Effect of substrate type on bacterial community composition in biofilms from the Great Barrier Reef


Correspondence: Verena Witt, Australian Institute of Marine Science (AIMS), PMB No. 3, Townsville MC, Queensland 4810, Australia. Tel.: +61 (0)7 47534483; fax: +61 (0)7 47725852; e-mail: v.witt@aims.gov.au


Natural and anthropogenic impacts such as terrestrial runoff, influence the water quality along the coast of the Great Barrier Reef (GBR) and may in turn affect coral reef communities. Associated bacterial biofilms respond rapidly to environmental conditions and are potential bioindicators for changes in water quality. As a prerequisite to study the effects of water quality on biofilm communities, appropriate biofilm substrates for deployment in the field must be developed and evaluated. This study investigates the effect of different settlement substrates (i.e. glass slides, ceramic tiles, coral skeletons and reef sediments) on bacterial biofilm communities grown in situ for 48 days at two locations in the Whitsunday Island Group (Central GBR) during two sampling times. Bacterial communities associated with the biofilms were analysed using terminal restriction fragment length polymorphism (T-RFLP) and clone library analyses of 16S rRNA genes. Findings revealed that substrate type had little influence on bacterial community composition. Of particular relevance, glass slides and coral skeletons exhibited very similar communities during both sampling times, suggesting the suitability of standardized glass slides for long-term biofilm indicator studies in tropical coral reef ecosystems.


Similar to coastal regions worldwide, local natural and anthropogenic impacts such as land runoff from agriculture deliver inorganic nutrients, sediments, freshwater and pesticides to the coastal and coral reef waters of the Great Barrier Reef (GBR) (Bell, 1991), and thereby influence the water quality of this ecosystem. Coral reefs harbour abundant bacterial biofilms that are crucial catalysts of biogeochemical nutrient cycling (Battin et al., 2003) and are therefore critical to reef ecosystem functioning. This underlines the necessity to understand community composition and function of microorganisms within coral reef-associated biofilms.

Marine biofilms are complex microbial communities comprising of surface-attached microorganisms embedded in an extracellular polymeric matrix (Mihm et al., 1981). The bacterial communities within biofilms respond rapidly to changing environmental conditions, and therefore bacterial community composition of artificially and field grown biofilms have previously been used as bioindicators for water quality in freshwater (Campbell et al., 2011), estuarine (Jones et al., 2007; Nocker et al., 2007) and temperate and polar coastal marine environments (Moss et al., 2006; Webster & Negri, 2006; Dang et al., 2008). In addition, biofilms may also be potential bioindicators for water quality in tropical coastal coral reef ecosystems (Kriwy & Uthicke, 2011).

Previous research addressed the composition of bacterial communities in marine biofilms in response to various environmental parameters such as the effects of nutrients (Chiu et al., 2008), tides (Dobretsov & Qian, 2006), water depth (Webster et al., 2004), salinity and temperature (Lau et al., 2005; Chiu et al., 2006). These studies have neglected to examine the effect that the settlement substrate has on the composition of the developing bacterial community and used artificial substrates, i.e. polystyrene dishes or glass slides only. Only two invertebrate larval settlement studies from harbour waters investigated the effect of different substrates and showed that bacterial communities in biofilms undergo temporal shifts from more different communities during colonization and early developmental stages to more similar communities over time irrespective of the initial substrate type (Huggett et al., 2009; Chung et al., 2010). These studies were, however, limited to only artificial substrates, i.e. glass slides coated in different chemicals to simulate different ‘wettability’ properties, deployed at one site only (Huggett et al., 2009) or subtidal biofilms on two substrates, i.e. granite and petri dishes, at one deployment time only (Chung et al., 2010). Therefore, although these studies have shed some light onto the effects of substrates on bacterial community compositions in marine biofilms, inferences on the suitability of various substrates for future studies cannot be drawn. This is especially the case for water quality bioindicator research, where substrates are required which on the one hand simulate or reproduce naturally occurring biofilm assemblages, but on the other hand are easy to deploy and sample and provide a standardized surface.

