Hema v. Lendu in Ituri (DRC)

Local memories depict an unequal relationship between Hema pastoralists and Lendu agriculturists since the 17^th century, when Hema gradually dispossessed Lendu of part of their domain (Pottier, Reference Pottier2008: 431–432), though this inequality was tempered by a mutually beneficial exchange system. Like elsewhere on the continent, the importance of ethnicity during this pre-colonial period was complicated and often relegated due to other, often cross-cutting identities, based on clan, class, location, spirituality, and gender (Lentz Reference Lentz1995).

However, violent confrontations occurred in 1911, 1921, 1966, 1971, 1981 and 1992–1993, before exploding into extremely brutal warfare around 2000 (Vircoulon, Reference Vircoulon2005: 137–138). It is important not to project our current understandings of identity and conflict back into the past: notions of what constituted ‘Hema’ or ‘Lendu’ were different then, as were the power structures of local, customary rule. The root causes of local tension and conflict were most likely inequality in land acquisition and access, along with the favouring of Hema in education, politics and the economy by the colonial state. Violence was the result of the exploitation, by local and regional actors, of deeply rooted conflict over economic opportunity and political power (Vlassenroot and Raeymaekers, Reference Vlassenroot and Raeymaekers2004: 387).

Colonial policies contributed to the exacerbation of older antagonisms, particularly by privileging Hema access to education and employment opportunities in the local colonial administration, the mines and the plantations. They could also consolidate their economic dominance through cattle ranching and commerce. Like elsewhere in the Great Lakes region, early European observers projected the ‘Hamitic myth’ on Ituri society. It stated that ‘everything of value ever found in Africa was brought there by the Hamites, allegedly a branch of the Caucasian race’ (Sanders, Reference Sanders1969: 521). Most Belgian administrators and missionaries regarded Hema as a ‘god-sent superior race’, qualified to assist them in their ‘civilising mission’ (Pottier, Reference Pottier2008: 433). From early colonial times, a process of ethnic labelling provided the scientific and moral foundation for foreign domination (Pottier, Reference Pottier2009: 48). The partnership between Hema elites and colonial administrators and settlers contributed to the stability of Belgian rule and consolidated the Hema’s predominance in the educational, political and economic spheres (Vlassenroot and Raeymaekers, Reference Vlassenroot and Raeymaekers2004: 388–390).

As their elites had easy access to the inner circles of the Mobutu regime, Hema economic and political dominance was further reinforced after independence. In 1966, a Lendu insurrection was violently repressed, while in 1973, a new land tenure law allowed Hema to consolidate their economic power. When ‘democratisation’ came on the agenda in 1990, in Ituri as elsewhere in Zaire, ethnic capital became a major political resource. Local politicians began to exploit the old grassroots tensions, and ethnicity became the most powerful instrument for political mobilisation. This created a climate of tension which degenerated into open, and even armed, conflict.

Several other social and political transformations took place in Ituri since the mid-1990s, accentuating ethnic conflict. First, the dramatic increase in artisanal mining, beginning with the liberalisation of gold and diamond mining in 1982, transformed the region’s political economy (Fahey Reference Fahey2013). Tens of thousands of young men flocked to the tin, tantalum, tungsten, and above all gold mines of eastern DRC, feeding raw ore into supply chains that had recently opened up toward southeast Asia, in particular toward the industrial belt of coastal China (Raeymaekers Reference Raeymaekers2014). Control of these mining areas and trade routes became one of the main points of contention among political elites in the region; the large profits accentuated existing conflicts, often crystallising along ethnic fault lines.

Secondly, and concurrently, there was a militarisation of the conflict. This began in earnest with the regional alliance that toppled Mobutu in 1997. Ugandan troops occupied Ituri and began to be dragged into land conflicts between Hema landowners and Lendu or Ngiti farmers (Vlassenroot and Raeymaekers, Reference Vlassenroot and Raeymaekers2004: 390–391). Militias began to appear around 1999, recruiting fighters along ethnic linesFootnote ¹ (Tamm Reference Tamm2013: 14–15). Hema elites claimed they acted in self-defence, arguing that, given their minority position, the best way to protect themselves was to fully control the Ituri province. For their part, Lendu leaders claimed that the formation of ethnic militias was a reaction against their historical marginalisation and Hema dominance (Vlassenroot and Raeymaekers, Reference Vlassenroot and Raeymaekers2004: 411).

