Fostering communities of practice for improved food democracy: Experiences and learning from South Africa

The highly concentrated formal agri‐food sector holds enormous power in the governance of the food system in South Africa. Yet the concept of food democracy holds that all people should have an equal opportunity to participate in (re)shaping their food system. To contribute to decision making, however, stakeholders not only need access to knowledge about the food system, but should also be able to inform what is considered relevant knowledge. Communities of practice—groups of people who share a common interest or concern and who deepen their knowledge and expertise in this area by interacting on an ongoing basis—provide novel spaces for stakeholder involvement in food governance. This article sets out the process of establishing and maintaining a community of (food) practice in the Western Cape province of South Africa. Through participant observation, analysis of unpublished documents, as well as in depth interviews, the core elements of the community of practice are charted over its 3 yr history. The article demonstrates the informal and evolving nature of communities of practice as social learning structures that can gain credibility and potentially develop into more formal democratic institutions over.


INTRODUCTION
The concept of food democracy holds that all people should have an equal opportunity to participate in (re)shaping the food system (Hassanein, 2003;Welsh & MacRae, 1998). Finding solutions to ecological, social, and economic problems, it is argued, should be undertaken through the meaningful participation and political engagement of citizens (Hassanein, 2008). To actively participate in decision making, however, individuals and communities need to be informed to the fore questions of "whose knowledge is to be recognized, translated and incorporated into action" (Nowotny, 2003). Central to the concept of food democracy is the need to foster place-based dialogue that crosses sectors, jurisdictions, and disciplines to extend policy networks to include all those who are governed and affected by the food system. De Schutter (2014) claims that "change can be expected neither from government action, nor from business initiatives alone, and grassroots innovations led by ordinary people have a limited impact. Only by connecting these different pathways for reform by food democracy can lasting food systems reform be achieved.' Conceived of in this way, food democracy is a call for a new method of decision making when values and interests come into conflict and when the consequences of decisions are uncertain (Hassanein, 2003). In these type of 'wicked' policy problems, such as food security, "what matters is arriving at decisions which are reasonable and appropriate in situations that are both morally and factually ambiguous" (Sanderson, 2002).
While scholars and activists recognize that democratization of our food systems is a growing imperative in contemporary food governance, there has been limited exploration of the meaning of food democracy both theoretically and practically (Hassanein, 2008;Shepard, 2008). Some scholars and international organizations have underlined the utility of developing novel spaces for public involvement in food policymaking (Government Office for Science, 2011;Hansen, 2010) to foster trust and engage the public in critical decisions that affect their everyday practices (Ankeny, 2016). This has led to a growing interest in food policy councils, initially in North America (Bassarab et al., 2019) and more recently in Europe (Sieveking, 2019). A well-known example that is often discussed in terms of food democracy is the Toronto Food Policy Council in Canada, which aims to provide a mechanism for citizens to actively participate in shaping the food system (Welsh & MacRae, 1998).
This article seeks to add to the literature exploring practical attempts to deepen food democracy by reflecting on a particular 'home-grown' initiative to inform more democratic decision making in local food governance in South Africa-the Food Governance Community of Practice (FG-CoP). The article aims to demonstrate that communities of practice (CoPs) (made up of individuals, organizations, and institutions that come together on a regular basis to share knowledge and experiences on a common issue or concern) can be a powerful, if indirect, approach to operationalizing food democracy. Communities of practice have been recognized as sites of social learning where practitioners and other stakeholders share and generate knowledge through conversations, network building, and joint activities (Wenger, 2009b). In contrast to food policy councils, CoPs do not aim to directly inform decision making and policy, as their informal make up rarely carries a government mandate, nor do they claim legitimacy to speak for a

Core Ideas
• Widening the scope of knowledge for decision making is essential for food democracy. • Multiple perspectives can help find workable solutions to complex problems. • Communities of practice provide spaces for wide participation in food governance.
certain community or geographical area. Rather, the participants learn from the experiences and perspectives of other community members, creating opportunities to inform their own practice as well as coproduce new knowledge to inform decision making through joint learning. The next section of the article outlines the three elements of CoPs that (when functioning well) help them create novel spaces to integrate diverse perspectives to inform food governance. The following section then sets out the background and methods used. The next section, describes how each element in the FG-CoP in South Africa contributes to its ability to develop and share knowledge about the local food system. The article ends with some thoughts on future direction and ongoing challenges for the FG-CoP as well as drawing conclusions about how this initiative contributes to food democracy.

