Coral Reefs

, Volume 27, Issue 1, pp 37–47

Cross-shelf variation in the role of parrotfishes on the Great Barrier Reef


    • Australian Research Council Centre of Excellence for Coral Reef Studies and School of Marine and Tropical BiologyJames Cook University
  • D. R. Bellwood
    • Australian Research Council Centre of Excellence for Coral Reef Studies and School of Marine and Tropical BiologyJames Cook University

DOI: 10.1007/s00338-007-0287-x

Cite this article as:
Hoey, A.S. & Bellwood, D.R. Coral Reefs (2008) 27: 37. doi:10.1007/s00338-007-0287-x


Herbivorous fishes are a key functional group on coral reefs. These fishes are central to the capacity of reefs to resist phase shifts and regenerate after disturbance. Despite this importance few studies have quantified the direct impact of these fishes on coral reefs. In this study the roles of parrotfishes, a ubiquitous group of herbivorous fishes, were examined on reefs in the northern Great Barrier Reef. The distribution of 24 species of parrotfish was quantified on three reefs in each of three cross-shelf regions. Functional roles (grazing, erosion, coral predation and sediment reworking) were calculated as the product of fish density, bite area or volume, bite rate, and the proportion of bites taken from various substrata. Inner-shelf reefs supported high densities but low biomass of parrotfishes, with high rates of grazing and sediment reworking. In contrast, outer-shelf reefs were characterised by low densities and high biomass of parrotfish, with high rates of erosion and coral predation. Mid-shelf reefs displayed moderate levels of all roles examined. The majority of this variation in functional roles was attributable to just two species. Despite being rare, Bolbometopon muricatum, the largest parrotfish species, accounted for 87.5% of the erosion and 99.5% of the coral predation on outer-shelf reefs. B. muricatum displayed little evidence of selectivity of feeding, with most substrata being consumed in proportion to their availability. In contrast, the high density of Scarus rivulatus accounted for over 70% of the total grazing and sediment reworking on inner-shelf reefs. This marked variation in the roles of parrotfishes across the continental shelf suggests that each shelf system is shaped by fundamentally different processes.


BioerosionHerbivorySediment reworkingCoral predationEcosystem processFunctional redundancy


It is widely accepted that herbivory plays a critical role on coral reefs, mediating the competition for space between corals and benthic macro-algae (e.g., Steneck 1988; Mumby 2006; Hughes et al. 2007). Roving herbivorous fishes, namely parrotfishes, surgeonfishes, rabbitfishes and drummers, have been identified as the key functional group performing this role. Irrespective of whether they are assimilating, or even ingesting algal material, the mechanical removal of this material through their feeding activities maintains the standing biomass of benthic algae at relatively low levels (Hatcher 1983; Klumpp and McKinnon 1992; Hughes et al. 2007). The role of these fishes is becoming increasingly important given the additional pressures to which corals reefs are being exposed, e.g., coral bleaching, crown-of-thorns starfish outbreaks, disease, eutrophication, sedimentation and overfishing. It is often a combination of these factors that ultimately leads to a phase shift from coral to macro-algal dominated reefs, as has been documented on both Caribbean and east African reefs (Hughes 1994; McClanahan et al. 2001; Mumby et al. 2006). In both of these regions the loss of large herbivorous fishes through overfishing underpinned the demise of these reefs. Even in the absence of additional pressures, the experimental exclusion of these fishes has been demonstrated to induce a shift to a macro-algal dominated state on the Great Barrier Reef (Hughes et al. 2007).

Despite the importance of roving herbivorous fishes on coral reefs, few studies have quantified the role of these fishes on ecosystem processes. The abundance and community structure of roving herbivorous fishes has been shown to vary latitudinally (e.g., Floeter et al. 2005), among habitats within a single reef (e.g., Russ 1984a), and between reefs from differing shelf locations (e.g., Russ 1984b). This spatial variation has often been cited as a causative factor in the distribution of marine macro-algae across similar scales, with large fleshy macro-algae (predominantly Phaeophyta) dominating on inshore or coastal reefs, and turf and coralline algae dominating on oceanic or mid- and outer-shelf reefs (e.g., Klumpp and McKinnon 1992; McCook and Price 1997). Many studies have assumed that herbivorous fishes all perform a similar role, or have a similar impact on the system (reviewed by Choat 1991). There is, however, considerable variation in feeding behaviour of these fishes both among and within families. The majority of the surgeonfishes (f. Acanthuridae), rabbitfishes (f. Siganidae) and drummers (f. Kyphosidae) tend to bite or “crop” the algae leaving the basal portions intact (Choat et al. 2002). In contrast, the parrotfishes (f. Labridae) possess a unique oral morphology that typically allows them to remove portions of the underlying substratum together with algae, leaving areas of bare substrata (Bellwood and Choat 1990). Within the parrotfishes three distinct functional groups have been identified based on the osteology and myology of their oral and pharyngeal jaws: croppers, scrapers and excavators (Bellwood and Choat 1990; Bellwood 1994). These functional groupings relate directly to the amount of the substratum that is removed through the feeding action. Croppers remove only algae and associated epiphytic material, whereas scrapers and excavators remove pieces of the substratum together with the algae, leaving distinctive bite scars. Scrapers leave relatively shallow bite scars (<1 mm), whereas excavators leave deeper bite scars and remove greater quantities of substrata with each bite. The ingested material is further processed by the pharyngeal jaws resulting in a reduction in the size of the ingested particles (Bellwood 1996). Through these mechanisms parrotfishes not only impact the standing biomass of algae but also provide areas of clean substrata for the colonization by benthic organisms, and contribute to the external erosion of the reef framework (Bellwood 1995a; Bruggemann et al. 1996).