This study therefore evaluates the effects of various substrates on the bacterial community composition in biofilms from tropical coral reef ecosystems with the aim of providing better rationale for future bioindicator studies of water quality in these types of ecosystems. The criteria for the choice of substrate include ease of handling and removal of biofilm from the substrate, standardized size and resemblance of developed bacterial communities to those found on ‘natural’ substrates. We specifically examined bacterial community compositions using the molecular fingerprinting method terminal restriction fragment length polymorphism (T-RFLP) on two ‘artificial’ substrates, i.e. ceramic tile and glass slides, which are frequently used in aquatic biofilm studies, and two ‘naturally occurring’ substrates that were collected directly from the coral reef sampling area, i.e. coral skeletons and reef sediments. Furthermore, the study extends previous knowledge by covering a more realistic time period for indicator biofilm development (i.e. 48 days), by incorporating temporal and spatial variability. Biofilm samples were collected in summer and winter (representing the annual water temperature extremes) at two locations (each at the end of a described inshore to offshore water quality gradient) to ensure that findings are not restricted to a single location or season. This investigation therefore results in recommendations on the best biofilm substrate for long-term water quality monitoring studies in coral reefs.

Materials and methods

Study site and biofilm development

Four different substrates (glass slides, coral skeletons, reef sediments and ceramic tiles) were deployed for biofilm development. Glass microscope slides (Sail Brand) were pre-cleaned with 70% ethanol and fixed in polyvinyl chloride frames. Reef sediment (approximately 50 : 50 carbonate, silicate mixture) was collected at 8 m depth from near-shore islands (Long, Lindeman, Repulse) in the Whitsunday Islands and sieved to a grain size of <100 and >63 μm. The sediment was autoclaved and dried at 60 °C over night. Sediment was glued onto microscope glass slides with aquarium grade silicone (Selleys), dried for 24 h and fixed onto PVC frames. Coral cores from Porites sp. (cylinders of 2 × 2 cm) were autoclaved, and unglazed ceramic tiles were sterilized by a 30 min UV treatment on each side. This study followed a hierarchical sampling design. Each substrate was deployed in duplicates at two replicate sites (25 m apart) at both Daydream Island (inshore, S 20°15.345′ E 148°48.729) and Deloraine Island (offshore, S 20°09.457′ E 149°04.183) (Fig. S1), and therefore making four samples per substrate for each island. These two islands were positioned at each end of a previously described water quality gradient in the Whitsunday Islands of the central GBR (van Woesik et al., 1999; Cooper et al., 2007; Uthicke & Nobes, 2008; Uthicke & Altenrath, 2010; Kriwy & Uthicke, 2011). Daydream Island (a permanent site of the long-term Reef Plan Marine Monitoring Program) was positioned inshore in ‘low’ water quality and Deloraine Island was positioned offshore in ‘high’ water quality (Table 1). All parameters measured were generally lower during the winter dry season than the summer wet season and higher inshore at Daydream Island compared with offshore at Deloraine Island, except light and salinity, which showed the inverse trend. The water quality measurements are consistent with data obtained from the same monitoring sites along the water quality gradient from previous years (Cooper et al., 2007; Schaffelke et al., 2010). Substrates were deployed on two separate times (48 days during austral winter of August–October 2008, average temperature 21 °C and austral summer of January–February 2009, average temperature 29 °C) to represent annual water temperature extremes. In summary, there were two islands with two sites each where duplicate substrates were deployed. These were sampled at two different times giving a total of 16 samples per substrate. Substrates were deployed at 6 m water depth (below the lowest astronomical tide level) for c. 48 days, and were vertically mounted approximately 40 cm from the underlying sediment on steel pickets (covered by ziplock bags to avoid effects from leached iron) and secured by cable ties. For sample collection after c. 48 days of deployment, much of the biofilm material was carefully scraped off the substrates into cryovials using sterile No. 11 scalpel blades (yield was usually >2 g), snap-frozen in liquid nitrogen and stored at −80 °C until further processing.

Table 1. Summary of analyses from the water column, irradiance variables and nearest distance from the coast of the Whitsunday Island sites. Water quality parameters for each year and season shown are chlorophyll a, turbidity, total suspended solids (TSS), temperature, light, salinity, dissolved inorganic nitrogen (DIN, includes NO3, NO2 and NH4), DIP. Mean (standard deviation)
IslandDistance from coast (km)Season and yearChlorophyll a (μg L−1)Turbidity (FLNTU)TSS (mg L−1)Temperature (°C)Light (mmol m−1 day−1)Salinity (ppt)DIN (μM)DIP (μM)
Daydream3Dry 20080.54 (0.12)1.31 (0.56)0.78 (0.21)22.13 (1.21)5.49 (17)36.09 (0.59)0.13 (0.07)0.11 (0.04)
Wet 20090.84 (0.11)2.80 (1.96)2.29 (0.41)28.43 (0.41)1.52 (30)32.63 (1.27)0.31 (0.12)0.08 (0.04)
Deloraine31.25Dry 20080.34 (0.20)0.78 (0.45)21.6 (0.58)8.39 (31)35.36 (0.43)0.08 (0.07)0.16 (0.05)
Wet 20090.53 (0.14)1.84 (0.60)27.71 (0.40)6.34 (30)35.20 (0.54)0.14 (0.09)0.10 (0.07)

