Bridging Indigenous and Western knowledge-systems in knowledge co-production with Amazonian Indigenous communities: a systematic realist review

ABSTRACT Researchers have sought to understand how Indigenous and Western knowledge systems can be bridged for development research to be (a) more inclusive of local actors and knowledge and (b) more relevant to the problems it aims to address. Research in Amazonian Indigenous contexts remains dominated by a ‘Western’ science struggling to accommodate bottom-up development research. This paper examines how Indigenous and Western knowledge is bridged in knowledge co-production with Amazonian Indigenous communities, and what outcomes such bridging incurs. The paper uses a systematic realist review analysing 22 studies. It finds that the bridging of Indigenous and Western knowledge, in co-production with Amazonian Indigenous communities, was based on partnership between Indigenous and scholarly communities. Emphasising mutual respect, power-sharing, and empowerment of Indigenous communities, this bridging contributes to generation of knowledge that is more legitimate and beneficial for all actors involved in the co-production process. However, the paper also observes a gap between knowledge co-production in theory and practice. Power-relations between Western scientists and Indigenous communities need to be further addressed in development research. More work is needed to enhance the legitimacy of knowledge, in order to further empower Indigenous communities to address local challenges according to their own contexts and needs.


Introduction
In the late 1980s, Edwards (1989) wrote an essay entitled 'The Irrelevance of Development Studies'.He focused on the relationship between research and development, observing that development research, controlled by Western academia, reinforces unequal power-relations and fails to facilitate local empowerment and participation that would enable progress towards development goals.He called for more participatory research, using 'local knowledge to explore local solutions to local problems' (Edwards 1989, 120).Three decades later, the call for such research continues (Chapman and Schott 2020).It is recognized that Western science alone is insufficient and that bottom-up approaches, aiming to bring development closer to people, are necessary (Norström et al. 2020;Sillitoe 2017).Researchers thus sought to understand how Indigenous knowledge-systems (IKS) and Western knowledge-systems (WKS) can be bridged (Abu, Reed, and Jardine 2020; Buell et al. 2020;Chapman and Schott 2020;Sillitoe 2017;Tengö et al. 2014).In this paper, the bridging of IKS and WKS represents a process wherein the two systems interface.Bridging is understood through the lens of epistemological pluralism, heeding the possibility of acknowledging and accommodating different ways of knowing in research (Abu, Reed, and Jardine 2020).
Bridging IKS and WKS can add intrinsic value by enabling multiple knowledge-holders to contribute to research.And it can add instrumental value by improving research processes as well as outcomes (Chilvers 2009;Sillitoe 2017;Tengö et al. 2014;Wyborn et al. 2019).Tengö et al. (2014, 579) observe that 'cross-fertilization among a diversity of knowledge-systems can contribute new evidence and also improve the capacity to interpret conditions, change, responses, and in some cases causal relationships in the dynamics of social-ecological systems'.Bridging can thus enable knowledge from both systems to provide alternative insights: enhancing research, helping address research questions (Tengö et al. 2017), contributing to more place-relevant and legitimised policies (Ban et al. 2018;Korhonen-Kurki et al. 2022;Obermeister 2017;Sillitoe 2017), and supporting solutions to such pressing issues as sustainability (Korhonen-Kurki et al. 2022;Norström et al. 2020;Tengö et al. 2017).Indeed, bridging has been used in contexts of climate change (Fernández-Llamazares et al. 2017;Hill et al. 2020;Makondo and Thomas 2018;Obermeister 2017), biodiversity (Tengö et al. 2017) and natural resource management (Sillitoe 2017).
We can agree, then, that calls for knowledge co-production are being heard.However, critics claim insufficient progress.It has been argued that development research remains prone to Western bias, and that other ways of knowing remain subordinate to WKS (Buell et al. 2020;Mazzocchi 2006;Sutherland 2016).Collyer (2018) deemed research persistently controlled by the global North.Chapman and Schott (2020), plus Virtanen (2021), observed research to be more relevant to, and unevenly advantageous for, non-Indigenous actors.Likewise, many researchers observe that development in Latin America as well as the Amazon were subordinated to Western paradigms and agendas.This has been seen to disadvantage rather than benefit Indigenous communities (Moran 2019; Vásquez-Fernández and Ahenakew pii tai poo taa 2020).In particular, Amazonian Indigenous communities experience policies that contribute to the erosion of their knowledge (Vásquez-Fernández and Ahenakew pii tai poo taa 2020) and researchers who provide insufficient information about research processes and results (Virtanen 2021).These observations illustrate that development research, in Amazonian Indigenous contexts, does not entirely correspond with the call for bottom-up, inclusive development research.
With this in mind, the present paper begins from the premise that knowledge co-production, a collaborative approach that bridges IKS and WKS for mutual benefit, might be an answer (Chapman and Schott 2020;Cooke et al. 2020;Latulippe and Klenk 2020).The springboard from which this paper starts is the following chief question: How are IKS and WKS bridged within knowledge co-production in the Amazonian context?The supplementary question, used to drive the paper's policy relevance, asks: What are the outcomes of this bridging for diverse actors included in the bridging process, in particular Indigenous communities?
In order to answer the chief and supplementary questions, the paper pays consideration to the following subquestions: (1) What approaches were used to bridge IKS and WKS in previous co-productive research with Amazonian Indigenous communities?
(2) How was Indigenous knowledge included in the research and at what stages?(3) What were the research outcomes?
The paper uses a systematic realist review to achieve this.It reviews scholarly and practitioner literature reporting on Amazonian research that seeks to bridge IKS and WKS using co-production modes.The following structure will be adopted.What follows, firstly, is a theoretical framework that defines knowledge-systems, navigates the harmonies and discords between IKS and WKS, and introduces co-production as a valuable bridging model.Second, the paper's methodology is outlined and the results disclosed.Finally, the academic and policy implications are discussed, including some closing thoughts on the value of bridging for current and future development research.
Before the article proceeds further, it would be prudent to offer a definition of development.There are many definitions of development, the conceptualization of which has been regularly subject to change.For our present purposes, the United Nations (1997, 1) definition of development is used, with development being understood as 'a multidimensional undertaking to achieve a higher quality of life for people', which includes 'economic development, social development and environmental protection'.

