Internally displaced people at Bangui airport Internally displaced people at Bangui airport Photo credit: S.Phelps/UNHCR
Humanitarian evacuations in the Central African Republic
by Josep Zapater September 2014

The situation in the Central African Republic (CAR) constitutes one of the most intractable humanitarian crises in the world. Hundreds of thousands of people have been displaced and are living in dire conditions. In a context of continuing violence, a failed state, insufficient international troops, limited capacity of humanitarian actors and the lack of a clear chain of command for non-state armed actors, protecting war-affected communities through humanitarian action has proved particularly challenging.


In March 2013, northern Seleka fighters took the CAR capital Bangui, ousting President Michel Bozizé. Seleka fighters were mostly Muslims, and included numerous Arabicspeaking mercenaries from Chad and South Sudan. The Seleka unleashed a wave of killing and looting in the capital and in the west and north of the country, targeting mostly non-Muslim communities. Many rural populations began reacting to Seleka violence, giving rise to the ‘antibalaka’ militias. By August 2013, the anti-balaka reaction had become a terror campaign against Muslims perceived as having collaborated with Seleka violence. As Seleka forces withdrew to the east of the country, Muslim communities fled their homes. Tens of thousands of people, mainly Peuhl (Fulani) cattle herders, embarked on a massive exodus to Cameroon. Many of those who could not flee joined Muslim communities trapped by the anti-balaka in urban centres in western CAR.

Stay – or go?

The Protection Cluster undertook to monitor this situation, publishing a weekly map and statistics on communities at risk and giving visibility to the crisis in the international press. The Cluster also prepared a strategy to protect these populations, in close collaboration with the Humanitarian Country Team and the Senior Humanitarian Coordinator. The strategy included daily monitoring through field presence, networking with humanitarian actors and local journalists. The Protection Cluster advocated weekly with international military actors to prioritise communities at risk when planning their presence on the ground. Protective presence and basic assistance, including food, health, non-food items and shelter, was planned, promoted and coordinated. More controversially the strategy, which was quickly adopted by the Humanitarian Country Team, also included the relocation of trapped communities as a measure of last resort, if the violence could not be stopped.

A number of international NGOs, including Mercy Corps, the Danish Refugee Council (DRC) and Search for Common Ground, argued that the strategy focused too much on relocation and that more emphasis should be given to reconciliation and to protecting communities in situ, to avoid further segregation and to give communities a real chance to stay. Mercy Corps took a finely nuanced position, which included proposing measures such as phone calls and visits to try to ensure social links between Muslim and non-Muslim communities after relocation, even on a cross-border basis. The Protection Cluster and UNHCR, while accepting the principle that efforts for social cohesion could happen in parallel, argued that many communities wanted to leave, that most of those trapped were already displaced and that relocation would facilitate their freedom of movement. They further contended that, given the continuing violence and animosity, the only way to avoid further killings was relocation – including, when necessary, across the country’s borders.

Objections to relocation from other quarters, including the government and the French embassy, were more political. There was concern that relocating Muslim communities to Seleka-controlled areas (which was the only realistic option) would reinforce the militia and advance its objective to divide the country into separate Muslimmajority and Christian-majority areas. It took the courage of the then Senior Humanitarian Coordinator, through a series of negotiations supported by UNHCR, the Protection Cluster, OCHA and other humanitarian actors, to resist this political pressure.

The relocation to northern CAR of the 1,200-strong Muslim community from the PK12 neighbourhood in Bangui, the most visible community at risk, was a messy affair. Detailed planning only started a few weeks before the relocation. Logistical hurdles, including the availability of adequate trucks, were enormous. While the displaced were boarding the trucks, anti-balaka militias surrounded the area, awaiting the departure of the military-escorted convoy in order to loot the area and defile the mosque. On a very bad dirt road, travelling through jungle, steppe and guerrilla-infested villages, it took the 20-truck convoy four days to reach its final destination. The convoy was attacked twice by anti-balaka, killing three people.

There were coordination and operational shortcomings among the actors involved (the International Organisation for Migration (IOM), the Protection Cluster, OCHA, the UN Multidimensional Integrated Stabilisation Mission (MINUSCA) and others). Protection, logistics and military aspects of the operation were not sufficiently integrated in planning. Important aspects of planning at the relocation site, such as impact on economic life and social cohesion, only started after the relocation had taken place. However, despite the tragic loss of life, and other problems with the operation, the decision to relocate people from PK12 was the right one, and relocation in general is a valid protection strategy. If relocations in Bossembele, Bossangoa and PK12 had not taken place, the Muslim populations in these enclaves would have been at high risk of generalised massacres by anti-balaka militias. Simply put, in these communities the humanitarian community had to assume that the human cost of remaining at displacement sites was much higher than the costs of relocation.

