Legitimate network governance in small-scale fisheries in northern Chile

ABSTRACT By introducing Territorial Use Rights for Fisheries (TURF), the Chilean government has devolved authority over the appropriation of benthic fisheries to local fishers’ organisations. Yet there is little evidence of how this local governance works for membership organisations. Drawing on the theory of lateral network governance, the role of legitimacy in governance outcomes is examined by conducting a comparative case study of two TURF networks in northern Chile. Counterintuitively, more effective governance outcomes were found in the TURF network characterised by a less favourable legitimacy structure of decision-making than the case with a better legitimacy structure. Considering context and network evolution, it is suggested that although organisational renewal and high membership turnover potentially fragment legitimacy, they also enable novel collective action and better governance outcomes. The observed divergence of actual legitimacy from formal governance structure underscores the need for dynamic analysis of collective resource governance beyond the formal chart.


INTRODUCTION
One aim of co-management is to provide organised networks of users with practical means to exercise authority and responsibility over the resources transferred to them by the state (Carlsson & Berkes, 2005).Co-management aims to involve social actors in regulation and decision-making who have previously been excluded from hierarchical management (Ayers & Kittinger, 2014;Cundill & Fabricius, 2010).Co-management is situated in a duality between the formal and the informal.It operates within a regulated legal framework, but at the same time leaves operational space for network governance between different actors to solve collective action problems (Carlsson & Sandström, 2008).
Co-management requires multilevel organisational structures and processes.Networks are often presented as a flexible organisational form for this purpose (Gutiérrez & Glückler, 2022;Sandström & Rova, 2009, 2010).Indeed, social networks are the basic organisational unit of co-management at multiple levels, especially in complex socio-ecological systems, and a key structure for the articulation of social learning (Berkes, 2009;Marín & Berkes, 2010).In this sense, researchers have shown networks to be conducive to information flow, shared understanding and problem-solving in hybrid systems, rather than governance being exercised exclusively by state hierarchies (Armitage et al., 2009;Finkbeiner & Basurto, 2015).
Motivated to close this research gap, this research drew on the lateral network governance approach (Bawarshi-Abarzúa & Glückler, 2023;Glückler, 2020;Glückler & Németh, 2012) and used methods of social network analysis to assess empirically the quality and structure of legitimacy in local fishers' organisations in Chile.One crucial element of effective government of the commons lies in establishing a legitimate framework that enables collective decisionmaking regarding the allocation and management of benthic resources, imposing binding obligations on all stakeholders.Moreover, this consistency not only impacts local social systems but also guarantees adherence to co-management strategies, including assigned quotas and fisheries monitoring, while promoting socio-ecological sustainability within the smallscale fisheries sector.
The rest of the paper is structured as follows.Section 2 reviews the literature on network governance and legitimacy in collective action contexts.Section 3 is devoted to a case study of fishers' organisations dealing with benthic species such as loco or Chilean abalone 1 (Concholepas concholepas, Muricidae), erizo or sea urchin (Loxechinus albus, Parechinidae), and huiro or Chilean kelp (Lessonia berteorana, Lessonia spicata and Lessonia trabeculata, Lessoniaceae) in the context of the Chilean Territorial Use Rights for Fisheries (TURF).In this section a research design and the SONA methodology employed in two local case studies are outlined.The analytical findings regarding the quality and structure of legitimacy in a comparative case and their coherence with effective governance outcomes are set out in Section 4, before drawing some conclusions in Section 5.

