OUP user menu

Diversities of coral-associated bacteria differ with location, but not species, for three acroporid corals on the Great Barrier Reef

Raechel A. Littman, Bette L. Willis, Christian Pfeffer, David G. Bourne
DOI: http://dx.doi.org/10.1111/j.1574-6941.2009.00666.x 152-163 First published online: 1 May 2009


Patterns in the diversity of bacterial communities associated with three species of Acropora (Acropora millepora, Acropora tenuis and Acropora valida) were compared at two locations (Magnetic Island and Orpheus Island) on the Great Barrier Reef to better understand the nature and specificity of coral–microbial symbioses. Three culture-independent techniques demonstrated consistent bacterial communities among replicate samples of each coral species, confirming that corals associate with specific microbiota. Profiles were also conserved among all three species of Acropora within each location, suggesting that closely related corals of the same genus harbor similar bacterial types. Bacterial community profiles of A. millepora at Orpheus Island were consistent in samples collected throughout the year, indicating a stable community despite temporal changes. However, DGGE and T-RFLP profiles differed on corals from different reefs. Nonmetric multidimensional scaling of T-RFLP profiles showed that samples grouped according to location rather than coral species. Although similar sequences were retrieved from clone libraries of corals at both Magnetic and Orpheus Island, differences in the relative dominant bacterial ribotypes within the libraries drive bacterial community structure at different geographical locations. These results indicate certain bacterial groups associated specifically with corals, but the dominant bacterial genera differ between geographically-spaced corals.

  • Keywords
  • Keywords
  • coral
  • symbiosis
  • host specificity
  • microbial community conservation


Bacteria associated with corals are both diverse and abundant (Shashar et al., 1994; Ritchie & Smith, 1995, 1997; Santavy, 1995; Kushmaro et al., 1996; Rohwer et al., 2001, 2002; Frias-Lopez et al., 2002); however, the current understanding of how this diversity varies among coral species is limited. Microbial communities occupy a range of niches on corals, from within the surface mucus layer (Ducklow & Mitchell, 1979; Paul et al., 1986; Ritchie & Smith, 1995, 2004; Bourne & Munn, 2005) to both on and within the coral tissue layers (Williams et al., 1987; Shashar et al., 1994; Kushmaro et al., 1996; Banin et al., 2000; Frias-Lopez et al., 2002); however, their functional roles in relation to the coral host are virtually unknown. A variety of mutualistic benefits have been suggested, including fixation and passage to the coral host of nitrogen and carbon (Williams et al., 1987; Shashar et al., 1994; Cooney et al., 2002; Rohwer et al., 2002; Lesser et al., 2004), as well as other nutrients (Knowlton & Rohwer, 2003) and secondary metabolites such as antibiotics (Castillo et al., 2001). In addition, a recent study has shown that some types of bacteria may exclude undesirable microbial organisms. Mucus from healthy Acropora palmata was found to contain antibiotic-producing bacteria and inhibit the growth of potentially pathogenic microorganisms (Ritchie, 2006). The growing recognition that microbial communities associated with corals represent an important component of coral symbioses highlights the need to better understand the nature and specificity of these microbial symbioses. Therefore, it is necessary to identify which bacteria are conserved as mutualistic partners and what factors might drive the structure of the coral microbial community i.e. geography, environmental factors or differences in coral physiology.

Culture-independent techniques, in particular retrieval of 16S rRNA gene sequences, have enabled investigators to identify a wide range of microbial groups associated with corals. For example, Rohwer et al. (2002) examined three Caribbean species and found 430 distinct ribotypes from 14 coral samples. Additional studies examining other coral species from different geographic regions have noted similar diverse bacterial assemblages associated with corals (Rohwer et al., 2001; Frias-Lopez et al., 2002; Bourne & Munn, 2005). Although the abundance and types of bacterial groups have varied among coral species, there is also some evidence that the same coral species from geographically separate reef environments host the same bacterial communities (Rohwer et al., 2002). These findings support emerging generalizations that corals harbor species-specific bacterial communities that are geographically consistent. However, current studies have focused on corals from different families, raising questions about specificity at the level of coral species. Bacterial communities associated with corals have also been shown to be dynamic in response to environmental changes (Ritchie & Smith, 1995, 2004; Pantos et al., 2003; Barash et al., 2005; Koren & Rosenberg, 2006; Bourne et al., 2008), although the significance of shifts in microbial composition to coral health is currently unknown. There is a growing need to evaluate the taxonomic level and geographic extent to which bacterial communities are conserved on corals; in other words, whether coral species maintain specific bacterial consortia.

