Breaking the legal link but not the law? The externalization of EU migration control through orchestration in the Central Mediterranean

ABSTRACT Transcending the established problem-solving perspective, this article proposes a novel conceptualization of orchestration as a strategy to escape legal responsibility. To test our conceptual argument, we study the case of EU migration governance vis-à-vis Libya in the Central Mediterranean. We show how legal constraints stemming from the 2012 Hirsi judgement of the ECtHR drove the EU from direct extraterritorial border operation to indirect orchestration via Libyan authorities. The EU used the orchestration techniques ‘assistance’, ‘endorsement’, ‘convening’ and ‘coordination’ to establish Libyan authorities like the coast guard as central intermediaries in the field of border patrol and search and rescue. These orchestration strategies increasingly replaced EU operations on the ground, thus posing new challenges for human rights protection.


Introduction
Recent scholarship has noted that the EU increasingly relies on orchestrating intermediaries to deal with external governance challenges (Genschel & Jachtenfuchs, 2018). This holds particularly true in the context of the so-called 'migration and refugee crisis', with UNHCR operating camps on Greek islands, NATO providing reconnaissance, monitoring and surveillance of illegal crossings and Turkey stepping up maritime interception in the Aegean (Heijer et al., 2016). 1 Orchestration involves orchestrators enlisting third parties as intermediaries on a voluntary basis, providing them with ideational and material support in pursuit of a certain governance goal (Abbott et al., 2015a(Abbott et al., , 2015b. As it is the case with the governance literature more broadly (e.g., Mayntz, 2004), existing scholarship tends to emphasize arguments about effective problem solving to explain motivations behind orchestration. In this view, orchestrating third party capabilities is faster, cheaper and requires fewer resource commitments than creating genuine EU capacities (Abbott et al., 2015b;Genschel & Jachtenfuchs, 2018).
Whilst we broadly agree, we maintain that political actors may also resort to orchestration as a strategy to escape legal constraints. Thus far, the role of law as a driver for orchestration has not been systematically studied. This is particularly surprising in the context of the EU, where law has played such a prominent role in the integration process (e.g., Stone Sweet, 2004). To test our conceptual argument about the legal dimension of orchestration, we focus on the empirical case of EU migration governance vis-à-vis Libya where we witness a growing shift from direct to orchestrated forms of EU border management. A problem solving-perspective struggles to explain why the EÙs orchestrated border management has increasingly relied on the authorities of a failed state as intermediaries, which have demonstrated severe shortcomings to perform key governance functions in an efficient manner whilst requiring substantive levels of EU-support. Rather than cost-savings and efficiency gains, we show that it was the aim to escape legal constraints stemming from the 2012 Hirsi judgement of the European Court of Human Rights (ECtHR) that pushed the EU towards orchestration. The Hirsi ruling made it difficult for the EU to continue its long-standing practice of extraterritorial push-back operations (Tondini, 2010). To balance a longstanding non-entrée policy (Hathaway, 1992) with the requirements of the Hirsi judgement the EU used various orchestration techniques to bolster the Libyan government, portraying Libya as a sovereign state with the right and duty under international law to perform border and search and rescue (SAR) operations within its territorial waters. Simultaneously, the EU has gradually suspended its direct involvement in extraterritorial maritime operations.
Being subject to growing tensions between tightening legal commitments and self-interested EU policy preferences, border management in Libya constitutes a 'typical case' (Gerring, 2007, pp. 91-93) to test the relevance of our theoretical propositions. Yet, we consider legal orchestration dynamics also relevant for other policy fields, not least given the growing legalization and judicialization of the EU's international relations (Müller & Slominski, 2017;Wessel, 2015). Evidence of orchestration occurs in numerous areas of the EU's external migration policy, where we increasingly witness non-hierarchical and soft agreements between the EU and third countries such as Turkey or Afghanistan (Carrera et al., 2017). Moreover, examples of orchestration strategies with the view to circumvent legal constraints can be observed in areas of external policy where the EU relies on international organizations, NGOs or private actors to pursue its political objectives (Bendiek & Bossong, 2019;Müller & Cornago, 2018;Van Dessel, 2019). Finally, our case also speaks to the growing literature on indirect accountability of extraterritorial human rights violations (Heupel, 2019).