Major violence erupted in mid-1999, killing thousands of people and displacing over 150,000. These conflicts occurred against the background of intensifying regional competition. From late 1998 onwards, Ugandan involvement profoundly exacerbated an already fragile situation. Different Ugandan army commanders backed Hema and Lendu groups,Footnote ² maintaining the chaos necessary to remain present in Ituri and exploit the region’s riches, gold in particular.Footnote ³ Later on, Rwanda and the government in Kinshasa armed Hema and Lendu militias, respectively, in a game of shifting alliances, resulting in massive human rights abuse. (Vircoulon, Reference Vircoulon2005: 136). During the ‘Battle of Bunia’ in March-June 2003, Lendu and Hema exchanged control of the trade, engaging in ethnic cleansing (Vircoulon, Reference Vircoulon2005: 134). The ICG noted ‘the progression from land-based communal violence, to land-related operations of ethnic cleansing, to repeated acts of genocide by both Hema and Lendu’ (International Crisis Group, 2003: 6).

By 2005, proxy war between Uganda, the DRC, and Rwanda in Ituri had dwindled under donor pressure, opening up space for the UN peacekeeping mission to conduct robust operations, contributing to the disarmament of most of the militiamen. While the bloody ethnic violence in Ituri made international headlines, the calm that followed is just as instructive. For around a decade, between 2007 and 2017, there was a sharp de-escalation as militia politics and ethnic tensions became peripheral. Several reasons are behind this (Stearns Reference Stearns2023: 194–224): Most armed groups in Ituri were never given the opportunity to develop deep local roots, as almost immediately after their emergence, they came under the sway of Ugandan and then Rwandan armies. When these armies withdrew, these groups became unmoored and easier to demobilise. Armed mobilisation in Ituri was, despite its brutality, also shorter-lived than in the Kivus, where it has simmered continuously since 1993, making it easier to nip these path-dependent dynamics in the bud.

Finally, local constituencies also pushed back against insecurity in ways they could not in the Kivus. While Hema landowners were initially deeply involved in backing the Union des patriotes congolais (UPC), the armed group’s more abusive and less responsive actions eventually disenchanted them. This contrasted starkly to the response in the areas around Beni, Butembo, and Goma, where two decades of conflict incrementally produced a new class of politicians and entrepreneurs invested in conflict and where local civil society and business were overwhelmed by the larger regional and political forces that opposed them – while Ituri was not of critical interest to Rwanda, for example, the Kivus were. By 2005, when a coalition of armed groups attempted to revive the conflict, the mood had soured, especially among the wealthy class of businesspeople and landowners who depended heavily on stability for cattle herding, cash crops, and long-distance trade. The 2005 rebellion was quickly dismantled.

Local and international peacebuilding initiatives also contributed to the relative calm that ensued. The international community wanted to make an example of this region, which was not only extremely violent but also geopolitically marginal – two factors that made it a prime candidate for interventions by the International Criminal Court (ICC), the European Union, and the United Nations. Here, outsiders could invest considerable resources without offending regional powers while claiming to have contributed significantly to a decrease in violence. The ICC indicted four local militia leaders, while the European Union deployed a peace enforcement mission, Operation Artemis, in 2003, to stabilise Bunia, the capital of Ituri (Stearns Reference Stearns2023).

Donors complemented these military and diplomatic efforts with humanitarian and development aid. A regional pilot scheme for the national demobilisation and reintegration programme was deployed in September 2004, the Disarmament and Community Reintegration (DCR) programme. Over 15,000 combatants, including 4,525 children, were demobilised (Amnesty International 2007). At the same time, donors funded a set of local peacebuilding projects intended to boost employment, reconciliation, and the local economy (Pax Christi and Haki na Amani 2009). According to several independent evaluations, these initiatives contributed to an improvement in relations between the Hema and Lendu communities (Gaynor Reference Gaynor, Biekart, Kontinen and Millstein2023, Samset and Madore Reference Samset and Madore2006)

Unfortunately, the government did not take advantage of this lull to address the structural drivers of violence in Ituri, in particular land conflicts (Vircoulon Reference Vircoulon2021; Crisis Group 2020). After a long period of relative calm, tensions between Hema and Lendu flared up in 2017, when they started attacking each other again. The violence was at first limited, but escalated when the Coopérative pour le développement du Congo (CODECO), a Lendu agricultural organisation turned militia, repeatedly attacked displaced people’s camps that were mostly hosting Hema. Despite attempts at pacification, including by the Forces armées de la République démocratique du Congo (FARDC) and Mission de l’Organisation des Nations Unies pour la stabilisation en République démocratique du Congo (MONUSCO), violence then escalated again (United Nations Group of Experts 2023) and Human Rights Watch (2023).