Greater knowledge of the bacterial communities associated with reef-building corals will aid our understanding of this multispecies mutualism and will help to identify which species play a key role in maintaining coral health. This study aimed to examine coral-associated bacterial communities to test assumptions about specificity in coral-bacterial associations. We compare bacterial profiles recovered from three closely related species of Acropora, Acropora millepora, Acropora tenuis and Acropora valida, to determine whether genetically similar corals differ in the structure of their bacterial communities. Bacterial profiles were compared between two locations on the Great Barrier Reef (GBR; Magnetic and Orpheus Island reefs) to identify which bacteria might be conserved across geographically distinct locations. Acropora millepora samples were additionally collected throughout 1 year on Orpheus Island to investigate whether temporal environmental changes would lead to natural variation in coral bacterial community composition. Three culture-independent 16S rRNA gene profiling methods [clone library construction, denaturing gradient gel electrophoresis (DGGE) and terminal restriction fragment length polymorphism (T-RFLP)] were used to cross validate our findings and provide an accurate assessment of bacterial community composition.

Materials and methods

Sample collection and processing

Three replicate healthy colonies of each of three coral species, i.e. A. millepora, A. tenuis and A. valida, were tagged on the reef flat (2–4 m depth) in Nelly Bay, Magnetic Island (19°10′S 146°50′E), and Pioneer Bay, Orpheus Island (18°35′S 146°29′E), Australia. Both are inshore, fringing reefs in the central GBR, although Magnetic Island reefs are more coastal and subjected to greater seasonal sea temperature variation, sedimentation and nutrients (Larcombe et al., 1995). All three coral species were sampled within the same transect at each site with replicate colonies of each species spaced 3–10 m apart to ensure that colonies were subject to the same environmental factors. Coral branches were collected from the center of each colony and placed in plastic bags underwater. Samples were rinsed with sterile artificial seawater (ASW) to remove loosely associated microorganisms from the water column, placed in new bags and frozen in liquid nitrogen for transport. Samples were later airbrushed (80 psi) with 2 mL of sterile ASW and the tissue slurry was aliquoted into two cryovials and stored at −80 °C. Samples used for species comparison were collected in May 2007 when seawater temperatures at the two sites were between 25 and 26 °C. In addition, replicate A. millepora colonies located on Orpheus Island were further sampled in July, October and December 2007 and February 2008 to assess temporal variability. Seawater temperatures were measured using data loggers positioned both at Orpheus Island and Magnetic Island and were associated with the Australian Institute of Marine Science long-term monitoring data collection. Temperatures during this time ranged between 20 °C in the Austral winter (beginning June) and 31 °C in the summer (ending in February).

DNA extraction and purification

DNA was extracted by suspending 200 μL of coral tissue slurry in 0.5 mL of buffer (0.75 M sucrose, 40 mM EDTA, and 50 mM tris, pH 8.3) and following the extraction protocol outlined in Bourne et al. (2008). The DNA pellet was suspended in 30 μL of sterile Milli-Q water and the total volume was loaded on a 1.2% low-melting agarose gel. DNA was purified using electrophoresis and cutting high-quality DNA (>2 kb) from the gel. The agarose was then removed from the sample using the QIAquick gel extraction kit (Qiagen, Hilden, Germany), following the manufacturer's instructions. DNA was recovered from the Qiagen column with two 30-μL washes of sterile Milli-Q water.

PCR amplification of bacterial 16S rRNA genes

Bacterial-specific primers 63f and 1387r (Marchesi et al., 1998) were used to amplify the 16S rRNA genes from extracted DNA for bacterial clone library construction. Amplification of the 16S rRNA genes for T-RFLP analysis was performed using the Beckman D4-labelled 63F primer and 1389R primer. The PCR mixtures (50 μL) contained 0.2 pmol μL−1 of each primer, 200 mM each deoxynucleoside triphosphate, 1 × PCR buffer [Tris-Cl, KCl, (NH4)2SO4, and 1.5 mM MgCl2], 0.08% (w/v) bovine serum albumin (BSA) and 1.25 U of Taq polymerase (Scientifix, Clayton, Vic., Australia). PCR was performed with an Applied Biosystems 2720 thermocycler and programmed with an initial 3-min step at 94 °C and 35 cycles consisting of 94 °C for 1 min, 55 °C for 1 min and 72 °C for 1.5 min, and a final extension for 10 min at 72 °C.

