Negotiating regime complexity: Following a regime complex in the making

Abstract This article broadens the understanding and empirical study of regime complexes by shifting the focus from the negotiation outcome to the processes of negotiating new international agreements. Although they are important to regime-complex formation and delimitation, the sites where states negotiate new agreements are rather neglected. We aim to enhance the methodological toolbox available to scholars studying global governance in two ways: (1) by demonstrating how dynamic relationships between states and international organisations (IOs) unfolding within the social space of international treaty negotiations contribute to regime-complex formation; and (2) how social network analysis (SNA) can help us to detect patterns in these relationships. Combining participant observation and collaborative event ethnography (CEE) with social network analysis, we present new empirical material illustrating how we delimited a regime complex and how IOs interact throughout the negotiation process. We applied our methodology to the case of marine-biodiversity governance and use observational data collected during three intergovernmental conferences (IGCs) (2018–19) on a new treaty for the conservation and sustainable use of marine biodiversity beyond national jurisdiction (BBNJ) for our analysis. We discuss the results in relation to our approach’s strengths and weaknesses and implications for future research on regime complexity.


Introduction
Global governance and international relations have witnessed a steady increase of international organisations (IOs) governing particular issue areas at the international level. 1 This phenomenon has been described as 'treaty congestion' , 2 'institutional linkages' , 3 'clusters of regimes' , 4 'conglomerate regimes' , 5 'correlated regimes' , 6 'networks of regimes' , 7 'regime interplay' , 8 a 'patchwork of international institutions' , 9 'organizational ecology' , 10 'fragmentation' , 11 and 'polycentricity' , 12 but at this point scholars of International Relations widely use the concept 'regime complexity' to capture the proliferation and fragmentation of institutions, agreements, and sites where states negotiate legal text to codify the values, norms, and rules that should govern a particular issue area. 13Empirical studies have shown that regime complexes are inherently dynamic, consisting of different phases 14 and levels of IO involvement, 15 where IOs are considered to be both important elements of regime complexes as well as central actors within them. 16uring regime-complex formation and the different stages of their consolidation, IOs tend first to compete and then to specialise, 17 which may lead to a division of work and create synergies between regimes and their elements, 18 but which could also encourage forum shopping or regime shifting. 19o far, empirical studies have largely produced descriptions of established regime complexes using analyses of mandates 20 or treaty text citations 21 to describe the units (e.g.legal agreements, 22 regimes, 23 IOs, 24 etc.) and connections (e.g.overlaps of governance topics, 25 conflict clauses in treaty text, 26 etc.), but central questions concerning the delimitation 27 of a sample of regimes into one analytical unit (a regime complex) and how the connections between the regimes form 28 remain unanswered.Although negotiations play an important part in regime-complex formation, 29 as actors tend to use intergovernmental negotiations, such as the Conferences of the Parties (COP) of the United Nations Framework Convention on Climate Change (UNFCCC), to exercise influence on the regime under discussion and its interlinkages with other fora, 30 negotiation sites themselves have remained rather neglected in debates about regime-complex formation and delimitation.This has come at the expense of knowledge about the delimitation and the agency of the involved IOs at the negotiation stage of regime-complex formation, where IOs seek a role and position in governing a specific issue area.
We advance this debate by showing how a regime complex forms through the interactions of IOs and states during negotiations, addressing the important role that intergovernmental negotiations play in the process of regime-complex formation. 31Thus, we see regime-complex formation as a process to which negotiations contribute through the delimitation by states and the agency of IOs.Conversely, we see negotiations as a process in which a regime complex can form because states mention which IOs they see as part of the regime complex, and IOs express how they relate to the issue area under discussion.This process has different phases itself (i.e.rounds of negotiations) and does not preclude that IOs become part of the regime complex before or after negotiations conclude.With a view on regime-complex formation happening during the negotiations, we aim to complement existing research which largely focused on negotiation outcomes such as legal documents and texts. 32ombining participant observation and collaborative event ethnography (CEE) to collect data on the actors and processes in negotiations with social network analysis (SNA), we present new empirical material illustrating (1) how a delimitation of the constituting elements of a regime complex based on states' statements takes place, and (2) how IO interaction is structured.We aim to strengthen the methodological toolbox available to IR scholars for studying global governance by demonstrating both how dynamic relationships and networks between states and IOs unfolding within the organised social space of negotiating legal text contribute to regime complex formation and how SNA can help us to detect patterns in the behaviour of state and non-state actors.
We applied our methodology to the case of marine-biodiversity governance and use observational data collected during three intergovernmental conferences (IGC) (2018-19) on a new treaty for the conservation and sustainable use of marine biodiversity beyond national jurisdiction 23 Stephen D. Krasner, 'Structural causes and regime consequences: Regimes as intervening variables' , International (BBNJ) 33 as an entry point for our analysis.The new legally binding instrument is an implementing agreement of the United Nations Convention on the Law of the Sea (UNCLOS) and has the objective to establish a coherent legal framework for the management and conservation of biodiversity in the high seas beyond the boundaries of nation-states. 34The BBNJ negotiations were distinctive in that, although the BBNJ negotiations took place under UNCLOS, it was widely recognised that the new agreement will interact with many existing treaties and IOs. 35As one state delegate put it, 'we need to be aware that we are creating a regime complex.This treaty will not function alone but will speak to other regimes and in relation to others' (own observation, 20 August 2019).Accordingly, the BBNJ negotiations were shaped by governmental attempts to embed the new treaty within the existing legal framework governing marine biodiversity and to define relationships to other IOs in this particular issue area. 36his article is structured as follows.The second section, by reviewing existing literature, demonstrates how new empirics can contribute to understanding how regime complexes form and discusses how we built on existing methodological approaches that focus on international negotiations.The third section describes the methods and data used for the empirical analysis, and the fourth presents the results, demonstrating how the IOs within the current marine-biodiversity regime complex can be identified, and how they interacted throughout three rounds of negotiations.Lastly, we discuss the results concerning our approach's strengths and weaknesses, and implications for future research on regime complexity.

