Corals Chemically Cue Mutualistic Fishes to Remove Competing Seaweeds

See allHide authors and affiliations

Science  09 Nov 2012:
Vol. 338, Issue 6108, pp. 804-807
DOI: 10.1126/science.1225748

This article has a correction. Please see:


Corals in the genus Acropora generate much of the structural complexity upon which coral reefs depend, but they are susceptible to damage from toxic seaweeds. Acropora nasuta minimizes this damage by chemically cuing symbiotic goby fishes (Gobiodon histrio or Paragobiodon echinocephalus) to remove the toxic seaweed Chlorodesmis fastigiata. Within minutes of seaweed contact, or contact from only seaweed chemical extract, the coral releases an odor that recruits gobies to trim the seaweed and dramatically reduce coral damage that would otherwise occur. In turn, chemically defended gobies become more toxic after consumption of this noxious alga. Mutualistic gobies and corals appear to represent a marine parallel to terrestrial ant-plants, in that the host provides shelter and food in return for protection from natural enemies.

Coral reefs are in global decline, with seaweeds commonly replacing corals. Coral cover has decreased by ~80% in the Caribbean (1) and by ~50% along the Great Barrier Reef (2). Drivers of decline are debated, but all major stresses—including overfishing of herbivores, pollution, ocean heating, acidification, and disease (3, 4)—suppress corals, enhance seaweeds, and result in greater seaweed-coral competition.

For reefs to flourish, rapidly growing, branching corals such as Acroporids are critical because they create much of the topographic complexity upon which other species depend (4, 5). Other species, such as herbivorous fishes, then enhance reef resilience by grazing on competing algae and facilitating the colonization and growth of corals after disturbances (3, 4, 6). In the Caribbean, when two dominant Acropora species declined, structural complexity was lost across the entire region with likely effects on fishes, fisheries, biodiversity, coastal protection from wave damage, and ecosystem function in general (7, 8).

Reef-scale herbivory facilitates coral growth and maintenance by removing competitively superior seaweeds (3, 4, 9, 10), as exemplified by herbivore-rich reefs and marine protected areas that are higher in coral and lower in macrophyte cover, whereas overfished reefs with fewer herbivores have fewer corals and more macroalgae (3, 9, 10). However, individual corals are damaged only by adjacent seaweeds. Thus, critical aspects of competition occur at coral edges, a spatial scale over which corals might exert influence. Recent studies of seaweed-coral competition emphasize effects of seaweed allelopathy (11, 12) (chemical suppression of competitors), seaweeds vectoring coral diseases (13, 14), and near-contact creating anoxic zones or enhancing detrimental microbes on corals (14, 15). These mechanisms all require close contact for seaweeds to damage corals. Thus, millimeter- to centimeter-scale differences in proximity may cause large differences in coral health (11, 12, 15). Just as mutualist ants on Acacia trees protect their host by removing nearby competitors (16), we reasoned that the goby or pomacentrid fishes that shelter in many Acroporid corals (17) might play a similar function and remove seaweed competitors from coral edges.

We therefore focused on the common coral Acropora nasuta and asked the following: (i) whether commensal fishes sheltering in Acropora suppressed an allelopathic seaweed competitor, (ii) whether different commensal fish species varied in the protection they provided the coral, (iii) whether the interaction was affected by a specialist crab that lives only in the allelopathic seaweed (18), (iv) if the coral chemically cued the fishes to remove toxic seaweed at sites of contact, and (v) whether mutualistic fish that consume the toxic seaweed become more toxic to generalist predatory fishes.

We assessed how coral-dwelling fishes affected seaweed-coral interactions in the field (19) by placing the allelopathic seaweed (12) Chlorodesmis fastigiata versus a control for shading and abrasion (an algal mimic made of nylon line) in contact with A. nasuta colonies occupied by four different commensal fishes (n = 20 corals per fish species). We then evaluated coral health at the coral-algal or coral-control area of contact using pulse-amplitude modulated (PAM) fluorometry to assess coral photophysiology as a proxy for coral health (12, 20). In corals occupied by the gobies Gobiodon histrio or Paragobiodon echinocephalus, C. fastigiata abundance declined by 30% over 3 days and the damaging effect of C. fastigiata on A. nasuta declined by 70 to 80% compared with A. nasuta colonies lacking gobies (Table 1 and Fig. 1). In contrast, the control had minimal effect (Fig. 1). The alga’s specialist crab (18), Cyphyra rotundifrons, had no effect on these interactions. C. fastigiata was found in the gut of 17 of 20 G. histrio and 0 of 20 P. echinocephalus from corals that were contacting C. fastigiata2 = 29.57, df = 1, P < 0.001); thus, G. histrio consumed C. fastigiata, whereas P. echinocephalus removed the seaweed but did not consume it. Guts of G. histrio or P. echinocephalus occupying corals not contacting C. fastigiata (n = 20 each) were devoid of C. fastigiata. Given that the allelopathic compounds from C. fastigiata are hydrophobic and the alga must contact the coral for these to be transferred (11, 12), gobiid removal of C. fastigiata filaments contacting the coral should lessen or prevent coral damage, which is what we found in our field experiments (Fig. 1). For this interaction to be broadly important, goby occupancy of A. nasuta would need to be frequent. We assessed this by running eight haphazardly placed 30- × 2-m transects across the reef and evaluating goby occupancy of all A. nasuta located in these transects. Gobies occurred in 81 ± 16% (mean ± 1 SD) of the 207 colonies assessed. An assessment in Australia also indicated common co-occurrence, with 1593 Gobiodon individuals occurring in the 1373 colonies of 11 Acropora species evaluated (17).

