RE-DESCRIBING AFRICA'S WARS

Transnational conflict can simply be defined as an armed conflict that extends or operates across national boundaries. Several datasets exist that can be used to examine transnational conflict in Africa. The Correlates of War (COW) project collects data on interstate wars (Small & Singer Reference Small and Singer1982; Sarkees & Wayman Reference Sarkees and Wayman2010). The Uppsala Conflict Data Project (UCDP) also collects data on interstate armed conflicts, but uses a lower threshold of conflict intensity (defined in terms of battlefield deaths) than the COW project (Pettersson & Wallensteen Reference Pettersson and Wallensteen2015: 536–50). The Militarized Interstate Dispute (MID) dataset compiled by the COW project focuses on low-intensity military confrontations between states (Jones et al. Reference Jones, Bremer and Singer1996: 163–213 Maoz Reference Maoz2005; Palmer et al. Reference Palmer, D'Orazio, Kenwick and Lane2015). Two major datasets have looked at interstate rivalries (Klein et al. Reference Klein, Goertz and Diehl2006: 331–48; Dreyer & Thompson Reference Dreyer and Thompson2012). The International Crisis Behavior (ICB) project looks at interstate crises (Brecher & Wilkenfeld Reference Brecher and Wilkenfeld1997). The UCDP external support dataset focuses on external support to conflict parties in the form of troops, funding, logistics, military equipment, intelligence and safe havens (Pettersson Reference Pettersson, Pettersson and Themnér2011: 41–60). More recently, San-Akca has also published data on state support to non-state armed actors (San-Akca Reference San-Akca2016).

These datasets suffer from several shortcomings. First, each of these different datasets catches only a specific type of transnational conflict and their data are not integrated to provide a comprehensive picture. Second, there are difficulties with missing data points. Third, there are definitional challenges.

To remedy these shortcomings, the method we adopt combines an effort to be comprehensive with ‘re-description’, an approach understood as ‘a problem-driven enterprise’ that develops and applies ‘close-to-the-ground’ descriptions and explanations of reality (Shapiro Reference Shapiro2005: 181; Twagiramungu Reference Twagiramungu2014: 10–30). In common with historians’ methodology, the TCA method values empirical accuracy in reporting incidents even though that may entail data gathering methods that are less consistent and generalisable.

The TCA dataset is built by combining, augmenting and revising several existing datasets each of which capture some elements of transnational conflict, including interstate wars, external state support in interstate wars, low-intensity confrontations between states, external interventions in civil wars, and external support to rebels or coup-makers. The TCA not only integrates existing data on armed conflict in Africa, but also includes new data for cases that fall beyond the temporal scope of existing datasets or that are overlooked in the existing datasets. This is based on country expert consultations, leading to identification of published findings in academic books, reports, memoirs and other non-contemporaneous sources. The method is an extensive scouring of published sources, followed by coding the incidents described as different kinds of action, by a state or a non-state actor, to engage in or support a domestic actor engaged in an armed conflict in the territory of another state.

A particular focus of the TCA method is finding and utilising sources that deal with covert or disguised military activities. Consequently we use more diverse sources than existing datasets. This is in contrast to, for example, the UCDP which uses contemporaneous media sources or NGO reports as the basis for its coding and does not revise datapoints when subsequent scholarly publications or memoirs are published or protagonists are interviewed. Those subsequent publications or interviews may indicate that the details in the public realm at the time were incomplete or misleading. There are obvious reasons why a state would not want the true account of its cross-border military activities to be known at the time, and why it may take some time for a truer account to become known. This poses a dilemma for those who compile and maintain datasets. Difficulties arise if a dataset remains open to revision at any subsequent time, when new information becomes available. Statisticians value consistency of method in data gathering and decry post-hoc tinkering with datasets as individual pieces of new data become available, and so they tend to maintain consistency at the expense of empirical accuracy. The TCA does the reverse.

