RESEARCH ARTICLE Disturbance to conserved bacterial communities in the cold-water gorgonian coral Eunicella verrucosa Emma Ransome1,2,3, Sonia J. Rowley2,4,5, Simon Thomas1, Karen Tait1 & Colin B. Munn2
1Plymouth Marine Laboratory, Plymouth, UK; 2School of Marine Science and Engineering, Plymouth University, Plymouth, UK; 3Smithsonian National Museum of Natural History, Washington, DC, USA; 4Bernice Pauahi Bishop Museum, Honolulu, HA, USA; and 5University of Hawai’i at Manoa, Honolulu, HI, USA Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021
Correspondence: Emma Ransome, Abstract Department of Invertebrate Zoology, National Museum of Natural History, Smithsonian The bacterial communities associated with healthy and diseased colonies of the Institution, Washington, DC 20560, USA. cold-water gorgonian coral Eunicella verrucosa at three sites off the south-west Tel.: (+1) 202 633 9075; coast of England were compared using denaturing gradient gel electrophoresis fax: (+1) 202 357 2343; (DGGE) and clone libraries. Significant differences in community structure e-mail: [email protected] between healthy and diseased samples were discovered, as were differences in the level of disturbance to these communities at each site; this correlated with Received 3 March 2014; revised 12 July 2014; accepted 27 July 2014. Final version depth and sediment load. The majority of cloned sequences from healthy coral published online 01 September 2014. tissue affiliated with the Gammaproteobacteria. The stability of the bacterial community and dominance of specific genera found across visibly healthy colo- DOI: 10.1111/1574-6941.12398 nies suggest the presence of a specific microbial community. Affiliations included a high proportion of Endozoicomonas sequences, which were most Editor: Patricia Sobecky similar to sequences found in tropical corals. This genus has been found in a number of invertebrates and is suggested to have a role in coral health and in Keywords the metabolisation of dimethylsulfoniopropionate (DMSP) produced by zoo- coral microbiology; Eunicella verrucosa; coral xanthellae. However, screening of colonies for the presence of zooxanthellae disease; Endozoicomonas spp.; DGGE; clone libraries. produced a negative result. Diseased colonies showed a decrease in affiliated clones and an increase in clones related to potentially harmful/transient micro- organisms but no increase in a particular pathogen. This study demonstrates that a better understanding of these bacterial communities, the factors that affect them and their role in coral health and disease will be of critical impor- tance in predicting future threats to temperate gorgonian communities.
such as nitrogen fixation, nitrogen cycling and sulphur Introduction cycling (Lesser et al., 2004; Raina et al., 2009). They also Recent marine disease epizootics have reduced the abun- confer resistance to the host by producing antibacterial dance of a variety of endangered, commercially valuable agents (Nissimov et al., 2009; Shnit-Orland & Kushmaro, and habitat-forming species (Harvell et al., 1999). This 2009), which are compromised during disease (Ritchie, includes a number of corals from tropical and subtropical 2006). Increased host susceptibility (Lesser et al., 2007) systems, where an increasing number of coral diseases are and increased pathogenicity of coral-associated microor- being described (Bourne et al., 2009). Investigations into ganisms (Rosenberg & Ben-Haim, 2002; Bruno et al., bacterial associations with corals have established that 2007) have both been proposed to be driving incidences these communities are often species specific (e.g. Ritchie of disease. In addition, disturbance to the fragile relation- & Smith, 1997; Littman et al., 2009), spatially and tempo- ships between the coral host and their bacterial commu- rally stable (Knowlton & Rohwer, 2003; Sharp et al., nities has been linked to a variety of environmental 2012) and different from surrounding water and sediment factors, including thermal abnormalities (Harvell et al., MICROBIOLOGY ECOLOGY MICROBIOLOGY communities (Frias-Lopez et al., 2002; Carlos et al., 2002; Rosenberg & Ben-Haim, 2002), increased nutrients 2013). There is strong evidence that these bacterial com- (Bruno et al., 2003) and sedimentation (Voss & Richard- munities have a beneficial effect, carrying out functions son, 2006). With an increasingly changing marine
