Ambiguous Outcomes of Returnees’ Land Dispute Resolution and Restitution in War-Torn Burundi

: Redressing land dispossession in the aftermath of violent conﬂicts is daunting and complex. While land dispute resolution and restitution are expected to promote return migration, this outcome is contingent upon the changing social, economic and political conditions under which return takes place. Drawing on qualitative data from Makamba Province in southern Burundi, this case study highlights the politically and historically shaped challenges underlying the resolution of competing and overlapping claims on land following protracted displacement. These include the undocumented and ﬂuid nature of customary land rights, institutional and legal pluralism and shifting land governance relations. This paper emphasises the centrality of the state in regulating returnees’ land dispute resolution and restitution processes. Violent conﬂicts and forced migration have enabled the state to expand its control over customary land tenure. The gradual exclusion or replacement of local authorities has shaped a competitive structure of jurisdictions and confused authority over land. National land restitution commissions have been used by the central government to shape land tenure and state–citizen relations and to exert pressure on land tenure institutions. Addressing competing claims on land following armed conﬂicts may fail if attendant struggles over public authority and changing political dynamics are insufﬁciently considered.


Introduction
Land governance in post-war settings can be contentious and complex. Considering the centrality of land and property to local livelihoods and poverty reduction, it is logical that land and property are a focal point for competition, disputes and tensions [1]. This has implications not only on sustainable return and peacebuilding but also on post-conflict land governance and state formation. For instance, studies in the Ugandan context have demonstrated how patterns of massive land dispossession by the government and military officers resulted in growing distrust and highly volatile relations between disenfranchised communities, customary institutions and government authorities, with the legal mandate to address land tenure issues [2,3]. Samuel Mabbike [4] further emphasised the importance of recognising the complexity of expropriation processes at multiple levels of governance. Post-war land policies and claims of belonging, entitlement and identity are sometimes used as strategies to exclude or mobilise individuals and groups. This situation can hamper peacebuilding efforts and the return process [1].
Ongoing debates are concerned with the urgency to address forced migration and increasing pressure on land in post-war reconstruction and development. Various land policies and tenure reforms seeking to resolve competing claims and enforce war victims' reparations have been implemented worldwide [5,6]. Proponents of land restitution argue that it can promote peace and the rule of law in post-conflict societies by supporting reconciliation and strengthening social and economic recovery [5,7]. Post-war land restitution can correct historical injustices by providing returning landholders or their descendants the necessary conditions to reclaim their pre-war identities, property rights and livelihoods [8]. The mechanisms and coverage of land dispute resolution and restitution initiatives, and the way they relate to legislation and customary land tenure, greatly differ from country to country [9].
In Burundi, armed conflicts and post-electoral violence (1965, 1972-1973, 1988, 1993-2005, 2015) caused massive and long-term displacement of rural populations [10,11]. Improving political and security conditions in the new millennium allowed refugees' return. By 2012, about 800,000 Burundian refugees had returned, reigniting grievances and competing claims over land [12]. Rampant land conflicts have affected rural communities and land governance. Land conflicts and land restitution became a prominent issue in national politics [12,13]. Following the 2015 post-electoral violence, over 300,000 people fled, of which only 55,000 refugees have returned [14]. With a new wave of return going on, land contestations remain a threat to the local sustainable and peaceful reintegration of returnees as well as to the national political stability and economic development [12,15].
Over 90% of the population depend on small-scale agriculture for their subsistence. At the same time, over 90% of litigation in local courts concerns land disputes [16]. Over the past twenty years, land dispute resolution and restitution have been addressed by different state and non-state actors. We will focus on how specific features of customary land tenure systems influence land relations after conflicts, and how different stakeholders involved in customary land dispute resolution and restitution processes compete for and contribute to constituting public authority in land governance. In doing so, we will attempt to shed light on the dynamic interactions between tenure insecurity, institutional and legal pluralism and the role of land politics in state formation. As will be observed, the multiplicity of actors has served to increase the presence and role of the central government in local land affairs, where undocumented land rights and contested tenure histories prevail. Land tenure reforms and the ability to endorse and challenge claims and to set up the (new) conditions or rules to be implemented imply momentous opportunities for public actors to exercise their authority and power. As will be shown in the case study below, some public authorities that hold up dominant political interests are maintained and rewarded, while other land governance authorities that seek local community interests are threatened and removed.
This article connects with the broader international literature on land restitution processes. Although there is a vast body of literature on post-war land restitution (e.g., [5,9,17,18]), land property restitution programmes in African customary tenure systems in general and particularly in Burundi are understudied. As demonstrated elsewhere, while significant changes in land policy may follow political transitions from authoritarian military to quasidemocratic rule, to break with the past and enforce new modes of governance, the outcomes of these changes may be mixed [17,19].
This article is organised as follows. First, we summarise the history of armed conflicts, forced displacement and land and property restitution programmes in Burundi. Second, we present a theoretical framework for analysing land restitution in conflict-affected settings. Third, we briefly present the research methodology used to collect data. Fourth, we present a case study of competing claims on customary land and restitution efforts in Makamba Province, located in southern Burundi, at the border with Tanzania to the east. Finally, we discuss the challenges and shortcomings of returnee-related land dispute resolution and restitution concerning state formation.

Violent Conflicts, Displacement and Land Restitution in Burundi: Historical Background
The outcomes and challenges of the land dispute resolution and restitution processes in Burundi are embedded in the complex history of war-induced forced displacement, multiple waves of partial return and many attempts to address competing land claims. This section shows that the aggregated effect of different approaches to the Burundian returnees' customary land issue since the 1970s has been a significant cause of confusion and tenure insecurity, with serious implications for a fragile and emergent peace.

Violent Conflicts, Forced Migration and Land Dispossession
Burundi is composed of three main ethnic groups, with the Hutu as the largest group, followed by the Tutsi and the Twa 1 According to René Lemarchand [20], the extreme violence in Burundi's post-independence era can be understood as the unforeseen outcome of a relentless competition among Tutsi and Hutu elites for control of the state and its resources. The roots of the first massive displacement of Burundians can be traced to the violent suppression of a 1972 Hutu insurrection by President Michel Micombero's regime. The 1972 bloodbath resulted in hundreds of thousands of people being killed and over 300,000 Hutu refugees. A legal dispossession scheme was enacted by the Micombero regime to punish anyone deemed involved in the Hutu insurrection. Over the 1972-1974 period, land and other assets, even bank accounts of the deceased, detainees and refugees, were seized and redistributed to army officers, UPRONA (Union pour le Progrès National) officials and their supporters and government authorities at the local, provincial and national levels [11].
As noted by Lemarchand, from this unprecedented wave of violence emerged a broken society in which the only elites and dominant class were the Tutsi; as in many areas including the southern regions, no educated Hutu men remained [20]. Until 1985, only Tutsi were granted access to wealth, power and authority. Access to the ruling regime, accumulation of land and other assets and engagement in violence hence became central elements of class formation [20]. The remnants of the Hutu society were subordinated and deliberately excluded from the civil service, military forces and the universities. Over time, Hutu shared memories of the massacre, dispossession and everyday experience of exclusion reinforced grievances and ethnic and political tensions over land.
Agrarian and legal reforms further promoted refugees' land dispossession and state control over resources and persons. Between 1973 and the 1980s, land redistribution was consolidated through the creation of state-owned development companies and implementation of agrarian reforms such as the Imbo Development Regional Company in 1973, the Nyanza-Lac Development Project in 1977, the Rumonge Integrated Rural Development Project in 1978 and the Rumonge Regional Development Corporation in 1980. Further, a Land Code was passed in 1986 besides other legal reforms [21,22]. These changes contributed to expanding legal land dispossession and state control over productive land, resources and people. In a parallel non-political process, the land and property of those who fled were appropriated by remaining community members, relatives, local authorities and early returnees, as coping mechanisms against growing demographic pressure and poverty [11,23].
In the early 1990s, the political balance of power shifted temporarily to a Hutu opposition party led by Melchior Ndadaye, enabling the return of 55,000 Hutu refugees. Their return sparked a large number of land conflicts and made land restitution and control over territory and people central issues in both national and local political discourses [20,24]. The assassination of President Ndadaye in October 1993 triggered a 12-year civil war which caused more population displacement and property loss. Over the subsequent years of violence, insecurity and chaos, contested land was passed to subsequent secondary occupants, both Tutsi and Hutu, often backed by statutory laws and policies. In addition, the authority and legitimacy of customary institutions in local land governance gradually eroded. Such a reshuffle of land tenure resulted in confusing landholding boundaries and complex and overlapping property claims. A fragile peace between 2005 and 2014 gave way to another civil uprising around the 2015 presidential election, with further forced displacement. This short political history illustrates the complexity of violent conflict legacies, population movements, competition to access land and contestations over public authority [25].