Banyarwanda/Banyamulenge v. ‘Autochthons’ in Kivu (DRC)

A telling controversy flared up in September 2020, triggered by the installation ceremony of the commune (municipality) of Minembwe (South Kivu). Local and national politicians decried the creation of a new administrative entity as an attempt by the Banyamulenge to usurp the authority of other ethnic groups and dispossess them of their ancestral grounds. They pointed to the leadership of the new commune, most of whom were Banyamulenge, and lamented that this entity would encroach on the power of customary chiefs from other communities (Mathys and Verweijen, Reference Mathys and Verweijen2020). That the national minister of decentralisation was a Munyamulenge himself added heft to the alleged conspiracy by Rwandophones or Banyarwanda to ‘balkanize’ the DRC and to put in place a ‘Tutsi-Hima empire’, an idea dating back to the 1960s, and again implicitly referring to the ‘Hamitic Hypothesis’. The idea was informed by discourses of ‘autochthony’ or ‘indigeneity’, opposing ‘first’ or ‘original’ inhabitants to ‘newcomers’ or ‘foreigners’ (Mathys and Verweijen, Reference Mathys and Verweijen2020). A large mobilisation with strong anti-Tutsi overtones forced President Tshisekedi to postpone setting up the Minembwe commune.

This incident was just one of the most recent in a long series revolving around the ‘Banyarwanda problem’ related to the status of Kinyarwanda speakers living in the Kivu region. The Banyarwanda of North Kivu are made up of many groups: the ‘natives’ who lived in the Kivus in pre-colonial days, the ‘immigrants’ and the ‘transplanted’ of the colonial period, the ‘infiltrators’ and the ‘clandestines’ before and after independence, and the Rwandan TutsiFootnote ⁴ and HutuFootnote ⁵ refugees (Jackson, Reference Jackson2007; Mathys Reference Mathys2014). The first violent conflict along these identitarian lines occurred in the 1960s during the so-called Kanyarwanda rebellion, when the Banyarwanda faced the threat of expulsion from North Kivu. Mobutu’s government then took steps to root out ethnic conflict on the periphery of the state – all the while he cultivated ethnic-based patronage networks at the centre – in an effort to consolidate power. Governors and administrators, for example, were not supposed to come from local communities, nor did they stay in place for long. This approach changed as Mobutu’s regime frayed, his resources dwindled, and he began to fan the flames of ethnic divisionism in an effort to stay in power. The ‘Banyarwanda’ thus came to the fore again during the National Sovereign Conference (1991–2), when representatives of the Kivu civil society raised the question of the ‘Zaireans of doubtful citizenship’, a coded expression referring to the Banyarwanda, who were largely excluded from the proceedings (Willame, Reference Willame1997, Ndaywel e Nziem Reference Ndaywel e Nziem1998).

These exclusionary dynamics during the democratisation period then triggered large-scale violence in North Kivu in March 1993. The conflict demonstrated the fluidity of ethnic categories and illustrated the phenomenon of ethnogenesis. Those who initially were the targets of attacks by ‘indigenous’ ethnic groups, such as the Hunde, Tembo, Nande and Nyanga, were the Banyarwanda, Hutu and Tutsi alike, seen as a single ethnic group. Only two years later, Congolese Hutu and Tutsi confronted each other in ‘ethnic’ strife. This showed the unstable nature of identities articulated in different and sometimes contradictory ways depending on changing contexts (Alida, Reference Alida2017: 43). While the 1993 conflict came to an end because of successful local mediation (Willame, Reference Willame1997: 124–131), in 1994 the massive influx of Hutu refugees from Rwanda completely transformed the politics of the region. Together the Banyarwanda and the Rwandan refugees suddenly constituted the majority of the population of the ‘petit Nord,’ the lower part of North Kivu province around Goma. The alliance of Hutu and Tutsi Banyarwanda broke up and, as in Rwanda, the two groups entered into violent conflict.

Similar dynamics led to conflict in South Kivu province, as well. The Banyarwanda there started calling themselves Banyamulenge in the late 1950s-early 1960s to differentiate themselves from the Tutsi refugees who started leaving Rwanda from 1959 onwards. Most historians date the arrival of the first members of this community to the second half of the nineteenth century, long before the waves of refugees who arrived around or after independence in 1960 (Depelchin Reference Depelchin1974; Weis Reference Weis1959). Ethnic tensions first boiled over during the Mulelist uprisings of the 1960s, when Banyamulenge sided with the national government against insurgencies that recruited largely from the Bembe community (Stearns Reference Stearns2011). Much like in North Kivu, ethnic conflict subsided until Mobutu’s grasp on power began to fray.

By calling themselves after Mulenge, a village in the highlands overlooking the Ruzizi Plain, the Banyamulenge sought to highlight their local roots (Stearns et al., Reference Stearns2013). As most of them immigrated to South Kivu from Rwanda before 1885, they possessed Zairean citizenship, even if the most restrictive legislation on nationality was applied. Nevertheless, in 1995, the political rights of ‘populations of doubtful citizenship’ were called into question in South Kivu, as had been the case in North Kivu. Tensions grew dramatically in mid-1996. The Banyamulenge were the victims of discrimination, violence and expulsion from their land engineered by local authorities and ‘autochthonous’ groups. The Banyamulenge armed themselves and allied with the new Rwandan government that was preparing the ground for a military operation that was to become the ‘rebellion of the Banyamulenge’ against the Mobutu regime (Reyntjens, Reference Reyntjens2009: 45–58).