Co-management
This paper addresses a general governance problem in which hierarchies, markets and governance networks function in a hybrid way in a collective coordination context (Pierre & Peters, 2000).Governance, conceptually, is concerned with the coordination of actions among multiple stakeholders who share a common goal and who are situated within diffuse boundaries of public-private action (Rhodes, 1996;Stoker, 1998).Following a recent debate, we argue that governance emphasises the processual nature of coordinating and moderating social interactions.Specifically, we define governance as the coordination of stakeholders to achieve consensual goals in situations where these parties are 'interdependently linked by a collective problem' (Glückler et al., 2020a, p. 4).
In the context of solving the tragedy of the commons (Hardin, 1968), that is, avoiding the overexploitation of natural resources, a particular form of governance based on comanagement has become established in many parts of the world to ensure the sustainability of resource use.As co-management is inextricably linked to several levels of organisational structures, it is a form of governance aimed at facilitating cooperative structures and processes between public, private, and community actors (Armitage et al., 2009;Sandström et al., 2014).Typically, co-management is based on a set of formal management arrangements and structures defined as the rights, roles and responsibilities shared between responsible government agencies and user communities (Ayers & Kittinger, 2014).
In Chile, the government responded to the overfishing of benthic resources in the 1970s and 1980s with a series of measures.It was only after several failed interventions that the TURF system, as it is known today, was regulated in 1995 and became operational two years later as the Management and Exploitation Areas for Benthic Resources (Áreas de Manejo y Explotación de Recursos Bentónicos -AMERB, hereafter referred to as 'TURF areas').In this case, the term 'benthic resources' refers to benthic communities that develop in the marine substrate or water column and that have economic value (Crespo & Pardal, 2020).In the Chilean case, the TURF areas system includes rights to access a variety of inshore benthic resources, that is, molluscs, algae, crustaceans and echinoderms (Arias & Stotz, 2020;Castilla & Fernandez, 1998).
In this context, the implementation of TURF transforms a non-excludable common good, such as fishing, into a club good (Buchanan, 1965) or a toll good (Ostrom, 2005) that clearly demarcates potential users (Uchida & Wilen, 2004).However, because it is difficult to effectively exclude third parties, local fishers' organisations repeatedly face problems of poaching, both by their own members and by individuals outside the established system (Chávez et al., 2018).The Chilean TURF comanagement has consequently given fishers' organisations more responsibility and authority, reflecting particular local organisational networks and also interactions of entities at different levels and scales (Albornoz & Glückler, 2020;Marín & Berkes, 2010).
Several organisations have effectively developed the two dimensions of this Chilean TURF as a co-management framework, that is, management and exploitation.In fact, each organisation is free to establish its own internal rules, such as for the provision of benthic resources or the sanctioning of poaching.Over two decades of implementation have generated a variety of internal organisations and rules, depending on the socio-ecological context (Palma & Chávez, 2006).There are currently over 500 TURF areas spread throughout Chile.Yet despite organisational development, these areas are generally not their users' main source of income (Arias & Stotz, 2020;Castilla et al., 2007;Palma & Chávez, 2006).The profitability of fisheries organisations is subject to fluctuations due to both changes in stocks and prices set by the market itself (Gelcich et al., 2017;Oyanedel et al., 2018).
Chile's TURF system, like most co-management approaches, is primarily focused on formal outcomes.Co-management is embedded in the operational structure of public agencies and in existing legal agreements (Sandström et al., 2014).This does not, however, mean that the informal dimensions and rules that emerge in this context are unimportant (Sandström & Rova, 2009).Network governance plays a vital role in the Chilean TURF system, as parties establish cooperative and negotiated links within a framework of willingness and equality between participants (Albornoz & Glückler, 2020;Marín & Berkes, 2010).

Network governance
Network governance is a form of governance based on processes and structures in which actors voluntarily participate in decision-making through negotiation processes (Carlsson & Sandström, 2008;Glückler, 2020;Keast, 2022).Researchers consider networks a more flexible organisational structure than hierarchical governance (Pahl-Wostl, 2015).The basic idea of network governance is to collectively engage in complex management processes that are difficult to control through rigid rules and hierarchies (Hirschi, 2010;Lauber et al., 2008).Within a co-management system, each participating entity can become a node in a network built around the value of interaction between different actors and different institutional arrangements (Carlsson & Berkes, 2005;Carlsson & Sandström, 2008;Gutiérrez & Glückler, 2022;Marín & Berkes, 2010).
In co-management systems, local organisations function as semi-autonomous entities that are granted authorisation by the state to undertake specific responsibilities and exercise a degree of discretion in decision-making (Bixler et al., 2016;Carlsson & Sandström, 2008).These systems can have structural variations that lead users to exhibit different collaborative responses to specific socio-ecological contexts (Albornoz & Glückler, 2020).At the community level, networks serve as a reliable and flexible structure for collaboration in which user expectations and formal management arrangements interact (Newman & Dale, 2005;Sandström & Rova, 2010;Tompkins & Adger, 2004).Local co-management associations are official organisations because they have formal rules and procedures.They fulfil the definition criteria of an organisation from an institutional point of view because they have specific goals, a regulated division of labour, and defined boundaries of organisational action (Mayntz, 2018).
Network governance usually implies that all its members are per se legally equal, regardless of their access to resources and/or power (Glückler, 2020;Lazega, 2000).At the local level, however, communities are not entities made up of homogeneous and monolithic actors.Instead, inequalities in control of organisations and resources are more pronounced (Nunan, 2006;Salmi & Muje, 2001).Indeed, scholars who have studied co-management systems at different levels have highlighted that there are asymmetries in the exercise of power in networks, with certain actors having certain advantages in assuming leadership and representational roles (Brewer & Moon, 2015;Finkbeiner & Basurto, 2015).It is therefore important to identify patterns that allow all parties to control user networks or hinder this goal, as this can affect the success of governance (Bawarshi-Abarzúa & Glückler, 2023;Glückler, 2020).It also helps to understand the principles of network governance that enable effective comanagement systems (Carlsson & Sandström, 2008).According to Sandström and Rova (2009), there is a knowledge gap about networks and how they function as collaborative structures at the local level.Researchers of network governance have suggested that networks are indeed appropriate governance structures, but more information about their internal governance patterns is still lacking (Glückler, 2020;Provan et al., 2007).

