Contribution
The first replicated herbivore-exclusion experiment to test herbivore–algae–coral interactions in the aftermath of mass bleaching, by the authors’ account. Before it, the role of herbivory in reef phase shifts was argued from Caribbean history and correlation (Hughes 1994; Mumby et al. 2006, a reserve comparison that did not report corals). It shows by manipulation that removing large fishes from a recovering, post-bleaching reef crest produces a macroalgal shift and suppresses coral recovery, and it decomposes the suppression into recruitment and mortality.
Key insight
After a mass mortality, a reef does not recover by its survivors regrowing. It recovers mostly by new colonies settling, and settlement and early survival need grazed, algae-free surfaces. Large herbivorous fishes keep those surfaces open. Take them away and the same disturbance that corals would otherwise recover from becomes the start of an algal reef.
Assumptions
- Site. Pioneer Bay, leeward Orpheus Island, inner Great Barrier Reef (18°36′S, 146°29′E). Turbid water (5–8 m visibility), tidal range 3–3.5 m, fishing banned since 1987. Coral cover and diversity had been sharply reduced by the 1998 bleaching. Alcyonaceans and branching Acropora were “virtually eliminated” from shallow sites.
- Treatments. Four fully meshed roofless cages excluding “all large and medium fishes”, four partial cages open along 50% of the perimeter as cage controls, and four open plots. Each plot was 25 m², 300 m² in all. Cages were 4 m tall, so no roof was needed. Mesh was 1 cm² in the bottom 2 m and 2 cm² above, scrubbed every 7–10 days. Diadema urchins were rare, and the three found in cages were removed.
- Caging artefacts. The authors argue against them because the cages were roofless and large, algae grew luxuriantly inside (so light was not limiting), and the partial cages showed no intermediate reduction in recruits.
- Timeline. Corals mapped in September 2000 and again in April 2003; cover from 16 photographic censuses; mesh removed after 30 months.
Key results
- Fish exclusion worked (Figure S1, as reported in text). Herbivore biomass was 0.45 ± 0.08 kg/m² per hour of video in cages, 4.29 ± 2.81 in partial cages and 3.12 ± 1.24 in open plots (F = 7.79, p < 0.001). This was measured after 28 months with 90 hours of remote video at 1 m² quadrats.
- Macroalgae (Figure 2A). Mean cover was 4.1% in partial cages and 1.7% in open plots (maxima 10% and 7%). In cages cover reached up to 91% and was 56% ± 21% after 30 months (repeated-measures ANOVA, F = 3.82, p < 0.05). Final biomass was 1363 ± 234, 146 ± 49 and 68 ± 28 g wet weight per m² (F = 20.8, p < 0.001). Sargassum grew 3 m tall, with more than 1000 holdfasts per 25 m² and up to 8.55 kg/m². Coralline algae composition also diverged (Figure S3).
- Coral cover (Figure 2B). Cages went from 6.0% ± 0.8% to 7.7% ± 1.0%. Partial cages reached 19.2% ± 2.3% and open plots 20.2% ± 2.2% (RM-ANOVA, F = 3.82, p < 0.05). Relative increases were 28%, 68% and 83%.
- Recruitment (Figures 3A, 4). There were 1062 recruits from 26 genera. Per 25 m² that was 39 ± 11 in cages, 108 ± 26 in partial cages and 118 ± 21 in open plots (F = 150.9, p < 0.001). Acropora gave 246 recruits (23%), about three times more outside the cages (F = 7.7, p = 0.011). Porites had 2 recruits in cages against 45 elsewhere (F = 12.49, p = 0.003). Fungia and Euphyllia, normally deep or shaded, were 18% of cage recruits against 3% elsewhere. The first two principal components explain 63% of the variation among the 12 replicates’ recruit assemblages. Partial cages and open plots were indistinguishable.
- Mortality (Figure 3B). Of the colonies present at the start, 24.2% died in cages, 9.8% in partial cages and 11.3% in open plots (RM-ANOVA, F = 4.29, p < 0.05). The net number of colonies changed by −72 ± 32 (−26%), +43 ± 21 (+16%) and +39 ± 24 (+14%) per 25 m².
