEM, quantitative and qualitative methodologies, and disciplinary cultures and values (Becher, 2001). For example, Randomized Controlled Trials (RCTs) as a public health ‘gold standard’ methodology stem from positivist epistemology which views knowledge as objective and value-free (Bryman, 2016, p. 24). Such ontology and axiology contrast with social constructivism or critical epistemologies where individual and collective values are made explicit in the co-construction of knowledge, e.g. in addressing social justice among marginalized groups (Creswell & Poth, 2018, pp. 34–35). In natural resource management, opposing environmental values among conservationist and economic use stakeholders are both necessary for longevity of the solution (Zietsma & Lawrence, 2010). Sustainability science scholars have proposed four perspectives to address differences in values, to enable moving beyond general importance-of-values discussions; reflect on the positionality of one's values; and the contextual operationalization of values (Horcea-Milcu et al., 2019).

• • Power dynamics: Consistent with values of intergenerational equity and commitment to ‘Leave No One Behind’, TD collaborations in LMIC settings need to be cognizant of power dynamics for the proactive avoidance of potential negative impacts including inadvertent erasure of existing knowledge systems (Littman et al., 2021; Reidpath & Allotey, 2019). Mapping and making sense of power dynamics has been suggested to increase the transformative potential of TD research as suggested in natural resource management (Hakkarainen et al., 2022), in sustainability transitions (de Geus et al., 2023), in global health consortia (Pratt & Hyder, 2017), and in PAR approaches (Littman et al., 2021). As posited in collaborative governance and other fields, knowledge is emancipatory in uncovering reified power relations and unacknowledged assumptions (Innes & Booher, 2018), which are critical for shifting power dynamics (Avelino & Wittmayer, 2016). Unequal power dynamics may also arise from persistent hierarchies among academic disciplines, and among HIC academic and local or indigenous knowledge (MacMynowski, 2007; Moser, 2016). Critical self-reflection and collective power reflexivity are important in encouraging equal valuing of diverse epistemologies and knowledge systems (Forester, 2013; Pineo et al., 2021) while offering greater variety of knowledge produced (Hopkins et al., 2020).

  Power is not simply a force one holds over others, but intersubjectively co-constructed within social and relational contexts, as demonstrated in the intersectionalities of power, privilege and disadvantage (Avelino, 2021; Crenshaw, 1989; Severs et al., 2016). Considering intersectionality is one way of mapping power dynamics and inequalities related to personal, professional, and social identities and the myriad discrimination that individuals and groups may face (Hankivsky et al., 2010; Larson et al., 2016). Internal factors, e.g. agency, knowledge and skills, and access to and control over resources and opportunities; and external factors, e.g. laws and policies, and social norms and exclusionary practices mediate stakeholders' ability to shift structural power through everyday practices (Avelino & Wittmayer, 2016; Sovacool & Brisbois, 2019). Explicitly acknowledging stakeholders' values, positionality, and contributions can help ensure meaningful engagement and equal valuing of all interests and contributions.

  Conducting research in LMIC settings is subject to a different set of power and resource conditions than in HICs. Most TD frameworks have emerged from well-resourced, HIC academic institutions in largely formal and legitimized contexts where societal and academic stakeholders can engage on equal footing, e.g. van Breda & Swilling (2018). Transnational and development studies scholars argue such resource and power differences are manifest in deeply uneven geographies of knowledge production, with HIC theories often universalized and flowing unidirectionally towards LMIC settings (Roy, 2009; 2016; Sassen, 2014; 2019; Sillitoe, 2018). These views are echoed in public and global health and justice research (e.g. PAR approaches), which question notions of expertise and thus usually conducted by and for communities (Corburn & Gottlieb, 2005; Littman et al., 2021), and in health where principles of equality of partners and primacy of end-user could safeguard against exploitation and ‘trickle down science’ (Heaton et al., 2016; Reidpath & Allotey, 2019). Our literature review of practice also revealed such unevenness in division of labor, with HIC partners involved in high-level agenda setting and research design, while LMIC partners tend to be involved in data collection (Gunasekara, 2020; Pryor et al., 2009).

