Between May and July 2023, at 18 points along the coast of Okinawa, a team of biologists collected 179 specimens of a small reef fish, the blue damselfish, Chrysiptera cyanea. Some sites faced shores still covered in vegetation. Others faced harbours, roads and housing. From the 90 juveniles they kept the whole body. From the 89 adults, 49 females and 40 males, they kept the liver.
Then they read the transcriptome, the set of messages the genes are sending at a given moment. DNA tells you what a fish can do. The transcriptome tells you what it is doing. If an animal is hot, hungry or fighting inflammation, the list of active genes changes.
The study is led by Emma Gairin and Vincent Laudet of the Okinawa Institute of Science and Technology, with colleagues at Academia Sinica in Taiwan, the CNRS station at Banyuls-sur-Mer in France and Indiana University. It was published in Nature Communications on 9 October 2026. The journal received it on 4 August 2025 and accepted it on 3 September 2026, after thirteen months of review.
Place matters a great deal. In juveniles the collection site explains 76 per cent of the differences between gene-expression profiles, in adults 67 per cent. In juveniles 2,410 genes change expression in more than a quarter of the pairwise site comparisons. In adult livers the figure is 251.
To measure the human footprint the authors chose a simple index: the share of natural land within one kilometre of each site, taken from the Japanese space agency's land-use map of Okinawa. Less forest and beach means more houses and tarmac. The authors treat it as a proxy for human presence. It does not measure what the fish breathe or eat.
The first driver, though, is temperature. 1,854 genes in juveniles and 1,312 in adults respond to it alone. At the four warmest sites the water was between 29.6 and 32.4 °C. Above 31.9 °C the signature of heat shock proteins appears, the molecules cells make to protect themselves from heat. Remove those four sites and the temperature-linked genes fall by almost two thirds.
The human footprint comes second, but it shows. 425 genes in juveniles and 585 in adult livers are associated only with proximity to urban land. They concern immunity, inflammation, programmed cell death and the cell cycle.
The oddest result concerns food. To recognise it in the genes, the authors kept juveniles and adult females in tanks, half fasted and half fed, with seven fish sequenced per group. Juveniles for three days, adults for fourteen. From those experiments they derived a genetic signature of feeding and looked for it in wild fish. Adults caught near towns come out better fed. In juveniles the difference does not appear.
The authors read this as an energy-rich diet combined with chronic stress. Survival in urban zones, they write, depends on a trade-off. The extra food says nothing about where it comes from, and the study does not measure it.
Another result runs against expectations. Many genes long used as pollution sentinels, tied to thyroid hormones, stress and the processing of foreign substances, do not track urbanisation. They respond mostly to heat and feeding. Anyone monitoring a coastline with those markers should know that a warm or well-fed fish can shift them with no pollutant involved.
The limitations are stated. The urban index is indirect. The tank experiments are short. Juveniles were not weighed in the field, and the models do not account for interactions between factors. The study covers one species on one island. It shows associations between genes and environment, not a chain of causes.
In the Mediterranean the blue damselfish's closest relative is the damselfish Chromis chromis, known in Italy as castagnola, a member of the same family, Pomacentridae. The same method could be pointed at Italian harbours, but none of these findings transfers directly to European coasts.
The research was funded by the Japan Society for the Promotion of Science, the Okinawa institute and the Iwatani foundation. The authors declare no competing interests. The sequences are deposited with the US NCBI database, the processed data and code on figshare. The paper carries no statement on the use of artificial intelligence.
References
Gairin E., Miura S., Chamot Z., Sautereau C. A., Zwahlen J., Takamiyagi H., Gibert Y., Herrera M., Laudet V., "Landscape transcriptomics reveals ecological constraints on fish under anthropogenic coastal change", Nature Communications, 17, 10218, 9 October 2026, open access (CC BY 4.0). doi: 10.1038/s41467-026-77751-2
Study data: NCBI BioProjects PRJNA1168661, PRJNA1271234, PRJNA1168266; processed data and R code on figshare, doi: 10.6084/m9.figshare.27966972; reference transcriptome, doi: 10.6084/m9.figshare.27143571