State-Led Land Restitution Initiatives in Historical Perspective
In Burundi, land restitution initiatives were part of a post-war political agenda to promote return migration and address dispute resolution and unsteady land tenure relations. Initial attempts to solve land restitution issues occurred under the government of Jean Baptiste Bagaza (1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987), which articulated a political discourse of national unity and post-war reconstruction which recognised the land rights of Hutu refugees. This political discourse was backed by legislation reforms regarding land governance and land restitution. A law incorporating customary land within the state land domain was issued in 1976, institutionalising statutory authority over customary tenure. The following year, the National Commission for the Rehabilitation of Returnees (Commission Nationale pour la Réhabilitation des Rapatriés) 2 was established to prompt refugees' return and address their land claims. Yet, the work of the 1977 Commission was criticised as it contributed to converting vacated customary land into the state land domain and formalising refugees' land dispossession [26]. While a small proportion of returnees managed to reclaim their customary properties through mediation within social and kinship networks, it was more difficult to regain access to ownership of land occupied by people closely linked to the ruling party or under the control of national agricultural development companies [23,26]. In 1991, President Pierre Buyoya (1987-1993, 1996 created the National Commission in charge of the Return, Reception and Reinsertion of Burundian Refugees (Commission Nationale chargée du Retour, de l'Accueil et de l'Insertion des Réfugiés Burundais) 3 to resolve returnees' land claims and resettle them on state land whenever possible. However, this had little impact as the majority of land occupation had been formalized, and development projects on state land left little to no room to accommodate the needs of Hutu returnees. Moreover, legislation eroded the authority of customary leaders in dispute resolution and increased the power of judicial authorities in regulating customary land disputes. Secondary occupants' claims were recognised and endorsed at the expense of returnees' claims. In this context, some returnees considered returning to former countries of refuge as a more viable livelihood, considering the comparative benefits they derived from humanitarian packages, including access to larger land plots (Interviews with former 1972 refugees, Makamba, 2013-2014).
Soon after his election in July 1993, President Ndadaye replaced the 1991 land restitution commission with the National Commission in charge of the Return of Refugees (Commission Nationale chargée du Rapatriement des Réfugiés). This change sparked widespread fears of property loss among both Tutsi and Hutu land occupants in southern territories, exacerbating social and political tensions [27]. The 1993 land restitution commission barely survived four months of operation and was cancelled right after the assassination of President Ndadaye.
The 2000 Arusha Peace Agreement (APA) revisited the returnees' land question by endorsing the creation of a new commission-the National Commission for Rehabilitating War Victims (Commission Nationale pour la Réhabilitation des Sinistrés, CNRS)-and the revision of the 1986 Land Code. The CNRS was a political compromise. Its main dispute resolution approach was reconciliation and mediation towards a formal/written landsharing arrangement between returning families and secondary occupants [28]. In a transition context characterised by rising population, land scarcity and rampant land disputes, land-sharing agreements were considered an acceptable and fair solution-at least temporarily.
However, soon after assuming power in 2005, President Pierre Nkurunziza promoted major political changes around land dispute resolution and restitution. President Nkurunziza was a Hutu, a 1972 refugee and leader of the former militia and current ruling party, the National Council for the Defense of Democracy and Forces for the Defense of Democracy (Conseil National pour la Défense de la Démocratie et Forces de Défense de la Démocratie, CNDD-FDD) [29]. One of the first reforms of his government was to suspend the CNRS and overrule its decisions on returnees' land disputes. In 2006, the CNRS was replaced officially by the National Commission on Land and other Assets (Commission Nationale des Terres et autres Biens, CNTB) 4 . The next sections focus on the 2006 land restitution policy, its amendments and the challenges of its implementation in practice.

Contemporary Framework of Returnees' Land Dispute Resolution and Restitution
According to the 2000 APA, all victims of previous armed conflicts are entitled to reparations from the state. The restitution scheme initiated in 2006 was guided by the principle of the gradual rectification of land and property loss, and injustices in land restitution committed under previous regimes. The CNTB was authorised by the central government to adjudicate not only the land claims of displaced people but also the state's claims to land wrongly allocated by former regimes [16]. Although restitution encompasses the return of the property lost, the programme was criticised for its poor coordination with existing land governance structures and for not providing means to formalise the recovered property rights [12,30]. The 2000 APA also provided for alternative restorative mechanisms such as restitution of equivalent land and monetary compensation when the restitution of the claimed property is not possible 5 . Yet, lack of sufficient funding and financial resources did not allow the state to provide such a remedy.
The Some of the main changes brought through these legal instruments include the following: These changes redefined the functional jurisdiction of land restitution structures and shifted public authority in dispute resolution from the local to the central government. They also resulted in confusion and increased competition over the responsibilities for settling land disputes and the rules to apply. In the next section, we provide a theoretical approach that supports the analysis of the data collected in Burundi.