When the Rwandan army attacked Zaire in September 1996 to dismantle the Hutu refugee camps, which were close to the border and constituted a serious security risk for Kigali, Rwanda used the Banyamulenge’s genuine grievances to provoke and support this rebellion, behind which it could hide its military action and thus avoid a claim that it violated international law by attacking its neighbour. While, on the one hand, Rwanda offered protection to the threatened BanyamulengeFootnote ⁶, on the other hand this close association confirmed in the minds of other ethnic groups that the Banyamulenge were foreigners or, at least, had treasonous loyalties. Within the nativist and ‘liberationist’ Zairean political discourse, they were an enemy within, a sort of fifth column (Ntanyoma and Hintjens, Reference Ntanyoma and Hintjens2022: 13). Although many Banyamulenge leaders later distanced themselves from Rwanda, this episode strengthened the bipolarity that made them vulnerable as a minority. Recently, it has even been claimed that the Banyamulenge are victims of ‘slow genocide through attrition’ (Idem: 14).

The Great Congo Wars of 1996–2003 and the subsequent low-grade violence have built on and cemented these ethnic tensions but also revealed their flexible nature. Throughout the Second Congo War (1998–2003), the Rwandan government backed the Congolese Rally for Democracy (RCD), which had many prominent leaders from Rwandophone communities. At the same time, however, the RCD had to contend with large rebellions from within the Hutu community of North Kivu, as well as from within the Banyamulenge community. The simple fact of being from a particular community, in other words, could shape but not definitively determine a political stance. Even during the subsequent National Congress for the Defence of the People (CNDP) (2006–2009) and M23 (2012–2013 and 2021-present) rebellions, which were led by members of the Congolese Tutsi communities and backed by Rwanda, significant strains emerged within the groups and between those groups and the Rwandan government (Stearns Reference Stearns2012; Stearns Reference Stearns2023). Armed groups led by Tutsi and Banyamulenge at times engaged in coalitions with groups from other communities, showcasing the importance of ethnic entrepreneurs (Lemarchand Reference Lemarchand2009). For example, the Banyamulenge dissident commander Patrick Masunzu joined forces with both Mai-Mai groups and Rwandan FDLR Hutu rebels, groups that had mobilised against the Banyamulenge previously.

The relative lull in violence in parts of the country during the Congolese transition between 2003 and 2006 is also instructive – although the attack on Bukavu by Nkunda and Mutebutsi was an exception to this. By 2006, around 130,000 combatants had been demobilised, and the number of internally displaced in the Congo had declined to 1.2 million, a third of what it had been just three years earlier (Stearns Reference Stearns2023). While we are not aware of any systematic study of ethnic relations undertaken during this period, it stands to reason that the decline in violence went hand-in-hand with an improvement in communal relations. Congolese Hutu and Tutsi officers were given senior command positions in the national army, and a Munyamulenge was elected to the senate.Footnote ⁷

Hima v. Iru in AnkoleFootnote ⁸ (UGANDA)

While the demographic structure in Ankole is similarly bipolar to that of the cases discussed earlier, the two groups have a different conflict history, and colonial interventions have exacerbated the antagonism. Ankole offers an a contrario illustration of our argument. Various trends – constituting together a ‘de-bipolarisation’ and ‘delocalization’ of the region – have significantly diminished conflict potential.

The myth of origin of Ankole’s two main ethnic groups is telling. Ruhanga, the kingdom’s creator, put his three sons, Kakama, Kahima and Kairu, to a competitive test that involved safekeeping a filled milk pot for a whole night. Kakama won and was charged with ruling the country. Kahima, who had given some milk to Kakama, was to look after the cattle, while Kairu, who had spilled his milk, was told to till the soil.Footnote ⁹ These were the ancestral eponyms of the kings (bakama), the Hima and the Iru. The premise of inequality between Hima and Iru is clear from the myth. While this is a stylised version of social divisions during pre-colonial days, it seems that ethnicity was an important marker during this period and that these groups engaged in different economic activities: cattle breeding for the Hima and agriculture for the Iru. In addition to these different occupations, social divisions were based on status and ethnicity (Doornbos, Reference Doornbos1978: 27). The two groups interacted mainly among themselves, and there were no extensive exchange relationships between them. Links were characterised by segregation and avoidance, rather than by symbiosis or serfdom (Karamura, Reference Karamura1998: 13; Elam, Reference Elam1974: 161–162). Occupational inequality showed in that Hima were convinced that Iru could not be allowed to own, keep or tend even one productive cow (Elam, Reference Elam1974: 164). Doornbos (Reference Doornbos1978: 44) notes ‘a basic and continued separateness and social distance between Bairu and Bahima’ during this period, and Hima dominated the political system; only in very rare circumstances could an Iru become a chief (Karamura, Reference Karamura1998: 14).