An example from our catalogue of incidents illustrates several of the methodological and definitional dilemmas. The case in point involves the Sudanese Armed Forces, the Ethiopian Armed Forces, the Sudan People's Liberation Army (SPLA), the Eritrean People's Liberation Front (EPLF) and the Oromo Liberation Front (OLF) during 1989–1990. De Waal describes how in November 1989, the SPLA and Ethiopian troops crossed the border at Kurmuk, over-running a Sudanese army garrison, and advanced along the banks of the Blue Nile and were poised to take the town of Damazin, and the nearby Blue Nile dam that generated Khartoum's electricity supply. The Sudanese army was in full retreat. It was saved only by a secret commando action by the EPLF, which defeated the SPLA and the Ethiopian army on both sides of the border in January 1990. The EPLF assault was accompanied by a small contingent of OLF fighters who were inserted into western Ethiopia to launch guerrilla operations there, though the main OLF combat over the following months was against SPLA troops who were serving as a proxy for the Ethiopian government in its domestic counter-insurgency (de Waal Reference de Waal and de Waal2004: 187–9; Reference de Waal2015: 46–9).

This incident illustrates the following points. First, an incident involving five armed actors is complicated and very hard to classify. This is especially so since the EPLF was the senior partner with respect to the Sudanese army in the counter-attack. Barely a year later, the EPLF became a state actor when Eritrea achieved de facto independence. Second, the two cross-border armed attacks were both clandestine and the involvement of Ethiopia in the first attack and the EPLF in the second was not reported in the media at the time. Based on contemporaneous media reporting, the UCDP codes Ethiopia and Sudan as having internal conflicts at that time; it does not code either of those conflicts as ‘internationalised’ and it does not include interstate conflict between Ethiopia and Sudan. Third, and closely related, the number of combat deaths has not been reliably reported so it is unclear whether the UCDP criteria are met for any of the conflict dyads. Indeed, although there was clearly a conflict between Ethiopia and Sudan, it appears that troops from the two national armies did not confront one another in combat in either of the two battles (though such encounters did occur on other occasions).

This is one example among many. Drawing on these, we identify the following problems with data points in the existing datasets: non-reporting (at the time or subsequently); mismatch between the political significance of a conflict and its military scale; definitional questions about the identities of actors and relations between them; and the question of external support for coups d’état and counter-coups. We cannot claim to have comprehensively corrected for these shortcomings, but we argue that our efforts indicate the value in the re-description, provide a foundation for making some broad comparisons and set a research agenda.

The first problem is the number of incidents that are missing because they are clandestine and only reported in subsequent histories or memoirs. This is very common, especially for cross-border logistical support to armed groups. One example is the crucial role played by Tanzania during the 1960s and 1970s, when it trained freedom fighters from southern African liberation fronts and facilitated contacts between the Congolese Simba rebellion and international fighters such as Che Guevara along with international sponsors including the USSR and China. This was in line with the policies of the OAU Africa Liberation Committee, a subsidiary organ of the OAU created in 1963 and based in Dar es Salaam for African states to provide funding, logistical support, military training and publicity to all liberation movements officially recognised by the OAU (Sahnoun Reference Sahnoun, Chachage and Cassam2010). A second example is Libya under Colonel Muammar Gaddafi, which is well known for having trained and armed hundreds of rebels who initiated or sustained civil wars across Africa, sometimes using intermediaries such as Burkina Faso and Uganda. Other instances occur in the sketched narrative of transnational conflict in the fourth part of this paper.

In some cases, clandestine direct armed action is missing from datasets. The 1989–90 battles on the Sudan–Ethiopia border are a case in point. From 1994–98, Ethiopian and Ugandan military officers were also present inside Sudan, not only supporting but sometimes commanding SPLA operations (de Waal Reference de Waal and de Waal2004: 206–7). Another covert operation absent from the UCDP data occurred on 9 December 1982, when the South Africa Defence Forces launched a raid against the homes of ANC members in the capital of Lesotho, Maseru, killing 32. Lesotho took its case to the UN Security Council, but it was not covered by the newspapers of record.