2.1
Communities of practice as a concept and body of literature A CoP is a group of people who share a common interest or concern and who deepen their knowledge and expertise in this area by interacting on an ongoing basis (Wenger et al., 2002). Communities of practice were originally developed as an analytical concept by Wenger and Lave to study learning through participation in networks by apprentices. Lave and Wenger (1991) argued that learning is not an individual but a social process situated in a cultural and historical context. However, the concept was quickly taken up by organizational and management studies that presented CoPs as vehicles to promote intrainstitutional learning. A myriad of articles now aim to guide the reader on how to cultivate and nurture a CoP (e.g. Cambridge et al., 2005;de la Rue, 2008;McDermott, 2004;Probst & Borzillo, 2008), while numerous articles also depict CoPs as dynamic structures constantly evolving and whose life cycle can be mapped out over time into several phases (e.g. Cambridge et al, 2005;Wenger et al, 2002). Still other articles set out a sobering array of causes of failure of CoPs (Archibald & McDermott, 2008;Probst & Borzillo, 2008).
Communities of practice can fulfil a variety of related functions. They can connect people who might not otherwise have the opportunity to interact, provide an opportunity to share information, help people organize around purposeful action, stimulate learning through the transfer of knowledge from one member to another, and generate new shared knowledge that helps people transform their practice (Cambridge et al., 2005). It is this latter function, as sites of knowledge integration and social learning, that is of most relevance for the coproduction of knowledge and food democracy.
The concept of the agora has been used to characterize this problem-generating and problem-solving environment in which actors and knowledge from inside and outside of science meet and the coproduction of knowledge takes place (Nowotny, 2003). It is populated not only by arrays of competing 'experts' and the organizations and institutions through which they bring their knowledge and experience to bear on decisions taken, but also variously jostling 'publics' (ibid). Rather than shifting the prioritization from one form of knowledge to another (whether indigenous or experiential etc.), realizing cognitive justice calls for the coproduction of knowledge-a collaborative process bringing together multiple kinds of knowledge and perspectives to construct an understanding based on an active recognition of a plurality of situated knowledges (Visvanathan, 2009;Oswald et al., 2016;). Expertise is therefore spread throughout society and democratized rather than in the hands of the elite (Nowotny, 2003).
While CoPs were originally thought of as intraorganizational learning structures, Cundill et al. (2015) argue for broadening our understanding of CoPs to recognize two types: intradisciplinary CoPs consist of people within a single discipline and are in keeping with the traditional definition, while transdisciplinary CoPs, on the other hand, can span several organizations and disciplines bringing together groups of people with very different expertise and experience. It is in the latter type of CoP that the coproduction of knowledge for complex or 'wicked' policy problems is most likely to occur (Cundill et al., 2015;Vincent et al., 2018).