Theoretical Framework: Researching Post-War Land Restitution
This article builds on concepts of legal pluralism and state formation to analyse land governance and its impacts on land tenure in war-torn settings. Drawing on Christian Lund's notion of rupture and legal and institutional plurality, we argue that moments of rupture following periods of conflict, displacement and return are central in showing that post-conflict land access and tenure do not simply emerge from public/political authority but also constitute political authority [31]. An empirical analysis of post-conflict governance processes related to land and property restitution involves scrutinising what Christian Lund labels 'the ruptures'-concerning "open moments' when opportunities and risks multiply, when the scope of outcomes widens, and when new structural scaffolding is erected' [31] (p. 1202). The use of the term 'post-conflict/post-war' here is to denote a break from a previous social configuration, without implying that the causes of insecurity and forced displacement have disappeared and that return migration is completed. Postconflict land governance processes during moments of 'rupture'-such as pre-electoral and post-electoral periods, or when forced migration occurs-are favourable moments for analysing the reordering of property, authority and legitimacy between public authorities and the society.
As Jon Unruh argues 'the confusion, competition, confrontation, yet importance of seeking secure access to rural lands during and following civil conflict results in a particularly problematic emergence of multiple normative orders for attempting to legitimise land access, claim, and use' [32] (p. 353). In war-torn settings, land governance is not the prerogative of one specific institutional actor. Institutional pluralism prevails, with different authorities in charge and referring to different normative frameworks. At the same time, competition may occur regarding who are the rightful owners of the land, who oversees land access and which rules apply. Post-war authority structures and sources of legitimacy are shaped and affected by factors such as legal and political changes, political awareness and mobilisation among groups with competing claims [33,34]. Post-war land restitution authorities function in diversified land governance systems in which rival claimants interact with different land governance authorities, and through which state authority is shaped or reinforced [35][36][37].
Consequently, the intersection between land restitution and land governance authorities can highlight struggles over authority and sources of legitimacy. Conflicts and changes in the social, economic, legal and political conditions further fuel institutional competition. In this article, we view land restitution programmes as arenas in which claimants enter processes of negotiation and litigation, and state and non-state actors interact. By defining those who are eligible and ineligible for restitution and endorsing certain claims, the postwar state and land policy reforms reshuffle land relations while (re)producing exclusionary relations [8]. This article presents a detailed account of how land dispute resolution aiming at property restitution works in practice and how land restitution authorities, existing public authorities-elected hill/sub-hill officials, customary elders (Bashingantahe), communal and provincial administrative officials, judicial officials-and community members interact in practice. As land restitution policy and legal reforms are passed at the central level, they create structural opportunities that legitimate land governance actors and influence the organisation of public authority in rural settings.
These processes of institutional competition are closely connected to state formation processes, including changes in legitimacy, authority, control, different interpretations of property and new practices of formalising land rights. Property is understood in this article as a relationship among social actors concerning the use or benefit of something of value [37]. Concerning land, property takes the form of legitimate claims to land whether legally sanctioned or not [37]. Struggles over property are as much about the scope and constitution of public authority in land governance as about access to land, because of their entanglement with contested authorities and jurisdictions [37,38]. In practice, land authority can be vested in diverse national, provincial, communal and local actors in the forms of territorial jurisdiction, functional jurisdiction over land or jurisdiction over persons about land issues [38]. When considering the interactions between levels of authority and patterns of land access and use in particular settings, struggles over land and authority must be addressed as dynamic political processes [39]. In the African context, while competing claims to land ownership may relate to changing land tenure grounded in longstanding and complex customary tenure histories, disputing parties can refer to different stakeholders, invoking different or overlapping jurisdictions in land governance, competing norms of tenure and different interpretations of the 'past' and 'space', which may blend in different ways [38,40].
Despite the enthusiasm in some studies on negotiability, scholars such as Pauline Peters are doubtful and indicate how certain actors and institutions are advantaged when it comes to land tenure forum shopping [41]. Diverse studies acknowledge the agency and the capacity of migrants to strategically challenge post-war land tenure authority structures and sources of legitimacy [5,32,42,43]. In ethnically diverse settings, competition over land is likely to be perceived locally in terms of higher-level contestations about identity, belonging, land control and authority, which may fuel broader political mobilisation and controversy [2,44]. As the data will show, land claimants, local government authorities, land restitution officials and powerful politicians are key players in shifting the power dynamics to influence local land restitution. Moreover, they demonstrate how sanctioning competing claims to land with the purpose of land restitution is often tied to asserting shifting polity and authority in local land issues. By highlighting the complexity of dispute resolution and restitution processes, we raise questions about the extent to which a focus on physical restitution of property effectively contributes to land dispute resolution and long-term return.

Methods
This article is based on data collected over a period of 19 months of ethnographic field research in Burundi (from June 2013 to November 2014, and in April 2019), as part of the wider research programme 'Grounding Land Governance -Land Conflicts, Local Governance and Decentralisation in Post-Conflict Uganda, Burundi and South Sudan' that aimed to document post-conflict land governance in a comparative perspective [45] 7 . Makamba Province, in southern Burundi, was chosen because it has been the site of violent conflicts, massive displacement and multiple waves of partial return since 1972. Since the mid-2000s, it has been one of the top three regions of the country with the highest influx of returning refugees and the highest number of returnee-related land restitution claims [46,47]. We conducted 95 individual interviews with key informants, including government officials involved in land restitution at the local, provincial and national levels, government authorities at various administrative levels (hill, zone, commune and province 8 ), returnees, secondary occupants, customary elders (also known as Bashingantahe), judicial officials and representatives of non-governmental organisations and communitybased organisations. In addition, we held 27 focus group discussions with members of the communities in which return took place. The interviews and focus group discussions took place in nine rural communities in Makamba Province.
The interviews and focus groups explored property claims and how these were addressed through dispute resolution and restitution and followed a semi-structured approach. This allowed similar research questions to be addressed while keeping enough flexibility when participants raised other relevant issues. During the discussions, research notes were taken in either French or the local language (Kirundi), which were later translated into French. The selection of participants was purposive. Interview transcripts were analysed in NVivo to organise the gathered information, generate key themes and highlight patterns. Direct quotations from interview transcripts were used to support the analysis and to preserve informants' voices. Field research was enhanced by secondary sources, including key policy documents, reports, published articles and newspapers about legal changes, the return process, land disputes and people's experiences of land restitution.

Findings: Returnees' Land Dispute Resolution and Restitution in Makamba Province
There is much at stake on the ground concerning the land dispute resolution and restitution processes in terms of institutional competition, political stability and household livelihoods. Interviewees, either returnees, secondary occupants (also called 'residents') or stakeholders involved in the settlement of land restitution claims, expressed disappoint-ment in the handling of return processes. In the early 2000s, government authorities assured Burundian refugees that they will recover all their land and other assets (houses, livestock, etc.) left before the exile, or receive equivalent plots or financial compensation upon their return (Interviews, returnees, civil society agents, 7 March, 13 August, 2 October 2014) [48]. Secondary occupants were told they would receive financial aid to implement community development projects, and formal written proofs of land ownership rights if they welcomed returnees peacefully and accepted land-sharing arrangements (Interviews, hill authorities, occupants, 16 November 2013, 6 December 2013, 26 February 2014). These promises raised unfulfilled expectations to both returning refugees and secondary occupants about the outcomes of land dispute resolution. In addition, both local-and higher-level land governance institutions were actively involved in the settlement of land restitution claims amidst ongoing legal changes and political challenges, which became entangled with the process of post-conflict state formation.