The second problem is mismatch between the political significance of a dispute and the number of fatalities that resulted. For example, the Nigeria–Cameroon dispute over the Bakassi Peninsula, which involved armed conflict in 1981, 1996 and 2005, only met the UCDP threshold of 25 battle deaths in one incident in 1996, but was politically significant for far longer (Price Reference Price2005). Another is the armed clashes between Ethiopia and Somalia during 1961–63 and again in 1975, at times when the conflict between the two was covert and most of the casualties were civilians (Africa Watch 1991: 73–4). A third is the recurrent confrontation between Egypt and Sudan over the Halaib Triangle, notably in 1995 and 2016, which rarely involved fatalities. This latter case is included in the MID, but until now has not been integrated into any comprehensive dataset. The mismatch problem also arises with cross-border technical and logistical support: it may make a telling contribution to the dimensions, lethality and outcome of an armed conflict, but not one that is readily quantified.

The third set of problems is definitional. Part of this is identifying parties. The conventional distinction between ‘state’ and ‘non-state’ actors is increasingly problematic in the contexts of protracted wars with multiple parties, in which the internationally recognised government may be no more powerful or durable than so-called insurgents. Chad is a case in point. Between 1979 and 1986, state power shifted back and forth between the warlords, Goukouni Wadeye and Hissène Habré, backed by Libya and Sudan respectively until a former aide to both of them, Idriss Déby, defeated all his rivals and consolidated his hold on power. Similar challenges arise in Somalia and the Central African Republic (CAR).

A comparable definitional issue also arises in terms of classifying the relationships among parties. There is a small but growing literature on proxy warfare (Hughes Reference Hughes2012; Mumford Reference Mumford2013), but little in the way of systematic research on this phenomenon in Africa. The literature assumes that state actors are senior partners to non-state actors, but the case of the EPLF operation through Sudan into Ethiopia described above, suggests that this cannot be taken for granted. During the 2003 conflict in CAR, the Movement for the Liberation of Congo (MLC) of Jean Pierre Bemba was more powerful than its protégé, the CAR government of Ange-Félix Patassé. More widely, the high prevalence of internationalised internal conflicts through the delegation of hostilities to a domestic actor in a target state blurs the definition of a civil war (Salehyan Reference Salehyan2010: 493–515).

A fourth complication is the question of military coups d’état mounted with external support or direction, or resisted with such support. While coups are excluded from most definitions of armed conflict, the security logics that link them have recently become the focus of study (Powell & Thyne Reference Powell and Thyne2011; Roessler Reference Roessler2016). Similarly, patterns of external engagement in coups and counter-coups are highly relevant to understanding transnational conflict.

In 1979, Tanzanian troops intervened in the Seychelles and contained the coup attempt led by mercenaries from South Africa on behalf of Colonel ‘Mad Mike’ Hoare, with a battalion remained to patrol Mahé International Airport and the coast to prevent a repeat attempt. A more complicated case occurred in 1981 when Senegalese troops intervened in Gambia to stop a coup against President Jawara. By the time the coup was defeated, Senegal was in de facto control of Gambia's territory, and it took the gamble of unilaterally establishing a confederation of the two countries, to be called Senegambia. This was opposed by many Gambians and did not succeed.