Knowledge-systems
The purpose of this theoretical framework is to offer an architecture in which the reader can understand how IKS and WKS are bridged in the context of knowledge coproduction.In order to understand this bridging, it is in turn necessary for us to grasp the power relations between knowledge systems and how they impact the bridging of knowledge.Let us first define a knowledge system as '(1) a social network of individuals (2) that exhibits structure and function (3) in the use of specific processes (4) to understand the conditions and causations of Nature' (Varghese and Crawford 2021, 8).A knowledge-system can thus be understood within the following two notions.First, it is a set of propositions or ideas, used to claim truth and knowledge credibility (Díaz et al. 2015;Cornell et al. 2013).Second, it should be framed within the socio-institutional relationships animating the processes of knowledge-production, organization, distribution, communication and application (Varghese and Crawford 2021;Wyborn et al. 2019) that determine what, how and by whom these processes are performed (Fazey et al. 2020).
Knowledge-systems therefore play important roles in society.They influence power-relations (Fazey et al. 2020) and can reinforce inequalities in knowledge-production, marginalising some whilst reinforcing the power of others (Collyer 2018).Power structures can be observed in the context of bridging IKS and WKS, where the latter are often perceived as dominant (Buell et al. 2020;Sutherland 2016;Mazzocchi 2006;Smith 2021).Such dominance often incurs epistemic marginalization of other ways of knowing (Keane, Khupe, and Seehawer 2017), serving to silence (or at least temper) Indigenous knowledge (Radcliffe, Raman, and Parissi 2021).Often-made assumptions about the need for Indigenous knowledge to correspond with Western-scientific standards (Grimm 2017;Löfmarck and Lidskog 2017;Mazzocchi 2018;Tengö et al. 2014) not only represent epistemological conflict but can cause further exploitation of Indigenous knowledgeholders (Latulippe and Klenk 2020).Radcliffe, Raman, and Parissi (2021, 137) argue that one way to address such epistemological conflict is to acknowledge 'the two contrasting epistemologies of science and Indigenous knowledge'.Understanding how IKS and WKS differ from each other is also important because it enables the articulation of the interface between the two systems (Andrews 2020).Some of the discords between IKS and WKS are discussed in what follows.
Indigenous and Western knowledge-systems IKS refer to socially-constructed knowledge, rooted inherently in communities, that provides understanding of the surrounding world and peoples' interactions therewith (Sandy and Bossaller 2017;Sillitoe 2017;Sutherland 2016).Such knowledge relates to what Indigenous people know and is termed, inter alia, Indigenous-, local-and traditional knowledge (Mazzocchi 2006;Sandy and Bossaller 2017;Sutherland 2016).For Sutherland (2016), IKS is characterised by its connection to the environment in which it is developed and used.Embedded in particular place, history and culture (Buell et al. 2020;Sillitoe 2017), it is inherently contextual (Magni 2017;Mazzocchi 2018).
On the other side of the dichotomy, WKS comprises knowledge 'justified by virtue of being acquired within, or endorsed by, formal research or education settings' (van Kerkhoff and Szlezák 2016, 4604).Consequently, it is deemed to be formalized around universities, research institutes and government (Fazey et al. 2020;van Kerkhoff and Szlezák 2016).Compared to IKS, WKS is often deemed universal, even global, in carrying 'broad, universally generic, intellectual ambitions' (Sillitoe 2017, 3).This particular/universal interplay between IKS and WKS has ethical implications, given the ongoing debate on the ethics of globalisation and the potential injustices sustained by attempts to impose universal 'common goods' on particular, and diverse locales (Weaver 2020).Ethical implications can also be observed in WKS' terminology.Whilst terms such as Western science, Western knowledge or Western-scientific knowledge indicate its relation with Western society (Mazzocchi 2018;Sillitoe 2017;Sutherland 2016), the term 'scientific knowledge system' might carry marginalising connotations.It can imply that Indigenous knowledge is non-scientific, and this in itself is open to claims of colonization.In this regard, Atalay (2020) rightly signposts us to the 2017 Indigenous Science Statement for the March for Science, which observes that Western science is only one science amongst many.Indeed in recent years, scholars compellingly discussed Indigenous science in the context of knowledge-bridging (e.g.Alexander et al. 2021;Atalay 2020;Whyte, Brewer II, and Johnson 2016).
With this in mind, Johnson and colleagues (in Alexander et al. 2021, 3) remind us that Indigenous science shares closer affinities with Western science than one might initially think.However, an important difference stems from the discussion of the nature of IKS' and WKS' scientific enquiry.Western science tends to isolate objects from their context and study them in controlled environments (Chapman and Schott 2020;Mazzocchi 2006).It tends to reify subject-object dichotomies (Mazzocchi 2018).'Rational' Western science, as such, has been seen to perpetuate Cartesian objectivism (Gregory 1994, 36;Weaver 2020).Whilst WKS is deemed reductionist and objective (Chapman and Schott 2020;Löfmarck and Lidskog 2017;Mazzocchi 2006;Sillitoe 2017), IKS operates from more holistic, relational perspectives based on human-nature-universe interconnections (Andrews 2020;Keane, Khupe, and Seehawer 2017;Magni 2017;Mazzocchi 2006;Singh and Major 2017;UNESCO 2016).This worldview is embedded in Indigenous practices, norms and beliefs.IKS carries practical value, enabling communities' responses to everyday challenges and representing their strategies for resilience and survival (Grimm 2017;Keats and Evans 2020;Magni 2017;Mazzocchi 2020;Sutherland 2016; UNESCO 2016).Indigenous knowledge stems from empirical observations of communities' daily practices in their immediate environment (Keats and Evans 2020;Löfmarck and Lidskog 2017;Magni 2017;Mazzocchi 2020;Sillitoe 2017).It is based on 'oral histories and cultural memory' (Andrews 2020, 4), resulting from centuries' intergenerational knowledge transmission (Chapman and Schott 2020;Magni 2017;Mazzocchi 2006Mazzocchi , 2020;;Sillitoe 2017).It is taught from birth, using story-telling, observation, direct instructions and rituals (Chapman and Schott 2020;Magni 2017).WKS, meanwhile, is data-driven and based on empirical research that observes, deduces and induces measurable 'truths' (Chapman and Schott 2020;Mazzocchi 2018).Whilst WKS transmits knowledge formally, IKS' knowledge transmission is informal (Sillitoe 2017).Yet it is crucial to avoid implicating us in the already critiqued dichotomising practices: it is essential that the paper avoids binary oversimplifications that polarise the two systems.

