Standard pollution-marker genes failed the test. Their activity often reflected temperature or diet, not human impact. Only when researchers examined the entire genome did a clear urbanization signal emerge.

Why might fish prefer polluted waters?

The gene-activity data suggest an answer. Urbanized sites may contain more organic matter, making food abundant and easily accessible for young fish. Adult fish from urban areas showed gene patterns resembling those of well-fed laboratory fish. Yet these same fish also displayed elevated inflammation and immune responses, hinting at a hidden cost.

"We suggest this is a 'junk food effect,'" said Vincent Laudet, head of OIST's Marine Eco-Evo-Devo Unit and senior author of the study. "It's like teenagers going to a fast-food chain: there is plenty to eat, it is cheap, but it may not be good for their health."

The finding may help resolve a long-standing question in reef ecology: why young fish settle in visibly degraded coastal areas when cleaner habitats lie nearby.

Could animals themselves become environmental sensors?

The researchers argue their approach could complement water-quality testing by tapping organisms as integrated recorders of their surroundings.

"What is exciting here is that instead of simply measuring the environment around an animal, we can ask the animal itself what it has experienced," Laudet said. "Genome-wide gene activity integrates the many different environmental influences acting on an organism and can reveal physiological effects that remain invisible when we measure water quality or simply count which species are present. There is no reason why this approach should be limited to fish or coral reefs. In principle, any animal, in any ecosystem, could become a living sensor of environmental change."

Gairin added: "Some of the sites we were sampling were really dirty and polluted, but still, there they were. Just because they are present doesn't mean they're not under stress."

What are the limits?

The study examined one species on one island. Whether the same gene signatures appear in other reef fish, or in other ecosystems, remains untested. The "junk food" interpretation rests on a comparison between wild urban fish and laboratory-fed fish, which cannot capture every variable wild fish encounter, including predators, disease and microplastic exposure. Still, the results point to chronic physiological costs that conventional population surveys would miss entirely.

The paper, "Landscape transcriptomics reveals ecological constraints on fish under anthropogenic coastal change," appears in Nature Communications (2026). DOI: 10.1038/s41467-026-77751-2.