Legitimacy in collaborative structures
Following Jentoft (2000), scholars have considered the legitimacy of fisheries co-management from a 'legitimate participation' perspective, examining either if a participatory comanagement system is legitimate or if stakeholder participation leads to greater legitimacy.Within or beyond these two approaches, questions range from user compliance (Hauck, 2011;Nielsen, 2003;Pinkerton & John, 2008) to how local users perceive co-management systems (Hoffman, 2009).Yet researchers have yet to define role of legitimacy in local governance processes.
Once designed, governance networks usually have a 'planned governance' structure, including the definition of formal rules and procedures, as well as positions and organs to enact these rules (Glückler, 2020).In reality, however, the actual workings of network governance often create a 'practised governance' structure to the fore that may eventually differ from the planned (Glückler, 2020).In this context, Lazega (2000) has contributed to the understanding of lateral control in professional partnerships.He has shown that the effectiveness of internal sanctions in an organisation depends not only on formal factors, such as seniority and access to power, but also on informal aspects, such as personal relationships among the members.Hence, a lateral control regime has elements of both planned and practices, formal and informal levels of governance.In the case of local co-management, this sheds light on why state-created organisational structures such as trade unions and cooperatives often deviate from the normative framework (Marín & Berkes, 2010;San Martín et al., 2010).
Social network analysis offers techniques that help scholars to assess the relational structure of transactions and relationships in a network.To study coordination in network governance, legitimacy is fundamental for members to accept decisions.Legitimacy refers to an actor's perception that a behaviour, structure, or decision is desirable and appropriate (Suchman, 1995).Following Weber's concept of domination (Weber, 1978), this implies the acceptance and justification of authority by particular actors.Therefore, one way to empirically assess the structure of governance in a network is to identify the relational legitimacy that members enjoy as well as the overall structure of the distribution of that legitimacy across the network (Flap et al., 1998;Glückler, 2020).Especially in the context of community-level co-management, resource user networks can reveal the origins of those who exercise control, such as local elites (Finkbeiner & Basurto, 2015;Singleton, 2000).Glückler (2020) emphasises the importance of legitimacy in decision-making processes to ensure compliance with collective decisions.He introduces the analytical framework of a 'legitimacy matrix' (Glückler & Németh, 2012), which he uses to capture the strength and structure of legitimacy assignments among network members.Whereas the strength is the total of nominations as a legitimate decision-maker that a member receives from the rest of the network, the structure of legitimacy reflects where these votes come from, for example, whether they stem predominantly from within one of the actor's home factions, or if they originate with members from other cohesive factions of the network.Using his methodology allows scholars to identify positions of local as opposed to global legitimacy.Bawarshi-Abarzúa and Glückler (2023) conducted a legitimacy analysis to assess the effectiveness of governance processes in Chilean rural water supply organisations.They observed a coherent relation between higher levels and more balanced distribution of legitimacy with better governance outcomes.In addition, they discovered that legitimacy was associated with higher levels of collaboration among network members, a finding other researchers have suggested as well (Sandström & Lundmark, 2016).However, Bawarshi-Abarzúa and Glückler (2023) report that dense collaboration is no guarantee for the emergence of legitimate representation.Such deficits can be explained by low internal trust, nascent organisational development, and a network's dependence on external collaborative partners (Ansell & Gash, 2008;Wossen et al., 2013).
Hence, in addition to the formal relational assessment of collaboration and legitimacy, the socio-ecological and organisational context is crucial to understanding the emergence of legitimacy, especially when it comes to access to livelihoods (Bustos-Gallardo & Irarrázaval, 2022;Ramírez-Sánchez & Pinkerton, 2009).For the purpose of this study, the legitimacy matrix is employed to discover actor patterns of legitimate decision-making power within a comanagement system.We consider relevant not only the resulting network structure, but also the socio-ecological and organisational contexts of collaborative relationships for the use of common goods (Finkbeiner & Basurto, 2015;Singleton, 2000).We therefore address the following research question: How are patterns of legitimacy in a governance network of natural resources linked to collaboration and effective governance outcomes?In the following section, we present the research design of a comparative mixed-methods network case study.

Study area and case selection
Those implementing the Chilean TURF system delegate exploitation rights and management tasks to local fishers' organisations, including trade unions, professional associations, cooperatives and Indigenous associations.The Undersecretariat for Fisheries and Aquaculture (Subsecretaría de Pesca y Acuicultura -SUBPESCA) sets catch quotas for each TURF area through annual or biennial legal decisions.The fishers' organisations must submit a 'fisheries biological assessment report' for each allocated TURF area (up to a maximum of four areas per organisation), together with their requested catch quota(s) for the coming season.The fisheries consultants are responsible for preparing these reports.In addition, the Fisheries and Aquaculture Service (Servicio Nacional de Pesca y Acuicultura -SERNAPESCA) controls fisheries exploitation and monitors landings and trade to ensure resource traceability.
This study is focused on the Tarapacá region in the coastal desert, the northernmost part of the TURF system in Chile, where 10 fishers' organisations operate (Figure 1).All have the organisational form of a 'trade union'.Approximately 300 registered users participate in these organisations, which corresponds to 2.1% of the National Register of Small-Scale Fisheries (Registro Pesquero Artesanal -RPA).Compared with other Chilean regions, for example, Coquimbo and Los Lagos, benthic landings are not remarkably high, accounting for 3.1% of the national total in the period 2000-20 (Sernapesca, 2020).In fact, the Tarapacá region has experienced a prolonged period of low productivity due to variable climatic and oceanographic conditions, as well as illegal fishing prevalent in the bays (Espinoza et al., 2020).
We worked closely with two fishers' organisations.To respect the confidentiality agreement, the aliases 'Phisqa' and 'Tunka' are employed, which mean 'five' and 'ten' in Aymara.Phisqa and Tunka share common criteria that are relevant for this study.Both organisations were among the first to participate in the TURF system, conducted biological assessments of the fishery, and regularly reported landings of loco and sea urchin.By 2018, Phisqa and Tunka had almost three decades of experience as fishers' organisations registered in the directory of formal organisations at the national level.In addition, both are an example of how the organisations established the informal rule to distinguish between members that can and cannot be part of the TURF catch quotas.
According to Ostrom (1990), these organisations would face the problem of appropriating a resource for all members, as catch quotas are often uncertain, and productivity may fluctuate over time.The second problem is provisioning, that is, the challenge of regulating demand for the resource to prevent depletion of stocks in the future.In Tarapacá, the fishers' organisations have addressed these problems by giving members with a certain seniority access to lucrative catch quotas, such as loco.As a result, not all members of a fishers' organisation have direct access to quotas in the TURF areas, that is, the oldest members belong to a TURF network and a formal organisation, whereas others belong only to the latter.