For DGGE, the bacterial 16S rRNA gene was amplified using primers 1055f and 1392R-GC (Ferris et al., 1996). PCR reactions (50 μL) consisted of 0.5 μM of each primer, 100 μM of each deoxyribonucleotide triphosphate, 0.08% (w/v) BSA, 1 × PCR buffer [Tris-Cl, KCl, (NH4)2SO4, and 1.5 mM MgCl2], 1.5 mM MgCl2 and 1.25 U Hotstar Taq (Qiagen). DGGE PCR reactions were carried out using an Eppendorf Mastercycler (Eppendorf, Hamburg, Germany) thermocycler. Temperature cycling was performed using a touchdown protocol (Ferris et al., 1996) with one cycle of 95 °C for 15 min, 10 cycles of 94° for 1 min, 53 °C (each cycle decreasing by 1 °C) for 1 min and 72 °C for 1 min, followed by 20 cycles of 94 °C for 1 min, 43 °C for 1 min and 72 °C for 1 min.

Clone library construction

The amplified bacterial DNA was ligated into the TOPO-TA cloning vector (Invitrogen, Carlsbad, CA), following the manufacturer's instructions. Ligations were submitted to the Australian Genome Research Facility for transformation, cloning and subsequent sequencing. Ninety-six clones were sequenced from each library using the M13f primer.

DGGE analysis

Bacterial profiling was carried out using an INGENY phorU-2 (Ingeny International BV, the Netherlands) DGGE system. PCR products were separated on gels containing 6.5% acrylamide with a 50–70% linear gradient of formamide and urea, using 0.5 × TAE buffer (0.02 M Tris base, 0.01 M sodium acetate and 0.5 mM Na2 EDTA; pH adjusted to 7.4). The buffer was preheated to 60 °C before sample loading, and the electrophoresis was run at 30 V for 20 min to draw the DNA into the gel before running the buffer through the system. Electrophoresis was then run at 60 °C for 16 h at 70 V. Gels were removed and stained for 10 min with an SYBR Gold nucleic acid stain (Molecular Probes Inc., Eugene, OR) in 1 × TAE buffer. Gels were destained by rinsing with 1 × TAE buffer and subsequently photographed using a UV transilluminator.

Clear bands were excised from the gel and placed in 100 μL of sterile Milli-Q water to elute DNA from the acrylamide gel. The DNA bands were reamplified and run on the DGGE gel to ensure correct migration and purity of the product. Products that showed one distinct band with the correct mobility on the DGGE were directly submitted for sequencing. Sequencing reactions were performed by Macrogen Inc. (Seoul, Korea) using the 1055F oligonucleotide as the sequencing primer.

T-RFLP analysis

Before restriction digestion, three replicate PCR products for each sample were purified using the QIAquik PCR Purification Kit (Qiagen). The restriction reaction mix was prepared with 1 × NEB Buffer 4 (Scientifix), 1 × BSA and 10 U of Hha1 and added to 15 μL of PCR product. Samples were incubated at 37 °C for 5 h and subsequently heat inactivated for 20 min at 65 °C. The digested DNA was precipitated from the solution by adding 2 μL of NaOAc (pH 5.2) and 50 μL of 95% ice-cold ethanol and centrifuged at 13 000 r.p.m. for 5 min at 4 °C. The ethanol was removed and the DNA was again washed with 100 μL of 70% ethanol. All ethanol was removed and the sample was air dried until residual alcohol had evaporated. The DNA pellets were resuspended in a solution containing 0.25 μL of size standard (600 bases) and 39.75 μL of SLS and loaded onto a 96-well sample plate. Each digest was then overlaid with one drop of mineral oil to avoid evaporation.

Digested samples were separated on a Beckman Coulter CEQ 8800 sequencer (Fullerton, CA). Fragment analysis was performed on an eight capillary array in fragment analysis mode using Beckman Coulter CEQ 8000 software. The following parameters were set: modified fragment 4, injection time 20 s with a ramp of 2 kV for over 2 min and run at 4.9 kV at 60 °C for 60 min. Peak size and retention times were analyzed using a quartic model. The threshold for relative peak height was set at 20% of the height of the second highest peak to remove any spurious artifact peaks. Replicate samples were compared using T-align (Smith et al., 2006) with a range of 0.5 peak area to determine the consensus peaks between duplicates. The list of fragments for each sample was then converted into a binary matrix.

Sequence analysis

Sequences were checked for chimera formation with the check_chimera software of the Ribosomal Database Project (Maidak et al., 1996). Sequence data were aligned to the closest relative using the blast database algorithm (Altschul et al., 1997). Sequence affiliations were determined by >97% identity to bacterial 16S rRNA gene sequences in the GenBank database.