Beyond polar comparisons of knowledge-systems
As evident in the above discussion, much of the literature that compares IKS and WKS posits them as binary opposites.However, the above contrasts pose challenges for bridging IKS and WKS (Harvey, Cochrane, and Van Epp 2018).Whilst the above distinctions are important, they raise the risk of misinterpretation.Undue binary distinctions may overshadow the common ground between the systems (Sillitoe 2017;Sutherland 2016).It should be acknowledged that Indigenous knowledge, constructed contextually, is unique to each community; the variation between different communities' knowledge can be significant (Buell et al. 2020;Magni 2017;Makondo and Thomas 2018;Mazzocchi 2020;Sillitoe 2017;Varghese and Crawford 2021).Differences in knowledge also exist within individual communities: knowledge depends on knowledge-holders' age, gender, occupation, caste, class, etc. (Sillitoe 2017).This itself presents challenges: Löfmarck and Lidskog (2017, 24) warn us against ignoring related 'issues of power-relations, legitimacy and gender', which might present risks not only for Indigenous knowledge but also for the bridging process' success (Cooke et al. 2020).Furthermore, differences not only exist within IKS but also within WKS.Sillitoe (2017) and Sutherland (2016), for example, argue that knowledge-production varies between disciplines through already-established research structures and perspectives.
We can thus observe that dichotomising IKS and WKS in such a binary fashion may be unhelpful.IKS and WKS 'may have similar rudiment and content' (Sillitoe 2017, 5).There may even be methodological harmonies between them: after all, knowledge in both systems is acquired through some form of empirical observation, experimentation and repetition (Chapman and Schott 2020;Sillitoe 2017).Löfmarck and Lidskog (2017) remind us that any knowledge, Indigenous or Western, cannot be entirely separated from the context within which it was acquired.Meanwhile, it is important to remind ourselves that in the age of globalization, both systems develop in concert with each other.Both systems are influenced by others' ideas; most actors will combine local, traditional knowledge as well as global (Sillitoe 2017).
Bridging approaches: knowledge co-production / Knowledge-weaving What we just assessed is the distinction between IKS and WKS, and the risks of perpetuating unhelpful binary dichotomies between them.Our task now is to explore how those differences can be bridged in knowledge co-production.An assessment of the relevant literature indicates many approaches to finding synergies between IKS and WKS: there is no single 'one size fits all' approach (Abu, Reed, and Jardine 2020; Buell et al. 2020;Sillitoe 2017).This is partly attributable to disciplinary differences, a multiplicity of actors bringing different perspectives and tools, variation in historical experiences, and substantive differences in knowledge systems (Buell et al. 2020;Sillitoe 2017).Whilst Tengö et al. (2014) identify three 'groups' of bridging approaches -(i) knowledge integration, (ii) cross-fertilization or co-existence, and (iii) co-productionthere has been a notable push recently towards knowledge co-production, or knowledgeweaving, of IKS and WKS (e.g.Alexander et al. 2021;Chapman and Schott 2020;Cooke et al. 2020;Henri et al. 2020;Korhonen-Kurki 2022;Latulippe and Klenk 2020).Knowledge co-production entails the 'contribution of multiple knowledge sources, ways of knowing, and perspectives from different user groups with the goal of co-creating knowledge' (Cooke et al. 2020, 90).This push towards co-production has, in part, resulted from the identification of insufficiencies and ethical shortcomings of knowledge integration and coexistence.Obermeister (2017), for instance, perceived the three approaches as occupying a spectrum from weak to strong.On the weak end, knowledge integration was seen to be often criticised for asymmetric power dynamics leading to Indigenous knowledge rejection, appropriation, exploitative extraction and cultural erosion (Chapman and Schott 2020;Latulippe and Klenk 2020;Löfmarck and Lidskog 2017;Mazzocchi 2018;Tengö et al. 2014).Somewhere between the weak and strong ends of the spectrum, coexistence situates IKS and WKS in close proximity with, and parallel to, one another.At the strong end of the spectrum, knowledge co-production, or weaving, is seen to promise an 'active partaking of diverse knowledge systems at all stages of knowledge-generation' (Obermeister 2017, 81).It seeks more holistic understanding of problems (Chapman and Schott 2020;Harvey, Cochrane, and Van Epp 2018) through partnership in knowledge-generation at all stages: from framing research problems, developing research designs, generating and interpreting data, through to practical knowledge application (Cooke et al. 2020;Harvey, Cochrane, and Van Epp 2018;Klenk et al. 2017;Tengö et al. 2014).
It is argued that co-production contributes to relevant and actionable knowledge that benefits not only scholars but also communities and policy-makers (Cooke et al. 2020;Harvey, Cochrane, and Van Epp 2018;Henri et al. 2020;Latulippe and Klenk 2020;Wyborn et al. 2019).Co-production can also help to rectify power inequalities.Established ways of Western knowledge-generation are challenged (Chapman and Schott 2020;Cooke et al. 2020;Harvey, Cochrane, and Van Epp 2018).Under co-production, Indigenous knowledge and its holders should not be marginalised, but empowered through respectful, equal collaboration (Chapman and Schott 2020; Latulippe and Klenk 2020).
However, co-production is not immune to critique.Some have questioned the approach's ability to truly overcome persistent power-inequalities, especially given that orthodox power-holders can maintain monopoly over collaboration, and can continue to define and frame problems without consulting Indigenous actors (Cooke et al. 2020;Wyborn et al. 2019).Impediments to co-production include, inter alia, differences in values, expectations, needs and interests; costs and funding; lack of global knowledge generalization; 'stakeholder fatigue'; and loss of researchers' independence (Cooke et al. 2020;Harvey, Cochrane, and Van Epp 2018;Korhonen-Kurki et al. 2022;Wyborn et al. 2019).However, some (e.g.Cooke et al. 2020) have shown that the way in which c; o-production is planned and delivered is essential.To help us understand how IKS and WKS are bridged in co-production, this paper identifies two important elements of a successful knowledge co-production delivery: the process and the principles of co-production.
With regard to process, we would do well to follow Tengö et al. (2017) in observing that any successful knowledge-bridging process should undertake five tasks: knowledge mobilization, where knowledge is organized and prepared for sharing; knowledge translation, where knowledge is communicated between actors in a mutually comprehensible way; negotiation of differences and similarities between all involved; synthesis of commonly-accepted knowledge into diverse forms; and application of knowledge within the included systems.The importance of following these steps has been observed elsewhere (Hill et al. 2020;Korhonen-Kurki 2022).
With regard to principles, it is worth joining Norström et al. (2020) in their identification of four tenets of coproduction.Firstly, co-production merits recognition of the importance of co-production being context-based (Norström et al. 2020;Vincent et al. 2021;Zurba et al. 2021), meaning it 'should be considered and situated within the particular social, economic and ecological context' in which it is embedded (Norström et al. 2020, 183).Secondly, the principle of plurality should be valued, because co-production brings together actors with different and diverse knowledge and expertise, and this merits recognition and respect of their different ways of knowing and doing (Norström et al. 2020;Vincent et al. 2021).Thirdly, the principle of goalorientation is encouraged, because whilst actors' agendas will differ, research goals should be shared and defined early on (Norström et al. 2020;Vincent et al. 2021;Zurba et al. 2021).This not only ensures that research is relevant for all actors, but strengthens actors' commitment to joint action, increasing the chances of achieving desirable outcomes (Norström et al. 2020;Vincent et al. 2021).Fourthly, frequent interaction between actors can help build mutual respect (Norström et al. 2020;Vincent et al. 2021;Zurba et al. 2021), resulting in more equitable relationships that allow 'for an increasingly open sharing of ideas, opinions and knowledge' (Vincent et al. 2021, 42).
`Valuable as Norström et al.'s (2020) four principles are, there is a case to be made that a fifth principle is necessary: empowerment.Empowerment is understood as managing the power dynamics that animate knowledgegeneration (Turnhout et al. 2020;Zurba et al. 2021).It has already been stated that power often resides in the domain of Western science (Chapman and Schott 2020;Collyer 2018;Vincent et al. 2021).And it has already been cautioned that power-relations between Indigenous and Western communities are not automatically balanced by co-production (Cooke et al. 2020;Turnhout et al. 2020).Empowerment of marginalised actors, such as Indigenous communities, is thus critical.Co-productive research should rise above assumptions of Western superiority, acknowledging, debating and addressing asymmetric power-relations that might surface (Turnhout et al. 2020;Vincent et al. 2021).
With these principles in mind, the paper will now focus on the Amazonian context.It will try to understand (a) how IKS and WKS are bridged within knowledge coproduction in the Amazonian context, and (b) the outcomes of this bridging for the diverse actors involved, and with particular concern for Indigenous communities.In order to reach the Amazonian context, though, the next section pays brief attention to the methods employed.

