In the decades after World War II, a one-eyed Irish missionary-surgeon named Denis Burkitt moved to Uganda, where he noted that the villagers there ate far more fiber than Westerners did. This didn’t just bulk up their stools, Burkitt reasoned; it also explained their low rates of heart disease, colon cancer, and other chronic illnesses. “America is a constipated nation,” he once said. “If you pass small stools, you have big hospitals.”

“Burkitt really nailed it,” says Justin Sonnenburg, a microbiologist at Stanford University. Sure, some of the man’s claims were far-fetched, but he was right about the value of fiber and the consequences of avoiding it. And Sonnenburg thinks he knows why: Fiber doesn't just feed us—it also feeds the trillions of microbes in our guts.

Fiber is a broad term that includes many kinds of plant carbohydrates that we cannot digest. Our microbes can, though, and they break fiber into chemicals that nourish our cells and reduce inflammation. But no single microbe can tackle every kind of fiber. They specialize, just as every antelope in the African savannah munches on its own favored type of grass or shoot. This means that a fiber-rich diet can nourish a wide variety of gut microbes and, conversely, that a low-fiber diet can only sustain a narrower community.

Sonnenburg, his wife Erica, and the graduate student Samuel Smits confirmed this idea in a recent experiment. The researchers started with mice that had been raised in sterile bubbles and then loaded with identical collections of gut microbes. They then fed these mice a high-fiber diet, before randomly switching half of them to low-fiber chow for seven weeks.

Predictably, the fall in fiber caused upheavals in the rodents’ guts. In the low-fiber group, the numbers of 60 percent of the local microbe species fell dramatically, and some remained low even after the mice returned to high-fiber meals. Those seven low-fiber weeks left lingering scars on the animals’ microbiomes.

These scars can cascade through generations. Mice regularly eat each others' poop, and pups often pick up their parents’ microbes in this way. Indeed, when Sonnenburg and Smits bred the mice from their first experiment, they saw that low-fiber parents gave birth to pups with narrower microbiomes, which lacked species present in the progeny of high-fiber parents. And if these bacteria-impoverished pups also ate low-fiber food, they lost even more microbes, especially those from the fiber-busting Bacteroidales group. As four generations ticked by, the rodents’ guts became progressively less diverse, as more and more species blinked out.

It also became increasingly hard to reverse these changes. If the fourth-generation mice switched to high-fiber meals, some of the missing microbes rebounded, but most did not. In other words, these species weren't just lying in wait in small numbers, waiting for the chance to bloom again; they had genuinely vanished. The only way of restoring these missing microbes was through a fecal transplant—loading them with the entire gut microbiomes of rodents that had always eaten a high-fiber diet.

These changes parallel those that have taken place over the course of human history. Many studies have now shown that the gut microbiomes of Western city-dwellers are less diverse than those of rural villagers and hunter-gatherers, who eat more plants and thus more fiber. The Stanford researchers’ experiment hints (but doesn't confirm) that this low diversity could be a lasting legacy of industrialization, in which successive generations of low-fiber meals have led to the loss of old bacterial companions. “The data we present also hint that further deterioration of the Western microbiota is possible,” the team writes.

“Given the infancy of the microbiome field, I think it is difficult to determine what specific impacts the loss of microbiota diversity has on the host,” says Kelly Swanson, a nutritional-science professor at  the University of Illinois at Urbana-Champaign. “But I think this paper provides even more evidence for including an adequate amount of dietary fiber in the diet.” For context, dietary guidelines recommend that women and men should respectively eat around 25 and 38 grams of fiber per day, but American adults eat just 15 daily grams on average.

This could be problematic for two reasons. First, without fiber, starving microbes often turn their attention to similar molecules, including those in the mucus layer that covers the gut. If they erode this layer sufficiently, they might be able to enter the lining of the gut itself, triggering immune reactions that lead to chronic inflammation.

Second, there’s evidence that a diverse microbiome can better resist invasive species like Salmonella or Clostridium difficile, while low diversity is a common feature of obesity, inflammatory bowel disease, and other conditions.

Still, no one has shown that a less-diverse microbiome is the cause of the health problems associated with low fiber intake.This means that it’s premature to talk about supplementing our microbiomes with those from communities that eat more fiber. Sonnenburg’s team writes, “It is possible that rewilding the modern microbiota with extinct species may be necessary to restore evolutionarily important functionality to our gut.” Sure, but first, they’d need to show if the microbial losses in their experiments matter, and to what degree.

After all, the diversity of the human microbiome has been falling long before industrialization. Even the rich gut communities of hunter-gatherers are a pale reflection of those of chimps and gorillas, whose diets are even richer in plants. The point is that animals tend to end up with the microbiomes they need; as our needs and habits change, so does our pool of partners.

Sonnenburg's concern is that these changes play out over millennia, and hosts and microbes have time to acclimate to their new relationships. By contrast, our modern diets and lifestyles are changing our microbiomes very quickly, leaving us with communities that we haven't adjusted to. “Our human genome is constantly trying to keep  up with this moving target of a microbial community,” he says. “If there are times when changes are exceptionally rapid, it might be problematic for host health.”