HEALTH
For twenty years the gut microbiome has been sold to the public as a garden. Feed it fibre, water it with fermented foods, avoid the weedkiller of unnecessary antibiotics, and the trillions of bacteria in your intestines will reward you with a flat stomach, a stable mood and a long life. It is your garden. Yours alone. A private ecosystem shaped by what you choose to eat.
A paper in Nature has just put a crack down the middle of that story.
A team led by the University of Vienna has found that specific populations of ordinary, harmless gut bacteria — not pathogens, not food poisoning, the everyday commensals living in the intestines of healthy people — can sweep across continents within a few decades. The pattern they leave behind in the genetic record looks, in the researchers' own words, like an epidemic.
And when the team lined those travelling populations up against human disease, some of them turned out to be strongly associated with colorectal cancer, inflammatory bowel disease, type 2 diabetes and advanced age — while their own close cousins, members of the very same bacterial species, were associated with the opposite.
The study was pushed into circulation on Monday and has been racing through the microbiome field all week. Its actual finding is more interesting, and more unsettling, than the headline it has generated.
THE PROBLEM WITH COUNTING SPECIES
To understand what the Vienna group did, you have to understand why almost every microbiome study you have ever read about might be measuring the wrong thing.
Standard practice is to sequence what is in a stool sample and count bacteria by species — or by something broadly species-shaped. Then you compare the counts between sick people and healthy people and report which species are enriched in disease. This is how the field produced its greatest hits: this bug is linked to obesity, that one to Crohn's, this other one to depression.
The trouble is that a bacterial species is an enormously baggy category. Two organisms filed under the same species name can differ in their gene content more than a human differs from a mouse. Lumping them together is like reporting that "mammals" are associated with living in the Arctic. Some are. Most are not. The average tells you nothing.
"If you don't just count species but take evolutionary adaptation into account, you can identify the biologically relevant units in the microbiome much more accurately," said lead author Xiaoqian Annie Yu of the Centre for Microbiology and Environmental Systems Science at the University of Vienna. "Even within the same bacterial species, some populations occur more frequently than others in certain diseases. When all are considered together, this often remains hidden."
So the team went looking for a better unit — not one defined by convention, but one defined by evolution itself.
HUNTING FOR BROOMS
The approach is called reverse ecology: instead of studying an organism's environment to work out how it lives, you read its genome and let the genome tell you what environment it has been shaped by.
The specific signature the researchers hunted for is a genome-wide selective sweep. It works like this. Somewhere in a bacterial population, one individual acquires a mutation that gives it a decisive advantage in its particular corner of the gut. It out-competes every close relative. Its descendants take over. Because the whole genome hitch-hikes along with the winning mutation, the surviving population is not just successful — it is genetically uniform, a single clone's family, which then slowly re-diversifies.
On a family tree, that leaves an unmistakable shape: a long bare branch with a dense bushy tuft on the end. The researchers call them brooms. A long handle, a thick head.
Finding brooms is harder than it sounds, because bacteria swap DNA promiscuously and because our collections of cultured bacteria are patchy and biased. Both can fake a long branch. So the team built a computational pipeline — PopCoGenomeS, for Populations as Clusters of Genome Sweeps — that first separates the parts of each genome inherited straight down the family line from the parts acquired sideways by recombination, then applies a statistical test known as the 5X rule: if the genetic distance between a cluster and its nearest relative is more than five times the diversity inside the cluster, the odds that the pattern arose by chance drift rather than selection are under one per cent.
They ran it across 16,864 bacterial genomes isolated from human guts, sorted into 176 species-level groups with enough genomes to analyse. That produced 377 candidate sweeps. Then, to check that the pattern was not an artefact of which bacteria happen to be easy to grow in a laboratory, they mapped 1,477 whole-community metagenomes — raw genetic soup from real human samples, drawn from 74 separate datasets spanning different ages, health states and countries — onto the candidates.
One hundred and twenty-four sweeps survived, across 66 bacterial groups spanning some 25 families. Forty-six of the 66 were common commensals. Eleven were pathogens. Nine were the sort of bacteria found in yoghurt and probiotic supplements.
