Lukas Vojir / Quanta

In the summer of 2011, the microbiologist Jon Sanders, then a graduate student, found himself in Peru’s tropical rainforest for the second time in as many years, lugging 60 pounds of lab equipment—a bulky fluorescence microscope and the generator to power it—up the Amazon River. Upon arriving at the remote field site, he quickly set about catching as many different ants as he could, eager to peer at the microbes that populated their guts.

In some of those ant species, he saw “this amazing, dense, packed cloud. It was like a galaxy of microbes,” he says. “They’d explode in your eyes when you looked at them” under the microscope. Which is what you might expect to find, given the extent to which we and so many other animals depend on the trillions of bacterial cells that reside within us—for processing food that we can’t otherwise digest, for providing key nutrients, for training our immune system to act effectively against infections. The microbiome is so crucial to our health and survival that some researchers even find it useful to think of animals as the sum of their microbial parts.

But when Sanders turned to the rest of the ants—about two-thirds of the different colonies and species he had collected—he was surprised to find that “you would be hard-pressed to find any cells in the gut that you could readily identify as bacteria,” he says. Food, debris, the cells of the insects’ gut lining—all were present. Microbes that might be engaged in the symbiotic relationships we take for granted, not so much.

As the tools to measure and analyze microbial communities have improved, it’s gradually become clear that the microbiome is nowhere near as ubiquitous and important across the animal kingdom as it’s often portrayed to be. Many animals seem to have more flexible or less stable associations with microbes; some don’t seem to rely on them at all. And ironically, these are the animals now allowing scientists to gain new insights into the mystery of how and why the microbiome evolves—its real importance, and the nuanced balancing act of pros and cons that lies at its core.

In the early 20th century, biologists began to uncover fascinating relationships between complex organisms and their microbes: in tubeworms that had no mouth, anus, or gut; in termites that fed on tough, woody plants; in cows whose grassy diet significantly lacked protein. Such observations generated excitement and prompted follow-up experiments. In those years, the absence of microbial helpers in an animal wasn’t considered particularly surprising or interesting, and it often received little more than a passing nod in the literature. Even when it was thought to merit more than that—as in a 1978 report in Science that tiny wood-eating crustaceans, unlike termites, had no stable population of gut bacteria—it ended up flying under the radar.

And so expectations quietly began to shift to a new norm, that every animal had a relationship with bacteria without which it would perish. A few voices protested this oversimplification: As early as 1953, Paul Buchner, one of the founders of symbiosis research, wrote with exasperation about the notion that obligate, fixed, and functional symbioses were universal. “Again and again there have been authors who insist that endosymbiosis is an elementary principle of all organisms,” he seethed. But counterexamples drowned in the flood of studies on the importance of host-microbe symbioses, especially those that made connections between human health and our own microbiome.

“The human microbiome has completely driven a lot of our thinking about how microbes work,” Tobin Hammer, a postdoctoral researcher in ecology and evolutionary biology at the University of Texas at Austin, says. “And we often project from ourselves outwards.”

But the human example is not a good model for what’s going on in a diverse range of species, from caterpillars and butterflies to sawflies and shrimp, to some birds and bats (and perhaps even some pandas). In these animals, the microbes are sparser, more transient or unpredictable—and they don’t necessarily contribute much, if anything, to their host. “The story is more complex,” Sarah Hird, an evolutionary biologist and microbial ecologist at the University of Connecticut, says. “More fuzzy.”

A transient, almost nonexistent relationship with bacteria was what Sanders saw in his tropical ants. He brought his samples back to his lab (then at Harvard University, although he is now at Cornell), where he sequenced the insects’ bacterial DNA and quantified how many microbes were present. The ant species with dense, specialized microbiomes had approximately 10,000 times more bacteria in their guts than Sanders found in the many other species he had captured. Put another way, Sanders says, if the ants were scaled to human size, some would carry a pound of microbes within them (similar to what humans harbor), others a mere coffee bean’s worth. “It’s really a profound difference.”

That difference, reported in Integrative & Comparative Biology in 2017, seemed to be associated with diet: Strictly herbivorous tree-dwelling ants were more likely to have an abundant microbiome, perhaps to make up for their protein-deficient diet; omnivorous and carnivorous ground-dwelling ants consumed more balanced meals and had negligible amounts of bacteria in their gut. Still, this pattern was inconsistent. Some of the herbivorous ants also lacked a microbiome. And the ants that did have one didn’t seem to have widespread, predictable associations with particular species of bacteria (although some sets of microbes were common to individual genera of the insects). That result marked a clear departure from mammalian microbiomes like our own, which tend to be very specific to their hosts.