Besides relevant academic literature our analysis draws on a broad range of primary sources including EU and national policy documents, parliamentary questions and answers, reports by NGOs and think tanks, legal texts as well as eight in-depth, semi-structured interviews with officials from EU-institutions, NGOs and EU member states. 2 The article first develops a theoretical framework that highlights the legal considerations behind the choice of orchestration and how legal aspects permeate and shape orchestration techniques, formulating key expectations that extend beyond established problem-based rationales. The subsequent empirical section, first, points to limitations of problembased arguments to account for the EÚs turn to orchestration in border management. Second, it shows how legal considerations have driven the EU to gradually strengthen its orchestration strategy, providing a detailed analysis of different orchestration techniques. In conclusion, we discuss key findings.

Orchestration assistance, endorsement, convening and coordination
Research has primarily pointed to the so-called 'goal/capabilities' dilemma, i.e., situations where governance actors lack sufficient capabilities to effectively perform their roles by themselvesto explain the choice for orchestration (Abbott et al., 2015a, p. 5). Addressing this dilemma, orchestration can create efficiency gains from 'specialization, pooling of resources and mutual learning' (Abbott et al., 2015a, p. 7;Genschel & Jachtenfuchs, 2018, p. 190). From this problem-solving perspective, the EU can be expected to engage in orchestration when it lacks certain capabilities needed to achieve its goals through other governance modes (Abbott et al., 2015a, p. 21).
Conversely, the literature has given little attention to the role of law in motivating orchestration strategies. To be sure, orchestration scholars have touched upon important legal and normative issues, e.g., acknowledging that diplomatic endorsement can also be legal in nature (Abbott et al., 2015a, p. 15) or that orchestration may allow governors to govern indirectly in areas beyond their territorial jurisdictions. Yet, thus far the question of how orchestration and its various techniques relate to the legal responsibility of a governor has not been studied systematically. This is surprising, as in related literatures legal considerations have been identified as important motivations behind the choice for specific governance arrangements. For instance, works on 'collusive delegation' showed how self-binding international cooperation may be employed to 'loosen the constraints imposed on governments in the domestic arena' (Koenig-Archibugi, 2004). Yet, our focus is not on arguments about 'two-level games', but on circumventing legal responsibility through orchestration, which is both indirect and soft. The literature that speaks most directly to our focus on the relation between orchestration and legal responsibility are works on 'outsourcing' that have developed in neighboring disciplines such as economics (Rawling, 2006, p. 3;Davitti, 2019). Still, outsourcing can be done in a number of different forms. Here the orchestration literature can improve our understanding of specific governance arrangements and techniques that actors deploy to evade legal responsibility.
Our conceptual argument draws on a synthesis of rationalist and constructivist assumptions that underpins 'strategic constructivism' (Saurugger, 2013;Schimmelfennig, 2003). Highlighting the social embeddedness of rational calculation, it maintains that EU-related actors act strategically on the basis of self-interested preferences, whilst their strategies are informed by a community environment defined by a collective identity, based on fundamental common values and norms (Manners, 2002). This also includes a commitment to the rule of law and the respect of the EU's (international) legal obligations. Hence, our emphasis on legal motivations behind orchestration is not meant to imply that EU policy-makers are apolitical actors mechanically following legal rules and case law regardless of their interests. Rather, we maintain that the EU might try to do morally problematic things whilst, mindful of its legal reputation, claiming that these actions are in accordance with the law (Johnstone, 2011).