PATTERNS OF TRANSNATIONAL CONFLICT IN AFRICA, 1960–2010

A comprehensive dataset that includes all types and reported instances of transnational conflict in Africa is thus needed. The TCA dataset is an attempt to fill this gap. This article's twin paper (Duursma et al. Reference Duursma, Twagiramungu, Berhe and de Waal2019) provides an overview of which datasets have been used to capture different types of transnational violence in the TCA. In brief, the TCA dataset records observations on conflict dyads in each calendar year between 1960 and 2010. A conflict dyad consists of a pair of actors that are engaged in a direct or indirect armed conflict. Through a process of research based on secondary sources and expert interviews, the TCA includes observations on cases that are missed in (or omitted from) the existing datasets because they fall outside of the temporal coverage of these datasets, observations on cases that do fall within the temporal coverage but have been overlooked or incompletely represented, and observations on cases that do not fit within the categories of the existing datasets. The TCA includes 2,977 transnational conflict dyad-years in total, of which 546 were new observations (not included in the existing datasets). This constitutes around 12.5% of the total of cases included in the TCA. The four maps in the online appendix provide a geographic overview of transnational armed conflict in Africa.

At the risk of overgeneralisation, historians of African conflicts are well-aware of the transnational dimensions of liberation struggles and post-colonial conflicts and have explored specific cases in their writings, while scholars within the disciplines of political science, international relations and security studies have marginalised or excluded transnational elements. Even the best comparative political science studies of African conflicts (e.g. Clapham Reference Clapham1996, Reference Clapham1998; Reno Reference Reno2011) do not systematically compare or theorise the transnational element (for an exception of an international relations study that does look at transnational conflict in Africa, see Tamm Reference Tamm2016: 147–81). The TCA dataset allows us to begin to develop an agenda for such an analytical comparative study of transnational conflict. This section provides the briefest of sketches of this project.

The story is one of African interstate collaboration and rivalry, that varies geographically across the African continent and changes over time. In this account, destabilisation and proxy wars are common, along with the threat and occasional use of force between states. To aid this story, we have represented the TCA data on a map, as figures in the supplementary data folder. Interstate war is displayed as a red borderline and a militarised interstate dispute is displayed as an orange borderline. Support to rebellions is displayed as purple arrows, with a dark shade of purple referring to troop support and the lighter shade referring to non-troop support. Dark green arrows refer to troop support to the government of a given country and light green arrows to non-troop support. Finally, those states that have experienced a civil war or coup in the period covered by the map are shaded.

During the years of decolonisation and the early post-colonial state (1960–1975), we see a gradual escalation of transnational conflict dyads, which overwhelmingly represents external support for conflict parties fighting a ‘civil war’. Most of this was African support to liberation movements in the Portuguese colonies and minority white regimes, and the counterpart destabilisation activities of the Apartheid state and its regional allies. A second factor is Cold War rivalries. In the Congo crisis, almost all the intervention was from outside the continent. The Horn of Africa showed a different pattern, where the states of the region intervened directly and by proxy in one another's affairs, in addition to external engagement.

The great majority of independent states experienced a civil war or a coup during this time period, with the exceptions of Zambia and Botswana. By contrast, there were only three conventional interstate wars: between Algeria and Morocco (1963–64), Somalia and Ethiopia (1964) and Egypt and Israel (1967–73).

During the subsequent 15 years (1976–90), transnational conflict remained at a high level, but its patterns shifted. While the liberation wars in southern Africa gradually came to an end, rivalries among independent African states emerged or sharpened. There were a handful of conventional interstate wars, notably Ethiopia–Somalia, Uganda–Tanzania and Chad–Libya. In each case, the conventional war was the most visible tip of a broader set of practices of either reciprocal destabilisation or external support for insurgents. A history of the Libya–Chad conflict is of necessity a narrative that shifts between a range of state actors (Libya, Sudan, Nigeria, Zaire, as well as France and the USA), the Chadian factions that moved in and out of government, often splitting and more rarely fusing, and various militia and proxy groups such as Libya's Islamic Legion and the Chadian–Darfurian Janjaweed (Burr & Collins Reference Burr and Collins1999). Such a narrative places rival claims to regional hegemony from Libya, Sudan and Nigeria, as well as Cold War geo-politics, on an equal footing with the dynamics of the internal contestation for power. In a comparable manner, the wars in Sudan, Ethiopia/Eritrea and Somalia were all entangled, not just in the Cold War politics, but also in mutual destabilisation and quests for regional hegemony.