British colonial rule reinforced this Hima dominance, drawing on the Hamitic hypothesis discussed earlier. In Ankole, the Hamites were Hima, said to have come from Somalia, Ethiopia or Egypt and therefore believed to be fit to rule. Colonial rule thus reinforced and perpetuated Hima oligarchy. A pattern of largely distinct communities became one in which a chiefly and administrative Hima elite functioned as a local ruling stratum. As Hima elites ruled the Ankole district, which was the extended colonial successor to the Nkore kingdom, ethnic inequality grew and antagonism increased (Doornbos, Reference Doornbos1978: 176). However, the conflict was symbolic rather than physically violent. It expressed itself in the lack of intermarriage and negative stereotyping, especially by Hima towards Iru, presented as uncivilised and ‘not knowing how to live in a community’.Footnote ¹⁰ In response to the dominance of Hima within local power structures, the Iru organised the Kumanyana – ’to get to know each other’ – Movement in the 1940s. The movement contributed to some extent to Bahiru emancipation (Refugee Law Project 2014).

Colonial rule introduced a second, religious cleavage in Ankole, which softened the ethnic divide. Missionary penetration split the population into roughly 50 per cent Protestants and 50 per cent Catholics. Virtually all Hima were Protestant, while Iru were roughly evenly Catholic and Protestant. In a Hima and British Protestant-dominated framework, Protestant Iru stood a better chance of gaining access to political rewards such as jobs and scholarships than their Catholic co-ethnics (Bøås, Reference Bøås2002). As seen elsewhere across the world, such cross-cutting cleavages alleviated their salience (Grove, Reference Grove1977). Over time, ethnic inequality was gradually eclipsed by socio-economic inequalities in which Iru elites joined the ranks, which earlier had been more exclusively occupied by Hima. As ethnic conflict slowly declined, new social and political alignments grew between Hima and Iru elites (Doornbos, Reference Doornbos1978: 176).

After independence, political and economic power became centralised in Kampala. In 1967, President Milton Obote got rid of chieftaincies and kingdoms in Uganda, decreasing competition between Hima and Iru. Before Museveni, himself a Hima Munyankole, came to power, the Banyankole, Hima and Iru alike were marginal in national power structures. This changed dramatically with the victory of the National Resistance Movement (NRM) in 1986. Hima Banyankole became the main beneficiaries of Museveni’s NRM favouritism (Lindemann, Reference Lindemann2010: 32). Despite their small demographic size – the Banyankole constitute around 10% of the country’s population, and the Hima are the minority within that group – Museveni’s co-ethnics came to dominate, for example at the top level of the army (Idem: 36). At the political level, positions of real power and influence were controlled by the Banyankole and others from western Uganda (Idem: 22). The embeddedness of Ankole in the larger Ugandan polity contributed to this loss of salience of local ethnic cleavages, increasingly superceded by tensions between Banyankole and other Ugandans.

Writing in 1978, Doornbos (Reference Doornbos1978: 176–177) finds that both the rise and decline of ethnic hostility in the case of Ankole thus appear explicable by changing situational variables, specifically of differential access to political and economic benefits for the two population groups. A factor mitigating ethnic tension was Ankole’s status as a sub-system of Uganda: the national arena offered alternative avenues for upward social mobility, slowing down local competition and thus helping to attenuate ethnic tension. Another explanation lies in the emergence of sub-cleavages, both religious and party political (Doornbos, Reference Doornbos and Lemarchand1977: 250–251). This logic has also persisted in NRM’s Uganda. Hima-Iru tensions have decreased as the local arena became embedded in a larger political environment and as cross-cutting cleavages gained purchase.Footnote ¹¹

Similarities and differences

Bipolarity, yes, but from where?: The case studies described here confirm the usefulness of ethnic polarisation as a lens through which to understand conflict. All these conflicts can be cogently analysed as the kind of bipolar conflicts described by Horowitz, albeit at the subnational level. However, our definition of polarisation does not match that employed by econometric analyses, which measure the relative size of the two groups (Montalvo and Reynal-Querol Reference Montalvo and Reynal-Querol2005). In contrast to this size-based polarisation, we describe how groups have come to perceive each other as inimical over time due to colonial ideology, economic practices, and political manipulation.

