The Climate Driven Migration Map Hidden in Ancient DNA

Charles Rotter

Every so often I find a climate paper that is a pleasure to read because it is not trying to sell me anything. No policy prescription, no call to action, no press release warning that we have five years to save the planet. Just four researchers, three of them at Harvard, asking a genuinely interesting question and answering it with a method that did not exist a decade ago. The paper is “Climate and Prehistoric Migration,“ a June working paper from the National Bureau of Economic Research by Peter Huybers, Marco Tabellini, Charles A. Taylor, and Francesco Toti. Huybers is a serious climate scientist, not a man with a skeptical axe to grind. I am going to report what they found and let you do what you like with it.

Abstract

What factors drove human migration before modern states, markets, and borders? We develop a sorting framework in which climate-specific subsistence knowledge depreciates with ecological distance. To test this, we use ancient DNA identity-by-descent segments to construct bilateral migration flows across Western Eurasia over the last 10,000 years. We document three main findings. First, migration flows decline with differences in growing degree days, precipitation, and
soil characteristics between origins and destinations. Second, the binding factor varies across subsistence systems: farmers exhibit strong thermal and soil matching, while pastoralists match most strongly on precipitation. Third, periods of warming increase farmer expansion while coolingincreases pastoral expansion in patterns that recover known archaeological migration episodes.

Migration also acts as a margin of climate adaptation: populations exposed to temperature change
move to destinations that partly offset the shift.

The method is the fun part

The question is simple: before borders, wages, and states, what moved people around the map? The answer they reach for is written in ancient DNA, and the trick is clever enough to explain slowly.

When two people share an identity-by-descent segment of DNA, a stretch of genome inherited intact, it means they share a recent common ancestor. If those two people were buried in different places, then someone in that shared lineage must have walked from one place to the other. Because every skeleton in the dataset is radiocarbon dated, you know which burial is older, so each genetic link is not just a connection but an arrow: older burial is the origin, younger burial is the destination. Do this across thousands of ancient genomes and you have something no one has had before, a directional map of who moved where across Western Eurasia over the last 10,000 years.

The distances are the first thing that stops you. Among the pairs sharing a long segment, meaning a common ancestor only ten to fifteen generations back, the average origin and destination sit 1,550 km apart. That is Berlin to Moscow, covered by a lineage in a few centuries. One Yamnaya man buried near Samara in southwestern Russia around 3200 BCE has documented descendants 2,600 km to the east in Siberia and 2,200 km to the west in Central Europe within a few hundred years. Three thousand years after that, the genetic threads still connect Central Europe to medieval Mongolia across more than 6,500 km. People moved, constantly, and far.

Finding one: people moved toward climates that felt like home

Here is the organizing idea. The knowledge that let you survive was local. A farmer who understood a cool growing season could not simply transplant that skill to a hot one, and a herder who knew arid grassland could not easily run his animals through a humid continental winter. That know-how was a kind of capital, and it lost value the farther you carried it from the conditions it was built for. So people should have moved, disproportionately, toward places that resembled the ones they left. The authors call these corridors of ecological familiarity.

The data agree. Feeding the flows through a standard gravity model, they find migration falls sharply as origin and destination grow less alike in growing-season heat, in rainfall, and in soil. A one-standard-deviation gap in growing degree days, precipitation, and soil each cut migration by roughly 20, 21, and 16 percent. To put that in walking terms, every one of those ecological gaps was worth an extra 570 to 770 km of travel, which is a lot set against an average move of 1,550 km. People did not just avoid distance. They avoided difference.

Finding two: it depended on how you made a living

This is the part that makes the result hard to wave away. The dimension of “home” that mattered depended on your subsistence system, in exactly the way the mechanism predicts.

