The muting and unmuting of caste across inter-linked action arenas: inequality and collective action in a community-based watershed group

Governing the commons in the face of socio-cultural heterogeneity is a challenge for academicians and practitioners alike. Some studies find that socio-cultural heterogeneity has a negative, a positive, or a non-linear relationship with collective action. Another set of studies finds that institutions can mediate the effects of heterogeneity that influence collective action for improving natural resource conditions. We build on the second set of studies to identify the underlying conditions under which these institutions promote successful collective action in a socio-culturally heterogeneous group with two different castes. Our case study suggests that under conditions that create equity, accountability, symbolic capital, and member capabilities, institutions can promote successful collective action. By using the concept of interlinked action arenas, we also show that although the community-based watershed group chose to mute caste in one action arena for the purposes of collective action, outside of this action arena, caste-related norms of untouchability are actively practiced. Group members have to continually switch between the muting and unmuting of caste from one action arena to another action arena, which continues to produce, reproduce, and maintain power asymmetries between members of both castes.


Introduction
Ever since Elinor Ostrom put institutions center stage in 1990, scholars have sought to understand the relationship between institutions and sustainable natural resource governance. Direct links from institutions to natural resource governance have been examined in a wide variety of settings over the years (Ostrom 1990(Ostrom , 1992Ostrom et al. 1994;Schlager and Blomquist 1998;Agrawal 2001;Dell'Angelo et al. 2017). While scholars sought to identify regularities across contexts, it became clear that the contexts themselves play an important role. In particular, the socio-cultural heterogeneity of a community is an important factor affecting collective action (Varughese and Ostrom 2001;Poteete and Ostrom 2004;Adhikari and Lovett 2006;Cochran and Ray 2009;Andersson and Agrawal 2011).
Scholars of common-pool resources (CPRs) and collective action have identified socio-cultural heterogeneity such as caste, ethnicity, and race as factors that may both enable and hinder community-level collective action for natural resource management. Most studies find that socio-cultural heterogeneity hurts collective action for managing natural resources, for several reasons. First, differences in social class, caste, and ethnicity create differences in preferences and interests, which makes organizing difficult. Second, such differences pose challenges to building agreements and enforcing norms of proper behavior towards the resource and other users. Third, differences can breed distrust and resentment towards others, hindering cooperation. Numerous studies have found these reasons to be barriers to successful collective action (Blair 1996;Baland and Platteau 1997;Kant 2000;Chakraborty 2001;Varughese and Ostrom 2001;Neupane 2003;Poteete and Ostrom 2004;Barros 2007;Naidu 2009;Shiferaw et al. 2009;Andersson and Agrawal 2011;Doss and Meinzen-Dick 2015).
In contrast, scholars have built on Olson's (1965) work to conclude that high levels of heterogeneity are associated with successful collective action (Baland and Platteau 1999). Other scholars have identied a non-linear relationship between heterogeneity and collective action (Fearon and Laitin 1996;Montalvo and Reynal-Querol 2005;Naidu 2009). By positing a U-shaped relationship between heterogeneity and collective action, scholars have found that low and high levels of heterogeneity are associated with lower levels of collective action, and medium levels of heterogeneity are associated with higher levels of collective action.
Findings from other studies do not support direct links from heterogeneity to collective action. Some have concluded that socio-cultural heterogeneity is not landowners could transition to Liquid Petroleum Gas (LPG) (Menon et al. 2007). These rules caused social harm to the poor but they did promote improved forest conditions.
Similarly, scholars have found that caste differences allow sub-groups to coerce others into following institutions that conserve the environment (Agrawal 1999). For instance, Kashwan (2016) discusses how power asymmetries limit weak actors from challenging unfair rules in one action arena so that leaders do not enforce additional costs in other institutional arenas. Going along with such rules allows the poor to benefit from other village institutions (Kashwan 2016) and accumulate social and symbolic capital (Kumar 2002;Cochran and Ray 2009). Such arrangements can produce inequitable social outcomes, but lead to positive evironmental outcomes (Agrawal 2001;Lele et al. 2010). Yet another study suggests that efforts to produce equitable outcomes can worsen environmental conditions (Engel et al. 2006). Thus, there is a tension between producing equitable outcomes and improving environmental conditions.
The above review demonstrates that this rich body of literature on heterogeneity and collective action points to the important role that institutions play in shaping resource governance outcomes under conditions of inequality. We draw upon the concept of network of adjacent action situations (NAAS) (McGinnis 2011) and inter-linked action arenas (Kashwan 2016), i.e. two action situations that are adjacent to each other where the outcomes from one action situation determine the rules of interaction that occur within the other action situation. Building on Cochran and Ray (2009) and Kashwan (2016), who examine reasons and incentives for the less powerful to contribute in inequitable action arenas, we focus on the effects of caste from one action arena that spill over onto another action arena that has supposedly muted caste to promote collective action. We demonstrate that even under conditions that promote successful collective action, caste lingers to reproduce and maintain relations of inequality despite the muting of caste. At the same time, participation in the group provides weak actors access to opportunities that broaden individual and collective capabilities that would not have been possible for them to pursue individually.
