Identity Conflict? Agri-Environmental Collectives as Self-Governing Groups of Farmers or as Boundary Organisations

Organisations of land managers in landscape management face the challenge of combining the need to foster bonding social capital within their member groups with the need to develop bridging social capital with other stakeholders and linking social capital with public authorities. This paper compares the concepts of self-governing groups, boundary organisations and quangos, to analyse how agri-environmental collectives in the Netherlands navigate their identity in interactions with public authorities and manage potential trade-offs between different forms of social capital. It shows the paradoxical situation that these self-governing collectives have to adopt characteristics of public agencies, in order to meet the demands of the Dutch government and EU legislation, required to gain the trust of the authorities for more room for self-governance. The resulting ‘professionalization’ and enlargement of agrienvironmental collectives is likely to reduce bonding social capital, which in turn is an important asset for effective landscape management. In order to prevent this counterproductive incentive of expecting self-governing groups to behave like public agencies, we recommend to nourish and protect the in-between identity of agri-environmental collectives, to acknowledge their variety, and to allow them to be self-governing groups as well as boundary organisations.


Introduction
Government is rarely completely absent in self-governance of landscapes (Menatti 2017). Rather, forms of interaction, co-operation and collaboration take place between self-governing groups, public authorities and in many cases also other actors (Tyson 2017;Plummer and FitzGibbon 2006). In this context, selfgoverning groups need to manage two types of relations: the relations within the group itself and the relations between the group and outside actors. In this article we explore the tension between these two tasks, felt by organisations that are formed by self-governing groups of land managers, in particular farmers.
Based on Westerink-Petersen (2016) we hypothesize that it is a difficult mission for a farmer organisation to combine the task of organising self-governance within a group of farmers with the task of enabling collaboration between the group and outside actors, particularly public authorities. These tasks can be conflicting as they involve different skills and rules of the game and diverging expectations from members and outside actors. A better understanding of this potential identity conflict is needed, because collaborative (deliberative, interactive, networking) forms of environmental governance are of increasing importance (Bodin et al. 2016;Schusler, Decker, and Pfeffer 2003;Prager, Reed, and Scott 2012;Robins, Bates, and Pattison 2011). When public authorities and other actors are seeking to collaborate more and more with groups of farmers in relation to land management, pressure will be put on those groups and their organisations to change and to adapt to the requirements for that collaboration (Westerink, Jongeneel, et al. 2017;Termeer et al. 2013). Public authorities as well as farmer groups need to be aware of the opportunities and risks associated with such changes.
In this article we zoom in on agri-environmental collectives in the Netherlands and the potential conflict between their identity as farmer groups and their identity as intermediaries between farmers and government. In agri-environmental collectives (from here on: collectives), farmers self-organise activities such as meadow bird protection and maintenance of landscape elements on farmland. In the Netherlands these collectives have increasingly become vehicles for organising collaboration between farmers and the government and the implementation of the agri-environment schemes (AES).
Self-organised landscape management in the Netherlands has been well-documented. For example, the study of Polman (2002) includes the various reasons why local stakeholders founded environmental cooperatives. Glasbergen (2000) investigated the potential contribution of these local, self-organised cooperatives to national policy goals. Stuiver (2008) followed specific groups defending their local environmental practices against national regulation. The strife for self-governance by these groups was further examined by Termeer et al. (2013).  and (2007) described the role of environmental cooperatives in joint submissions and in enhancing relations between farmers and citizens. Jongeneel and Polman (2014) analysed these groups from a political economy perspective in relation to group contracts. Westerink, Melman, and Schrijver (2015) showed how a landscape approach combines with self-governance and Westerink, Jongeneel, et al. (2017) considered agri-environmental collectives as taking on governance tasks in implementing the agri-environment scheme, including spatial coordination of actions by individual farmers. The study of Prager (2015) included contributions such groups can make to coordinating and mediating agri-environmental activities as well as to awareness raising and behavioural change in rural communities. De Vries et al. (2019) studied the role of trust in collective agri-environmental management. However, an analysis of the agri-environmental collectives from the viewpoint of a potential identity conflict as a result of their in-between position is lacking.
In the following section, we conceptualise our concerns about this potential identity conflict in theoretical terms and formulate our research question. After a brief methods section, we illustrate our concerns with a case study narrative on the evolution of local environmental cooperatives to agri-environmental collectives in the Netherlands. After that, we analyse the case study through the lens of the theoretical concepts. The conclusions are dedicated to broader implications for collaboration between public authorities and farmer groups in landscape management.

