At a glance:
- Applying new analytic methods to nearly 16,000 ancient genomes reveals natural selection has acted on hundreds, not dozens, of genes in West Eurasia over the last 10,000 years.
- More than half of the genes have known links to disease risk and other traits today, although it's not yet clear what made each gene advantageous in prehistoric contexts.
- The work demonstrates the power of ancient DNA to illuminate human biology and medicine in addition to history.
A massive study of ancient DNA from nearly 16,000 people across more than 10,000 years in West Eurasia reveals that natural selection has shaped modern human genomes far more than previously thought.Before now, studies of ancient human DNA had identified only about 21 instances of directional selection — the type of natural selection that occurs when one version of a gene that confers an extreme form of a trait, such as lactose tolerance after infancy, proves advantageous enough for survival and reproduction that it gets passed on to more offspring than less advantageous versions of the gene and rapidly rises in frequency across a population. The dearth of evidence suggested that directional selection has been rare since modern humans arose in Africa some 300,000 years ago and began to split into different population groups around the world.
Combining an unprecedented amount of ancient genomic data with novel computational methods, the new analysis shows instead that directional selection has driven the spread or decline of hundreds of gene variants in West Eurasia since the end of the Ice Age and that selection has actually accelerated since people transitioned from hunting and gathering to farming.
The work demonstrates the power of ancient-DNA research to illuminate human genetic adaptation and other fundamental principles of evolutionary biology.
Many of the identified gene variants have known links to complex physical, psychological, and social traits, including risk for type 2 diabetes and schizophrenia. Delving into the evolution of these traits could deepen understanding of behavior, health, and disease and inform treatment efforts. However, the way we define some of the traits today, such as household income, doesn't translate to prehistoric contexts, and the current analysis can't speak to what made a variant beneficial for survival when it first arose.
The findings, led by Harvard University researchers, are published April 15 in Nature .
"With these new techniques and large amount of ancient genomic data, we can now watch how selection shaped biology in real time," said Ali Akbari , first author of the study and senior staff scientist in the lab of Harvard geneticist David Reich. "Instead of searching for the scars natural selection leaves in present-day genomes using simple models and assumptions, we can let the data speak for itself."
"This work allows us to assign place and time to forces that shaped us," said Reich , professor of genetics in the Blavatnik Institute at Harvard Medical School, professor of human evolutionary biology in the Harvard University Faculty of Arts and Sciences, and senior author of the study.
10,000 ancient genomes, new computational methods
Since 2010, when the first genome-wide data was recovered from ancient human remains, ancient-DNA research has expanded understanding of the relationships among people living in different time periods and regions of the world.
But geneticists struggled to realize the technology's promise to illuminate how natural selection has shaped human genetic variation even over the last 10,000 years, when there is enough well-preserved genetic material to support large-scale studies.
The new study broke through that barrier using two innovations.
First, the Reich Lab spent seven years building a collection of DNA sequences from ancient people living in West Eurasia — what is now Europe and parts of the Middle East — that would be comprehensive enough in size and time span to support the work.
"If the goal is to uncover changes in the frequency of genetic variants in the last ten millennia that are greater than can be expected by chance, then we need to detect subtle effects, which requires having thousands of genomes spanning that time period," explained Reich, who is also a member of the Broad Institute of MIT and Harvard and a Howard Hughes Medical Institute Investigator.
The lab collaborated with more than 250 archeologists and anthropologists to report new DNA data from 10,016 ancient individuals from West Eurasia. They added those to another 5,820 published ancient sequences and 6,438 modern ones.
"This single paper doubles the size of the ancient human DNA literature," Reich said. "It reflects a focused effort to fill in holes that limited the power of previous studies to detect selection."
The regions from which ancient and recent human DNA samples were studied in this work. Image: Akbari A et al., "Ancient DNA reveals pervasive directional selection across West Eurasia," Nature (2026)
Alt text: A map of Europe and western Asia, with regions marked in five colors. Countries span Iceland and Russia in the north to Spain and Iran in the south.
The second innovation — and even more important to the success of the study, Reich said — was Akbari's development of computational methods to isolate the signal of directional selection from other causes of gene frequency changes, such as human migration, population mixing, and random genetic fluctuations that occur in small populations.
"Ali developed a powerful technique that could zoom in on the patterns that actually mattered," said Reich.
In the end, it was a faint signal indeed that Akbari detected. By the team's calculations, directional selection accounted for only about 2 percent of all gene frequency changes.
What has natural selection selected for?
Two percent still encompasses a lot of DNA. Akbari identified 479 gene versions, or alleles, that were strongly selected for — or against — in West Eurasian genomes.
He and colleagues were able to ascertain when and where some of the alleles began to spread through or be pushed out of the West Eurasian gene pool. They also calculated an overall rate at which selection seemed to occur and detected changes in that rate. They found that selection accelerated after the introduction of farming, reflecting how different traits became advantageous as people shifted to agricultural environments and behaviors.