ª 2014 Federation of European Microbiological Societies. FEMS Microbiol Ecol 90 (2014) 404–416 Published by John Wiley & Sons Ltd. All rights reserved Disturbed bacterial communities in E. verrucosa 405 environment and an increase in the frequency and inten- ecology of the benthic environment in which it is found sity of disease (Garrabou et al., 2009), it is essential to (Hall-Spencer et al., 2007). Colonies were monitored for understand the progression of disease and its link to envi- disease at three sites, chosen for differences in depth and ronmental stress. Documenting the alteration of coral substratum, off the SW coast of England, during 2008, bacterial communities is necessary to provide accurate and healthy and diseased colonies were sampled in June diagnosis of coral disease for researchers and ecosystem and September of 2008. Sedimentation, temperature and managers (Ainsworth et al., 2007). irradiance were also recorded. Recently, disease outbreaks have been noted in gorgo- nian corals (Order: Alcyonacea) in temperate waters of Materials and methods the NW Mediterranean (Cerrano et al., 2000; Martin et al., 2002; Garrabou et al., 2009) and in SW England Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021 Field observation and sample collection (Hall-Spencer et al., 2007). Mass mortality by tissue necrosis has been observed for several species and Two sediment traps were deployed at each site for although the cause of this tissue loss has not been clearly 12 days in June and September 2008 (English et al., defined, opportunistic pathogenic bacteria have been 1997). Seawater temperature and irradiance were mea- implicated in a number of gorgonians (Cerrano et al., sured at each site every 15 min throughout the monitor- 2000; Harvell et al., 2001; Martin et al., 2002). Using cul- ing period (HOBO data loggers, Onset, MA). On 7 ture-based methods Hall-Spencer et al. (2007) found an June, E. verrucosa colonies at each site were evaluated for increase in the diversity of culturable bacteria from visual signs of necrotic tissue, epibiont cover and fouling healthy to diseased Eunicella verrucosa tissue, many of to determine colony health (see Table 1 for site details) which were a close match to Vibrio splendidus. Members via a stratified videographic survey using closed circuit of the family Vibrionaceae are associated with disease in rebreather diving technology (AP Diving Inspiration Clas- other coral species (e.g. Godwin et al., 2012), including sic). Each individual colony encountered within four per- infections correlated with temperature stress (e.g. Kush- manent 10 m 9 2 m belt transects per site was filmed maro et al., 1996; Ben-Haim & Rosenberg, 2002). A Vib- face on with a scaled back board aligned appropriately rio strain showing thermo-dependent virulence has also for scale. Fouling was determined by the percentage cover been isolated from diseased colonies of the gorgonian per colony quantified from videographic frame grabs in coral Paramuricea clavata (Bally & Garrabou, 2007) and ImageJ64 (Abr amoff et al., 2004) and was defined as has been shown to be involved in mass mortality events established, and therefore nontemporary (e.g. not snagged of this coral in the NW Mediterranean (Vezzulli et al., on colony temporarily due to water current dynamics) 2010). Hall-Spencer et al. (2007) demonstrated that these matter covering the coral axis. Abundance data were Vibrio isolates induced tissue necrosis at 20 °C but not at quantified in ImageJ64 from scaled frame grabs (using 15 °C in the laboratory, suggesting a possible link belt transects, as above) and counts of colonies ≤ 2cm between E. verrucosa disease and temperature. tall were used to determine the previous years recruit- Details of the nonperturbed microbial communities ment (Munro, 2004). Five healthy colonies and two dis- thought to associate with temperate anthozoans are scant eased colonies (depending on presence) of similar size in comparison with tropical anthozoans (Bayer et al., from each site were tagged. Evaluation of visual condition 2013), as is information concerning how these communi- of colonies was repeated on 16 September. Recruitment ties change with the progression of disease. Given a greater of E. verrucosa was assessed using abundance data from knowledge of these communities we may further our 2007 and 2008 surveys. understanding of multispecies mutualism, the effect that Branches 4 cm in length were collected from each of environmental conditions have on these associates and aid 37 tagged colonies for analysis. In June, five healthy colo- our identification of species that play a key role in both nies from each site and one diseased colony from site 1 maintaining coral health and progressing disease. Further, (16 in total) and in September, five healthy colonies and documenting shifts in microbiota, if they occur prior to two diseased colonies from each site (21 in total) were signs of visible stress, may also allow the use of microbiol- collected. Only one colony found at site 1 showed signs ogy as a bio-indicator of both environmental change and of disease during the June survey, and only two diseased disease (Pantos et al., 2003; Bourne & Munn, 2005). colonies from each site were found in September. The aim of this study was to investigate microbial com- Branches were placed in plastic bags underwater and munities associated with the pink sea fan, E. verrucosa, immediately taken to the surface where they were washed using molecular based methods. This cold-water gorgo- in 0.2 lm filtered, sterile phosphate-buffered saline to nian coral is on the international ‘red list’ of threatened remove loosely attached microorganisms. The branches species and is known to be important for the functional were then placed in RNAlater (Life Technologies, Paisley,
FEMS Microbiol Ecol 90 (2014) 404–416 ª 2014 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved 406 E. Ransome et al.