These three functional groups have profoundly different impacts on reefs as a consequence of their variation in morphology and feeding mode. The aim of this study therefore, was to quantify the roles of parrotfishes on reefs in the northern Great Barrier Reef, and to describe how these roles varied both across the continental shelf and among habitats within a single shelf location. In particular, the roles of parrotfish in grazing and eroding the reef substrata, coral predation, and sediment reworking were examined.

Materials and methods

Distribution of fishes

To quantify the distribution of parrotfishes a series of timed swims were conducted between November 1998 and February 1999 in the northern section of the Great Barrier Reef (GBR). Three reefs were censused in each of three cross-shelf locations: inner-, mid-, and outer-shelf (8–21, 22–36, and 46–55 km from the mainland, respectively). Details of the reefs surveyed are provided in Bellwood and Wainwright (2001). Four habitats (slope, crest, flat and back; see Bellwood and Wainwright 2001 for definitions) were censused on each of the mid- and outer-shelf reefs. The absence of a clearly defined crest on the inner-shelf reefs precluded the distinction between slope and crest habitats and subsequently only three habitats (slope/crest, flat, and back) were censused on each of the inner-shelf reefs.

Four censuses were conducted in each habitat on each reef, with adjacent censuses being separated by a minimum of 20 m. Where possible, censuses in each habitat were conducted within the same cross section of reef. Each census consisted of a diver (always DRB) swimming parallel to the reef crest for 20-min and recording all parrotfishes greater than 10 cm total length (TL) within a 5 m wide transect that extended from the reef substratum to the surface of the water. This census method was selected to minimise diver effect and maximize the area covered to ensure detection of larger parrotfish species. Individual fishes were identified and placed into 5 cm size categories. Care was taken not to re-census fish that left and subsequently re-entered the transect area. The numbers per unit effort were converted to densities per unit area by estimating the length of each transect using a differential GPS (mean = 235 m, see Bellwood and Wainwright 2001). Density estimates were converted to biomass using known length–weight relationships for each species (Kulbicki et al. 1993, 2005).

Data analysis

Variation in parrotfish abundance and biomass across shelf locations, habitats and reefs was analysed using two three-factor ANOVAs. Type IV sums of squares were used to adjust for missing habitat on inner-shelf reefs. Assumptions of the ANOVA were examined by residual analysis. Abundance was log10 transformed, and biomass \( \sqrt[4]{{}} \) transformed to improve normality and homoscedasticity.

To characterize major parrotfish faunas, a principal component analysis (PCA) was used to examine any differences in community composition between habitats and reefs. This analysis was based on the mean number of individuals per 100 m2 at each habitat within each reef, 33 “sites” in total (three shelf locations, three reefs per shelf location, three to four habitats per reef, mean of four transects per habitat). Data were standardized to individuals per 100 m2 to account for differences in transect length. The analysis was based on the covariance matrix of log10(x + 1) transformed data. To provide an objective description of the site grouping a Ward’s hierarchical cluster analysis was performed on the squared Euclidean distances of the log abundance data. The significance of the clusters was estimated following Sandland and Young (1979). This method represents a more conservative approach to the identification of clusters than simply relying on hierarchical clustering methods (e.g., Williams and Hero 1998).

Functional roles

Parrotfishes have previously been classified as being excavators (Bolbometopon muricatum, Cetoscarus bicolor and Chlorurus spp), scrapers (Hipposcarus spp and Scarus spp), and croppers (Calotomus spp, and Leptoscarus vaigiensis) based on the osteology and myology of their oral and pharyngeal jaws (Bellwood and Choat 1990; Bellwood 1994). There is considerable variation in the maximum size of the excavating species that is directly related to the amount of material they remove from the reef. Subsequently we further divided the excavators into three sub-groups: B. muricatum (to 1,200 mm total length, TL), Chlorurus microrhinos and C. bicolor (to 800 mm TL), and Chlorurus bleekeri, Chlorurus japanensis and Chlorurus sordidus (to 400 mm TL). Cross shelf and cross reef variations in two biological roles (area grazed and coral predation) and two physical roles (erosion and sediment reworking) were examined for the scraping and three excavating functional groups. Only adult fishes (i.e., B. muricatum >500 mm TL; Chlorurus spp, C. bicolor, Hipposcarus longiceps, and Scarus spp >150 mm TL) were included in the analyses to provide conservative estimates of all functional roles. The extremely low number of cropping individuals censused precluded analysis of the group.