Understanding how regime complexes form
The proliferation of international treaties and organisations 37 has been followed by literature using different concepts to address the increasing complexity of global governance.This literature is based on observations that IOs are 'embedded in a larger web of international rules and regimes' . 38austiala and Victor introduced the concept of a 'regime complex' , 39 which has found popularity in the field of global environmental governance. 40A regime complex is defined as 'an array of partially overlapping and non-hierarchical institutions governing a particular issue area' . 41Literature on regime complexes has identified as topics to be addressed: (1) a lack of clarity on the delimitation of elements; 42 (2) a gap in describing the dynamics between IOs; 43 and (3) a lack of active fieldwork, as most studies so far have relied on desk research. 44This section reviews relevant literature to clarify our contributions.

The problem of regime-complex delimitation
One defining part of a regime complex is the 'array of … institutions' , 45 raising the question of how to delimit this array: where does it start, where does it end -or which IOs are in and which are out -and what makes them a regime complex?When the boundaries between regimes become blurred, 46 it is difficult to define a delimitation of a regime complex 47 or to assess which IOs are 'in the same population' , 48 which has led to varying mappings of the same regime complex. 49ddressing these questions, Hollway shows that it is the dependence among regimes that makes a set of regimes complex. 50Orsini, Morin, and Young argue that 'perceptions draw the boundaries of the complex, indicating which regimes are recognized or not as elements of a complex' . 51erception becomes key when approaching the delimitation problem in the phase of intergovernmental negotiation.In this phase, the regime complex is not yet codified in adopted legal text and the elements 'governing a particular issue-area' are still up for debate. 52It is not yet clear which IOs will govern the issue area, but instead IOs compete for governance tasks and states favour diverging IOs according to their interests. 53he delimitation of the regime complex is not predetermined but is often the outcome of international negotiations or other state or organisational deliberations such as forum-shopping. 54In any case, a delimitation that considers only institutions that possess formally codified overlaps risks falling short of understanding that the regime complex exists not necessarily only through legal documents but also through the perception of actors, 55 and that regimes may develop in interdependence to each other even without codified relationships. 56espite the scholarly recognition of these factors, the delimitation problem of regime complexes has not been sufficiently discussed in the literature, which has largely focused on studying established regime complexes.These accounts relied on arguably rather static legal documents 57 and did not consider social data from within negotiations, thus not accounting for the dynamics of international negotiations. 58IOs may interact and compete during negotiations without becoming part of the codified regime complex.In summary, there has been a lack of work supported by empirics that can broaden the delimitation of elements of the regime complex to acknowledge the IOs that governmental negotiators perceive and consider as relevant. 59This study thus aims to contribute to the existing literature by bringing actor-oriented data from negotiations to the analysis in order to complement studies that used formal documents.