TURF networks
In the Chilean context, fishers' organisations operating under the TURF system tend to adhere to the principles outlined by Ostrom (1990).These principles refer to the exclusion of third parties and the establishment of norms and rules for appropriation.The term 'TURF network' is used to refer to the subset of members of fishers' organisations that actively participate in decision-making and activities in TURF areas, as opposed to those that are merely members of the formal organisation.It is important to emphasise that this division arises because members (whether or not they are part of a TURF area) may also have access to other fishing regimes, such as pelagic fisheries.The clear distinction between TURF members and non-members is evident in our results: 31 of 39 members in Phisqa and 20 of 24 members in Tunka were part of the respective networks. 2To illustrate this definition, Ostrom and Crawford's (2005) concept of position and boundary rules is used to explain how the TURF network facilitates agreements between its members.The position rule is based on Chilean fisheries law, which grants each member an official status within the organisation.It determines the member's seniority and the specific badge they have acquired in the RPA.Formally, through their RPA registration, each member may have one or more licences, which certify them as boat owners, fishers, divers, free divers, seaweed collectors, and beach collectors.In practice, however, we found that many members also engaged informally in other fishing activities that are part of their daily routine.
The boundary rule is used to determine member access to specific TURF catch quotas.Although the goal is to achieve equitable distribution among all members of the TURF network, the practical application is somewhat nuanced.The fishers' organisations typically use two methods to allocate quotas.First, the loco quota, which has the highest economic value, is distributed equally.Because not all members have the necessary equipment and permission to harvest, they rely on the diving group and pay a fee for their help.Secondly, they give catch quotas for seaweed and sea urchins to members who want to collect these resources mainly from the inshore.These quotas are usually not transferable to other members.The number of beach collectors in the TURF areas varies according to price fluctuations, especially for seaweed.

Data
For the purpose of our research goal -to capture the local governance context and assess the relational structure of legitimacy -the methodology of situational organisational network analysis (SONA) was adopted, which integrates qualitative, network analytical and visual methods (Glückler et al., 2020b).SONA was developed to emphasise the importance of connectivity and contextuality in revealing structures, actions and social processes in networks, and includes several analytical steps.Our primary fieldwork included preliminary interviews with members of the organisations, participant observation in the organisations' fishing operations, and conducting in-depth interviews with key informants, such as members and consultants (Table 1).In total, 15 interviews were carried out and two group discussions were facilitated, which were subsequently audio-recorded, transcribed and subjected to thematic coding using MaxQDA (VERBI Software, 2021).This information was used to design an appropriate network questionnaire, and then to conduct a network survey on all members of the two TURF networks.After submitting all data to qualitative content analysis (interviews) and social network analysis (network survey responses), we sought for communicative validation of our findings in focus groups.The network survey (see Survey S1 in the supplemental data online) was conducted in December 2018 surveying each member of the TURF network individually at their home.The questionnaire contained 13 questions divided into five categories: • Personal profile: date of birth; level of education; place of origin and residence; and additional occupations in the last five years.• Affiliation profile: year of entry in the National Register of Small-Scale Fisheries (RPA) and formal RPA licence(s); area TURF occupation(s); year of affiliation to the fishers' organisation (trade union); year of affiliation to the TURF areas (as a member of the network); and income from TURF areas in the last five years.
• Collaboration in TURF: participation in fishing labour and/or management activities in the last five years; and mention of members with whom they have collaborated.
• Representation: type(s) of position(s) held as president, secretary or treasurer and duration of each mandate throughout the history of the fishers' organisation.The statutes of both organisations stipulate that representatives' term of office is two years.• Central question of legitimacy: delegation of decision-making authority among members in the TURF network.