All functional roles were calculated using bite scar dimensions, bite rates, feeding day lengths and fish densities. The volume of the bite scars (n = 288) of scraping species were estimated by examining small pieces of coralline algae collected from the reef crest and reef flat on Lizard Island, northern GBR. Both scraping and excavating parrotfishes leave conspicuous feeding marks and can be readily distinguished from each other. Scraping species typically leave two parallel feeding marks on the substrata with the length being approximately ten times the width. Excavating species leave scars that are approximately twice as long as they are wide, with four to six deep grooves running parallel to their major axis (Bellwood and Choat 1990). The length and width of each bite scar was measured using dial calipers. The depth of each bite scar was measured by fracturing the substrata perpendicular to the bite scar and viewing the cross-section under a stereo dissecting microscope with an eyepiece graticule. The density of the coralline algae means that these are conservative estimates of depth. The bite scar dimensions for all excavating species, bite area for scraping species, and the bite rates and feeding day lengths for all species were taken from the literature (Bellwood and Choat 1990; Bellwood 1995a; Bellwood et al. 2003; Fox and Bellwood 2007). Where possible species-specific rates were used, when these were not available rates for congeners were used. Overall means and error terms were calculated using an expanded three-term version of Goodman’s estimator following Bellwood (1995a). The compound error term combines all estimated components of variation from each parameter into a single estimate of the overall sample variance, thereby providing an objective assessment of the variation around the overall means. The use of this compound error term precluded statistical comparisons among means.

Biological roles

The total area grazed by each species in each habitat was calculated as the product of daily feeding rate (i.e., feeding rate × feeding day length), bite area, fish density, and the proportion of bites on turf- and coralline-algae covered substrata. Similarly, coral predation was calculated as the product of daily feeding rate, bite volume, fish density, and the proportion of bites on live coral. The proportion of bites on each substrate type for all species was taken from the literature (Bellwood and Choat 1990; Bellwood 1995b). Feeding selectivity was also quantified for B. muricatum, the only parrotfish considered to be a major coral predator (Bellwood et al. 2003). Twelve 30 m transects were laid haphazardly on the reef crest of two outer-shelf reefs, Yonge and Hicks reefs, and an area approx. 50 cm either side of each transect was recorded using an underwater digital video camera. The reef crests were chosen as they are the preferred feeding habitats for B. muricatum (Bellwood et al. 2003). The substrata immediately under the transect tape was identified and recorded as a proportion of the 30 m transect length. Live corals were placed into the following categories based on their growth form: Acropora “humilis” group, Acropora“isopora” group, Montipora spp, Pocillopora spp (primarily Pocilloporaverrucosa), and other coral. B. muricatum feeding scars within each 1 m × 30 m transect were identified and the substratum from which they were taken was recorded. The selectivity of feeding was estimated using a resource selection ratio (“the forage ratio”: Savage 1931 in Manly et al. 2002). The forage ratio compared the proportion of bites taken from each substratum type to the proportion of each substratum type available.

Physical roles

Reworking of sediment was calculated as the product of daily feeding rate, bite area, fish density, proportion of bites on turf- and coralline-algae and the sediment load for that reef zone. The sediment loads for each reef zone were taken from Purcell and Bellwood (2001). These estimates were based on a single mid-shelf reef and are likely to over- and under-estimate the relative sediment loads for outer- and inner-shelf reefs, respectively. Bioerosion was calculated as the product of daily feeding rate, bite volume, carbonate density, and fish density. Gut content analyses were used to compare variation in the size distribution of the sediment produced by each group. Five B. muricatum and sixteen Scarus spp individuals were collected from reefs in the vicinity of Lizard Island using spears. Preparation of the gut contents for particle size analysis follows Bellwood (1996). Sediment size distributions for C. microrhinos and Chlorurus spp were taken from Bellwood (1996).