• • Diversity and interdependence: Following collaborative and environmental governance scholarship, collaboration necessarily serves diverse and interdependent stakeholders who contribute relevant and complementary knowledge and resources, but also rely on other stakeholders to achieve common and respective interests (Innes & Booher, 2018; Moser, 2016). Diversity and inclusion help ensure relevance and legitimacy, including those who benefit and those potentially harmed, stronger and weaker interests, deal makers and deal breakers, and contrarian and disadvantaged stakeholders (Forester, 2006). Collective reflexivity can feed the potential for creativity and innovation, reciprocity, and discovery of mutual benefits (Wittmayer & Schäpke, 2014).

  Recognizing interdependence was identified as a way to equally value LMIC stakeholders' knowledge and contributions, who generously supported the research despite relative lack of resources (Wardani et al., 2023). Research field workers, community, and government stakeholders contributed local contextual knowledge and resources crucial to intervention design, and facilities, time, and moral support for household surveys and data and sample collection. Interdependence was also found to be a key ingredient underlying stakeholders' motivation to engage. Case study participants' responses to the meaning of collaboration, such as ‘you can't do it alone,’ ‘all stakeholders need each other’ and ‘they won't engage if they don't perceive to get something out of it’ underlined such interdependence, where diverse participants must rely on each other to achieve a common goal (quotes from Wardani et al., 2023). Likewise, interdependence is reflected in the collaborative governance and co-production of public services literature (Innes & Booher, 2018; Ostrom, 1996).

• • Boundary spanning and communication: Upon reflection of diversity in values, power dynamics, epistemologies, and local contexts, the collaboration must consider the resources needed for boundary spanning. Used in the sustainability transitions field and drawn from institutional theory (Zietsma & Lawrence, 2010), boundary spanning include ‘strategies that connect different worlds,’ such as facilitating dialog, negotiation of interests, and reconciliation of potential tensions (Smink et al., 2015). Water sustainability scholars have also proposed the T-shaped concept highlighting three core group of skills to be developed by boundary spanners, including technical and functional understanding of one's own and collaborators' disciplines; organizing and management; and influence leadership (McIntosh & Taylor, 2013). Such efforts can be a time- and resource-intensive endeavor and individuals acting as boundary-spanners tend to be underappreciated, but are necessary to build mutual understanding, acceptance, and trust in relationships (Brown et al., 2019; Clark et al., 2016b; Harris & Lyon, 2013; Innes & Booher, 2018; Moser, 2016; Schneider & Buser, 2018). Deep and meaningful engagement and respect for diverse stakeholders' perspectives and interests, are essential to collaborations (Hakkarainen et al., 2022), and should not be avoided at the expense of efficiency (Littman et al., 2021). These represent transaction costs of facilitating a process from complexity towards mutual understanding, which need to be built into budgets and timelines, especially when working across multiple scales and diverse geographies (Brown et al., 2019; Moser, 2016; Wardani et al., 2023).

  In power-diverse settings, individual and collective reflexivity, itself a form of boundary spanning, is required to encourage mutual learning and accountability, yielding and wielding of power, and using values and vision alignment as a guide (Brown et al., 2019; Hakkarainen et al., 2022; Littman et al., 2021; Tawake et al., 2021). Integration of diverse knowledges require boundary spaces, conducive for social interactions, multi-way communication, relationship building knowledge exchange, and social learning (Bos et al., 2013; Marzano et al., 2006).

• • Aligned vision: Another form of boundary spanning, an alignment of vision redraws an inclusive boundary and drives stakeholders towards a common direction and purpose (Brown et al., 2015; Brown et al., 2019). It is an essential part of collaboration bringing together diverse stakeholders towards ‘a common aim’ alongside ‘different classes of outcomes’ (Wardani et al., 2023). This is reflected in recent TD practice in EHD fields, which identified a shared vision, common ground, group safety and transparency as enabling factors (Black et al., 2018). Forging a shared mission requires visionary leadership in engaging stakeholders in Phase 1: Predevelopment & Initiation (Co-Design), but also facilitative co-leadership in Phase 2) Implementation (Adaptive Co-management) phase to encourage transparent governance and for others lead and develop the process (Wardani et al., forthcoming).