Statistical analysis

Statistical analyses were carried out using past statistical software (Ryan et al., 1995). T-RFLP profiles were examined using nonmetric multidimensional scaling (nMDS) to determine whether sample profiles grouped according to species or location. A Euclidean similarity measure was used to generate a two-dimensional plot. A principal components analysis (PCA) was used to analyze clone libraries to determine which dominant bacterial ribotypes contributed to the observed differences in coral bacterial community profiles. Any ribotype constituting 5% or more (arbitrarily assigned as dominant) of each clone library was included in PCA.


Clone library analysis of bacterial diversity

The bacterial diversities associated with the three coral species, A. millepora, A. tenuis and A. valida, were highly similar for samples derived from the Magnetic Island site. Sequences from duplicate clone libraries of each species were consistent at the class level and dominated by Gammaproteobacteria, which represented approximately half of the sequences retrieved (Table 1). The second most abundant class was the Alphaproteobacteria, which comprised between 10% and 33% of retrieved sequences, while a small proportion of clones affiliated within the Deltaproteobacteria class, constituting between 1% and 12% of the libraries (Table 1). Comparisons of identical coral species at a second location (Orpheus Island) displayed similar microbial diversity profiles at the class level with both the Alpha- and the Gammaproteobacteria-affiliated sequences dominating the libraries. The Gammaproteobacteria constituted between 18% and 40% of libraries while the Alphaproteobacteria affiliated sequences represented 16–32% of all the libraries, except for one sample dominated by these sequences (71% of OItenuis-1 library). Unlike the Magnetic Island samples, Betaproteobacteria-affiliated sequences comprised a large proportion of some of the Orpheus Island libraries (between 3% and 19% of sequences; Table 1).

View this table:
1 Proportions of bacterial taxonomic classes for each clone library
Bacteria classificationMagnetic Island clone libraries (%)Orpheus Island clone libraries (%)
A. tenuisA. milleporaA. validaA. tenuisA. milleporaA. valida

Investigation of the sequences retrieved from the clone libraries at the lower taxonomic levels of genera and family also displayed consistent microbial diversity within coral species at a particular site. Libraries derived from Magnetic Island corals were dominated by sequences affiliated with Spongiobacter, Marinobacter, Acinetobacter, Roseobacter and Anaeromyxobacter species, with the relative proportion of ribotypes within these libraries being similar for all coral species (Fig. 1). In contrast, libraries derived from Orpheus Island corals were dominated by sequences affiliated with a different suite of bacterial genera, i.e. Achromobacter, Brevundimonas and Caulobacterales species. Similar to the bacterial communities associated with Magnetic Island corals, the relative proportions of the ribotypes within the Orpheus Island derived clone libraries were consistent for each coral species, the exception being A. valida-derived libraries, which additionally contained sequences affiliated with Idiomarina- and Flavobacteria-related organisms (Fig. 1). Despite differences in many dominant groups within libraries from Magnetic and Orpheus Island corals, ribotypes affiliated with Stenotrophomonas species were consistent between libraries at the two sites. Vibrio-affiliated sequences were also consistently retrieved from corals at both sites, with the exception of the Orpheus Island A. millepora samples.


Dominant bacterial sequence affiliations for Magnetic and Orpheus Island clone libraries. Sequences were grouped into dominant ribotypes at the genera and class level. Only groups representing 5% or more of the clone libraries are represented in the figure. *This group represents sequences comprising <5% of each library and unclassified bacteria.

PCA taking into account dominant clone types within each library (>5% of clones) displayed separation of libraries based on location, with the dominant bacterial ribotypes driving these differences (Fig. 2). The first two principal components described 79% of the variation in the relative proportions of dominant bacterial ribotypes. Libraries derived from Magnetic Island corals grouped together and correlated strongly with Spongiobacter- and Marinobacter-affiliated clones (Fig. 2). Other sequences affiliated with Acinetobacter, Vibrio, Spirulina and Anaeromyxobacter, weakly correlated with the Magnetic Island samples. In contrast, Orpheus Island coral microbial communities appeared to be dominated by a different consortium of bacterial genera. Libraries derived from Orpheus Island corals correlated strongly with Brevundimonas-, Stenotrophomonas- and Flavobacter-affiliated sequences and weakly with Delftia-, Idiomarina-, Pseudoalteromonas- and Pseudomonas-affiliated sequences. Roseobacter-affiliated sequences were present in all libraries derived from Magnetic Island samples, constituting between 5% and 29% of sequences (Fig. 1). Within the libraries derived from Orpheus Island corals, Roseobacter-affiliated sequences were only retrieved from A. millepora libraries at low levels (1–5% of sequences), but dominated the A. tenuis library 1 (63% of sequences retrieved). This resulted in a separation of this library from other tightly grouping Orpheus Island-derived libraries in the PCA analysis (Fig. 2).