Table 1. Site descriptions
Mean value SD (where appropriate)
Site 1 Site 2 Site 3 Site Month June September June September June September Latitude 03°58.1160W – 04°07.2610W – 04°08.8810W – Longitude 50°17.1020N – 50°18.1680N – 50°20.0210N – Depth (m) 24–27 – 15–25 – 6–9 – Sediment type Medium-fine grain – Coarse grain – Fine grain – Substratum type Wreck – Rocky reef – Artificial reef – Average temperature (°C) 14.81 0.44 16.72 0.05 14.97 0.41 16.68 0.10 15.58 0.37 16.83 0.08 Temperature (max–min) (°C) 13.85–15.47 16.62–16.81 14.04–16.81 16.52–17.00 14.61–17.48 16.52–17.57 Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021 Average light (lux) 119.4 – 547.4 – 475.5 – Sedimentation (cm 2 day 1) 5.84 2.08 2.31 0.64 4.34 0.57 3.08 0.48 12.36 1.81 16.12 7.1
UK) and kept on ice before transporting back to the lab- Taq Flexi (Promega) and 0.4 ng of DNA using the follow- oratory and storing at 80 °C until analysis. ing conditions: 1 cycle at 96 °C for 4 min; 35 cycles at Water samples were not collected for microbial analysis 96 °C for 1 min, 53 °C for 1 min and 72 °C for 1 min and at coral sites during June and September surveys. There- one final extension at 72 °C for 5 min. Each PCR was con- fore, twenty litres of water collected from the coastal ducted in triplicate. Following successful amplification, monitoring site L4 in June and September of 2008 PCR products were diluted 1 : 10 with dH2O and primers (8 miles from E. verrucosa sites; 50°15.000N, 4°13.020W), 341f (50-CCTACGGGAGGCAGCAG-30) with a 40-bp GC at 10 m depth, was used to assess the microbial commu- clamp and 907r (50-CCGTCAATTCMTTTGAGTTT-’3) nity in the water column at the time of coral sampling. were used to amplify a 566-bp section of the 16S rRNA There is no record of E. verrucosa at this site. Water was gene (Muhling€ et al., 2008). The reaction mixture con- prefiltered [140 mm diameter, 1.6 lm GF/A filter (What- tained 0.5 lM each primer, 0.15 mM each dNTP, 1.5 mM man, Maidstone, UK)] and then applied directly to a MgCl2 and 0.15 Units of Go Taq Flexi (Promega) in a total 0.22 lm Sterivex filter (Millipore, Billerica, MA). Follow- volume of 60 lL. The amount of diluted DNA added to ing filtration, each Sterivex was pumped dry, frozen in each reaction depended on band brightness from agarose liquid nitrogen and stored at 80 °C until analysis. gels from the first PCR round (from 0.1 to 2 lL of diluted PCR product was added). Temperature cycling for PCR amplification was 1 cycle at 94 °C for 5 min; 22 cycles at DNA extraction and purification 94 °C for 1 min, 65 °C for 1 min (decreasing by 0.5 °C DNA was extracted using the DNeasy Blood and Tissue every cycle) and 72 °C for 1 min, 15 cycles at 94 °C for Extraction Kit (QIAGEN, Manchester, UK) according to 1 min, 55 °C for 1 min and 72 °C for 1 min and one final the manufacturer’s protocol for animal tissues, with an extension at 72 °C for 7 min. PCR products were pooled, extension of the 56 °C incubation to overnight. To cleaned using the QIAquick PCR purification kit (QIA- extract DNA from water column samples, the methodol- GEN) and total DNA quantified by Nanodrop spectropho- ogy of Neufeld et al. (2007) was used. Following the tometry (NanoDrop Technologies, Delaware). DGGE was extraction, total nucleic acids were eluted in 200 lLof performed using the DCodeTM System (BioRad) with nuclease-free water and total DNA was quantified by 800 ng of cleaned PCR products on an 8% polyacrylamide spectrophotometry (Nanodrop 1000, Thermo Scientific) gel with urea and formamide as denaturants (30–60% gra- and diluted to 20 ng lL 1. dient), at 60 °C and a constant voltage of 60 V for 18 h. Subsequently, gels were stained with SYBR Green 1 (Molec- ular Probes, Eugene, OR) for 30 min and then washed with Denaturing gradient gel electrophoresis deionised water for 30 min. This process was repeated (DGGE) twice to check for reproducible results. A nested PCR approach was used to amplify the 16S rRNA gene for DGGE. DNA was amplified using 0.2 lM of prim- Clone library construction, sequencing and ers 9bfm (50 GAGTTTGATYHTGGCTCAG-30) and phylogenetic analysis 1512uR (50 ACGGHTACCTTGTTACGACTT-30)(Muhling€ et al., 2008), 0.2 mM of each dNTP, 19 PCR buffer (Pro- PCR products amplified using primers 9bfm and 1512uR mega, Southampton, UK), 2 mM MgCl2, 0.25 Units of Go (as above) were pooled by site, month and health status,
ª 2014 Federation of European Microbiological Societies. FEMS Microbiol Ecol 90 (2014) 404–416 Published by John Wiley & Sons Ltd. All rights reserved Disturbed bacterial communities in E. verrucosa 407