3.2.2 Structural factors | stakeholder contexts

Structural factors may be difficult to change, but may also be assets to the collaboration. With stakeholders as the focus of collaboration, structural factors relate to stakeholders' disciplinary, institutional, and cultural contexts. Not least important are funding institutions at the fulcrum of change with financial and knowledge resources as leverage to institutionalize collaboration (Abson et al., 2017; Wardani et al., 2022). Funders play an important role in the evaluation, design and mechanisms of international TD collaborations; reflexivity is required to explicitly map values and power held by global funders vis-à-vis recipients of funding and communities benefiting from an intervention.

Disciplinary and institutional contexts of stakeholders can affect propensity towards collaboration, epistemological and axiological standpoints, and power dynamics within and across HIC and LMIC settings. Disciplines considered academic purists may view interdisciplinarity as risky, while those more applied and practical or formed at the boundary of two disciplines may be more encouraging of inter- and transdisciplinarity (Becher, 2001; Klein, 1996). Institutions that identify as ‘boundary organizations’ at the interface of science, policy and practice, such as policy think-tanks and applied research centers may be more experienced with facilitating interactions and mutual understanding among diverse epistemological and values perspectives (Gustafsson & Lidskog, 2018; Kivimaa et al., 2019). Within HIC academic settings, institutional ranking and traditional hierarchies (e.g. HASS vs. STEM) may create power dynamics that require reflexivity to identify and address proactively for balanced engagement (MacMynowski, 2007). LMIC settings are not homogenous, with geographical, resource and training disparities across countries and complex power dynamics within each context (Gunasekara, 2020).

Relational contexts can also predetermine power dynamics and value clusters among stakeholders. Process initiators may draw from existing networks and prior collaboration in identifying partners, as a preference over the steep learning curve required for establishing new relationships amidst managing funding uncertainties (Moser, 2016). Prior collaboration may mean sufficient mutual understanding and trust, while little prior knowledge and shared experience requires greater intensity and facilitation of interactions (Harris & Lyon, 2013; Schneider & Buser, 2018). Existing relationships may bear significant power and values clustering that may be a barrier for ‘newer’ partners, while complementarity and interdependence should be considered in balance, i.e. some stakeholders are critical, while others may add value but also complexity (Wardani et al., forthcoming).

3.2.3 Input factors | stakeholder contributions

Consideration of inputs is iterative with that of stakeholder engagement and structural contexts. Additional stakeholders may be required to provide relevant knowledge and as such prompt reflection on foundational considerations and structural contexts. Different stakeholders may also have different understanding of the societal challenge being addressed and different concepts depending on disciplinary and cultural backgrounds. Facilitating exploration of such differences and coming to a shared understanding is important in the initiation stage (Moser, 2016). Further, openness and ability to adapt to changes must be maintained throughout, as stakeholders may contribute differently than initially expected. In the framework diagram (Figure 2), this is reflected in the lighter shading of Phase 1: Predevelopment & Initiation (Co-Design) and in the need for an adaptive approach to co-management (Hakkarainen et al., 2022; Norström et al., 2020).

As found in the empirical case study, collaboration depended on a variety of stakeholder contributions, including tangible and intangible contributions (Wardani et al., 2023). Tangible contributions, such as funding and material resources, were typically contributed by HIC funders and researchers, while intangible ones, such as time, commitment, moral and political support, existing relationships and use of existing facilities for gatherings, sampling events, and laboratories were typically contributed by LMIC academic, government, and community stakeholders. Scientific and technical knowledge and skills tend to come from HIC stakeholders, while LMIC stakeholders contribute locally relevant applications, and contextual knowledge that may not be easily identified, described, and valued. Nonetheless, due to interdependence, without LMIC stakeholders' contributions, by completing surveys, providing biological and environmental samples, and providing community land tenure information, the research could risk implementation failure. Highlighting such interdependence could help equalize power dynamics (Wardani et al., 2023).