Unclear Boundaries and Competing Oral Tenure Histories
Land dispute resolution and restitution face several challenges on the ground. The features of the customary land tenure systems are among them. Customary land tenure in Burundi is based on traditional land occupancy and orally transmitted ownership rights with boundaries of lineage and family lands demarcated using natural features such as hardy perennial plants, trees and rocks. The unclear and undocumented nature of land boundaries and the erosion of tenure histories, due to protracted displacement and reconfiguration of land occupancy, are major constraints in post-return land access and the resolution of returnees' land disputes. In general, there are two main categories of returnees, namely: returnees 'without references' (rapatriés sans références) and returnees 'with references' (rapatriés avec références). The first category refers to descendants and relatives of long-term refugees who were not able to remember or locate their (parents') places of origin. They either were accommodated in peace villages by the Burundian government (see, e.g., Falisse and Niyonkuru [49]) or developed coping strategies to access land for residential and farming purposes (rental, sharecropping, purchases, borrowing from local churches, etc.). The second category of returnees was able to remember and identify their (parents') places of origin and hence make claims on family/lineage land.
Returnees' land claims are generally ambiguous and difficult to resolve satisfactorily. This ambiguity stems from the difficult task of locating and recognising the land formerly occupied by returning lineage/family heads and inherited by returnees from their parents or relatives. Upon their return, all interviewees who (or whose parents) left during the 1972 conflict had found their land occupied by others, as illustrated by an elected hill official: There are many land conflicts here because, in 1972, many people fled to Tanzania, Congo and elsewhere. After a few years, the state urged people from Bururi, Kayanza, Gitega and Muramvya provinces to come and settle on the vacant land. There were also people from the provinces of Rutana and Bujumbura rural who had arrived of their own accord since 1981. As soon as the refugees started to repatriate, they found their properties occupied by other people (Interview, elected hill official, 16 November 2013).
In addition, in all nine communities we visited, natural landmarks demarcating customary lands before displacement either had been dislocated or had disappeared. Even when the returnees could identify their hill/place of origin (or that of their (grand-)parents) and locate the family land, the loss of memory or a reorganisation of the space that occurred during the exile generally prevented the boundaries from being established. Additionally, elders and former neighbours were sometimes reluctant to come forward as witnesses to customary claims. Even when witnesses were available, stakeholders in dispute resolution were confronted with conflicting testimonies from the disputing parties. Contradictory testimonies forced dispute resolution actors to rely heavily on personal judgment while making decisions.
For example, in a land dispute case, three Hutu returnee brothers (and their thirty sons) who returned in 2011 made claims to several hectares of land on which over 170 households' subsequent occupants bought plots of land from a former Tutsi local government official. The latter used his position within the post-1972 communal administration and connections with higher-level politicians to occupy the vacated land. During the CNTB hearings, the returnee brothers and their witnesses had declared the land was granted to their father several decades ago by former local representatives (batware) of the Burundian king (Mwami) 9 . However, occupants argued that, when they investigated the land tenure history before purchasing the land, different persons testified that a portion of the land was the property of the state, and another portion was grabbed by the returnees' father from the widow of a former foreign trader. Additionally, the returnees' witnesses had stated that the land belonged to the state in earlier CNTB hearings, but they had changed their testimonies to back the returnees' land histories in further rounds of CNTB hearings (Interviews, occupants, 8, 15, 17 July 2014). As there was no official local land registry to corroborate or challenge these claims, in 2014, the CNTB ruled in favour of full restitution to the returnees. Yet, several occupants surmised that the witnesses and CNTB officials had received bribes from the returnees (in the forms of money and pieces of land) and that the CNTB ruling was biased and influenced by the membership of the returnees to the ruling party (Interviews, occupants, 8, 15, 17 July 2014).
The lack of documentary evidence to prove ownership of land, memory loss and biased testimonies of some returnees about their land claims created confusion. This lack of clarity is an opportunity, on one side, for returnees to reinvent the 'past' and legitimise claims in strategic ways and, on the other side, for occupants to shape their narratives and to legitimate counterclaims in ways meant to conceal dubious land acquisition. To illustrate such confusion in the restitution claims, a key informant shared the following: The piece of land where the Pentecostal Church is built is also in dispute. The Church received it after 1972, but the church representative has claimed that the claimant is a liar, because the land did not belong to him. There is a returnee who claimed a part of the land occupied by the Catholic parish, saying that this part belonged to him and the CNTB decided that the parish must return it to him. We have also heard that there is a returnee who is claiming the land on which the commune's office is built (Interview, occupant, 17 July 2014).
In some cases, land occupants invoked legitimate claims because their land acquisitions were backed by government officials. Yet, in certain cases, CNTB officials did not recognise the claims of so-called 'bona fide' purchasers of the land claimed by the returnees, although they provided documents to prove the legitimacy of their land acquisition. The truthfulness of these documents was deemed problematic, as the national land commission continually questioned the authenticity and validity of the formal documents which were issued by previous administrative authorities, under Tutsi-dominated UPRONA governments that had condemned the Hutu landowners to death, forced many into exile and then ordered the seizure and redistribution of their property. Furthermore, as argued by the CNTB chairman during a workshop on land governance, official written land-related papers issued during the past years of armed conflicts and forced displacement could in no way provide legal certainty to secondary occupants (Field notes, Makamba, 9 June 2014). As more returnees, especially those who previously agreed to land divisions with occupants, became aware of this CNTB perspective on occupants' claims, they started to encroach over and even remove the land boundaries with occupants. This situation created a new round of claims in which local authorities intervened with mixed results, as noted by a key informant: The other day, four elected local officials, and three bashingantahe, including myself, met to resolve a land dispute between the head of a returnee family and the head of a resident family. These two people had divided the contested land into two parts and had obtained CNTB official attestations. But the returnee's son, while cultivating the field, exceeded the boundaries between the two plots that the CNTB had placed. The resident then filed a complaint before the first elected hill official, and the latter called some elected hill and sub-hill officials and bashingantahe to join him in settling this issue. The head of the returnee family explained that it was his son who had encroached on the land boundaries. When we asked his son, he said that he considers that the plot belongs to his father and that he does not know the resident. We advised them to respect the land boundaries set by the CNTB, and that whoever is not satisfied with this decision should transfer the case either to the court or to the CNTB. The two heads of household agreed that we put back the boundaries of the plots in their place (Interview, customary elder (mushingantahe), 19 March 2014).