One of the main such factors in these case studies was colonialism. In Ankole and Ituri, while ethnic inequality existed in pre-colonial days, the application of the Hamitic hypothesis consolidated the superior status of Hima and Hema, respectively, which led to greater inequality and ethnic conflict. Although colonisers drew on pre-colonial social identities and histories, ethnicity was deeply shaped by colonial administrators, priests, and scholars, sometimes radically so, and rendered more rigid. In North Kivu, colonial policies regarding immigration were extremely consequential: between the 1930s and the 1950s, the Belgian colonial government orchestrated a transfer of populations from Rwanda on the order of 200,000 or 300,000 people (Bucyalimwe Reference Bucyalimwe1990; Pabanel Reference Pabanel1991). This population became the demographic majority in places like Masisi territory, changing local and provincial politics considerably. Finally, in South Kivu, the colonial impact on the ethnic tensions discussed here was perhaps most notable due to what it did not do: since the Banyamulenge community was pastoralist and relatively mobile, Belgian administrators did not create an administrative entity for them, as they did for most other communities in 1933 – the anecdote above about the creation of the commune of Minembwe shows the lasting importance of this decision today.Footnote ¹²

Our analysis also confirms the importance of access to state power, as suggested by the EPR analyses. However, the state plays an ambiguous role. In the case of the Banyamulenge community of the DRC, for example, both they and their neighbouring communities were long excluded from national and provincial positions of state power – it was rather access to land at the local level that initially fomented conflict, before it was further exacerbated by armed mobilisation (Verweijen and Vlassenroot Reference Verweijen and Vlassenroot2015). While state actors played a role in these dynamics, they did so in complicated ways. At times, it was political elites unrelated to the ethnic groups in question that manipulated tensions to whip up popular support or to distract from other problems. At other times, especially during the first decade of Mobutu’s reign, the state intervened to keep a lid on ethnic competition and conflict, perceiving this as a challenge to its authority.

These different ways that colonial and post-colonial governance impacted ethnic relations highlight a key limitation of the quantitative datasets described at the outset: they highlight useful correlations between polarisation and conflict, but say little about how these tensions came about in the first place – why these two groups and not others? What were the mechanisms? And how does ethnicity interact with other power dynamics in society – land governance, the distribution of wealth, and gendered aspects of power?

How conflicts can de-escalate

Studies of ethnic conflict can sometimes make the mistake of selecting on the dependent variable. Since they are mostly interested in understanding the causes of ethnic conflict, they spend comparatively less time understanding why conflicts fizzle out or do not appear in the first place (Tang, Reference Tang2011). And yet, as recent studies suggest, ethnic conflict seems to be on the decline (Cederman et al., Reference Cederman, Gleditsch and Wucherpfennig2017). Both the individual case studies and the comparative analysis leave us with a greater sense of possibility and optimism than many studies of ethnic conflict. Images of deep ethnic divisions obfuscate less publicised and visible periods of de-escalation and cohabitation between groups.

The malleability of ethnicity: The importance of this kind of nuanced analysis can also be seen through our second point: the malleability of ethnic antagonisms and the importance of political elites in stoking these resentments. As much as these conflicts draw on age-old myths and stereotypes, these divides and their intensity can shift abruptly. There are many examples of this. In 2005, the former Lendu and Hema armed groups in Ituri, under pressure from robust UN peacekeeping operations, joined forces in the Revolutionary Movement of the Congo (MRC) coalition. In 2001 and then again in 2004, the dissident Banyamulenge commander Patrick Masunzu collaborated with Rwandan Hutu rebels from the Democratic Forces for the Liberation of Rwanda (FDLR), even though members of that group were allegedly involved in the 1994 genocide against Tutsi (Rafti Reference Rafti2006: 13). In 2012, the veteran North Kivu rebel leader Kakule Sikuli Lafontaine, whose main justification for rebellion had been resistance against Tutsi and Rwandan invaders, abruptly abandoned his FDLR allies and decided to join forces with the M23 rebellion, which was led by Tutsi and backed by Rwanda.

This malleability appears most clearly among armed groups, as the above examples demonstrate. This underscores the importance of distinguishing between various actors. While the term ‘ethnic conflict’ suggests entire communities opposed to each other, it is very small percentages of these communities who are engaged in actual violence, often with large degrees of discretion. These ethnic entrepreneurs vary between the cases studied here. In North Kivu, for example, the first wave of violence against Banyarwanda in 1993 was spearheaded by local and national political leaders. After an incendiary speech by the provincial governor, who denied Zairean citizenship of the Banyarwanda and threatened them with extermination, around 15,000 people were killed and over 300,000 displaced in 1993 (Amnesty International, 1993). At around the same time, conflict was brewing in South Kivu, and again, state authorities stoked the fire. The Uvira commissioner, the deputy provincial governor and the deputy speaker of the transitional parliament (Haut Conseil de la République – Parlement de transition) openly contested the Banyamulenge’s Zairean citizenship (Stearns Reference Stearns2011). Retaliatory massacres were then carried out not by the Banyamulenge in general, but by organised rebellions in which the Banyamulenge participated – in particular, the Alliance of Democratic Forces for the Liberation of Congo-Zaire (AFDL) and the RCD, which killed hundreds of civilians in places like Kasika, Makobola, and Walungu (United Nations 2010).