3.2.4 Process factors | stakeholder interactions

Strategically incorporating foundational considerations into process design and management can help enable collaboration, as the project level can span structural, relational, and individual factors and offer multi-level opportunities (Wardani et al., 2022). Organizational conditions, for example, can be established to encourage follow-on effects in stakeholder interactions (Process Factors), for example by forging common vision, values, and identity which also develops commitment and ownership (Brown et al., 2019). Likewise, creating a culture of openness and transparency in decision-making, listening and respect for diverse perspectives, and group psychological safety which help ensure equity in negotiating power dynamics (Black et al., 2018; Edmonson, 2019; Littman et al., 2021).

Another example of a structural Process factor that can be established include clear and equitable division of roles and responsibilities, institutional support for dedicated staffing and a base for a Project Management Unit (PMU), clear rules and policies, and information and communication technology (ICT) (Bark et al., 2016). Clear roles and responsibilities were highly cited as enabling, an absence of which creates ambiguity and confusion leading to misunderstanding, tension and conflict (Nix et al., 2018). Equitable division of roles means LMIC stakeholders' involvement is not limited to data collection, but should include representation in governance, agenda -setting, co-design, data analysis and authorship (Gunasekara, 2020; Pratt et al., 2016).

Strong facilitative leadership is likely to enable collaboration (Wardani et al., forthcoming), along with power-reflexive co-governance structure including sectoral and HIC-LMIC representation (Littman et al., 2021). Co-governance helps ensure relevance and legitimacy and lack of engagement of societal stakeholders could compromise dissemination and impact (Heaton et al., 2016). Processes of governance, team building, learning, and innovation must be established, to effectively orchestrate stakeholder interactions. These processes are pivotal in creating the conditions, space and time for authentic dialog, boundary spanning, and build trust, understanding and relationships (Hakkarainen et al., 2022; Harris & Lyon, 2013). Facilitative leadership helps encourage the sharing of power with leaders showing willingness to step back and yield to emerging leaders, allowing them to take greater ownership (Tawake et al., 2021). Facilitation can help with developing consensual theoretical, methodological, and evaluation frameworks, establishing common language for shared understanding (e.g. evolving text for negotiation) (Innes & Booher, 2018).

Facilitation can help build team cohesion by providing semi-formal spaces for social interactions, social learning, and creative cross-fertilization. This helps build trust through familiarity and repetition (Wardani et al., forthcoming), which helps achieve the conditions for authentic dialog where stakeholder interactions are mutually comprehensible, accurate, sincere, and inclusive (Bracken & Oughton, 2006). Social learning can be facilitated by encouraging reflexivity, listening, openness, and valuing of different perspectives (McIntosh & Taylor, 2013) to achieve triple-loop learning and systemic change (Bos et al., 2013). Creative cross-fertilization is necessary for innovation, producing knowledge and solutions through bricolage, borrowing of concepts, and looking at problems through complementary lenses (Klein, 1996).

In addition, facilitation can help stakeholders have equal access to knowledge, and that their knowledge and interests are being equally valued. High complexity, as proxied by degree of contestation and diversity requires careful design, planning and facilitation, and sound knowledge of power dynamics and stakeholder interests for knowledge exchange and production to happen (Schneider & Buser, 2018). Sustaining engagement through shared understanding, trust, and relationships is important due to the inherent uncertainties and ambiguities (Harris & Lyon, 2013). Collaborative governance scholarship note stakeholders may engage initially for instrumental reasons, but over time sustain their motivation for learning complementary viewpoints and personal friendships (Innes & Booher, 2018).

