Gut bacteria don't inherit family lines the way genes do

A study by Suzuki et al., discussed in Source 1, had argued that dozens of species of gut bacteria evolved in parallel with different human population groups over long stretches of history, a process called codiversification. The claim rested on finding statistical similarities between family trees built from human genomes and family trees built from the genomes of gut bacteria carried by the same people. An accompanying commentary went further, proposing that some gut bacteria may have stayed faithful to particular human lineages for thousands of generations.

A new re-analysis in Source 1 pushes back on that stronger picture. Comparing the same data, the author reports that the statistical links between human and microbial family trees are much weaker than the links between two unrelated halves of a person's own genome. Because different chromosomes should be equally good markers of a person's ancestry, that gap suggests the bacterial results are not capturing a tight evolutionary partnership between microbe and host.

The author also shows that even fake correlations can appear where none should exist. In two theoretical examples, geographic patterns alone — with no shared evolutionary history between humans and microbes at all — produced statistical scores as strong as the largest ones reported in the original study. That means a significant result on this kind of test does not, by itself, prove that a bacterial species and its human hosts evolved together.

Suzuki et al., as described in Source 1, used a method called PACo to measure how closely a bacterial family tree matched the human family tree of the people carrying it. The test works by shuffling which host goes with which bacterial strain and checking whether the real pairing looks more similar than the shuffled ones. Under this test, 36 of 59 bacterial species examined showed a stronger-than-chance match to human ancestry.

The re-analysis in Source 1 adds a comparison point that the original study lacked: two unlinked halves of the human genome, compared against each other the same way. Different segments of a person's own DNA are inherited somewhat independently within a local population, so any correlation that remains between them reflects only the broadest features of shared ancestry — mainly which continent or country someone's ancestors came from. Repeating the PACo test on these two genome halves produced an effect size of about 0.7, far larger than the strongest bacterial effect size, around 0.07, found by Suzuki et al. In other words, two pieces of a person's own genome resemble each other in ancestry far more than any bacterial species resembles its human host.

The gap held up under direct inspection of the underlying genetic distances, not just the PACo scores. For species with the strongest reported codiversification signals, such as Prevotella copri, hosts with more distant ancestry did carry slightly more different bacterial strains on average — a statistically detectable pattern. But that average difference was small next to the variation in bacterial strains found within a single country, or even between the most closely related hosts, such as identical twins.

The re-analysis attributes the weak correlation partly to how freely gut bacteria travel between unrelated people. Earlier work cited in Source 1 found that the genetic difference between bacterial strains from different continents is often no larger than the difference between strains from the same location, and that unrelated hosts on different continents can sometimes carry nearly identical strains. Even identical twins, who share the same genome, often end up colonized by more distantly related bacterial strains over their lifetimes.

Separate work in Sources 2 and 3 points to a mechanism that could explain this mobility: horizontal gene transfer, the movement of genetic material between organisms that are not parent and offspring. Source 2 describes the gut as an especially favorable environment for this kind of exchange, citing its stable temperature, constant food supply, and the sheer density of microbial cells and viruses packed together, all of which create many opportunities for bacteria to swap DNA directly. Source 3, a longitudinal study of 338 people from the Lifelines-DEEP cohort in the Netherlands sampled roughly four years apart, identified 5,644 high-confidence gene transfer events among gut bacteria, with 61.2% of the reconstructed bacterial genomes involved in at least one transfer.

Source 3 further reports that some of these transfers happened very recently: an estimated 154 events showed DNA matches so close that they must have occurred within the past 100 years, including the spread of tetracycline resistance genes across four different bacterial genera at essentially identical sequence, consistent with the rise of tetracycline antibiotic use starting in the 1950s. This kind of rapid, recent exchange is a different process from the vertical, generation-to-generation inheritance that codiversification would require, and Source 1 argues that frequent exchange of this sort would tend to blur any signal of long-term parallel evolution between a bacterial species and its human hosts.

Source 1 does not claim that human and bacterial ancestries are completely unrelated. For Prevotella copri, the analysis found a statistically significant, if small, tendency for host genetic distance and bacterial genetic distance to increase together. The disagreement is about how to interpret that kind of signal: a statistically significant match under the PACo test, the source argues, does not by itself demonstrate the close evolutionary partnership implied by traditional codiversification, because geographic clustering of both human and bacterial diversity can produce similar-looking results on its own.

The re-analysis also notes a technical complication: frequent recombination — the shuffling and mixing of genetic material — within both human and bacterial genomes on these timescales makes any single family tree an imperfect summary of ancestry to begin with. It states that further caveats and details are addressed in supplemental methods and figures not included in the passage reviewed here. The author frames the findings as pointing toward the need for different strategies to quantify how much humans and their gut microbes have truly coevolved, rather than settling the question outright.