IO behaviour in regime complexes
Research has shown that the 'density of international rules and institutions' 60 has effects on actors' behaviour 61 and on global governance in general. 62Understanding IO behaviour requires theoretical guidance on the process of regime-complex formation.Orsini, Morin, and Young identified four stages in the life cycle of regime complexes.The first is 'atomisation': elemental regimes exist separately.The second stage is 'competition': the complex takes shape, and the elemental regimes compete for strategic positions.To move from the first to the second stage, actors need to perceive the complex and become aware of its externalities.IOs tend to 'compete' for authority and potentially engage in conflict 63 because they need to secure support from their constituents. 64ence, scholars understand IOs under regime complexity to be 'self-interested bodies that are predominantly interested in increasing their individual resources and competencies' . 65Alter and Meunier noted that the negotiation of an international agreement belongs to the stage of competition, during which actors' strategies concerning the preferred legal, normative, and institutional setting become visible. 66In the third stage, labelled 'specialisation' , actors create common metaprinciples that increasingly manage competition. 67This means that over time, competition between IOs is expected to decrease as they increasingly accommodate each other's positions in the regime complex, 'specialise' , 68 and defer to each other. 69The fourth and final stage is 'integration': the complex becomes unified and stable. 70ence, the authors see competition and specialisation as specific stages in the evolution of a regime complex, 71 but this scheme has so far rarely been filled with empirics.Further, assuming that IOs compete and specialise requires an approach that acknowledges the agency of IOs in global governance, which in turn raises the question of how to observe this agency.We argue that competition and specialisation are inherently relational processes and propose to look at the statements of IOs during the negotiations to observe these interactions and to use SNA to analyse patterns within them.

Negotiations as social spaces to study regime complexity
The view that regime complexes are social systems is not new, and literature on regime complexes has grown through the application of network analysis,72 complexity theory, 73 and empirical ethnographies to capture the relational dimensions of regime complexity. 74Using SNA to study regime complexes brings about analytical synergies, 75 such as a shared focus on structure and interactions. 76Therefore, a growing body of literature describes conflictual or synergic 77 relationships between international institutions 78 at different levels.By highlighting that regime complexes are networks of regimes, 79 authors have demonstrated the network structure of the international environmental governance system, 80 where overlaps are operationalised as references that connect legal texts 81 or as collaborative links between organisations' secretariats. 82However, network scholars have often encountered the problem of accessing detailed relational data to describe and analyse regime complexes at different stages of their formation, and arguably the methodological approach to do so.Subsequently, many researchers conducting SNA relied on legal documentsa data source that arguably is particularly 'unsocial' and not interactive but rather the outcome of social interactions -to identify the elements of a regime complex and record overlaps between them. 83n order to address this shortcoming, we built on a body of literature that has produced empirical ethnographies of international negotiations and other sites of global governance. 84Negotiations were found to be an important step in the formation of a regime complex, 85 and an issue area is understood to be a set of issues 'dealt with in common negotiations' . 86Further, ethnographic studies of global governance have established that international negotiations are largely social events 87 and have emphasised the relational and interactional aspects of negotiations 88 -both central themes for regime complexity and network analysis.Hence, scholars have repeatedly identified the potential for ethnographic studies to shed light on inter-institutional complexity by considering the perceptions of negotiators and participants of international negotiations, 89 as well as by providing a wealth of relational and social data for network analytic approaches. 90Regime complexes may emerge through various developments such as international crises, 91 state strategies, 92 the activity of secretariats, 93 or the influence of small groups of people, 94 but the study of negotiations on a new treaty can offer a window to observe the formation of regime complexes. 95articularly, CEE, which collects data at intergovernmental negotiations, has become a widely applied research method in international environmental politics. 96CEE has the advantage of dispatching a team of researchers to highly complex and dynamic negotiation sites, which enables the researchers to collect and connect various data sources. 97Similarly to CEE literature, we approach the negotiations as sites of social interaction 98 and make use of collecting data 'within' the 85 Morin and Orsini, 'Regime complexity and policy coherency'; Morin and Orsini, 'Policy coherency and regime complexes' . 86 negotiations, being sensitive to social and relational interactions in the negotiations. 99Approaching regime complexity with a CEE view, we focus on the processes rather than the outcomes of international negotiations in which governmental and non-governmental actors shape the regime complex through social and political struggles. 100owever, in contrast with most CEE contributions, we use participant observation from the CEE toolkit as a data collection method and SNA as an analytical tool.Separating data collection and data analysis methods had a number of advantages: first, it allows different researchers to contribute to the same database of observations, which is used for different research purposes.This enabled us to systematically track the involvement of IOs throughout negotiation rounds, even when the researchers conducting the SNA were not present themselves.Second, in this way, the data and the method underlying the network analysis can be published to make the analysis reproducible and comprehensible.While the proposed combination of CEE with SNA in this way brings several advantages, it does render the 'relational' nature of the data relatively thin compared to a full ethnographic account of relationality.Building on the ethnographic and network-analytic work in the area of regime complexity, the next section describes a methodology to combine both: data collection through participant observation and CEE at international negotiations and network analysis using the collected data.