In the next section we discuss the theoretical framework of the conditions that promote collective action in socio-culuturally heterogeneous groups. Section 3 describes the study context and methods. Section 4 presents the results of field work on the conditions that promote collective action and the motivations of the non-elite actors to participate in action arenas that are embedded in a broader, unfair system, which are discussed in Section 5. Section 6 concludes with implications for future research on the potential for social transformation in socioculturally heterogeneous context.

Collective action in socio-culturally heterogeneous groups
While socio-cultural heterogeneity, by itself, often poses barriers to collective action, these barriers may be reduced by institutions (Varughese 2000;Chakraborty The muting and unmuting of caste across inter-linked action arenas 229 2001; Gautam 2002;Janssen and Ostrom 2006;Somanathan et al. 2007). Here we examine the key conditions under which institutions can promote collective action in a socio-culturally heterogeneous group of resource users. Prior research suggests that under conditions of equity, symbolic capital, accountability, and member capabilities, institutions may promote collective action in such communities. Moreover, prior research suggests competing motivations for weak actors to participate and follow rules that disadvantage them in heterogeneous settings.
Equity is the distribution of decision-making responsibilities, rights, costs, burdens, benefits, and resources in social systems that are considered to be "fair" by the stakeholders (Cook and Hegtvedt 1983;McKean 1992;Poteete 2004;Lawrence 2007). There are two dimensions of equity: procedural and distributive equity. Procedural equity refers to fair decision-making procedures that avoid the concentration of power in any one individual (Poteete and Ostrom 2004). Institutions for procedural equity include representation of marginal groups in resource management bodies and opportunities for marginal groups to influence decision-making through deliberation, membership, authority, power, leadership, rules and regulations, customs and laws (Poteete 2004).
Distributive equity refers to the fair distribution of benefits (Jacobs 1989;Anand and Sen 2000;Poteete 2004;Sunam and McCarthy 2010). Distributive equity is particularly important for collective action because it increases perceptions of fairness among members (Poteete and Ostrom 2004;Adger et al. 2016;Loft et al. 2017). Analyses of distributive equity usually focus on an economic evaluation of costs and benefits dependent on the resource outcome (e.g. Ostrom's design principles that include the congruence between the economic costs and benefits of participation). A few studies have focused on social costs and benefits (Kumar 2002), whereas others have examined symbolic benefits (Cochran and Ray 2009).
In considering equity, a challenge is to distinguish between institutions that genuinely create equity and seemingly consensual institutions that are followed because of the inability of less powerful actors to challenge unfair rules. Bourdieu's theory of practice is useful here because it turns our attention to the production and reproduction of relations of power that are maintained over time through institutions that structure social interactions (Bourdieu 1977). Thus, institutions followed without explicit coercion may seem to be equitable, but may actually be repositories of power that continue to be produced and maintained in social interactions, activities, and practices (Kashwan 2016). Through such consensual institutions, the politically and economically powerful can influence the promotion of certain norms to secure their interests (Poteete 2004;Kashwan 2016).
Symbolic capital is closely connected to equity, but differs in subtle ways. Symbolic capital goes beyond the examination of equity as an economic evaluation of costs and benefits, and instead examines material and symbolic goods that, otherwise, may not be considered economically valuable (Bourdieu 1986). Symbolic capital consisting of non-material benefits or feelings such as honor, pride, respectability, unity, and collective identity can help build strong bonds among actors to promote cooperation (Menon et al. 2007;Cochran and Ray 2009;Cohen 2013). These benefits can include " 'fair words' or smiles, handshakes or shrugs, compliments or attention, challenges or insults, honor or honors, powers or pleasures, gossip or scientific information, distinction or distinctions etc." (Bourdieu 1977, 177-8). Actors accumulate symbolic capital to use at another time through their everyday interactions and participation in networks and groups. Both the powerful and less powerful may contribute to collective action motivated by the need to maintain collective identity and project a positive image of their village even in the presence of internal differentiations and inequalities (Menon et al. 2007;Sangameswaran 2008). The poor can also be selective about engaging in activities to build a collective identity by choosing to participate depending on who benefits from the outcomes of the collective labor (Menon et al. 2007). Incremental reductions in social barriers among different sub-groups are possible even though symbolic notions of pride and unity do not address internal inequalities in a community (Menon et al. 2007;Wilshusen 2009).
