Main

Coral reefs harbour abundant and diverse marine invertebrates that perform important ecosystem functions such as: calcification, bioerosion, consolidation and benthic-pelagic coupling (Glynn and Enochs, 2011). Animal–plant/microbe symbioses are vital to these ecosystems as they facilitate photosynthetic productivity, mineral recycling, nutrient provision to the host and secondary metabolite production (Smith and Douglas, 1987). Although patterns of microbial diversity and putative symbiotic functions have been well explored in corals and sponges (Sunagawa et al., 2009; Mouchka et al., 2010; Webster and Taylor, 2012; Bourne and Webster, 2013), there is a lack of data on microbial associations in other reef taxa including Bivalves, Foraminifera and Ascidians.

The diversity of microbial communities associated with corals and sponges is known to be influenced by host interactions (Wegley et al., 2007; Kimes et al., 2010; Raina et al., 2010; Fan et al., 2012), the production of antimicrobial compounds (Ritchie, 2006; Shnit-Orland and Kushmaro, 2009) and environmental conditions (Hong et al., 2009; Ceh et al., 2011). Recent studies, however, indicate that other members of the coral holobiont (in particular Symbiodinium dinoflagellates) also influence microbial community structure through release of complex carbon-containing exudates including dimethylsulfoniopropionate (DMSP; Ikeda and Miyachi, 1995; Raina et al., 2009, 2010). DMSP can be degraded to dimethylysulphide, a central molecule in the global sulphur cycle, which diffuses from the ocean into the atmosphere where it influences cloud formation, with consequences for atmospheric chemistry, local climate and water temperature (Ayers and Gras, 1991; Andreae and Crutzen, 1997). A complex array of other organic exudates including amino acids and polysaccharides can also influence invertebrate-associated microbiomes, which may affect holobiont fitness. For example, Symbiodinium spp. have been shown to influence the response of bacterial communities to thermal stress, which affects the susceptibility of the holobiont to bleaching (van Oppen et al., 2009; Stat et al., 2012), disease (Stat et al., 2008) and colonisation by opportunistic potential pathogens (Littman et al., 2010).

In this study, we used 16S ribosomal RNA (rRNA) gene amplicon pyrosequencing (Supplementary Methods) to characterise the microbiomes of 16 common Great Barrier Reef marine invertebrate species representing five invertebrate families (Table 1). These families included 11 species that host photosynthetic symbionts (Symbiodinium and diatoms) and five species that do not host these symbionts. The microbiomes for three replicate samples from each invertebrate species and seawater controls were characterised. Briefly, 16S rRNA gene amplicons generated using primers 63F and 533R (Engelbrektson et al., 2010) were subjected to 454 pyrosequencing. Sequences were checked for chimeras using UCHIME ver. 3.0.617 (Edgar et al., 2011), denoised using Acacia (Bragg et al., 2012) and then parsed using the QIIME pipeline with default settings (Caporaso et al., 2010). We tested the hypotheses that: (1) the presence of photosynthetic symbionts influences the diversity of marine invertebrate-associated microbiomes, and (2) that the diversity of marine invertebrate-associated microbiomes differs between host species.

Table 1 List of samples, phylogentic classification, associated pyrosequence reads and symbiont type

The presence of photosynthetic symbionts influenced the composition (Figure 1), but not the species richness, evenness and phylogenetic diversity (P>0.05, linear regression; Supplementary Table S1) of invertebrate-associated microbiomes. At the class level, the presence of photosynthetic symbionts explained 21% of variation in the composition of microbial communities between samples (PERMANOVA, F1,44=11.37, P<0.001). At the level of operational taxonomic units (OTUs), defined as groups of sequences that shared 97% nucleotide sequence similarity (‘species’ level), the presence of photosynthetic symbionts explained a significant, albeit smaller (6%) proportion of variation in the composition of microbial communities between samples (PERMANOVA, F1,44=2.66, P<0.001). Unifrac analysis based on OTUs also confirmed that the presence of photosynthetic symbionts influenced microbial community composition (unweighted P=0.002, weighted P<0.001).