Water quality

Water quality samples were obtained and analysed as described in detail in Schaffelke et al. (2010) and Cooper et al. (2007). In short, duplicate samples from two depths at each location per sample time were analysed for dissolved inorganic nutrients (DIN, includes NH4, NO2, NO3), dissolved inorganic phosphorus (DIP), total suspended solids (TSS), chlorophyll a and salinity. For particulate nutrients and chlorophyll a analysis, water samples were collected on pre-combusted glass fibre filters and analysed after acetone extraction. Samples for determining TSS were collected on pre-weighed 0.4 μm polycarbonate filters, and TSS concentrations were determined gravimetrically. Salinity was determined using a Portasal Model 8410A Salinometer (Guildline). Autonomous water quality instruments (Eco FLNTUSB Combination Fluorometer and Turbidity loggers; WET Labs, Philomath, OR) recorded turbidity (optical backscatter) and in situ temperature data. Light was measured using Odyssey light loggers equipped with wiping units as described in Uthicke & Altenrath (2010).

Genomic DNA extraction

Total DNA was extracted from 0.5 g (wet weight) of each biofilm sample using the MoBio UltraClean Soil Kit (MoBio Laboratories, Solana Beach, CA) according to the manufacturer's protocol with the following modifications. Bead-beating (Mini-Bead-Beater, Biospec Products, Bartleville, OK) (2 × 30 s) cycles were performed, 900 mL of S3 buffer was used and DNA was eluted from the column with 2 × 50 μL of 1 × TE buffer. DNA extracts were examined using standard 1% agarose gel electrophoresis and quantified using a Nanodrop Spectrophotometer (Thermo Fisher Scientific, Waltham, MA).

PCR amplification, cloning and sequencing

Bacterial 16S rRNA genes were amplified by PCR using the general bacterial 16S rRNA gene primers 63F (5′-CAGGCCTAACACATGCAAGTC-3′) and 1389R (5′-ACGGGCGGTGTGTACAAG-3′) (Sigma-Proligo, The Woodlands, TX) (Marchesi et al., 1998). Each sample was amplified in triplicate 25 μL reactions containing 2.5 μM non-acetylated bovine serum albumin (New England Biolabs, Biolabs, USA), 2 μM (2 mM each) dNTP (Astral Scientific, Australia), 2.5 μM forward primer 63F, 1.25 μM reverse primer 1389R, 1 μM MgCl2 (Qiagen, Germany), 1.25 U HotStar Taq (Qiagen), 2.5 μL HotStar Buffer (Qiagen) and c. 2 ng of template DNA. Amplification was performed with an initial incubation at 95 °C for 15 min, followed by 30 cycles of 94 °C for 1 min, 55 °C for 1 min, 72 °C for 90 seconds and a final extension at 72 °C for 10 min.

As T-RFLP profiles from glass slides and coral skeletons were very similar, only communities from glass slides were cloned. Four clone libraries of bacterial 16S rRNA genes amplified from DNA extracted from biofilms grown on glass slides were constructed, and represent one library for each season at each location. Therefore, the clone libraries represent (i) inshore at Daydream Island during summer, (ii) inshore at Daydream Island during winter, (iii) offshore at Deloraine Island during summer and (iv) offshore at Deloraine Island during winter. Triplicate PCR reactions were performed for each of the four replicate biofilm samples from each of these representative two sampling locations (total of eight) and two sampling times (overall total 16), and were pooled accordingly for construction of the four clone libraries. Samples were then purified using the MinELUTE PCR Clean-Up Kit (Qiagen) and cloned using a TOPO-TA Cloning Kit (Invitrogen) according to the manufacturer's instructions. Afterwards, blue-white screening colonies were checked for correct insert size using a colony PCR method using primers 63F/1389R. Per clone library, 96 randomly picked clones were then dispersed in LB media and 10% glycerol in 96-well plate format and sent to the Australian Genome Research Facility Ltd. (Brisbane, Australia) for purification and sequencing using an ABI3730 XL Automatic DNA Sequencer.