Accountability increases perceptions of fairness to facilitate collective action (Saravanan 2002;Adhikari and Lovett 2006). Accountability is critical for monitoring whether socially marginal groups receive a fair share of benefits (Agrawal and Ribot 1999;Oyugi 2000;Young 2002;Lebel et al. 2006;Ostrom et al. 2014), and it is particularly important for creating checks on power. Agrawal and Ribot (1999) define accountability as "the exercise of counter power to balance arbitrary action." Accountability can result from information sharing, transparency, independent monitoring, polycentricity, separation of powers, legal recourse, budget control, free media, regular meetings, and maintaining records and meeting minutes (Klooster 2000;Kerr 2002;Ribot 2002;Agrawal et al. 2006;Gruber 2010;Pandit and Bevilacqua 2011).
Finally, group members may have diverse individual skills and capabilities, and such interdependence can be an asset, if pooled together, to group functioning (Wade 1988;Baland and Platteau 1996). Members can be encouraged to provide their specialized skills such as writing and bookkeeping, and resources of time, money, and knowledge to support collective action despite existing socio-cultural heterogeneity (Quiggin 1993;Varughese and Ostrom 2001;Poteete and Ostrom 2004). Thus, the contribution of capabilities, despite prevalent socio-cultural heterogeneity, can facilitate collective action.
Reasons for participating in inequitable action arenas are closely tied to equity, symbolic capital, and capabilities. Kashwan (2016) suggests participation in inequitable action arenas is because of the inability of the less powerful actors to challenge rules that disadvantage them and the additional costs that will be accrued in challenging powerful actors. Both Kashwan (2016) and Cochran and Ray (2009) suggest that less powerful actors participate and contribute in one action arena to accumualte economic and symbolic benefits in other action arenas. Cochran and Ray (2009) find that participation in inequitable action arenas is also motivated by an aspiration to equality that is manifested through the weak actors contributing as much as the elite in village development projects in a bid to be seen as equals to the village elites. Thus, less powerful subgroups may participate in inequitable action arenas to aspire to future equality, claim equal rights in a village, individual dignity, the right to participate in community life, build village unity, build social capital, and fulfill their dharm (religious duty) (Bourdieu 1986;Kumar 2002;Menon et al. 2007;Cochran and Ray 2009;Wilshusen 2009).
Amartya Sen's (1979) concept of capabilities enables an examination of what individuals are able to do because actors with less power can use their capabilities to negotiate access to and benefits from action arenas that would be previously restricted to them individually. Thus, the poor may have limited access to goods and resources individually, but as a group, the capability to access goods and resources increases (Sen 1979). This increased choice allows them to overcome entry barriers to other economic activities and opportunities, thus incorporating new doings and beings into their set of capabilities and achieve states that would not be possible individually (Pelenc et al. 2013).
As described above, examining conditions under which institutions promote collective action in a socio-culturally heterogeneous group shifts focus to equity, accountability, symbolic capital, and capabilities as well as points to the motivations of weak actors to continue participating in action arenas despite being embedded in a disadvantageous broader system. In this study, we investigate the institutions of a long-standing watershed management group in Karnataka, India with two castes through the lens of two interlinked action arenas: the broader action arena of village life and the narrower action arena of group meetings of the watershed management group. We base this study on the following research questions: 1. Do institutions promote collective action under conditions of sociocultural heterogeneity? 2. What are the conditions under which institutions promote collective action given existing socio-cultural heterogeneity? To what degree does such collective action spill over to affect socio-cultural heterogeneity in village life? 3. Why do weak actors participate in collective action when the broader system in which they are embedded in continues to disadvantage them?

Study context
To examine the conditions under which institutions promote collective action in a socio-culturally heterogeneous group, we turn to a community group comprised of two different castes. This case occurs in Karnataka, India, amidst the backdrop of community-based watershed development programs. Watershed management in India is the holistic management of natural resources and diversification of agro-based activities to create sustainable livelihoods (Common Guidelines for Watershed Development Projects 2008). In the early 1990s, government and non-government agencies engaged in community-based watershed management, conceived community as a group sharing a common history, ethnicity, norms, values (Chhotray 2004). Policy makers assumed that such commonalities would motivate individuals to self-organize. Villages with "harmony," i.e. villages without factional relationships or radical social movements, were chosen over villages with different caste groups (Baviskar 2001;De Souza 2001;Chhotray 2004). Such assumptions of community downplayed differences within seemingly "homo geneous" communities (Agrawal and Gibson 1999). Socio-cultural divisions of wealth, caste, and gender are never far away in Indian society, and these divisions determine access to and control of natural resources, participation of users in watershed programs, institutional development for resource management, decision-making, and benefit sharing within a group (Oakerson 1986;Bromley and Cernea 1989;Mehta 1997;Arnold 1996;Agrawal 2001;Poteete and Ostrom 2004). The socially-legitimized nature of caste hierarchies and inequalities over centuries allows elite castes and classes to control natural resources unchallenged by other social groups (Mehta 1997). Although the caste system and its practices are outlawed in India and are punishable offences, untouchability continues to disadvantage Dalits and tribal populations that form the most impoverished segment of Indian society. Thus, the search for community in watershed development must grapple with systemic socio-cultural heterogeneity.