These two points – the historical reach and the malleability of ethnic identity – represent a dialogue. Political and military elites are important, but also operate within deeper, structural constraints. As Hoffmann et al. argue (Reference Horowitz2020), ethnicity is neither merely the result of material interests and instrumental manipulation, nor is it ‘infinitely elastic,’ as that would neglect the very real material stakes that are tied to ethnicity. Citizenship and access to land and power are linked to ethnicity through laws and social ties (Vlassenroot Reference Vlassenroot2013), and even if identities are constructed, they can have deep roots. In the Congo, a series of polls in 2003, 2016, and 2023 have shown that acceptance of Congolese Hutu and Tutsi communities to be remarkably and consistently feeble: only about a quarter of respondents thought that Hutu and Tutsi could be Congolese (Weiss and Carayannis Reference Weiss and Carayannis2004; BERCI and Congo Research Group 2016; BERCI, Congo Research Group and Ebuteli 2023).

Pathways of de-escalation: The case studies provided here suggest several pathways toward de-escalation. We call these top-down peacebuilding, bottom-up peacebuilding, and changing administrative frameworks.

Ituri has showcased the importance of both elite and grassroots efforts to de-escalation. Beginning in 2002, the international community adopted perhaps its most forceful approach in Ituri, deploying a military force, pressuring neighbouring countries to withdraw troops, and funding large-scale demobilisation exercises. This top-down approach benefitted from the lack of deep roots that this conflict had in Ituri – the militias were largely dependent on governments in Rwanda, DRC, and Uganda; once their leaders were arrested and those countries scaled down their involvement, space was opened up for NGOs and community organisations to carry out bottom-up peacebuilding (Vircoulon Reference Vircoulon2021; International Alert 2012). One outside evaluation of these efforts pointed to significant progress in communal relations during the 2005–2017 period. Ituri, by far the most violent part of the DRC in the 2002–2005 period, became relatively peaceful.

The subsequent escalation, however, is also instructive. Triggered by the killing of a Lendu priest in June 2017, new militias were formed among both communities, but in contrast to past periods of escalation, customary chiefs and community leaders denounced the violence, saying it had been instigated by unnamed outsiders (Crisis Group 2020). Nonetheless, even local peacebuilding organisations argue that the main flaw of these initiatives was their inability to address the ‘structural determinants’ of conflicts, in particular the unequal distribution of land and the lack of interest of elites in Bunia and Kinshasa to address this and governance issues (Gaynor Reference Gaynor, Biekart, Kontinen and Millstein2023; International Alert 2012).

This account of violence in Ituri demonstrates the nuanced pathways of escalation. The demography of the region – the main variable examined in studies of ethnic polarisation and fragmentation – is a poor predictor of the rise and fall in ethnic conflict. Even polarisation and socio-economic marginalisation – the predictors used by Cunningham and Weidmann (Reference Cunningham and Weidmann2010), Bleaney and Dimico (Reference Bleaney and Dimico2017), and Wimmer (Reference Wimmer2017) – cannot explain these trends, because they involved elites from outside Ituri, but also because they involved elites manipulating ethnic identity to further their interests. Herein lies perhaps the greatest blind spot of quantitative or formal theoretical studies of ethnic conflict: if the activation of ethnic cleavages is exogenous to social relations, if ethnic feuds can be ignited or quelled through outside manipulation, then these approaches can at best provide contextual analysis. This is not to reduce ethnic conflict to manipulation – as Varshney (Reference Varshney2003) has argued, it will always be a blend of value rationality and instrumental rationality – but to highlight the importance of careful process tracing in understanding the pathways of mobilisation.

The case of Ankole, where the manipulation of ethnicity did not result in violent confrontation, provides another such pathway. Here, the cross-cutting cleavages (Hima v. Iru; Protestant v. Catholic; socio-economic distinctions) seem to have been critical in preventing violent ethnic conflict. This confirms the large-N analyses provided by other scholars that show that ‘cross-cuttingness’ – the degree to which members of the different ethnicities have ties in terms of class, religion or other identities – reduces the likelihood of ethnic civil war (Gubler and Selway Reference Gubler and Selway2012; Siroky and Hechter Reference Siroky and Hechter2016). Equally important, the incorporation in the broader Ugandan system made Hima and Iru aware of their common interests as Banyankole. Even though most of the Banyankole who rose to power with the NRM in 1986 were from the Hima minority, this does not appear to have aggravated conflict with the Iru – in part because there were not ethnic entrepreneurs who sought to accentuate these divides, but also because Iru elites saw their interests aligned with the NRM in a country where power has firmly been in the hands of one political party.