Communities of practice as a structural model
While intradisciplinary and transdisciplinary CoPs may be different in some important characteristics (such as the level of trust and shared values between members), they are both still comprised of three basic elements described in the literature that help CoPs to become dynamic social learning structures and distinguish them from other groups or networks, namely, a shared domain, a community of people, and a joint practice (Wenger, 2009a).
First, a CoP is organized around a shared domain of interest, a common concern or issue that the members wish to learn about and potentially solve. The domain is the raison d'être of the CoP and brings people together and guides their learning (Wenger et al., 2002). Membership therefore implies a commitment to that domain and also a shared expertise in that domain that distinguishes members from other people (Wenger, 2009a). The domain "inspires members to contribute and participate, guides their learning and gives meaning to their actions" (Wenger et al., 2002). Knowing the boundaries (and the leading edge) of the domain enables members to decide what is worth sharing, how to present their ideas, and which activities to pursue (ibid). A community may be more or less explicit about what their domain includes but the members' shared understanding of their domain-its purpose, its resolved issues, its open questions-allows them to decide what matters. The domain guides the questions they ask and the way they organize their knowledge. It helps them sort out the trivial ideas and the one with real promise. A domain is not a fixed set of problems and can evolve along with the world and the community. As these problems are solved, new ones can appear, new challenges can arise in the real world, or the next generation of members can bring a new perspective.
Second, CoP members engage in joint activities, interacting with each other and sharing information with an objective of improved mutual understanding and empowerment. During this process, relationships are built that enable members to learn from each other. This process develops a community of people that care about the domain. It is the community that creates the social fabric of learning by fostering relationships and a willingness to share ideas and questions (Wenger et al., 2002). "Having others that share your overall view of the domain and yet bring their individual perspectives on any given problem creates a social learning system that goes beyond the sum of its parts" (Wenger et al., 2002, p34). While the relationships within CoPs are often said to be built on mutual respect and trust, Wenger et al. (2002) argue that effective communities are not necessarily without conflict. Rather, strong communities are better able "to handle dissension and make it productive' and even use conflict as a way to deepen relationships and learning." Third, CoP members are practitioners and not merely spectators. They actively test and use ideas and learnings in their daily lives or work, usually through developing a shared repertoire of communal resources and ways to address recurring problems (Wenger et al., 2002). Whereas a domain denotes the topic the community will focus on, the practice is the specific knowledge the community develops, shares, and maintains (Wenger et al., 2002). In short, the community develops a shared practice in order to be more effective in their domain (Wenger, 2009a). Wenger et al. (2002) explain that successful practice depends on a balance between joint activities exploring ideas and the coproduction of 'things': It involves an ongoing interplay of codification and interactions, of the explicit and the tacit. . . ..On the one hand, the goal of documenting and codifying focuses community activities, and on the other hand, these activities give life and legitimacy to the documentation.
Shared practice therefore includes reification (making object) of learning between the community members through books, articles, knowledge bases, websites, and other repositories that members share (Wenger, 2009b). It embodies a certain perspective on the problem or domain, a thinking style or even in some cases an ethical stance. An effective practice evolves with the community as a collective product and is integrated into the member's work.
According to Wenger et al. (2002), when these three structural elements of the CoP function well together, they make a CoP an ideal knowledge structure-a social structure that can assume responsibility for developing and sharing knowledge. These three structural elements of CoPs are used in this article to help reflect on the experiences of establishing and maintaining the FG-CoP to inform more democratic knowledge production to inform local food governance in South Africa.

Food insecurity in the Western Cape, South Africa
Whatever measure or indicator is used, about one-quarter of South African households experience food insecurity (Statistics South Africa, 2019)-one of the clearest reflections of poverty and inequality in the country. In the Western Cape province, as a result of the high level of urbanization, twothirds of households vulnerable to hunger live in urban areas (Statistics South Africa, 2019). Cape Town, the capital city of the Western Cape, concentrates 64% of the population of the province. Its residents experience particularly high levels of food insecurity, with 54% of households across the city reported as food insecure (Crush et al., 2018) and up to 89% in the poorest areas (Battersby, 2011). Food insecurity is compounded by a rapid nutrition transition (Steyn & Mchiza, 2014), exacerbating the double burden of malnutrition (the coexistence of under nutrition alongside overweight and obesity). This is reflected by stunting in 22.9% of children under five and simultaneously high levels of overweight and obesity affecting 43.7% of men and 73.7% of women in the province [National Department of Health (NDoH), Statistics South Africa (Stats SA), South African Medical Research Council (SAMRC), & ICF, 2017]. These figures are attributed not only to the wider social issues mentioned above, but also to the highly industrialized agri-food production and to economic concentration in manufacturing and retail, which extends its reach through a broad periphery of informal retailers while offering limited livelihood opportunities (Greenberg, 2017).

Food governance in the Western Cape
Food security researchers interpret these nutritional and economic aspects as the outcomes of a complex food system dominated by the interests of capital (Pereira & Drimie, 2016). State capabilities to govern the food system are fragmented vertically by three spheres, or levels, of government and horizontally by multiple departmental mandates. The policy space is dominated by rationalities rooted in deregulation of agricultural production, processing, retail, and international trade. Food security most often continues to be relegated to a narrow agricultural, productionist framing spearheaded by agricultural departments, while engagement with the nutritional and health impacts is confined to the ambit of politically weak departments of health (Thow et al., 2018). The highly concentrated formal sector holds enormous de facto power in the governance of the food system and exerts significant policy influence. Civil society engagement with food issues is oriented primarily to food aid and to the promotion of small-scale, agroecological production and urban food gardens. Because of the social and political complexity of the issue, reaching food security is frequently framed as a 'wicked problem' that defies technical, hierarchical, and state-centric governance approaches (Candel, 2014;Pereira & Drimie, 2016;May, 2017). This systemic, 'wicked' problem framing has informed calls for forms of governance that are systemic, cross-boundary, adaptive, inclusive, and oriented toward systemic transformation (Termeer et, al.,2018). Two recent policy developments make the province, and particularly the city of Cape Town, a responsive governance landscape for multiple-stakeholder dialogue. First, in September 2016 the Western Cape Government released a draft Western Cape Government household food and nutrition security strategic framework for public comment (Government of the Western Cape, 2016). Second, in August 2019 the City of Cape Town published a resilience strategy that included food insecurity as one of 14 major stresses on the city and proposed an action of "establishing a food systems programme to improve access to affordable and nutritious food" (City of Cape Town, 2019). Both of these policy documents were preceded by extensive stakeholder workshops and consultation as well as sustained lobbying by food systems researchers, which perhaps prepared the ground for these policy shifts. The documents indicated an openness by these two levels of government to a variety of perspectives and served to strengthen connections between stakeholders.
The FG-CoP was initiated in this specific context. It developed from a series of stakeholder workshops organized by academics in 2017 which included stakeholders who had previously taken part in the workshops associated with the provincial draft strategic framework. The suggestion to loosely institutionalize the meetings within the framework of a CoP came from a government official within the meetings and was taken up by the local academics who took on the role of FG-CoP secretariat.