Avoiding legal responsibility through orchestration is not the same as breaching law in pursuit of self-interested policies. It is concerned with a strategic understanding of law, treating law as a normative resource that cannot easily be ignored. Relying on intermediaries, the EU as orchestrator is able to address its governance priorities without becoming directly involved. This is important, as direct engagement often has consequences in terms of liability and legal responsibility. However, not all forms of indirect governance have the same legal implications (Abbott et al., 2019). Importantly, principalagent delegation relies on hard, hierarchical and legally-binding instruments (Dür & Elsig, 2011). Whilst this ensures greater control over agents, it also implies a closer association with the agent's performance and, by extension, greater legal responsibility of the principal. Other forms of indirect governance, such as trusteeship, are also typically based on a formal grant of authority, making it difficult for the trustor to escape legal responsibility for the acts of the trustees. Similarly, co-optation is also not an attractive option to evade legal obligations (Abbott et al., 2019), as it is based on the exercise of hierarchical authority in decision-making and control that closely associates the co-optor with the actions taken by a co-optee. Conversely, orchestration is voluntary and non-hierarchical, both of which is crucial to blur or even break the legal link between the intermediary and the orchestrator (ibid.; Majone, 2001). As such, orchestration neither requires a formal delegation or grant nor does the intermediary own its authority to the orchestrator.
We expect governance actors to be likely to rely on orchestration strategies to avoid legal responsibility if three key conditions are met. First, it is attractive in situations where an orchestrator is obliged to observe high legal standards which create tensions with its political preferences (e.g., Fazal, 2018). Such tensions may arise in cases where preferences have changed, legal standards have developed or orchestrators are confronted with unintended consequences of an established legal regime. The more demanding these legal standards are, the greater the incentive for circumvention. Second, orchestration is likely in situations where alternative strategies to avoid legal responsibility (e.g., creative interpretation or suspending the law) are considered too costly or are not available (Gammeltoft-Hansen & Aalberts, 2018). Third, the orchestrator needs an intermediary that is legally competent to perform governance functions the orchestrator seeks to offload. 3 Legal requirements may come from a variety of sources (e.g., international, EU or domestic law) depending on the status of the intermediary (state or a private actor) as well as the jurisdiction and policy field in which the intermediary operates. The less demanding these requirements are, the more likely the intermediary will be chosen by the orchestrator.
If the intermediary does not meet relevant legal requirements, the governor can rely on the orchestration techniques 'assistance', 'endorsement', 'convening' and 'coordination' to ensure that intermediaries fulfill them. 'Assistance' involves material support for an intermediary, such as training and capacity building or the provision of financial and administrative support. Assistance has an obvious problem-solving dimension, as it helps to enable an intermediary to carry out certain governance objectives more effectively. In this view, the governor uses its own financial and administrative resources to build up the capacity of intermediaries, which is particularly attractive when doing so leads to lower costs and greater effectiveness than direct involvement. Still, 'assistance' has also a distinct legal dimension. There are domains in which only intermediaries who are sufficiently equipped and resourced may be legally entitled to execute certain tasks the orchestrator seeks to outsource (Humphrey, 2012;Van Damme, 2001). Second, an orchestrator may enable an intermediary through 'endorsement', providing ideational and rhetorical support that 'increases the social authority intermediaries can bring on targets' (Abbott et al., 2015a, p. 15). Endorsement is political, with the governor emphasizing the capacity and political legitimacy of an intermediary in performing certain governance functions. Still, it can also serve a legal function. Sometimes intermediaries have to be formally recognized before they are legally allowed to act. For example, the formal recognition of a certain political group as the legitimate government of a state serves as a precondition allowing the respective government to act as an intermediary in a legally meaningful way (Lauterpacht, 2012). Legal endorsement also occurs when an orchestrator formally acknowledges that a certain practice or standard applied by the intermediary fulfills legally required conditions.
Orchestration may, furthermore, rely on 'convening', seeking to empower intermediaries by bringing them in contact with other influential actors in a given governance domain. Convening differs from the governor's own endorsement of an intermediary as it aims at legal recognition and legitimacy by relevant stakeholders at large. Here, orchestrators can use their privileged access to other important actors at the international level (e.g., states or international organizations), which have the power to enhance the legitimacy and legal status of the intermediary.