(e.g. five replicates from site 1 healthy colonies, in June), data corresponding to each branch of a dendogram (1000 cleaned, quantified and diluted to 20 ng lL 1. Twelve permutations, significance level 5%). To assess variability clone libraries were constructed, two for water samples as a measure of disturbance to coral bacterial communities from June and September, and ten from coral tissue multivariate dispersion (MvDISP) analysis was performed (healthy pooled samples: June site 1, 2 and 3, September on DGGE and clone library data. PERMANOVA analysis estab- site 1, 2 and 3; diseased pooled samples: June site 1, Sep- lished the importance of site, month and health in deter- tember site 1, 2 and 3) using a pGEM -T easy cloning mining bacterial community composition for DGGE data kit (Promega) and E.coli JM109 (Promega) competent (Primer-E v6). The Shannon–Weaver (H’) test investigated cells, using the manufacturer’s protocol. Clones contain- diversity in DGGE and clone library analyses, Pielou’s (J’) ing inserts were verified using the vector primers M13F test investigated evenness in clone libraries and average 0 0 0 (5 -GTAAAACGACGGCCAG-3 ) and M13R (5 –CAG taxonomic distinctness (D+) took phylogenetic relatedness Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021 GAAACAGCTATGAC-30). Temperature cycling for PCR of different classes into account for clone library data. amplification was 1 cycle at 94 °C for 3 min; 30 cycles at 95 °C for 30 s, 59 °C for 30 s and 72 °C for 30 s and Results one final extension at 72 °C for 5 min. Forty clones were randomly selected for each coral library and each water Environmental perturbations library and directly sequenced using a BigDye Terminator v3.1 cycle sequencing kit (ABI) and the M13F primer for The environmental conditions at each site are described the sequencing reaction. Sequences were analysed on an in Table 1. In June, seawater temperature at site 3 was ABI3100 automated sequencer. Only one strand of the significantly higher (ANOVA F = 12.68, P < 0.001) and in DNA fragment was sequenced, proving sufficient for tax- September, seawater temperature at site 1 was signifi- onomic identification. The 16S rRNA gene sequences cantly lower (ANOVA F = 7, P < 0.01) than other sites. were compared to sequences stored in GenBank (NCBI Light intensity (lux) was significantly lower at site 1 database) using the BLAST algorithm to identify bacteria (square root transformed data; ANOVA F = 25.76, and archaea associated with E. verrucosa and aligned P < 0.001), and sediment trap analysis showed site 3 to using CLUSTALW in MEGA 4 (Molecular Evolutionary Genet- have a significantly higher sediment load in June and ics Analysis; Tamura et al., 2007), with the closest rela- September (ANOVA F = 20.70, P < 0.005; F = 10.63, tives from BLAST searches. Sequences were submitted to P < 0.01), compared with site 1 and 2. GenBank under accession numbers KF180619–KF181098. Coral recruitment and fouling Presence of zooxanthellae Visual signs of disease in E. verrucosa were evident in Sep- PCRs were performed using dinoflagellate-specific primers tember at all three sites; however, the numbers of colonies symITSFP (50-CTCAGCTCTGGACGTTGYGTTGG-30) affected were much smaller compared with previous years, and symITSRP (50-TATCGCRCTTCRCTGCGCCCT-30) where 9 per cent of colonies at 7 of 13 sites were found as described by Van Oppen et al. (2001) to amplify zoo- with disease (Hall-Spencer et al., 2007). In this study, only xanthellae ITS1 region from coral tissue DNA samples. one or two colonies were affected at the sites investigated. Affected colonies had patches of necrotic tissue that were soft and white with areas of exposed black gorgonian skele- Statistics ton where the tissue was sloughed. Healthy gorgonians typ- GelCompar (Applied Maths) was used to identify DGGE ically had tough, orange–pink coenenchyme (the bands within the bacterial profiles and construct a binary mesogloea surrounding and uniting the polyps in anthozo- matrix based on presence and absence of aligned bands. ans). In two-factor ANOVAs with Post hoc tests of the effects Levene’s test for equal variances of environmental variables of site and season on E. verrucosa recruitment and degree between sites was performed in MINITAB 6.0 (Minitab) as of fouling, site 1 had significantly higher recruitment and were ANOVA’s and Tukey tests to test for differences significantly lower fouling than site 2 and 3 (F = 30.985, between sites. In Primer-E6 v6 (Clarke & Gorley, 2006), a P < 0.001; F = 23.087, P < 0.001). Bray-Curtis similarity matrix of DGGE bands (presence/ absence data) and of clone libraries (standardised by total, Microbial community changes: DGGE analysis square root transformed) with nonmetric multidimen- of the bacterial 16S rRNA gene sional scaling (MDS) enabled visualisation of similarities between samples. Hierarchical clustering (group average) Thirty-seven E. verrucosa branches and two water samples with SIMPROF tests tested for structure in each subset of were collected for analysis of their associated microbiota.
