More than 10,000 stool samples, one clear split
Researchers working on the American Gut Project in 2018 analysed over 10,000 stool samples and found a consistent pattern: people who ate more than 30 different plant types a week carried more diverse gut microbiomes than people who ate 10 or fewer plant types a week.
That gap showed up regardless of whether someone identified as vegetarian, vegan or omnivore. The trial found that plant variety predicted microbiome differences more strongly than any of those diet category labels.
This matters because low diversity is one of the few gut microbiome signatures that shows up again and again in people described as unwell in other research, including in studies of inflammatory bowel disease and obesity. It is an association, not a diagnosis, but it is a pattern researchers keep coming back to.
Why more plant types might matter more than more plant volume
Different plants carry different fibre structures, and different gut bacteria specialise in breaking down different fibre types. A diet built around three or four staple plants, even in large quantities, offers a narrow set of fibres. A diet spanning 30 or more plant types offers a much wider set of fermentable substrates.
The proposed mechanism is that each additional plant type feeds a slightly different subset of bacterial species. More available food niches is associated with more species persisting in the gut community, at least in cross-sectional data like the 2018 American Gut Project analysis.
This is a plausible mechanism, not a proven one. The 2018 study was observational, based on self-reported food frequency data linked to a single stool sample per person. It shows association between plant variety and diversity, not that adding plants causes diversity to rise in any individual.
What the antibiotic-resistance gene finding actually showed
Beyond diversity scores, the American Gut Project researchers also looked at antibiotic-resistance genes carried by gut bacteria. They found that people eating more than 30 plant types a week carried fewer of these resistance genes than people eating 10 or fewer.
This finding was associated with plant variety in the same 2018 dataset, using the same self-reported categories. The study did not test why fewer resistance genes turned up in this group, and it did not track whether this had any bearing on antibiotic treatment outcomes for these individuals.
It is a correlation worth noting rather than a mechanism worth explaining. Resistance genes can spread between bacteria independent of diet, and a single cross-sectional dataset cannot separate plant variety from other lifestyle factors that often travel together, such as overall diet quality or antibiotic use history.
What this evidence does not tell you
The American Gut Project relies on self-reported eating habits and a single stool sample per participant, which is a limited window into a system that changes day to day. It cannot show what happens if one person changes their plant intake over months or years.
No randomised controlled trial in this dataset assigned people to eat 30-plus plant types and then measured what changed. Without that kind of intervention study, the direction of cause and effect stays an open question: it is possible that people with already-diverse gut communities are simply more inclined to eat varied plants, rather than the plants creating the diversity.
The practical takeaway is narrow but concrete: counting plant types across a week, rather than counting grams of fibre or picking a diet label, is the metric this particular 2018 dataset found most closely tied to microbiome diversity. Whether widening that variety changes an individual's own gut community remains untested in this research.