Farmers, whose crops live and die by accumulated warmth and by the ground underfoot, matched most strongly on growing-season heat and on soil, and cared less about rainfall. Pastoralists, whose herds follow the grass and the grass follows the rain, matched most strongly on precipitation and less on temperature. Foragers, whose knowledge was tied to specific landscapes, paid the steepest penalty of all for moving anywhere unfamiliar. The travel-cost numbers line up with the same logic: each additional 100 km of terrain-adjusted distance, about 80 hours of walking, reduced pastoral migration by only 2 percent, against 7 percent for farmers and 13 percent for foragers. A herder’s wealth walks itself to new pasture. A farmer’s wealth is cleared fields and improved soil that he cannot carry. The steppe cultures bear this out directly: the Yamnaya, mobile pastoralists, show little sensitivity to either travel cost or thermal matching, while farming cultures show both.

The authors make a sharp point about why this matters for belief. No single hidden variable can explain matching on all three dimensions at once, because temperature runs with latitude, rainfall with coastal proximity, and soil with geology. And any confounder that somehow did would still have to sort itself by whether the people involved were farmers, herders, or foragers. That is a hard thing to fake.

Finding three: climate change moved people, on schedule

Then they let the climate change over time and watch what the migrations do. Warming widens the zone where cereals will grow, so farmers and foragers spread out as more of the map becomes viable. Cooling shifts the balance toward grassland, so pastoralists spread as the steppe opens up. Same logic, opposite temperature sign.

When they feed realized Holocene temperatures into the model, it reproduces the actual archaeological record. Farmer migration peaks between 5500 and 4500 BCE, during the warm Holocene Climatic Optimum, the very window when Neolithic farming swept across central Europe. Pastoral migration climbs during the Neoglacial cooling after 3000 BCE, coinciding with the Yamnaya, Sintashta, and Scythian expansions that every archaeologist already knows.

A stripped-down model that knows nothing but temperature recovers the timing and direction of the great prehistoric migrations. That is the kind of result that makes you trust the mechanism underneath it.

And people adapted by moving

The last piece is the one worth sitting with. Migration was not only pushed by climate. It was a way of coping with it. Populations hit by warming tended to relocate toward cooler ground, and those hit by cooling drifted warmer. The size is concrete: a 1°C warming at the origin is associated with descendants settling about 0.4°C cooler than their ancestors, offsetting roughly 40 percent of the local temperature change over the course of centuries. When the climate shifted, people followed the conditions they knew, and in doing so canceled out a good share of the shift.

They go one step further and ask whether the migrants remade the places they arrived. Using fossil pollen, they find, cautiously, that destinations receiving lots of newcomers from a given origin drifted toward that origin’s vegetation, consistent with farmers bringing their crops and clearing with them. Along the way the analysis recovers a familiar Holocene pattern, the one where warming favors tree cover and cropland at the expense of open grassland.

The caveats, which they supply themselves

I will flag the limits, though the authors beat me to most of them, which is itself a mark of the paper’s temperament. It is a working paper, not yet peer-reviewed, and it says so on the front page. The coverage is Western Eurasia, because that is where the ancient DNA is; whether the patterns hold on other continents is an open question that better data will settle. The adaptation result may partly reflect survivorship, since we only sequence the lineages that made it into the ground and back out again, and successful movers may be overrepresented. The vegetation-feedback finding, they are careful to say, is suggestive rather than causal. This is what careful looks like. They are not overselling, and neither will I.

What you do with it is up to you

I said at the top I would report it and let you draw your own conclusions, so I will keep my thumb off the scale. But a few things in this paper are simply facts about the human past, and they are worth stating plainly.

Climate has always changed, long before anyone was burning anything, swinging from the warmth of the Holocene Optimum to the chill of the Neoglacial. Those warm stretches were not eras of collapse; they were when farming spread across a continent and when forests advanced on grassland. And for 10,000 years, the human answer to a shifting climate was not to perish in place. It was to adapt, most often by moving, and in moving to cancel out a large fraction of the change. The authors close by noting that understanding how people historically adapted through migration can inform how societies respond to future climate change. That is as far as they push it, and it is a good deal more measured than most of what gets written on the subject.