A total of 5,843 adult parrotfish from 24 species were recorded, equating to a mean density of 377 individuals per hectare. Parrotfish abundances exhibited a significant interaction between shelf location and habitat (F5,115 = 0.463, P = 0.008). The densities of adult parrotfishes were highest in the back reef and slope/crest habitats (6.7–8.5 ind. 100 m−2) of the inner-shelf reefs and generally decreased with distance offshore (Fig. 1a). The inner-shelf reef flat and four mid-shelf reef habitats displayed similar densities of parrotfishes (3.3–4.5 ind. 100 m−2). The lowest densities recorded were on the outer-shelf reefs with the slope, crest and back reef habitats having similar densities (1.8–2.3 ind. 100 m−2), and the reef flat having the lowest density of all habitats censused (0.6 ind. 100 m−2). This pattern is primarily driven by the greater abundance of scraping parrotfishes (i.e., Scarus spp) on inner-shelf reefs, especially the deeper back reef and slope/crest habitats where they accounted for 91.6 and 94.1% of the total parrotfish densities, respectively. Surprisingly, of the seventeen scraping species recorded across all transects only three species (Scarusrivulatus, Scarusghobban and Scaruspsittacus) had higher densities on inner-shelf reefs when compared to mid- and outer-shelf reefs. Collectively these three species accounted for approximately 94% and S. rivulatus 80% of all scraping parrotfishes on inner-shelf reefs. Of the remaining fourteen scraping parrotfish species eleven had highest densities on mid-shelf reefs, and three species had highest densities on outer-shelf reefs. The abundance of excavating parrotfishes was greatest on mid- and outer-shelf reefs, especially the shallow reef flat and crest habitats. Parrotfish biomass was found to differ significantly among shelf locations (F2,115 = 47.12, P = 0.046) and habitats (F3,115 = 99.77, P = 0.0003). Further details of the ANOVA results can be found in the electronic supplementary material. The biomass of adult parrotfishes was greatest on the shallow outer-shelf reef flat and crest habitats where it was approximately three to eight times greater than that of any other habitat (Fig. 1b). B. muricatum, the largest of the excavating species, accounted for the majority (88%) of the biomass in both shallow outer reef habitats and was not recorded in any other habitat (Fig. 1b). There was little variation in the total biomass of parrotfishes between inner- (0.7–2.0 kg 100 m−2) and mid-shelf (1.1–2.0 kg 100 m−2) reefs. Scraping parrotfishes accounted for 87% of the total biomass on inner-shelf reefs and only 50% on mid-shelf reefs. In contrast the relative contribution of C. microrhinos to total biomass increased from 10% on inner-shelf reefs to 42% on mid-shelf reefs.
Fig. 1

Variation in a mean density and b mean biomass of parrotfish across the continental shelf in the northern Great Barrier Reef. Each mean is based on four transects from each of three reefs (n = 12). The overall means are divided into the relative contributions of each of the four groups identified. Black, Bolbometopon muricatum; hatched, Chlorurus microrhinos; grey, Chlorurus spp; open, Scarus spp; B back reef; F reef flat; C reef crest; S reef slope

The principal component analysis showed clear cross-shelf trends in parrotfish community composition, with the first two components explaining 36.7 and 26.2% of the total variation, respectively (Fig. 2). There was a clear separation of inner-shelf habitats from both mid- and outer-shelf habitats along the first principal component (Fig. 2a) and this division was supported by the cluster analysis (P = 0.0005). There was some separation of outer- and mid-shelf habitats along the second principal component (Fig. 2a), which was again supported by the cluster analysis (P = 0.0327). The community composition of two of the three outer-reef slopes and one outer-reef crest was found to be more similar to that of the mid-shelf habitats than the other outer-shelf reef habitats (Fig. 2a). Inner shelf reefs were characterised by three scraping species; S. rivulatus and to a lesser extent S. ghobban and S. psittacus (Fig. 2b). Outer-shelf reef flats and two of the outer-shelf reef crests were separated from all mid-shelf habitats by the presence of B. muricatum (Fig. 2b). The mid-shelf reefs were characterised by a diversity of both scraping and excavating parrotfishes. Analysis of parrotfish community composition using non-metric multidimensional scaling (MDS) and analysis of similarities (ANOSIM) revealed similar cross-shelf patterns to the PCA and cluster analysis (see electronic supplementary material).
Fig. 2

Principal component analysis showing the relationships among parrotfish assemblages across nine reefs in the northern Great Barrier Reef. a Ordination plot showing the relationship between 33 sites. Each site is based on four 20-min transects. Ellipses represent significant grouping identified by the cluster analysis. I inner-shelf reef; M mid-shelf reef; O outer-shelf reef; B back reef; F reef flat; C reef crest; S reef slope; numbers refer to replicate reefs. b Species loadings showing the relative contribution of each species to the observed differences in parrotfish assemblages. Species names: Bolbometopon muricatum, Chlorurus microrhinos, Scarus rivulatus, Scarus ghobban, Scarus psittacus, Scarus altipinnis, Chlorurus sordidus, Scarus schlegeli, Scarus frenatus, Scarus niger, Scarus oviceps

Biological roles

There was a general decrease in parrotfish grazing rates from inner- to outer-shelf reefs, although there was considerable variation among habitats within inner- and outer-shelf locations (Fig. 3a). Estimates of grazing on inner-shelf reefs suggest that each square metre of reef is completely grazed 8.5–11.0 times per year on the back reef and reef slope/crest, and 4.9 times per year on the reef flat (Fig. 3a). The lowest grazing rates were on the reef slope, flat and back reef (1.9–2.4 times per year) of outer-shelf reefs. Grazing rates were relatively constant on mid-shelf reefs ranging from 5.2 to 6.3 times per year across all habitats (Fig. 3a). S. rivulatus alone accounted for >70% and collectively scraping parrotfishes (Scarus spp) accounted for 90–97% of the total grazing on inner-shelf reefs. The proportion of the total area grazed by scraping parrotfishes decreased to 54–74% on mid-shelf reefs. The contribution of scraping species on outer-shelf reefs was highly variable, accounting for 11.5–25.8% of total area grazed in the reef crest and flat habitats, and 68.1–84.9% in the reef slope and back reef habitats.
Fig. 3