Figure 1 The proxy wars of the Horn, 1991–2000.

Following the end of the Cold War, transnational conflict in Africa exhibited divergent and contradictory trends. Following a brief decrease in such conflict dyads in the early 1990s, there was a sharp increase by the middle of the decade. Extra-continental powers engaged much less in military terms during the decade 1990–2000. However, by the mid-1990s, regional rivalries again became a driving factor in a resurgence of transnational conflict.

This pattern was particularly notable in north-east Africa, where the fall of the Somali and Ethiopian military governments in 1991 made that Horn's ‘year zero’ (Clapham Reference Clapham2017: 60). For a brief moment, transnational conflicts declined. But they soon returned with a vengeance, initiated by the Sudanese government's overreach in exporting Islamism and the new Eritrean state's readiness to dispatch its army to contested border areas, starting with an unexpected conflict with Yemen over the Hanish Islands in 1995. The pattern that emerged was rivalry among the major states (Eritrea, Ethiopia, Sudan and Uganda), which intervened reciprocally in one another's affairs, with direct military action (covert and occasionally open) and extensive support to domestic armed groups.

During the period 1994–98, the three ‘frontline states’ of Eritrea, Ethiopia and Uganda coordinated their covert efforts against Sudan, which included both dispatching their own troops on combat missions and support for the SPLA and other rebels. This was in response to Sudanese support for jihadist groups in Eritrea and Ethiopia and the Lord's Resistance Army in Uganda. In 1998, the outbreak of war between Eritrea and Ethiopia reconfigured this alignment, with the two belligerents withdrawing their support for Sudanese rebels, and instead supporting their respective proxies in Somalia. Although these conflicts are documented, the regional rivalries that underpin them are not foregrounded in analyses (Clapham Reference Clapham2017; Bereketeab Reference Bereketeab2013).

In November1996 the Rwandese Patriotic Front (RPF), which had recently come to power in Kigali, took the lead in an invasion of Zaire with the intention of putting in place a friendly government in Kinshasa. It planned the invasion in coordination with Eritrea, Ethiopia and Uganda (de Waal Reference de Waal and de Waal2004: 208) and under the cover of support for a domestic insurgency, the Alliance of Democratic Forces for the Liberation of Congo (ADFL). Uganda mounted parallel operations; Eritrea contributed special forces. The Ethiopian role was indirect: it provided commanders and heavy equipment to military operations in southern Sudan, jointly mounted with the SPLA and Uganda, to capture the town of Yei and cut off Sudanese government reinforcements to the Zairean regime forces (de Waal Reference de Waal and de Waal2004: 209). Angola joined once the operation was underway. It had its own reason for wanting to topple the Mobutu regime: to cut off external support from Zaire to the União Nacional para a Independência Total de Angola (UNITA). In order to keep its role covert for as long as possible and not provide UNITA with an excuse to re-start the war following the conclusion of the Lusaka Agreement in 1994, the Angolan government began providing massive support to the anti-Mobutu forces through an airlift of heavy material, engineers and intelligence officers to Rwanda rather than directly attacking Zaire from its own territory. Angola became the second biggest foreign contributor to the anti-Mobutu forces, with only Rwanda providing more support (Roessler & Verhoeven Reference Roessler and Verhoeven2016: 200).

This First Congo War which ended with the replacement of the Mobutu regime in May 1997 sowed the seeds for the second war. The anti-Mobutu alliance had been divided over strategy, with Rwanda and Eritrea arguing for rapid regime change and Uganda and Ethiopia advocating for a more protracted struggle so that the ADFL could build up its political capacity. The latter's fears were proven correct: the ADFL government was inept and its former patrons turned against one another in a fight to secure the spoils of victory. The second Congo war lasted from August 1998 to July 2003.