PCA of clone library sequences. Sequence affiliations included in analysis constituted >5% of each library.

Acropora millepora colonies located at Orpheus Island were again sampled in July 2007 (winter) and February 2008 (summer) to compare seasonal changes in microbial diversity. The clone libraries were pooled because the relative abundance of ribotypes was consistent between duplicate samples derived from each sampling time point. Direct comparisons of the diversity patterns between winter (July) and summer (February) were similar (Fig. 3). The dominant bacterial ribotypes in each library, which were also consistent with the previous sampling in May 2007 (Fig. 1), included Brevundimonas sp. (18.5%, 15.7%), Achromobacter sp. (7%, 1%) and Stenotrophomonas sp. (6%, 2%) for, July- and February-derived libraries, respectively (Fig. 2). Other dominant sequences consistent between summer and winter samples, such as Idiomarina sp. (7.6% and 11.8%), Pseudoaltermonas sp. (12% and 16.7%) and Vibrio sp. (17.4% and 13.7%), were not, however, retrieved from May samples. Besides Shigella-like species recovered from July libraries, all other less-abundant ribotypes were isolated from libraries from winter and summer time periods, indicating that temporal changes had little impact on the bacterial communities of A. millepora on Orpheus Island despite large temperature shifts between winter (July average daily seawater temperature 20 °C) and summer (February average daily seawater temperature 30 °C). A summary and comparison of all sequences retrieved from all libraries is presented in Supporting Information, Table S1, along with the closest affiliated sequence in the GenBank database under accession numbers FJ489682FJ489837.


Dominant bacterial sequence affiliations for summer (February 2008) and winter (July 2007) Orpheus Island clone libraries. Sequences were grouped into dominant ribotypes at the genera and class level. Only groups representing 5% or more of the clone libraries are represented in the figure. *This group represents sequences comprising <5% of each library and unclassified bacteria.

Bacterial fingerprinting of coral-associated microbial communities

Bacterial community profiles of coral species assessed by DGGE displayed almost identical banding patterns for replicate samples of the same coral species (Fig. 4). These profiles were also similar between coral species located from the same reef site (Magnetic Island or Orpheus Island), suggesting a high level of similarity between dominant bacterial ribotypes associated with these three species of Acropora. Although DGGE bacterial profiles were similar between replicates and species within sites, the banding patterns differed between colonies located around Orpheus Island and Magnetic Island. The sequences retrieved from Magnetic Island profiles affiliated with Roseobacter, Anaeromyxobacter, Pseudomonas, Spongiobacter, Vibrio and Marinobacter species (Table 2). Sequences isolated from Orpheus Island corals affiliated with Brevundimonas-, Achromobacter- and Serratia marcescens-related organisms. The retrieved sequences from DGGE analysis of the respective corals at each location matched the dominant ribotypes recovered from clone libraries derived from these sites, cross-validating our findings. DGGE bacterial fingerprints of A. millepora corals repeatedly sampled from Orpheus Island throughout the year showed nearly identical profiles between replicate samples and for each month examined (Fig. 5). The sequences of dominant bands (Table 2) were again consistent with dominant ribotypes in the clone libraries and provide further evidence of consistent bacterial communities associated with corals despite temporal temperature changes.


DGGE for Magnetic and Orpheus Island coral samples. Band numbers correspond to the sequences retrieved in Table 2.