The reasons why would become clearer as case studies of other organisms started to trickle in.

At around the same time that Sanders was examining ants in Peru, Hammer was in Costa Rica on an independent search for a microbiome in caterpillars. (“What better insect to have obligate relationships with bacteria than these cows of the insect world?” Sanders commented.) But try as he might, Hammer couldn’t find much bacterial DNA in the gut and fecal samples he collected. “Something really weird was going on,” he says.

When, after months of “frustrating lab work,” he realized that the animals might simply not have a stable microbiome, “it was a shift in thinking for me that was not expected at all.” He and his colleagues ultimately found that, like so many of Sanders’s ants, caterpillars had much, much lower quantities of microbes than was considered the norm. Moreover, those microbes were simply a subset of the ones found in the animals’ plant diet—“which supports the idea that they’re transiently passing through and some of them are getting digested, essentially,” Hammer says. “They’re not establishing stable populations within the gut.”

To determine whether those transient bacteria benefited the caterpillars, the researchers eliminated them using antibiotics. In other insects and animals, such a treatment tends to stunt development or kill the host outright. But it had no effect whatsoever on Hammer’s caterpillars.

Deepa Agashe, an ecologist and evolutionary biologist at the National Centre for Biological Sciences in Bangalore, India, saw something similar in insects that her team collected from several locations near the greenery of their campus. The microbes they found in dragonflies and butterflies strongly correlated with the insects’ diets rather than a particular insect species or developmental stage. The majority of the dragonflies’ bacterial communities seemed to have come together by chance. “Most of the bacteria were just there because they got there,” Agashe says. The insects “do not seem to be selecting for particular species of bacteria or a particular kind of bacteria.”

Repeated experiments that disrupted the butterflies’ microbial populations yielded no effect on the hosts’ growth or development. Neither did reintroducing the bacteria to their guts. “Really,” Agashe says, “they don’t seem to care about their microbes at all”—even though the butterflies feed on toxic plants and seem like perfect candidates for a full-fledged, functional microbiome that could detoxify their meals.

Like Hammer and Sanders, “initially we were scratching our heads,” Agashe says. “It was a surprising result, and actually it took us a while to wrap our heads around it.”

But maybe it shouldn’t be so surprising. As the scientists realized, when microbiomes are present, they’re often found in specific tissues—and they involve specific bacteria that influence specific traits at specific times. The bobtail squid, for example, has a symbiosis that’s limited to one species of luminous bacteria, which is sequestered in a single, light-producing organ while the squid’s gut and skin remain microbe-free. Adult honeybees have important relationships with their bacteria, but the larvae don’t.

So it’s not much of a leap to think that some animals might not have such relationships at all, or have relationships that play by different rules. “I think there’s now an increasing realization that there’s this whole spectrum of kinds of associations that you might find,” Agashe says.

Hammer agreed. “We’re just getting a glimpse at the tip of the iceberg,” he says.

And “it’s not a dichotomy between cows and caterpillars,” he added. “There’s a whole range of different kinds of lifestyles that are going to be really complicated.” Perhaps the transient, low-abundance microbes are doing something more nuanced, or perhaps they represent an early step in the formation of a more stable evolutionary relationship. Maybe they remain neutral most of the time and become functional only in certain contexts. Some researchers, for instance, posit that these microbes could protect a host from infections simply by taking up space in the gut and blocking out pathogens. Furthermore, bacteria that have adapted to a toxic plant or other hazard might be helpful even if they’re acquired just temporarily, without ever engaging in a formal symbiosis.

“Even if a transient microbiome is not associated with you,” says Alison Ravenscraft, a microbial ecologist and entomologist at the University of Texas at Arlington, “if you’re swallowing bacteria adapted to the environment, it’s possible that you could still derive a benefit from them. It would just be much harder to measure.”

Even in humans, she points out, the microbiome (including transient microbes) can shift with changes in diet or behavior. Studying living systems that don’t depend on a stable microbiome could help scientists disentangle the effects of those shifts. It could also allow them to better pinpoint the costs of having a microbiome and gain new insights into its evolution.

