The research team, led by Matthew M. Carter, Zhiru Liu and Matthew R. Olm of Stanford University and the University of Colorado Boulder, together with Benjamin H. Good and Justin L. Sonnenburg, faced a formidable technical challenge. Bacterial genomes are constantly reshuffled by homologous recombination, a process in which strains exchange stretches of DNA with one another. This genetic mixing produces mosaic patterns of ancestry that cannot be captured by a single phylogenetic tree, which is why conventional evolutionary methods often fail when applied at the strain level. To get around this problem, the researchers developed population genetic analyses that explicitly account for recombination, allowing them to distinguish ancient co-migration from more recent microbial exchange driven by geography or lifestyle.

The study began with an extraordinary sequencing effort. The team performed deep metagenomic sequencing on 133 fecal samples collected from 85 Tsimane adults in 2009 and between 2012 and 2013, reaching a median depth of 31.9 gigabases per sample. From these data they recovered 12,746 metagenome-assembled genomes, or MAGs, representing 1,408 distinct bacterial and archaeal species. They then compared this collection with an analogous set of 32,034 MAGs previously obtained from 137 Hadza hunter-gatherers in Tanzania. The result was striking: 87.4 percent of the species found in the Tsimane cohort, a total of 1,231 species, were also present in the Hadza, despite the enormous geographic distance and the absence of any historical contact between the two groups.

Most of these shared species are not cosmopolitan gut dwellers. Roughly 60 percent of them, some 848 species, are rare in or completely absent from the fecal microbiomes of industrialized populations. A few, such as the starch-degrading bacterium Ruminococcus bromii, are found worldwide, but the majority belong to a category of microbes that previous work has labeled VANISH taxa, meaning they are volatile in prevalence or negatively associated with industrialized societies. The new study provides the strongest evidence yet that these vanishing species were not recent acquisitions by non-industrialized peoples but rather ancient companions of the human lineage that have been lost in the West.

To establish the antiquity of this partnership, the researchers calculated the average nucleotide identity, or ANI, between every pair of genomes within each of 636 species that had at least four MAGs in both populations. Because gut bacteria accumulate mutations at a rate of at least 10 to the power of minus 7 per site per year, the typical ANI of around 98 percent between Tsimane and Hadza genomes implies that they shared a common ancestor within the past 100,000 to 200,000 years. This timeframe places strong constraints on the evolutionary history of these microbes: the ancestors of the Tsimane did not acquire an entirely new suite of gut organisms during or after their migration into the Americas. Instead, they carried ancient African lineages with them.

The team then probed the high-similarity tail of the distribution, focusing on pairs of genomes sharing more than 99.75 percent average nucleotide identity. These nearly clonal strains contain large stretches of identical DNA that have not yet been overwritten by recombination, and their age can be estimated from the fraction of core genes with identical sequences. By calibrating this signal with known mutation rates, the researchers determined that genome pairs sharing more than 10 percent identical genes correspond to strains that diverged within roughly the past 5,000 years. Across all 636 species, 7.8 percent of genome pairs met this clonal criterion, but 93.8 percent of those pairs came from within the same human population. Moreover, 545 of the 636 species, or 85.6 percent, showed no recent strain sharing between the Tsimane and Hadza at all, indicating that very little microbial transmission has occurred between the two groups over the past five millennia.

There were telling exceptions. Species such as Akkermansia muciniphila and Bacteroides ovatus showed Tsimane-Hadza strain sharing rates comparable to their within-population baselines, and these same species tend to be enriched in industrialized populations. When the team expanded the analysis to include 8,855 additional MAGs from Europe, Asia and North America, these recently shared species turned out to have close relatives distributed globally, suggesting they were introduced into the Tsimane and Hadza through more recent contact with other human populations. The researchers also identified a third class of microbes, which they call StIL species, stable independent of lifestyle, that are prevalent across all populations yet still bear the genetic signature of ancient Tsimane-Hadza isolation, exemplified by common species such as Agathobacter rectalis and Ruminococcus bromii.

An independent line of evidence came from tracking horizontal gene transfer, the movement of DNA between bacterial strains. When two strains recombine, the transferred segment initially appears as a long tract of identical DNA that is gradually shortened by subsequent mutations, creating a predictable relationship between tract length and the age of the transfer event. The researchers measured the 99th percentile of identical tract lengths, called the L99 metric, for each species and found that between-population tracts were systematically shorter for the Tsimane and Hadza than within either population, a hallmark of genetic isolation. By contrast, the same comparisons between industrialized populations in Europe, Asia and North America showed much weaker barriers to gene flow, indicating that ongoing exchange in modern societies has blurred the historical signal.

Using these L99 lengths, the team estimated when the gut microbes of the two populations last exchanged DNA. The median split time for the Tsimane-Hadza comparison was 17,090 years, with a standard error of 863 years, dramatically older than the split times inferred between industrialized populations, which ranged from a median of 1,109 years for Europe-North America to 4,637 years for Asia-North America. A second, entirely orthogonal approach based on the distribution of single-nucleotide variant frequencies, applied through demographic modeling in 15 species, broadly confirmed these estimates. Notably, the inferred onset of microbial genetic isolation falls roughly within the window between the initial human migration out of Africa, around 60,000 to 70,000 years ago, and the settlement of the Americas, around 16,000 to 23,000 years ago, mirroring the timeline of human population splits.

The implications extend far beyond evolutionary history. The loss of VANISH taxa with industrialization has been linked to an increased risk of several chronic metabolic and inflammatory disorders, and the authors suggest that reintroducing these co-migrating species might one day help restore microbial functions that co-evolved with human biology. Such an endeavor, they caution, will require thoughtful and inclusive dialogue among researchers, ethicists and Indigenous communities, and the study itself was conducted under protocols approved by the Gran Consejo Tsimane and Bolivian institutions, with data governance guided by the CARE Principles for Indigenous Data Governance. As industrialized life continues to erode microbial diversity, the guts of the Tsimane and Hadza preserve a living archive of the ancient partnership between humans and their microbes, one that our species carried across continents and that modern societies are now quietly letting go.

Subject of Research: Prehistoric co-migration of human gut microbiome species traced through population genetics of the Tsimane and Hadza

Article Title: Prehistoric global migration of vanishing gut microbes with humans

Article References: Carter, M. M., Liu, Z., Olm, M. R., Martin, M., Sprockett, D. D., Ghadermazi, P., Trumble, B. C., Kaplan, H., Stieglitz, J., Rodriguez, D. E., Relman, D. A., Sonnenburg, E. D., Gurven, M., Good, B. H., & Sonnenburg, J. L. (2026). Prehistoric global migration of vanishing gut microbes with humans. Nature. https://doi.org/10.1038/s41586-026-11106-1

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11106-1

Keywords: gut microbiome, Tsimane, Hadza, metagenomics, human migration, co-evolution, microbial diversity, horizontal gene transfer, VANISH taxa, industrialization, population genetics, bacterial evolution