Can resident fish help wounded corals heal?

Could a fish living among coral branches change how quickly damaged tissue closes? Vega and colleagues tested that question in 2024 at the Richard B. Gump South Pacific Research Station, then reported the experiment in 2026. Fish presence was assigned to whole aquaria; wound size varied among coral fragments within each aquarium.

Study at a glance: six 95-litre flow-through aquaria, with yellowtail damselfish in three and no fish in three. Each aquarium held twelve Pocillopora fragments: four unwounded, four with a small wound, and four with a large wound. The observation period lasted 21 days.

The evidence

Read the fish comparison at the aquarium level: three tanks with fish versus three without fish.

The central caveat: the fish treatment had six independent aquaria—not 72 independent coral fragments—with only three tanks per fish condition. The team also did not record which of the six source colonies supplied each fragment, so donor lineage could not be balanced or included in the analysis.

What remains unresolved?

The randomized fish assignment, standardized feeding, two wound sizes, and multiple response measures make this a direct experiment rather than a field association. The size-dependent response is also informative: the absolute fish effect on healing rate was about 3.3 times larger for large wounds than for small ones.

Several observations support the healing result without identifying its mechanism. By day 21, 16 of 24 wounded fragments in fish tanks had fully closed, compared with 9 of 23 in fish-free tanks. Of the eight wounds that still carried algae, only one came from a tank containing fish. Yet the researchers did not observe the fish eating that turf. The direction could run the other way: advancing tissue may simply have covered bare skeleton before turf became established. A third fish-mediated process could also have affected both outcomes.

The experiment cannot separate nutrients released by fish from water movement, oxygen delivery, microbial change, deterrence of other organisms, or another feature of fish occupancy. It used one damselfish species and one biomass level. Coral identity was resolved only to Pocillopora spp., and source-colony identity was not retained. Those limits narrow how far the aquarium result can be generalized.

One fragment in the fish-free, small-wound group lost additional tissue and was excluded from the healing-rate model, leaving 47 wounded fragments in that analysis. All rate estimates used 21 days as the denominator. An early-closing wound was therefore recorded as though it needed the full interval, making its rate a lower bound rather than an observed time to closure.

What this means

What the study found: under this six-aquarium protocol, assigning resident Dascyllus flavicaudus to tanks increased wound-closure rates and preserved higher photosynthetic efficiency near wounds, with the largest measured differences after larger injuries. Fish presence did not produce a detectable change in area-adjusted calcification over 21 days.

Reef Cloverson interpretation: the treatment supports a causal conclusion about fish presence in this setup, but not about any single pathway. A useful next experiment would manipulate one candidate pathway while holding the others as constant as practicable—for example, separating a dissolved-waste treatment from the physical presence of fish—using independently controlled tanks and donor-balanced fragments. That would test mechanism rather than simply repeat the fish/no-fish contrast.

Could we replicate this?

Published protocol: feasible for a research aquarium with animal-care approval, regulated access to organisms, six independent flow-through systems, and consistent photography. The independent unit for fish presence was the aquarium. Randomization occurred both when fragments were distributed among tanks and when the three wound treatments were assigned within tanks.

The original experiment used 72 standardized branches cut from six mature colonies, twelve fragments per 95-litre aquarium, and four fragments per wound class in each aquarium. Small wounds averaged 1.48 cm² and large wounds 3.82 cm². Six adult damselfish were placed in each fish tank. All tanks received two grams of flakes three times daily, uneaten food was removed after ten minutes, and tanks and coral plugs were cleaned daily. Wounds were photographed with a scale and measured in ImageJ; buoyant weighing, wax-dip surface-area estimates, and a Diving-PAM fluorometer supplied the other endpoints.

Proposed adaptation—not performed by the authors: a smaller pilot could retain the fish-present and fish-free comparison while recording donor lineage, distributing fragments from each donor across conditions, and photographing every wound on repeated dates. Each tank—not each fragment—must remain the treatment replicate. A pilot without PAM could evaluate closure and visible algal occupation, but it could not reproduce the photosynthetic-recovery endpoint. A design that does not separate fish excretion, swimming-generated flow, and other fish activities still could not identify the causal pathway.

Major constraints include animal-welfare review, collection and holding permits, suitable fish–coral pairing, independent environmental control, and the need to avoid confounding tank effects with treatment. Using a different coral, fish species, fish biomass, or closed recirculating system would be an adaptation, not an exact replication.

U.S. research resources: the author affiliations connect this work to the University of California, Santa Barbara; the University of Georgia; and the University of Kentucky. The authors also deposited their analysis materials in a public Zenodo archive. These are public starting points for methods review, not evidence of available specimens, facilities, access, or collaboration.

Methods: How they did it

In August 2024, the team collected 72 branches from six mature Pocillopora colonies in Mo’orea. Branches were cut to four centimetres, attached to ceramic plugs, allowed three days for the cutting margin to regrow, and then randomized among six flow-through aquaria receiving unfiltered seawater at six litres per hour. An airbrush and rectangular stencils produced either no new wound, a small wound, or a large wound.

Eighteen adult yellowtail damselfish, averaging 5.52 ± 0.70 g, were randomly assigned to three aquaria; three aquaria remained fish-free. The researchers followed wounds for 21 days. They photographed a rotating subset daily and every wounded fragment at the endpoint, quantified wound area from images, measured buoyant weight at the start and end, estimated surface area by wax dipping, and measured dark-adapted Fv/Fm near the wound on day 21.

The healing analysis used mixed models with fish presence, wound size, and their interaction as fixed effects and aquarium as a random effect. Separate models addressed calcification and Fv/Fm. The paper reports likelihood-ratio tests and model-derived means with confidence intervals. Source-colony identity could not be modeled because parent IDs were not recorded.

Funding source

The paper acknowledges support from the Keck Foundation; U.S. National Science Foundation awards OCE-1851510 and OCE-1851032; and the NSF-funded Moorea Coral Reef Long Term Ecological Research program, including OCE-1637396 and prior awards. The acknowledgments separately thank Gump Station staff for assistance with field and wet-laboratory operations. Funding identifies support for the work; it does not by itself establish the outcome or remove the design limitations above.

Sources

Vega, Osenberg, Seifert, Munk & Stier (2026), Biology Letters, 22:20260177

PubMed record and complete citation

Public data and analysis archive · Zenodo

Supplementary material · Royal Society Figshare