Methods
The account of the FG-CoP is based on three sources of evidence. First, participant observation by the authors, three of which formed the secretariat of the FG-CoP and the fourth a companion from the start of the FG-CoP process. Second, documentation of the 15 FG-CoP gatherings and other unpublished documents generated by the FG-CoP such as its design principles and outputs from specific research processes initiated by the FG-CoP. Third, semi-structured interviews with 15 FG-CoP members. A purposeful sampling technique was used with the interviewees selected on the basis of their active involvement in the FG-CoP. In addition, interviewees were selected in order to represent a range of different types of members such as academics, government officials, privatesector representatives, and practitioners. Interviewees were asked questions relating to three general themes: (a) motivations for joining and continuing to participate in the FG-CoP, (b) perceptions of the power relationships within the group, and (c) the knowledge shared and gained in the meetings. The interviews were conducted face to face by the authors in Cape Town between December 2019 and February 2020. The recordings were transcribed and coded by hand in line with the three themes covered in the interviews. Ethical clearance for data collection was obtained from the Humanities and Social Science Research Ethics Committee of the University of the Western Cape (ethics reference number HS18/5/ 13).

RESULTS: EXPERIENCES OF A COMMUNITY OF (FOOD) PRACTICE IN SOUTH AFRICA
This section sets out how each element of the FG-CoP contributes to the FG-CoP's ability to develop and share knowledge about the local food system and relevant alternative ways to organize it.