3.2.5 Output factors | stakeholder integration

Through facilitated interactions, the collaboration may start to see intermediate outputs within Phase 2: Implementation (Adaptive Co-Management) and into Phase 3: Monitoring & Refinement (Co-Monitoring). Occurring in conducive boundary spaces, social learning and creative cross-fertilization can bring stakeholders to discover reciprocity and interdependence amongst their interests, and innovative problem-solving beyond initial expectation, which may snowball into greater motivation and cohesion for mutual support and accountability (Innes & Booher, 2018). Repeated social interactions build familiarity, mutual understanding, trust and acceptance, which eventually develop into team cohesion, and social and political capital (Putnam, 2000; Sabatier, 2005). Experiencing the initial uncertainties of the collaborative process together may build stakeholders' adaptive capacity to solve future problems, a sign of transformative triple-loop learning whereby stakeholders recalibrate their perspectives through collective decision-making.

These Output Factors are expected in parallel with specific knowledge outputs which may be the formal ‘deliverables’ of the project, including physical and policy innovation for the societal challenge at hand, a jointly developed conceptual framework, and academic co-publications. Innovative solutions and heuristics serve as boundary objects relevant and legitimate to all stakeholders, held together by ‘communicative rationality’ (Innes & Booher, 2018). For research equity, LMIC stakeholders should be involved in governance, leadership, coordination, and analysis activities, including co-authorship of research outputs (Gunasekara, 2020; Pryor et al., 2009). Building trust and shared heuristics, like achieving conditions for authentic dialog, requires formidable effort, time and resources (Innes & Booher, 2018). However, diversity should not be foregone in the interest of efficiency (Littman et al., 2021), and excluding certain stakeholders may create obstacles later on in the process (Wardani et al., 2023).

3.2.6 Outcomes | stakeholder & system transformations

As TD action research seeks to address a societal challenge, a litmus test for success is sustained improvements in human health, the environment, and social equity – a whole system transformation (Abson et al., 2017) or systems adaptation through innovation (Innes & Booher, 2018; Luederitz et al., 2017). Co-creation of solution-oriented knowledge (Lang et al., 2012) towards nature- and health-supportive development involves a shift in the power dynamics in decision making. A systematic shift in power dynamics is crucial in upending deeply entrenched legacies of colonialism and reification and imposition of ‘universalized’ HIC values, knowledge and cultures to LMIC contexts (Odora Hoppers, 2011; Tawake et al., 2021). Sustainability scholarship and PAR approaches emphasize the reflexive role of HIC researchers and stakeholders in not only recognizing the different thought styles and power dynamics (Christian Pohl et al., 2010), but also in yielding power and centering LMIC interests in such collaborations (Littman et al., 2021). Power reflexivity can help avoid inadvertent exclusion of certain stakeholders' interests and subsequently, the knowledge or resource they contribute. Socio-economic wellbeing and intra- and intergenerational equity are expected (Luederitz et al., 2017).

Sustained benefits in health and environment include socio-ecological integrity, resource maintenance and stewardship (Luederitz et al., 2017; Sabatier, 2005), and a more integrative appreciation of the interdependence between nature and health for all stakeholders (Boyden, 2016). Examples of development mechanism meeting health, environmental, and social objectives include the Green New Deal, prioritizing renewable energy, with positive health impact through improved air quality, reduced carbon emissions, and investments in inclusive upskilling centered on traditionally disadvantaged communities (Calhoun & Fong, 2022).

Knowledge produced collaboratively is hoped to meet the CRL criteria the notion of ‘socially robust knowledge’ (Clark et al., 2016a; Nowotny et al., 2003) – or knowledge transformation. Adoption and sustainability of the intervention are important outcomes to monitor, as suggested by implementation science (Peters et al., 2013) and attests to knowledge CRL. Legitimacy implies that all stakeholders' interests are satisfactorily communicated, listened to and addressed, even if they were not fully met; otherwise, long-term sustainability is compromised. Transformation of the current system towards the desired state needs to abide by the CRL and equity (CRL + E) criteria if we are to avoid decision making by a powerful elite at the disadvantage of certain groups; and such decision making rely on a transformation of praxis described below.