Multiplicity of Actors and Changing Relations in Dispute Resolution
The management of restitution claims involves a diversity of stakeholders and institutions with different jurisdictions and changing relations. At the local level, elected hill officials (conseillers/chefs collinaires) and other community leaders-including the chefs de Dix ménages (ten-household chiefs), chefs de sous-colline (elected sub-hill officials), chefs de zone (appointed area chiefs) and customary elders (bashingantahe, singular: mushingantahe)mediate local land disputes. In this capacity, they support the operations of the CNTB on the ground. According to the statutory guidelines, returnees should initially bring their claims to the local authorities. Elected hill officials are authorised to allocate a small piece of land, temporarily, to accommodate returnees' households pending the outcome of the dispute settlement process.
The ubushingantahe institution is a precolonial and customary institution of 'wise men' (bashingantahe, elders) whose main role was-and still is-to manage and resolve disputes that arise within the community, through conflict resolution mechanisms including negotiation, mediation, reconciliation or arbitration [50]. Although the Municipal Law of 2010 10 divested the bashingantahe of any legal power, function or responsibility concerning local dispute resolution, their role in managing and resolving conflicts remains important. In practice, they are still involved in local dispute resolution processes. Their legitimacy and authority at the locality are reinforced when they are invited by the elected hill officials to participate in the land dispute negotiations. In some instances, sub-hill and hill officials also held positions within the bashingantahe councils. When disputes cannot be resolved at the community level, claimants are directed to government representatives at the area level, and further at the communal level to the chefs de zone (area/zone chiefs), the autorités administratives communales (communal administration authorities) or the judges at the tribunal de résidence (local court). When land disputes are not settled at the communal level, claimants are directed to higher-level government authorities, as shown in Figure 1 below. Some claimants may again turn to local authorities when they lack financial resources to further their claims to higher-level institutions. Local authorities are not remunerated for their role in the mediation of returnees' land disputes and their involvement in the enforcement of rulings rendered by higher-level institutions.
The dispute resolution and restitution processes have proven to be extremely demanding in terms of institutional capacities and coordination, resulting in an institutional saturation. Each case is nuanced, usually involving a diversity of claimants and stakeholders, and multiple layers of restitution claims. In one case, a returnee family filed multiple claims to recover its land. Initially, the household of six members (husband, wife, three sons and one daughter) fled to Tanzania due to the 1972 conflict. In 1988, the elder son temporarily returned to locate and claim back the family land that was occupied by two families. Unfortunately, his claim was denied by local authorities as he was wrongfully accused of malicious intent against the government. He left for Tanzania and later returned in 2008 with his parents, 2 brothers and over 30 family members (wives, children, grandchildren). Although the head of the returnee family did not know the size of his pre-war land holdings and exact boundaries, he was able to identify the two primary occupant families and to retrieve a portion of the land from them, with the help of local authorities and CNTB officials. However, one occupant refused to sign the written CNTB land-sharing attestations on the ground that the division was not equitable. While the second occupants filed a claim at the CNTB provincial office for a new division of the land, the returnee family intended to make a new claim at the CNTB provincial office to recover the whole land, including plots that were sold to third parties by the early occupants and not disclosed during the first round of land division (Interviews, returnees, 12 February 2014; occupants, 26 February, 24 October 2014). Coordination between different levels of governance was also a challenge. Each case of restitution claims was attended by stakeholders with their own experience of past conflicts and displacement. Whether the stakeholders involved were former refugees or not had implications for the perceptions of opponents in land disputes and the (un)fairness of the terms of land-sharing arrangements. For instance, several returnees interviewed affirmed being cheated in land allocations, as they supposedly received the smaller share or the less fertile part of the land (Focus group, 12   Coordination between different levels of governance was also a challenge. Each case of restitution claims was attended by stakeholders with their own experience of past conflicts and displacement. Whether the stakeholders involved were former refugees or not had implications for the perceptions of opponents in land disputes and the (un)fairness of the terms of land-sharing arrangements. For instance, several returnees interviewed affirmed being cheated in land allocations, as they supposedly received the smaller share or the less fertile part of the land (Focus group, 12 December 2013; interviews, returnees, 13 November The policy and legal changes between 2009 and 2019, unclear tenure histories and blurred land boundaries were linked to the changing relationships between land restitution officers and other state and customary stakeholders. For instance, the active role of local elected officials and customary elders in the settlement of returnee-related land disputes was gradually minimised by amendments to the CNTB legislation since 2010. The 2013 policy changes which reinforced the power of the national commission in charge of the resolution of returnees' land disputes and promoted full restitution by the CNTB created momentum for returning groups to disregard local mediation processes, a huge popular demand for revising previous sharing arrangements and even fraudulent claims on large land areas [51]. As illustrated by this quote from an interview transcript with an elected hill official, this situation opened the terrain for institutional competition, fraud and increasing informal land transactions: When the returnees came, they were welcomed and hosted temporarily by their friends or family who had settled here, or by local churches. They would then make inquiries to find out where their plots were located and who was occupying them. They later made themselves known to the land occupant families or their representatives, who directed them to the local elected officials. We collaborated with the bashingantahe of the hill to help these people to share the claimed land. We would collect information about the land history from the local elders who had never fled because they were the ones who would know the realities of land tenure. At the end of the investigation and mediation process, the claimed plots were divided among the complaining families. In the beginning, the CNTB agents came to support us and monitor our practices. When they saw that we were used to doing this, they reduced their visits. But, in 2010, when there was a change in leadership at the CNTB, the new CNTB agents accused us of having done poorly in the sharing of plots, of having committed injustices, and of having received bribes. Since then, the commission has informed us that it will be the only one to settle disputes involving returnees and to decide on plot divisions. In 2013, the commission changed its approach once again. There is no longer any question of land sharing arrangements between returnees and occupants/residents. The commission demands that residents return all occupied plots... Therefore, the returnees who had previously accepted land sharing agreements submitted new complaints to the CNTB directly, demanding the recovery of entire family plots . . . Today we observe powerful people who have never fled or who arrive from Tanzania, who go to the CNTB to falsely claim the restitution of plots of land that are occupied by other people. They bring false witnesses and the CNTB assigns these plots to them. They then sell the plots (Interview, elected hill official, 7 January 2014).
While ongoing legal changes prolonged the land dispute resolution process, they provided opportunities, to some extent, for local institutions to step up in local land governance as land claimants redirected their claims to local dispute resolution arenas as a last resort against the central government.
Changes to the CNTB and CSTB regulations often contradicted national legislation. The shifting role of the land restitution commission emerged as a cause of growing criticism from civil law experts. By endorsing the land restitution commission as the sole state agency with the initial and final decision on land disputes, various law experts consider the 2010 and 2013 CNTB laws as anti-constitutional in the Burundian context [52,53]. Aimé-Parfait Niyonkuru has shown that the failure to assess the nature and eligibility of claims and the biased interpretations of legal instruments cast doubts on the authority and functional jurisdictions of the CNTB and CSTB in certain land dispute/restitution cases [54].
With the termination of the UNHCR funding to support the CNTB's operations, the CNTB's operations considerably slowed down in 2015. Nevertheless, the commission relaunched its activities on the ground in late 2016. The 2019 CNTB legislation set the deadline of 13 March 2021 for returnees and occupants to register their claims at the commission and re-established judicial courts in dealing with returnees' land claims after this delay. Notwithstanding the official CNTB statistics stating that, since 2006, 44,142 dispute cases have been settled, in May 2021, there remained about 25,437 pending cases including 22,374 cases at the provincial level and 3063 cases waiting to be dealt with at the head office [55]. In the meantime, while return migration is ongoing, returning refugees can no longer file their claims with the CNTB and are confronted with changing rules and roles in dispute resolution and restitution.

Deepening Elusiveness and Insecurity in Land Tenure
Throughout the history of land restitution programmes, the issuing of written attestation papers as a means of formalising dispute resolution decisions has gained pace. Our findings show that the state, through the issuance of CNTB land attestation papers, has contributed to uncertainty in land tenure in ways that enhance unclear land rights and land disputes. CNTB-led dispute resolution processes led to the issuance of two forms of land attestation papers carrying different types of information, signed and stamped by CNTB officials, and printed on A4-sized papers. The first type of CNTB attestation paper was issued when land dispute opponents agreed on a land-sharing arrangement through mediation led by elected hill officials. The second type of CNTB allocation paper was issued when a party was dissatisfied in the first round of dispute resolution at the community level and appealed to the CNTB provincial or central officials to settle the land disputes. In this way, the CNTB endorsed two levels of authority over land disputes.
In the first case, the elected hill officials, with the support of local authorities and bashingantahe, would oversee the land allocation process and write a short notice (minutes of the local resolution process) that the claimants and their witnesses would bring to the communal administration office. A communal administration official who served as the CNTB contact person was mandated to write the land-sharing report to be signed by the different parties, their witnesses and two to three representatives of the communal administration. This document was transmitted to the CNTB provincial office which issued the official, signed attestations of the land-sharing agreement. Copies of this document were sent back to the land dispute opponents through the communal administration office. Although many returnees were not satisfied with these agreements in the first round of dispute resolution, the CNTB attestations of the land-sharing agreement provided a sense of 'security' in the perspective of future rounds of dispute resolution.
In the second case, the CNTB officials were directly involved in the investigations, with or without collaborating with elected hill officials. As land dispute opponents often did not agree on the land division, this situation contributed to reinforcing the authority of CNTB officials and legitimising their role in making and enforcing decisions that produced winners and losers. A quotation from a CNTB attestation paper issued on 25  Has noted that: Since BL addressed the commission saying that he fled in 1972, he was repatriated in 2008 and he found that part of his plot was occupied by CE and that he [CE] sold another part to BJM; Since CE explains himself by saying that he requested the plot and that it was allocated to him [by the 1970s local government officials]; Since CE claims that the sharing of this land plot had been carried out by the hill elected officials, but that the commission sees that this was done by an institution that has no jurisdiction; Since CE does not deny that he sold this plot and this is confirmed by BJM; Since after having bought this land plot, BJM built a house where he lives until today; Since BL affirms that he has no conflict with the buyer [ Concerning the remaining plot, which BJM has bought from CE, they will agree (the buyer BJM and the seller BL).

