As global climate change affects livelihoods across the world, migration patterns are also changing. In a recent study published in Global Environmental Change, Clark Gray, Valerie Mueller, and I found that since the 1970s, climatic variations have been increasing internal migration across many South American countries, with few exceptions. And many people are headed to cities.

Using data from 25 censuses conducted in eight countries between 1970 and 2011, we found that both above- and below-average temperatures were associated with an overall increase in the likelihood of migration between provinces. We focused on migration among adults aged 15 to 40, the ages when migration is most common, that did not cross international borders. By linking these data – over 21 million observations in total – to historical climate records, we could apply statistical models to estimate the association between variability in temperature and precipitation and the likelihood of migration between provinces, controlling for a set of other factors that may have affected migration.

Each month that average temperatures were two or more standard deviations above normal – a low probability event – the odds of out-migration increased three percent on average across the countries in the study: Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Paraguay, and Uruguay. Comparable exposure to below-average temperatures increased the odds of migration by an average of eight percent.

We speculate that these patterns may reflect the disruptive effects of anomalous temperatures on agricultural production and economies dependent on agriculture in the region. We did not identify such mechanisms in our data, but this hypothesis is consistent with evidence that both cold and warm temperatures reduce crop yields.

In addition to these overall effects, anomalous temperatures had a stronger and more consistent positive effect on migration to urban destinations compared to rural destinations. We were not able to identify whether migrants moved from rural or urban areas, so could not quantify the extent to which climatic variability was contributing to urbanization. But it likely has had some effect. And our findings do suggest that climate change is contributing to increases in the overall population of some South American cities, whether migrants are coming from rural places or other urban areas.

The links between climate change and movement to urban areas raises questions about the planning needs of cities on the receiving end of such migration flows. It also raises questions about migrant incorporation into these cities, which may be a particular challenge for individuals moving from rural areas, who suddenly face a very different setting and new livelihood prospects.

Demographic Differences and Trapped Populations

The association between climate change and migration did vary across demographic groups. Gradual increases in temperature increased migration among women, but not men, and among individuals who did not complete primary school, but not among those who did.

However, exposure to more extreme changes in temperature – when monthly temperatures were two or more standard deviations above or below the mean – had significant positive effects on migration among all groups. Differences in climate effects across groups are seemingly a matter of degree than of kind.

Finally, analysis of how climate affects migration in each individual country revealed a major exception to our overall findings. In Bolivia, higher temperatures were associated with decreased migration, the opposite of what we found in other countries. This “trapped population” dynamic has been observed elsewhere and is expected to be most common among very poor populations. This may help explain its presence in Bolivia, where the very poor make up a larger share of the population than other countries in our study.

While we were not able to identify the particular characteristics that explain the different results across demographic groups and countries, the variability in climate-migration links is consistent with the high degree of variation in wealth, dependence on agriculture, and geography across South America. Future research on these characteristics – essentially, how the interaction between socioeconomic conditions and climate change produces these results – may allow policymakers to proactively identify countries where climate-related changes in migration are most likely.

This was one of the first studies to examine climate-related migration patterns across multiple countries using comparable data and measures. The fact that it identified significant climate effects on migration adds further support to the expectation that climate change will disrupt human population movements. Yet the variation in these climate effects also underlines the need to anticipate nuances and tailor analyses, conclusions, and policy recommendations to specific contexts.