With increased collective capacity for problem solving and social learning, collaborative experience is hoped to bring about transformation of praxis, a change in the system and practices of knowledge production. Following Giddens' structuration theory (1984), our practice framework centers upon stakeholders as agents of change, and knowledge production and innovation practices conceived and embedded within its socio-political context, are more likely to yield transformative solutions and lead to the institutionalization of new norms for sustained systemic transformation. Power reflexivity, and centering historically marginalized and LMIC interests, need to be core to praxis to reverse and avoid further harm caused by colonialism and neoliberalism with enduring negative systemic effects (Littman et al., 2021; Pratt & Hyder, 2017).

3.3 Reflexivity & co-learning

Continuous and iterative reflexivity and co-learning is important in Phase 1: Predevelopment & Initiation (Co-Design) and at multiple touchpoints throughout. This is depicted in a purple band encircling the entire process in Figure 2. Initially during stakeholder engagement, explicitly mapping out values clusters and power dynamics in a stakeholder analysis (Littman et al., 2021), and spending time understanding the local socio-cultural, geographical, political, economic, and historical contexts through lived experience and/or learning the LMIC language, can lead to deeper understanding of potential opportunities and constraints (Gunasekara, 2020; Sillitoe, 2018). In environmental sustainability, reflexivity is a type of social learning supporting TD collaboration through self-positioning, acknowledgement of values and epistemic worldviews, and increasing mutual understanding of a complex natural system (Hakkarainen et al., 2022).

Engaging a diversity of stakeholders at various stages would likely bring some differences that must be reconciled, and values and vision alignment can be useful mitigation strategies (Littman et al., 2021). Due to mismatched institutional logics, stakeholders do not always agree on reasons, goals, and values; but importantly need a shared understanding of the problem to be addressed and direction to be taken, and trust that their shared and interdependent interests can be met through collaboration (Harris & Lyon, 2013; Innes & Booher, 2018; Smink et al., 2015). Developing shared understanding is another form of co-learning; here, academic researchers can provide mutually credible, high-quality systems, target, and transformative knowledge (Schneider & Buser, 2018). Joint agenda setting is important to reach common ground and ensure diverse stakeholders' interests are met, especially LMIC priorities that may not always be central in international research consortia (Pratt & Hyder, 2017). Discussions around target knowledge can help determine the common vision, through mapping out power dynamics and practicing reflexivity for more equitable outcomes (Littman et al., 2021).

During all phases, skilled facilitation is highly recommended across EHD, in planning and coordinating knowledge exchange, learning, and deliberations around key decisions. Facilitators may encourage self-reflexivity around power, diversity, and interdependence, and provide spaces for creative cross-fertilization (Bos et al., 2013; Innes & Booher, 2018). Orchestrating stakeholders' contributions and responsibilities in a fair and equitable way is another important role of a facilitative project manager, creating the conditions for boundary spaces for all stakeholders (Touati et al., 2019; Zietsma & Lawrence, 2010).

Reflexivity and co-learning could occur during Phase 3: Monitoring & Refinement (Co-Monitoring), through a facilitated process to develop an implementation and monitoring framework observing intended and unintended outputs and outcomes. These include formal project deliverables and lessons learned on the process of collaboration and implementation of solution, through reflexive reporting which some funding institutions have begun to adopt, e.g. Most Significant Change (MSC) monitoring and reflection method (Davies & Dart, 2005) which could yield immediate learnings for Phase 2: Implementation (Adaptive Co-Management) and implementation. These learnings and refinements could include technical improvements, additional stakeholders with needed knowledge and skills, or improvement in the process of stakeholder interactions.