Article 3:
Whoever feels aggrieved by this decision must appeal to the higher authority of the National Commission in charge of the Resolution of Land and other Property Conflicts, within two months since he/she is aware of this decision, as provided for in article 17 of law n • 1/01 of 4 January 2011.
If the deadline is exceeded and no one has lodged an appeal, the decision shall be enforced.

Implementation order:
"The President of the Republic orders and directs all administrative and police authorities to assist in the execution of this decision whenever required to do so." Although the CNTB attestation papers were intended to inform landholders and land tenure authorities that land plots had been (re-)allocated to claimants, they failed to clarify which property rights had been recovered or allocated. They did not specify or remained evasive about the contested and restituted land location and area, the nature of the recovered land rights, the names of family members entitled to the allocated family/lineage land and any compensation to the parties for losses (crops, trees, houses, etc.).
This uncertainty was further exacerbated when the CNTB instructed and authorised the winning parties to claim financial compensation from their opponents, should they wish to maintain their access and use of the restituted land. As the CNTB attestation papers provided no instructions regulating how these negotiations should be organised (e.g., second article in the quote above), this provided room for manoeuvre for the winning parties to request exorbitant fees from the losing parties. Shockingly, taking advantage of the overall low literacy of rural populations, these transactions were sanctioned by some sort of 'informal/small papers' in which relevant information-about the nature of the transferred rights, the location and size of the transacted land, the names of the family members entitled to the transferred/allocated land-was (purposefully) left out and written by 'secretaries' receiving payments from the winning parties. The latter role was controlled, informally, by some communal, area or hill administrative officials.
For example, a large land area in Makamba Province occupied by over 75 households was claimed in 2011 by the descendants of a 1972 refugee who died in Tanzania. Most land occupants were government civil servants and small business owners, who bought land plots from a former local government official. After receiving the land through the land redistribution government scheme after the 1972 armed conflict, he divided the land area into smaller plots and sold them to new settlers originating from other provinces. After a cumbersome and contested dispute resolution process, the CNTB ruled in favour of full restitution to the returnees in May 2014. Occupants wanting to maintain use of the land plots were ordered to negotiate financial compensation with the returnees. Fees were set by the winning parties, often at the disadvantage of losing parties, and to be paid in cash. Although many occupants managed to meet this requirement, the informal written proofs of land transactions generally did not specify relevant information, such as the nature of the transaction (loan, sale/purchase, etc.), the nature of transferred land rights, the transaction fees, land location, land area or signature of losing parties. For instance, although a widow and small businesswomen compensated the returnees over 500,000 BIF 11 for two adjacent land plots, the content of the informal written paper she received was very puzzling, as illustrated below: [Title of the small paper:] Agreement made for a plot of land which is in RN's property: I, RN, and my elder brother MR confirm that the plot is now in the hands of Mummy X. who is NC. As of today [15 May 2014], she has the right to it. She can build or use it as she wants.
Moreover, returnees were often not eager to go to the communal administration office to formalise these transactions to avoid paying the 10% fees on land sales required by statutory legislation. Across the nine communities of the study, winning parties in a dispute were unenthusiastic about the idea that losing disputing parties might register and formalise their land rights retained on a share of the contested land or after making payments. Some feared land registration might prevent them from appealing contentious dispute resolution decisions, while others were concerned that land registration might lead to the individualisation of family land property in a context where some relatives/siblings have not yet returned from exile.