Cross-shelf variation in the biological roles of parrotfish on the northern Great Barrier Reef: a proportion of substratum grazed per year; b mass of live coral consumed per year. The mass of live coral refers to the mass of live skeleton removed. Overall error terms calculated using Goodman’s estimator following Bellwood (1995a). Definitions of axis labels and species are given in Fig. 1

Coral predation rates showed marked peaks on the outer-shelf reef flat and crest habitats, 12.7 and 15.0 kg m−2 year−1, respectively (Fig. 3b). This high rate of coral removal was solely attributable to B. muricatum. Coral predation rates in all other habitats were at least two orders of magnitude lower. Live scleractinian coral (predominantly Acropora “isopora” group, Pocillopora spp and A. “humilis” group) accounted for almost fifty percent of the total B. muricatum bites recorded on the outer-shelf reef crests (Fig. 4). The remaining bites were taken from epilithic algae (37%) and coralline algae (15%). The feeding of B. muricatum displayed very little evidence of selectivity, with most substratum types being bitten in similar proportions to their availability. The exception to this was Montipora spp that was consistently avoided (Fig. 4). Despite constituting 13% of benthic substrata on the outer-shelf reef crest, only 3.5% of all B. muricatum bites were recorded on this coral.
Fig. 4

Mean proportion of bites taken from seven substrata categories by Bolbometopon muricatum on outer-shelf reef crests. A = Acropora. Other coral included Acropora formosa, Acroporanasuta, Millipora sp., Stylophora sp., Favites sp., Porites sp., Goniastrea sp., and Pocillopora eydouxi. Each mean is based on 12 30 m × 1 m transects. Asterisk denotes substratum for which the selection ratio was significantly less than one (i.e., the substratum was avoided)

Physical roles

The estimated mass of sediment reworked displayed high variability among habitats within each shelf location, with an overall trend of decreasing mass with distance from the coastline (Fig. 5a). The highest estimates of sediment reworking were recorded for the back reef (41.7 kg m−2 year−1) and slope/crest (25.6 kg m−2 year−1) of inner-shelf reefs and the back reef of mid-shelf reefs (30.4 kg m−2 year−1). The lowest rates (0.7–0.9 kg m−2 year−1) were recorded for the reef crest on the mid- and outer-shelf reefs. Scarus spp were responsible for most of the sediment reworking on the inner- and mid-shelf reefs (91.5 and 60.2%, respectively), while only accounting for 41.7% of the sediment reworking on outer-shelf reefs. Within each shelf location the back reef had the highest and on the mid- and outer-shelf reefs the reef crest the lowest estimates of sediment reworking (Fig. 5a).
Fig. 5

Cross-shelf variation in the physical roles of parrotfish on the northern GBR: a mass of sediment reworked per year; b annual rates of primary erosion. Overall error terms calculated using Goodman’s estimator following Bellwood (1995a). Definitions of axis labels and species are given in Fig. 1

Estimated erosion rates showed a steady increase from inner- (0.9–3.8 kg m−2 year−1) to mid-shelf (5.2–8.4 kg m−2 year−1) reefs, and then a marked increase on the outer-shelf reef crest (32.3 kg m−2 year−1) and flat (23.1 kg m−2 year−1, Fig. 5b). There was very little erosion on the outer-shelf reef slope or back reef habitats (0.8–1.8 kg m−2 year−1). Patterns in erosion were almost solely attributable to two species; C. microrhinos and B. muricatum. C. microrhinos accounted for 67.3 and 90.4% of all erosion on inner- and mid-shelf reefs, respectively. Despite being only present in two outer-shelf reef habitats B. muricatum accounted for 87.5% of the erosion on outer-shelf reefs and 53.9% of the total cross-shelf erosion. The third group of excavators, Chlorurus spp, only accounted for 3.1% of the total cross-shelf erosion.

In terms of sediment production, the gut contents of Scarus spp and C. sordidus displayed similar sediment size distributions, with the highest fraction being the mud (<63 μm) component and very little sediment greater than 1,000 μm for both groups (Fig. 6). In contrast, C. microrhinos had the highest fraction in the 125–1,000 μm size classes, and approximately eight percent of the total sediment was greater than 1,000 μm. B. muricatum displayed an even distribution of sediment among the various size classes, with over 27% of total sediment greater than 1,000 μm (Fig. 6).
Fig. 6

Particle size distributions of sediments from the intestines of Bolbometopon muricatum (n = 5), Chlorurusmicrorhinos (n = 6), Chlorurussordidus (n = 8), and Scarus spp (n = 16). Values for C. microrhinos and C. sordidus were taken from Bellwood (1996)