Shared domain
The shared understanding of the domain and purpose of the FG-CoP was developed jointly by the members over two early meetings in November 2017 and March 2018 (Unpublished data, 2017(Unpublished data, , 2018. The common issue or problem that brought the members together was their shared concern over the persistently high levels of household food insecurity in the province. The domain has a strong social justice aspect, with members of the FG-CoP sharing a high commitment to "making a difference in the access of nutritious food for disadvantaged sections of the society" (Appendix 1, Interviews 6 & 13). The importance of values in shaping the FG-CoP, such as justice, vulnerability-centered, inclusivity, asserting socioeconomic rights, and representing diversity, came out strongly in break-away discussions in the first formal meeting as well as several pre-FG-CoP meetings, which eventually led to calls to establish the CoP (Unpublished data, 2017). Closely linked to these values was a shared belief by FG-CoP members in the need for fundamental structural transformation within the food system: Sometimes we don't agree on the analysis of the problem but there is a shared understanding that there is a problem and that the food system is broken and that we need to do something about it. And I don't think that outside of the room that this is something that people necessarily understand. (Appendix 1, Interview 11) At the same time, there was a clear understanding that there was potentially more room for maneuvering to shape the food system in the Western Cape than at the national level. The initiative was thus informed by a systemic framing of the issue that recognized opportunities for governance innovation across multiple domains spanning production, processing, distribution, and retail as well as multiple actors from different social sectors including state, private sector, civil society, and academia.
In these early meetings, the purpose of the FG-CoP was also deliberated. The community was seen as an opportunity to contribute to the transition to a just and sustainable food system through the establishment of a collaborative network, which would promote, exchange, and connect different types of knowledge and action for policy change, draw on different perspectives, inform decision making, create awareness, understand broader food challenges, and identify research gaps (Unpublished data, 2017). The research aspect was initially a strong part of the purpose of establishing the FG-CoP. As a government official working in the food governance space recounted: As soon as we started engaging with food [policy] it became clear how fractured the knowledge base was and also how large the gaps were and that there were no active spaces where that was knowledge shared but also the gaps being identified and responded to. So I think that was one of the original impulses to create space like that so over time we could better understand the policy environment. (Appendix 1, Interview 8) At the same time, the researchers, who originally organized the series of workshops in 2017, which eventually led to the FG-CoP, were strongly motivated by a desire to mold their research agenda closely with the user needs in local government and beyond.
The domain, values, and purpose of the FG-CoP formulated in these initial meetings were recorded in a short document "CoP Design Principles" (Unpublished data, 2018). However, Wenger et al. (2002) reminds us that the domain is not a fixed set of problems and can evolve and grow along with the world and the community. For a CoP to thrive it must continue to intersect with the passion and aspirations of its members but also be of strategic relevance to the context (e.g. organization or governance landscape) in which the CoP is situated if it is not to be marginalized and have limited influence. In the case of the FG-CoP, the original series of topics discussed in the meetings were mainly chosen by the academics and government officials that initiated the process leading to the community as a way to cover various perceived research gaps and needs.
However, over time, the topics were increasingly guided by two additional considerations. First, as participation in the FG-CoP grew and became more inclusive, more nongovernmental organizations (NGOs), civil society organizations (CSOs), and practitioners joined the community, and the topic of the meetings was widened to include the interests of the CSOs. For example a specific meeting focused on the role of CSOs in food governance, and a parallel meeting was set up to map CSOs working, directly or indirectly, in the food space. The format of the meetings also evolved from mainly expert talks and plenaries to include group work and feedback in a bid to allow more voices from the floor to emerge (Appendix 1, Interviews 10 & 11). Nongovernmental organizations, CSOs, and practitioners consequently started to play a much more central role in the FG-CoP meetings asking "what does this mean for us, and what are the practical challenges for us?" (Appendix 1, Interview 3).
Second, in March 2020, as South Africa went into one of the strictest lockdowns in the world in a response to the Covid-19 pandemic, food security rose up in the political agenda and into the consciousness of citizens with a chilling urgency. Three million people were estimated to have lost their jobs in the country between February and April (Wills et al., 2020).
As a result of school closure, the government stopped essential services such as a school nutrition program feeding nine million children (Seekings, 2020); food prices rose (Pietermaritzburg Economic Justice & Dignity Group, 2020), and consequently, levels of food insecurity more than doubled (Wills et al., 2020). In the Western Cape alone, government officials feared that the pandemic might push as many as three million people into food insecurity (Maguire, 2020). The lockdown also created a completely novel governance environment where new national regulations needed to be interpreted, their impacts on different constituencies considered, and their legitimacy contested. Through online platforms, the FG-CoP leveraged its broad networks to respond rapidly to these disruptions and channel information into local and provincial governments. The frequency of the online meetings increased from quarterly meetings to monthly formal online meetings and other informal smaller online gathering between FG-CoP members and government officials. The topic of the meetings also shifted to cover the most pressing issues at particular points of time such as the impacts of lockdown on informal food trade, emergency food aid, the rise of civil society in food issues, the consolidation of the corporate sector under lockdown, and the role of urban agriculture in broader food system change after Covid-19. The FG-CoP was also awarded funding by provincial government through the channel of the Centre of Excellence in Food Security to support its more dynamic role during this time and in particular to get guidance on crisis management and beyond and capture some of the knowledge generated into policy briefs.