Among the commensals alone — the least contaminated signal, because their spread cannot be explained away by disease outbreaks or the marketing of probiotics — the team confirmed 77 sweeps in 46 bacterial groups. Every single one carried the population-genetics fingerprint of recent selection. And in 43 of the 77, at least one gene showed evidence of ferociously strong positive selection, changing far faster than chance would allow.
THE BIT THAT LOOKS LIKE AN EPIDEMIC
Then the researchers asked where these winning populations live. The answer is: everywhere.
Of the 77 commensal sweep clusters, 54 turned up on more than one continent. Of those 54, forty-two showed no geographical preference at all — they were as likely to appear in one part of the world as another, which suggests they are responding to the same selective pressure in wildly different human populations, on wildly different diets.
Then came the clock. Using established estimates of how fast bacterial genomes accumulate mutations — one to ten changes per genome per year — the team dated the sweeps. The range is enormous: from roughly a decade old to thousands of years old, a near thousand-fold spread. They cross-checked those dates using a second, independent method, tracking mutations that accumulated in the same clusters in people sampled more than a year apart, and in twins who had lived separately since adulthood. With two exceptions, the two methods agreed to within an order of magnitude.
And using the middle-of-the-road clock rate, 26 of the intercontinental sweep clusters came out younger than a century.
That is the sentence that has the field talking. In 47 of the 54 intercontinental clusters, bacteria sampled on different continents were no more genetically different from each other than bacteria sampled on the same continent — the hallmark of repeated, rapid, long-distance transfer.
"Our findings show that gut bacteria are also more dynamic than previously thought," said study leader Martin F. Polz of the University of Vienna. "Well-adapted strains can spread internationally and occupy new ecological niches."
Until now, that kind of global genetic homogeneity has been the signature of pathogens — cholera, tuberculosis, drug-resistant staph. Harmless gut bacteria were assumed to be inherited from your mother, shaped by your dinner, and largely parochial. The Vienna data say that at least some of them travel like infections, just more slowly. The paper is careful about this: commensals seem to need decades rather than months to go global, because the transmission chains between human guts are slower and less efficient than a cough.
THE HUNTER-GATHERER GHOST
Buried in the results is one finding with the eerie quality of an archaeological discovery.
A single sweep cluster of Bacteroides uniformis — a workhorse gut bacterium found in most people on earth — turned up in exactly three human populations and nowhere else: the Baka and Beti peoples of Cameroon, and the Matsés of Peru.
The Baka and the Matsés are remote forest communities on different continents who have never had contact with each other. The only plausible explanation for them sharing the same clonal population of gut bacteria is a chain of transmission running through other, less isolated human groups — a bacterial relay that once connected central Africa to the Amazon.
And its absence from everyone else is the sharper point. This cluster is the oldest sweep the team found in that species, dated at roughly one hundred to a thousand years. It appears to have spread before industrialisation, and then to have gone extinct almost everywhere as diets, cities and antibiotics rearranged the human gut. It survives in the two places the modern world reached last.
Somewhere between the invention of the flush toilet and the arrival of the supermarket, a population of bacteria that had circumnavigated the planet quietly died out of most human beings. Nobody noticed, because nobody had a name for it.
WHY THIS MATTERS FOR DISEASE
The commercial and clinical stakes sit in the final section of the paper.
The researchers scaled their method up to 3,941 metagenomes from 17 datasets and asked whether individual sweep clusters — rather than whole species — tracked with human conditions. They tested two: colorectal cancer, and advanced age. Thirty sweep clusters showed statistically robust associations with colorectal cancer and twenty-seven with old age, spanning multiple independent datasets with no geographical bias. Another 28 and 26 respectively showed associations within specific countries.
The killer detail concerns Bacteroides fragilis, one of the most studied bugs in the human gut. As a species, B. fragilis showed no significant association with either colorectal cancer or ageing. Null result. Move on.
Except it was not null. The species contained six distinct sweep clusters, and they pointed in opposite directions — some associated with the condition, others protectively against it. Three up and two down for ageing; one up and one down for cancer. Averaged together at species level, they cancelled each other out into a tidy nothing.
