In the nine Tsimane villages along the Maniqui River and in a nearby forest region of the Bolivian Amazon, none of the families in the study had plumbing, used pit toilets or diapers, or owned a refrigerator when researchers began handing participants sterile urine cups for a gut microbiome study. The collectors returned between 7 a.m. and 9 a.m., carried the specimens to a field laboratory in coolers with reusable ice packs, homogenized each one in its cup, divided it with non-sterile wooden tongue depressors, and dropped the portions into liquid nitrogen. The stool came from 85 people, sampled in 2009 and again in 2012 and 2013, and it waited in freezers in the United States until a sequencer could read it deeply enough to reach the rare bacteria in it, at a median of 31.9 gigabases per sample.
Anyone who thinks about the gut microbiome at all tends to treat it as a reflection of diet and place, a community rebuilt in each generation from local food, soil, water and neighbors. On that view the Tsimane, who are forager-horticulturalists in the Amazon, and the Hadza, who are hunter-gatherers in Tanzania, should carry quite different bacteria, since their ancestors separated tens of thousands of years ago. The paper, published in Nature1 on 7 October 2026 and led by Stanford University groups that include Justin Sonnenburg of the Department of Microbiology and Immunology and Benjamin Good of the Department of Applied Physics, reports that 1,231 of the 1,408 microbial species recovered from the Tsimane (87.4%) also occur in the Hadza, whose samples were sequenced in an earlier study published in Cell in 2023.
Sharing a species proves little by itself, and the authors say so: a bacterium found on two continents could have been carried by ancestors or picked up last century from the environment or from other people. Some of the shared species are found everywhere. Ruminococcus bromii is one. But 848 of the 1,231 (60.2%) are rare in or absent from the gut microbiomes of industrialized populations, which makes them the candidates for something older. To separate inheritance from recent exchange, the team narrowed to the 636 shared species for which each cohort yielded at least four metagenome-assembled genomes (MAGs, bacterial genomes reconstructed from sequenced stool), then asked how closely related the strains were within each population compared with between them. They also flag a limit of their own: differences in sequencing depth or in strain-level heterogeneity between the two cohorts may have affected the recovery of some rare MAGs.
The first test looked at the last few thousand years. Pairs of MAGs sharing more than 10% identical genes were treated as clonally related, and by the team’s calibration that means a common ancestor within roughly the past 5,000 years. Of 78,746 such pairs across the 636 species (7.8% of all comparisons), 93.8% came from within a single population.
Of the 636 species, 545 (85.6%) showed no recent strain sharing between the Tsimane and the Hadza.
