Microbial consortia of gorgonian corals from the Aleutian islands


  • Editor: Patricia Sobecky

Correspondence: Christina A. Kellogg, US Geological Survey, 600 4th St. South, St. Petersburg, FL 33701, USA. Tel.: +1 727 803 8747, ext. 3128; fax: +1 727 803 2031; e-mail: ckellogg@usgs.gov


Gorgonians make up the majority of corals in the Aleutian archipelago and provide critical fish habitat in areas of economically important fisheries. The microbial ecology of the deep-sea gorgonian corals Paragorgea arborea, Plumarella superba, and Cryogorgia koolsae was examined with culture-based and 16S rRNA gene-based techniques. Six coral colonies (two per species) were collected. Samples from all corals were cultured, and clone libraries were constructed from P. superba and C. koolsae. Cultured bacteria were dominated by the Gammaproteobacteria, especially Vibrionaceae, with other phyla comprising <6% of the isolates. The clone libraries showed dramatically different bacterial communities between corals of the same species collected at different sites, with no clear pattern of conserved bacterial consortia. Two of the clone libraries (one from each coral species) were dominated by Tenericutes, with Alphaproteobacteria dominating the remaining sequences. The other libraries were more diverse and had a more even distribution of bacterial phyla, showing more similarity between genera than within coral species. Here we report the first microbiological characterization of P. arborea, P. superba, and C. koolsae.


In the Aleutian Islands, extending in an arc from southern Alaska, as much as 85% of commercially important fish species, such as rockfish and Atka mackerel, are associated with deep-sea corals (Heifetz, 2002; Stone, 2006). The significant economic importance of the fishing industry in the Bering Sea, the importance of corals as fish habitat, and the provisions of the Magnuson–Stevens Fishery Conservation and Management Reauthorization Act of 2006 necessitate a comprehensive understanding of these unique deep-water ecosystems. Seafloor surveys in the Aleutian Islands have revealed trawl scars covering as much as 39% of the seafloor observed (Stone, 2006), supporting concerns regarding impacts to deep-sea corals worldwide (Roberts et al., 2006).

The Aleutian archipelago supports diverse coral communities, with representatives of stony corals (Scleractinia), hydrocorals (Anthoathecatae), black corals (Antipatharia), and octocorals (Alcyonaria) (Heifetz et al., 2005; Stone & Shotwell, 2007). Aleutian corals exhibit a high degree of endemism, with >50 species found nowhere else (Stone & Shotwell, 2007). Much of what is known about the spatial distribution of corals in the Aleutian Islands has come from information on bottom trawl by-catch samples. These data indicate that gorgonians (octocorals) are major contributors to the habitat structure and the most common corals in the Aleutian archipelago (Stone & Shotwell, 2007).

One of the first published studies on bacterial associates of cold-water or deep-sea corals investigated gorgonians (bamboo corals) and antipatharians (black corals) from Gulf of Alaska seamounts (Penn et al., 2006). Since then, most of the focus has shifted to the reef-forming scleractinian, Lophelia pertusa (Yakimov et al., 2006; Neulinger et al., 2008, 2009; Kellogg et al., 2009; Schöttner et al., 2009). Both stony corals and octocorals have been shown to harbor diverse microbial communities, whether investigated by traditional culture and FISH (Brück et al., 2007; Neulinger et al., 2009) or purely by culture-independent means (Penn et al., 2006; Yakimov et al., 2006; Neulinger et al., 2008; Hansson et al., 2009; Kellogg et al., 2009; Schöttner et al., 2009). Cold-water corals are not immune to disease (Hall-Spencer et al., 2007), and, as is the case with their warmer-water counterparts, a complete understanding of the natural microbiota of the coral holobiont is critical to understanding the coral holobiont as a whole. Additionally, identifying and characterizing the often novel bacterial associates of these nonphotosynthetic corals may reveal insights into the symbiotic roles microorganisms play in coral biology. Coral microorganisms may provide other ecosystem functions such as cycling carbon, fixing nitrogen, chelating iron, producing protective antibiotics, and other beneficial activities yet to be described.

The objective of this study was to characterize the bacterial associates of three common gorgonian corals of the Aleutian archipelago using culture- and molecular-based approaches. This is the first description of the microbial communities associated with Paragorgia arborea, Plumarella superba, and Cryogorgia koolsae.

Materials and methods

Sample collection

Coral samples were collected using the ‘Kellogg II’ sampler and the two-person research submersible Delta, deployed from the RV Velero IV between June 26 and July 7, 2004. The Kellogg II sampler is a simplified version of the original Kellogg sampler (Kellogg et al., 2009) that was built specifically to be used with the Delta. The simplified sampler, consisting of two sealable acrylic containers, enabled the isolation and preservation of collected samples at depth to prevent cross contamination from contact with other samples and minimized thermal stress to samples as the submersible surfaced. The sampler allowed collection of paired duplicate samples from each coral colony. One was preserved at depth as in Kellogg et al. (2009) to fix in situ microbial diversity for molecular analyses; the second was retrieved alive for culture-based analyses.