Through discussions with non-governmental organizations (NGOs) working in India, the principal investigator identified a community-based group with sociocultural heterogeneity that has sustained itself over time. The establishment of the group was catalyzed by the Mysore Rehabilitation and Development Agency (MYRADA), a non-profit organization that works in drought-prone regions of south India. MYRADA's approach has been to build the capacity of the poor and vulnerable to craft their own local-level institutions for resource management by developing rural credit systems of self-help groups (SHGs) (Fernandez 1994(Fernandez , 1999. In 1991, MYRADA identified three villages, Guru Malaya (GM) Doddi, Arepalyam, and Bailur, for a watershed development project. Within these villages, MYRADA invited farmers with land in the catchment area, but who lacked access to irrigation facilities to be a part of the project, funded by Misereor, a German funding agency. Misereor provided 80% of the funds and loaned the remaining 20% of the funds to the farmers with the stipulation that the farmers would create a common fund for post-project maintenance of the watershed management structures by repaying the loans.
Deeply rooted norms of caste untouchability hindered group formation. A significant challenge in establishing a group was to convince the Lingayats, a politically powerful and wealthy caste, and the Scheduled Castes (SCs), a historically marginalized caste, to participate in the watershed development program together. Although they had adjoining farm land that would benefit from the proposed watershed intervention, caste and its associated norms that prohibited intermingling of the Lingayats with the SCs was a major barrier towards forming a water-shed organization with non-cooperation from members belonging to both the castes. The GM Doddi Watershed Association (GMDWA) was finally formed in 1994, after 68 meetings and 6 participatory planning exercises. When MYRADA started conducting capacity-building sessions, it also developed a rule that farmers from both castes must participate together, which then gave way to members of both castes eating their meals together during the sessions. Previously, such interaction between the two castes would not have been permitted.
In 2002, members from Arepalyam and Bailur split from the GMDWA. There were three reasons for this split. First, fortnightly meetings were held in GM Doddi, two kilometers (3.2 miles) away from Arepalyam and Bailur, an inconvenience where most transportation is by foot. Second, members from Arepalyam and Bailur claimed that members from GM Doddi rarely attended meetings because they had received private benefits on individual farms, while farmers from Arepalyam and Bailur had received shared benefits. A third reason was that the local bank does not provide loans to savings groups as large as the GMDWA (over 20 members).
In 2002, the AB Prakruti Watershed Association (ABPWA) began functioning with 17 Lingayat and SC members from Arepalyam and Bailur. Today the ABPWA has 14 members, after three members exited the group. Nine SCs are small and marginal farmers with land ranging between two and five acres. Five Lingayats have land ranging between six and twelve acres. Kannada is the common language spoken among villagers, including those in the ABPWA. The index of caste heterogeneity for Arepalyam and Bailur is 0.45, 1 which is considered to be moderately heterogeneous (Adhikari and Lovett 2006). See Table 1 for the demographic characteristics of Arepalyam and Bailur.

Data collection and analysis
Data for this case study were collected from December 2014 to August 2015 in five visits that included two month-long stays from December 2014 to January 2015 and from July 2015 to August 2015. Interviews lasted 45 to 60 minutes each and took place at the watershed office, homes of members, fields, or other loca- 1 We used the method suggested by Varughese and Ostrom (2001), which computes heterogeneity as, 2 ( 1) 1 (P ) , where Pi is the proportion of the total population in the ith race. A varies from 0 to 1 and measures the probability that two randomly selected persons will be from the same race. tions familiar to the interviewees. Starting with an ABPWA leader, interviews were conducted until all the members and the ex-members were contacted. Interviews were conducted with 14 ABPWA members, one ex-member of the GM Doddi Watershed Association and one ex-member of the ABPWA (the other two ex-members declined to be interviewed), five residents from Arepalyam and Bailur, and 6 external agency personnel knowledgeable about the group and context. Interviews with external agency personnel were conducted in person or over the phone and included members from Syndicate Bank, MYRADA, Community Managed Resource Center (CMRC) -an organization linked to MYRADA, and the Krishi Vigyan Kendra (KVK), Kollegal -the extension agency of the Department of Agriculture for Chamarajanagar. Interviews covered questions about the group origin, effect of caste on group formation, group structure, membership, decisionmaking, leadership, benefits of participation, monitoring and accountability, interactions of members within and outside of the group, and skills of the members. The interviews were conducted in Kannada and Tamil, recorded in a notebook, transcribed into a Microsoft Word document, and then into a Microsoft Excel spreadsheet. The principal investigator, who speaks Tamil and English, translated the interviews into English. A field assistant, fluent in Kannada and English, translated interviews conducted in Kannada. The meeting minutes from 2002 to 2015, originally in Kannada, were translated into English by a Research Associate fluent in Kannada and English. Meeting minutes provided information about the decisions and discussions in the ABPWA and were used to conduct follow-up interviews. Following Yin (2003), we triangulated data across 27 semistructured interviews, informal conversations, 55 pages of meeting minutes from the previous 13 years, and observations in four meetings with the ABPWA.
