Plague is usually told as a story about cities. Crowding, granaries, rats, fleas, trade routes: the standard account holds that Yersinia pestis needed dense sedentary populations before it could cause epidemics, and that the Neolithic agricultural transition was the precondition for everything that followed.
A 2026 study reports plague outbreaks that break every part of that framing. They occurred among mid-Holocene hunter-gatherers near Lake Baikal in southeast Siberia, beginning about 5,500 years ago, in communities with no agriculture, no cities and no commensal rats. And they were lethal — not the mild, ambiguous infections that early Y. pestis strains had been assumed to cause.
The evidence comes from the Nature paper Lethal plague outbreaks in Lake Baikal hunter-gatherers 5,500 years ago, reporting infections across four hunter-gatherer cemeteries with a 39% detection rate.
The short answer: early plague strains recovered from four Siberian cemeteries document two phases of outbreak from about 5,500 years ago, with plague DNA detected in 39% of tested individuals. Reconstructed pedigrees show small family groups affected in patterns consistent with human-to-human spread, and the first outbreak unfolded within a single generation. Mortality was acute, especially among children aged 8 to 11. The strains diverge ancestrally to known Y. pestis, placing its emergence before roughly 5,700 years ago.
The problem these strains were supposed to have#
Yersinia pestis has been recovered from Eurasian skeletons for the better part of a decade, going back to the Late Neolithic and Bronze Age. Those early lineages, sometimes grouped as the LNBA strains, are missing genetic components that the historically documented bubonic form depends on — most importantly the ymt gene, which allows the bacterium to survive in a flea's gut and be transmitted by flea bite. That capability appears around 3,800 years ago.
That left an unresolved question. Early plague existed; it was in people; but without flea transmission, how sick did it actually make them, and could it spread far enough to matter? A plausible reading was that these were sporadic, largely dead-end infections — the bacterium present but not yet epidemic.
The Baikal study is the first to answer with a population rather than with scattered positives.
What "39% detection rate" means#
Across four Late Neolithic cemeteries near Lake Baikal, 18 individuals tested positive for Y. pestis, giving a detection rate of 39% among those screened.
That number is extraordinary in context. Ancient pathogen work usually reports single-digit percentages, because a pathogen must be in the blood at the time of death, its DNA must survive burial, and the sample must be tested. A 39% rate implies something close to the ceiling of what the method can detect — consistent with a substantial fraction of a community dying while actively infected.
The word to avoid is "mortality rate". The figure describes detection in tested skeletons, not the proportion of a living population that died. But it is difficult to reconcile with occasional isolated infections, and that is the point.
Pedigrees turn a cemetery into an epidemiology#
The methodological core of the study is that it does not treat the burials as a list of individuals. It reconstructs kinship pedigrees from the human genomes and then maps the infections onto them.
That converts a set of positive results into something an epidemiologist can read. The pattern that emerges is of small familial groups affected together, which is what human-to-human transmission looks like — an infection moving through the people who shared shelter, food and care.
Two further results follow from the same analysis:
- The first outbreak occurred within a single generation. This was an event, not a slow background presence.
- Mortality was concentrated among children aged 8 to 11. Age-structured mortality is itself evidence of an acute infectious process rather than of chronic illness or ordinary attrition.
Reconstructing relatedness from ancient genomes is now routine — it underpins work such as the Avar family networks and the kinship results from Bronze Age Aegean burials. Applying it to a pathogen dataset is what lets this study argue about transmission rather than merely about presence.
Two phases, and a superantigen#
The outbreaks fall into two distinct phases, separated in time and represented by strains that differ genetically. Some of those differences are functional, and the one the authors highlight is at the ypm superantigen locus — a gene also present in Yersinia pseudotuberculosis, the less dangerous relative from which Y. pestis descends.
Superantigens provoke a massive, poorly targeted immune response. The presence of a functional ypm locus in these early strains raises the possibility that they harmed people through a mechanism different from the one that made later bubonic plague so lethal — a different route to the same outcome, in a bacterium that had not yet acquired the flea-borne toolkit.
That is a hypothesis grounded in gene content rather than a demonstrated mechanism. Nobody can measure symptoms in a person who died five and a half thousand years ago.
Pushing the origin back#
Phylogenetically, the Baikal strains diverge ancestrally to known Y. pestis: they branch off before the lineages recovered elsewhere. That position constrains the timing of the bacterium's emergence, indicating that it had already separated from its ancestor before roughly 5,700 years ago.
| Milestone | Approximate date |
|---|---|
| Y. pestis emerges as a distinct lineage | before ~5,700 years ago |
| Lake Baikal outbreaks, phase one and two | from ~5,500 years ago |
| Flea transmission (ymt) acquired | ~3,800 years ago |
| Justinianic Plague | 541 CE |
| Black Death | 1346–1353 CE |
The gap between the first and third rows is the interesting part. For nearly two thousand years Y. pestis circulated in human populations without the adaptation usually credited with making it epidemic — and, on this evidence, killed people anyway.