FEMS Microbiol Ecol 90 (2014) 404–416 ª 2014 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved 408 E. Ransome et al.
While the high stress of the 2-dimensional MDS plot of the DGGE presence/absence data did not emphasise the distinction between water and coral samples, SIMPROF groupings showed these bacterial communities to be dis- tinct (Fig. 1). In June, DGGE fingerprints from colonies at site 1 and 2 grouped tightly on the MDS and showed homogeneity with SIMPROF tests (Fig. 1). In contrast, the bacterial profiles from colonies at site 3 shared similar structure to that found in coral samples from site 1 and 2 in September. In September, bacterial profiles of colonies at site 1 and 2 clustered separately to the same colonies Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021 sampled in June, suggesting a shift in community struc- ture. Site 3 profiles were mostly homogeneous but clus- tered separately from other sites in September, with one Fig. 1. Multidimensional scaling of microbial community DGGE colony profile clustering with diseased samples. Interest- bacterial profiles in E. verrucosa colonies. Labels represent June (J) ingly, this colony showed no visual signs of disease. Dis- and September (S) sites (1–3) for healthy samples. Diseased (D) eased samples from all three sites, whether in June or samples and water (W) samples are also represented. Symbols September, also grouped closely on the MDS and clus- represent Simprof test results, showing structure in bacterial tered in the SIMPROF test, away from healthy colonies, communities between samples and clustering represents similarity suggesting the presence of a distinct bacterial community between samples (50%). (Fig. 1). While June samples (Site 1 and 2) and diseased samples showed tight clustering on the MDS, September more similar to that of water samples (Table 2). At site 3, samples were more dispersed (Fig. 1). Multivariate dis- diversity was also high in September samples. Pielou’s persion analysis (MvDISP) also highlighted this (June: site evenness values (J’) showed June libraries (and site 1 Sep- 1: 0.354, 2: 0.885, 3: 1.056; September: site 1: 1.095, 2: tember) to be dominated by few bacterial classes, whereas 1.146, 3: 1.464). diseased colonies, water samples and September site 3 colo- PERMANOVA analysis (factors: site, month and health) nies showed a more even distribution of bacterial classes showed health of colonies to be the determining factor in within their communities (Table 2), with no detection of a bacterial community composition (Pseudo-F = 2.86, single dominant taxonomic group when the coral suffers P < 0.01). However, significant interactions were found from disease. Average taxonomic distinctness (D+)of between site x health and site x month (Pseudo-F = 1.87, genus level data confirmed high diversity and unrelatedness P < 0.05; Pseudo-F = 2.83, P < 0.001), indicating the of water samples. D+ was also lowest in June samples and complex nature of this data set. Without diseased samples highest in September diseased samples, with diversity val- complicating the analysis, significant differences in com- ues more similar to that of water samples, and with no munity structure were also found between site (F = 2.59, detection of a single dominant ribotype under disease con- P < 0.001) and month (F = 4.84, P < 0.001) in a two- ditions (Table 2). factor ANOVA. In pairwise tests, differences between site 1 Water libraries were dominated by the Alphaproteobac- and 3 and site 2 and 3 were responsible for overall site teria and the Bacteroidetes, but also contained a range of differences seen. other bacterial groups. Within these groups, no one spe- cies on the NCBI database was dominant, with the major- ity of clones representing distinct NCBI hits. In contrast, Microbial community composition: clone healthy coral tissues were dominated by the Gammaprote- library analysis obacteria, representing 80.0–90.0% of clones in June. In MvDISP based on genus and class level phylogeny con- September, this dominance varied, with 80.0%, 67.5% firmed DGGE results: June samples were less dispersed than and 20% of clones at site 1, 2 and 3, respectively. In Sep- September samples, indicating higher disturbance in Sep- tember, the Alphaproteobacteria increased at sites 2 and 3 tember. MvDISP also indicated an overall greater distur- and the Bacteroidetes and the Cyanobacteria at sites 1 and bance to bacterial communities at site 3 (1.52) compared 3, in addition to the Firmicutes at site 2 and the Plancto- with site 1 (0.60) and site 2 (1.19). Calculated diversity mycetes and the Deltaproteobacteria at site 3 (see Table 3 indices (Shannon–Weaver index; H’) based on class level for closest NCBI database affiliations). data from NCBI closest relatives for sequence data, high- Analysis of clone libraries from visually diseased colo- lighted an increase in diversity from June to September. nies revealed the dominance of clones affiliated to the Diversity peaked in diseased samples at all sites and was Gammaproteobacteria (site 1, June 57.5% and September
ª 2014 Federation of European Microbiological Societies. FEMS Microbiol Ecol 90 (2014) 404–416 Published by John Wiley & Sons Ltd. All rights reserved Disturbed bacterial communities in E. verrucosa 409