The roles of parrotfishes on reefs in the northern Great Barrier Reef display clear cross-shelf variation. Inner-shelf reefs were characterised by high densities but low biomass of parrotfishes, with high levels of grazing and sediment reworking, whilst outer-shelf reefs had high biomass but low densities of parrotfish, with high rates of coral predation and erosion (predominantly on the shallow reef crest and reef flat). Much of this variation in functional roles amongst shelf locations could be attributed to the variation in community structure at each scale. Inner-shelf reefs were characterised by high abundances of scraping parrotfishes, S. rivulatus, and to a lesser extent S. ghobban and S. psittacus. The outer-shelf reef crests and flats were characterised by a single species, B. muricatum. In contrast, the mid-shelf reefs supported a higher diversity of both scraping and excavating species with 15 of the 24 species censused having highest densities on mid-shelf reefs.

The higher levels of grazing and sediment reworking on inner-shelf reefs was directly attributable to the higher abundance of scraping parrotfishes, particularly S. rivulatus, on those reefs. Density estimates from the present study contrast markedly with those of earlier studies that reported significantly lower densities of parrotfishes on inshore reefs when compared to mid- and outer-shelf reefs in the central GBR (Williams and Hatcher 1983; Russ 1984b). Whilst these latter density estimates were based on both timed visual census (Russ 1984b) and explosive sampling (Williams and Hatcher 1983) they only sampled a single habitat (3–9 m depth) on the inshore reefs. All three habitats surveyed in the present study supported densities of parrotfishes that were equal to, or greater than those on mid-shelf reefs, and greater than those on outer-shelf reefs. Variation in underwater visibility on inshore reefs, which typically ranges from 2 to 10 m, may have contributed to the differences in density estimates on inshore reefs between the present study and those of Russ (1984b). However, it is unlikely that such differences could explain the lower densities of parrotfish recorded on outer-shelf reefs in the present study. The present study was conducted on reefs situated approximately 400 km north of the reefs sampled in these earlier studies. Latitudinal variation in the community structure of herbivorous fishes has been well documented in the Atlantic Ocean (Floeter et al. 2005), and may explain the differences between these regions on the Great Barrier Reef.

Variation in grazing intensity has often been cited as a significant factor in shaping the succession and productivity of algal communities (e.g., Scott and Russ 1987; Steneck 1988). Models of algal community succession predict that reduced levels of grazing will result in a shift from grazing resistant crustose coralline algae to highly productive algal turfs and ultimately to slower growing climax species such as Sargassum, Tubinaria and Padina (Steneck and Deither 1994; McClanahan 1997). Interestingly, the highest estimated rates of parrotfish grazing were recorded in the only habitats where large fleshy macroalgae (primarily Sargassum spp) is seasonally abundant; inshore slope/crest and back reef habitats. The lowest estimated levels of grazing intensity were recorded on the outer-shelf reefs where the cover of crustose coralline algae is generally the highest (Fabricius and De’ath 2001). This apparent contradiction may reflect the grazing mode of parrotfishes. Recent evidence suggests that parrotfishes have only a limited role in controlling the proliferation of fleshy macroalgae as they display a limited capacity to graze larger, or established, macroalgae (Bellwood et al. 2006; Mantyka and Bellwood 2007). Furthermore, when considering variation in grazing intensity in structuring algal communities all herbivorous groups should be considered. Density and biomass estimates of acanthurids are at least one order of magnitude greater on outer-shelf reefs than mid- and inner-shelf reefs in the central GBR. Those of siganids, while considerably lower than acanthurids, are generally greatest on inshore reefs in the same region (e.g., Williams and Hatcher 1983; Russ 1984b). In addition, many environmental factors also vary across the continental shelf which may influence the relationship between grazing intensity and algal community composition. These include sedimentation, eutrophication, wave action and water flow.

Estimates of grazing intensity on inner-shelf reefs suggest that each square metre of reef would be completely grazed every 4–5 weeks on the slope/crest and 10–11 weeks on the reef flat (0.36 and 0.14% h−1, respectively). Fox and Bellwood (2007) reported similar rates on the reef crest and outer reef flat of Pioneer Bay, a leeward inshore reef in the central GBR (0.46 and 0.23% h−1, respectively). Overall rates across the GBR range from 0.004 to 0.46% h−1 (Fox and Bellwood 2007). These stand in marked contrast to Caribbean estimates. Computer models of Caribbean reefs estimated parrotfish grazing to range from 0.6% h−1 in Jamaica to 1.2–2.5% h−1 in Belize (Mumby 2006). Given a conservative 10 h feeding day this would equate to the reef being completely grazed every 17 days in Jamaica and every 4–8 days in Belize. These grazing rates are extremely high compared to GBR estimates, especially given that Jamaica has experienced systematic overfishing of herbivorous fishes (Hughes 1994) and estimates of total parrotfish biomass only range from 2.1 to 3.7 g m−2 on mid-depth forereefs (Williams and Polunin 2001). In the relatively unexploited reefs of Belize biomass estimates of 7.5–21 g m−2 (Mumby 2006) are broadly comparable to those reported in the present study (6.5–73 g m−2). Differences in parrotfish assemblages may explain some of the variation in grazing intensity between these biogeographic regions. But given the similar parrotfish biomass estimates the differences in estimated grazing rates are likely to reflect computational differences between studies.