Systematic realist review
To address the above questions, this paper follows Castleden et al. (2017) in adopting a systematic realist review: a combination of systematic review and realist review methods.While the former enables a systematic search of relevant literature and ensures rigor, replicability and transparency, the latter helps to interpret why, for whom and under what conditions some processes succeed or fail (Castleden et al. 2017;Stefanelli et al. 2017).For this reason, a systematic realist review is beneficial for the present attempt to examine bridging approaches using knowledge co-production, and to explore the opportunities and challenges contributing to their success or failure.The RAMESES (Realist And Meta-narrative Evidence Syntheses: Evolving Standards) publication standards for realist reviews, developed by Wong et al. (2013), were followed.
However, it is important to acknowledge one's positionality, alongside the fact that the systematic realist review is itself a Western method (Stefanelli et al. 2017).Using this method was necessary because the COVID-19 pandemic prevented the author from conducting fieldwork in the Amazonian region.In order for the project to avoid at least some elements of Western bias, the present author used the research tools developed by Castleden et al. (2017).It should be clarified that Castleden et al. (2017) developed their research tools in collaboration with Indigenous knowledge-holders.Their work specifically addressed the question of knowledge-bridging; they worked with Indigenous knowledge-holders to design their inclusion and exclusion criteria and their reporting tool.It is this very tool and these very criteria that the present author used, in an attempt to ameliorate the risks of Western bias at a time when fieldwork was not possible in this time-limited and resource-limited project.The author acknowledges their Western academic positionality, but reaffirms that this was the most appropriate and feasible way that they could modestly contribute to the decolonization agenda in development research.

Literature search
The initial literature search was conducted in the databases Web of Science, Scopus and Google Scholar, focusing on the terms 'Indigenous people', 'Indigenous knowledge', 'knowledge co-production' and 'participatory research'.The search strategy focused on peerreviewed scholarship and practitioner literature, published in English from January 1980 to July 2021.The word-strings are presented in Appendix 1.All citations, including Authors, Title and Abstract, that were returned from Web of Science and Scopus, were exported in EndNote X9, de-duplicated and screened for relevance.All Google Scholar results were exported to EndNote and screened for relevance, to the point where the whole results page was irrelevant.

Selection and appraisal of literature
Literature was screened in two stages.First, titles and abstracts were scanned for relevance.Second, the full text was assessed.In both stages, the same inclusion/ exclusion protocol, presented in Table 1, was used.The protocol was led by the following questions: (1) Is the research conducted in the Amazon?; (2) Does the study involve Indigenous people?; (3) Is the study describing collective production of knowledge, where Indigenous communities are fully included in the research process?; and (4) Is the study reporting on primary research?.
Literature selection was accompanied by literature appraisal based on the study's relevance and rigor.With regard to relevance, the author assessed whether the study sufficiently described the process of collaboration in knowledge-generation (Pawson et al. 2005;Wong et al. 2013).With regard to rigor, the author assessed the trustworthiness of the collaboration methodology and evidence provided (Pawson et al. 2005;Wong et al. 2013).The data of all included literature is found in the subsection of this paper, 'References of Resources Included in the Systematic Realist Review'.

Data extraction, analysis and synthesis
A reporting tool (see Appendix 2) was used for data extraction from the included literature.The tool

Limitation
Whilst the paper has already acknowledged the author's positionality and the necessity of following Castleden et al. (2017), some brief coverage of the paper's methodological limitations is also warranted.
Compared to traditional systematic review, the realist approach cannot be standardised and is not replicable in the same way, due to 'explicitness and reflexivity on the part of the person/s undertaking the review' (Pawson et al. 2005, 32).In order to ensure greater chances for replicability and transparency, the paper has sought to elucidate, above, the steps that it took to assess and safeguard the suitability of the data sample it used.The author would also wish to stress that any conclusions, drawn from this review, are based on the analysis of a limited number of resources, as only English-language resources were included and potentially valuable literature might have been missed out (Figure 1).

Document characteristics
22 resources were analysed using the above methods.
The earliest was published in 1998; the latest 2021.Figure 2 illustrates that the%age of published resources is the largest in the year bracket after 2015.Almost a third (32%) of resources were published in 2020 (n = 3) and 2021 (n = 4).Regarding resource type, Figure 3 shows that journal articles constitute the largest proportion of resources (n = 18).They are complemented by notably fewer doctoral theses (n = 3) and one conference paper.All resources' first authors were identified as researchers; 64% were affiliated to institutions outside their studies' countries, while only 36% were affiliated to institutions in their studies' countries (Figure 4).A significant proportion of first authors did not self-identify as Indigenous (n = 19, 86%) within their published manuscript.Compared to first authors, the proportion of co-authors not self-identifying as Indigenous (n = 10, 45%) is smaller.Thus there is a larger proportion of Indigenous co-authors or co-authors representing Indigenous communities, organizations and authorities (n = 7, 32%) than first authors (n = 3, 14%) (Figure 5).

Context
Study's country and field.Figure 6 shows that the included studies represent all but one country in the transboundary Amazon.The largest proportion of studies were conducted among Indigenous communities in the Peruvian Amazon (n = 9, 41%), followed by the Brazilian Amazon (n = 7, 32%).Communities from the Amazon in Suriname and Guyana were covered only by one study in each area.The studies focused on various fields, with a higher proportion focusing on ecology (n = 6, 27%), health and research (n = 4, 18%), and culture (n = 3, 14%).The higher proportion in these fields could illustrate Indigenous peoples' realities and needs as well as stronger academic interests in specific areas of Indigenous knowledge (Figure 7).
Research actors and relationships.Aligning with the inclusion criteria, Indigenous communities or their organizations were identified as being involved in all included studies.Other involved actors were government (n = 4, 18%), international organizations and institutions addressing Indigenous issues (n = 3, 14%), local and national non-governmental organizations (n = 8, 36%) and other stakeholders (n = 6, 27%) such as entrepreneurs (Figure 8).
Inclusion criteria focused not only on inclusion of Indigenous individuals, communities or organizations in research, but also on the collaboration between Indigenous people and researchers.In all resources it was possible to identify some form of partnership.However, the terminology was not always one of overt partnership.Figure 9 represents three main groups of language indicating more or less collaborative relationships.More collaborative terms used for Indigenous partners were found in the largest proportion of resources (n = 9, 41%), including terms such as 'partners', 'collaborators' and 'allies'.23% of resources (n = 5) used less collaborative terms like 'participants'.Other neutral terms included 'students', 'researchers', 'stakeholders', 'internal agents' or community names.
Moreover, to understand the context of bridging and co-production, this study sought to identify who was involved as Indigenous partners.Whilst some studies identified multiple partners, a significant proportion of resources reported partnerships with Indigenous communities or members (n = 14, 64%), followed by Indigenous leaders (n = 9, 41%).Other partners were Indigenous   researchers (n = 7, 32%), Indigenous organizations (n = 5, 23%) and Indigenous representatives, identified as partners in only one study (Figure 10).
Objectives and purpose of study.The included studies' objectives can be categorised in three general themes: b) Bridging-orientated objectives The objectives, included in this category, focus on the assessment and description of approaches that aim to increase Indigenous participation, bridging gaps between Indigenous and scholarly communities, or the inclusion of IKS and WKS in research.These objectives present the aim of discussing applied methods and their outcomes, benefits, costs and challenges.Bridging-related terminology, used in these approaches, includes bridging differences among actors (Amazonas et al. 2019), worldview bridging and knowledge inclusion (Ayaviri Matuk van Maurik 2020) as well as partnership and collaboration (Vallejos 2014).