Table 2. Diversity indices for class level and average taxonomic distinctness for genus level clone library analysis of bacteria associated with water samples, healthy coral colonies in June (J) and September (S) and diseased (D) colonies at three sites
Water Site 1 Site 2 Site 3
J S J S J (D) S (D) J S S (D) J S S (D) Number of clones 40 40 40 40 40 40 40 40 40 40 40 40 OTUs 30 33 8 7 15 15 14 10 24 12 17 31 Shannon–Weaver diversity (H’) 2.2 2.5 0.4 0.7 1.3 1.4 0.7 1.1 2.1 0.8 2.3 2.2 Pielou’s evenness (J’) 0.77 0.89 0.31 0.47 0.71 0.64 0.37 0.65 0.79 0.43 0.90 0.79 Average taxonomic distinctness (D+) 77.4 77.2 66.7 73.8 71.6 75.2 71.0 72.2 75.2 70.2 77.7 75.7 Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021 60%; site 2, 42.5%; site 3, 27.5%), although this repre- the libraries (Site 3 September and diseased; Site 2 dis- sented a drop in number compared to healthy libraries, eased) were also found in Acropora millepora from the with the exception of site 3 in September. In diseased Great Barrier Reef (Bourne et al., 2008). Similarly, samples, the Alphaproteobacteria (site 1 June, 15%; site 3, sequences matching Achromobacter sp. were found in 30%), the Bacteroidetes (site 1 September, 22.5%; site 2, E. verrucosa as well as A. millepora (Bourne et al., 2008). 15%) and the Verrucomicrobia (site 2, 12.5%) also domi- nated. Further, clones affiliated most closely with the Presence of zooxanthellae Betaproteobacteria, the Deltaproteobacteria, the Planctomy- cetes, the Cyanobacteria and the Chloroflexi, which The presence of Spongiobacter prompted us to investigate appeared in diseased libraries, were not present in their whether E. verrucosa tissue contained zooxanthellae as it healthy counterparts (see Table 3 for closest NCBI data- has been suggested that Spongiobacter species may break base affiliations). down the high levels of DMSP produced by endosymbi- Similarities and patterns between clone libraries can be otic dinoflagellates in coral tissue (Raina et al., 2009). visualised in Fig. 2 which shows June and September However, no sequences were amplified from the tissue of libraries to the bottom left of the plots, with an exception E. verrucosa using dinoflagellate-specific primers sym- of the September site 3 library, which groups more closely ITSFP and symITSRP (Van Oppen et al., 2001) for the with diseased and water libraries, demonstrating a shift in zooxanthellae ITS1 (data not shown). community structure. While there is no pattern to the distribution of many bacterial classes over the MDS plots Discussion (e.g. the Betaproteobacteria), patterns of change can be seen in the Gammaproteobacteria, the Alphaproteobacteria Conserved bacterial communities inhabiting and the Flavobacteria (Fig. 2). the gorgonian coral E. verrucosa Diseased libraries did not show an increase in specific genera when the community changed; however, healthy This study elucidates, for the first time, that bacterial samples had a large proportion of Gammaproteobacteria communities associate with E. verrucosa. It also confirms clones that affiliated with one of two sequences in the that this shallow, cold-water coral has microbial consortia NCBI database; Spongiobacter sp. (FJ457274) and Endozoi- that differ from the surrounding environment, like those comonas montiporae (FJ347758) (Table 4). None of the seen in warm-water species (e.g. Rohwer et al., 2002; Litt- clones from water libraries matched these sequences in man et al., 2009), and endorses recent evidence that tem- the NCBI database. Interestingly, there were differences perate octocorals also contain specific bacterial genera between the number of clones affiliated with these two (Bayer et al., 2013; La Riviere et al., 2013). Previous stud- sequences found in June, September and diseased ies have shown that warm-water coral-associated bacteria libraries, by site, with total numbers originally lower at can be species specific (Bourne & Munn, 2005) and also site 3 (Table 4). A number of other clones grouped with spatially specific, with similar bacterial communities sequences retrieved from previous studies of microbial inhabiting different species of closely related coral in the diversity associated with corals. These included two same location, but differing between location (Littman unknown bacteria originally cloned from Eunicella cavo- et al., 2009). However, little is known about microbial lini, which increased from 2.5% to 15.0% at site 1, from diversity associated with temperate gorgonians (La Riviere 7.5% to 20% at site 2 from June to September and were et al., 2013). DGGE analysis in this study provides evi- also present in site 1 diseased samples (June; 12.5%) and dence of conserved bacterial genera associating with site 3 June and diseased samples (2.5%). Sequences that E. verrucosa colonies at the same site and demonstrates con- most closely affiliated with Delftia sp., found in three of servation of these bacteria between sites (1 and 2). While
FEMS Microbiol Ecol 90 (2014) 404–416 ª 2014 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved 410 E. Ransome et al.