4. Application of the practice framework

Without unpacking the process of collaboration and inquiring into stakeholder interactions and dynamics, the design and context of TD collaboration remains a black box and risks failure to deliver the outcomes and aspirations of system transformations. The proposed practice framework aims to shed light on stages of the stakeholder collaboration process and the factors influencing it. Figure 2 outlines how these stages come together, alongside the phases of TD research, and is to be used in tandem with Table 3, a matrix of reflexive practice questions providing specific guidance throughout the cycle of research collaboration. Although each collaboration will differ in specifics, the questions can facilitate the creation of a boundary space for diverse stakeholders to practice individual and collective reflexivity, discuss potential roles and contributions; governance, leadership, and culture; and alignment of aims, objectives, and team expectations (Brown et al., 2015, 2019; Hakkarainen et al., 2022; Wardani et al., 2023).

We propose that the framework would be most effectively applied from at start at or prior to Phase 1: Predevelopment & Initiation (Co-Design) by process initiators and funding institutions assessing TD process design proposals for potential funding. As previously identified, funding institutions are at the fulcrum of transformative shifts with significant leverage to influence process design (Wardani et al., 2022), and hence can use the questions to guide their assessment of TD funding proposals. Early application of the framework at Phase 1: Predevelopment & Initiation (Co-Design) would proactively set the stage for the collaboration. The framework can also be used by stakeholders subsequent joining the process, as reference point for discussing the complex, multi-faceted dimensions of collaboration. This helps create transparency by providing a view of the process as a whole, and thus aligning expectations. While outputs and outcomes will be observed later during Phase 3: Monitoring & Refinement (Co-Monitoring), and may be less apparent initially, the reflexive practice questions in Table 3 can pre-empt important factors to be considered early on.

5. Conclusion

Despite increasing recognition of the importance of TD approaches in producing credible, relevant, and legitimate (CRL) knowledge and solutions for ‘wicked’ and complex sustainability challenges (Clark et al., 2016a; Innes & Booher, 2018; Rittel & Webber, 1973), progress towards the SDGs remains slow and multi-stakeholder processes need more substantial guidance in its design and implementation. As LMICs and the world's poorest will face the greatest threats and disproportionate burden from climate and environmental degradation (Thiery et al., 2021), indigenous peoples and knowledges have proven most effective in conservation and management efforts (Dawson et al., 2021). We propose that additional emphasis on addressing knowledge inequities by practicing reflexivity, consciously mapping power dynamics, and reconfiguring the collaborative process (Forester, 2013; Hakkarainen et al., 2022; Littman et al., 2021; Zeinali et al., 2020).

The research and analyses leading to the development of this practice framework included (1) a theoretical meta-analysis of existing TD frameworks in the EHD fields; (2) a literature review of enabling and constraining factors synthesized from recent practice in these fields (Wardani et al., 2022), and (3) an empirical case study (Wardani et al., 2023). Such theoretical and practical syntheses were helpful in connecting across diverse fields' perspectives; while the in-depth case study offered a unique and instructive context to understanding a broad, large-scale collaboration, and a focus on the early stages of conceptualization and implementation in an LMIC setting. These contribute towards a good starting point for such a framework guiding TD practice, but further empirical research is needed to continue refining the framework and more fully understand such collaborative process, as well as its design and implementation.

The application and audience of this practice framework is envisioned to be in TD collaborations involving HIC and LMIC stakeholders. However, with its particular attention to power dynamics and stakeholder engagement, the framework may also be applicable in resource-poor, power-diverse, and vulnerable settings within HICs and in LMIC-LMIC collaborations. These may include indigenous communities or low-income or culturally and linguistically diverse groups in areas of public health and sustainability in HICs, or in informal settlements or refugee communities in LMICs. Such settings are vulnerable to well-intentioned collaborators bringing resources and associated power from outside the community (Avelino, 2021), and requires power reflexivity to avoid unintended consequences.

Admittedly, a broad application of the framework may raise limitations in meeting the specificity required in practice; however, we hope the framework offers an expansive space to carefully reflect upon a broad diversity of stakeholders and their potential interests and contributions. Moreover, further testing and refinement through reflexive practice over time could increase the relevance and enhance the usefulness of the framework for specific contexts. For the foreseeable future, more prioritization of LMIC perspectives is needed to shift the balance towards knowledge equity.