View this table:
2 Affiliation of DGGE bacterial sequences retrieved from coral colonies for Magnetic Island (n=9) and Orpheus Island (n=9)
Band no.Closest relative and database accession numberAlignment (bp)Similarity* (%)Taxonomic description
Magnetic Island
Band 1Roseobacter sp. (DQ412059.1)277/28198Alphaproteobacteria
Band 2Anaeromyxobacter dehalogenans (EU331403.1)302/30798Deltaproteobacteria
Band 3Marinobacter hydrocarbonoclasticus (EU624424.1)303/30898Gammaproteobacteria
Band 4Pseudomonas sp. (EU770402.1)303/30999Gammaproteobacteria
Band 5Spongiobacter nickelotolerans (AB205011.1)297/31295Gammaproteobacteria
Band 6Uncultured bacterium (EU628068.1)309/31299
Band 7Vibrio sp. (EU143360.1)272/27798Gammaproteobacteria
Band 8Pseudomonas sp. (EU770402.1)277/28198Gammaproteobacteria
Band 9Gammaproteobacterium (DQ351795.1)309/31299Gammaproteobacteria
Band 10Spongiobacter nickelotolerans (AB205011.1)305/31397Gammaproteobacteria
Orpheus Island
Band 11Muricauda aquimarina (AY445076.1)301/30897Bacteroidetes
Band 12Uncultured bacterium (EU540610.1)296/29998
Band 13Achromobacter xylosoxidans (EU827475.1)287/29098Betaproteobacteria
Band 14Achromobacter xylosoxidans (EU827475.1)277/28796Betaproteobacteria
Band 15Brevundimonas sp. (AB447548.1)265/26998Alphaproteobacteria
Band 16Achromobacter xylosoxidans (EU308119.1)277/28796Betaproteobacteria
Band 17Uncultured bacterium (AJ428146.1|UBA428146)296/29998
Band 18Serratia marcescens (FM163485.2)287/29098Gammaproteobacteria
Band 19Uncultured bacterium (EU540603.1)293/29898
Band 20Gammaproteobacterium (AY868014.1)293/29999Gammaproteobacteria
  • * *Sequences were aligned to the closest relative using blast (Altschul, 1997). The similarity was calculated with gaps not taken into account.


DGGE for Orpheus Island samples collected from May 2007 to February 2008. Band numbers correspond to sequences retrieved in Table 2.

Comparisons of T-RFLP patterns of replicate samples from each coral species at each site were consistent. Peak profile patterns (based on the presence or the absence) incorporated into a nMDS analysis demonstrated grouping of the coral bacterial patterns within locations (Fig. 6). Similar to clone library and DGGE analysis, bacterial diversity patterns of the closely related coral species were also similar and clustered together in the nMDS analysis (Fig. 6). Although profiles derived from Magnetic Island samples were not tightly clustered, they were separated from profiles derived from Orpheus Island samples.


Representation of nMDS plot of T-RFLP profiles. The data were obtained by distance matrix analyses of T-RFLP fingerprints.


Previous studies have highlighted high bacterial diversity associated with corals, but have also suggested that some bacteria form relatively stable and species-specific associations (Rohwer et al., 2001; Frias-Lopez et al., 2002; Bourne & Munn, 2005). Through poorly understood complex interactions, corals structure these microbial partners, which are suspected to play an important role in the coral host's health (Reshef et al., 2006). Observations of conserved communities have contributed to the coral holobiont principle, the concept of a structured symbiosis between the coral host and its microbial partners (Knowlton & Rohwer, 2003; Zilber-Rosenberg & Rosenberg, 2008). This study investigated the principle of conserved bacterial communities associated with three closely related species of Acropora and demonstrated that not only were the bacterial profiles of replicate coral samples of the same species highly similar, but microbiota of closely related species within the same site were also conserved. These conclusions were based on consistent observations in dominant bacterial ribotypes recovered from clone libraries of each species within the locations as well as consistent DGGE and T-RFLP profiling fingerprints.

In the current study, we were particularly interested in the dominant ribotypes and sequences retrieved from the respective 16S rRNA profiling techniques. For clone libraries, ribotypes representing >5% of each library were used as a cut-off to indicate a dominant member of the microbial community. Using this principle of dominant clone types within the libraries, a PCA plot indicated that the bacterial communities associated with corals at two sites separated by 40 km, an inshore fringing reef (Magnetic Island) and a coastal reef (Orpheus Island), displayed different microbial diversity patterns. nMDS analysis of bacterial T-RFLP profiles and DGGE fingerprints supported this observation. T-RFLP profiles, like clone libraries, were selected based on a threshold for relative peak height to account for dominant sequences within duplicate PCRs, while DGGE is known for providing only dominant profiles within a community (Muyzer et al., 1993; Muyzer & Smalla, 1998; Gelsomino et al., 1999; Maarit Niemi et al., 2001; Nikolausz et al., 2005). Sequences retrieved from both DGGE and clone library analyses were consistent, supporting and cross validating the conclusions that differences in the dominant bacterial community members exist on corals located at the two reef systems.