Theoretical concepts
In this article we combine a number of existing bodies of literature. To characterise the identity of a group of land managers we distinguish three types of organisations: self-governing group, boundary organisation and quango (semi-autonomous agency). The concepts of bonding and linking social capital are helpful to understand the risks and strengths of these identities, while the groups are managing the relations within the group as well as those between the group and public authorities. The theoretical concepts and their literatures are introduced below, but we start with conceptualising the interface between self-governing groups and public authorities as a boundary.

Managing the boundary of self-governance
Well-defined boundaries are central to Elinor Ostrom's conception of well-functioning self-governance of common pool resources (Ostrom 1990(Ostrom , 1999(Ostrom , 2009. Not only does she recommend to define the boundary of the resource and of the group managing it, she also recommends to draw a clear line between the rights to self-organise and government intervention (Ostrom 1990). This boundary of self-governance needs to be recognized and respected by the authorities, if self-governance is to function well (Anthony and Campbell 2011;Ostrom 1990). Boundaries are socially constructed distinctions between categories (Jones 2009). Boundaries enable ordering and making sense of the world, but they also constrain actors in their endeavours (Barth 2000;Hernes 2004). The boundary of self-governance enables self-organisation, but the actions of the group are also constrained by it. This constraint can be severe when governance by the government enters into the domain of self-governance that the group had defined for itself, that is, when the government and the group have different perceptions of the scope of the group's rule-making rights (Westerink, Melman, and Schrijver 2015). Likewise, collaboration between self-governing groups and public authorities is enabled and at the same time constrained by the boundary of self-governance (O'Mahony and Bechky 2008;Emerson, Nabatchi, and Balogh 2012;Sternlieb et al. 2013;Westerink et al. 2016). On the one hand, access to each other's complementary qualities and resources is an important reason to collaborate. On the other hand, different procedures, norms, values, interests, language and culture can be a barrier to collaboration. The enabling and constraining properties of the boundary between self-governance and government intervention evoke boundary management by the government as well as by self-governing groups. Boundary management includes defining, challenging, defending, changing and bridging the boundary as well as coordinating interests and tasks (Miller 2001;Rhodes 1996;Van Broekhoven et al. 2014). A major strategy for boundary management is to appoint or to set up organisations for that purpose.

