Plague was already striking Siberian families 5,500 years ago

Ancient DNA from teeth at four Lake Baikal cemeteries carried Yersinia pestis in 18 of 46 people — the oldest directly dated plague yet found, in mobile hunter-gatherers long before cities or farming. An unusual spike of child deaths packed into a few decades is strong circumstantial evidence that these early, flea-less strains were lethal. But the study can't prove plague caused each death, and its own test for whether infection clustered within families was not significant.

12 October 2026 · How we rate evidence

In a grave at a site called Bratskii Kamen, on the bank of the Angara River in southeast Siberia, three little girls were buried together. They were between four and nine years old. Their DNA shows they were close relatives on their mothers' side — two of them most likely cousins. And all three carried Yersinia pestis, the bacterium that causes plague. They seem to have died at about the same time, more than 5,000 years ago.

The claim going around. When most people picture plague, they picture the Black Death: medieval cities, rats, fleas, mass graves. So finding it in small, mobile hunter-gatherer bands thousands of years earlier is surprising. It was widely covered when it appeared in Nature in June 2026, and again in September, when a University of Copenhagen press release recirculated under the headline "Plague was killing entire families thousands of years before the Black Death." The releases said nearly 40% of the people tested carried plague — "a detection rate even higher than that reported at some medieval plague burial sites" — and that the strains carried a toxin-making genetic factor that could make infection "much more dangerous."

What they did. An international team led by Ruairidh Macleod and Eske Willerslev extracted ancient DNA from the teeth of 46 people buried at four hunter-gatherer cemeteries near Lake Baikal, then screened it for pathogens. Y. pestis turned up more than anything else — in 18 of the 46. Radiocarbon dating sorted the cases into two waves: an early one around 5,500 to 5,300 years ago, and a second four to six centuries later. At Ust'-Ida I, the largest cemetery, the dates cluster so tightly and the family relationships line up so neatly that the researchers argue most of those people died within a single generation.

What they found. The 18 cases were spread across the four sites, with most at Ust'-Ida I.

The most striking pattern is who died. At the two cemeteries with the most cases, Ust'-Ida I and Bratskii Kamen, deaths peak sharply among children aged roughly 7 to 11 — a spike that stands out as statistically unusual against seven other Baikal hunter-gatherer cemeteries. Between 65% and 75% of the people buried at those two sites were children under 15. Adults in their prime are oddly scarce, and parents almost absent from the family trees; the sexes died in roughly equal numbers. The strains themselves are the oldest directly dated plague found so far, and sit at the very base of the plague family tree — ancestral to all later strains, including the medieval ones. The groundwork is careful: kinship calls were cross-checked, and the data and code are public.

What it doesn't show. The plague DNA is real; that plague killed these people is an inference, and a few gaps sit under it. Detecting the bacterium in someone who died is not the same as proving it killed them, and not everyone at these sites tested positive — though, as the lesson box notes, many infected people would leave no trace. The child-mortality spike is genuine and unusual, but the study cannot rule out that an outbreak simply coincided with a cemetery already skewed toward children.

The transmission story is thinner than the headlines suggest. The image of plague "killing entire families" rests on co-buried relatives who both tested positive — the three girls, an aunt and her nephew. But when the authors ran the formal test — whether infection clustered among relatives more than chance would produce — the answer was no: six related pairs were both positive, but the result was not statistically significant. These early strains also lack ymt, a gene later plague needed to spread by flea bite, so any human-to-human spread would have been by coughing or contaminated food — plausible, but inferred, not observed.

The "much more dangerous" toxin is a hypothesis, not a result. The strains carry ypm, a superantigen gene that in living relatives of plague can provoke violent immune reactions, and it resembles the most virulent known form. That is a suggestive clue to why children might have suffered most, but the authors say its actual effect "requires functional studies." The paper is careful here; the release was not. The emergence date — plague splitting from its ancestor before about 5,700 years ago — is a model estimate the authors call a lower bound. How firmly to rate the "oldest, most basal" finding turns on which half you weigh: the directly dated, ancestral strains rest on hard measurements, while the exact age plague split off is only a model estimate built on very few genomes.

What it means beyond the study. The human thread here is not medieval history but where deadly new diseases come from. Plague still spills into people in this very region today, from marmots — large burrowing rodents the Baikal groups hunted. The likeliest reading is that these outbreaks began as spillovers from a wild rodent reservoir. What makes that striking is the setting: no cities, no farming, no crowding — the conditions usually blamed for turning a microbe into an epidemic. A deadly zoonosis was already possible in small, mobile bands, a reminder as the world worries about climate change reshuffling which animals meet which people. An independent team, weeks later, reported a roughly contemporary East-Asian plague genome and a rapid trans-Eurasian spread — pointing, like this work, to an eastern origin.

Why it's still interesting. For years, researchers debated whether the earliest plague strains were dangerous at all; some argued they caused little more than a mild gut infection. This study is the first to set the genetic evidence beside the graves — who died, when, and how they were related — and argue that even these primitive, flea-less strains were lethal, and hardest on children. Whether or not every link holds, that combination is a new way to read an ancient epidemic, and a reminder that a disease does not need a crowded city to kill.

Open questions.

Did plague cause the spike in child deaths, or merely coincide with it? Testing it means looking for plague DNA specifically in the child skeletons that drive the mortality peak, at more cemeteries with and without such peaks. It matters because the whole "lethal outbreak" reading turns on it: for now the disease and the deaths are associated, not shown to be cause and effect.

How did it spread — a fresh spillover each time, or person to person? Denser sampling could show whether cases form tight genetic chains within families, as human-to-human transmission would predict; the current family-clustering test was too small to settle it. The answer would tell us whether small, mobile groups could sustain an outbreak at all, or whether every case traces back to a hunted animal.

Did the superantigen actually make these strains deadlier to children? That is an experiment: reconstruct the ancient toxin and test how human immune cells respond to it. It would separate a real virulence factor from a genetic coincidence — and sharpen how confidently anyone can read "dangerous" off an ancient genome.

A note on timing. The study appeared in Nature on 17 June 2026 and was covered widely then. The ScienceDaily version this article addresses is a recirculation of the press release on 14 September, with no change to the science.