The two Congo wars are a very well-documented case of transnational armed conflict (Clark Reference Clark2002; Stearns Reference Stearns2011; Roessler & Verhoeven Reference Roessler and Verhoeven2016). The second Congo war soon expanded to include the armed forces of most of the neighbouring countries and some further afield including Libya, Namibia and Zimbabwe. The rivalries of neighbouring states were played out, directly or by proxy, within the territory of the DRC. The Congolese state was not an actor in cross-border armed activity itself. However, one Congolese non-state actor (the MLC) engaged in military action in CAR in support of the then government of CAR.

The Congolese wars are generally interpreted as the exception that proves the rule: Africa's one big case of international conflict. However, the TCA dataset shows that the Congolese war was less of an exception, and more as the accentuation of an existing pattern. This can be seen in the fact that the state actors that came together in 1996 to support the Rwandese invasion of Zaire consisted of leaders who either had a history of supporting Pan-African causes (Tanzania's Julius Nyerere, Uganda's Yoweri Museveni) or were already engaged in military action in Sudan (Eritrea's Isseyas Afewerki, Ethiopia's Meles Zenawi and Uganda's Museveni) and saw the invasion of Zaire as an extension of that projection of power.

In West Africa, there was a different pattern, with the inception of the wars in Liberia and Sierra Leone – conflicts which were closely entangled with one another – and the eruption of war in Côte d'Ivoire. These conflicts also had significant transnational elements, including the sponsorship of insurgents by Burkina Faso and Libya, and the active role of Nigeria in seeking regional hegemony, in part through its leading role in peace operations under the auspices of the Economic Community of West African States (ECOWAS).

During the first decade of the millennium, the pattern of transnational conflict shifted again. Notably, non-African support to African conflict parties was ramped up during this period. While the sole recipient of US support for pursuing an armed conflict in the 1990s was Rwanda, with the launch of the Global War on Terror in 2001, many more countries received US military support to conduct conflicts. France also resumed military support. The legacy of the previous decade of regional rivalries, pursued with military tools, also continued, with some interesting differences. One of the striking features of this period is how transnational armed interventions have been replaced in part by coercive regional peace enforcement. It is striking, for example, that some cases of African countries contributing troops to African Union (AU) peace operations in their neighbourhoods, replicate the same patterns of non-consensual military intervention of earlier years.

CONCLUSION: IMPLICATIONS FOR RESEARCH AND POLICY

This paper shows that the conventional wisdom that Africa has experienced little interstate conflict is misleading. The majority of African conflicts should correctly be re-classified as internationalised internal wars, while other forms of interstate violent or coercive actions are not uncommon. Under the narrow definition of a civil war – a war that is fought solely between a sovereign government and armed non-state actor within a country (Sambanis Reference Sambanis2004: 814–58) – there are far fewer cases of civil war in Africa than typically assumed.

This error appears to have arisen in part because of the interaction between a norm for international behaviour and the methods used in generating the most authoritative conflict dataset, namely the UCDP. The norm in question is that the use of force to settle disputes between states is prohibited, and especially in Africa this norm is reinforced by a prohibition on altering inherited colonial territorial boundaries. It follows that insofar as an African government wishes to use force as an instrument of policy against another state, it will do so clandestinely. In turn it follows that any data-gathering method that relies on contemporaneous public sources, such as major international news media or NGO reports, will understate clandestine military activities. As a result, the UCDP has unwittingly misrepresented the nature of African conflicts. Historians and country experts whose own writings do not make this error have not until now sought to correct the error. For the academic and policy communities, the absence of datapoints for transnational conflict is interpreted as evidence for its absence, and the norm of non-use of violence between states thereby erroneously appears to have been upheld.

The exercise in classifying and cataloguing transnational armed conflict incidents allows for a systematic analysis of this phenomenon. Additional research can further document, categorise and theorise in a more rigorous and systematic way the dynamics of transnational armed politics in post-colonial Africa. Four particular sets of research questions arise.