a) Indigenous-orientated objectives
The purposes, included in this category, seek to contribute to actions that would benefit Indigenous communities.The research in this category also seeks to understand the Indigenous perspective of the problem in order to support local solutions (Bussalleu et al. 2021;Calegare et al. 2020;Santafe-Troncoso and Loring 2021;Sirén 2004;Vásquez-Fernández et al. 2021).The resources aim to strengthen Indigenous governance (Fisher et al. 2020;Schmidt et al. 2021;Vásquez-Fernández et al. 2021), build community capacity (Fisher et al. 2020), address local issues and develop interventions that reflect communities' realities, empower communities and enable cultural preservation (Athayde et al. 2017;de Toledo, Giatti, and Pelicioni 2012;Sirén 2004).

c) Less Indigenous-orientated objectives
In contrast with the previous two categories, the objectives here are based on WKS rather than specifically IKS.Anticona (2012) and Anticona et al. (2013) aim to examine the factors of lead exposure and pathways for elevation of blood lead levels; Bussalleu et al. (2020) aim to examine the presence of bacteria in liquids consumed by Indigenous communities; and Sirén (2004) examines the environmental impact of natural resource exploitation.However, while these objectives can be perceived as originating from WKS, in all cases they accompany objectives benefiting  Indigenous communities, as discussed in the previous theme.
Indigenous rights.The included resources were assessed for mentioning relevant treaties or Indigenous rights.While 32% of resources did not mention Indigenous rights at all, 68% discussed Indigenous land-, cultural-, social-and territorial governance rights.Some such rights were directly linked to the conducted studies in three ways.First, the struggle for recognition and respect of Indigenous rights represents a motivation for Indigenous communities to initiate or consent to collaborative research (de Toledo, Giatti, and Pelicioni 2012).Second, Indigenous rights represent a safeguard for Indigenous communities to be involved in projects and research (Aikman 1998;Anticona 2012;Espinosa and Duque 2018).Third, the collaborative research outcomes support Indigenous peoples' claims for Indigenous rights (de Toledo, Giatti, and Pelicioni 2012; Vásquez-Fernández et al. 2018;Vásquez-Fernández et al. 2021).

Mechanisms
Methodology and methods.The paper will now present the methodologies and methods that researchers used in the analysed resources (Figure 11).
Participatory research was the most common methodology used in the resources.91% of authors used participatory research in different forms, including community-based participatory research, 'ecohealth' research and action research.32% of resources combined two or more methodologies.91% of resources used more than one method.The most common method was the interview (73%), followed by participatory observation (32%), surveys (27%) and focus groups (27%).Oral history and storytelling appeared in 23% of papers; participatory mapping in 18%.Other methods from science as well as IKS were discussed, for instance, remote sensing, microbiological study, sharing circles, rich pictures, participatory video and photography.
While the variety of above methods might indicate that the bridging of IKS and WKS requires multiple approaches to address the needs of diverse knowledge-systems, only 32% of studies distinguish between Indigenous and Western (scientific) methodology.However, Indigenous methodology is understood, in the resources, as a methodology emanating from Indigenous epistemologies and their holistic worldviews (Ayaviri Matuk van Maurik 2020; Vásquez-Fernández et al. 2018).It is initiated and designed by Indigenous researchers, affording culturally safe and respectful research which helps empower Indigenous communities (Aikman 1998;Vásquez-Fernández et al. 2018).Terminology, used in relation to Indigenous methodology, includes Indigenous, originary, indigenist, indigenising and decolonising methodology (Athayde et al. 2017;Vásquez-Fernández et al. 2018;Vásquez-Fernández et al. 2021).The most common characteristic of such methodology was respect for Indigenous oral tradition, preferring such methods as storytelling and oral history (Athayde et al. 2017; Ayaviri Matuk van Maurik 2020; de Toledo, Giatti, and Pelicioni 2012; Green, Green, and Neves 2003;Vásquez-Fernández et al. 2021).Other methods recognised were empirical   Indigenous knowledge.Analysis of the studies showed that differentiation between Indigenous and Western methodology is rare, occurring in less than a third of cases.However, authors of 50% of the papers distinguished Indigenous from Western scientific knowledge.These studies often describe Indigenous knowledge as a complex set of values, practices, technologies and innovation, which is socially regulated and presents practical value to communities (Angarita-Baéz et al. 2017;Athayde et al. 2017;Santafe-Troncoso and Loring 2021;Vásquez-Fernández et al. 2021).Another characteristic, frequently emphasised by these papers, is the holistic nature of Indigenous knowledge, which does not separate humankind from nature, and which includes mythical as well as spiritual elements (Aikman 1998;de Toledo, Giatti, and    Collaboration processes and knowledge co-production approaches.This review sought to analyse knowledge-bridging processes in co-production.It focused on collaboration, relationships, approaches and limitations, each of which is discussed below.

a) Defining collaboration processes
The resources defined and discussed collaboration in knowledge-production through six main themes.One of these described collaboration through partnerships, wherein all actors through dialogue have opportunities to equally contribute knowledge, share decisionmaking, and share responsibilities (Anticona et al. 2013;Athayde et al. 2017; Ayaviri Matuk van Maurik 2020; Bussalleu et al. 2021).
The other two frequently mentioned aspects were trans-disciplinarity and participation.The former emphasises the need to respect and integrate different worldviews, knowledge types and research methods (Angarita-Baéz et al. 2017;Anticona et al. 2013;Athayde et al. 2017;Ayaviri Matuk van Maurik 2020;Calegare et al. 2020).The latter discusses Indigenous participation with other actors during the different stages of research (initiation, problem-identification, design, knowledge-development, solution-development and outcome-validation).Not all resources discussed all stages, but most discussed the importance for meaningful participation beyond data extraction.Other themes, apparent in the definition of collaborative processes, focused on solving common challenges (Anticona et al. 2013;Schmidt et al. 2021); mutual benefits (Berardi et al. 2014;Vallejos 2014); and collaborative learning (Athayde et al. 2017;Sirén 2004).

b) Bridging IKS and WKS
An important element of knowledge-bridging is the application of IKS and WKS to research.The resources were analysed for the way they combined Indigenous and Western-scientific methods and ways of knowing.73% of studies explicitly described at least some aspect of knowledge-bridging.For instance, the papers described IKS and WKS being balanced and used complementarily (Aikman 1998;Schmidt et al. 2021).Moreover, the importance of knowledge-sharing between Indigenous peoples and academics was emphasised (Sirén 2004;Vásquez-Fernández et al. 2018).The studies used classifications from IKS and WKS (Schmidt et al. 2021;Vallejos 2014), and integrated this in research findings, interventions and recommendations (Calegare et al. 2020;de Toledo, Giatti, and Pelicioni 2012).
Papers frequently reported using various methods to obtain Indigenous and Western knowledge (Angarita-Baéz et al. 2017;Anticona et al. 2013).They emphasised that Indigenous as well as Western worldviews need inclusion at all stages of research (theory, research question, methodology) and that checking throughout research is needed to ensure that these views are respected (Fisher et al. 2020;Green, Green, and Neves 2003).However, other studies disclosed that different research stages focused on either IKS and WKS, depending on needs (de Toledo, Giatti, and Pelicioni 2012; Espinosa and Duque 2018).For instance, IKS might be applied at data collection and results validation, whilst WKS would be used in data analysis and dissemination.The latter was often linked to preparation of reports for authorities and policy-makers.

