When humans started farming around 10,000 years ago, they reshaped their landscapes, diets, and social structures. A common assumption in evolutionary biology held that those cultural changes acted as a buffer, shielding human biology from the pressures of natural selection. A sweeping ancient-DNA study published in Nature in May 2026 argues the opposite happened: selection did not ease off. It intensified, leaving its mark on many hundreds of genetic variants scattered across the genome.
The largest ancient-DNA dataset yet assembled
A team led by David Reich at Harvard Medical School analyzed genomes from 15,836 ancient people who lived across West Eurasia, the broad region stretching from Ireland to Iran. According to the paper, 10,016 of those are newly reported individuals whose sequence data have been deposited in the European Nucleotide Archive. The full analytical package, including imputed genomes and supplementary tables, is archived in the Harvard Dataverse.
That sample size matters because earlier ancient-DNA studies, while groundbreaking, lacked the statistical power to separate real selection from background noise. A 2022 analysis of Black Death survivors, for example, showed that plague-resistance alleles shifted measurably in just a few generations. But West Eurasia also experienced some of the most dramatic population turnovers in human prehistory: the spread of Anatolian farmers beginning roughly 8,000 years ago and the later expansion of Steppe pastoralists around 5,000 years ago. When one group replaces another, allele frequencies shift for reasons that have nothing to do with fitness. Without careful controls, migration can masquerade as natural selection.
A new statistical framework
The core innovation in the paper is a method that models migration, admixture, and genetic drift explicitly, then tests whether observed allele-frequency changes over time exceed what those neutral forces alone would predict. A 2024 study in Nature Communications by Mathieson and colleagues laid out the conceptual case for why such strict demographic controls are essential when hunting for selection signals in ancient European DNA. The new paper scales that logic to a dataset roughly ten times larger.
“Previous methods could not reliably distinguish selection from the massive population movements that characterize European prehistory,” David Reich and co-authors write in the paper. By accounting for those movements directly, the team recovered “many hundreds of alleles” under strong directional selection, a count far exceeding earlier estimates that typically identified only a handful of clear targets.
Iain Mathieson, a population geneticist at the University of Pennsylvania who was not involved in the study, told Nature News that the new framework represents “a real step forward” in disentangling selection from demographic history, though he noted that independent replication on separate datasets will be important for confirming the scope of the signals.
What the signals point to, and what remains unclear
The study does not claim to have pinpointed every gene’s function. An interactive web application hosted by the Reich Lab displays allele-frequency trajectories and selection-coefficient summaries for individual loci, giving outside researchers a starting point for deeper investigation. But connecting a statistical signal to a specific adaptation, whether related to lactose tolerance, immune defense, skin pigmentation, or metabolic efficiency, requires functional work that goes beyond what the paper presents.
There is also the question of intensity. Detecting selection on hundreds of alleles does not reveal how strong each episode was. Selection coefficients can range from barely detectable to fierce, and the distribution across those loci has not been broken down in detail in publicly available summaries. The difference between weak, widespread nudges and a few bursts of intense pressure carries very different implications for how rapidly human populations can reshape themselves biologically.
Geographic scope is another limitation the authors acknowledge. The dataset covers West Eurasia exclusively. Whether the same acceleration pattern holds in East Asia, sub-Saharan Africa, or the Americas is an open question. The Allen Ancient DNA Resource, maintained by the same lab, curates genotype data from a broader global set of ancient and present-day populations, but the 2026 analysis did not extend its detection framework beyond Europe and western Asia.
Some archaeological metadata associated with the dataset is listed as embargoed in the Harvard Dataverse, which means independent teams cannot yet cross-reference selection signals against specific burial contexts, associated artifacts, or radiocarbon dates. That gap may narrow as embargoes lift, but for now it limits how far anyone can push the interpretive story.
Why independent replication matters
The findings rest, for now, on a single team’s analysis of a single dataset, however large. That is standard for a newly published study, and the open data deposits are designed to invite exactly the kind of outside scrutiny that will test whether the signals hold up under alternative demographic models. The peer-reviewed Nature paper, with its detailed methods and supplementary materials, is the primary evidence. Accompanying coverage in Nature News and institutional outlets like Phys.org provides useful framing but draws from the same underlying work and should not be mistaken for independent verification.
Independent population geneticists will likely focus on two questions in the coming months: whether the method’s demographic corrections are conservative enough to avoid false positives, and whether the selection signals replicate when applied to ancient-DNA datasets from other labs and other regions.
What accelerating selection means for modern health research
If the method survives that scrutiny, the implications stretch well beyond prehistory. Understanding which genes were under recent, strong selection can illuminate why certain disease risks vary across populations today. It can also inform how quickly human biology responds to new environmental pressures, a question with relevance to everything from emerging infectious diseases to rapid dietary shifts in industrializing societies.
For now, the study’s clearest contribution is methodological. The tools for detecting selection in ancient DNA have matured to the point where hundreds of signals can be pulled from a single well-powered dataset. That alone changes the scale of the conversation about recent human evolution, turning a field that once debated whether post-agricultural selection was even detectable into one that must now explain why it was so pervasive.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
