Before It Could Ride Fleas: Siberian Teeth Catch the Plague at an Earlier Stage
Some 5,500 years ago, on the banks of a river that feeds Lake Baikal in southeast Siberia, small bands of hunter-gatherers began burying their dead in unusual numbers, and the dead were unusually young. Three girls were laid in a single grave. Elsewhere an aunt and her nephew were interred together. To an archaeologist, a cluster of children and close kin dying at once is the signature of something moving fast through a community, faster than accident or age can explain. The teeth of those dead have now named the culprit, and it is an old and familiar one: Yersinia pestis, the bacterium behind the plague, caught long before it became the killer of medieval cities.
The DNA pulled from those teeth pushes the earliest documented plague outbreak back to roughly 5,500 years ago, about two centuries earlier than the previous record. The work was published in Nature on 18 June 2026 by a team led by Ruairidh Macleod, a postdoctoral fellow at All Souls College, University of Oxford, with the ancient-DNA group of Eske Willerslev. What makes the strain remarkable is not just its age. It is where it sits on the family tree of the plague, and what it could not yet do.
A plague that could not ride a flea
Modern plague is, above all, a disease of fleas. When a flea bites an infected rodent, the bacterium forms a sticky blockage in the insect's gut; the starving flea bites again and again, regurgitating bacteria into each new host. That mechanism — the engine of the bubonic plague that killed tens of millions during the Black Death — depends critically on the ymt gene, which lets Y. pestis survive the hostile interior of a flea.
The Lake Baikal strain does not have it. Sequencing the ancient genomes, the researchers found bacteria that "sit on a very early branch of the Y. pestis family tree" and "diverge before all previously known Y. pestis diversity," as they describe it, a lineage older and more basal than anything sampled before. It lacks ymt and the other virulence factors that later made the bubonic form possible. Those adaptations, the wider genetic record suggests, did not arrive until roughly 3,800 years ago. In other words, this is plague from before it learned to travel by insect, a snapshot of the pathogen at an earlier rung of its long climb toward becoming a pandemic disease. The find even pushes back, by around 2,000 years, the estimated split between Y. pestis and its milder ancestor Y. pseudotuberculosis.
So how did it spread at all? The most likely reservoir, the team argues, was the marmot, a large burrowing rodent that still carries plague around Lake Baikal today. People probably picked it up while butchering the animals, through raw organs or contact with infected hides, and then may have passed it on directly through close contact within a household. By reconstructing kinship from the genomes, the researchers showed the disease striking small family groups, with the first outbreak burning through the population within a single generation. That is the pattern you would expect from person-to-person spread, not from an insect vector fanning out across a region.
The evidence in the teeth
The signal is not faint. Of at least 46 children and adolescents examined from cemeteries near the Angara River, 18 tested positive for Y. pestis, about 39 percent of the graves sampled. The dead skewed young, with mortality falling hardest on children between roughly eight and eleven years old, according to an analysis by the Center for Infectious Disease Research and Policy at the University of Minnesota. The primary site, Ust'-Ida, sits on the Angara's banks; a second cluster of cemeteries farther upstream records what looks like a separate outbreak some 300 years later, suggesting the pathogen returned to the same landscape more than once.
That so lethal a disease could take hold in a mobile, low-density population of hunter-gatherers is the part that unsettles the researchers most. Epidemiological theory tends to hold that crowd diseases need crowds: dense, settled populations of the kind that agriculture and cities later produced. "The fact that we're finding this happening in an isolated group of prehistoric hunter-gatherers is really, really extraordinary to me, and challenges a lot of that epidemiological theory," Macleod said. The plague, on this evidence, was killing people in small numbers long before there were cities for it to empty.
What it does and does not tell us
The study is peer-reviewed, published in Nature, and rests on genomes recovered directly from the ancient remains rather than on inference from later strains, a comparatively firm footing for a claim about deep time. (A related version by the same team had circulated earlier as a bioRxiv preprint; the Nature paper is the reviewed record.) Ancient DNA of this age is always fragmentary, and reconstructing how a pathogen behaved millennia ago from the genes it left behind involves careful interpretation. But the central results (the age, the position on the evolutionary tree, the absent ymt gene) are the kind of concrete, testable findings that ancient genomics does well.
What emerges is less a story about a single outbreak than about a beginning. Long before the Justinian plague, before the Black Death, before rats and grain stores and trade routes gave the disease its highways, Yersinia pestis was already lethal to human beings. At this stage it was working without the flea it would later come to depend on. The teeth of a few Siberian children have caught it in the act of becoming what it would eventually be.