Three organisations managing the boundary of self-governance
We propose that groups of farmers collaborating with public authorities on the management of landscapes can be characterized in three ways: as self-governing groups defending the boundary of self-governance, as boundary organisations bridging and coordinating across the boundary, or as semi-autonomous agencies (quangos) defining the boundary: as bodies executing public policy (see Table 1). How we interpret the literature, the three types of organisations all have an in-between position but they represent a range of increasing 'publicness' (more characteristics of a public agency), greater distance to the farmers, and decreasing autonomy.
As self-governing groups, farmer organisations need to work on the relations within the group, and to establish good institutions for the self-organisation of landscape management (German 2018). Most of Ostrom's principles for self-governance institutions (Ostrom 1990) concern the relations within the group (clearly defined boundaries, congruence, collective-choice arrangements, monitoring, graduated sanctions, conflict-resolution mechanisms and nested enterprises). In addition, she included a principle concerning the relation with public authorities: these should not undermine the institutions for self-governance (minimal recognition of the rights to self-organise). Part of the role of a self-governing farmer organisation is to defend the boundary of self-governance against governmental policies that interfere with its internal rules and relations. Bonding social capital (see below) is crucial for self-governing institutions to succeed, and institutional arrangements can foster or hinder the development of social capital (Anthony and Campbell 2011;Ostrom 2003;López-Gunn 2012).
A second way of looking at farmer organisations in landscape management is to view them as boundary organisations that focus on the relation between farmers and government. Boundary organisations are formed to manage the boundary between two worlds (Miller 2001). They can be formed by actors on either side  (Westerink-Petersen 2016). By mediating between the two worlds they enable collaboration (Guston 2001). They coordinate joint actions and they translate between actors on both sides with the aid of boundary objects and boundary concepts (Guston 2001;Parker and Crona 2012). These are material and non-material carriers of various and developing meanings, enabling translation and collaboration between groups at both (or more) sides of the boundary (Star 2010;Opdam et al. 2015). Boundary organisations are accountable to all sides of the boundary, but those sides may have diverging expectations regarding the role of the boundary organisation (Carr and Wilkinson 2005). Inevitably, the boundary organisation is involved in the debate about the delineation of the boundary, and therefore, its own role. This makes the position of a boundary organisation a difficult one (Parker and Crona 2012). A third type of organisation engaged in managing the boundary between government and nongovernment is the semi-autonomous agency or quasi-autonomous non-governmental organisation (quango) (Overman, van Thiel, and Lafarge 2014;Antonsen and Beck Jørgensen 1997). Such agencies are founded to carry out public tasks, often ' at arm's length' from a parent ministry, and mainly in an attempt to improve the quality of public services (Overman, van Thiel, and Lafarge 2014). They can be organisations according to public or private law (Van Thiel and Van der Wal 2010). Quangos have a contractual relationship with the parent ministry but the extent to which the ministry exerts control over the agency can differ (Van Thiel and Yesilkagit 2014). Quangos that are closer to the parent ministry are more bureaucratic and have less autonomy: means of control used by the ministry include goals, rules about production and dedicated budgets (Antonsen and Beck Jørgensen 1997). Van Thiel and Van der Wal hypothesised that value congruence between quangos and the parent ministry increases compliance and mutual trust (Van Thiel and Van der Wal 2010). These values concern public sector values such as accountability, expertise, reliability, efficiency and transparency.

Social capital
To gain an understanding of the potential identity conflict that a group of farmers may experience, while managing internal as well as external relations, we use the concepts of bonding and linking capital. There are multiple conceptions of social capital (Wolz et al. 2010;Patulny and Svendsen 2007). We consider social capital as the soft qualities of networks and relationships that enable groups to accomplish things together, including trust, access to knowledge and support, shared values and the capacity to learn and innovate as a group (Bodin and Crona 2008;Lopez-Gunn 2003;Pretty 2003;Van Bommel et al. 2009). Various authors distinguish bonding, bridging and linking social capital (Van Dam 2016;Granovetter 1983;Taylor 2000;Putnam et al. 2004;Guribye 2013;López-Gunn 2012). Bonding social capital evolves within groups of people that are close to each other, such as families and communities, or within more or less homogeneous groups of people with similar interests, such as farmers. Bridging social capital evolves across boundaries between different groups with similar positions of power, such as between farmers and citizens. Linking social capital describes the quality of the relationships between groups with different power positions, such as farmers and public officials. Examples have been described in which improving the relations within groups also enhanced relations with other groups (López-Gunn 2012; Mc Dougall and Ram Banjade 2015). However, bonding, bridging and linking social capital do not always reinforce each other: one form can develop at the expense of another (Guribye 2013;Patulny and Svendsen 2007;Ostrom and Ahn 2008). Ostrom and Ahn (2008) speak of the 'dark side' of social capital when strong bonding ties lead to deterioration of relations with others through for example exclusion, oppression or crime. For this article, the bonding social capital within groups of farmers and the linking social capital between groups of farmers and the government are the most relevant.
The importance of bonding social capital in self-governance is reflected in Ostrom's (1990) design principle of nested enterprises. When groups become too big, ties between members of the self-governing group become weak, because members do not know each other anymore and loose the possibility to review and correct each other's behaviour. Nested enterprises, a structure of collaborating subgroups, may preserve bonding social capital in large groups. The overarching enterprise in which the collaborating subgroups are nested, may then take on the task of dealing with the outside world.
We can thus frame our concerns in theoretical terms. We suspect that self-governing groups find difficulty in combining that identity, and the task of fostering bonding social capital, with the identity of a boundary organisation, and the task of developing linking social capital. We use the concept of quango as a reference in terms of 'publicness' and related public sector values. Our research question is: what are risks and opportunities in terms of social capital of combining the identity of a self-governing group with that of a boundary organisation?

