Proc. Natl Acad. Sci. 109, 18344–18349 (2012)

Previous studies have shown a positive impact of climatic variability on the risk of conflict in sub-Saharan Africa, with temperature as the strongest driver of violence. However, the lack of thorough geographical considerations challenges those results.

John O'Loughlin of the University of Colorado, Boulder, USA, and colleagues examined climate–conflict relationships with, for the first time, a geographically disaggregated approach. Using a conflict database that contains 16,359 individual geo-located violent events for East Africa from 1990–2009, they considered a wide range of socio-economic, geographic and political factors at a 1° gridded resolution. Unlike earlier studies, they found that — after controlling for the non-geographic factors — precipitation levels that are two standard deviations above the long-term average decrease the risk of violence by 30.3%, whereas drier and normal periods have no effect. Results also show that temperatures two standard deviations higher than the average raise the risk of violence to 29.6%, with normal and cooler temperatures not affecting incidence of violence.