Retrieved sequences were trimmed and analysed manually using Chromas Lite 2.33 (Technelysium Pty Ltd., Australia), and submitted to the Greengenes NAST Aligner (DeSantis et al., 2006) for alignment of sequences to the Greengenes database. Greengenes NAST-aligned 16S rRNA gene sequences were checked for chimeras using bellerophon Version 3 (Huber et al. 2004), and identified chimeras were excluded from further analysis. The NAST-aligned 16S rRNA gene sequences were submitted to the Greengenes batch sequence classifier [http://greengenes.lbl.gov/cgi-bin/nph-classify.cgi], and taxonomic assignments for each sequence were recorded using NCBI taxonomy. All sequences were submitted to the GenBank Database (Accession numbers: JF261700JF262029).

Terminal restriction fragment length polymorphism analysis

Bacterial 16S rRNA genes were PCR amplified using the same reaction mixture and conditions as outlined for clone libraries, except that fluorescently labelled 5′Cy5-labelled 63F (Sigma-Aldrich) was used (adapted from Wilson et al., 2008). Each individual biofilm sample was amplified in three replicate PCR reactions. The amplicons were pooled, purified and quantified as above. Each purified product (150 ng) was digested with the restriction enzyme MspI (New England Biolabs) according to the manufacturer's instructions. Digested fragments were desalted using the DyeEx 2.0 Spin Kit (Qiagen) and vacuum dried for 40 min at low temperature in the dark. Terminal restriction fragments (T-RFs) were resolved and visualized using the CEQ 8800 Genetic Analysis System (Beckman-Coulter, Fullerton, CA) with a 600 bp size standard (Beckman-Coulter). Replicate samples were compared using the software T-align (Smith et al., 2005) with a range of 0.5 bp peak area to determine the consensus peaks between duplicates. The relative fluorescence intensity of the peak area of T-RFs was used as a relative abundance measure of dominant T-RFs in further statistical analyses detailed below. For verification of T-RFs, purified DNA from individual clones were analysed by T-RFLP using the same protocol as for environmental samples, except that 75 ng of digested PCR products generated from each clone was used. Each clone produced a single peak (T-RF) that was then manually matched with T-RFs identified from whole community T-RFLP analyses. Prior to statistical analyses, T-RF peak area values were third root transformed and standardized. Principal Component Analysis (PCA) was used to determine whether bacterial assemblages in samples grouped by substrate, location and/or season. The significances of assemblage dissimilarities between substrates, seasons and locations were tested by applying one-way Analysis of Similarity (anosim) based on permutation procedures using the Bray–Curtis distance measure. The contributions of each taxon to the total dissimilarities of treatments were analysed using the Similarity Percentage (SIMPER) routine. All analyses were performed using the past statistical software (Hammer et al., 2001). One-way analysis of variance (anova) was performed using the ncss 2007 (NCSS) statistical software to determine significant differences between relative abundances (peak area) for taxa at different locations.


The effect of substrate type on bacterial community structure in biofilms was examined using T-RFLP for the whole dataset (pooled from both sampling times and locations). Biofilm communities were very similar, regardless of the settlement substrate. PCA analysis showed that bacterial communities were largely overlapping for all substrates. PCA analyses also suggested that biofilms grown on glass slides and coral skeletons were most similar to each other, whereas the reef sediments displayed the highest variability between replicate samples (Fig. 1). For the global dataset, no significant differences in community structure among substrates could be detected using anosim analysis (R = 0.039, P = 0.090). PCA analyses also suggested that similar community structures occurred among different substrates when sampling times were analysed separately (Fig. 2a and b), although small, but significant differences in bacterial community structures on different substrates within both sampling times were determined (anosim summer: R = 0.122, P = 0.0316; winter: R = 0.175, P = 0.0059). For samples collected in winter, post hoc tests revealed that the only significant difference was between ceramic tile in comparison to reef sediments and coral skeletons (Table 2). Although the overall anosim test of different substrates for the summer was significant (R = 0.122, P = 0.037), post hoc tests showed no significant effect between individual substrates (P > 0.05) (Table 2). When the substrate data was compared for each location, the four substrates were statistically indistinguishable (anosim P = 0.0949) offshore at Deloraine Island, whereas inshore at Daydream Island significant differences were found for reef sediments compared with the remaining three substrates (P = 0.0009) (Fig. 3a and b).