Methods
We present an exemplar of the social phenomena conceptualised above by means of a case study narrative (Flyvbjerg 2006). The case study builds on previous research by the authors on agri-environmental policy in the Netherlands (Termeer et al. 2013;Nieuwenhuizen et al. 2016;Nieuwenhuizen et al. 2014;Westerink, Jongeneel, et al. 2017). One of the authors was involved in the process of revision of the Dutch AES and the formation of agri-environmental collectives. We used qualitative methods for data collection and analysis as we are dealing with 'how' questions rather than 'how many' (Pratt 2009;Silverman 2006). For the purpose of this article, 32 confidential interviews with in total 43 stakeholders were analysed that were conducted between 2013 and 2015 (see Table 2), before the implementation of the new AES in 2016. Most respondents gave their permission for making a sound recording of the interview. If not, notes were taken during the interview. Due to budget constraints verbatim reports could not be made, the often extensive reports were sent to the respondents for comments. The corrected reports were submitted to a content analysis (Miles and Hubeman 2013). The reports were coded for the theoretical concepts elaborated above, complemented with bottom-up coding for additional insights (see Annex 1). The analysis of the interviews and the experience of the authors was supplemented by consulting policy documents and research literature.

Case study Founding and development of local environmental cooperatives
Farmers in the Netherlands formed Local Environmental Cooperatives (LEC) and more recently Agri-Environmental Collectives to self-organise landscape management (Glasbergen 2000;Polman 2002;Westerink, Opdam, et al. 2017). The first local environmental cooperatives in the Netherlands appeared in the nineteen eighties, and many more were founded during the nineteen nineties (Oerlemans, Hees, and Guldemond 2006). Many LEC were founded in response to agri-environmental policies, including AES (Oerlemans, Hees, and Guldemond 2006;Polman 2002). By means of such policies, the Dutch government tries to influence rural landscape management towards conserving biodiversity in line with the EC Birds and Habitats Directives (EC 2009(EC , 1992. A few of the first LEC originated from protests against environmental policies; not so much against their goals, but rather against the way these were implemented (A54) (Polman 2002;Stuiver 2008). In addition to agri-environmental issues, many LEC were (and remain) active in broader rural development projects, such as the promotion of rural tourism (L5, L6). Between 2000 and 2003, LEC functioned as intermediaries between farmers and government, drawing up collective management plans for comprehensive areas with meadow bird protection and taking care of recruitment and payment of participants (A31). However, in 2003 the European Commission forced the termination of this arrangement, because according to the regulations of the common agricultural policy (CAP), the end-beneficiaries of the subsidies could only be individual farmers, not private intermediate organisations (Oerlemans, Hees, and Guldemond 2006). Other common self-organised activities of LEC included training of participants and monitoring ecological results, often with groups of local volunteers. Some LEC welcomed such citizens as members.
In 1999, the first regional umbrella organisation of LEC was founded ('In Natura'), followed by a national umbrella organisation 'Natuurlijk Platteland Nederland' in 2003 (Oerlemans, Hees, and Guldemond 2006). The wish to regain self-governance in the implementation of the AES in their working areas led the LEC and their umbrella organisations to keep discussing this with the Ministry of Agriculture.