c) Relationships
This paper sought to analyse co-production relationships as a discrete element of knowledge co-production mechanisms.First, it is observed that not all papers discussed how the relationship between researchers and Indigenous knowledge-holders was developed.The common terminology used in the resources was trust (n = 8), respect (n = 4), and dialogue (n = 3).In the majority of resources, the relationship was based on partnership and collaboration (Bussalleu et al. 2021;Vallejos 2014;Vásquez-Fernández et al. 2021).The relationship between Indigenous actors and academics was either based on pre-established relationships (Angarita-Baéz et al. 2017) or presented as part of the knowledge-production process (Bussalleu et al. 2021;Fisher et al. 2020).In the resources, relationships also appear as limitations and challenges, especially linked to power-relations and Indigenous engagement.These aspects are, however, discussed in the below section discussing limitations identified by authors.

d) Inclusion of Indigenous individuals, communities and organizations in the research process
The paper also sought to examine how researchers ensure inclusion of Indigenous people in research.The resources described this mechanism mainly through involvement of Indigenous individuals, communities or organizations at different stages of the knowledge-production process.Most resources (64%) discussed Indigenous inclusion at three or four stages within the same study; only one resource reported on Indigenous inclusion at all stages of the research process.
36% of resources reported on the study being initiated by Indigenous communities.In some cases, Indigenous communities invited academics to conduct research in their territories (Aikman 1998;Calegare et al. 2020;Vásquez-Fernández et al. 2021).Reported cases of Indigenous involvement increased at the stage of designing research questions (45%) and research design (59%).While the former focus on establishing common understandings of problems, in order to reflect Indigenous as well as Western science needs (Ayaviri Matuk van Maurik 2020; Fisher et al. 2020), the latter focus on designs that would be appropriate to Indigenous communities (Athayde et al. 2017;Vásquez-Fernández et al. 2021).One element of this stage was the use of suitable methods, reflecting Indigenous epistemologies and enabling Indigenous strategies (Amazonas et al. 2019;Angarita-Baéz et al. 2017).The second was respect for local customs, protocols and language (Bussalleu et al. 2021;Vásquez-Fernández et al. 2021).
Further, the resources reported on joint research implementation (45%).This stage, especially, is reported for collaboration through local interlocutors, individual members, representatives or leaders who represent links between communities and researchers, and thus need good understandings of Indigenous as well as Western worldviews (Aikman 1998;Schmidt et al. 2021).
The following stage of data collection reported the involvement of Indigenous communities in 59% of studies.Contrarily, collaboration in the next stage of data analysis is included in only 27% of papers.Likewise, joint dissemination of results appears in only 23% of resources.However, Indigenous communities are more often involved in data co-validation, reported in 50% of studies.
Other aspects of Indigenous inclusion include financial and technical support for capacity-building (Athayde et al. 2017;Fisher et al. 2020); regular visits to, or living with, communities, to develop better understandings of cultural differences (Calegare et al. 2020; Santafe-Troncoso and Loring 2021); communityresearcher dialogue (de Toledo, Giatti, and Pelicioni 2012;Vásquez-Fernández et al. 2021); and the obtaining of community consent and ethical approval.

e) Limitations and challenges
The most common themes, in the studies' coverage of limitations and challenges, were epistemological differences between IKS and WKS; the issue of representation in research; Indigenous-scholarly relationships; and lack of time, funding and geographical accessibility.I will address these points in turn.
Second, with reference to representation, authors were concerned with limited representations of whole communities.Knowledge co-production sometimes involved only leaders and representatives, influencing what and whose knowledge was included (Anticona et al. 2013, Sirén 2004).Community power-relations enabled some individuals to disproportionately influence contributions, and this is especially evident in gender contexts (Ramirez-Gomez, Brown, and Fat 2013).Further, representation can be limited by the choice of methods that might exclude some community members (Amazonas et al. 2019;Sirén 2004).
Third, with regard to academic-Indigenous relationships, two elements need addressing.First, many resources discussed the challenge of unequal powerrelations, where researchers with influential funding and affiliations hold more decision-making power (Ayaviri Matuk van Maurik 2020; Vásquez-Fernández et al. 2021).Second, authors often observed mistrust between Indigenous communities and academics (Anticona et al. 2013;Vásquez-Fernández et al. 2018).Some found it challenging to distinguish private from research relationships (Vallejos 2014).Sirén (2004) observed that concurrently being an 'insider' and 'outsider' presents limitations.
Fourth, another common challenge of co-production was insufficient funding and time, plus geographical inaccessibility (Ayaviri Matuk van Maurik 2020; Ramirez-Gomez, Brown, and Fat 2013).These elements present challenges for researchers and community members.Calegare et al. (2020) observe that collaborative research requires significant time commitments from Indigenous communities.Funding is often provided only for limited durations, hindering projects' sustainability (Calegare et al. 2020).
On top of the above most commonly reported challenges, the others, reported less often, include hidden and changing agendas that affect research and outcomes (Anticona et al. 2013;Anticona 2012); changing leadership and membership of long-term research, impacting participation and commitment (Anticona et al. 2013); and unevenly shared benefits within communities (Green, Green, and Neves 2003).

Outcomes
The main academic outcomes arising from the resources were better understandings of participatory methods and knowledge co-production, including how to involve Indigenous communities more effectively (Anticona et al. 2013; Ayaviri Matuk van Maurik 2020); new knowledge or understandings of the topics under investigation (Athayde et al. 2017;Ramirez-Gomez, Brown, and Fat 2013); enhanced understandings of IKS (Bussalleu et al. 2021; Santafe-Troncoso and Loring 2021); and knowledge development for policy-makers (Anticona 2012;Fisher et al. 2020).
Most resources discussed outcomes that could benefit Indigenous communities.The most common outcome discussed was empowerment and support for self-determination, including self-governance and autonomy (Aikman 1998;Ayaviri Matuk van Maurik 2020).Collaborative research is said to increase the validity of Indigenous knowledge among authorities and policy-makers (Angarita-Baéz et al. 2017;Ramirez-Gomez, Brown, and Fat 2013).This improves Indigenous opportunities to claim rights (de Toledo, Giatti, and Pelicioni 2012;Ramirez-Gomez, Brown, and Fat 2013).The resources also produced recommendations for interventions to benefit Indigenous communities (Bussalleu et al. 2021;Vásquez-Fernández et al. 2021).Another benefit discussed was capacity-building, enabling Indigenous communities to address local issues (Aikman 1998;Schmidt et al. 2021), negotiate with authorities (Espinosa and Duque 2018;Calegare et al. 2020), and conduct and participate in research (Anticona et al. 2013;Espinosa and Duque 2018).The resources also reported on physical knowledge products being generated, including dictionaries, documentaries and maps, providing educational or tourist resources for Indigenous communities (Ayaviri Matuk van Maurik 2020; Vallejos 2014).
Resources also issued recommendations for future Indigenous-academic collaboration.The recommendations focused on epistemological differences.The resources emphasised the need for acknowledging similarities and differences between IKS and WKS that would enable mutual understandings.Moreover, it was evident that researchers need to better understand, and pay more attention to, Indigenous worldviews and avoid sequestration of Indigenous knowledge (Amazonas et al. 2019;Athayde et al. 2017).The papers expressed the need for methods going beyond WKS methodology and preventing exclusion of members, especially women (Athayde et al. 2017;Vallejos 2014).Research was said to need ethical sensitivity, meaningful community engagement, and respect (Ayaviri Matuk van Maurik 2020; Calegare et al. 2020).It was said to require better coordination between actors, reflection on actors' agendas, and identification of common goals (Anticona 2012;Ayaviri Matuk van Maurik 2020).Angarita-Baéz et al. (2017) emphasised the need for more time and funding to address such challenges.