Figure 1.

PCA incorporating relative abundances of T-RFs (using the relative fluorescence peak intensity matrix) showing bacterial assemblages of each substrate independent of location and season (1, reef sediment: filled triangle; 2, coral skeleton: square; 3, glass slide: filled square; 4, ceramic tile: triangle). Fifteen per cent of the vectors are shown in the biplot and are labelled with the bacterial taxon.

Figure 2.

PCA incorporating relative abundances of T-RFs (using the relative fluorescence peak intensity matrix) showing bacterial assemblages of each substrate at different sampling events (a) winter and (b) summer. (1, reef sediment: filled triangle; 2, coral skeleton: square; 3, filled glass slide: square; 4, ceramic tile: triangle).

Figure 3.

PCA incorporating relative abundances of T-RFs (using the relative fluorescence peak intensity matrix) showing bacterial assemblages for all substrates at one (a) inshore location (Daydream Island) and one (b) offshore location (Deloraine Island). (1, reef sediment: filled triangle; 2, coral skeleton: square; 3, glass slide: filled square; 4, ceramic tile: triangle).

Table 2. Table showing R and P values of the analysis of similarity (anosim) on the relative abundance of T-RFs (using the relative fluorescence peak intensity matrix) as derived using T-RFLP grouped by substrate type for each sampling time (summer and winter). R and P values for the summer are above the diagonal, and those within the winter are below. Overall anosim for the summer: R = 0.122, P = 0.030; winter R = 0.175, P = 0.006. P-values <0.05 are highlighted for clarity
 Reef sedimentCoral skeletonGlass slideCeramic tile
Reef sediment R = 0.154 P = 0.065R = 0.146 P = 0.080R = 0.064 P = 0.239
Coral skeletonR = 0.146 P = 0.063 R = 0.054 P = 0.198R = 0.093 P = 0.163
Glass slideR = 0.073 P = 0.165R = 0.077 P = 0.147 R = 0.203 P = 0.102
Ceramic tileR = 0.371 P = 0.009R = 0.416 P = 0.004R = 0.131 P = 0.1044 

Although it was not the focus of the study, differences in bacterial community structures between the two sampling locations were examined to determine if the T-RFLP method is able to detect differences among bacterial assemblages that are assumed to be due to differences in water quality. A PCA clearly separated the bacterial assemblages between the two locations and the two sampling times (Fig. 4). Replicates from each location were more variable during summer than winter, and more variable offshore than inshore (Fig. 4).

Figure 4.

PCA incorporating relative abundances of T-RFs (using the relative fluorescence peak intensity matrix) showing bacterial assemblages for different locations (inshore Daydream Island and offshore Deloraine Island) and sample times (winter and summer). (1, inshore winter: square; 2, inshore summer: filled square; 3, offshore winter: triangle; 4, offshore summer: filled triangle). Fifteen per cent of the vectors are shown in the biplot and are labelled with the bacterial taxon.

This result was confirmed using anosim, which revealed significant differences between locations (R = 0.544, P = 0.0177) and sampling times (R = 0.299, P < 0.0001).

The length of the species-vectors in the PCA biplot and a SIMPER analysis consistently indicated that T-RFs representing the Roseobacter clade (Roseobacter and Silicibacter), Erythrobacter, Hyphomonas, Gammaproteobacteria and diatom plastids contributed mostly to the dissimilarities (54.9%) between substrates at different seasons and locations (Fig. 1) and between locations and sampling times despite substrate type (Fig. 4).

Overall, 37 T-RFs were identified, of which, 89.2% could be assigned to clones that were taxonomically identified from the clone libraries (within ±0.5 bp) (Supporting Information Table S1), and thus could be assigned to a bacterial taxon. All T-RFs detected were present in the glass slide profiles.