Following qualitative data analysis techniques, data were summarized and coded into themes (Miles and Huberman 1994). First, the main points of the interview were summarized into short notes. Next, the interviews were transcribed along with notations for comments to follow up in future interviews. The transcribed data were hand-coded into broad categories of equity, accountability, symbolic capital, and capabilities. Thus, the data were coded during as well as after data collection (Miles and Huberman 1994). After the first iteration of coding, two additional coding cycles were employed. Following Mackey and Gass (2005), the principal investigator re-coded the data three times, separated by three months' time to increase rater reliability. The rater reliability scores were above 95% across these coding iterations. An iterative approach was followed by using a theoretical lens to guide coding as well as identifying concepts in the data to link back to theory (Strauss and Corbin 2008).

Perceptions of fairness, pride, and capabilities promote collective action
Under conditions of equity and accountability, institutions in the ABPWA create perceptions of fairness. These institutions promote collective action through exclusive membership, inclusive decision-making, impartiality, responding to concerns of the SCs, rotating leadership positions and providing training opportunities, separation of powers, and transparency. Two of these institutions, exclusive membership and transparency, also create community pride that motivates members to continue participating in the ABPWA. Training opportunities also enable members to learn and contribute their capabilities towards group functioning.
Exclusive membership has created trust among members of the ABPWA. Only original members of the GMDWA from Arepalyam and Bailur are members of the ABPWA. The ABPWA rejects requests of interested villagers to join the group, "even if we are very good friends" as one member stated. The only exception to this rule is when a current member dies, the next of kin inherits his/her position. Members said that they do not feel comfortable allowing new members into their group because the group has accumulated a large amount of savings over the past 15 years and that they do not trust new members to repay their loans back to the group. This rule of exclusive membership is now a source of contention in the group.
In January 2015, a new farmer from the Lingayat caste was allowed to participate in one meeting, which gave impetus to a new round of discussions in the ABPWA. The group agreed on two conditions for new members: First, new members have to pay a membership fee of Rs. 1000 ($15) and second, new members have to make deposit of Rs. 25,000 ($384) before applying for a loan. This rule has led to mixed reactions among the members. One of the members said that the new rule has created a fear that if new members are allowed into the group, they will immediately ask for loans. This was driven by the fact that the new member did not attend the next meeting because although she has farmland in the village, she lives in Kollegal, a small town, 60 km away from Arepalyam and Bailur. Others argued that the rule to deposit Rs. 25,000 would prevent new members from joining the group because it is not possible for farmers to make contributions of such a high amount in six months. One member commented, "It is easy to make a rule like this to show that the group is ready to take in new members. In reality, they are just making it harder for new people to enter the group." The current membership rule creates an exclusive group that retains the core membership of the group -those with a shared history who have come to trust each other over time despite socio-cultural heterogeneity. The new membership rule, if it passes, seeks to maintain exclusivity by imposing high costs of entry into the group. Even as there is much contention within the group regarding new membership, the exclusive nature of the ABPWA, in addition to the requests that the ABPWA receives from other villagers, is also a source of pride for the group because it is an indication of its well-functioning nature. Pride, engendered from such exclusivity, contributes towards the collective identity of the group to sustain collective action. Members said that their group is the only functioning watershed group across three villages, which is why there are frequent requests to join the group. Conversations with other villagers revealed that their requests to join the ABPWA have always been rejected.
According to the ABPWA's inclusive decision-making rule, meetings are only conducted if a 90% quorum is present to pass decisions through unanimous agreement among members. Disagreements, if any, are discussed until the group achieves consensus. As one member said, "Even if 5 people say yes and 10 people say no, we won't leave the issue unresolved. We find out why people say yes or no, we find out their opinions about it, so that there is no domination in the group." Impartiality through the equal enforcement of attendance in the ABPWA was created after the split from the GMDWA. A fine of Rs. 25 ($0.39) was decided for non-attendees. The rule was enforced by appending the fine to the loan amount that members had to repay back to the group. One member said, "When it is time to pay the fine, the caste of the member, or the importance of the member in the village does not matter." One Lingayat member paid a fine of Rs. 725 ($11.17) for missing 29 consecutive meetings, while traveling for work outside of the village. The member who paid the fine said, "Although my fine was high, I've received many benefits from the group. If I want to continue in the group, I need to follow the rules of the group." Thus, an equal enforcement of the attendance rule creates fairness because sanctions are implemented irrespective of caste.