In contrast to abundance and grazing, biomass, bioerosion and coral predation were greatest on the outer-shelf reef crest and flat, and decreased markedly at deeper outer-shelf habitats and all mid- and inner-shelf habitats. These patterns were largely driven by the presence of a single species, B. muricatum, on the shallow outer-shelf reef flat and crest. B. muricatum does occur on both inner- and mid-shelf reefs in the northern GBR (ASH, DRB personal observation), and as such we will have underestimated its role in those habitats. However, they typically occur in extremely low densities in these habitats and are unlikely to influence the overall trends. During the present study no B. muricatum individuals were seen on inner- and mid-shelf reefs despite surveying over 1.4 ha of reef within each habitat at each shelf location (approximately 10 ha in total).

The rate of parrotfish bioerosion on outer-shelf reef flats and crests (23.0–32.3 kg m−2 year−1) approaches the highest estimates of calcification, or growth, recorded on Indo-Pacific reefs (23–35 kg m−2 year−1; reviewed by Barnes and Chalker 1990). This suggests that parrotfishes are consuming most, if not all, of the annual gross carbonate production in these habitats. Estimates of coral reef calcification range from 0.5 to 35 kg m−2 year−1, but are highest in shallow seaward habitats and decrease with both increasing depth and decreasing coral cover, with average rates being in the vicinity of 3–10 kg m−2 year−1 (Barnes and Chalker 1990). The lower bioerosion rates on mid- (5–8 kg m−2 year−1) and inner-shelf reefs (1–4 kg m−2 year−1) may reflect the lower gross calcification rates in these habitats. Excavating parrotfishes avoid flat and concave surfaces when feeding (Bellwood and Choat 1990; Bellwood 1995b; Bellwood et al. 2003). Subsequently almost all of the erosion on inner- and mid-shelf reefs and approximately half of the erosion on outer-shelf reefs would be concentrated on dead corals and other algal covered protuberances on the reefs surface (the remaining erosion on outer-shelf reefs is from live coral). The loss of excavating parrotfishes through overfishing on many central Pacific reefs has been identified as a major contributing factor to the persistence of large areas of dead coral skeletons following disturbances (Wilkinson 2002; Bellwood et al. 2004).

Sea urchins and endolithic organisms (primarily sponges, bivalves, sipunculids, and polychaetes) also contribute to the erosion of reef substrata (e.g., Hutchings 1986; Risk et al. 1995). Erosion by these organisms differs markedly to that of parrotfishes in that they burrow into the reef matrix, compromising the structural integrity of the reef (Bellwood et al. 2004). The proliferation of sea urchins following the overfishing of herbivorous fishes has been well documented on both Caribbean and Indo-Pacific reefs (e.g., Hughes 1994; McClanahan et al. 2001). On these heavily impacted reefs the density of sea urchins may exceed 10 ind. m−2 (Scoffin et al. 1980; Glynn 1988), equating to average annual erosion rates of 4–10 kg m−2 year−1 (Bak 1990; Peyrot-Clausade et al. 2000). Whilst there are currently no estimates of sea urchin density or bioerosion on the GBR several studies have reported that they occur in low densities (Hutchings 1986; Done et al. 1991). Consequently they are unlikely to contribute significantly to bioerosion. Similarly, bioerosion by endolithic organisms is much lower than that by parrotfishes, with estimates ranging from 0.04 to 0.82 kg m−2 year−1 across the continental shelf in the northern GBR (Osorno et al. 2005; Tribollet and Golubic 2005). Low external erosion rates in the latter study (0.004–0.77 kg m−2 year−1) probably reflect the use of flat artificial substrata in relatively deep water (7–10 m). Overall, parrotfishes appear to be the primary bioeroders in all shelf habitats with rates closely matching patterns of calcification.

Annual coral predation rates of 13–15 kg m−2 on the outer-shelf reef flat and crests are likely to have significant impacts not only on the growth, mortality rates, and colony morphology of corals, but also on the benthic community composition. The largely non-selective feeding by B. muricatum, the only major coral predator within the parrotfishes, may promote coral diversity by reducing the abundances of competitively dominant or faster growing species, such as Pocillopora and the tabulate acroporids (cf. Lang and Chornesky 1990; Dullo 2005). The apparent avoidance of Montipora spp by B. muricatum may be a function of its encrusting/foliose growth form, which is dominated by flat and concave surfaces. By selectively removing the upper portions of the coral colonies, predation by B. muricatum will reduce colony drag and may substantially increase the resistance to dislodgement. Such coral predation may increase the resilience of reefs to damage by storms and exceptional wave energy (cf. Madin and Connolly 2006).