Relocation or reconciliation?

It is important to give more background on discussions around reconciliation in CAR. By January 2014, many within the international community believed that violence in CAR was of both a communal and a religious nature. This led to highly publicised attempts to support dialogue between Muslim and Christian religious leaders as a means to reduce violence. These attempts had little or no effect on the ground, for a number of reasons. First, the roots of the conflict were much more nuanced than religious enmity, and they were also very localised. In Boda, non-Muslims claimed that they resented the Muslim majority’s domination of the gold and diamond economy, which enabled them to buy local political power from the central government.

Second, the ‘reconciliation’ approach ignored the fact that individuals, not communities, were guilty of atrocities. Even at the height of the violence, some Muslim and non-Muslim communities continued to trade, talk and live together in Bangui and other towns. For those who lost loved ones and property, or who were injured in the violence, reconciliation could not be achieved without justice. Third, whereas there are many opportunists and petty criminals among their ranks, many anti-balaka leaders stated that they had to uproot the Muslim population to destroy any social basis for more Seleka killings, in a sort of community-based counter insurgency strategy. They also believed that they had saved the country from the brutality of the Seleka, and that their sacrifices had not been rewarded with political power or moral recognition. None of the reconciliation initiatives acknowledged or attempted to address their grievances – justified or not, but real even so.

Violence led to fear and fear led to more violence, creating displacement and segregation which closed down communication channels between Muslim and Christian communities. The lack of communication increased mistrust and fear, further feeding the cycle of violence. The humanitarian community largely lacked the tools to address the social aspects of the violence. The UN Department of Political Affairs deployed a mediation team to CAR to help the government develop a reconciliation strategy. Of course, the strategy could only be as valid and legitimate as the government which formally adopted it, and lacked in the short term mechanisms to implement it at field level.

By May 2014 the DRC had developed and field-tested a methodology for emergency mediation, seeking to reduce tensions in communities in the short term so as to enable humanitarian actors to operate. The model included intervention by trained third-party mediators between armed actors, or between armed actors and communities, in a neutral, impartial and non-judgmental way. The Protection Cluster adopted and supported these efforts as part of attempts to protect communities at risk. However, this all took time and had to be tested: when the community in PK12 had been relocated, there had been no systematic attempt aside from a military presence to prevent violence there. The Protection Cluster and OCHA also led the preparation and implementation of localised humanitarian action plans in some other communities at risk, with varying degrees of do no harm analysis, but these efforts were not able to cover all communities at risk, largely because of the lack of professional protection staff, in particular in remote areas.


It is too early to draw firm lessons and conclusions from protection strategies for communities at risk in CAR. Two can perhaps be advanced. First, protection strategies in the midst of conflict need to resist political pressure, be pragmatic and actionable and ruthlessly prioritise the worst protection problems. As an example, there was some discussion in the Protection Cluster on how far to prioritise protection mainstreaming in existing aid efforts, while huge swathes of the population, in particular in rural areas outside of Bangui and in communities at risk, were completely devoid of aid. The leadership of the Cluster took the position that, for a limited period, Protection Cluster advocacy for the extension of basic aid to remote areas should take priority over efforts to mainstream protection in areas where aid was being distributed. Providing effective protection needs professional human resources, including in remote areas. These resources need to integrate personal maturity and skills in protection in conflict-affected areas, negotiation, facilitation and mediation.

Second, the protection community needs to develop approaches to reducing the social aspects of violence, including mediation practices that can be used locally and in the short term. Even in an emergency, humanitarian actors need to be able to analyse the social, economic and even anthropological aspects of violence. As an example of other contexts, UNHCR could not have assisted indigenous communities in Colombia to develop self-protection strategies without the help of professional anthropologists. In CAR, incipient efforts by a few humanitarians to understand how people felt about violence led, at least, to the realisation that high-level reconciliation efforts were not effective in communities at risk and that, at least in the short term, localised, professional mediation efforts needed to be developed and tested. The humanitarian community needs to continue exploring and drawing lessons, even painful ones, from efforts to address violence in CAR.

Josep Zapater was Protection Cluster Coordinator in the Central African Republic between February and May 2014. The opinions expressed in this article are the author’s and do not necessarily reflect those of the Protection Cluster or UNHCR.