'Coordination', in turn, aims at coordinating activities among different intermediaries in a given governance space to increase their effectiveness and avoid overlapping activities. Coordination of intermediaries is of legal relevance in a context of regime complexity, which is shaped by different regimes and jurisdictions (Alter & Meunier, 2009;Gammeltoft-Hansen & Aalberts, 2018). Here, coordination measures involve the allocation of competences or exercise thereof between different intermediaries. Clearly delimited competences are not only important for preserving the coherence of the relevant legal framework, but are also a precondition for ensuring legal accountability of intermediaries. Moreover, intermediaries may themselves be composed of different actors. From a legal perspective coordination may serve the purpose to unite and integrate different actors under a formally recognized authority, which can be particularly relevant when dealing with situations of fragmented actorness and authority. The simple orchestratorintermediary model becomes even more complex, when intermediaries turn into orchestrators creating further (sub-) intermediaries (Abbott et al., 2015a, p. 19).

The EU's turn to orchestration in external migration governance
The EU has a longstanding reputation for shifting the control of its external border further away from its territory (Lavenex, 2006), which has always posed a serious challenge to human rights law (Aalberts & Gammeltoft-Hansen, 2015;Slominski, 2013). Externalization of migration control traditionally heavily relies on both third-country cooperation and direct involvement of the EU and its member states in extraterritorial border operations. In the case of Libya, Italy provided patrol boats to Libya and was also directly involved in extraterritorial activities. Whilst the donated boats were sailing under the Libyan flag, they consisted of mixed Italian-Libyan crew (Marianna, 2011, p. 25). In addition to the Friendship Treaty, Italy and Libya concluded two Protocols, which contained further evidence that Italian officials were directly involved in patrolling and push-back operations (ECtHR, 2012: para. 19).
Recently, however, we witnessed a notable shift within externalization from the EÚs direct involvement to orchestration. Not only did the EU strengthened its orchestration approach by bolstering the Libyan coast guard (see below), but it also substantially decreased its direct involvement in maritime operations. In October 2014, Italy suspended its large 'Mare Nostrum' operation which included a considerable SAR component and has been described as part of an initial humanitarian EU response to an intensifying migration crisis that also included close coordination with NGOs that became active in the SAR domain (Cusumano, 2019). Yet, given the lack of any meaningful burden-sharing agreement among EU member states, Italy received nearly all migrants who were rescued by SAR NGOs on the Central Mediterranean route. Against this backdrop, Italy not only nearly ceased its SAR activities but also started to take restrictive measures towards SAR NGOs (Cuttitta, 2018a(Cuttitta, , 2018b, which intensified during Giuseppe Conte's first government (Cusumano & Gombeer, 2018).
The subsequent 'Triton' mission operated by the EU border and coast guard agency Frontexwhich was launched in 2014 and replaced by mission 'Themis' in 2018prioritized border control and anti-smuggling tasks, whilst lacking sufficient assets required for effectively substituting the SAR activities of Mare Nostrum (Cusumano, 2019, p. 9). Similarly, neither of the other two main EU missions operating in the Mediterraneanthe Italian-led operation 'Mare Sicuro' and the EU Naval Force Mediterranean (EUNAVFOR MED) operation 'Sophia'have a specific SAR mandate (UN Dispatch, 2017). The latter, which recently became a maritime mission without any vessels, is emblematic for the EU's disengagement from operations on the ground.

The limits of problem-based accounts
From a problem-solving perspective, the EÚs progressive turn to orchestration could be seen as an effort to externalize costs and make border management more effective. Here, orchestration should serve as a strategy for the EU to offset certain capability deficits by enlisting an intermediary that possesses complementary capabilities (Abbott et al., 2015a, p. 21). However, in the case of Libya this argument is anything but convincing. The EU decided to increasingly rely on the Libyan coast guard as a key intermediary in border management even though it did not have much to offer in terms of complementary capabilities. As our analysis below shows, crucial capabilities of the Libyan coast guard for managing the Mediterranean Sea border have to be provided by the EU and its member states, including equipment and boats, surveillance infrastructure and information, and training activities. What is more, the EU has outsourced governance tasks to Libyan authorities, whicheven after receiving the EÙs supportstill lack the necessary resources, political will and cohesion to perform the outsourced tasks more effectively than the EU. Instead of relying on 'costsaving' arguments, we show that legal considerations, notably the Hirsi ruling, have more explanatory power in accounting for the EÚs growing turn to orchestration.