Negotiating a collective agri-environment scheme
The upcoming reform of the CAP of 2014 opened a window of opportunity. Four leading LEC convinced the ministry of the need to experiment with self-governance in relation to CAP (A32, A34, A51, C1, M1, N4). These so-called 'CAP pilots' included: the design and monitoring of new management options, spatial coordination of measures on multiple farms, and control and payment organised by the LEC (Terwan and Rozendaal 2014). That these experiments were ongoing, strengthened the ministry in its negotiations with the European Commission about the design of a collective AES for the Netherlands (M1, M2).
A very strong argument for reforming the Dutch AES was the lack of effectiveness in ecological terms (RLI 2013; ECA 2011). The influential Dutch council RLI (Raad voor de Leefomgeving en Infrastructuur) advised to concentrate agri-environmental management in large, comprehensive areas and to improve the quality of the management. To enhance effectiveness, the government focussed the new scheme on specific species and on areas where management could be expected to be successful. After a fierce debate between ecologists and farmers about the cause of the 'failure', the LEC incorporated the effectiveness argument in their plea for more self-governance. The LEC argued that their knowledge of local situations and the farmers, and their presence in the area would be indispensable for making the AES more effective through spatial and temporal coordination of the measures (A52, L2, L3, L4, L6, L7, L11, N3, P3, S1). Because of the social capital they have in their areas, LEC have been reported to be able to enhance participation in the AES as well as the quality of the landscape management ). In addition, more self-governance seemed an opportunity to save on public expenditure. The four LEC that undertook the CAP pilots predicted that self-governance by LEC would significantly reduce transaction costs in the AES, because they expected that LEC would be less bureaucratic and therefore more efficient than the paying agency Rijksdienst voor Ondernemend Nederland RVO (A32, L7). On the one hand, respondents were concerned that control and sanctioning by smaller LEC would be hampered by strong internal ties (L6, L7, M1, P3). On the other hand, respondents felt that control and sanctioning were likely to be accepted more easily from a trusted LEC than from a distant public agency (A1, A31, A21, A33, A51, A51, C, L4, P3).

Larger and more professional agri-environmental collectives for more self-governance
The four LEC of the CAP pilots claimed that they were ready to take on responsibility in the implementation of the new AES (A1, A32, and also LEC1: L11, L12, P2). However, these four were relatively large and wellinstitutionalised bodies with professional staff. By 2012, in almost all areas of the Netherlands a LEC was active, but some had started only recently. Boards and staff of most LEC consisted of volunteering farmers and many LEC had one paid staff member, a (part-time) field coordinator. According to the ministry of agriculture, LEC in the Netherlands would need to 'professionalise' in order to be capable to handle the responsibility of self-organising the implementation of the new scheme (Dieleman, Van Drooge, and Manhoudt 2013). Thereby, it was argued that professional LEC could only be efficient, if large enough in terms of number of members and the amount of public money managed. Hence, the ministry set a target for the LEC to form about 40-60 larger agri-environmental collectives that together would cover the whole area of the Netherlands. This urged LEC in regions to discuss collaboration or mergers (A1, L4, L5, L7). This was not always easy because of fear of competition within and between collectives (L6, L7, P1, U1). Most collectives hired staff to take care of the increased workload and to improve the coordination. See Table 3 for a comparison of LEC and collectives.
The LEC of the CAP pilots proposed a role division between the public agencies and the collectives which they illustrated with the metaphor of 'front door and back door' (see Figure 1). The 'front door' represents the relation between the public agencies and the collectives. The government would no longer make detailed agreements with individual farmers about agri-environmental management, but with collectives, based on a general collective management plan composed by each collective for its working area (this plan spatially specifies goals and management actions). The 'back door' represents the relation between the collective and the participating farmers, who must be or become members to be able to participate. The collective would select participants for carrying out the collective management plan and would make contracts with individual farmers about agri-environmental measures on their farms. It would take care of controlling the compliance of the farmer's actions with the agreement, sanctioning and payment. No government Working area Local-regional Regional