Measuring legitimacy
Glückler's (2020) analytical framework of the legitimacy matrix was adopted, along with the recommended methodology for empirical assessment.Legitimacy is empirically observed as the self-reported delegation of decision-making authority to peer members.His central questionnaire item was tailored to the local context of the TURF networks and asked: Imagine that you are unable to participate in an important decision regarding your TURF areas.Which other member(s) of the fishers' organisation would need to be present during the decision-making process for you to accept the agreed outcome?Responses to this question were coded as a one-mode matrix, where each member was recorded as either delegating decision-making authority to the other (= 1) or not (= 0).In a subsequent step, two measures were used to map each member in the legitimacy matrix (Glückler, 2020): • Strength of legitimacy: Strength was measured as the degree of each member, that is, the number of mentions or votes that they have received from all other members on the legitimacy item in the questionnaire.The maximum indegree for each node in a network of size N is N -1.Theoretically, a high indegree reflects the prestige and influence of that actor in the network (Borgatti & Brass, 2019).• Source of legitimacy: Source was measured using the E-I index of each node in the symmetrised network.This index is built on the assumption that the network consists of two or more cohesive groups (factions) and that legitimacy can originate from within or from across a faction.Using the E-I index means taking into account the number of external links (EL) and internal links (IL) for each node, which can range from −1 to +1, where a value close to +1 means that all legitimate links come from outside of a node's home faction, whereas −1 means that all legitimacy comes from within the home faction.
The formula for calculating the E-I index value for each node is: (EL -IL)/(EL + IL; Krackhardt & Stern, 1988).
Factions are defined as a means of dividing the network into n cohesive subgroups that are maximally internally connected and minimally externally connected (Parker & Parker, 1979).Initially, the number and size of factions were determined using the software UCINET (Borgatti et al., 2002) and its method for faction analysis.In addition, the statistical results on the factions were validated with insights from our qualitative research to understand the nature of subgroup cohesion and each member's empirical positions (Zachary, 1977).Finally, with the legitimacy matrix one plots each node according to the source (horizontal axis) and the strength (vertical axis) in a two-dimensional space of legitimacy.Actors characterised by above-average strength and positive E-I values enjoy global legitimacy, whereas those with above average strength and negative E-I values enjoy only local legitimacy from their home faction (Figure 3).The lower two quadrants contain actors with only few or no nominations, indicating their limited decision-making authority (Glückler, 2020).

Methods of analysis
Multivariate regression analysis (ordinary least squares -OLS) was used to assess how the individual members' personal and professional characteristics impacted the strength and source of legitimacy within each TURF network.Several models were developed with the variables listed in Table 2, selecting these variables on the basis of certain observations.For descriptive statistics of the used variables, see Tables S2.1 and S2.2 in the supplemental data online.Researchers have previously shown that members who have been in representative positions for longer and have more experience are more likely to achieve prestigious status in organisations (Bodin, 2017;Bodin et al., 2017;Lazega et al., 2006;Panitz & Glückler, 2020).Those studying Chilean TURF organisations have also shown that members who currently or previously held representative positions have a greater capacity to influence decision-making in fishers' organisations (Gelcich et al., 2006;Rosas et al., 2014).Hence, the year of membership in the organisation and years as a representative were included in the models.
In traditional organisational settings, life experience and seniority based on age often influence social status (Brewer & Moon, 2015).Accordingly, the year of birth was included in the models.In addition, education serves as another pathway to social status through external qualification systems (Arias & Stotz, 2020;Chávez et al., 2018).However, in the context of the TURF networks studied, a strong correlation was observed between years of education and year of birth, suggesting that younger individuals spent more time in the education system (see Tables S3.1 and S3.25 in the supplemental data online).Therefore, these variables were included in different models.
Finally, it is assumed that the tasks and practices that a particular group performs over time influence the creation of trust relationships among peers (Ansell & Gash, 2008;Panitz & Glückler, 2020).Some individuals owned a larger number of official RPA licences, offering them the opportunity to carry out various fishing tasks (Cancino et al., 2007).Therefore, both the number of RPA licences and the number of collaborating members (indegree of collaboration) a member had between 2014 and 2018 were included in the models.