Previous studies have reported that similar bacterial populations have been found on the same coral species that are geographically separated (Rohwer et al., 2001, 2002), suggesting that corals associate with certain bacteria despite geographic location. While the current study demonstrates that bacterial profiles separate corals based on geographic location, these conclusions are based on the dominant profiles driving the differences. When we look at individual sequences that represent minor components of retrieved 16S rRNA sequences from clone libraries, we see consistent sequences between the corals (Table S1). For example, Roseobacter- and Stenotrophomonas-affiliated sequences are observed in all libraries. While Spongiobacteria-affiliated sequences are dominant in the Magnetic Island samples (12–27% within all libraries), they are also retrieved at a low relative number in the Orpheus Island clone libraries (1%). This indicates that similar phylotypes are associated with corals and the multitude of coral diversity reports in the last 10 years similarly report common coral-associated bacteria (Cooney et al., 2002; Rohwer et al., 2002; Bourne & Munn, 2005). The observations in the current study indicate that while the same bacterial ribotypes are observed within all the corals investigated, dominant members of the community differ between locations. However, what drives this apparent difference is unknown.

Reshef et al. (2006) proposed the coral probiotic hypothesis, which states that corals potentially adapt to new environmental conditions by altering their specific symbiotic bacterial partners. Environmental variables such as water quality, light exposure or temperature may be determining factors driving the difference in dominant coral-associated bacteria at these two similar shallow in-shore reef environments on the GBR. Klause et al. (2007) demonstrated that water depth in varying proximity to coastal pollution can affect the bacterial composition on healthy colonies of Montastrea annularis. While both coral reefs are proximally located to the shore, Orpheus Island is not inhabited, apart from a small research station and resort, while Magnetic Island is populated and some land-derived pollutants are discharged onto the reef (Larcombe et al., 1995; Muslim & Graham Jones, 2003). Factors such as light intensity (Kuta & Richardson, 2002; Richardson & Kuta, 2003), water temperature (Gil-Agudelo & Garzon-Ferreira, 2001; Kuta & Richardson, 2002; Ben-Haim et al., 2003; Cervino et al., 2004; Rosenberg & Falkovitz, 2004; Bruno et al., 2007) and increased pollution (Bruckner et al., 1997; Kim & Harvell, 2002; Patterson et al., 2002; Kaczmarsky et al., 2005) have all been correlated with shifts in coral microbial associations that lead to coral disease outbreaks. However, all corals in this study were visibly healthy at the time of sampling. It is unlikely that bacteria within the water column are driving the observable differences in the dominant bacteria communities because sequences affiliated with Spongiobacter, Roseobacter and Stenotrophomonas, to name only a few, were a component of coral clone libraries at both locations, suggesting that many of these bacteria are present in both environments. One aspect of the coral holobiont that should be investigated is the potential for the photosynthetic dinoflagellate Symbiodiunium to drive differences in associated bacteria communities. This algal partner provides the coral host with the majority of its photosynthate nutrient sources, with different genotypes harbored by coral from separate locations (Baker, 2003; Berkelmans & van Oppen, 2006). Variation in photosynthetic activity, producing variable bioavailable nutrient sources, may influence the composition of coral mucus and thus indirectly the coral microbiota (Klause et al., 2007).

Previous studies have suggested that bacterial communities may be dynamic in healthy corals in response to seasonal temperature fluctuations. For example, Koren & Rosenberg (2006) examined the bacteria associated with Oculina patagonica in the summer and winter to establish the temporal changes in bacteria on healthy corals. While the dominant group (35% of sequenced clones) was Vibrio splendidus in both summer and winter, the next 10 most abundant clusters of bacteria differed between seasons and temperature was one factor identified as causing these microbiota shifts. The results from this study demonstrate that the dominant members of the bacterial community associated with A. millepora were relatively stable throughout the seasons on Orpheus Island despite temperature fluctuations of >10 °C. While there were differences in some dominant sequences retrieved from May sample libraries, DGGE profiles were consistent for all sampling periods, and direct comparisons between libraries from winter (July) and summer (February) showed a highly similar abundance of dominant ribotypes. The seasonal conditions are less variable along the GBR compared with the 20 °C fluctuation in the Mediterranean Sea, which may provide one explanation for the differences observed by Koren & Rosenberg (2006) and the stable microbial communities observed associated with A. millepora within one site and throughout one year.