External EU border management and the Hirsi ruling
Prior to the 2012 Hirsi ruling, the EU's extraterritorial border and push-back operations in the Central Mediterranean rested on the assumption of the extraterritorial non-applicability of established human rights standards (Human Rights Watch, 2009;Tondini, 2010). In 2003, then German interior minister Otto Schily suggested that migrants could be intercepted in international waters and then returned to third countries because European human rights standards were not applicable outside of the EU (Bundesministerium des Inneren, 2005). To justify extraterritorial maritime operations, the EU referred to ambiguous provisions of the Schengen Borders Code, international law, and bilateral agreements with third countries such as the 2008 Treaty of Friendship between Italy and Libya, which was designed to provide the legal cover for these maritime operations (Camera dei Deputati, 2010, p. 72; Giuffré, 2012).
Yet, in 2012 the legal situation changed when the ECtHR issued its landmark Hirsi decision, declaring that EU member states also had to observe their obligations under the ECHR even if they were conducting extraterritorial border operations. The Hirsi judgement was regarded as a mayor development that necessitated a substantive re-evaluation of the EU's border management approach (interviews 2, 3, 4). A direct involvement of EU related actors in extraterritorial push-back operations was no longer feasible without violating established human rights law. 4 Similarly, it no longer was adequate that donated border patrol vessels sailed under the flag of a third country while EU officials remained on board of those boats, as the ECtHR was clear that such direct extraterritorial practices constitute 'jurisdiction' and by extension legal responsibility.
The adjustment of EU external border management to the post-Hirsi legal environment has been particular evident in the domain of SAR operations. From an international law perspective, people in distress at sea have to be rescued and delivered to a 'place of safety'. 5 Under post-Hirsi conditions, a policy of non-entreé became difficult to reconcile with international SAR obligations, because any migrant rescued by an EU vessel is protected under the ECHR. To evade the new legal constraints set by the ECtHR in Hirsi, the EU had to refine its externalization practices. Besides strengthening its orchestration approach by bolstering the Libyan coast guard (see below), the EU also substantially decreased its direct involvement in maritime operations thereby avoiding 'contact' with migrants crossing the Mediterranean (see also Moreno-Lax & Giuffré, 2019).
Simultaneously, the EU intensified its orchestration efforts to fill the void of not being present on the ground. Rather than EU actors, who are subject to international human rights law, Libyan authorities, which are not bound by similar obligations, were expected to intercept, 'rescue' and finally return migrants to Libya. To escape legal accountability, the EU not only sought to avoid direct involvement in external border management, but also preferred a 'thin' relationship with its intermediaries. Accordingly, the EU did not opt for 'hard' delegation or granting authority to Libyan authorities. Rather, the Libyans were enlisted on a voluntary basis through soft means both in terms of form and content (Abbott & Snidal, 2000), allowing the EU to portray their activities as a result of Libyan sovereignty.
Within the EU, Italy has been spearheading many activities aimed at externalizing border management to the GNA (interview 1 and 2). In February 2017, Italy signed a Memorandum of Understanding (MoU) with the GNA on cooperation in the fight against 'illegal immigration' and on 'reinforcing the security' of their borders (MoU, 2017). Unlike the 2008 Treaty of Friendship, the 2017 MoU is not legally binding and was adopted without the formal approval of the Italian parliament (Camilli, 2017). This informality was complemented by its rather generic and imprecise language, which makes it even more difficult to hold Italy legally accountable for violations of human rights law (Palm, 2017). In 2019, an Italian MP admitted that the MoU was designed to make Libya do things Italy was legally not allowed to do . Subsequently, the EU endorsed the MoU in its legally nonbinding Malta Declaration, in which the European Council also agreed to 'take additional action to significantly reduce migratory flows along the Central Mediterranean route' (European Council, 2017Council, , 2018.