Figure 1
figure 1

Redundancy analysis (RDA) summarising variation in the composition of marine invertebrate-associated microbial communities that was attributable to the presence–absence of photosymbionts. The filled shapes represent individual samples collected from each invertebrate species. The black crosses represent bacterial OTUs. For clarity, taxonomic affiliations are shown for the most discriminating OTUs only. The distance of an object (sample or OTU) from the origin is proportional to its variance along an axis and its angle relative to the axes reflects its correlation with those axes. Full sample collection and processing details can be found in the Supplementary Methods and the sequence data set deposited in the NCBI Sequence Read Archive (SRA) database with the accession number SRA4494953.

Alphaproteobacteria and Gammaproteobacteria were the dominant classes of bacteria associated with reef invertebrates. Invertebrates without photosynthetic symbionts (with the exception of one replicate Bryozoan sp.) were associated with a larger abundance of Alphaproteobacteria, whereas those with photosynthetic symbionts generally hosted a higher relative abundance of Gammaproteobacteria (Supplementary Figure S1). The only exceptions were Seriatopora hysterix and two of the Sinularia sp. samples in which Flavobacteria were particularly abundant. Other bacterial classes including the Deltaproteobacteria, Sphingobacteria and Cyanobacteria differed between invertebrate species but were not influenced by the presence of photosymbionts (Supplementary Figure S1). Although present in all samples, the Cyanobacteria were particularly abundant (8–24% relative abundance) in Heterostegina depressa and in one Marginopora vertebralis sample (17% relative abundance). The composition of microbial communities associated with the sponge Rhopaloeides odorabile was different to those associated with other invertebrates, although this community pattern is consistent with a previous investigation (Webster et al., 2010).

Most dominant OTUs (that is, >5% relative abundance) were affiliated with bacterial populations previously retrieved from marine environments including corals and sponges (Figure 2). Invertebrates that host photosynthetic symbionts were positively correlated with OTUs related to Oceanospirillales spp., a Roseivirga sp., an Alteromonas sp., Pseudoalteromonas spp., Halomonas spp., Pseudomonas spp. and Flavobacteriacae spp. (Figure 1). Indicator species analysis (Dufrene and Legendre, 1997) confirmed these OTUs were significantly correlated with the presence of photosynthetic symbionts by having high relative abundance and frequency of occurrence (Figure 2). These OTUs are all affiliated with species implicated in the metabolism of complex organic molecules such as DMSP and dimethylysulphide. For example, previous studies have identified abundant bacteria within the Oceanospirillales that are able to metabolise DMSP in the coral Acropora millepora (Raina et al., 2009). Halomonas spp. have been shown to be capable of metabolism of DMSP and its breakdown product acrylic acid (Todd et al., 2010), while members of the Flavobacteriacae respond rapidly to high DMSP concentrations in phytoplankton blooms, although the genetic pathways for metabolism of this compound in this group of bacteria is unknown (Howard et al., 2011). In the marine environment, DMSP has been the focus of considerable attention because of its fundamental role as carbon and sulphur sources for bacteria (Sievert et al., 2007). Coral reefs are one of the largest producers of DMSP with the source thought to be derived from marine invertebrates harbouring symbiotic dinoflagellates (Broadbent et al., 2002; Van Alstyne et al., 2006). In fact, the concentrations of DMSP and its breakdown products dimethylysulphide and acrylate in reef-building corals are the highest recorded in the marine environment (Broadbent and Jones, 2004). These results further support the concept that sulphur-based organic compounds derived from photosymbionts influence the microbial communities of marine invertebrates by providing nutrient sources readily available for metabolism by associated microbiomes. Although compounds such as DMSP appear to have a role in structuring microbial communities associated with the host organism, there are likely to be many other organic exudates derived from photosymbionts that also influence microbial associations. In addition, host animal factors can have an important role in structuring microbial communities. Results from this study and other recent reports highlight that members of the Oceanospirillales, specifically, Endozoicomonas spp. are commonly found in marine invertebrates with and without photosymbionts and potentially have important functional roles within their host species (Yang et al., 2010; Nishijima et al., 2012; Speck and Donachie, 2012).