Administrative burden
Low High intervention or control would be needed ' at the back door'. This was believed to reduce transaction costs considerably, because the collectives could draw on their bonding social capital and local knowledge (Dieleman, Van Drooge, and Manhoudt 2013). The umbrella organisations formed a project organisation Stichting Collectief Agrarisch Natuurbeheer (SCAN) to support the collectives in their transition (see Figure 2). This project organisation developed a large number of tools and standard documents, including a handbook for boards of collectives, contract templates and a geographical information system for administering the agri-environmental measures (L6, M2, P3). These tools facilitate the collectives in their work, including the communication with and the transfer of information between public agencies and farmers. An independent foundation was set up (Stichting Certificering SNL) and a certification scheme was developed to set standards for the quality of collectives in terms of landscape management, organisation and administration (BIJ12 2020). Only certified organisations can apply for the subsidies. Accountability was a main issue in the terms of reference for certification (Vullings 2012). This seems to imply that a certain degree of 'publicness' was expected from collectives (A2, L4, L7, L12, P1, P3, R1). Nevertheless, both the ministry and the LEC wanted to avoid that the collectives would become quangos (S1). In that case, the administrative burden for collectives would be very heavy. The collectives were to be private law organisations with members (mostly associations). The contracts with the government were considered as public law agreements and the detailed contracts between the collectives and the farmers as private.

Disappointments for collectives and their members
Working towards implementation of the new AES in 2016 involved the sorting out of many details between the Ministry of Agriculture, the paying agency RVO, the provinces who since the decentralisation of nature policy in 2013 were officially responsible for forming agreements with the collectives, and the European Commission (M1, M2, P2, R1). The results were in part disappointing for LEC and their members. At first, the LEC were told that control of individual farmers by the public agencies would not be needed for certified collectives. However, this was not accepted by the European Commission because of rules on financing, management and monitoring of the CAP (EU regulation 1306/2013, EU delegated regulation 640/2014, EU implementing regulation 809/2014). The European Commission demanded detailed administration of landscape management at the level of field parcels. As a result, there is now more bureaucracy around the 'back door' than intended (A2, L7, L12, M1, M2, P3, see also Figure 3). Collectives and participating farmers experienced this as frustrating, while it resulted in more work and higher transaction costs for the collectives than anticipated. The collective as well as public agencies control individual farmers. This government interference with the 'back door' is by the collectives experienced as an intrusion into the domain of self-governance.
Another disappointment was the exclusion of farmers from participation in areas with little chance of ecological success of landscape management (L5, L6, L7, L11, U1). The national government had made the choice to focus on feasible areas to enhance the effectiveness of agri-environmental management, but the message was to be brought by the collectives because of their responsibility for recruitment. Nevertheless, the collectives themselves wanted to improve the results of the management and urged their members in eligible areas to choose the more effective management options. As this implied the higher priced management options, and the new scheme came with budget cuts (L2, L4, L6, L7), not all management proposals by farmers could be funded. Despite the restrictions, the collectives started implementation of the  new AES in 2016. According to an assessment of the collective management plans, the collectives intensified and diversified the landscape management, improving conditions for recovery of biodiversity (Melman et al. 2016). The first formal evaluation of the new scheme is expected in 2020.

Agri-environmental collectives as self-governing groups
The example of the agri-environmental collectives illustrates that shifting boundaries of self-governance can be a long-term process which requires patience and endurance (Termeer et al. 2013). By 2016, several of Ostrom's conditions for well-functioning self-governance were met: (i) the collectives work in a clearly defined region and the members define the group; (ii) these members can take part in the design of the internal rules of statutes and bylaws and appoint the board; (iii) the recruitment of participants for the agrienvironmental management, monitoring of results, data management, control of compliance, sanctioning, internal conflict resolution and payment are all done by the collectives (see Annex 2 for governance tasks that respondents attributed to collectives); (iv) the collectives are organised in nested structures of the old LEC or working groups, the regional collectives, and the national umbrella BoerenNatuur (see Figure 2). However, not all the rulemaking is an internal affair: the main part of the rules of the AES, including the rules for reporting, is still made by the government (EU and national). In addition, the government still carries out controls at farm level. This constrains the room for self-governance by the collectives; farmers and government disagree on the extent of the right of the farmers to self-organise. The boundary of selfgovernance is still under construction.