Responding to the concerns of the SCs creates perceptions of fairness in the ABPWA. Until 2013, the maximum principal amount for loans was Rs. 20,000 ($302) with an annual interest rate of 2%. The SCs, also the poorest members in the ABPWA, requested a decrease in the interest rate of 2%. After six months of deliberations, the principal amount was increased to Rs. 30,000 ($452) and the interest rate was decreased to 1%.
In 2006, the ABPWA developed a rule to rotate the three leadership positions so that every member has an opportunity to lead the group. Members are also sent regularly for training on record-keeping, auditing, accounting, developing maps, and interpreting legal documents. Once members attend such training, they contribute their skills towards group functioning. These rules of inculcating leadership skills simultaneously create perceptions of fairness as every member has an opportunity to lead the group and attend training activities and also allow members to contribute their capabilities to the group.
Institutions of separation of powers between the Chairperson and the representative, as well as maintaining records and the presence of an external monitor for accounts, create accountability in the ABPWA. The representatives and the members monitor whether other members repay their loans on time and whether the group leaders carry out their duties. An external auditor from the CMRC audits the accounts of the ABPWA every six months in the presence of the members of the ABPWA after members discovered discrepancies in audits conducted by the CMRC. Thus, members are quick to respond to lapses in accountability and put in transparency mechanisms to avoid discord in the group.
The ABPWA also uses graduated sanctions in case members do not repay their loans. Meeting minutes indicate that the ABPWA regularly reminds its members to repay loans. The group also restricts access to government schemes and subsidies until members have repaid their loans. Occasionally, the ABPWA provides a grace The muting and unmuting of caste across inter-linked action arenas 237 period to the members in case of poor harvests from a crop. If members default on their loans, fines are attached to the interest that they are repaying on the loan. Even after this, if members default on their loans, group members visit the loan defaulter's home to demand payment in front of the neighbors. Interviewees said that they have had to resort to the last sanction twice in the past to demand for loan repayment from members who have left the group since then.
An unintentional outcome of accountability is the emergence of symbolic capital in the form of community pride. For example, while describing ABPWA's activities, the members of the ABPWA emphasized the "neat" and "meticulous" maintenance of the transactions, account books, and the receipts. The ABPWA considers the maintenance of accounts as evidence of their success, which has created a sense of pride in the ABPWA. Although the ABPWA has not intentionally crafted such a strategy to generate pride, accountability strategies that create visibility of success do so in this heterogeneous group.

Inability to challenge caste across interlinked action arenas
Since the formation of the ABPWA, interviewees said that there are fewer caste restrictions between the Lingayats and the SCs in the ABPWA and increased bonding between members from both the castes. Yet, caste continues to shape feelings of powerlessness. One SC member said, There are no overt problems in the village due to caste, but mentally there is the feeling that we are not equals. Untouchability is practiced in the village. For example, I cannot enter a Lingayat's house. They will not eat at our weddings. But the group is different. In the group, we sit down and eat together and it feels okay for at least that moment. The group system is different from how things are in the village.
Even within the group, members operate within the boundaries of caste rules. For example, not all SC members speak freely in the group because of norms that discourage SCs from speaking up in front of the Lingayats. During the discussion on adding new members to the group, one SC member deferred to the dominant members of the group, but was immediately encouraged by other members to give his opinion. Some SCs said that before the formation of the GMDWA, SCs would not speak in front of the Lingayats. Now, in the ABPWA, the SCs consider their situation to have improved since 1994.
Even though the ABPWA is free from caste-related conflict and hostilities, the effects of the norms of untouchability and caste rules are reflected in the interactions between the Lingayats and the SCs in the ABPWA. SCs spoke about untouchability only in the absence of Lingayat members and said that a fear of the Lingayats creates hesitancy to speak up during meetings. Even today an SC cannot venture too close or linger outside a Lingayat's house, in case someone from the house might accidently view their faces, which contributes to a feeling of powerlessness. When a member from the SC community accompanied the principal investigator to a Lingayat's home for an interview, he remained behind the principal investigator so that his face would not be visible to the people inside the house. The doors did not open until the SC member announced our arrival to the residents and left.
Arepalyam and Bailur are segregated along caste lines. Soliga tribes live outside the boundaries of the villages of Arepalyam and Bailur, although both SCs and Soligas are employed as laborers on the fields of the Lingayats. Since SCs in the group are employed by Lingayat farmers as laborers, maintaining the norms of untouchability serves to benefit the SCs who might not want to risk forfeiting the additional income that comes from working as laborers on Lingayat farms by speaking up in front of the Lingayats.