“If you think about it, there’s lots of reasons not to have an established microbiome,” Agashe says. “It’s actually not surprising that there are animals that have gone a different route. … But the key thing is, we don’t know why”—what factors lead to and enable the formation and maintenance of a microbiome, and conversely, what factors might prevent those relationships.

Caterpillars, dragonflies, certain ants, and other animals provide a way to investigate the potential disadvantages of long-lasting symbiotic relationships with live-in microbes; such disadvantages tend to be difficult to measure and test. Researchers suspect that these animals might be selectively avoiding certain potential penalties of symbiosis: Bacteria might compete with their host for nutrients, for instance, or aggravate the immune system.

For some animals, those risks might outweigh the potential benefits. If they have already evolved whatever enzymes or behaviors they need to live on their own, they’re no longer bound by selective pressures to acquire a microbiome. That might be the case for Hammer’s caterpillars, which eke out their herbivorous lifestyle simply by eating massive amounts of plant material. A microbiome might theoretically enable the caterpillars to manufacture additional important nutrients or go after more nutrient-dense vegetation, but the insects can make up for quality with quantity.

Another factor that might bear on the presence or absence of microbes seems to be anatomy (although Agashe does not consider it a plausible explanation, given the blurred line between cause and effect). Many of the organisms carrying few bacteria have a short, simple gut structure, essentially a tube through which food gets rapidly swept and processed. That doesn’t give microbes the time or space to gain a foothold and grow.

Ecological factors must also be considered. “If you think about how a symbiosis should or could get up,” Agashe says, “it’s actually pretty incredibly amazing.” Generation after generation, an organism has to encounter another species often enough to start a partnership that’s consistently and mutually beneficial, even under changing conditions. Agashe speculates that because her butterflies and dragonflies are constantly flitting from place to place, consuming diets that change with the location and the season, they may not meet up frequently enough with the same bacteria to establish a stable microbiome.

The researchers emphasize that there’s probably no single unifying rule or principle governing the evolution of the microbiome. “Evolution is incredible and idiosyncratic, and in lots of different organisms it proceeds along totally different routes,” Sanders says.

Hird agreed. “Most of our assumptions about microbiomes are based on mammal research,” she says, “when it might be the case that mammals are the weird ones. Maybe we don’t have stability even on a day-to-day basis in something like fish or birds or caterpillars.”

Even among mammals, there’s diversity in how the microbiome presents itself. Although most mammalian species seem to associate predictably with specific bacteria, a recent study by Sanders and his colleagues found that bats do not. In fact, their microbiomes were more transient and random—and bore a far closer resemblance to the microbiomes of birds than to those of fellow mammals. The researchers posit that this difference might relate to an evolved need for both bats and birds to be as light as possible to enable powered flight. Perhaps they couldn’t afford to carry any additional baggage.

Regardless, the findings illustrate that there’s much to learn by comparing species—and that there’s much to lose by making premature assumptions about what their relationships with bacteria might look like. At the very least, this means proceeding with greater caution when translating studies done in flies, mice, and other model organisms to humans. (Already, significant differences have been found in the gut microbiomes that develop in wild mice versus lab-bred mice—and the former have often proved to be a more accurate model of how certain experimental drugs might function in humans.)

“Every organism that exists has three and a half billion years of evolutionary history behind it, many millions or tens or hundreds of millions of which are not shared with organisms that we use as models,” Sanders says. Scientists’ emerging awareness of the diverse relationships that animals share with microbes “should make us really cautious about drawing inferences using fruit flies as models for gut microbiome importance or interactions, because fruit flies might be operating from a very different fundamental starting point compared to humans. It’s the same thing with mice.”

“We need to keep our eyes and ears open,” he added. “There’s still a lot to learn from natural variation and diversity.”

While scientists are beginning to acknowledge many cases that differ greatly from our own when it comes to microbiomes, “I think they’re still considered as kind of the oddballs,” Hammer says, “weird quirks that are totally unusual”—so much so that Sanders and others have found it challenging to get some of their work published.

The sooner that attitude changes, the more we’ll be able to learn. “We are barely starting to wrap our heads around the complexity of having hundreds of species in a very small amount of space, interacting with each other and interacting with their environment and interacting with the host,” Hird says. “Every scenario is probably still on the table.”

This post appears courtesy of Quanta Magazine.

We want to hear what you think about this article. Submit a letter to the editor or write to