Figure 2
figure 2

Heatmap of OTUs (97% sequence identity, averaged within each invertebrate species) that represent >5% of sequence tags within a particular sample type or represented an OTU with a significant indicator value (representative of high relative abundance and high relative frequency of occurrence in photosymbiont-bearing invertebrates). OTUs with highest indicator values are represented by an asterisk (*). The closest sequence match determined in a BLAST database query (Altschul et al., 1997) and its corresponding accession number and derived sample source are also represented. If these affiliated sequences are represented in a published study the associated reference can be found in Supplementary References.

Indicator species analysis demonstrated that no OTUs were significantly correlated with invertebrates that do not host photosymbionts, although Rugeria-, Rhodobacteraceae- and Rhodospirillaceae-related sequences were more commonly retrieved in these samples as observed in the redundancy analysis (Figure 1). Microbial communities associated with the seawater samples were distinct from those associated with both photosymbiont and non-photosymbiont-bearing invertebrates (P=0.002, PERMANOVA). This difference was related to a larger abundance of the ubiquitous bacterioplankton Candidatus pelagibacter (SAR11) comprising ∼66% of sequence reads from this control group (Figure 1).

The composition of microbial communities also differed between invertebrate groups (Foramaninfera, Scleractinia, Octocorallia, Bivalvia, Bryozoa and Ascidiacaea; P=0.001, see Supplementary Figure S2) and this was further supported by both unweighted and weighted unifrac distances (P<0.001, redundancy analysis). Many replicate microbiomes from the same invertebrate species grouped well at both class and OTU taxonomic assignment levels and was reflected by the composition of microbial communities being different between invertebrate species (P<0.001, PERMANOVA). Most within-species variability existed for the samples derived from A. millepora, Sinularia flexibilis, Tridacna spp. and Bryozoan sp. (Supplementary Figure S2). Although the redundancy analysis and Heatmap/Cluster analysis generally group individual specimens from one species closely together and showed significant relationships at the taxa level, there is no apparent higher phylogenetic grouping. However, both analyses clearly separate samples by presence or absence of photosymbionts. Further studies comparing microbiomes among taxa with and without photosymbionts will be useful in further clarifying the strength of these relationships and the role photosymbionts have in driving microbial associations.

Rarefaction analysis demonstrated that all three Foraminifera species hosted the largest bacterial diversity among the invertebrate samples (Supplementary Figure S3), which may reflect their lifestyle closely associated with reef rubble, filamentous algae and reef sediment. Heterostegina depressa was the only diatom-bearing invertebrate species and richness of these samples (1123 OTUs) exceeded that of all other invertebrate taxa. The two Octocoral species, Sarcophyton sp. and Sinularia flexibilis hosted the lowest microbial richness (80–180 OTUs, respectively), which may relate to the high antimicrobial activity previously identified in these species (Kim, 1994; Jensen et al., 1996; Harder et al., 2003). Many invertebrate species including all Foraminifera, Bivalvia, Bryozoa and the Scleractinian corals A. millepora and Pocillopora damicornis hosted higher bacterial richness than the surrounding seawater (Supplementary Figure S3). Rarefaction analysis of the Scleractinian corals in this study is consistent with earlier estimates of coral microbial diversity (Sunagawa et al., 2010).

The documented decline in coral reef ecosystems worldwide (Wilkinson, 2008) has prompted research into understanding how changing environmental conditions affect the close symbiotic associations of marine invertebrates (Webster et al., 2001; Bourne et al., 2008; Vega Thurber et al., 2008; Webster et al., 2008; Littman et al., 2010, 2011; Webster et al., 2011). Only by studying marine invertebrates as holobionts (the host and all associated microbial communities) and better characterising the forces that structure their microbial associations will we be able to fully assess their capacity to adapt or acclimatise to environmental stress. From this study, 16S rRNA gene amplicon pyrosequencing revealed high diversity of bacterial symbionts within 16 common Great Barrier Reef species. Importantly, although microbial composition was related to host species, a significant amount of the variation in community composition was attributed to the presence or absence of photosymbionts. These results highlight the importance of photosymbionts in structuring reef bacterial symbioses.