Discussion
Now that the paper has outlined its chief results, it would be prudent to discuss their implications.The above findings resulted from analysing a sample of 22 scholarly and practitioner resources, in an attempt to identify knowledge-bridging approaches in research conducted with Amazonian Indigenous communities.The work was concerned with knowledge co-production processes and outcomes.It sought to understand the drivers of their (in)effectiveness, and their implications for the actors involved in the research sample, particularly Indigenous communities.The sample represented studies from various disciplines and all but one country of the transboundary Amazon.
The review shows that, despite more than three decades of calls for the bottom-up approach to development, knowledge co-production with Amazonian Indigenous communities became more prominent in development research only in recent years.And still, only a third of all resources published in 2020/21 indicate the strengthened calls for research decolonization, producing knowledge that better reflects Indigenous realities (Smith 2021).
The majority of scholars in the sample self-identified as non-Indigenous, with affiliation outside Indigenous contexts.This aligns with similar findings from other regions (Alexander et al. 2021;Castleden et al. 2017;Stefanelli et al. 2017), confirming those critiques that deem research on Indigenous issues to be dominated by non-Indigenous actors (Castleden et al. 2017;Collyer 2018;Smith 2021;Stefanelli et al. 2017).The troubling question of asymmetric power-relations is also raised in the proportion of Indigenous co-authors being larger than Indigenous first authors.But some hope should be found in Turnhout et al.'s (2020) observation that, when dominant research actors step back, the redefinition of roles is possible and knowledge coproduction strengthened.
The review found, in chief, that the umbrella co-production approach is the collaborative partnership between researchers and Indigenous actors.This is unsurprising, given that partnership is a central mechanism for bridging IKS and WKS in knowledge co-production (Cooke et al. 2020;Stefanelli et al. 2017).Indigenous participation, in the sample, was frequently enabled through participatory research, which was reported as a methodology in 91% of resources.A close relation between knowledge co-production and participatory research therefore exists: both advocate a 'more interactive arrangement between academic and non-academic actors' such as Indigenous peoples (Norström et al. 2020, 183).
Further, the results illustrate how such partnership interaction follows the aforementioned principles of knowledge co-production.First, the studies mostly discussed the principle of empowerment in terms of shared decision-power and responsibilities throughout the research process.Second, studies identified the need for pluralism and the importance of inclusion of different worldviews and methodologies in research.Third, they emphasised the principle of frequent interaction for collaborative relations.Last, the principles of context dependency and goal-orientation are discussed in terms of defining shared challenges and working towards mutual benefits.
But whilst acknowledging these principles beckons potential for co-production, the gap between discussion and realisation of principles varies within the sample.This finding is broadly coherent with Castleden et al. (2017), who note that the principles of co-production provide an effective framework for collaborative and empowering research, but that the collaborative approaches are still on the periphery of, in their context, Canadian research.Similar challenges are identified in this review.
First, context dependency as a co-production principle can be observed as part of all other principles (pluralism, empowerment, goal-orientation and frequent interaction).Second, and in relation to pluralism, research often sought to include Indigenous worldviews DEVELOPMENT STUDIES RESEARCH and methodologies.But epistemological differences were often cast as limits to co-production: as hindrances to communication between actors.Only half the research teams discussed epistemological and ontological differences, and even fewer acknowledged methodological differences.This aligns with findings from Latin America and the Caribbean (Iwama et al. 2021) plus Canada (Castleden et al. 2017).In contrast with these findings, though, many studies in the present sample, which did discuss epistemological and methodological differences, indeed valued Indigenous worldviews, including their cultural and spiritual elements, and sought to incorporate them in their reports.This demonstrates that acknowledging differences between knowledge-systems contributes to the disruption of hegemonic power-relations, enabling IKS to be perceived as equally important (Iwama et al. 2021).And whilst most resources' methods originated from WKS, mixed methods were often used to obtain knowledge from both IKS and WKS.More than half the studies described the application of knowledge from both systems.
Second, and with regard to the principle of empowerment, the sample described steps that were taken to establish collaborative approaches to research where decision-power and responsibilities are shared between academic and Indigenous communities.As this paper discussed above, and as evidenced in Figures 12 and 13, all of the resources used in this literature review were seen to include Indigenous participants in multiple stages of those studies' research processes.Whilst it is beyond the scope of this present paper to evaluate the exact degree to which the resources' authors took due account of their Indigenous collaborators' contributions, what we can at least observe is that knowledge-bridging appears to have been more frequent at the stages of research design and data collection, followed by results co-validation, design of research questions and research implementation.This trend indicates a form of knowledge-generation moving beyond extraction of Indigenous knowledge, which is a characteristic of the less-empowering integration approach (Latulippe and Klenk 2020;Tengö et al. 2014).Nevertheless, only one study reported Indigenous inclusion at all stages of research.This in itself demonstrates the risk of persistent colonising practices in Amazonian development research.What is also evident is a less frequent inclusion of Indigenous communities in data analysis and dissemination.This again serves to validate Castleden et al.'s (2017) observations.Likewise, the sample frequently reported on the need for WKS validation, when knowledge is presented to authorities and policy-makers.And again, this reinforces the presence of asymmetric power-relations.Powerrelations within communities are often ignored in research with Indigenous communities (Löfmarck and Lidskog 2017), which can further challenge the process of co-production (Cooke et al. 2020;Vincent et al. 2021).This review found that partnership can often favor privileged actors such as Indigenous leaders, Indigenous organizations' delegates and Indigenous researchers, rather than including all community members.This is problematic, as it can determine whose voices are heard and what/whose knowledge is included.Empowerment can be selective, limited and limiting.Already marginal voices may be 'silenced' further, and prevalent 'truths' and dispositions reinforced and perpetuated by 'louder' voices that are more often heard.The risk is that biased, unrepresentative knowledge permeates the research and decisionmaking process, impeding less 'amplified' but nevertheless crucial perspectives (Vincent et al. 2021).
Third, further gaps open between theory and practice concerning the goal-orientation principle.In the sample, this principle links to collaboration within the development of research questions, so that research goals reflect both Indigenous and Western science needs.But the review found that research was initiated by Indigenous people in only approximately a third of cases, and that Indigenous communities participated in designing research questions in less than half.This is problematic not only because the lack of Indigenous peoples' inclusion at these stages fails to ensure equal power-sharing.It might also fail to represent Indigenous peoples' needs.This resonates with Virtanen's (2021) discussion of research relevance in Indigenous territories, where research is often observed to be relevant to WKS communities but less to Indigenous ones.Moreover, the lack of collaboration in establishing common goals might affect Indigenous commitment to knowledge co-production (Norström et al. 2020;Vincent et al. 2021).This was observed in one of the examined cases, where actors changed during research and new actors, not participating in research question design, were less interested in collaboration (Anticona et al. 2013).
Fourth, the principle of frequent interaction is discussed within Indigenous-scholarly partnerships.Whilst Castleden et al. (2017) find face-to-face interactions throughout their own sample, only approximately a quarter of the Amazonian studies here described how this principle was achieved through regular visits or living with communities.In other cases, frequent interaction was limited due to funding, inaccessibility, and distance between Indigenous and academic communities.This had negative impacts on trust.Conversely, authors purporting to interact regularly with Indigenous communities did not deem community mistrust a challenge.These findings correspond with the theoretical framework highlighting the value of frequent interaction for relationship building and collaborative learning (Norström et al. 2020;Sutherland 2016;Vincent et al. 2021;Zurba et al. 2021).Regarding the latter, the studies described how regular interactions enabled better understanding of differences between knowledgesystems.This, in turn, is claimed to support communication, representation of Indigenous culture and customs, and contribution of credible, salient and legitimate knowledge (Norström et al. 2020;Zurba et al. 2021).Now that we have discussed the variation in these principles' rationales and realities, another important observation needs to be stressed, namely the variation between the outcomes of co-production as well.The first important observation is that distinctions between Indigenous and scholarly outcomes were not always clear.This could indicate that co-production in many cases was orientated towards common research goals, aiming to benefit Indigenous and scholarly communities.The latter benefited from generated knowledge that addresses gaps in the literature and is actionable for policy-makers.Likewise, Indigenous communities benefited from the co-production process and the knowledge generated, as they developed resources and skills with which to address their particular issues and needs.This exemplifies how bridging in co-production can result in knowledge that is legitimate and actionable for scholars and Indigenous people.These findings resonate with the benefits of co-production identified by many researchers (Cooke et al. 2020;Harvey, Cochrane, and Van Epp 2018;Latulippe and Klenk 2020;Wyborn et al. 2019).However, it is important to note that not all resources explicitly discussed co-production outcomes for Indigenous communities.In addition, as Castleden et al. (2017) asserted, it is difficult to translate Indigenous communities' benefits into action.Moreover, Amazonian Indigenous communities were claimed to be empowered and equipped with skills and resources, but implementation was often left as a step beyond collaboration.This illustrates that work on co-production is, and indeed should be, still ongoing.
To conclude this discussion, we can observe that the bridging of Indigenous and Western knowledge, in coproduction with Amazonian Indigenous communities, was largely founded upon a form of partnership that emphasised mutual respect, power-sharing, and empowerment of Indigenous communities.Such bridging was seen to contribute to a knowledge-generation that is more legitimate and beneficial for all actors involved.However, clear challenges remain.The paper observed a discrepancy between the theory and practice of knowledge co-production.Further work is required to address the potentially asymmetric power-relations between Western researchers and Indigenous communities.And in turn, more work is needed to enhance the legitimacy of knowledge, in order to further empower Indigenous communities to address local challenges according to their own contexts and needs.