Combining observational data collection and social network analysis
While much of the work on regime complexes has relied on desk research, 101 surveys, or document analysis to identify the elements of the complex, 102 we propose studying the regime complex from within the negotiations.We used the ongoing BBNJ negotiations to collect data via participant observation as well as document analysis, making use of the rich network data collection opportunities offered by IGCs. 103During the second and third two-week IGC sessions, three researchers conducted CEE at the New York headquarters of the United Nations (UN).During plenaries, working-group sessions, and side events, we made systematic observations on the delegation of the speaker, the section of the draft text, and the statement's content. 104Regarding the first IGC, which the research team did not attend, we watched the proceedings via the UN webcast and made observations with the same system. 105Furthermore, all draft treaty texts were included in the document analyses.Our data collection and variables are summarised in Table 1.We scanned the more than 10,000 observations for either (1) state references to IOs, to delimit the regime complex according to the statements of negotiators; 106 or (2) IO statements, to analyse their behaviour. 107n analysing the observations, we combined network analysis with qualitative interpretations, 108 following the identified need to mix quantitative and qualitative methods in the description of 99 Barnett and Duvall, 'Power in international politics'; Weiss and Wilkinson, 'From international organization to global governance'; Suiseeya and Zanotti, 'Making influence visible' . 100Hughes and Vadrot, 'Weighting the world'; Hughes, Vadrot, Allan et al., 'Global environmental agreement-making' . 101Gomez-Mera, Morin, and Van de Graaf, 'Regime complexes' . 102Nasiritousi and Faber, 'Legitimacy under institutional complexity' . 103Paterson, 'Using negotiation sites'; Hughes, Vadrot, Allan et al., 'Global environmental agreement-making' . 104Our notes are close to transcripts of the statements.For example, on 26 March 2019, the representative of Tonga highlighted the similarities between the ISA and the BBNJ negotiations: the ISA was a deal between conservation and use of resources; he added that the ISA legal framework enabled cooperation with other instruments.Another observed IGO activity was the FAO representative pointing out that the FAO already delivered technical assistance and cooperated closely with many of the states that were in the room (26 August 2019). 105The UN webcast is available at: {https://www.un.org/bbnj/content/webcast}. 106See 3.2.Delimitating the 'Constituting Elements' . 107See 3.3.Competition and Specialization during the Negotiations. 108Paterson, 'Using negotiation sites'; Wayne W. Zachary, ' An information flow model for conflict and fission in small groups' , regime complexes. 109We mapped the IOs that are associated with the marine-biodiversity regime complex and how they overlap as network graphs and used qualitative insights from our observations to interpret them.The observations contain relational data because IOs enter into a relational structure when they overlap regarding the treaty provision on which they make a statement.To analyse and uncover patterns in the structure of these relations, we propose using SNA.We differentiate between the three IGCs to emphasise the dynamics of the negotiation process. 110In the following sections, we explain our methodological approach for the two main contributions of this paper: (1) the delimitation of the constituting elements of the regime complex; and (2) the description of patterns of interaction between them. 109 According to Raustiala and Victor, 'regimes' are the elements of a regime complex. 111To shed light on the behaviour of elements of the regime complex, we need to take into account their agency and acknowledge that non-state actors also play an active part 'in the design and construction of the institutional framework of global governance' through their active involvement. 112Thus, this study operationalises the conceptualisation by Raustiala and Victor in a way that makes regimes and their interactions in negotiations observable.We adopt the definition of elements of the regime complex by Langlet and Vadrot that IOs contain (1) a normative framework (normally codified in a treaty); (2) member states (or other international organisations as constituting members); and (3) a body such as a secretariat with staff, budget, and a registered office that embodies the normative framework and sends representatives to the negotiation site. 113Further, we take into account organisations of the UN system that represent certain normative frameworks and can compete in governing a certain issue area, 114 but we do not include NGOs as elements of the regime complex, although some have argued that 'clubs, private regulations and transnational initiatives' 115 or other non-governmental actors such as cities 116 are indeed elements of a regime complex.In our case, they cannot be mandated by states to govern a particular issue area of international law and are thus excluded from the analysis.Because we aim to find IOs that govern a 'particular issue area' , 117 we concur that issue areas are sets of issues 'dealt with in common negotiations' . 118Operationalised, this means that issues are the individual provisions of the draft treaty text, which in total is the issue area dealt with in common negotiations.To advance rules for the delimitation of our sample of elements of the regime complex, we scanned our observations for state references to IOs.Any time a state mentioned an IO in a statement, we recorded this for the delimitation of the elements of the regime complex.For each IGC, these references were compared to the IOs that were mentioned in official draft texts to contrast a delimitation based on official documents with a delimitation based on perceptions and interactions.
These considerations helped us to combine various data sources (Table 1) to delimit the boundaries of the regime complex through the following rule: an IO must have: (1) made a formal statement; (2) be referred to by states in the room; or (3) be mentioned in a draft text.