Differential governance outcomes
Given our interest in effective governance outcomes, different patterns of collaboration, and legitimacy in co-management structures, we here offer qualitative and quantitative insights into the differences of organisational context and governance outcomes between the two fishers' organisations under study.
The Phisqa network consists of 31 members, 16 of whom were founding members of the fishers' organisation (trade union) itself, and TURF areas, both of which were established in the 1990s.In fact, this network had admitted only two new members in the last ten years.The membership structure was marked by a strong attachment to benthic fisheries even before the TURF regulation.By conducting the network survey, it was discovered that the founding partners came from other regions south of the Tarapacá region, coinciding with the most intense period of exploitation of the loco through the notion of open access in the 1980s.In 2018, the average age of members in the Phisqa network was 55.7 years.
This situation led to a range of relationships between members with different formally enrolled RPA licences.There were boat owners and hookah divers, who traditionally occupy a central position in extractive activities.In later years, seaweed collectors also became more involved, especially due to the boom in Chilean kelp in the second half of the 2000s (Sernapesca, 2020).However, they differed in that they worked individually, almost separately from other members, as opposed to members who were organised in a technique that requires a crew.The limited influence of the seaweed collectors' group is illustrated by the remarkable fact that during the almost three decades that have passed since the organisation's founding, only one member of this group has ever taken on a representative role (see Table S4a in the supplemental data online).
Examining Phisqa's experience over a decade (2009-18) reveals a contrasting trend.Loco landings decreased by 95.1% between the two five-year periods, whereas sea urchin landings increased by 472.7%.Yet loco's relatively high price point makes it attractive for exploitation by fishers' organisations.During the same period, the regional average prices per unit were CLP 380 (US$0.55)and CLP 60 (US$0.09),respectively. 3To overcome the financial challenges of low income, Phisqa negotiates annually with private companies to receive compensation for potential environmental damage, as the TURF areas lie close to a mining port.The network uses the resulting funds for fisheries' biological assessment of the TURF areas and other activities related to small-scale fishing.
The Tunka network consists of 20 members, including two founding members of the fishers' organisation and with them seven other people who are among the founding members of the TURF areas.Like Phisqa, most of the Tunka residents who have settled in the village originate from the coastal regions of central Chile.However, some members also came from the neighbouring city of Iquique, the region's capital.According to the survey data, the average age of Tunka members was 55.1 years.Crucially, activities within the Tunka network were marked by a higher level of integration, albeit with a division of labour between members.In addition, some network members who were divers and boat owners were actively involved in bay harvesting activities.
Tunka had gone through two processes to admit new members.Five new members joined the network from 2004 to 2008; another four joined from 2009 to 2018.In the five years prior to our network survey, the intake of new members was notable, especially as two young members holding only collector licences had been elected as representatives.These individuals played a crucial role in the establishment of a network-owned seafood processing plant (see Table S4b in the supplemental data online).
In Tunka, according to landings statistics provided by the organisation, loco landings had decreased by 89.7% in a decade , whereas sea urchin landings had increased by 21.2%.As a consequence, the new seafood processing plant helped to raise additional income.The plant is managed by the members themselves.From 2016 to 2019, they were able to process 87.4% of loco landings and 45.6% of sea urchin landings in the TURF areas of Tunka.In addition, they processed landings from other nearby bays (see Table S4c in the supplemental data online).
Loco is the most valuable resource in both networks and requires specialised equipment such as oxygen compressors for diving.Loco quotas are shared equally (sub-quotas) among members, but non-divers must cede their share to harvesters, for example, divers and boat owners.In the Tarapacá region, these harvesters usually charge 70% of an individual member's sub-quota as a fee (see Table S4d in the supplemental data online), resulting in minimal profits for non-divers and little incentive for legal work.To counteract this, in 2016 Tunka centralised the landing of locos, reduced catch fees by 30% and allowed on-site processing.This reform also led to new business relationships with local buyers.In the same organisation, members also discussed the centralisation of sea urchin quotas and landings (see Table S4e online).
Both organisations were established as legal entities and had a formal organisational structure, including different formal positions.Formal position and status have often been associated with the people who control resources, such as boats or other fishing materials.However, we observe differences in the two TURF networks' collective actions, which led to different collective outcomes.Qualitatively and quantitatively, Tunka produced more extensive and larger numbers of collective actions and outcomes than Phisqa.It exploited its benthic resources more regularly than its counterpart; its members managed to articulate a collective structure to reorient the organisation from mere primary exploitation to generating added value for their harvests by establishing a new processing plant.Phisqa succeeded, though to different degree, in maintaining mandatory biological monitoring of privately funded fisheries, although benthic landings had declined over the past decade.Finally, the fact that Tunka's performance and actions were closely observed by Phisqa's members and became one of the group's main discussion points is more evidence of Tunka's pioneering and more effective governance outcomes.

The network structure of legitimacy
To understand how members legitimise each other for collective decision-making, we here offer an analysis of the empirical legitimacy structure and the deviation from the structure of planned governance (formal reporting lines).
Figure 2 illustrates the structure of planned and practised governance as a star-shaped TURF network.Theoretically, a member's decision-making authority (legitimacy) is based on the formal structure of the fishers' organisation (trade union), where the president, secretary, and treasurer assume management responsibility and decision-making power over all tasks of the TURF areas.In Phisqa, the three representatives at the time were founding partners.They had already served several terms in their capacity as representatives.In Tunka, on the other hand, two of the representatives completed their first term as representatives and their first five years in the organisation in 2018.
By assessing centralisation and density for the planned and practised governance structures, it was discovered that legitimacy is not only concentrated in the three current formal representative positions.In contrast to high rates of centralisation of formal legitimacy of around 90% in both cases, the practised governance features much lower values for Phisqa and Tunka, namely 24.3% and 18%, respectively.The observed network density is also lower than the formal structures would lead one to expect -9.7% in Phisqa and 15% in Tunka -whereas the density in practised governance is 6.5% and 9.2%, respectively.Despite these differences, however, planned and practised governance structures are statistically correlated (p < 0.05).
Running the faction analysis, we identified different numbers of potentially cohesive subgroups (factions) in each case of legitimacy structures.From our field observations and validation work with the organisations, we inferred that four factions (internally cohesive subgroups) were empirically appropriate and meaningful for Phisqa, and two for Tunka. 4 In the case of Phisqa, the composition of the factions corresponds with differences in the use and possession of RPA licences and materials: faction 1 includes seaweeds and beach collectors; faction 2 includes fishers; faction 3 includes divers and boat owners; and faction 4 comprises various fishing techniques.In Tunka, the existence of two factions fits with the distinction discussed in this TURF network: faction 1 involves divers and boat owners trading with intermediaries, whereas faction 2 includes seaweed and beach collectors with boat owners working in the seafood processing plant (Figure 3).
In both cases, observed moderate levels of legitimacy delegation were observed.On average, these TURF networks' members named two members as potential delegates.Because the number of outgoing and incoming ties in a network is identical, delegates were also named as legitimate delegates by two persons on average (see Table S5 in the supplemental data online).Moreover, some members were frequently named as legitimate delegates (Figure 3), whereas others were not named at all.Eight members in Phisqa and three members in Tunka were mentioned as potential delegates by more than three members.Regarding inter-factional legitimacy, negative values of the E-I indices of legitimacy were observed in both cases (Phisqa E-I = −0.273;Tunka E-I = −0.600),from which we conclude that members tend to grant legitimacy predominantly to members of their own faction.In both cases, the attainment of global legitimacy depends on acceptance and recognition among factions.
Regarding the network of collaboration, members collaborated in operations with an average of 2.7 people at Phisqa and 5.3 people at Tunka.This difference can be explained by the establishment of a seafood processing plant at Tunka.Members from different backgrounds worked in this plant, resulting in more collaborative ties.In the collaboration networks, we observe a larger tendency for inter-factional collaboration with higher E-I indices (Phisqa E-I = 0.429; Tunka E-I = −0.019)than in the legitimacy networks.This means that there is no tendency to collaborate only with people from one's own faction.
Although legitimate decision-making follows formal organisational structure, several deviations of practised from planned governance have been observed.Additional, there is evidence of a group bias in legitimacy delegation.Members tend to confer legitimacy on members of their own faction.Tunka displayed a denser collaboration network and a larger variety of collective outcomes and actions than Phisqa.Tunka, however, has not conveyed anyone who enjoys global legitimacy.This observation contradicts the expectation that the higher the number of globally legitimate persons in a network, the better the governance outcomes.Searching for an explanation of this counterintuitive observation, in the following section we offer insights into the factors that shape legitimacy positions in both organisations.