Table 1

Effects of coral-associated gobies and an alga-associated crab on abundance (volumetric displacement) of the alga C. fastigiata after 3 days of exposure (± SEM) (N = 20).

View this table:
Fig. 1

Effects of gobies (A) G. histrio and (B) P. echinocephalus on algal-coral interactions. Mean (± 1 SE) percent difference in effective quantum yield of the coral holobiont when exposed to the seaweed C. fastigiata or an inert mimic compared to the control location on the coral that was exposed to no treatment. P value is from a split-plot analysis of variance (ANOVA) (arcsine-transformed data). Letters designate significant groupings.

Because G. histrio produces a toxic skin secretion, whereas P. echinocephalus does not (21), we tested the effect (22) of G. histro mucus on two model predators that consume a variety of invertebrates and small fishes to see if potency of G. histro secretions increased after consumption of C. fastigiata. We placed mucus from one disturbed G. histrio into 300 ml of seawater with the cardinal fishes Ostorhinchus nigrofasciatus (n = 20) and Nectamia similis (n = 10). The mucus of both control and C. fastigiata–exposed G. histrio produced significant effects; however, secretions of G. histrio from coral heads contacting C. fastigiata caused predators to lose equilibrium (falling forward or sideways) more than twice as fast (66 ± 4 and 65 ± 8 s, respectively) as mucus of G. histrio from corals without C. fastigiata (143 ± 9 and 162 ± 11 s, respectively; P < 0.001 for each species, t test). The toxins producing the effects are unknown, but thin-layer chromatograms of extracts from C. fastigiata and from G. histrio mucus did not show secondary metabolites from C. fastigiata in G. histrio mucus.

In contrast to the advantage that both gobies provide coral by removing C. fastigiata, the coral-sheltering damselfishes Dascyllus aruanus or Chromis viridis provided no advantage (fig. S1). All D. aruanus and C. viridis abandoned C. fastigiata–treated A. nasuta within 24 to 48 hours; none abandoned colonies treated with the mimic alone (P < 0.001 for each species, n = 20, Fisher’s exact test). Initial presence or absence of fish had no effect on photosynthesis of the coral holobiont when in contact with C. fastigiata (D. aruanus F1,38 = 2.221, P = 0.144; C. viridis F1,38 = 2.234, P = 0.147). By day 3 of the experiment, contact with C. fastigiata had damaged corals and suppressed effective quantum yield by ~80%, whether or not damselfishes had been present. In contrast, algal mimics suppressed photosynthesis by only ~5% (F1,38 = 2280.66, P < 0.001 for both fish species), which indicated the primacy of chemical, as opposed to physical, effects.

To determine whether fish were responding to chemical cues from the seaweed or the coral, we used 60-ml syringes to pull in situ seawater from: among the filaments of C. fastigiata alone, the C. fastigiata–A. nasuta contact area with C. fastigiata still present, the C. fastigiata–A. nasuta contact area after removing C. fastigiata 20 min earlier (allowing loss of algal odor but retention of odor from the damaged coral), and the water column well away from the benthos (as a control) and then slowly released these odors into corals containing G. histrio. Olfactory cues from C. fastigiata alone generated no response by the goby. In contrast, odors from the coral-algal contact point or from the stressed coral alone caused 17 and 19, respectively, of the goby pairs in 20 separate A. nasuta colonies to move toward the odor source. Thus, the goby responds to chemical cues from the host coral, not to cues from the seaweed (Fig. 2, Υ = 559.12, df = 2, P < 0.001; G test,).

Fig. 2

Response of the goby G. histrio to chemical cues from: the alga C. fastigiata alone, C. fastigiata in contact with the coral A. nasuta, or the damaged coral that had been in contact with C. fastigiata but with the C. fastigiata removed 20 min before the odor was collected. Right side of graph is the same experiment with the same types of cues from Acropora millepora introduced to gobies living in A. nasuta. P value from a G test. Letters indicate significant groupings.

The same experiment conducted with odors from Acropora millepora produced no responses from G. histrio living in A. nasuta (Fig. 2). Thus, gobies responded to cues from their host species, but not to odors from a closely related coral, even one that G. histrio sometimes occupies.