Outside of the watershed meetings, the Lingayats referred to their caste with pride. For instance, some interviewees frequently referred to the illustrious lineage of the Lingayat caste and made a hand gesture to depict the dominance of the Lingayats, while indicating that the other castes are non-vegetarians -implying their lower caste status. One interviewee said, "XXX is my best friend and we are members of the SHG. But XXX cannot enter my house. We are Lingayats and we have to follow these traditions. I cannot invite him inside my home. Nobody will allow that and there is no chance that that will happen! The village rules are different." One member said, "The everyday dealings in the village follows the dictates of the village elders who are particular about caste and religion because their ancestors did it before them. It is difficult to express such things to them." Other SC members said that the improved relationship between the Lingayats and the SCs is merely an external demonstration of "friendships." The ABPWA's rules that create perceptions of fairness and symbolic capital in the form of pride have not been able to subvert the social conditioning and the highly-entrenched norms of untouchability that hinder the SCs from speaking up freely in the presence of the Lingayats. Why then do members of both the castes continue participating in the ABPWA?

Enhancing benefits and capabilities
All 14 members mentioned that the watershed structures such as bunds and waste weirs augmented groundwater and reduced soil eroding from their farms, which continue to provide improved agricultural yields even today. Members have benefited from loans, government schemes, information from external agencies. Both Lingayat and SC members also work together to avail of government schemes and group subsidies for transitioning from fuelwood to LPG cylinders, and to purchase cattle and seeds. This access to government schemes and subsidies appears to be an important benefit independent of the watershed management activity.
The ABPWA also works with the local government to build toilets in the village, desilt the lake, initiate community fishing in the village, and build roofs for the school and the village temple; something they could not have accomplished individually, helping to build group capability. The ABPWA's contributions to village activities produces symbolic capital in the form of reputation and community pride. MYRADA and the Department of Agriculture frequently refer to ABPWA as a "model" group in light of its contribution to village development as well as its ability to function despite the presence of two different castes and have given awards to the ABPWA. One interviewee noted, "Our group activities shone. We could function despite having two different castes in our group. Government officials from the agricultural department saw what we were doing and emulated it in other parts of the state. They started inviting us for conducting communitybuilding exercises." Local newspapers also featured the ABPWA's efforts at community-building. Other groups in the region frequently approach the ABPWA to request training on community-building and organic farming, which helped to build member as well as group capabilities. Conducting such events enhances ABPWA's reputation, community pride, and also brings in funds for the ABPWA.
An important aspect of member capabilities is their dynamic interaction with collective action. The ABPWA's success has resulted in other farmer groups, government and non-government agencies inviting the ABPWA to provide training on community-building and organic farming. Even as skills are seen as important for the ABPWA success, the ABPWA's success also helps members develop skills when they participate in opportunities to train other groups. These training sessions generate income for the ABPWA, which in turn reinforces the reputation of the ABPWA as well as broadening the capability of members (by developing skills). Such recognition and the associated economic gains encourage members to continue cooperating in the ABPWA, who take more pride in contributing to the group's activities.

Aspirations to equality
The recognition that comes to the ABPWA has been particularly beneficial for SCs in enhancing their individual capabilities. For instance, an SC is contacted to provide training to other farmer groups as well as government agencies and MYRADA. He is viewed by all members as the most influential person in the ABPWA. He writes the meeting minutes in the ABPWA, attends the Panchayat (village government) meetings, informs the ABPWA of the discussions in these meetings, and represents the group at press meetings. He said, "For years we were told that we are not good enough; that they somehow are better than us. But, the abilities that I have matter more to the group than my caste. People can see what I do, I, a SC, could do, and hopefully, it sets an example that we can also have a chance of doing better, even if we do not belong to a powerful caste." Thus, being a part of the ABPWA provides opportunities to SCs to use their skills and pursue economic activities, creating an aspiration to be considered as equals to the Lingayats. The ABPWA's decision to continue sitting together by keeping aside caste barriers, initially formulated by MYRADA, has led to successful outcomes for the ABPWA in the form of community pride, recognition, honor, collective identity, as well as economic benefits. In the absence of overt caste conflict, reification of the symbolic sense of community and community pride benefits the members from the watershed management activities as well as through the opportunities to broaden the collective and individual capability. Opportunities to pursue economic activities create an aspiration towards equality among SC members, partially attenuating the widespread sense of powerlessness in their community.

Discussion
A key theoretical finding is that institutions foster group activities in a socioculturally heterogeneous context through equity, symbolic capital, accountability, and capabilities in the ABPWA. The group has been successful in setting aside caste for collective action, providing benefits to members, creating impartiality, maintaining accountability, and broadening the capability of its members, but success in challenging the effects of caste-related norms in the group has been gradual and is not yet reflected in every institution of the ABPWA. In the slow and (sometimes absent) challenge to such norms, some of the ABPWA's institutions continue to reproduce and maintain unequal power relations (e.g. where SCs still hesitate to vocalize their opinions in the presence of Lingayat members). In this context, the group's sense of community, community pride, honor, and reputation continue to mask caste-related inequalities.