Similar dynamics have played out elsewhere, suggesting that political salience of cleavages depends on their importance in the broader political context. Posner (Reference Posner2004b) has shown how relations between the Chewa and Tumbuka communities in Malawi are antagonistic, since those communities make up a sizeable share of the overall population and are therefore politically important; meanwhile, in neighbouring Zambia, where these groups are marginal, they live in relative harmony. Similarly, in the DRC, when the current day provinces of South Kivu, North Kivu, and Maniema formed one large Kivu province before 1962, the main ethnic competition during the political tumult around independence was between the Kusu and Shi communities, two of the largest ethnic groups (Verhaegen Reference Verhaegen1966). When the province was split into three in 1962, those tensions abruptly disappeared and were replaced by competition between the Rega and Shi in South Kivu and between the Nande and Banyarwanda in North Kivu.

From the examples above, we can see that cross-cutting identities can fade to the background or be highlighted, based on the context and the rhetoric employed by ethnic entrepreneurs. When debates over citizenship began in the 1990s in Zaire, the ‘foreignness’ of Congolese Hutu and Tutsi was highlighted, leading them to bond together; when Rwandan Hutu refugees and militia spilled across the border following the genocide in 1994, this coalition was split apart as their ethnic identity came to the forefront. Similarly, the Banyamulenge community of South Kivu became briefly more associated with the Congolese nation than with local ethnic rivalries during the rebellion of Patrick Masunzu in 2002–2003, given the bloody standoff with Rwanda that forced many to reassess their perception of the Banyamulenge (Stearns et al. Reference Stearns2013). Unfortunately, in a similar way as in Ituri, the government in Kinshasa did not build on this goodwill to address the deeper communal grievances – reconciliation for past violence, disputes over local administrative structures, and conflicts between pastoralists and farmers (Vlassenroot Reference Vlassenroot2002).

This discussion adds to recent debates around the dynamics of restraint in mass violence (Straus Reference Straus2015, Bergholz Reference Bergholz2019). One such factor, which has been raised by Straus (Reference Straus2012), is the economic costs produced by violence, which can lead elites to step back from the brink. There is evidence for this in Ituri – once regional actors withdrew from the conflict arena, local elites quickly tried to forge the stability necessary for trade and cattle farming. Other factors that Straus highlighted that we find in the examples here are the importance of strong institutions at the national level to mitigate or arbitrate regional tensions in Uganda, and the involvement of external peacebuilders – in Ituri, this was the European Union, the ICC, and the United Nations, in particular.

But our reading of these dynamics adds other factors, as well, that are important in bringing an end to violence. Straus proposes a variety of mechanisms at the local, meso, and macro levels that can lead to restraint. Our account suggests that it is in the feedback loops between grassroots and elite actors that sustainable de-escalation can take hold. This is in contrast with arguments by other scholars on the region, such as Autesserre (Reference Autesserre2010), who have placed an emphasis on local actors. However, as Gaynor (Reference Gaynor, Biekart, Kontinen and Millstein2023: 267) has argued in the case of Ituri, ‘the biggest limitation, and one which has reportedly led to the return of widespread violence and displacement since 2017, is the inability of local groups – by virtue of their very localism and powerlessness in the face of broader structural constraints – to address these same constraints’. What is needed is a theoretical understanding of how elite manipulation interacts with grassroot grievances to lead to escalation. In the case of the eastern DRC, the space opened up by internationally-led peace processes has not led to the structural reforms necessary: land tenure legislation, national debates over citizenship, education and accountability for past violence, and sanctioning of spoilers have all been lacking (International Alert 2012).

The importance of subnational analysis: This point leads us to the importance of scale: understanding the linkages between local, national, and regional levels of escalation. The cases studied here show that, contrary to a widespread bias that favours national approaches, the subnational level must not be overlooked or underestimated. This level remains under the radar in the large-N studies that predominate in the literature, as pointed out in the introduction. Therefore, ‘thick’ detailed case studies are necessary to generate a more complete, complex and reliable understanding of fragmentation levels and the risks of conflict they can entail.

Case studies give us a different kind of purchase on causality than large dataset analysis: they are more attentive to causal mechanisms and temporal processes, such as critical junctures. Crucially for ethnic conflict, they are also able to capture the formation and the nature of identities and interests. In the case of the Kivus, it shows the importance of the colonial legacy in terms of hardening ethnicity, tying it to citizenship and land, and generating a discourse of autochthony that has become one of the main cleavages of Congolese politics, at least in the east of the country. And yet, paradoxically, it has also shown the malleability of ethnic identity by highlighting the role that ethnic entrepreneurs play, in particular, leaders of armed groups.