Conclusion and recommendations
This paper began with Edwards' (1989) call for a development research that was more relevant and more inclusive of Indigenous views and perspectives.Now that we have taken stock of contemporary attempts to meet this challenge in the Amazonian context, it is worth drawing some final conclusions.The chief finding is that whilst laudable progress was achieved by scholars seeking to bridge IKS and WKS, there is still work to be done.We have seen that co-production carries significant potential for development research, and corresponds with the bottom-up approach that seeks to include marginalised communities in a collaborative search for local solutions.
We saw, from the data sample, that bridging knowledge-systems in co-production is largely partnership-based.Participatory methodologies and mixed methods take precedent, and this helps enable Indigenous peoples' inclusion.The emphasis is firmly on mutual respect, power and benefits-sharing, and empowerment of often-marginalised voices.
What we have also observed in the above account, is the attempted inclusion of Indigenous communities at multiple stages of the research process, where they shared decision-power with scholars and stakeholders.However, it is important to avoid naively concluding that Edwards' challenge has been met entirely.Indigenous actors are rarely included at all stages of the research process, and it will prove valuable for researchers to further address, and seek to overcome, the persistently asymmetric power-relations evident within co-production practices.This is not to say, however, that positive outcomes eluded this paper.The key positive outcome, identified across the data sample, was the legitimisation of a knowledge that was more beneficial for all actors involved in the research process.Of course, there is room for further improvement, not least with regard to post-research action and 'aftercare'.
Finally, the paper found that by following the key principles of co-production, the chances of mutually beneficial knowledge-bridging can be enhanced.Whilst it is clear that such attempts are impeded by conceptual, logistical and technical barriers, a commitment to building relationships and developing mutual understandings of IKS' and WKS' harmonies and discords can reap valuable rewards for both the academic and Indigenous communities.This can help overcome conceptual misunderstandings and enable a greater sense of inclusiveness in contemporary development research.Second, and on that point, the paper cannot stress too keenly the practical value and ethical purchase of Indigenous inclusion.Researchers must discuss projects' possible collaboration mechanisms with Indigenous actors.Doing so will enable meaningful collaboration, corresponding with the context in which research occurs.Sensitive dialogue is essential.Moreover, it is important that researchers are aware of power-relations within the Indigenous community and seek to accommodate marginal 'voices' therein.Third, a key finding from the above review is the extent to which linguistic and environmental barriers can hinder effective communication and interactions between academic researchers and Indigenous knowledge-holders.While insufficient funding might limit collaboration, it is important to find creative ways to ensure frequent communication, an essential prerequisite for collaboration.Lastly and most importantly, researchers need to be aware of their own positionalities and decision-power, given their financial and institutional connections.A crucial lesson for development researchers, seeking to effectively bridge IKS and WKS, is to reflect on the research process throughout, and consider where their decision-power can be shared with Indigenous actors more equitably.
In the final analysis, the bridging of IKS and WKS, using co-production, is rich in potential.Co-production is not straightforward; it requires much creativity and flexibility to ensure desirable outcomes.However, it enables the generation of knowledge that can address local challenges and empower local communities, so that they themselves can find solutions for their local issues and seek development according to their own contexts and their own community needs.

Figure 2 .
Figure 2. Proportion of analysed resources by year bracket.

Figure 3 .
Figure 3. Proportion of analysed resources by resource type.

Figure 1 .
Figure 1.Flow diagram for systematic realist review.

Figure 4 .
Figure 4. Proportion of analysed resources by first author's affiliation.

Figure 5 .
Figure 5. Proportion of analysed resources by first authors and co-authors' indigenous identity.

Figure 6 .
Figure 6.Proportion of analysed resources by study's country.

Figure 7 .
Figure 7. Proportion of analysed resources by study's field.

Figure 8 .
Figure 8. Proportion of analysed resources by involved actors.

Figure 9 .
Figure9.Proportion of analysed resources by terminology used to name Indigenous partners.

Figure 10 .
Figure 10.Proportion of analysed resources by the representation of Indigenous actors.

Figure 11 .
Figure11.Proportion of analysed resources by methodology used in studies.

Figure 12 .Figure
Figure 12.Proportion of analysed resources by Indigenous peoples' inclusion at represented research stages.

Table 1 .
Inclusion and exclusion protocol with scale rating.