Detecting competition and specialisation during negotiations
In addition to delimiting the elements of the regime complex, this paper looks at relational patterns between IOs to analyse how they influence global governance structures through their involvement in negotiations. 119To create a network, we scanned our observations for statements of IOs and connected each statement to the relevant provision of the draft treaty.These overlaps between IOs, i.e. when two or more IOs make statements in the negotiations on the same provision of the treaty draft text, represent the links between IOs in the emerging network.
To justify that these overlaps are indeed relational, we analysed the IOs' statements.We found that IOs generally make similar statements to each other in which they underline their own competence, mandate, and interest in working in relation to a certain provision of the treaty.UNEP, for example, by referring to its regional seas programmes120 or its ocean-monitoring center,121 indicated that it is competent to be tasked to manage aspects of marine biodiversity.The IOC declared that it was ready to provide member states with technical assistance to develop a proof of concept for the clearing-house mechanism, 122 and WIPO emphasised that it already had the mandate to govern intellectual property aspects of the new agreement. 123While most of the analysed statements follow this communicative style, some statements indicated more directly a competitive relationship.For example, the FAO, by highlighting that existing bodies should be respected, underlined its authority over fishing-related matters and urged other IOs present in the room to refrain from interfering with 'its' topic. 124On the other hand, we could not observe a single statement in which an IO advocated for the recognition of another IO's mandate.Overall, IO statements express a conscious decision to intervene on a specific provision and thus, if two intervene on the same provision, a shared interest between two or more IOs.In this way, we argue that IOs enter a relational structure when both make a statement on the provision.As a next step, we conducted SNA to uncover patterns in the relational structures between IOs, which are displayed in the following section.This approach limits the scope of our analysis to IOs that were 'present' during the negotiations and took the floor to make statements.
We suggest two readings in the relational structure: first, overlaps -when two (or more) IOs make statements in the negotiations on the same provision of the treaty draft -can be read as competition, because competition among IOs has repeatedly been observed in regime complexes 125 and IOs are expected to compete when experiencing complexity because they need to secure their position, financial resources, and governance tasks. 126We argue that when states negotiate a new treaty, and with it additional financial resources and governance tasks, statements highlighting IOs' competence and relevance in relation to the targeted treaty provision can be viewed as competition when a second IO makes a similar statement on the same provision.Competition may not be directly expressed in the statement (which is expected in diplomatic settings where potentially conflictual statements are often formulated in very diplomatic language) and may not even be conscious behaviour, but competition emerges when two or more IOs make similar statements on the same treaty provision.
Second, it is suggested that specialisation takes place when IOs make statements in relation to treaty provisions in which no other IO appears interested.Just as with competition, specialisation is a behaviour that has been observed when IOs face complexity. 127When IOs highlight how their competences and mandates could contribute to the objectives and tasks of the agreement and no other IO makes a similar statement on the same treaty provisions, they seem to recognise the other IOs' competence and specialisation.