Magnetic Island corals displayed highly similar bacterial profiles when comparing duplicate libraries of the same species. While duplicate samples derived from Orpheus Island were similar, there were some differences particularly within the A. tenuis libraries, indicating within-colony variation, or patchiness of microbial associates. This is consistent with results from Rohwer et al. (2002), which demonstrated that one bacterial ribotype resided only in the branch tip, but not in other portions of Porites furcata. Furthermore, Bourne & Munn (2005) showed that different microenvironments exist within the layers of Pocillopora damicornis in which the associated bacterial communities differed. Over-representation of retrieved Roseobacter sp. affiliated sequences in the A. tenuis library 1 caused separation in PCA analysis of dominant ribotypes within the libraries. This result highlights the potential for bias in PCR-based bacterial profiling techniques, also indicating the strength of cross-validating the conclusions using three different profiling techniques, each with different bacterial-targeted PCR primer sets.

For corals located at Magnetic Island, sequences related to Spongiobacter species were of particular interest. These related sequences have been derived as the name suggests from sponges, but have also been observed previously as dominating a culture-independent assessment of A. millepora coral species at Magnetic Island (Bourne et al., 2008). Bourne et al. (2008) showed that Spongiobacter sp.-related sequences dominated bacterial profiles before and after a bleaching event, shifting away from this stable microbial association during the stress period. Within the libraries of this study, Spongiobacter-related sequences accounted for between 17% and 35% of clone libraries and was the dominant clone type. Being such a dominant member of the coral-associated microbial community indicates that Spongiobacter play a key role in the holobiont-functioning system, though at present little is known as to what role this may be. Other sequences dominating libraries for coral at Magnetic Island such as Roseobacter (9–38%), Stenotrophomonas (4–13%) and Marinobacter (17–24%) species also potentially perform necessary functions. For instance, Marinobacter sp. are aerobic halophilic bacteria that can utilize a range of carbon sources as well as degrade hydrocarbons (Huu et al., 1999; Nicholson & Fathepure, 2004; Marquez & Ventosa, 2005), which may be vital to the carbon cycle within the holobiont. The bacterial inhabitants of corals from Orpheus Island that comprised the greatest portion of the clone libraries, i.e. Brevundimonas sp. (4–22%), Stenotrophomonas sp. (8–26%) and Roseobacter sp. (5–38%), are possible candidates for investigation in coral nutrient cycling or production of antimicrobial properties. For example, Brevundimonas sp. and other Caulobacterales are anaerobes found ubiquitously in water and are believed to play important roles in carbon cycling within their respective habitats (Staley et al., 1987; Abraham et al., 1999), and therefore may have a function in the host's nutrition. The marine Roseobacter clade comprises several genera of marine bacteria with heterogeneous physiological properties including sulfite reduction (Sorokin, 1995), metabolism of dimethylsulfoniopropionate (Zubkov et al., 2001, 2002), production of toxins (Gallacher et al., 1997; Hold et al., 2001) and aerobic anoxygenic photosynthesis (Algaeir et al., 2003). As such, Roseobacter species are suspected to have important functional activities within the coral holobiont, including the ability to produce secondary metabolites (Lafay et al., 1995; Gram et al., 2002). Some species have been found to produce antagonistic activity against other marine bacteria and algae (Brinkhoff et al., 2004; Roa et al., 2005), which may also serve as a protection mechanism in coral excluding potential pathogenic organisms (Ritchie, 2006).

Unresolved questions

As we begin to unravel what the players are in coral microbial communities, several questions arise as to what implication these associations have for coral health. Future studies need to explore what the primary holobiont factors are that may drive these consistent microbial associations and how environmental factors such as temperature, light and eutrophication can influence and subsequently change the microbial consortia. An understanding of the fragility of these microbial associations and how their functions in the holobiont supports coral health is essential to predict and understand what effect environmental change will have on coral health. Determining the functional roles of bacteria as well as other microorganisms such as Archaea, the endolithic community, algae and protozoans in coral associations will also allow us to distinguish whether there are symbiotic relationships between the coral host and certain microorganisms. In addition, we can determine whether there are a consortia of many bacteria interacting synergistically to form a multispecies partnership with the host, or whether there are bacteria merely taking advantage of the host as a suitable habitat without any reciprocated benefit.


We would like to thank the ARC Centre of Excellence for Coral Reef Studies and the Australian Institute of Marine Sciences for their contributions to this research. We also thank Dr Bryan Wilson for help in analyzing T-RFLP data, Tim Simmonds for help in preparing figures and Dr Lone Høj for critical comments on the manuscript.


Additional Supporting Information may be found in the online version of this article.

Table S1. Affiliation of bacterial sequences retrieved from 16S rRNA gene clone libraries

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.


  • Editor: Patricia Sobecky


View Abstract