This is the paper's real bombshell, and it is a methodological one. It suggests that some unknown fraction of the microbiome literature — the negative results as much as the positive ones — may have been washing out real biology by measuring at the wrong resolution. Two organisms with the same species label, living in the same organ, can be doing opposite things.
As for how they differ, the first clue is chemical camouflage. When the team looked for genes unique to each B. fragilis sweep cluster, the enrichment was in machinery for building and modifying capsular polysaccharides — the sugary coat a bacterium wears on its surface. That coat is exactly what pathogens use to dodge immune recognition, and B. fragilis capsules are already known to be able to damp down inflammatory disease. The implication is that these populations may be differentiated not by what they eat, but by how they negotiate with your immune system.
THE CAVEATS, AND THE WORD 'CONTAGIOUS'
The framing that has spread fastest this week — that your microbiome is contagious — is a fair headline but a loose one, and the distinction matters.
The Vienna team did not watch bacteria move from one person to another. Nobody swabbed a household. This is an inference drawn from the shape of family trees and the ticking of a molecular clock, and molecular clocks in bacteria carry substantial uncertainty; the authors say so explicitly. "Twenty-six clusters younger than a century" is a best estimate with wide error bars, not a stopwatch reading.
The transmission itself, though, is not really in doubt — that groundwork was laid in 2023, when an international team analysed more than 9,700 human metagenomes and documented over ten million instances of bacterial strain-sharing between people. Infants share more than half their gut strains with their mothers. People who live together share strains at rates far above people who do not. Nicola Segata, who led that work, is a co-author on this one.
What is new here is not that bacteria move between people. It is that specific, competitively superior populations appear to be riding those chains of transmission to global dominance, on a timescale short enough to have happened within living memory.
Two further limits deserve flagging. The analysis is constrained by which bacteria humans have bothered to culture — the researchers note that the number of sweeps they found in a group correlates with how many samples exist for it, meaning there are almost certainly more sweeps hiding in the families that are hard to grow. And every disease link in the paper is an association. None of it establishes that a sweep cluster causes bowel cancer. It is entirely possible that a tumour changes the gut environment and a particular adapted population moves in afterwards.
WHAT THIS MEANS FOR YOU
Practically, today: nothing. Do not change your diet on the strength of this. Do not buy a probiotic because of it — and note, drily, that nine of the sweeps the team detected were in exactly the sort of bacteria sold in supplements, which means the products themselves may be moving these populations around the globe.
What it changes is the shape of the promise. The microbiome has spent two decades generating more excitement than clinical benefit, and one plausible reason is now on the table: the field may have been measuring the wrong unit. If the meaningful thing is the sweep cluster rather than the species, then microbiome diagnostics that have underperformed might work far better at higher resolution — and a stool test that reports which populations you carry, rather than which species, could be a genuinely different product.
It also reframes what a microbiome actually is. If well-adapted bacterial populations spread between people the way fashions or accents do, then your gut is not only a product of your choices. It is partly a product of who you live with, where you have travelled, and which populations happened to be sweeping through the human species while you were young. Strains, once acquired, tend to stay: on average, a strain persists for more than a third of a human lifespan.
The garden metaphor was always slightly wrong. This is less a private plot than a shared commons, with traffic running through it.
THE BOTTOM LINE
A Vienna-led team has shown that genome-wide selective sweeps are a routine, pervasive force reshaping the bacteria in human intestines; that they carve familiar species into distinct populations with distinct ecologies; that more than two-thirds of the ones they confirmed have crossed continents, some of them within the last hundred years; that populations inside a single species can point in opposite directions on disease; and that one of them appears to be a relic surviving only in two isolated forest communities half a world apart.
The next step, the researchers say, is to work out which genes actually separate these populations and what those genes do. That is where a diagnostic — or a therapy that boosts one population and suppresses another — would have to come from.
For now, the finding stands as a quiet correction to a very loud field. The bacteria in your gut are not a static garden you tend. They are a population under selection, competing, winning, and moving — and on the evidence of this week, moving considerably faster than anybody thought.