Why hunter-gatherers matter to the argument#
The Neolithic hypothesis for plague is intuitive: settle down, crowd together, store grain, attract rodents, and epidemic disease follows. Ancient plague genomes from Neolithic Europe fitted that story comfortably enough that it rarely needed defending.
Lake Baikal does not fit it. These were mobile hunter-fisher-gatherer communities, well outside the sphere of Late Neolithic Europe, without agriculture or permanent dense settlement. The authors' conclusion is direct: higher population densities and the lifestyle changes of the agricultural transition were not prerequisites for plague epidemics.
There is a broader lesson in that, and the pre-contact leprosy work in the Americas makes the same one from a different continent. Assumptions about which diseases existed where, and under what social conditions, have repeatedly been overturned once anyone screened the skeletons instead of reasoning from first principles.
Limitations#
| Limitation | Why it matters |
|---|---|
| Detection rate is not mortality | 39% describes positive tests among screened individuals, not deaths in a living population. |
| Cause of death is inferred | Pathogen DNA shows infection at death, not that the infection was fatal. |
| Ancient genomes are partial | Reconstructed bacterial genomes may miss genes present in the living strain. |
| Gene content is not symptoms | The ypm locus suggests a mechanism; it does not demonstrate one. |
| Four cemeteries, one region | Whether such outbreaks were widespread elsewhere is untested. |
| Screening is uneven globally | Absence of early plague elsewhere may reflect absence of testing. |
Frequently asked questions about the Lake Baikal plague outbreaks#
How old are these plague strains?#
The outbreaks begin about 5,500 years ago, and the strains' phylogenetic position indicates that Yersinia pestis had emerged as a distinct lineage before roughly 5,700 years ago.
How many people were infected?#
Eighteen individuals across four Late Neolithic cemeteries tested positive, a 39% detection rate among those screened — very high for ancient pathogen work.
Was this bubonic plague?#
No. These strains lack the genetic components required for flea-borne transmission, which appear around 3,800 years ago. They were lethal by some other route, possibly involving the ypm superantigen locus the study highlights.
How do we know it spread between people?#
Kinship pedigrees reconstructed from the human genomes show infections clustering within small family groups, and the first outbreak unfolding inside a single generation. That pattern is consistent with person-to-person transmission rather than repeated independent infections from an animal source.
Why is it significant that these were hunter-gatherers?#
Because the standard account holds that plague epidemics required the population densities and animal contact that farming brought. These communities had neither, which means the preconditions were less restrictive than assumed.
Does this affect plague risk today?#
No. Y. pestis still exists in rodent populations in several parts of the world and causes a small number of human cases each year, all treatable with antibiotics. This study concerns the deep history of the bacterium and has no bearing on present-day risk or treatment.
A pathogen older than the conditions it supposedly needed#
Ancient pathogen genomics has spent a decade dating diseases earlier than expected. Plague in Bronze Age Eurasia, tuberculosis in the pre-contact Americas, hepatitis B across prehistoric Europe: in each case the organism turned out to have been present long before the historical record noticed it.
The Baikal study goes further, because it is not only about presence. By combining pathogen genomes with human pedigrees, it argues about transmission and consequence — who infected whom, over what interval, and who died. That is a genuinely different class of claim, and it is what allows the paper to say that early plague was not merely present but lethal.
The implication for the standard narrative is uncomfortable in a useful way. If a mobile hunter-gatherer population in Siberia could suffer a plague outbreak that killed a substantial share of a community within one generation, then the link between farming, density and epidemic disease is weaker than the textbooks have it. The Neolithic may have amplified what followed. It did not make it possible.
Sources and further reading#
- Macleod, R., Seersholm, F. V., De Sanctis, B. et al. (2026). Lethal plague outbreaks in Lake Baikal hunter-gatherers 5,500 years ago. Nature 654, 697–705. DOI: 10.1038/s41586-026-10540-5.
- Rascovan, N., Sjögren, K.-G., Kristiansen, K. et al. (2019). Emergence and spread of basal lineages of Yersinia pestis during the Neolithic decline. Cell 176, 295–305.
- Andrades Valtueña, A., Neumann, G. U., Spyrou, M. A. et al. (2022). Stone Age Yersinia pestis genomes shed light on the early evolution, diversity, and ecology of plague. PNAS 119, e2116722119.
Editorial note: this article was written as a source-based synthesis and distinguishes pathogen detection from cause of death throughout. Its hero and section artwork was generated with AI as an interpretive archaeological scene, not as scientific evidence.