The following metadata were recorded for each specimen collected: time, date, depth (m), location (latitude and longitude), tentative coral identification, and any special notes. The coral samples were numbered according to the official dive number of the Delta submersible, and then the individual sample number. Isolated bacterial colonies were designated by a dive number, plus a ‘C’ for ‘cold incubated (4 °C)’ or ‘R’ for ‘room-temperature incubated (21 °C)’ and an isolate number (e.g. 6201-C15). Clones were designated with a ‘B’ for ‘bacterial amplification’ and sequential numbers (e.g. 6205-B5). Table 1 lists the metadata for the sampled corals.

Table 1.   Summary of sites, samples, and 16S rRNA clone libraries
Site descriptionLat./long.Depth
SampleCoral sp.SeqsFGII
Sing., doub.,
trip., cluster
Sing., doub.,
trip., cluster
  • *

    Cultured isolates only, no clone libraries.

  • lat./long., latitude and longitude; Seqs, number of sequences; FGII, fastgroupii; sing., singletons; doub., doublets; trip., triples; cluster, four or more; OTU, operational taxonomic unit.

Cape Moffet Garden51°51.694′N/176° 49.996′W1406201P. arboreaNA*      
Little Tanaga Strait51° 52.230′N/176° 15.994′W1386203P. superba403633, 2, 1, 03.543082820, 5, 2, 1
Little Tanaga Strait51°52.073′N/176°15.249′W1106205C. koolsae494036, 3, 0, 13.532562817, 6, 4, 1
Little Tanaga Strait51°51.991′N/176° 15.147′W966206P. superba66148, 1, 3, 51.944674, 1, 0, 2
SW Tanaga Island51°34.304′N/177° 57.243′W866215C. koolsae21117, 2, 0, 11.972763, 1, 1, 1
Wall off Kasatochi Island52°10.863′N/175° 36.930′W1036229P. arboreaNA*      

Sample processing

Once aboard the support vessel, the sample pairs (preserved at depth and live) were processed as follows. Small sections were preserved in DMSO/EDTA/NaCl solution (Dawson et al., 1998) as an archive. Bacterial community DNA was extracted in triplicate (where possible) using the MoBio PowerSoil kit according to the manufacturer's instructions and immediately frozen at −20 °C.

From the unfixed samples, tissue and skeleton were cut, diced, and blended into a slurry with 1–2.5 mL of sterile phosphate-buffered saline (PBS) using flame-sterilized shears. This slurry was diluted in sterile PBS and plated in duplicate onto thiosulfate–citrate–bile salts (TCBS) agar to selectively target Vibrionaceae (Kobayashi et al., 1963) and glycerol artificial seawater agar (GASWA) as a nonselective media (modified with Instant Ocean in place of Rila Salts; Smith & Hayasaka, 1982) chosen for its common use in isolation of bacteria from shallow-water corals (e.g. Ritchie, 2006). These plates were incubated at 21 or 4 °C.

Upon return to the lab, all plates were counted, and isolated colonies were picked and archived in 50% glycerin at −80 °C. DNA was extracted from bacterial colonies using the Qiagen DNeasy Tissue kit. Extracted DNA was amplified by PCR using the 16S rRNA gene primers Eco8F (Edwards et al., 1989) and 1492R (Stackebrandt & Liesack, 1993) and sent directly to Northwoods DNA Inc. (Becida, MN) for unidirectional sequencing using the Eco8F forward primer.

Extracted DNA from the coral samples preserved at depth was also amplified using 16S primers Eco8F and 1492R. The amplicons were cleaned using the Qiagen PCR cleanup kit and ligated into the Qiagen pDrive vector at 4 °C overnight. The clone libraries were constructed following the manufacturer's instructions. Cultures frozen in glycerol were placed in 96-well plates and sent to the Genome Center at Washington University for sequencing with the Eco8F forward primer.

Sequences were base-called by phred (Ewing & Green, 1998; Ewing et al., 1998), and trimmed with Greengenes (DeSantis et al., 2006). Trimmed sequences were dereplicated with fastgroupii (Yu et al., 2006) using the percentage sequence identity with gaps option at 97%, as well as mothur (Schloss et al., 2009) and compared against GenBank with blastcl3 (Altschul et al., 1990). Sequences were nast aligned using Greengenes to collect nearest-neighbor sequences. For phylogenetic analysis, representative sequences from the clone libraries >400 bp and nearest neighbors were aligned by clustalx (Larkin et al., 2007) and trees were built in mega (Kumar et al., 2008).

The clone library sequences generated in this study are archived in GenBank with accession numbers HM173176–HM173275. The sequences from bacterial isolates are archived with accession numbers HM173276–HM173317.