Orchestration in EU border management: offloading legal responsibility
To facilitate the progressive evolution from direct to orchestrated forms of border management, the EU has aimed at strengthening the status and capacity of Libyan authorities to engage in sovereign maritime border and SAR operations in accordance with international law. As we argue below, the EU has supported Libya's Government of National Accord (GNA) and its coast guard through 'assistance' and 'endorsement', and has integrated key Libyan intermediaries and external stakeholders into a common approach through its 'convening power' and 'coordination' activities.
Empowering the Libyan coast guard through 'assistance' and 'endorsement' Following Gaddafi's fall from power in 2011 and the resulting instability the EU had a clear interest in finding a reliable and internationally recognized partner to resume cooperation on migration and border control (interview 6). With the Hirsi ruling, building-up the capacity of Libyan actors and developing close bilateral cooperation in the domain of border management acquired new urgency. In the words of an EU-official 'with the Hirsi judgement it became clear that Italy could not be involved directly and that we needed a different track of capacity building' (interview 2; see also interviews 3 and 4).
The EU has given strong diplomatic backing to the GNA and took various steps to build-up Libyan authorities (e.g., Council Decision 2013/233/CFSP). For the EU, it was an important step that in 2015 the GNA was internationally recognized as the sole legitimate government of Libya (see UN Security Council Resolutions 2259 and2278;EEAS, 2016b). The international legitimacy granted to the GNA allowed the EU to 'sustain genuine Libyan ownership' and make formal requests to allow EU-funded capacity building and training of its border guard as well as necessary information sharing (Council of the European Union, 2016;EEAS, 2018;Frontex, 2016). This has been described by Admiral Enrico Credendino, commander for operation Sophia, as follows: (w)e will create a Libyan system capable of stopping migrants before they reach international waters, as a result it will no longer be considered a push-back because it will be the Libyans who will be rescuing the migrants and doing whatever they consider appropriate with the migrants. (quoted in Liguori, 2019, p. 12) Notwithstanding the weak state of Libya's political institutions, the EU has regularly portrayed Libya as a sovereign state, which has every right to protect its territory and borders and to conduct appropriate maritime operations as it sees fit. Framing it as the sovereign obligation of Southern Mediterranean countries, Italy's minister of interior at that time, Angelino Alfano stated that 'under the new system more burden would be placed on North African countries such as Egypt and Libya to look after search and rescue operations in their own waters' (cited in The Guardian, 2014: see also Merkel, 2018). Simultaneously, the EU stepped up its material support for capabilities and training of the Libyan coast guard through a number of EU schemes, including operation 'Sophia' and the EU Border Assistance Mission in Libya (EUBAM). Additional EU support was channeled through programs like the Trust Fund for Africa and the Seahorse Mediterranean Network program (Monroy, 2018a). Important elements of the EÚs assistance aimed at enabling Libyan authorities to meet legally prescribed requirements for carrying out specific border management tasks, including in the area of SAR. According to the SAR Convention, a state with an own SAR region holds the primary obligation of bringing migrants who are rescued in that very SAR region to a 'place of safety' (Giuffré, 2012, p. 706). While Libya has been a party to the SAR Convention since 2005, it did not have an own SAR region or an MRCC (Gombeer & Fink, 2018, p. 8). In August 2017, the Libyan GNA had first announced its own SAR region, a move that was endorsed by the EU. Although Libya lacked the adequate SAR infrastructure, Italy supported the move and even felt legally obliged not to interfere with the Libyan SAR region (Loschi et al., 2018, p. 20). Against this backdrop, the Libyan coast guard began to intercept or 'rescue' migrant boats, whilst simultaneously warning NGOs involved in SAR activities to stay out of Libyan waters and even engaged in hostile acts towards them (interviews 7 and 8; Mail Online, 2017;Refugees Deeply, 2017;The Maritime Executive, 2017). However, in December 2017, Libya provisionally withdrew its initial IMO application to determine its SAR zoneafter an implication from the IMO that in the absence of a rescue coordination center core requirements for the SAR zone were not metbut resubmitted a new declaration a few days later (Gombeer & Fink, 2018, p. 8). This unilateral declaration was endorsed by the EU. Former DG Home Commissioner Dimitris Avramopoulos stressed that the declaration 'represents a constitutive, legitimate act' that was 'in accordance with the provisions of the SAR Hamburg Convention' (European Commission, 2018a).