Community
The members of the FG-CoP were initially participants in a set of three workshops on food governance organized by academics in 2017 before the FG-CoP was established. However, new members joined (at first on the invitation of the secretariat or other FG-CoP members) and others dropped out over time. The number of people that participated in each meeting was initially intended to be relatively modest (20-35) to facilitate face-to-face dialogue. The number of overall FG-CoP members was higher, as not all FG-CoP members participated in every meeting. Half-day meetings took place 2018 and 2019 roughly four to five times a year and were based on a program of thematic topics that were identified by the FG-CoP members initially in the two early meetings in late 2017 and early 2018 laying down the core elements of the CoP. Meetings usually included a session of formal presentations with questions and answers followed by a session of group work and feedback. Time was reserved at the end of the meetings for housekeeping and reflection on the FG-CoP itself and each meeting was followed by a lunch in which members were encouraged to continue the discussions informally. During the Covid-19 lockdown and immediate aftermath, the FG-CoP shifted to shorter (2.5 h) online meetings consisting of two to three short presentations followed by panel responses from different perspectives. The meetings also switched to open invitation that were not anymore limited to the Western Cape (even if the majority of participants remained from the province) and numbers attending each meeting became far higher (sometimes as high as 100 people or more) with many participants joining for a single meeting and not considered FG-CoP members. A fairly stable core group of 25-30 FG-CoP members, however, join each meeting, often taking an active role in the planning, presentations, and deliberations.
The FG-CoP meetings were embedded in ongoing activities of the secretariat that included documenting meetings, reflecting on key insights, managing contact databases for dissemination of invitations, conceptualizing future themes, identifying and recruiting prospective panelists and invitees, and interviews with individual stakeholders. Thus, the dense interactions and knowledge exchanges that took place in the CoP gatherings were entwined with a more diffuse and continual process of knowledge coproduction. Various functions or roles within the community developed over time, mostly unofficially; the organization of the FG-CoP activities was mainly taken on by the secretariat and regularly discussed with a government official who has been a companion of the process in his personal capacity. A secondary group of core FG-CoP members took on a 'thought leadership' role presenting in several meetings and advising on other speakers and also playing an active role in the deliberations during the meeting. This core group comprised of other academics, government officials from both spheres of government (local and provincial), as well as one CSO. These core members of the FG-CoP later became formalized in a reference group that guided the topics and content of the upcoming meetings initiated in response to the new situation created by the Covid-19 pandemic, and they also oversaw the production of learning and knowledge outputs such as policy briefs.
In the course of these activities, the core community members identified several joint activities, or learning projects, that were then carried out in smaller groups often in separate activities or processes that were later reported back to the main FG-CoP. These activities were usually to address knowledge gaps or specific needs in building commitment to the shared learning agenda. One of the early activities was the development of a 'good food declaration' laying out the values and principles that the community members would like to see underpin the food system in the Western Cape. It was originally anticipated that this declaration would be a precursor to the more far reaching activity in the construction of a local food charter for the Western Cape. However, a scoping exercise developed in parallel (Even-Zahav et al., 2020) raised concerns of credi-bility and legitimacy and recommended a process of building relationships and awareness with CSOs first. In light of this, another joint activity was organized to map the CSOs in the province working in the food space. Other learning projects undertaken under the auspices of the FG-CoP with support of the Centre of Excellence in Food Security included a study of the local government mandates for shaping the food system, a place-based community food sensitive planning workshop, a digital story workshop exploring lived experience of food environments in a local township, the development of infographics on the food system, and a policy brief setting out the (food) governance landscape in the Western Cape. All of these learning projects resulted from ideas and learning gaps identified by the community members during discussions and the results were reported back to the community members in subsequent FG-CoP meetings.

Joint practice
A shared practice has taken time to develop within the FG-CoP, in part, possibly in response to the wide and complex nature of the domain as well as the diversity of the perspectives of the community members. The meetings were documented from the official start of the FG-CoP and these records made available in a shared online repository. Presentations from the meetings as well as key articles or reports were also stored centrally and made available. The outputs from joint learning projects were fed back to the members through presentations in the FG-CoP meetings and also through working papers, research reports, infographics, and digital stories. In some cases, research gaps identified by the FG-CoP were filled and documented through the academic theses of post-graduate research students supervised by members of the community. During the Covid-19 phase of the FG-CoP, additional resources were sourced so that the secretariat could initiate smaller specialized groupings of FG-CoP members to continue discussions after specific meetings and formulate short policy briefs that integrated as far as possible the different knowledges and perspectives on each meeting topic. While a dedicated website and more-elaborate knowledge repository for the FG-CoP was raised as a desirable joint activity several times by FG-CoP members in the early stages of the community, a lack of time by the secretariat and budget constraints related to creation and maintenance of a website has kept this on a back burner. A Slack forum was established to facilitate information sharing immediately after the implementation of the lockdown and despite initial activity, rapidly became dormant. Instead, the secretariat prioritized real-time virtual gatherings using the Zoom platform advertised through multiple channels including a rapidly growing email database, social media posts, and various WhatsApp groups. Wenger et al. (2002) remind us that these explicit aspects of the community's knowledge embody a certain way of behaving, a perspective on problems and ideas, a thinking style, and even in many cases an ethical stance-"in this sense a practice is a sort of mini-culture." Learning, therefore, essentially involves becoming an insider as they acquire the community's subjective view point and learn to speak its language (Brown & Duguid, 2001). In our experiences in the FG-CoP, learning to become an insider is most apparent in the experiences of the students who are new to the food governance space, as one student member explains: I am beginning to feel more of a community of practice member . . . Every single time I come I learn something new. . . For example when I heard X speak . . . the next day I read her book and incorporated her views into my views and then I heard another speaker with another angle and it feels like I am building a castle and I am adding things to it so that I can have a well-informed opinion. . . because I am fresh in this space . . . . (Appendix 1, Interview 6) As less experienced participants, such as this interviewee, create an opinion and understand the topics from what they hear in the meetings, both their confidence and identification with the group grows.
For other members of the FG-CoP, however, the development of a shared perspective and understanding of the food system is more tenuous, in part, as a result of the heterogeneity of the community members (as part of a transdisciplinary CoP). There are, however, certain areas of commonality that can be seen reoccurring in the FG-CoP discussions, which began to appear to indicate the beginning of a shared understanding such as an understanding that local and provincial government has a (often unrecognized) mandate for food governance, access to nutritious food is as important as the availability of food locally, including a diversity of perspectives on the policy problem is essential to finding workable solutions, and the necessity of connecting bottom up grassroots knowledge and initiatives to various levels of government. In addition, the heterogeneity of FG-CoP members means that they operate in substantially different areas of practice and so apply this emerging shared understanding and knowledge in varied arenas such as science, government departments, and civil society.