Agri-environmental collectives as boundary organisations
Agri-environmental collectives are not only self-governing groups of farmers, they are also boundary organisations that enable collaboration between farmers and government (A1, A2, L1, L2, L6, L11, L12, N2, P2, P3, R1, S1). The collectives use boundary objects for exchange and translation between farmers and government such as geographic information systems and contracts. The metaphor of 'front door and back door' illustrates their two-sided accountability (L11, M1, M2, N2, R1). The collectives have an agreement with the government and agreements with the farmers (see Figure 3). These two sides hold diverging expectations regarding the collectives. Respondent L1 was worried that in future the collectives will be mangled between farmers and government. It is known that the two-sided accountability and the hybrid identity make it hard for boundary organisations to 'do things right', to please their stakeholders at both sides of the boundary, to position themselves and to communicate their successes (Parker and Crona 2012).

Agri-environmental collectives as quango's?
The demands by the government with regard to 'professionalism' as defined in the certification scheme for the collectives, expressed typical public sector values such as transparency, accountability and efficiency in handling public funds. 'Professionalization' of collectives implied development of administrative skills, implementation of a GIS tool, appointment of paid staff, improving the quality of the collective management plans, and control of compliance of participants according to certain standards. In effect, even though the government tried to avoid an official quango status for the collectives, it was expected that collectives would adopt characteristics of a public agency. However, even when some tasks of the collectives are considered 'public' -for example because the delivery of public goods such as landscape and biodiversity is at stake, or because the environmental management is funded by tax payers' money -it does not follow automatically that public tasks can best be carried out by (semi-)public organisations, or according to their ways.

Implications for social capital
The collectives have the ambition to embody the self-governance of farmers in agri-environmental management (A32, A43, A54, L6, L7, L12, S1). However, defending the boundary of self-governance, and resisting the governmental interpretation of that boundary, may harm the collaboration with the government. In order to gain the government's trust and to develop the linking social capital that was needed to be given more responsibility in the governance of the scheme and gain more room for selfgovernance, collectives had to adapt and to develop characteristics of a public agency. However, because of that, there is a risk that farmers can no longer identify with or feel represented by their collectives. In that case, the bonding social capital that forms the basis of the success of collectives in organising collective agri-environmental management could become eroded.
This bonding social capital is already under pressure as a result of the transition from smaller LEC to larger collectives. Several respondents expressed their concerns about this (A1, A32, L3, L4, L5, L7, M1, S1), although some also see the advantages of a bigger size (A31, A33, A51, L6, P3, U1). Ostrom (1990) pointed at the risk of losing bonding social capital and self-governing capacity in larger groups, and she saw that many groups solve this through nested structures. For exactly that reason, many agri-environmental collectives preserved their predecessing LEC, either as full organisations or as subgroups within the collective. Other collectives chose to fill in the gap with so-called 'regional coordinators' for subareas, who often have a central role in communicating with individual farmers and in knowledge transfer (A1, A32, A33, L5, L6, L7, S1). As long as these subgroups and regional coordinators manage to maintain bonding social capital, the collectives can remain successful.
The case of the Dutch agri-environmental collectives suggests that developing one form of social capital may occur at the cost of another form of social capital. The development of linking social capital in the relation with the government came with risks for bonding social capital within the groups of farmers. An additional risk is that the emphasis on developing linking social capital, which involved the investment of large amounts of time (A32, A51, L6, M2, P1, R1), could endanger the development of bridging social capital of collectives with other parties in their area, such as citizen groups and nature organisations (A1, L4, L6, L11, N2, N3, N4). Because of fragmented landownership and multiple interests in most rural landscapes, such bridging social capital might be needed for successful landscape governance.