Delimiting the 'constituting elements'
In this section, we describe the formation of the new marine-biodiversity regime complex in the course of the three IGCs and describe its delimitation based on three graphs displaying the elements of the complex in each of the three IGCs -as proposed by states and as codified in the draft text.
During the IGCs, states refer to IOs they perceive to play a role in relation to the topic that is being negotiated.In the follow-up of an IGC, the presidency drafts a new version of the treaty text   based on the statements that were made during the negotiations.As this codifies the current status of the negotiations (and the role of IOs within them) at the end of each negotiation round, this is the last step of each round and the first of the next round accordingly.The draft text serves as the basis for negotiations and structures the discussion on specific provisions.
Figure 1 visualises the chronology of the data presented in Figures 2-7 from IGC 1 to IGC 3. Figures 2-4 show a network of IOs connected through references made by state delegates.By recognising different IOs in relation to the same treaty provision, states express diverging political interests that shape the constitution of the regime complex.Technically, if states mentioned two or more IOs in the same treaty provision, these will be connected through this provision.We present the negotiation of the regime complex as a network in which IOs become connected by being jointly mentioned in the same treaty provision.Figure 2 is the first step in this phase of regime-complex formation in which states delimit which IOs should be part of it and which not.
Figure 2 shows that in IGC 1, states refer to a considerable number of IOs -27 in total.States perceive UNDOALOS, CBD, ISA, IOC, FAO, and IMO as central parts of the regime complex, and, accordingly, they appear in the resulting draft text after IGC 1.This shows how the marinebiodiversity regime complex evolves formally into legal text when the IOs that states propose are mentioned in treaty provisions.We see that while 14 of the IOs that were mentioned by states appear in the draft text (in blue), 13 do not (in red).
Figure 3 shows that states refer to 17 IOs in IGC 2. We see that states mention fewer IOs as they start to agree on the delimitation of the regime complex that is represented by a thinning out and a simultaneous consolidation of the centre of the network.Now, only 5 out of the 17 IOs are mentioned in the following draft text.We see that IOs such as FAO and IMO moved away from the centre of the network (e.g. were mentioned less by states) and subsequently were deleted from the draft text.A similar fate can be observed for IOs that were mentioned rather peripherally in IGC 1 and which do not appear in the next step of the emerging regime complex in IGC 2, such as UNEP, GCF, and CITES.On the other hand, some peripheral IOs such as WIPO and the GEF remain part of the codified regime complex, although they are only mentioned in relation to a few treaty provisions.
The thinning out of the regime complex continues during IGC 3. In IGC 3, states referred to only 15 IOs, of which only 3 are mentioned in the following draft text.The three IOs that remain in the draft text occupy a central position, meaning that states perceive them to be important to many different aspects of the treaty.
In total, throughout the three IGCs, we can observe how states first propose many elements of the emerging regime complex, out of which some become part of the codified regime complex.Here, we want to highlight the different delimitation methods.Regime-complex descriptions relying only on codified documents such as treaty texts would not be able to observe the IOs that states consider and propose to play a role but that do not make it into the draft text.This means that research approaches looking only at the treaty draft text would only conceive IOs in blue to be part of the regime complex.We can see from Figure 2-4 that in fact the majority of IOs do not appear in the draft text, although states mentioned them and thus seemed to consider them important for the future governance of the regime complex.

Detecting competition and specialisation during negotiations
After states made their statements, IOs took the floor and issued statements on specific BBNJ treaty provisions to express their competence, existing mandate, and interest in relation to specific provisions.
During the first IGC, 12 IOs made 38 statements in relation to 14 different provisions of the draft treaty.The following networks of IOs show that several IOs sought to be recognised within the future BBNJ agreement.Figure 5 demonstrates that all IOs (with the exception of WIPO) overlap in their statements in relation to treaty provisions.IOs enter a relational structure by making statements, and if overlaps are interpreted as competition, the results confirm to some extent literature that expects competition among IOs128 that use negotiations to attempt to increase their individual resources and competences.129This pattern, however, seems to change as the negotiations move on.During IGC 2, we can observe that IO behaviour begins to diverge.While the number of total statements remains equal at 38, now 10 different IOs relate to 17 different treaty provisions, and two opposing trends emerge.While the separation of the CBD from the main component and the disappearance of WIPO and the CMS indicate that there may be a specialisation taking place, overlaps between IOs in the centre seem to be aggravated.
In IGC 2, the CBD focuses solely on provisions that no other IO appears to be interested in, and WIPO, which had shown a similar behaviour in IGC 1, does not make any statement.In this way, some IOs find their position in the new regime complex, and other IOs then focus on other parts of the treaty -IOs specialise or find their niche in organisational ecological terms. 130However, among a few IOs, particularly the closely connected UNEP, IUCN, FAO, and IMO in the centre of the network, competition appears to persevere.These IOs are also the ones that were taken out of the draft text in IGC 2 (as indicated in red), which additionally suggests that they needed to highlight their work in relation to the treaty.
During IGC 3, only 8 IOs made statements in relation to 16 provisions (Figure 7).However, the overall number of statements increased to 53.We can see a continuation of the trends identified in Figure 6: some IOs (NEAFC and NPFC) cease to make statements.The IOC and ISA move away from the centre of the network where contestation persists and seem to specialise in specific provisions.Interestingly, both IOs are mentioned in the draft text (indicated by the blue colour).This suggests that they may have found their niche, which is codified in the draft text, and in response move away from competing.On the other hand, we also observe a further increase of overlaps in the centre, particularly among the four IOs (UNEP, IUCN, FAO, and IMO) that have been at the centre throughout all three IGCs.The thickness of their links indicates that the number of overlaps has increased, possibly indicating increased competition.

Discussion
We set out in this paper to (1) show how a methodological focus on international negotiations combining observational data collection methods and SNA can contribute to understanding the formation of a regime complex by obtaining and analysing new data; (2) demonstrate how this data can be used to delimit the regime complex based on the perception of actors, rethinking traditional ways to delimit the regime complex; and (3) advance understanding of competition and specialisation among IOs during the ongoing negotiations.The following section first discusses the methodological contributions and their implications for regime-complex delimitation and then evaluates the results regarding the competitive and specialising behaviour of IOs, while giving an outlook on further research.