Controversial Roles of the CNTB in (Re-)Activating Tensions and Violence
The failure of the CNTB to address competing and overlapping land claims has further poisoned the tenuous relationships between citizens and the state. As a result, violent conflicts and land dispute resolution continue to play a critical role in the broader land policy in Burundi.
Towards the 2015 presidential election, the government's failure to address competing land claims was used by political opponents to frame narratives of ethnopolitical bias [12,56]. CNTB authorities invoked polarising narratives about the role of previous violent conflicts in the systematic dispossession of Hutu refugees and framed the claims of secondary occupants on contested land as illegitimate [46]. In contrast, civil society activists and opposition parties argued hotly for a distinction between legitimate and illegitimate secondary landholders. From their perspective, legitimate secondary landowners included people who were relocated through the state resettlement programmes, and illegitimate secondary landowners referred to people who encroached on refugees' land through a non-political process, using informal/illegal means (such as boundary encroachment, illegal sales and land grabbing).
Over the period 2010-2014, the CNTB leadership persistently argued for the rejection of land-sharing arrangements in favour of full land restitution as the most desirable remedy to correct Hutu returnees' claims. This shift in land dispute resolution increased tensions, intimidation and violence within rural communities. The inflaming role of the national commission is illustrated in the following extract from a blistering speech of the CNTB chairman in Nyanza-Lac Commune on 13 March 2014, as occupants were threatened with being evicted from the land unless they gave it back to returnees: In the past, there was a Kirundi proverb, that is no more relevant nowadays, which says: 'A virile man is the one who eats his share and that of others.' A person behaving like this today is like a dog clinging to a bone that does not belong to him. Therefore, occupants must return to their communities of origin and give back returnees' land and properties! Otherwise, the CNTB will evict them by force! (Interviews, returnees, residents and local authorities, Makamba, 13 March, 10 April, 16 April, 5 May, 29 October 2014. Informal conversation, Makamba, March-October 2014).
A few days later, key informants reported that this speech emboldened returnees to trespass into residents' plantations to steal farm products. Returnees, however, shared they had heard rumours of secret meetings among residents to prevent future evictions. Returnees' houses were targeted in the evenings; stones were thrown at their roofs and windows; anonymous threatening messages were left at their doors; and the authority of returnees who were elected as local officials in 2010 was contested. Unrest was further promoted by UPRONA politicians, judicial officers and civil society activists through press conferences, radio programmes and community events [52,57]. Ultimately, this situation evolved into violent resistance against the enforcement of CNTB judgments and eviction notices in various communities. Residents were allying against returnees, CNTB authorities and their escort (local officials and the police).
Most interviewees considered the CNTB as a political instrument of a CNDD-FDD hidden agenda for (re)claiming territory and (re)producing dispossession and exclusion. Moreover, occupants/residents marginalised in dispute resolution regarded Hutu returnees who cancelled previous land-sharing agreements and filed new claims to retrieve full land rights as allies of the ruling party. The growing turndown of occupants' claims fueled a sense of mistrust and disenchantment against the national land commission and the state, as observed by a key informant: The land-sharing approach enforced by the CNTB before the nomination of Bishop Serapion as the CNTB chairman was good and well appreciated by local people. However, taking the land away from us has offended us a lot. We are not angry against the returnees, but against the state which is dividing us . . . There is no future! There is no peace! The only hope is that the CNTB will change its approach because it does not help people to be reconciled; instead, it is dividing people (Interview, land occupant, 17 July 2014).
In July 2014, a group of occupants addressed a joint letter to the representative of the Secretary-General of the United Nations in Burundi, with copies to government representatives at the communal and provincial levels and to President Nkurunziza to denounce the detrimental outcomes of the CNTB intervention in local communities. This communication also warned about imminent bloodshed in case immediate actions would not be taken to reverse the land restitution approach. Unfortunately, this action did not receive the expected support, which encouraged rural community members, particularly secondary occupants, to mobilise and obstruct the enforcement of the CNTB eviction notices between 2014 and early 2015 (Interviews, BNUB (United Nations Office in Burundi) agent, local government officials, occupants, Makamba, July 2014).
To defuse rising tensions and violence, local government authorities organised community meetings to sensitise local populations on reverting to locally based dispute resolution solutions in the form of informal land-sharing arrangements. Additionally, returnees were encouraged to withdraw their claims and appeals from the CNTB provincial and central offices. This initiative received the support of bashingantahe and the Makamba governor, despite the disapproval of the CNTB authorities (Interview, Communal administration, Makamba, 10 April 2014).
Yet, in Makamba, as the 2015 election approached, the central government was forced to take serious actions to calm down local populations and CNTB authorities. President Nkurunziza stepped in to pacify the situation. In February 2015, the CNTB central office ruled for full restitution in 240 land dispute cases in the communes of Mabanda, Nyanza-Lac and Kibago in Makamba Province. These decisions ignited and intensified violent clashes opposing occupants to returnees, CNTB officers and police forces [58]. To prevent the conflict from escalating, the governor issued a directive to temporarily stop the implementation of the CNTB eviction notices in the province. In addition, judges and magistrates of the Makamba High Court blocked the CNTB order to convict the arrested occupants involved in violent clashes. These decisions against the CNTB operations received the support of several parliamentarians originating from Makamba and Rumonge Provinces, and ultimately President Nkurunziza [56,59].
In response, the CNTB chairman held a press conference on 9 March 2015 where he vividly criticised the interferences of the Makamba governor and the central government, stressing that the CNTB must continue its mission of bringing justice to returnees [60]. Unsurprisingly, this last public defiance led to his firing on 20 March 2015. In addition, a presidential decree was issued to nullify (at least until after the 2015 elections) all CNTB full restitution and eviction decisions across the country, with the purpose of 'avoiding any source of tensions and insecurity during the elections' [61]. A month later, a new CNTB chairman was appointed from within the CNDD-FDD party [62] 12 . Between 2015 and early 2022, four CNDD-FDD partisans have occupied this position.
Despite these changes in the CNTB leadership since 2015 and the leading role of the central government in local land governance, land restitution has remained a hot topic in Burundi. Occupants continue to be labelled 'illegal' settlers and are encouraged to voluntarily return contested land to their 'rightful' owners (now referring to the past returnees who did not flee in 2015) (Focus group discussions, Makamba, April 2019). Post-2015 hill, communal and provincial officials, most of whom are close to the ruling party, have aligned with CNTB directives, reducing the risk of emerging counter-narratives from local government officials about land restitution.

Discussion
This article examined the land dispute resolution and restitution processes and their ambiguous outcomes in war-torn Burundi, highlighting the political challenges of adjudication and enforcement following a long history of forced displacement. A history of armed conflict and forced migration enabled the state to build a new legal framework around land restitution and expand its control over customary land tenure. This study also highlights how the contested nature of land relations is historically and politically shaped. State changes in land policies and laws have played and still play an important role in the trajectories of land disputes. Furthermore, the lack of clarity and documentation of customary land tenure, the proliferation of stakeholders in dispute resolution and shifting land governance relations have deepened insecurity in land tenure. Specifically, the role of the CNTB has rekindled tensions and violence over land in Burundi.
In this section, we discuss these outcomes-drawing on theoretical insights from Lund [31,40], Lund and Boone [38] and Berry [39] to demonstrate how ruptures, complex tenure histories, different conceptions of tenure and other underlying legal, institutional and political processes that support land dispute resolution with the aim of restitution of land rights to people affected by armed conflicts and displacement have determined and continue to determine competing claims over land in the Burundian context. Congruent with Lund [31], earlier ruptures of civil wars and political transitions enabled the state to enter customary land tenure and redistribute land. The rupture of peace in the early 2000s enabled the state to present itself as the arbiter of land restitution and to consolidate its power over customary tenure by sidelining customary authorities. Moreover, different interpretations of the 'past' and 'space' emerged in this analysis [40]. In the context of the widespread contestations over land when refugees return due to the erosion of land boundaries, memory loss over many years of violent conflicts and dislocation and the undocumented nature of tenure histories, rival parties in land disputes and land governance actors are inclined to interpret the 'past' in different ways that serve their interests to retain or regain access to and control over land. While returnees upheld their claim to land through what Lund [40] refers to as an enduring traditional 'past' regarded as 'how things have always been done before forced migration', occupants relied upon a past shaped by significant historical events, actions and transactions. Occupants based their claim to land on the active role of previous central governments in fostering internal migration and land redistribution/occupation following the 1972 conflict, and their long history of using and valorising the land.
Stakeholders in dispute resolution interpreted 'space' as a territory governed by them. We argue that the various legitimation strategies employed by disputing parties, and the ongoing contestations of the terms of sharing agreements, were stimulated by the decline of customary land tenure systems and contested statutory control. Similarly, in the Colombian context, land restitution was confronted with several challenges, including the diversity of land claimants invoking complex and divergent tenure histories [63]. In Burundi, as in Colombia and other post-war settings, for some refugees and displaced persons, there can be greater insecurity related to returning to their so-called 'community of origin' than remaining in or going back to their places of refuge. Returnee men were more likely to voluntarily re-migrate in quest of better livelihoods, leaving behind their wives and children. Even youth were re-migrating to Tanzania, leaving behind their parents and younger siblings, as observed by a returnee family: Since our return, life has been difficult because we have no plot of land to cultivate. Also, we are not at peace because we are persecuted for the simple reason that we claim our family plots from the occupants. Moreover, our children accused us of having brought them here to Burundi to die. They asked us to make them return to Tanzania. My seven children have already returned to Tanzania. They are my four sons, my daughter-in-law and two grandchildren. We had 15 children; now we have 8 children left (Interview, a returnee family including two brothers and their wives, 28 October 2014).
Moreover, in respect to the 2015 new wave of forced displacement, humanitarian organisations have reported that a large number of Burundian refugees have yet to return, which implies that many refugees would consider a return to be a choice rather than a necessity [64]. In the current situation where the delay to file claims on pre-displacement land to the CNTB has passed, we anticipate that most Burundian refugees might not return to their places of origin.
The data demonstrate that land restitution policies alter local land governance. The proliferation of stakeholders and their changing relations were key elements shaping the outcomes of land restitution in the Burundian context. Jurisdiction over returnees' land disputes has been controlled by subsequent central governments through legal instruments on land and property and new (although contested) interpretations of legitimacy. Successive central governments have endowed authority in land dispute resolution and restitution to different institutions through what Lund and Boone [38] call a structure of jurisdictions that is hierarchical and divided into distinct functional and territorial jurisdictions. In the case study, mediation produced positive results for returnees and community members in some cases, due to the good reputation of local authorities and the endorsement of competing claims by the CNTB leadership as valid.
However, most disputes remained unsettled, as disputing parties were able to navigate institutions at different governance levels and in diverse jurisdiction realms to make their claims. Institutional pluralism and changing dynamics of power relations impeded the settlement of land disputes. This is illustrated by the disregard of local government officials and customary leaders in the early round of dispute resolution. Their gradual exclusion or replacement is particularly worrying as it shapes the competitive structure of jurisdictions about dispute resolution, as well as dynamics of multiple and fragmented loci of authority over land among government institutions. As observed by Kobusingye et al. in northern Uganda [65], complex dynamics of institutional competition raise serious concerns about the role of the state in promoting policies and legislation resulting in certain institutions being legitimated or contested and new social tensions and violence.
The poor coordination, the controversial role of dispute resolution stakeholders and their failure to accurately formalise the decisions and agreements resulting from a land dispute settlement in a context of prevailing uncertainty were other important challenges to the effective reduction in local land disputes. The state, through the introduction and dissemination of new forms of 'official land-sharing/restitution papers' and fostering, to a certain extent, further informal arrangements between disputing parties, increased ambiguity and uncertainty in land tenure. The role of the CNTB (and any future national land restitution commission) in the formalisation of agreements and decisions on contested lands without proper zoning and registering of customary lands as family-owned properties, and in a context where competing histories of land acquisition and ownership prevail, is deeply troubling. The Burundian land restitution legislation fails to acknowledge the importance of the existing cadastral, land titling and localised land rights registration government agencies in the formalisation of customary land rights resulting from land-sharing agreements. Hence, the CNTB transferred ill-defined land rights to 'new legitimate owners' without an in-depth historical contextualisation of land occupation dynamics and without offering any compensation to occupants. This also sets the stage for the new land attestation papers to be tools in future rounds of claim making and dispute resolution.
Over time, returnees' land dispute management and property restitution reproduced state power through the establishment of a new hierarchy of land governance officials, new roles and new laws. Land restitution policy reforms redefined the scope of the local and central authority. Customary elders and local elected officials legitimised and negotiated their local authority regarding their roles in land dispute resolution and the enforcement of rulings rendered by other institutions. The state's land restitution administration was further built around political allies whose authority was legitimised by the state's legislation and the president's orders. Returnees' land claims evolved as a key arena in which the central government operates and establishes its power and control over society, through legal instruments, statutory land institutions and even the use of force. As demonstrated in other war-torn settings in sub-Saharan Africa (e.g., Rwanda [66,67], South Sudan [68]), the Burundian state (central government) emerges as the key authoriser and ultimate arbiter in (customary) land governance.

