Long-term environmental change and geographical patterns of violence in Darfur, 2003–2005

Many publications that investigate connections between environmental change and violence refer to the case of Darfur (Benjaminsen et al., 2012, Raleigh and Kniveton, 2012, Salehyan, 2008). Qualitative studies emphasize that decreasing access to water and fertile land has played an important role in the violence in western Sudan (e.g. De Waal, 1989, Flint and Waal, 2005, Harir, 1994). Quantitative research, on the other hand, generally concludes that natural resource availability has not been a primary cause of the civil war (e.g. Brown, 2010, Kevane and Gray, 2008, Olsson and Siba, 2013). This case-specific debate mirrors the more general research on the links between environmental change and violence: while some studies find correlations, most find evidence against such a connection (see overviews in Bernauer et al., 2012, Gleditsch, 2012).

This paper aims at contributing to this debate. It argues that previous research has not adequately considered the role of spatial variations in longer-term ecological developments. Studies have focused on the role of natural resource availability in general terms, on short-term variations, or on geographically aggregated longitudinal trends. Environmental change can, however, affect subnational regions in different ways. The increase in the availability of natural resources in some areas versus a decrease in others can, in the long run, shape the geographical distribution of livelihood opportunities and lead to migratory processes that may in turn affect conflict risks.

This paper investigates these claims in the case of Darfur. More specifically, it analyses whether longer-term changes in the subnational distribution of water and vegetative resources can contribute to explaining the spatial distribution of violence in the early phase of the civil war (2003–2005). The empirical analysis presented in the paper consists of two parts. The first uses qualitative and descriptive quantitative information to demonstrate the plausibility of the argument in the case of western Sudan. More specifically, it presents evidence that corroborates the assumed causal steps linking environmental change to the geography of violence. The second consists of a quantitative analysis that treats the argument more systematically. The main explanatory and outcome variables are based on information extracted from satellite imagery of the region. The Humanitarian Information Unit (HIU) of the US State Department provides a data set on villages that have been damaged or destroyed during the civil war. I use remote sensing information on changes in vegetation to construct variables of environmental change in the 20 years prior to the beginning of the civil war in 2003. I complement both data sources with additional socio-economic data to construct a data set of artificial grid cells covering all populated areas of northern and southern Darfur. Estimations using these grids as units of analysis yield robust correlations between the probability and intensity of violence and long-term changes in natural resource availability.

Natural resources and violent conflict

Associations between environmental change and violence have received considerable attention over the past 10 years (Bernauer et al., 2012, Gleditsch, 2012). Empirical studies can roughly be attributed to one of three different approaches, depending on how they consider temporal variation: The first focuses on natural resource availability in general terms. The second approach focuses on short-term variations in environmental conditions – namely rainfall. The third investigates the role of

Environmental change, migration and violent conflict

The main argument of this paper consists of three causal steps that link long-term environmental change to violence: First, environmental change can lead to migration. Second, migration can change demographic constellations and increase resource competition in areas of high in-migration. Third, both factors taken together can increase the risk of violent interethnic resource conflict in subnational regions that experience long-term increases in the availability of water and vegetative resources

The case of Darfur

The beginning of the civil war in Darfur was not directly related to resource issues. It evolved out of rebel groups' struggle against economic and political marginalization by the Sudanese central government (Flint and Waal, 2005, Flint, 2010a). The underlying grievances were shared by most parts of the population in western Sudan. Nonetheless, organized resistance initially and most strongly emerged among non-Arab groups (Tanner & Tubiana, 2007).

At that time in 2003 the government was heavily

Qualitative evidence for the role of environmental change

The following subsection presents evidence that underscores the plausibility of the causal chain between environmental change in the period from 1982 to 2002 and observable patterns of violence in Darfur. This subsection primarily draws on visual inspections of geographical patterns and anecdotal qualitative information. These sources alone do not allow for strong empirical claims. In combination with subsequent quantitative analyses, however, they can provide important insights into

Quantitative analysis

The next section investigates whether the probability and intensity of violence was actually higher in areas that experienced favourable ecological developments in the 20 years prior to the beginning of the civil war in Darfur.

I rely on artificial grid cells as units of observation.⁵ More specifically, I use a grid cell net that I have modelled according to an extensive environmental and livelihoods vulnerability mapping (ELVM) that was undertaken in Darfur in 2010 (International Organization

Conclusions

Migration has been emphasized as an essential element of the causal pathway linking environmental change to violence (Barnett and Neil Adger, 2007, Homer-Dixon, 1994, Homer-Dixon, 1991, Reuveny, 2007). The deterioration of livelihood opportunities can lead to population movements. These processes can alter demographic constellations and increase resource competition in areas that attract large-scale in-migration. The resulting risks of violence may be particularly high when migration confronts