Table 3. Affiliations in the NCBI database for clones from E. verrucosa tissue, shaded if present in June (J), September (S) and diseased (D) libraries; shades representing ≤ 10 clones (lightest grey), 1–20 clones (light grey), 20–30 clones (dark grey) and ≥ 30 clones (darkest grey) Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021
(u) indicates sequence from uncultured bacterium. communities at site 3 do not cluster as closely with site 1 sequence affiliation from water samples (dominated by and 2 on the MDS plot, similarities can be seen between the Alphaproteobacteria and the Bacteroidetes), to healthy replicate colonies at site 3 (Fig. 1 and Table 2). This sug- E. verrucosa libraries, which were dominated by the Gam- gests that although bacterial populations associating with maproteobacteria. Many studies of warm-water coral tis- E. verrucosa are conserved, environmental factors may sue have found the Gammaproteobacteria to be abundant cause bacterial populations to differ to some degree by site. (e.g. Bourne & Munn, 2005; Ainsworth et al., 2006; Although the clone libraries used in this study were Ritchie, 2006; Wegley et al., 2007; Correa et al., 2013), as small, phylogenetic analysis confirms the presence of con- have recent investigations of temperate gorgonians in the served bacterial groups, displaying distinct differences in NW Mediterranean (La Riviere et al., 2013; Vezzulli et al.,
ª 2014 Federation of European Microbiological Societies. FEMS Microbiol Ecol 90 (2014) 404–416 Published by John Wiley & Sons Ltd. All rights reserved Disturbed bacterial communities in E. verrucosa 411
2013). In culture-based studies, the dominance of the range of marine hosts and maintain their communities by Gammaproteobacteria amongst isolates is thought to be preventing the proliferation of competing or invading related to their ability to produce antibiotics that limit microorganisms, as suggested by Bourne et al. (2008). the growth of other bacteria, in addition to their resis- La Riviere et al. (2013) recently suggested that these End- tance to antibiotics (Long & Azam, 2001). In coral and ozoicomonas associates may also have differentiated to form other marine invertebrate hosts their role is still some- a stable symbiotic complex specific to a cnidarian taxon. what unclear. This was suggested due to the 16S rRNA gene sequence Of the Gammaproteobacteria found within the clone similarity of Endozoicomonas associates from the gorgo- libraries, a high proportion of sequences affiliated with the nians P. clavata and Gorgonia ventalina (GU118518; 96%) genus Endozoicomonas (also found under the name Spon- when compared to Endozoicomonas sequences from hexa- giobacter; Bayer et al., 2013) in all healthy libraries coral species (93%), suggesting that these associates (e.g. Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021 (Table 3). Members of the family Hahellaceae, order GU118518) may be a new Hahellaceae genus, that is, Oceanospirillales, this genus has been found to associate adapted to gorgonian hosts (La Riviere et al., 2013). While with the ascidians Cystodytes dellechiajei (Mart ınez-Garcia this may be true, this study shows E. verrucosa sequences to et al. 2007) and Ciona intestinalis (Dishaw et al., 2014), the be more closely related to Endozoicomonas montiporae marine sponge Halichondria okadai (Nishijima et al., 2013) (FJ347758; 97% similarity), originally isolated from Monti- and the nudibrach Elysia ornata (Kurahashi & Yokota, pora aequituberculata, an encrusting pore coral (Yang et al., 2007). It has been found to dominate DGGE bands from 2010). This finding suggests that the similarities between the sea anemone Metridium senile (Schuett et al., 2007) various Endozoicomonas sp. are not always reflected in sim- and clone libraries in pre- and post-disease corals, for ilarities between their coral hosts and calls for further example, in Acropora millipora from the Great Barrier Reef research to establish the forces that may drive the composi- (Bourne et al., 2008), and three other Acropora species tion of these coral-associated microbial communities. (Littman et al., 2009). It has also has been isolated from Montipora aequituberculata (Yang et al., 2010) and Pocillo- Spatial differences to bacterial diversity pora damicornis (Kurahashi & Yokota, 2007). More recently is has been found in the Caribbean gorgonian Gor- In this study, three sites that differed in depth, and thus gonia ventalina (Sunagawa et al., 2010) and the Mediterra- temperature and light intensity, as well as in substratum nean gorgonians Eunicella cavolini (Bayer