Combining observational data collection and SNA at international negotiations to fill the data gap
We studied the emerging marine-biodiversity regime complex by collecting data at ongoing negotiations.We consider negotiation sites as organised social spaces, in which the regime complex forms through interaction and recognition between state and IO actors that can become codified in legal text.This study addressed the gap in active fieldwork, as regime complex research has relied 'too heavily on desk research' . 131We filled in detail two empirical gaps: (1) the delimitation occurring before a regime complex is codified in legal documents; and (2) IO interactions during the negotiations of the regime complex.
We demonstrate how data collected at intergovernmental negotiations can be used to apply social network analyses, 132 which in turn are interpreted qualitatively.We believe that this combination of observational methods of data collection and SNA shows great potential for further research on regime complexes to complement work that so far has used shared membership, 133 treaty text citations, 134 or participants' biographies 135 to gather data that captures elements of relations such as structured interactions between actors.It uses the strengths of CEE to gather data collectively 130  in highly interactive settings, while being sensitive to the minutiae of social interaction and the strengths of qualitative SNA, focusing on structures that emerge from interactions.In this way, SNA helped to uncover interactional patterns that would have been invisible to a network-insensible observer and guided our qualitative interpretations, adding a social dimension to the delimitation problem of regime-complex research. 136hile this approach has advantages when it comes to identifying the elements and overlaps of a regime complex, it also suffers from some limitations.First, the collected data was used to study a regime complex in the making, offering a live window into regime-complex emergence, but failing to capture the whole life cycle of regime complexes.Once the negotiations are finished and the new treaty is in place, one would need to adapt the research approach by (1) looking at potential conference of the parties (COPs) meetings of the new regime or (2) looking at other relevant meetings and negotiations to track the involvement of the IOs identified in this paper.Secondly, CEE requires access to negotiation sites and shared fieldnote-taking practices, which is resource-and time-consuming and may not be a feasible research design for single researchers or scholars who do not get access to the negotiation site. 137

The problem of regime-complex delimitation
We show that a regime complex can be delimited through state references to IOs.States recognise different IOs in relation to the issue area, expressing different preferences regarding the development of the marine-biodiversity regime complex.In this way, the regime complex can be delimited by going beyond rather static legal texts and considering the perception of actors and the varying levels of integration of elements in the regime complex.We hereby contribute to defining the boundaries of the 'array of … institutions' . 138We demonstrate that IOs play a role -to a varying extent -in relation to the treaty because representatives of states and IOs perceive them to be a part of the regime complex even if they do not become legally codified elements of it. 139This different way of delimiting the regime complex becomes crucial when analysing the final treaty text.By looking only at the text of the finalised agreement, one would perceive the GEF, IOC, ISA, IMO, and FAO as part of the regime complex but would potentially fail to recognise the key role played in the eyes of states by the CBD, UNEP, or regional organisations such as OSPAR in relation to the governance of marine biodiversity.In this way, our method can make visible what is not in the final text but, nevertheless, continues to influence the governance of the marine-biodiversity regime complex.However, no delimitation will ever be final, as the regime complex also continues to evolve outside or after international negotiations.
IOs are involved to different extents -some make statements in the negotiations, some are perceived by states to be important, and some may receive a formal mandate under the new treaty.However, the fact that the final text mentions fewer IOs than previous drafts and much fewer than mentioned by states, and all IOs except for the GEF in only one provision (on the clearing-house mechanism), may indicate that, with increasing complexity, international documents tend to have fewer references to IOs, as it could become too contested and 'messy' .This outcome, both the lack of mention of many IOs and the mention of four IOs in one provision, raises the question of how a division of work may materialise in the future.
The marine-biodiversity regime complex is particular in its awareness and explicit recognition of inter-institutional dependencies, which may limit the applicability of the proposed delimitation 136  method.Still, we argue that some aspects can be considered for future research into regime complexes in other issue areas: it is not only the legal overlaps in treaty texts but also the perception of policymakers that makes the regime complex. 140This implies that there is not one correct way to delimit the regime complex -rather, there are different levels of integration and interdependence, 141 and the right delimitation may depend on the interest and analytical focus of the researcher.While we acknowledge that state statements and, hence, the delimitation of the regime complex reflect larger geopolitical interests and conflicts, we do not address them in this paper but encourage research into the geopolitical conflicts and configuration of authority and legitimacy in complex international governance that may be represented in the delimitation of a regime complex. 142