Explaining legitimacy positions
In this section, we use multivariate regression models to test for the effects of several factors on an individual's legitimacy position in the legitimacy matrix.In the multivariable regression analysis, experience as an official representative -measured by the number of years in the position -is positively correlated with the strength (indegree) of legitimacy in Phisqa (Table 3, model 1: r = 0.646, p < 0.01).Several members (14 of 31 members) here have held formal representative positions in the last 25 years.Four of them held one position for more than two terms.These people are also among those with the highest indegree of legitimacy.
In Tunka, years of representation had a positive effect on the strength of legitimacy (Table 3, model 1: r = 0.202, p < 0.05).With model 2, however, it is demonstrated that this effect was not stable across the board, as Tunka underwent a high membership turnover with a renewal of half of its members in the last 15 years.The new representatives and those who had reached representative positions had not accumulated as many years as the longest serving members of the organisation, from which one can read the positive correlation between year of birth and the strength of legitimacy (Table 3, model 1: r = 0.068, p < 0.05) (see Table S4c in the supplemental data online).This finding highlights the difference in underlying legitimacy processes between the two TURF networks.Contrary to the assumption that education might influence legitimacy, no significant effect was identified.
From the third set of models in Table 3 (r = 0.643, p < 0.01) and in Table 4 (r = 0.093, p < 0.05), one can see that the degree of collaboration influences both the strength and source (global) of legitimacy in the case of Phisqa.This is probably because new fishers' organisation members had to undergo a multi-year process to join the TURF network.As a result, collaboration can be very restrictive.Indeed, there was controversy within the network over an agreement to collectively harvest locos for the benefit of all members (see Table S4d in the supplemental data online).
In contrast to Phisqa, from the third set of models in Tables 3 and 4, featuring strength and global legitimacy as dependent variables, one can discover no significant effects for Tunka.We argue that this is due to a break in the historical evolution of collaboration, which had  changed over the last years.The new generation of members that had joined the network took over the organisational restructuring by agreeing on new rules for the appropriation of catch quotas and introducing the seafood processing plant.Yet these newcomers' assumption of positions of authority and active engagement did not necessarily imply that incumbent members immediately viewed them as legitimate (see Table S4e in the supplemental data online).
In Phisqa, collaboration among organisation members played a crucial role in promoting long-term legitimacy relationships.Such legitimacy ties had been built over decades by almost the same members of the TURF network.Further, formal positions in the network promoted informal legitimacy.In Tunka, on the other hand, the integration of new members into the organisation appeared too recent to have yet affected the legitimacy structure.The network included no globally legitimate leaders.Given the high membership turnover and the dynamics of collective action in the case of Tunka, even previously repeated collaborations had lost their effect on building legitimacy ties.This observation underlines the importance of context and the development phase of networks when studying legitimacy structure in comanagement organisations.