More than 60 percent of the individual DNA variants that were flagged as being strongly selected for — most of them single nucleotide polymorphisms, or SNPs — have documented links with present-day human traits, such as:
- Light skin tone
- Red hair
- Risk of celiac disease and Crohn's disease
- Immunity to HIV infection and resistance to leprosy
- Lower chance of male-pattern baldness
- Lower risk of rheumatoid arthritis and alcoholism
- Having the B version of the proteins on red blood cells that confer A, B, and O blood types and influence resistance to infection with bacteria and viruses
In some cases, groups of SNPs were under selection together to influence polygenic traits. Some changes raised the frequency of beneficial traits, including some that are interpreted today as:
- "Health span" traits such as faster walking pace
- Measures of behavioral and social status or cognitive functions, such as scores on intelligence tests, household income, and years of schooling
Other changes reduced the frequency of harmful traits, such as those that are interpreted today as:
- Reduced risk of bipolar disorder and schizophrenia
- Lower body fat percentage, waist-to-hip ratio, and body mass index
- Less susceptibility to tobacco smoking
Still other SNPs, such as some that today are associated with susceptibility to tuberculosis and multiple sclerosis, at first rose and then fell in frequency over the millennia, indicating shifts in environmental pressures and the traits that prove beneficial, the team found.
Some of the links seem logical, others counterintuitive, like the major genetic risk factor for gluten intolerance spiking after people began farming wheat.
However, the authors emphasize that there are several crucial factors to understand before interpreting SNP associations like these.
First: What a variant is associated with now is not necessarily why an allele propagated in the West Eurasian gene pool. Reasons for this include:
- Some of the traits that SNPs are associated with in modern societies did not exist in ancient contexts and therefore can't explain why an allele was originally advantageous or detrimental. A variant that now correlates to household income or years of schooling had to have meant something different in the Stone Age. So these results do not mean that Europeans evolved to be smarter or healthier.
- The fact that an allele shapes a particular trait today also does not automatically mean this trait was important in the past. Perhaps having red hair was beneficial 4,000 years ago, or perhaps it came along for the ride with a more important trait.
- Some SNPs affect multiple traits, so what a genomic database tags a SNP as affecting may not capture everything it's doing. Today, for instance, we know that the same gene variant that raises risk of sickle cell disease also protects people from malaria, so what looks like natural selection for one disease may be selection against another.
- It's possible that a flagged SNP is actually in a gene next to the one that natural selection was targeting — another way of coming along for the ride.
- Present-day traits a SNP influences may not yet be known or included in the databases the team analyzed.
Second: Just because an allele, SNP, or trait swept into or out of West Eurasia during this time doesn't mean this happened only in West Eurasia. Researchers can use the new computational methods to look for directional selection in other populations worldwide that have enough ancient DNA sequences and construct a clearer picture of what's unique to different groups and what generalizes across populations.
Reich expects that future studies will show that shared selective pressures acted on some of the same core traits across diverse human groups, even as those groups split off and migrated to different parts of the world over tens of thousands of years.
What comes next
The team has made its data and methods freely available to spur new research.
One avenue is to investigate other possible signals in the data. Akbari said he and colleagues identified more than 7,600 genetic locations that have better than a 50/50 chance of "being real examples of directional selection" and warrant follow-up.
Using the new methods to explore other groups and further back in time are the most exciting avenues for Reich.
"To what extent will we see similar patterns in East Asia or East Africa or Native Americans in Mesoamerica and the central Andes?" he asked. "If we can't use ancient DNA to study the most important period in human evolution 1 million to 2 million years ago, then at least we can study selective pressure on human genomes during more recent periods of change and learn broader principles."
It will also be crucial for scientists to conduct molecular studies to better understand the health consequences of selected alleles.
It's possible the results could point scientists to new genetic factors in health and disease that improve experts' ability to assess disease risk, prevent illness, and develop new medicines. Researchers developing gene therapies might consider whether the gene they're targeting was flagged in the study as being advantageous, Akbari said.
"You could speculate that if the variant someone wants to knock out was strongly selected for, it's probably not the best idea," he said.
Scientists could also use the new methods to study natural selection in other species. Such work could uncover alleles that have made cattle or chickens well-suited to domestication, Akbari suggested, or that have helped animals adapt to changes in climate.
The possibilities are enticing for deepening our appreciation of human diversity, history, and health, Reich said.
"This paper shows how complex selection can be and provides an opportunity to consider the richness of variation in human populations," he said.
Authorship, funding, disclosures
Additional authors are Annabel Perry, Alison R. Barton, Mohammadreza Kariminejad, Steven Gazal, Zheng Li, Yating Zeng, Alissa Mittnik, Nick Patterson, Matthew Mah, Xiang Zhou, Alkes L. Price, Eric S. Lander, Ron Pinhasi, Nadin Rohland, and Swapan Mallick.
This research was supported by the John Templeton Foundation (grant 61220), the Allen Discovery Center for Human Brain Evolution (a Paul G. Allen Frontiers Group advised program of the Allen Family Philanthropies), the Howard Hughes Medical Institute, the National Institutes of Health (grant HG012287), a private gift from Jean-François Clin, and the European Research Council (grant 834087, COMMIOS). The research was conducted using the UK Biobank resource under Application 16549. The authors also acknowledge support from the Research Computing Group at HMS.