Because G. histrio effectively defended its host and was the most common goby in A. nasuta, we conducted assays evaluating how rapidly A. nasuta cued its goby symbionts and whether the coral would signal in response to the seaweed’s chemistry alone. C. fastigiata damages Acropora species via hydrophobic compounds including acetylated diterpenes (12). We obtained the hydrophobic crude extract from C. fastigiata via extraction in methanol followed by partitioning between water and ethyl acetate, removed the solvent in vacuo, redissolved the ethyl acetate partition in ether, coated this lipid-soluble extract onto algal mimics at natural volumetric concentration (12), and placed extract-treated mimics against A. nasuta–harboring G. histrio. Control mimics treated with the same solvent but without the algal extract also were placed against the coral. Gobies rapidly moved to the site of contact between the extract-treated mimic and the coral (Fig. 3). Fifteen minutes after contact, 70% of the 20 goby pairs were beneath the treated mimics, this increased to 95% by 30 min; movement to the control ranged from 0 to 10% (P = 0.001 at 30 min; Kolmogorov-Smirnov test). Patterns in Fig. 3 indicate that the coral signaled in response to C. fastigiata compounds alone and that the signal was produced within 5 to 15 min of coral exposure.

Fig. 3

Response of G. histrio to A. nasuta in contact with algal mimics treated with natural concentrations of the lipid-soluble extract from C. fastigiata (treatment) versus control algal mimics treated with solvent only and placed against the coral host. P value from a Kolmogorov-Smirnov test of the 30-min data.

Thus, the gobies serve as bodyguards for host corals, and the coral chemically cues gobies to attract them to the site of coral-algal contact where they begin removing the alga within minutes of seaweed contact (or contact by the seaweed’s hydrophobic extract alone). Gobies are not attracted to cues from C. fastigiata alone nor to cues from related corals in contact with C. fastigiata; they respond only to odors from their host species. Symbiotic gobies that spend their adult life in a single coral played this protective role; damselfishes did not. Just as terrestrial plants release volatile signals that attract predators of herbivores (23), A. nasuta releases chemicals that cue symbiotic gobies to remove a competing, allelopathic seaweed. We could find no previous example of a species chemically cuing consumers to remove its competitors.

As corals have declined and seaweed cover has increased on reefs over recent decades, understanding seaweed-coral competition has become more important (3, 24). Because Acroporid corals are major builders of topographic complexity on coral reefs, they play critical roles as foundation species (5, 7), creating critical habitat that is associated with the diversification of numerous lineages of reef fishes (25). Coral-dwelling gobies facilitate persistence of these corals despite increased competition from seaweeds. These small, inconspicuous fishes may have effects considerably larger than their mass would predict.

The coral-goby relation appears similar to terrestrial ant-plant symbioses as exemplified by ants and Acacia trees (16). Ants receive food and shelter from their host Acacia and protect the host from competitors and consumers. Symbiotic gobies have a similar relationship with Acropid corals. Several gobies consume coral tissue (5 of the 19 species of Gobiodon are corallivores), but no species feeds exclusively on coral (26, 27). Most Acroporid coral colonies host at least one pair of gobiids (17), these fish remain in the same coral colony for most of their adult life, and death of host corals is commonly correlated with goby population decline (17). Gobies consume coral tissue [they also consume copepods (26) and algae growing against the coral base (27)]. Thus, like ants on Acacia, they receive shelter and food from their host, which they protect from a damaging competitor.

Both gobies we investigated protected their host by removing C. fastigiata; however, only G. histrio consumed the alga, which contains metabolites that deter feeding by numerous reef herbivores (18, 28). These findings may explain why P. echinocephalus removes, but does not consume, algal tissue in contact with its host coral. Consumption of this chemically noxious alga may benefit G. histrio by making its skin secretions more noxious to predators. However, metabolites from C. fastigiata are unlikely to be a primary source of skin toxins because G. histrio not exposed to C. fastigiata were also toxic, just less so.

As reefs continue to convert from coral to macroalgal dominance, there is increasing need to understand interactions that enhance coral resilience or suppress seaweed impacts on corals. Symbiotic gobies play a key role in defending Acroporid corals from an allelopathic alga, with chemical signals and cues mediating responses of both the coral (Fig. 3) and fish (Figs. 2 and 3). A worrisome recent discovery is that chemically mediated behaviors, such as these, that often are critical to reef function can be disrupted or even reversed (i.e., attraction to predator odors) by changes in ocean pH (29, 30). With ocean acidification (24), critical aspects of chemical communication in the sea may be destabilized, with the attendant loss of key processes underlying reef resilience.

Supplementary Materials

Materials and Methods

Fig. S1

References (3133)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: Support provided by NSF grant OCE-0929119, NIH grant U01-TW007401, and the Teasley Endowment to the Georgia Institute of Technology. We thank the Fijian government and the Korolevu-i-wai district elders for collection and research permissions; B. Devine, J. White, G. Longo, R. Bonaldo, and D. Rasher for field, lab, or statistical assistance; J. Kubanek, D. Rasher, P. Munday, J. P. Hobbs, and anonymous reviewers for comments on the manuscript; and V. Bonito for assisting scientifically and culturally. The data reported in this manuscript are presented in the main paper and supplementary materials.
View Abstract

Navigate This Article