Four other Indo-Pacific parrotfish species have been observed to graze on the surface of live Porites spp (Bellwood and Choat 1990). However, grazing on live coral only accounted for a low proportion of the total bites for each species (C. microrhinos < 2%, C. bicolor, S. frenatus, and S. rivulatus < 1% of total bites; Bellwood and Choat 1990, Bellwood 1995b). This, coupled with their smaller bite volumes, means that these species had very little influence on overall patterns of coral predation. Feeding by other corallivorous taxa differs markedly to that of parrotfishes as they typically only remove the living tissue, leaving the coral skeleton largely intact (e.g., crown-of-thorns starfish: Keesing and Lucas 1992; butterflyfish: Harmelin-Vivien and Bouchon-Navaro 1983).

Variations in grazing and erosion rates had a direct influence on rates of sediment production. Total sediment production by parrotfish was high on the back reef of inner- and mid-shelf reefs (44.6 and 35.8 kg m−2 year−1 respectively) and on the reef flat and crest of the outer-shelf reefs (27.7 and 33.0 kg m−2 year−1 respectively). However, the source and size of the sediments produced differed markedly across the continental shelf. Reworked sediment accounted for 91% of total sediment production on inner-shelf reefs. Sediment trapped within epilithic algae is ingested and triturated in the pharyngeal jaws, reducing the particle size of the sediment (Bellwood 1996). The majority of the sediment reworking on inner-shelf reefs was performed by scraping parrotfishes (Scarus spp) with the sediment produced being dominated by the smaller particle-size classes (<63 μm). These smaller sediment fractions may be differentially lost from the reef by hydrological transport (Bellwood 1996). Ctenochaetus striatus (f. Acanthuridae) has also been reported to reduce the size of ingested sediments (Nelson and Wilkins 1988). In marked contrast to patterns of reworking, erosion accounted for over 90% of the total sediment production in the shallow outer-shelf reef habitats. The sediment produced by B. muricatum, the major excavating species in these habitats, was dominated by larger particle-size classes (>250 μm) with only a small fraction of the smaller particle-size class. Given the faster settling rate of larger particles (Dyer 1986) much of the sediment produced is likely to fall on the outer-shelf reef slope and crest, the common defecation sites for B. muricatum (ASH, DRB personal observation).

The results of the present study show marked cross-shelf variation in both the community composition and functional roles of parrotfishes in the northern GBR. Whilst previous studies have described cross-shelf variation in taxa (Williams and Hatcher 1983; Russ 1984b), feeding modes (Russ 1984b), and ecosystem processes (Williams et al. 1986; Scott and Russ 1987; Bellwood et al. 2003) among herbivorous fishes, this study is the first to quantify the extent of variation within a single clade. Given the observed differences in the feeding modes and diets of species within other herbivorous groups (e.g., Choat 1991; Purcell and Bellwood 1993; Choat et al. 2002) we may expect similar cross-shelf variation in the roles of these groups. An understanding of these patterns of variation in reef processes is critically important, as the management and preservation of coral reefs is based on maintaining processes. It is clear that the three regions of the GBR are shaped by fundamentally different biological processes.

Overall the GBR shows clear transitions from inner- to outer-shelf reefs in both taxa and, more importantly, processes. There is no such thing as a typical GBR reef. Each shelf region appears to operate with a fundamentally different set of dominant processes. Inner-shelf reefs were characterized by high grazing intensity and sediment reworking, and low rates of erosion and coral predation. Conversely, outer-shelf reefs exhibited the highest rates of erosion and coral predation, and the lowest rates of grazing intensity and sediment reworking. Mid-shelf reefs displayed moderate levels of all functional roles examined. Surprisingly the majority of this variation in functional roles was attributable to only 2 of the 24 parrotfish species censused. The presence of B. muricatum on the outer-shelf reef crest and flat, albeit in low densities, was almost solely responsible for the high rates of erosion and coral predation on outer-shelf reefs. In contrast, the high densities of S. rivulatus were primarily responsible for the high grazing intensity and rates of sediment reworking on inner-shelf reefs. This marked variation in the roles of parrotfishes across the continental shelf and the lack of functional equivalents highlights the importance of this group in reef processes. Furthermore, the apparent reliance on just two species suggests that there is low functional redundancy within the parrotfishes on inner- and outer-shelf reefs in the northern GBR, a pattern that is becoming increasingly common in coral reef ecosystems (Bellwood et al. 2003, 2006). As our understanding of reef processes increases the role of biodiversity in ecosystem processes becomes increasingly restricted to a few key players.


We thank M. Marnane, P. Osmond, S. Blake, S. Collard, J. Elliot, S. Job, E. Vyotpil for and Lizard Island research station staff for field support, the Australian Research Council for financial support (DRB) and J. Hoey, S. Wilson and S. Wismer and two reviewers for helpful comments or assistance.

Supplementary material

338_2007_287_MOESM1_ESM.doc (160 kb)
ESM1 (DOC 160 kb)

Copyright information

© Springer-Verlag 2007