Conclusions
Organisations of land managers in landscape management face the challenge of combining the need to foster bonding social capital within their member groups, with the need to develop bridging social capital with other stakeholders and linking social capital with public authorities. In this article we have looked at the evolution of the Dutch agri-environmental collectives, which try to combine the identity of a self-governing group of farmers with the identity of a boundary organisation that enables collaboration between farmers and government. However, in order to develop linking social capital (gaining the trust of the government to be able to self-govern landscape management) the collectives had to change: they had to become larger and adopt characteristics of a public agency. This, combined with government interference into what the collectives had defined as the domain of self-governance, came with a risk for bonding social capital within the collectives.
For self-governing groups of land managers collaborating with public authorities in landscape management the following lessons can be drawn. It may be necessary to adopt the identity of a boundary organisation in order to develop the linking social capital needed to acquire the desired room for self-governance. When developing linking social capital, care should be taken not to neglect bonding social capital which is the most important resource for self-governing groups. Nested structures can help to maintain bonding social capital in larger groups. When considering their role and identity, we recommend that farmer groups prioritise the safeguarding of bonding social capital within the group of farmers. In that respect, the identity of an organisation of self-governing farmers should prevail over the identity of a boundary organisation. However, developing as a boundary organisation is not necessarily a bad thing: ways need to be found to reconcile and navigate both identities. Awareness of the dilemma is a first step. After that, farmer groups can learn to position themselves more deliberately, depending on the situation (see also Parker and Crona 2012). At times, strengthening bonding social capital or defending the boundary of self-governance may be required; other instances may demand the development of linking social capital and require mediation, translation, collaboration and coordination in the relation with public authorities. For public authorities collaborating with self-governing groups of land managers there are also a couple of lessons. There is an increased acknowledgement of the value of bonding social capital within selfgoverning groups of land managers in preservation of landscapes and their biodiversity. Trying to make use of that resource, public authorities develop policy arrangements such as financial incentives for collective management. With most power and financial resources at the side of government, public authorities can make demands towards the self-governing groups. However, it can be counterproductive to expect selfgoverning groups to behave like public agencies (see also Van Dam, Duineveld, and During 2015). The case of the Dutch agri-environmental collectives suggests that this may threaten bonding social capital within the self-governing group, which was the resource that the government wanted to tap into in the first place. It may also threaten the collaboration: if the identity conflict of self-governing groups/boundary organisations is deepened because of governmental demands, conflicts between such groups and the government are likely to increase. In turn, this also threatens linking social capital in the sense of land managers trusting the government. It can foster the dark side of bonding social capital: a kind of group think that sees confirmation of the idea that the government cannot be trusted. Therefore, similar to selfgoverning groups, also public authorities need to carefully navigate the boundary. It can be wise to take trust as point of departure in the collaboration with farmer groups (De Vries et al. 2019). In the case of the collectives, the controls of individual farmers by the government on top of controls by the collective was felt as a lack of trust in the collective. Self-governing groups must be given time to develop the skills of a boundary organisation. Anthony and Campbell (2011) provide suggestions as to how public authorities can support well-functioning self-governance: by providing legitimacy to common pool resource management rules, helping to monitor resource use, and providing platforms for conflict resolution. However, as the Dutch agri-environmental collectives illustrate, needs and capacities of self-governing groups are likely to differ. Support by public authorities therefore must be agreed by, and tailored to the needs of the selfgoverning groups.
Further research could look into the development of the Dutch agri-environmental collectives since the implementation of the collective agri-environmental system in 2016; whether the collectives succeeded in maintaining bonding social capital as well as in developing linking and bridging social capital with public authorities and other parties, and which mechanisms explain this performance.

Annex 1: Codes
Bonding social capital, bridging social capital, linking social capital, farmer group/representation, boundary organisation, boundary objects, quango, identity conflict, self-governance ambition, self-governance tasks, responsibility for ecological results, front door/back door, self-governance vs government, risk of selfgovernance, public authorities find it hard to let go, role of field coordinators, disappointments, CAP pilots self-governance, negotiations with 'Brussels', collaboration of LEC in collectives, competition, expectations from collectives by government, enlargement needed, risk of enlargement for bonding social capital, budget cuts.
Annex 2: Envisioned self-governance tasks of agri-environmental collectives according to respondents A1, A31-34, A4, A51, A54, C1, L6, L12, M1, N2, N3, P2, P3, R1. Achieve results/ecological quality, develop a vision, develop a code of conduct/clear rules, develop and adapt collective management plans, define management options, local tailoring, recruit/select and contract participants, coordinate landscape management in time and space, extension, supervision of participants, administration of management, monitoring of results, payment of participants, control, sanctioning, set up an appeals committee, organise meetings for members, be transparent for members, discuss plans with members, develop knowledge, maintain the network/develop bridging social capital, safeguard the boundary of self-governance.