Under international law the establishment of a Libyan SAR region also required the creation of a fully-fledged Libyan Maritime Rescue Coordination Centre (MRCC), responsible for coordinating all activities in the SAR region (Annex Chapter 4.3 SAR Convention). A MRCC is required to be properly equipped and staffed with trained personnel with a working knowledge of the English language (Annex Chapter 2.3.3 SAR Convention). To meet these legal requirements, the EU encouraged and assisted Libya to establish an own SAR region, with the view to hand over SAR responsibilities of EU member states, notably Italy, which typically assumed these responsibilities prior to the establishment of an official Libyan SAR region in the area (Cuttitta, 2018a). Funded under the EU's Trust Fund for Africa and largely implemented by the Italian coast guard, the EU's support for building-up a Libyan MRCC involved a feasibility study on the MRCC, capacity building measures and setting up operational rooms, as well as support for developing a Libyan communication network (AP News, 2017;European Commission, 2018a).
With the Libyan MRCC not expected to be operational before 2020, the Libyan government created a provisional coordination facility close to Tripoli's airport, the so-called Joint Rescue Coordination Centre (JRCC) (European Parliament, 2019). Previously, communication equipment of a Tripoli-based Italian naval ship was used to coordinate SAR operations at sea. Information collected by EU missions Triton and operation Sophia about distress calls and potential SAR activities were also shared with Libya's coast guard (Cuttitta, 2018a).
Highlighting Libya's ownership (European Commission, 2018a), EU officials expressed the view that Libya had been granted a SAR zone and, under international maritime law, was in charge of it (interview 1, 2 and 5). Simultaneously, the Italian MRCC in Rome increasingly referred responsibility of emergency calls to the Libyan coast guard (Monroy, 2018b). However, despite the EU's considerable assistance and diplomatic backing of the Libyan coast guard both reports by external organizations as well internal EU assessments exposed significant shortcomings of Libyan SAR capabilities, let alone the fact that rescued migrants are brought back to Libya where they are exposed to severe threats against life, freedom and safety (Cusumano & Pattison, 2018;Engert, 2019).
Bringing other stakeholders into the EU's orchestration strategy: coordination and convening To forge international and local support for the Libyan coast guard in external border management, the EU, moreover, relied on 'coordination' and its 'power to convene'. Within Libya, the EU promoted the integration of several local groups under the umbrella of the Libyan GNA, which remains caught up in a civil war with militias aligned with the Tobruk-based House of Representatives that controls vast territories and competes for power with the GNA. The EU's coordination efforts focused on strengthening the authority of the Libyan coast guard, as the GNA's lack of political authority undermined the credibility of the Libyan coast guard (interview 1; Global Initiative, 2017). In an effort to strengthen their support for the Libyan coast guard and to create a broad alliance for border management, Italy engaged with local authorities and non-state actors such as tribal and militia leaders. This included several meetings hosted by the Italian ministry of interior in Rome as well as meetings convened within Libya (Amnesty International, 2017, p. 49). Italian officials reportedly promised financial and other assistance to local Libyan actorsincluding access to the EU Trust Fund for Africain exchange for their contribution to a coordinated approach to border management.
To build international support for a Libyan SAR region the EU also used its coordination and convening powers to bring Libyan intermediaries in contact with relevant international actors (European Commission, 2017, p. 10). An important precondition of establishing an own Libyan SAR region was its formal recognition by the IMO and its delimitation vis-à-vis neighboring states. This formalization was necessary as it defines and allocates the competence of different national SAR regions, allowing Libya to fulfill its sovereign SAR obligations. In June 2018, the IMO accepted the establishment of a Libyan JRCC and published relevant information about it in its Global Integrated Shipping Information System, including the geographical coordinates of the Libyan SAR region (IMO, 2019;Santer, 2019).