The first set of questions concerns the nature and patterns of interstate rivalry. These issues are familiar to historians of Europe and international relations scholars studying Asia and the Middle East, and there are questions over whether and how to adopt concepts of regional hegemony to sub-Saharan Africa. Given the deepening influence of North African states south of the Sahara, and the penetration of the rivalries of the Arabian Gulf into the Horn of Africa, there are also immediate applications for the understanding of current regional political dynamics.

A second set of issues relates to the instruments used in transnational conflicts. For example, conventional interstate war appears to occur most often in the context of wider support for proxies and covert interventions. Questions arise as to why and how support is extended to domestic armed groups and recognised governments; and under what conditions such covert engagement escalates to the point of intervention. It would be interesting to investigate what other instruments are used, such as diplomatic and economic pressure.

A third avenue for research is the linkage between transnational armed conflict and internal armed conflict. We observe that most cases of conventional interstate war occur in the context of interstate rivalry that is conducted in part by support to domestic insurgents. We have not investigated any sequential or causal relationships, however. Nor have we examined whether transnational conflict typically follows internal conflict in one or other of the states involved, or whether internationalised internal conflicts should be considered as a different type of conflict to solely internal wars, with distinct causes and trajectories.

A fourth cluster of questions relates to the relationship between transnational conflicts, the drivers of such conflicts, and the African peace and security architecture. Have African inter-governmental mechanisms such as the AU and regional economic communities sought to prevent, manage and resolve transnational conflicts as such? If so, have they been successful? Have African intergovernmental security mechanisms reinforced regional state power hierarchies or reconfigured them? Has the development of multilateral instruments, both coercive and non-coercive, supplanted the unilateral forcible measures used by states in previous decades?

Lastly, we caution against the use of existing datasets such as the COW and UCDP without careful attention to their methodological and definitional limitations. Noting that a comparable critique has been mounted of economic data in Africa (Jerven Reference Jerven2015), this calls into question the solidity of many of the quantitative analyses of the links between economic and conflict variables that abound in the literature.

This re-interpretation of the patterns of African conflicts also has important consequences for conflict prevention, management and resolution policies.

The first point is that conflict resolution must not be seen as solely an internal matter for the country concerned, but also as an exercise in mediating the political interests of neighbours and regional hegemons see World Peace Foundation 2016. It follows that, when the neighbours’ interests are in alignment or they have been resolved through process of regional mediation, a resolution to the internal conflict is far more likely, than when those neighbouring countries have conflicting interests. Thus, the common positions of the member states of the Intergovernmental Authority on Development (IGAD) facilitated Sudan's Comprehensive Peace Agreement in 2005, while the differing positions of the neighbouring countries hindered the peace process in the DRC and stood in the way of African leadership on peace missions in that country.

Second, the neighbours involved in peace missions (political and military) already have a stake in the country concerned. This may strengthen their ability to influence the outcome of the process, but it also means that there is a greater risk that their involvement is not impartial, and that the peace negotiations and any peace enforcement operation is an extension of national security interests or hegemonic ambitions. The roles of Nigeria in Liberia in the 1990s and Ethiopia and Kenya in Somalia in the last decade are cases in point.

The third important implication is that African states and the AU face a dilemma in developing their political and security strategies for the ‘shared spaces’ of the Mediterranean–North Africa and the Red Sea–Gulf of Aden. In the Middle East and North Africa, interstate rivalries and contests for regional hegemony are overt, and the instruments for pursuing them are well-developed. The conflict in Libya in 2011 witnessed both a clash of interests and of political norms between the AU (which advocated unsuccessfully for a negotiated settlement) and the Arab League and NATO (which pressed for regime change by military means) (de Waal Reference de Waal2013). This appears to have been a harbinger for how Middle Eastern and North African states have subsequently pursued their interests south of the Sahara and in the Horn of Africa. This kind of transactional realpolitik threatens to reverse the fragile gains made by the African peace and security architecture, and demands a strategic response from the AU.