et al., 2013) and and sediment accumulation, were chosen with the aim of Paramuricea clavata (La Riviere et al., 2013). understanding if bacterial populations associated with The dominance of this group in so many marine inver- E. verrucosa were stable and conserved. While these vari- tebrates across diverse habitats suggests that they are ables are insufficient to capture the extent of variation important members of marine invertebrate microbial between the three sites, they give clues as to some of the communities and that they play a key role in the function environmental factors at work in moulding these bacterial of the coral holobiont, supporting the hypothesis that communities. While sites 1 and 2 have similar bacterial these animals may shape their microbial partners for communities, site 3 communities are less similar, most nutritional or protective benefits. Raina et al. (2009, diverse (H’) and most disturbed (Fig. 1 and Table 2). 2010) suggested that this group may play a role in Site 3 had a significantly higher temperature in June and degrading DMSP, produced by symbiotic zooxanthellae, had significantly higher sediment loads in both months, which are known to associate with a number of corals. when compared to the other two sites (Table 1). Site 3 was This prompted an investigation into the presence of zoo- also the shallowest site (at 6–9 m). A number of factors xanthellae in E. verrucosa, with no positive result. Correa have previously been shown to correlate with changing et al. (2013) similarly found Endozoicomonas spp. domi- bacterial diversity associated with corals, including nating the Caribbean octocoral Pseudopterogorgia elisabet- increased nutrients (Bruno et al., 2003) and temperature. hae, which also lack zooxanthellae; adding weight to the Both have been shown to disrupt bacterial populations and possibility of a different or additional role for this genus increase the severity of disease in a number of corals in marine invertebrates. One such function may be pro- (Harvell et al., 2007; Ward et al., 2007). However, an tecting the host from potentially invasive microorganisms increase in temperature was only seen at site 3 in June of (Klaus et al., 2007) by competing for nutrients and niche this study, and all sites had higher temperatures in Septem- allocation (Koh, 1997; Klaus et al., 2007) and producing ber, suggesting that temperature was not a contributing antibiotics (Ritchie, 2006). For example, Ritchie (2006) factor to the increased disturbance seen at site 3. Black demonstrated that bacteria within the mucus of healthy band disease has also been shown to be more prevalent at corals inhibit the growth of other bacteria 10-fold. In this shallow sites (Kuta & Richardson, 2002), thus a factor scenario, this genus may be able to inhabit a diverse related to depth, such as nutrient load, may contribute
FEMS Microbiol Ecol 90 (2014) 404–416 ª 2014 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved 412 E. Ransome et al. Downloaded from https://academic.oup.com/femsec/article/90/2/404/2680460 by guest on 27 September 2021
Fig. 2. MDS bubble plots based on class level clone library data showing relative abundance of bacterial classes at each site. Diameter of circles (1, 2, 3, 4) represents the level of the variable for Gammaproteobacteria (40, 28, 16, 4), Betaproteobacteria (3, 2.1, 1.2, 0.3), Alphaproteobacteria (20, 14, 8, 2) and Flavobacteria (8, 5.6, 3.2, 0.8).
Table 4. Two most common affiliations in the NCBI database for clones from E. verrucosa
Spongiobacter sp. Endozoicomonas sp. Closest Relative FJ572274 FJ347758 Accession number Similarity (%) % of clone library Similarity (%) % of clone library Total % Library Site 1 June ≥ 96 30.0 ≥ 97 50.0 80.0 September ≥ 96 27.5 ≥ 97 45.0 72.5 Diseased (J) ≥ 96 17.5 ≥ 97 22.5 40.0 Diseased (S) ≥ 96 22.5 ≥ 97 30.0 52.5 Site 2 June ≥ 95 52.5 ≥ 97 10.0 62.5 September ≥ 96 22.5 ≥ 97 15.0 37.5 Diseased (S) ≥ 95 17.5 ≥ 97 10.0 27.5 Site 3 June ≥ 96 15.0 ≥ 97 22.5 37.5 September ≥ 96 2.5 ≥ 97 7.5 10.0 Diseased (S) – 0 ≥ 97 2.5 2.5 towards disturbance at this site. Finally, Voss & Richardson sediments may also play a role in coral infections. Rogers (2006) have observed increased sedimentation rates at sites (1990) estimated that mean sediment rates for reefs not of disease when compared to healthy sites, indicating that subject to stresses from human activities are <1to
ª 2014 Federation of European Microbiological Societies. FEMS Microbiol Ecol 90 (2014) 404–416 Published by John Wiley & Sons Ltd. All rights reserved Disturbed bacterial communities in E. verrucosa 413