DISCUSSION AND CONCLUSIONS
The collective action problem of coordinating the use of common goods and resources motivates the question of how resource users can establish effective network governance.One problem is the difficulty of coming to collective agreements, and researchers have proposed the concept of legitimacy and legitimate decision-making as a viable solution for acceptable collective decisions.Building on this notion, we have examined the role of empirical patterns of legitimacy in two governance networks and how they were related to collaboration and effective governance outcomes.
Building on the concept of the legitimacy matrix (Glückler, 2020), the network patterns of legitimacy strength and structure were examined.By analysing two fishers' organisations, we have shown that although they were similar in formal organisation and geographical location, both networks differed in governance performance, collective actions and legitimacy structures.Probably the most surprising result is that the fishers' organisation Tunka, characterised by more collective activities and a better collective performance, had fewer globally legitimate members than the other fisher's organisation, Phisqa.The two fishers' organisations were not only characterised by different expressions of legitimacy patterns, but their observed legitimacy structures also differed from the formal governance structure.Whereas the history of previous collaboration and formal offices accounted for the legitimacy structures in the context of Phisqa's stable organisational environment, it was less helpful to understand the fragmentation of legitimacy in the case of Tunka, which had been affected by high rates of membership and organisational renewal.
We conceive legitimacy as a relational concept that is sensitive to organisational contexts.Accordingly, an in-depth research design was chosen to explore and compare two cases in detail, rather than testing hypotheses over a large-N sample of cases with stylised observations.By combining qualitative interview data and formal network survey data on legitimacy has contributed to bridging the gap between relational structure and meaning (Glückler & Panitz, 2021).Concretely, data on formal network structure as well as qualitative insights from the interviews were used to identify meaningful and valid factions (cohesive subgroups) within the two empirical networks.Similarly, the qualitative in-depth insights provide a more comprehensive evaluation of organisational performance.In turn, due to the limited number of cases covered in this study, the findings at this point cannot be generalised.Instead, we encourage future researchers to examine the impact of quality and structure of legitimacy on governance performance by looking at larger numbers of cases as well as other contexts of governance, including governance of other natural resources (e.g., Bawarshi-Abarzúa & Glückler, 2023), but also of innovation, civil engagement, etc.This study has demonstrated how legitimacy as a meaningful concept can be submitted to empirical measurement and also be included in deductive large-N research designs.
Our findings have led us to three important implications that deserve further attention.First, multiple actors with global legitimacy are no guarantee of consistent governance processes.The role of legitimacy in governance outcomes should not be assumed as deterministic.Instead, legitimacy creates an opportunity for more viable decision-making and more acceptable decision outcomes.Legitimacy is created through long-term interactions based on members' collective efforts.However, certain groups, such as divers and boat owners, may have greater control over the organisation, which occasionally leads to their individual interests taking precedence over collective goals.In fact, formal roles, such as serving as a representative, gave them greater importance than other members (Bodin, 2017;Bodin et al., 2017).The key is whether these legitimate actors with collaborative connections can effectively join with others for global collective action within the network and contribute as relevant nodes in an effective network governance structure (Bodin, 2017;Newman & Dale, 2005;Obstfeld, 2005;Sandström & Lundmark, 2016).
Second, with our case-based analysis we have revealed uneven distribution of legitimacy across factions in the network.The factions with lower legitimacy had little incentive or confidence to engage in collective action.Conversely, gaining global legitimacy requires the actors to maintain formal and informal activities that are visible to the entire network.In one of our cases, however, the collaborative structure reinforced the global legitimacy of some actors who practised fishing techniques that were simultaneously restrictive to other members.Further research is needed to examine the importance of group homophily in legitimacy-based interactions (Newman & Dale, 2005) and the role of different factions in the co-management process (Finkbeiner & Basurto, 2015;Singleton, 2000).
Third, it is important to highlight the potential of Glückler's (2020) concept and methodology of the legitimacy matrix to capture legitimacy dynamics within networks over time.This study covered only network snapshots of two organisations at a particular point in time, and it has recognised that the influence of multiple factors and events, such as high turnover rates in membership or drastic fluctuations in the stocks of key resources, are important for understanding the evolution of member legitimacy.We hope that our discussion will help other researchers to better understand the interaction between attributes in networks and outcomes in the sustainable governance of the commons (Bodin et al., 2017;Carlsson & Sandström, 2008;Gutiérrez & Glückler, 2022;Ramírez-Sánchez & Pinkerton, 2009).

Figure 1 .
Figure 1.Context of Territorial Use Rights for Fisheries (TURF) areas in the Tarapacá region.There are currently 504 areas in force granted to 354 fishers' organisations nationwide.

Figure 2 .
Figure 2. Comparison between planned and practised governance in the organisational networks of the Territorial Use Rights for Fisheries (TURF) area.Note: Quadratic Assignment Procedure (QAP) correlations.Number of permutations: 5000; and random seed: 18,294.

Figure 3 .
Figure 3. Distribution of legitimacy and collaboration in Territorial Use Rights for Fisheries (TURF) networks.Note: QAP correlations.Number of permutations: 5000; and random seed: 18,294.

Table 1 .
Mixed-methods research design and data collection.

Table 2 .
Variables of the network survey on the delegation of decision-making authority in Territorial Use Rights for Fisheries (TURF) networks.
fisher, diver, free diver, seaweed collector and/or beach collector) a Note: The E-I index of legitimacy is the only one that is continuous variable; the rest are discrete variables.a Checked with available official documents.

Table 4 .
Multivariate regression analysis of the E-I index of legitimacy as dependent variable.

Table 3 .
Multivariate regression analysis of indegree of legitimacy as dependent variable.