We had also posed the question of why actors with less power participate in action arenas that disadvantage them. Literature suggests that actors with less power participate in unfair action arenas to 1) avoid costs from other inter-linked action arenas, 2) gain benefits from inter-linked action arenas, and 3) aspire to equality. Norms of untouchability continue to be followed in spirit, if not openly, in the group. The hesitation of the SCs in vocalizing their opinions in front of the Lingayats maintains and reinforces the existing power relationships that exist between the two castes outside of the ABPWA. Even though the norms of untouchability are muted in the group, SCs continue following them to avoid costs of other inter-linked action arenas (for instance, speaking up against a Lingayat in the ABPWA might prevent an SC being hired as a laborer on a Lingayat's farm). This finding supports Kashwan's (2016) argument that the effects of power asymmetries are inherent even in non-coercive institutions and the effects of power asymmetries go beyond the visible outcomes of institutional development and institutional change. One reason that the ABPWA has not been able to completely eradicate this remnant of untouchability from its group is because untouchability is actively practiced and championed outside of the group in both the villages, even among members of the ABPWA.
Scholars have found that communities can be successful in muting and transcending caste towards developing a collective identity (see Lansing 2006;Menon et al. 2007). In the ABPWA, we find caste is muted only in the ABPWA, during the bi-monthly meetings, but not fully. This muting does not address internal differentiations, inequalities, and power asymmetries. Although the ABPWA has been successful in bringing together the Lingayats and the SCs for watershed management and in giving equal space in decision-making to the traditionally marginalized SCs, the legacy of caste and its effects still lives on in the interactions between members in and outside of the group, which prevents complete participation and empowerment of the SCs, despite successful collective action.
Non-material benefits such as visibility of its work, pride, reputation, and collective identity are important to group members because the success of the ABPWA also provides opportunities for SCs to seek economic opportunities, echoing findings by Cochran and Ray (2009). Less powerful actors participate in an action arena where feelings of powerlessness still exist because SC members benefit from the broadening of capabilities and contributing these skills to the group. For instance, an SC member is recognized as an informal and influential leader of the group. Such recognition also gives the SCs the right to participate in village life, which, otherwise, is denied to them through conditional participation in village festivities, and outright exclusions from the homes and spaces near the homes of the Lingayats. Participation in the ABPWA, contributing towards village development activities, developing and applying their capabilities, and pursuing economic opportunities provide SCs with opportunities to be considered as equals, at least in the ABPWA, if not the broader context of the villages. Without the right and ability to fully participate in village life, SCs have found other mechanisms to strive for equality by honing their individual skills and collective capabilities in the ABPWA.

Conclusion
The aim of this study was to examine the underlying conditions under which institutions promote collective action in a socio-culturally heterogeneous group. We built on previous studies that examined the incentives for the participation of less powerful in inequitable action arenas. This work contributes to an analysis of how power asymmetries are reflected in institutions that shape social interactions across multiple action arenas, even while promoting successful collective action. First, we found that institutions that create equity, symbolic capital, accountability, and capabilities can promote collective action, but they do so without addressing caste-related inequalities. Second, we found that the muting of caste in one action arena does not necessarily mute it in other action arenas, where the effects of caste continue to reproduce and maintain power asymmetries. Since caste is such an age-old custom, it spills over into every aspect of village life. Since SCs have to switch between the frequent muting and unmuting of caste in the interlinked action arenas, they fall back into following norms of caste in action arenas, even where caste is muted, out of sheer habit, and because they expect to encounter ABPWA members in the interlinked action arena of village life.
Third, despite the lingering effects of caste in the ABPWA, there have been gradual transformations where both Lingayats and SCs consider an SC as an informal leader. The group's success allows the SCs to pursue economic opportunities that would otherwise not be available to them. This change of attitude, even if it is within one narrow action arena, does have a small but positive impact on the SCs. SCs therefore continue participating in the ABPWA, despite feelings of powerlessness to avail of such benefits as well as an aspiration to be considered as equals. Thus, a question for future analysis is to examine the potential for social transformations that are made possible through the broadening of capabilities occurring in one action arena to spill over into other action arenas that may still be beset with highly entrenched norms of untouchability and power asymmetries that structure social interactions.
Finally, of particular interest is the dynamic nature of symbolic capital and capabilities to not just initiate collective action, but also to sustain it. For instance, successful outcomes of collective action generate community pride that sustains collective action. Community pride can, therefore, become more effective after the initiation of collective action to sustain the collective action effort and can be developed as a by-product of collective action, which then fosters further collective action in a dynamic feedback loop. In another dynamic process, member capabilities can foster collective action, and the collective action can further build member capabilities. Such dynamic processes make it more challenging to tease out cause and effect relationships, but they more realistically represent the social processes and outcomes of collective action.