In January of 2011, Ilana Brito maxed out her credit cards and booked a trip to Fiji. At the time, she was a postdoctoral fellow at Columbia University, with a particular interest in how infectious diseases move throughout human communities and the environment. Nine months earlier, Brito had attended a talk in New York by Stacy Jupiter, a marine biologist at the Wildlife Conservation Society’s program in Fiji, who discussed the link between watershed health and human health. In Fiji, cases of typhoid typically spike after cyclones and other serious storms: pools of standing water form, providing safe harbor to Salmonella typhi, the bacterium that causes the illness. Like many other infectious diseases, typhoid can be very difficult to track in humans, but Brito—undaunted by the challenge of this epidemiological quandary—began to wonder whether it was possible instead to follow the bacterium as it travelled. She approached Jupiter after the talk with the idea of piloting the Fiji Community Microbiome Project (FijiCOMP). Soon, she was on her way.

In Fiji, Brito sought to understand not only how bacteria move between human communities but also how genes move between bacterial communities. Unlike most human cells, bacteria have a flexible genome, meaning that they can pick up stretches of genetic material from viruses, other bacteria, or dustings of DNA in their immediate environment. “It’s as if a giraffe can pass to an elephant its gene for a long neck,” Brito told me. These genetic vagabonds, known to scientists as mobile genes, can equip microbes with a wide range of new skills. Some are advantageous to humans: the gut bacteria of many Japanese people, for instance, share genes that allow them to digest seaweed, a staple of the Japanese diet. But other mobile genes are more pernicious. They can spread drug resistance, making a bug like Staphylococcus aureus, which was eminently curable about seventy years ago, almost invulnerable to common antibiotics such as methicillin. Brito’s study would investigate what caused these and other genes to mobilize—why some remained specific to certain populations while others dispersed worldwide.

Fiji proved the perfect setting for the project. “Here, people live in very close family units, which allows you to look closely at the transfer of genes within the same household,” Jupiter, who has worked in the country for more than eight years, told me. “And, within separate villages, you can look at what specific environmental factors or behaviors can influence the types of mobile genes.” Still, the logistical aspects of FijiCOMP were daunting. “The idea of one person single-handedly doing this project in the developing world is pretty extreme,” Eric Alm, one of Brito’s former academic advisers, told me. While the Human Microbiome Project, a similar effort launched in 2008 and funded by the National Institutes of Health, drew several hundred investigators from thirty-four institutions around the United States, FijiCOMP was largely a one-woman show.

Brito set out to survey a total of four remote villages, first in the lush Bua Province, then in the agricultural region of Macuata, often staying with her subjects in their modest houses. Over six weeks, Brito did her rounds, collecting hand swabs and stool, saliva, and soil samples, which she placed in pre-labelled test tubes. Microbiologists typically freeze these samples immediately to preserve an accurate snapshot of the microbial population and avoid contamination—the bacteria would otherwise thrive and divide rapidly in Fiji’s warm, humid climes—but none of the villages had enough electricity for proper refrigeration. In preparation for these conditions, Brito arrived armed with three liquid-nitrogen freezers, which weighed just over fifty pounds apiece fully charged. She eventually collected data on three hundred people. By comparison, the Human Microbiome Project amassed samples from only two hundred and forty-two Americans by the time researchers submitted a paper, three years after its launch.

Once Brito was back in the United States, she solicited the help of her colleagues at M.I.T., where she had moved from Columbia, and the Broad Institute, and began genetically sequencing the samples. If two distantly related organisms within the same sample shared a segment of DNA, Brito identified it as a mobile gene. And, since she wanted to see whether the genes were specific to the environment and conditions in Fiji, she compared them with results from the Human Microbiome Project. Her findings were recently published in the journal Nature. They show that diet, in particular, has a strong effect on the types of mobile genes in disparate populations. Since Brito’s subjects primarily ate root vegetables, such as taro, plantains, and cassava, the bacteria in their guts were likely to trade genes involved in digesting complex, starchy foods. “Our microbiome needs to be fed to stay alive and is at the mercy of what we ingest on a daily basis,” Gautam Dantas, a microbiologist at Washington University in St. Louis, told me.

But what was concerning, from a public-health perspective, was the prevalence of antibiotic-resistance genes among the Fijians, even though they had never been exposed to some of the powerful antibiotics that are frequently prescribed in the affluent West. Brito found bacteria in her subjects and the soil, for instance, that were immune to cephalosporins, which are used to treat pneumonia, meningitis, and other infections. One way for these resistance genes to have gotten to Fiji, she said, was in the body of a visitor from abroad. But it’s also feasible that they arose on their own. Microbes naturally produce antibiotics, which their neighbors then must develop resistance to in order to survive. “How these antibiotic genes from bacteria in soil can wind up in the human gut isn’t well understood,” Brito said. “Since many mobile genes can move together, it’s possible these cephalosporin-resistant genes hitchhiked with other antibiotic-resistant genes.” By protecting themselves from their toxic neighbors, the bacteria may have unknowingly limited the types of drugs that would effectively treat infections in their human hosts.

Brito’s findings add a twist to how scientists have traditionally studied the microbiome. They suggest that, rather than merely taking a roll call of the microbes that are involved with different states of health and disease, we ought to remember their genetic dynamism, too. “If you’re just looking at the types of bacteria in a microbiome, you’re missing a part of the story,” Alm said.