Moreover, the EU used its convening power to promote SAR arrangements between the Libyan coast guard and the border authorities of other North African countries, which constitutes a recommended practice under the Maritime SAR convention (European Commission, 2019). According to an EEAS official, the EU sought to encourage Libya to conclude bilateral SAR agreements with neighboring countries (Interview 2; European Commission, 2018a). Its active encouragement notwithstanding, the EU was careful to stress that the legitimate Libyan authority had initiated this process 'in accordance with international law' (European Commission, 2018b). Besides relying on bilateral channels, the EU also engaged with Mediterranean partner countries through multilateral fora such as the 'Shared Awareness and De-confliction in the Mediterranean' forum, hosted by operation Sophia, to promote cooperation with Libyan border authorities. The EU's various convening and coordination activities once more testify of its ambition to promote Libyan border authorities as legitimate actors in the SAR domain, which has been crucial for offloading legal responsibility through orchestration.

Conclusion
Exploring how external EU border governance practices are shaped by legal considerations, this article transcends the existing focus on orchestration as a tool for effective problem-solving. Our case study on EU migration governance in Libya provides detailed evidence for core expectations related to legal orchestration dynamics. Shedding new light on the legal motives and techniques behind the EU's turn from direct governance practices to orchestration, we hope to stimulate further research.
To avoid legal responsibility deriving from the 2012 Hirsi judgement, the EU has pursued a comprehensive outsourcing of maritime border control and SAR activities in the Mediterranean Sea, whilst simultaneously reducing its own direct involvement. This distinct version of 'externalization through orchestration' has relied on indirect techniques such as 'assistance', 'endorsement', 'convening' and 'coordination'. The EU has provided the Libyan intermediaries with considerable material, technological and financial assistance, allowing them to fulfill important legal requirements under international maritime law. Moreover, it endorsed the legal status of border management authorities of the GNA and strengthened their authority through its coordination and convening activities, both internally by engaging with various Libyan groups, and externally, by bringing Libyan representatives in contact with other external actors. This comprehensive orchestration approach allowed the EU to portray Libya's border control and SAR operations as a reflection of Lybian sovereignty and as in accordance with international law so that they cannot legally be linked to the EU.
Overall, the EU's main challenge in pursuing a non-entrée policy has not been the need to compensate for a lack of capacity for effective external border control. In fact, the EU selected an intermediary with the Libyan coast guard which lacked essential capacities, cohesion as well as an intrinsic political will to effectively assume key governance tasks outsourced by the EU. Rather, EU orchestration aimed at complying with a legal environment shaped by the ECtHR in a way that allowed it to continue its restrictive migration agenda (Wilde, 2017).
In May 2018, the Global Legal Action Network submitted a complaint to the ECtHR claiming Italy's legal responsibility for outsourcing border operations to the Libyan coast guard (ECRE, 2018). In another lawsuit filed at the International Criminal Court, human rights lawyers sued the EU for its migration policies in the Central Mediterranean (Branco & Shatz, 2019). It thus remains to be seen whether the EU's orchestration strategy will stand the test of time.
Notes 1. We use the term 'migrants' for individuals traveling irregularly across the Mediterranean towards Europe, whilst we do not want to exclude the possibility that he or she may be a refugee or asylum seeker (see Human Rights Watch, 2009, p. 22). 2. The interviews were carried out in the period between May 2018 and April 2019 (see Annex). 3. If not 'intrinsically motivated', intermediaries may be 'made' like-minded by an orchestrator so that they act as if they pursue a joint governance goal (Abbott et al., 2015b, p. 723). 4. The ECtHR found Italy guilty to return 24 people from international waters to Libya although they were already on board of the Italian armed forces and thereby under 'the continuous and exclusive de jure and de facto control of the Italian authorities' (ECtHR, 2012, para. 81). 5. Several international treaties such as the UN Convention on the law of the Sea (UNCLOS), the 1974 Safety of Life at Sea (SOLAS) Convention and the 1979 Search and Rescue (SAR) Convention specify the obligation to rescue people in maritime distress situations.