IO behaviour in regime complexes
This paper proposes a structural view of competition and specialisation among IOs during negotiations and places its insights into the discussion on IO behaviour in regime complexes.While the analysis suggests that overlaps in statements mean competition, this interpretation leaves room to further study the behaviour and intentions of IOs in negotiations.For example, a more systematic qualitative analysis of the overlaps shown in Figures 2-7 may highlight different strengths and qualities of statements, giving insights about the underlying intentions.Conducting interviews with IO representatives could also further strengthen analyses of their behaviour at negotiations.The results of this paper can support the view that IOs engage in competitive behaviour 143 when facing regime complexity but add important depth to the dynamics during the negotiation phase.We show that the negotiation phase is a process itself, and dynamics change over time.After the first round of negotiations, the number of overlaps in IO statements decreased, suggesting that IOs started moving towards a division of labour and specialisation (Figures 5-7). 144At the same time, while some (WIPO, CBD, ISA, IOC) specialised, other IOs (UNEP, IUCN, FAO, IMO) seemed to intensify competition.This IO behaviour seemed to correlate with their being mentioned in the draft text.IOs tended to specialise when states identified a role for them and codified this in the relevant treaty provisions.IOs that were continuously not mentioned in the draft text seemed to have intensified their competition, making an increased number of statements in overlapping provisions This contradiction between making statements and being mentioned in the text extended to the final text.The GEF -the IO that did not make any statement during the three analysed IGCs -is mandated a special role as part of the financial framework of the new treaty.But two of the IOs that made many statements are not mentioned at all (UNEP and IUCN), and four other IOs are mentioned as potential examples and contributors to the clearing-house mechanism of the BBNJ instrument.
These findings speak to Orsini, Morin and Young and Morin and Orsini, 145 who theorised that in the formation of a new regime complex, elements first compete, then manage their competition and specialise, before becoming a stable regime complex.While our findings support the notion that a regime complex evolves, we suggest that the view of the different phases may be too simplistic, as both competition and specialisation already seem to occur simultaneously during the negotiations.We also add that not all IOs specialise, but some may increase their competition instead.Negotiations may be a particularly interesting moment to study these dynamics because the regime 140 Orsini, Morin, and Young, 'Regime complexes' . 141Hollway, 'What makes a "regime complex" complex?' . 142Green and Auld, 'Unbundling the regime complex'; Steffen Bauer, 'Does bureaucracy really matter?The authority of intergovernmental treaty secretariats in global environmental politics' , Global Environmental Politics, 6:1 (2006), pp.23-49. 143Alter and Meunier, 'The politics of international regime complexity'; Morin and Orsini, 'Regime complexity and policy coherency' . 144Downie, 'Competition, cooperation, and adaptation' . 145

Figure 2 .
Figure 2. IOs are connected through provisions in which states referred to them in IGC 1.In blue are IOs that appear in the draft text after states mentioned them, in red IOs that did not appear in the draft text.

Figure 3 .
Figure3.IOs are connected through provisions in which states referred to them in IGC 2. In blue are IOs that appear in the draft text after states mentioned them, in red IOs that did not appear in the draft text.

Figure 4 .
Figure 4. IOs are connected through provisions in which states referred to them in IGC 3. In blue are IOs that appear in the draft text after states mentioned them, in red IOs that did not appear in the draft text.

Figure 5 .
Figure 5. IOs are connected by making statements regarding the same treaty provision in IGC 1.In blue are IOs that appear in the draft text after states mentioned them, in red IOs that did not appear in the draft text.

Figure 6 .
Figure 6.IOs are connected by making statements regarding the same treaty provision in IGC 2. In blue are IOs that appear in the draft text after states mentioned them, in red IOs that did not appear in the draft text.

Figure 7 .
Figure 7. IOs are connected by making statements regarding the same treaty provision in IGC 3. In blue are IOs that appear in the draft text after states mentioned them, in red IOs that did not appear in the draft text.
The British Journal of Politics and International Relations (2022), p. 13691481221115937; Helfer, 'Regime shifting in the international intellectual property system' .

Table 1 .
Overview of data collection.
Figure 1.Chronology of data.
Kim, 'The emergent network structure'; Kim and Mackey, 'International environmental law'; Merry, 'Global legal pluralism and the temporality of soft law' .
137Campbell, Corson, Gray, MacDonald, and Brosius., 'Studying global environmental meetings to understand global environmental governance: Collaborative event ethnography at the Tenth Conference of the Parties to the Convention on Biological Diversity' .138RaustialaandVictor,'The regime complex for plant genetic resources'; Orsini, Morin, and Young, 'Regime complexes' .139Orsini,Morin, and Young, 'Regime complexes' .