Looking forward
A number of challenges and choices face the FG-CoP as it continues to mature. An immediate concern is the growing size of the community; even prior to the meetings going online and food security rising up the political agenda during the Covid-19 lockdown, the numbers participating in the meetings were steadily increasing. Since lockdown, there have been several meetings with more than 100 participants. A community is not just a network where people may only rarely meet each other. It is a group of people that interact, build relationships, and develop a sense of belonging. If a community becomes too large (and goes online) this can inhibit direct interaction. Break-out groups were trialed in one online meeting with some success, as they allowed for more intimate discussions between members. However, organizing breakout groups online has its own set of logistical, technical, and resource demands that compete with the secretariat's academic demands of teaching, researching, and publishing. Another way to facilitate more direct interaction between CoP members has been to organize smaller subgroups of members to continue discussion on particular issues and focusing interaction around jointly producing media articles and policy briefs. However, retaining and nurturing a sense of belonging and common purpose in a larger online community remains a challenge. Moreover, the legitimacy of such an undertaking depends on its inclusivity, and since much of the activity has shifted into virtual spaces, the digital divide and the economic constraints to internet access pose real hurdles for the democratization ambitions of the FG-CoP.
As the community and its activities grow, practical questions arise about how to consolidate funding and spread leadership across a wider group of FG-CoP members. A healthy CoP does not depend on the leadership of one person or even a few people; rather, leadership is distributed and is a characteristic of the whole community (Wenger et al., 2002). There is now a core reference group of FG-CoP members that help steer the activities and especially the meeting topics and presentations. The reference group internally critiqued its elite and overwhelmingly white membership (the racial divide being a very sensitive issue in the South African context) and has since reached out to diversify its membership. However, in this busy world, it is difficult to find FG-CoP members that can take responsibility to lead each of the subgroups and take these activities forward. If funding is secured each year, it does not cover salary costs beyond some limited contribution to the secretariat.
Communities of practice evolve and mature overtime. One manifestation of this process is their relationship with relevant organizations, which Wenger et al. (2002) argues, can range on a continuum from unrecognized (often invisible to the relevant organizations) through supported (providing direct resources) to institutionalized (given official status). While the Western Cape FG-CoP was at first unrecognized, including to the members themselves for the first few (pre-FG-CoP) meetings in 2017, it has been gaining visibility with different stakeholders and organizations as it grows and gains resources and legitimacy. Most notably has been its increasing visibility and support from both provincial and local government and growing interest from stakeholders from other parts of the country who are joining the CoP dynamic with the development of online meetings. Looking forward, it is possible to see several paths for the FG-CoP: one would be to broaden out as a larger independent network with multiple territorial and thematic focus areas; and another would be to become fully institutionalized, possibly as a more formal body (e.g. a food policy council), with an official mandate to provide diverse perspectives and policy advice to local governments.

