A new Rice University study is shedding light on a long-debated question: Can climate variability influence the risk of armed conflict? The answer, researchers say, is yes — but in more nuanced and region-specific ways than previously understood.
Led by Rice statistics doctoral student Tyler Bagwell , with climate scientist Sylvia Dee and statistician Frederi Viens , the study uses high-resolution data and empirical modeling to examine how large-scale climate patterns shape the probability of civil conflict and war. The research is published in the Proceedings of the National Academy of Sciences .
"We wanted to understand whether armed conflict risk is linked to these climate patterns, and whether local conflict risk scales with how strongly the patterns influence local weather," Bagwell said. "This would allow us to identify regions particularly susceptible to climate-driven political or social instability."
The research focuses on two major climate patterns: the El Niño-Southern Oscillation (ENSO), a well-known driver of global weather variability, and the Indian Ocean Dipole (IOD), a more regionally focused system influencing climate across the Indian Ocean basin. Both the ENSO and IOD are climate patterns driven by ocean temperature anomalies that seesaw east and west across the tropical Pacific and Indian Oceans, respectively, affecting global weather patterns.
"Crucial to our study is the fact that the extreme phases of ENSO and the IOD are each associated with distinct, often opposing, local climate impacts," Bagwell said. "For instance, ENSO oscillates between two phases — La Niña and El Niño — and each is linked to specific weather impacts around the world. Exploiting these differential impacts, we then found statistical relationships between when and where armed conflicts occurred and whether ENSO was in El Niño versus La Niña and whether societies experienced dry versus wet ENSO impacts."
While previous studies have linked El Niño to increased conflict risk, the Rice team took a more detailed approach. By building a novel, high-resolution dataset of more than 500 conflict onsets in the period 1950-2023, pinpointed in both space and time, the researchers were able to move beyond country-level aggregation and examine local relationships.
"That level of spatial detail documenting conflicts over such a long time span hasn't really existed before in a dataset," Bagwell said. "It allowed us to look at how climate variability affects conflict risk at a much more local scale over decades."
Creating the dataset itself was a major undertaking. Rice undergraduate researchers Anna Stravato and Divya Saikumar manually analyzed primary sources, including news reports in multiple languages, to geolocate each conflict event, a process that could take up to an hour per case.
One of the study's most significant findings is that not all climate impacts are equal.
"Similar to previous studies, we found that global armed conflict risk is greater during El Niño relative to La Niña, but we also found that increased conflict risk during El Niño is primarily linked to regions that experience drier conditions," Bagwell said. "In areas where El Niño is associated with wetter conditions, we do not find a credible relationship."
That distinction helps clarify a lingering debate in climate-conflict research, which has struggled to identify consistent causal pathways that link climate conditions to conflict. Rather than a simple global pattern, the study suggests that drought-related stress, such as reduced water availability or agricultural strain, may play a larger role in elevating risk.
The findings also provide evidence that conflict risk associated with El Niño does not grow in proportion to how exposed a region is to ENSO beyond some baseline exposure, indicating possible threshold effects.
The study also identifies a previously undocumented link between the IOD and conflict risk. Unlike ENSO, where one phase (El Niño) is associated with higher risk, both positive and negative phases of the IOD were found to increase conflict risk in regions whose climates are strongly coupled to the IOD, particularly in the Horn of Africa and parts of Southeast Asia.
"That's a very different pattern," said Dee, who is an associate professor of Earth, environmental and planetary sciences. "The Indian Ocean Dipole operates on shorter timescales and can shift rapidly, creating climate 'whiplash' that may disrupt already vulnerable regions."
The study provides strong statistical evidence that climate variability can act as a "threat multiplier," amplifying existing social and economic pressures. Importantly, both the ENSO and IOD can be predicted months in advance, offering a potential window for preparedness.
"These climate modes are predictable on seasonal to annual timescales," Dee said. "That means there's an opportunity to use this information as part of early warning systems."
The authors say the findings could help inform policymakers, humanitarian organizations and peacekeeping efforts by identifying when and where risks may be elevated.
"We can't say definitively that climate causes conflict," said Viens, who is a professor of statistics. "But we can say that some climate patterns change the probability of conflict. And understanding those shifts in risk is valuable for planning and mitigation."
"These results shed light on important connections between climate and conflict," Bagwell added. "With both American and European meteorological agencies predicting the emergence of El Niño by the end of this year, and even some forecasts anticipating a super El Niño, our findings are especially timely."
The study originated through a Rice Faculty Initiatives grant designed to foster interdisciplinary collaboration, bringing together expertise from climate science and political science.
"This is a great example of how combining different skill sets crossing our fields can lead to new insights," Dee said. "It took years to bring the right team together, but the result is a dataset and analysis that opens new directions for critical climate-conflict research."