- Reversal. After the mesh came off, macroalgal cover fell from 53% to 13% in 12 days and to about zero by 30 days. Juvenile fishes in the former cages fell 98% in 3 days.
- Sublethal effects. Coral tissue thickness (64 Porites cylindrica colonies) and reproductive output (90 Montipora digitata fragments: egg size, eggs per polyp, reproductive polyps) were measured. The results are only in the supplement and were not read here. The summary’s “suppressed the fecundity” rests on them.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Excluding large and medium fishes from a post-bleaching reef crest produces a macroalgal phase shift within 30 months | strong within its scale: replicated, controlled for caging, large effect | Figures 1–2, F-tests above |
| C2 | Without large fishes, coral recovery after bleaching is largely suppressed | strong within its scale | Figure 2B |
| C3 | The suppression works through both lower recruitment and higher mortality of established colonies | strong for the two demographic rates; the mediator (macroalgae, by shading) is inferred | Figure 3; recruit composition |
| C4 | Post-bleaching recovery in the presence of fishes came mainly from recruitment of locally extirpated taxa | moderate: stated in the summary and supported by recruit counts; the split between growth and recruitment in cover is not quantified | Summary; Results |
| C5 | The operative interaction is herbivory by roving fishes | moderate: the treatment excludes all large fishes; supported by algal responses and rapid grazing after reopening | Results; Conclusions |
| C6 | Managing fishing locally gives some insurance against large-scale disturbance such as bleaching | weak as stated: an extrapolation, since bleaching was a single prior event and was not manipulated | Conclusions |
| C7 | Loss of resilience can be measured by depletion of functional groups, recruitment and sublethal impacts | informal (methodological recommendation) | Conclusions |
Concepts
- resilience — “the ability of reefs to absorb recurrent disturbances … and rebuild coral-dominated systems”.
- phase shift (regime shift) — a change “to an alternate assemblage that is typically characterized by hyperabundances of fleshy seaweeds or other opportunistic species”. The paper does not require that the alternate assemblage be self-stabilising.
- trophic cascade / top-down control — changes in food-web structure and composition from reducing medium and large fishes.
- partial cage — a cage control open on half its perimeter.
Connections
- Kaniewska et al. (LIT-710). Same reef system, a shared author (Hoegh-Guldberg), and the other half of the recovery story: recruitment needs larvae, which come from synchronised spawning. Neither paper cites the other.
- Boilard et al. (LIT-712). Their model ends traumatic dysbiosis in algal “phase shifts”. This experiment shows a phase shift following bleaching through the loss of herbivory, with no microbial step.
- Borsboom (LIT-547) and THEORY-058. The record’s alternative-stable- state account of disorder rests on hysteresis. The rapid reversal here shows a shifted state that was not self-maintaining at plot scale. It is a contrast in the meaning of “shifted state”, not a test of that theory.
- Wilson & Barker (LIT-168). The reef is their example of a group that is a living agent but not a Darwinian individual. This paper is evidence about what keeps such a group in its state, not about whether it is one individual.
Bearing on the record
- Source of the theory that coral recovery after bleaching on a reef crest depends on large herbivorous fishes keeping macroalgae down, through recruitment and survival (THEORY-128).
- It is the record’s first experimental ecology. The vocabulary has no
word for community or ecosystem ecology (resilience, regime shifts,
trophic interactions).
natural-sciencesandcomplex-systemsare the honest tags, and the gap is reported to the coordinator. - No ML instruction; nothing for the anthology.
Limitations
- One site, one reef zone, one post-disturbance window, four replicates per treatment.
- Plot scale. 25 m² plots draw larvae and grazers from the surrounding reef. A real reef-wide loss of herbivores would also change larval supply, which the experiment holds fixed.
- Fish exclusion is not herbivore exclusion.
- The supplement was not read, so the fecundity and tissue claims are taken from the summary’s wording only.
- Two small errors in figure references. The text cites Figure 3B for recruitment where Figure 3A shows it, and Figure 1C for the cages, which it does not show.
Open questions
- Does the effect hold at reef scale, where larval supply and herbivore populations are themselves affected?
- At what herbivore density does the shift occur, and is there hysteresis in the threshold? Plot-scale reversal in 30 days suggests none here.
- How much of the cage effect is herbivory and how much is the loss of predators of small fishes?