Conclusions
The various land restitution commissions established by successive Burundian governments have failed to address the returnees' land question over time in a satisfactory manner.
The case of Makamba Province shows that although land governance reforms aiming to correct the land and property rights of returnees were undertaken in Burundi, they produced ambiguous outcomes. Despite a strong presence of the state in local communities, dispute resolution processes that aimed at land and property restitution have been contested either overtly or discreetly at the community level. Land restitution remains a critical, complex and highly politically sensitive and possibly destabilising issue in Burundi [13,64]. However, state control over local restitution dynamics has increased, because of subsequent policy and legal changes, and the massive return of refugees. The reinforcement of the state presence at the local level emerged largely because of opportunities provided by new land policies, shaping new jurisdictions over land dispute resolution, and a narrative of restorative justice. Moreover, state authority was exercised in the distribution of new roles, and the elimination or creation of governance authorities. The reordering of land governance authority took advantage of openings and barriers in the local and national political arenas to shift narratives of dispossession and impose a status quo in the land restitution process. This effectively strengthened the consolidation of state authority in local land governance, even though this is dependent upon wider social and political dynamics.
This analysis aligns with previous analyses that question the preference for specific remedies such as restitution over other alternatives, especially when displacement has been protracted, as changes in land tenure relations can result in major resistance and antipathy between community members who did not flee or returned earlier, secondary land occupants and returnees [6,9]. This article argues for the importance of an approach to comprehend and regulate post-war land dispute resolution and property restitution that explores, in more depth, the (re)construction and contestation of land governing authority. Such a perspective must take into consideration the legacies of violent conflicts, trajectories of customary land tenure and the ambiguous role of the statutory institutions in enhancing state control over territory and its people. In a continually changing and elusive socio-political context, it is important for policymakers to acknowledge the coexistence, relationships and struggles between different authorities positioned differently at various levels of governance. Rather than a focus on formulaic post-war remedies to protracted displacement and land conflicts, the emphasis should be on finding durable, fair and locally acceptable ways in which competing land claims may be settled and land rights secured.

Informed Consent Statement:
This article is based on the research programme 'Grounding Land Governance' (cf. section Methods) which was implemented by researchers and PhD affiliates of the African Studies Centre (ASC) in Leiden, from 2011 to 2017. The research was initiated when the ASC was operating as a multidisciplinary and independent research institute on Africa in the Netherlands, outside the jurisdiction of an institutional ethics review board. It is only in 2016 that the ASC became an interfaculty institute of Leiden University. However, this study was conducted in compliance to the ASC research guidelines. Prior to all interviews and focus group discussions, participants received complete information about the research; their questions were answered; and verbal agreement to participate in the study was obtained. Researchers have taken all necessary measures for protecting participants' identity, maintaining confidentiality, and storing data safely.

Data Availability Statement:
The data are not publicly available due to privacy restrictions. 1 The Twa (also identified as the Batwa) represents a deeply marginalised minority group that can be found in Burundi, Rwanda, the Democratic Republic of Congo and Uganda. They used to derive their livelihoods from hunting, gathering, and pottery. They are underemphasized in most literature on violent conflicts and land tenure in Burundi. Although the Burundian Constitution recognises and entrenches their political rights through the consociational representation model in government structures, this legal provision has less impact on Batwa lives and everyday experiences in practice. Their land tenure rights are weak and insecure; they have been shaped by long processes of dispossession as well as social and political exclusion, thereby resulting in extreme poverty and landlessness.  6 Both the CNTB and CSTB are national-level government agencies which members are appointed by the President of the Republic of Burundi. 7 The research in Burundi focused on post-conflict land governance in relation to dispute resolution, restitution and the formalisation of customary land rights. In addition, it took into account how land policy reforms were experienced at the community level and how dispute resolution and land restitution have been shaping public authority and legitimacy in everyday land governance. 8 The territory is officially divided into provinces, communes, zones, hills, sub-hills, and units of 'ten households'. Authorities at these different levels were involved in the sample. For simplication purpose, the term 'local authorities' will be used as a cluster word to identify authorities at the zone, hill, sub-hill and ten-household levels.