Terence P. Hughes, Maria J. Rodrigues, David R. Bellwood, Daniela Ceccarelli, Ove Hoegh-Guldberg, Laurence McCook, Natalie Moltschaniwskyj, Morgan S. Pratchett, Robert S. Steneck and Bette Willis (2007), Current Biology 17(4):360–365 — DOI-10.1016/j.cub.2006.12.049
Key takeaways
- Removing large fishes after bleaching induced a phase shift. Inside the cages, herbivorous fish biomass fell to 0.45 kg/m² per hour of video, against 4.29 in partial cages and 3.12 in open plots. Macroalgal cover reached up to 91% and averaged 56% ± 21% (SE) after 30 months, against means of 4.1% and 1.7% in the controls. Final algal biomass was 1363, 146 and 68 g wet weight per m². Sargassum, absent from the crest before, grew 3 m tall.
- Coral recovery ran through recruitment and survival, and both were suppressed. Coral cover rose from 6.0% to 7.7% in cages, but to 19.2% and 20.2% in partial cages and open plots. Recruits per 25 m² were 39, 108 and 118. Mortality of colonies established before the experiment was 24.2%, 9.8% and 11.3%. In the cages the number of colonies fell by 26%; elsewhere it rose 14–16%. The control recovery was “primarily because of recruitment of species that had been locally extirpated by bleaching”: Acropora gave 246 of the 1062 recruits.
- The shifted state was not self-maintaining at this scale. When the mesh was removed, macroalgal cover fell from 53% to 13% in 12 days and to about zero in 30 days under grazing (citing Bellwood, Hughes & Hoey 2006).
- The management inference is the authors’ extrapolation. That local control of fishing gives “some insurance” against bleaching is argued from the experiment, but bleaching was not manipulated or repeated within it.
Standing in the record
Filed on 2026-10-04 at the owner’s request, as one of four readings on corals as composite systems (with LIT-709, LIT-712 and LIT-710). No anthology topic holds reef ecology. Its management conclusions are about fisheries, not machine-learning practice.
The owner’s question. The owner’s note: “Experimental evidence that reef-level recovery depends on ecological interactions, rather than simply the individual persistence of its corals.” The core of that is well supported, and more strongly than the note says. Recovery in the controls came mostly from new colonies, recruits of taxa the bleaching had wiped out on that crest. Excluding large fishes cut recruitment and also doubled the mortality of the colonies that had survived bleaching. So even the “individual persistence” the note contrasts with depended on the interaction. Four qualifications:
- “Reef-level” is a 25 m² plot level. The experiment covers 300 m² in one bay of one inshore island. The plots’ recovery depended on larvae arriving from the wider reef, which the experiment does not manipulate.
- The manipulated interaction is fish exclusion. The cages kept out all large and medium fishes, predators included. Herbivory is identified as the operative interaction by the macroalgal response and by the rapid grazing after the cages were opened. It is not isolated from predation, and small fishes increased inside the cages.
- How macroalgae suppress corals is inferred. The authors read the shift in recruit composition toward shade-tolerant Fungia and Euphyllia as an effect of shading. The sublethal and fecundity effects are in the supplement, which was not read.
- “Phase shift” here does not mean an alternative stable state. The algal state collapsed within 30 days once fishes returned. The experiment therefore shows that recovery depends on a continuing interaction, not that the reef has two self-maintaining states. The authors do not claim hysteresis, and the reversal is evidence against it at this scale.
Where it meets the record. The record holds no ecology of communities
or ecosystems, and has no topic word for it; complex-systems is the
nearest true word for a state of the whole maintained by an interaction
among its parts. Wilson and Barker (LIT-168) place the coral reef as a
group that is a living agent but not a Darwinian individual. This paper is
evidence that what keeps a reef coral-dominated after a disturbance is an
interaction among its members. That is the kind of fact a physiological
reading of the reef as one living system would need, and also the kind that
a reading of the reef as an ecosystem of separate organisms explains
without strain. It does not decide between them. The record’s account of
disorder as an alternative stable state held by self-reinforcing loops
(Borsboom, LIT-547) uses the same vocabulary of shifted states and
resilience. This paper’s reversal within a month is a reminder that a
shifted state need not be self-sustaining. No relation is declared: the
paper builds on, tests or rivals nothing the record holds.