4.5
The potential to deepen food democracy The FG-CoP provides an example of bottom-up democratization dynamics because it was initiated by nongovernmental actors at the local level, and the (growing) government involvement was indirect and came initially through the interest and dedication of a few officials in their personal capacity rather than officially mandated to be part of the community. This informal interest and involvement by government in the FG-CoP is changing as is demonstrated by the direct investment in the community by provincial government more recently in order to gain policy guidance based on multiple perspectives. The involvement of stakeholders from across the sectors has led to the perspectives of consumers, small-scale farmers, and practitioners from CSOs and the private sector being placed alongside those of government officials and academics. Having said that, it is inevitable that some power imbalances persist in any such multiple-stakeholder dialogue, which can significantly impact on social learning and knowledge coproduction if left unchecked. The challenge is to prevent the process from being highjacked by particular groups of stakeholders and to rather make sure that no social actors are privileged over what other disciplines and social actors contribute (Dewulf et al., 2005). In our experiences of the FG-CoP, we found that asymmetries in power need to be actively identified and mitigated (Adelle et al., 2020). Risks of having the process dominated by a few groups can be managed in part by opening up the invitations to a wide range of stakeholders, bringing on board their concerns and challenges. Given the recent emergence of the FG-CoP and the complexities of tracing policy impacts, it is not yet possible to assess the community's outcomes in terms of impact on foodrelated policymaking in the Western Cape. There are, however, some indications that the FG-CoP is at least part of a growing mobilization and networking of food-system stakeholders in the Province. This mobilization started before the formation of the FG-CoP but the community has developed, diversified, and gained support from both levels of government (province and city). During the lifetime of the FG-CoP, the food security debate has been revitalized in the Province and City of Cape Town and has progressively been incorporated in government strategies. This has resulted in a pilot food program in the City of Cape Town, in a nongovernmentalgovernmental emergency food aid coordination group, as well as in cross-department food coordination groups initiated within both levels of government (i.e. city and province). The FG-CoP members sit on each of these groups and this 'fertilization' from the bottom of the food governance debate contributes to the much needed desegmentation of the food policy space.

CONCLUSION
If food is a site of struggle between those economic forces de facto controlling the system and those citizens seeking to create more sustainable and democratic food systems (Hassanein, 2008), then widening the democratic scope of knowledge on which our governance decisions on food are based is an essential component of food democracy (Adelle, 2019). A diverse group of stakeholders not only need to be informed about food systems and potential alternatives, but they also need to have an opportunity to shape the discourse by which the system is represented, understood, and interpreted, and thus the knowledge on which decisions about food are made. The integration of different types of knowledge for decision making is not only a moral and political necessity with regard to economic and social justice but also a pragmatic one. Multiple perspectives can contribute to the local capacity to find workable solutions to some of society's most complex problems. Nowotny (2003) describe this as a shift from generating merely scientifically reliable knowledge toward more socially robust knowledge. For complex, cross-cutting or 'wicked' policy problems, therefore, the coproduction of knowledge to inform decision making is one possible pathway toward greater food democracy. Communities of practice represent one way of operationalizing food democracy by creating social learning structures in which different types of knowledge can be integrated and fed into decision making. The empirical account of establishing and running the FG-CoP in the Western Cape, South Africa, shows how CoPs are not static over time and can be influenced by both changes within the community itself and changes in the real world context in which they are situated and are attempting to informas directly illustrated by the Covid-19 crisis. In the case of the FG-CoP, the increasing interest in food security by local government in its bid to put together a resilience strategy, coupled with the urgent need to better understand the food system created by the lockdown, led to a new dynamism within the community. Paradoxically this new phase may, in part, lead to the FG-CoP mutating over time away from its CoP beginnings and into a larger food governance network or, even in a future stage, institutionalized as a food policy council.
While the latter is only a tenuous possibility, it does underline the potential for building more recognized and formal food democracy institutions from below. It shows how incubating these structures informally can help gather momentum and legitimacy but also implies the need to develop inclusive strategies that facilitate participation by marginalized stakeholders.

A C K N O W L E D G M E N T S
We would like to thank the members of the FG-CoP that have taken the time to engage with each other and embark on this process of knowledge sharing and social learning. In particular we are grateful to the FG-CoP members who shared their reflections on this (often messy) process with the authors in interviews. We would also like to thank two anonymous reviewers for their helpful comments on an earlier draft of the manuscript. This research is funded by the DSI-NRF Centre of Excellence in Food Security, South Africa.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest O R C I D Camilla Adelle https://orcid.org/0000-0003-4500-0496 Florian Kroll https://orcid.org/0000-0003-1244-914X