T-RFLP, cloning and sequencing of 16S rRNA genes revealed that coral reef-associated biofilms comprised of complex bacterial and microalgal communities. Relatively similar, although not always identical bacterial community structures were present on different substrate types over two sampling times (during a summer and a winter). Bacterial community composition on reef sediments differed significantly from the other substrate types at the inshore location that was influenced by pronounced changes in water quality during different seasons. Reef sediments also showed the largest variability in bacterial community composition among all investigated substrates. This suggests that reef sediments may have low reproducibility and is therefore not suitable for bioindicator studies in coral reefs in comparison to other more ideal substrates. Relatively variable bacterial community compositions were also identified on ceramic tiles in comparison to the other substrates during winter, suggesting that ceramic tiles are also not ideal substrates for bacterial biofilm bioindicator studies. In contrast, glass slides and coral skeletons substrates produced comparably stable and highly reproducible community compositions independent of sampling time and/or location.

Another aspect of substrate choice is the practical requirement for a simple method for the removal of total and/or near complete biofilm biomass from the actual substrate. Removal of biofilms from rough, uneven surfaces such as those of the ceramic tile, reef sediment and coral skeleton is rather difficult. Although coral skeletons represent the most natural of all tested substrates, when regarding the ease of handling and removal of the biofilm, glass slides have the clear advantage in that their smooth, flat surfaces enable simple and rapid removal of most of the biofilm biomass. Considering that bacterial community structures on coral skeletons and glass slides were not significantly different, we propose the use of glass slides for future bioindicator studies.

Both spatial and seasonal influences (i.e. changes in water quality including light, salinity, turbidity, chlorophyll α) on bacterial community structures may have been responsible for some of the variability among certain substrates, rather than the actual substrate type. We suggest that all of the substrate types used in this study have relatively little influence on the bacterial community composition when examined after the relatively long deployment period (c. 48 days). Types of bacteria initially colonizing and settling on specific substrates may be different depending on the surface properties of the substrate, however, biofilms undergo distinct temporal shifts, where the effect of substrate type diminishes, and tend to form more similar community structures over time (Huggett et al., 2009; Chung et al., 2010). In the present study, distinct bacterial communities were identified at the two different locations suggesting that discrete bacterial communities develop in response to the different environmental parameters found at the different locations rather than different substrates. As our study sites were positioned at either ends of a clearly formed water quality gradient that is known from a continuous long-term monitoring program (Uthicke & Altenrath, 2010; Uthicke et al., 2010; Kriwy & Uthicke, 2011) and from recently measured data (Table 1), we propose that this response was caused by differences in water quality at the two locations. The rationale to collect samples from two islands (representing extremes of a previously studied water quality gradient) and at two sampling times (representing the annual extremes in water temperature) was merely to test for substrate differences under a variety of environmental conditions, and thus extends the validity of this study.

Given that differences between the bacterial community compositions at different sites could be easily detected, reproducible patterns among replicates were produced, and tentatively 89.2% of the taxonomic affiliations of the T-RFs after comparison to sequence data produced from clone libraries were identified. This study therefore suggests that T-RFLP is a suitable and rapid, high-throughput fingerprinting method for detecting spatio-temporal and water quality-induced bacterial community shifts. Further support is given by the fact that dominant bacterial taxa identified using this method (e.g. Roseobacter, Rhodobacteraceae) were similar to those found in previous aquatic biofilm studies using glass slides (Dang & Lovell, 2000; Jones et al., 2007).

In summary, this study suggests that when biofilms are subjected to long-term deployment (weeks to months), as presented here, simple glass slides enable the formation of bacterial biofilm communities that are highly similar to other ‘natural’ substrates such as coral skeletons or reef sediment grains. Additional advantages for the use of glass slides include a standardized size, low cost, ease of handling and the formation of relatively reproducible bacterial community structures among replicates. This study therefore also provides further evidence that monitoring bacterial communities associated with coastal biofilms may find application as a bio-monitoring tool for environmental management for examining local and regional changes in water quality in the long-term. Future work should include more in-depth studies of the bacterial communities grown in different water qualities over replicate seasons.


We thank C. Humphrey, C. Reymond, F. Patel and J. van Dam for assistance in the field and the crew of the R.V. Cape Ferguson for the assistance during fieldwork. The water quality data were collected as part of the Reef Plan Marine Monitoring Program, which is supported by the Great Barrier Reef Marine Park Authority (GBRMPA) through funding from the Australian Government's Caring for our Country and by the Australian Institute of Marine Science (AIMS). We are grateful to I. Zagorskis for summarizing the water quality data and K. Wasmund for his critical and helpful comments on the manuscript. This project (project 3.7.1) was funded by the Australian Government Marine and Tropical Sciences Research Facility (MTSRF).