Countries where the majority of the population lives in areas that become much warmer in El Niño years (red) are more likely to experience wars than those where temperatures are less affacted (blue). Click for larger image. Credit: Hsiang et al. Nature

Click here for larger image

Natural climate cycles seem to have a striking influence on war and peace around the equator. Tropical countries face double the risk of armed conflict and civil war breaking out during warm, dry El Niño years than during the cooler La Niña phase of the El Niño/Southern Oscillation (ENSO), according to an analysis published today in Nature1.

The study throws light on the hotly contested issue of whether climate change has any notable effect on violence and societal stability, particularly in poor countries. The authors of several popular books have previously proposed a link, but there are disagreements within the scientific literature over whether a robust climate signal can be detected in conflict statistics.

Previous studies have focused on the question of how anthropogenic climate change might increase conflict risk. A 2009 study2 by economist Marshall Burke at the University of California, Berkeley, and his co-workers found that the probability of armed conflict in sub-Saharan Africa was about 50% higher than normal in some unusually warm years since 1981. But critics point to statistical problems — for instance when linking possibly random local temperature and rainfall variations with outbreaks of civil war — that may have resulted in a false appearance of causality.

To overcome this problem, Solomon Hsiang, an economist currently at Princeton University in New Jersey, and his colleagues opted to look at how historical changes in the global, rather than local, climate affect conflict risk1.

Clear signal

The team designed a 'quasi-experiment' for which they divided the world into regions strongly affected by the ENSO — the tropical parts of South America, Africa and the Asia–Pacific region, including parts of Australia — and regions only weakly affected by it. They then searched for a link between climate and armed conflicts that arose in the first group between 1950 and 2004.

A very clear signal appeared in the data. The team found that the risk of annual civil conflict doubles, from 3% to 6%, in countries of the ENSO-affected, or 'teleconnected', group during El Niño years relative to La Niña years. In many cases, conflicts that might have broken out anyway may have occurred earlier owing to the effects of El Niño, Hsiang suggests.

Civil conflicts have been by far the most common form of organized political violence in recent decades, Hsiang says. Globally, one-fifth of the 240 or so civil conflicts since 1950 could be linked to the 4–7-year climate cycle originating in the southern Pacific, the study concludes. The results were unaffected by any modification to the statistical set-up of the analysis — such as excluding particularly crisis-prone African countries — which the team performed to confirm the robustness of their findings.

"A doubling of risk is a very strong effect," says Halvard Buhaug, a conflict researcher with the Peace Research Institute Oslo, who was not involved in the study.

Buhaug, who has previously criticized3 claims such as Burke's, says he feels "surprised and a bit puzzled" by the results. He grants that the study is "very competently executed" but adds that the issue is nonetheless far from being settled. "I don't dismiss that a correlation exists, but it is a correlation we so far don't understand," he says. "I remain sceptical about any potential causal connection."

A more detailed analysis of the 'narratives' of historical conflicts that have occurred during El Niño years is needed to establish whether any factors that may have caused these conflicts — such as harvest failures that led to food shortages — can be traced to El Niño events, he says.

Greenhouse effects

The authors of the study are aware of its limitation and of the difficulties involved in establishing a causal link between climate and conflict. But, says Hsiang, case studies are ongoing at Columbia University in New York and elsewhere on how El Niño events might link to local outbreaks of violence.

"Different hypotheses have been proposed as to how one phenomenon causes the other, and we aren't sure yet what the correct narrative is," he says. "It could be that agricultural income in El Niño years drops to levels that can trigger violence. Furthermore, psychologists think that aggressive behaviour gets generally more widespread during exceptionally warm conditions."

El Niño events, he adds, are by no means the sole factor leading to conflict. But although these natural climate cycles do seem to play a part in the peacefulness of nations, the authors warn against rushing to the conclusion that anthropogenic greenhouse warming will lead to more armed conflict and political instability.

Climate models give ambiguous projections as to how ENSO will change in a warming world. "El Niño is different in structure than anthropogenic climate change," says Hsiang. "It would therefore be hard to map our results onto future changes."

But Burke, who was not involved in the study, says that the work could be useful for at-risk countries. "The fact that ENSO is itself somewhat predictable makes their findings policy-relevant," he says. "If we think an El Niño is coming, then governments in teleconnected regions could put in place measures and safety nets to try to reduce the risk of conflict in that year."