Six coral colonies (two P. arborea, two P. superba, and two C. koolsae) were sampled from the submersible Delta. All samples were cultured on GASWA and TCBS, and clone libraries were constructed for samples 6203, 6205, 6206, and 6215. Sampling locations and other metadata for the sampled corals are listed in Table 1.

Clone libraries

Bacterial diversity was assessed by the construction of 16S rRNA clone libraries in the four samples 6203 and 6206 (P. superba) and 6205 and 6215 (C. koolsae). Diversity in the clone libraries was estimated by calculating the Shannon and Chao1 diversity indices, and bacterial clones were dereplicated and grouped using both fastgroupii and mothur (Table 1). The fastgroupii collated sequences were used for further analyses. The diversity indices indicated greater diversity in samples 6203 and 6205 (Table 1). The greatest diversity, including both species richness and evenness, was derived from sample 6205 (S-W index of 3.53, Chao1 of 256), as compared with the least-diverse sample, 6206 (S-W index of 1.94, Chao1 of 46). The maximum recovered dereplicated sequence numbers from the samples was 39 (6205), with the rarefaction plots for 6203 and 6205 especially indicating unsampled diversity (Fig. 1). Sample 6206 had begun to level off in number of ribotypes, indicating that sequence diversity of the sample had been mostly explored. Sample 6215 returned too few valid sequences to fully explore sequence diversity.

Figure 1.

 Rarefaction curves for all clone libraries, as calculated by fastgroupii. OTU, operational taxonomic units.

After grouping and removal of short, chimeric, nonbacterial or repeated sequences, 100 clones were used for the construction of the phylogenetic trees and comparison of the clone libraries. Figure 2 shows the distribution of bacterial taxa and their comparative proportion of the cloned communities. In samples 6203 (P. superba) and 6205 (C. koolsae), although dominated by Proteobacteria (42% and 45%, respectively), the distribution of bacterial taxa was much more even than in 6206 (P. superba) and 6215 (C. koolsae), which were both dominated by Tenericutes (64% and 70%, respectively). In sample 6203 (37 sequences after fastgroupii grouping), the largest number of sequences was obtained from the Gammaproteobacteria (24%), followed by Bacteroidetes (22%), Alphaproteobacteria (16%), Tenericutes (3%), plastids and Chloroflexi (5% each), Actinobacteria (10%) and Verrucomicrobia (8%), Spirochaetes and Acidobacteria (3% each). The distribution of sample 6205 (39 sequences after fastgroupii grouping) was Gammaproteobacteria and Alphaproteobacteria (23% each), Tenericutes (13%), Acidobacteria (8%), Verrucomicrobia and Actinobacteria (10% each), Chloroflexi (8% each), Bacteroidetes and plastids (3% each). In addition to the 64%Tenericutes, sample 6206 (14 sequences after fastgroupii grouping) contained representative sequences from the Alphaproteobacteria (21%), Bacteroidetes and Spirochaetes (7% each). Sample 6215 (10 sequences after fastgroupii grouping) had only three taxa represented in its clone library, the Tenericutes (70%), Alphaproteobacteria (20%), and Gammaproteobacteria (10%). Although the number of sequences, once grouped, was drastically different between the libraries, more valid sequences were recovered from library 6206 (n=66) than from either of the more diverse libraries of samples 6203 (n=40) and 6205 (n=49). The differences in the clone libraries obtained from these corals did not vary with the species, i.e. C. koolsae samples 6205 and 6215 did not have the most similar clone sequence distributions to each other, nor did P. superba samples 6203 and 6206.

Figure 2.

 Composition of the clone libraries showing the percentages of all taxa recovered. Groups with fewer representative sequences than 10% of the entire library are grouped into ‘Other.’ This includes plastids, Spirochaetes, Verrucomicrobia, Chloroflexi, Acidobacteria, and Actinobacteria.

As stated, all four clone libraries contained Tenericutes. These sequences were primarily mycoplasmas. Figure 3 shows the relations between this study's sequences and the other Mycoplasma-like clones and isolates.

Figure 3.

 Neighbor-joining tree of clone library Tenericute sequences, created in mega using the maximum comprehensive likelihood model. Percentages of 1000 bootstrap replicates are shown if they are >50%. Sequences in bold are from the present study; clones representing more than one sequence are indicated by the number of clones in parentheses. Reference sequences are labeled with their clone designation and accession number (in parentheses). The upper section of the tree shows three related clades of coral-associated sequences. The scale bar indicates 0.02 substitutions per nucleotide position.

Both Alpha- and Gammaproteobacteria were prominent in the clone libraries and are shown in Fig. 4. Clones clustering with groups of psychrophilic Gammaproteobacteria were scattered through the tree, including Colwellia and Glaciecola, as well as Psychrobacter. Alphaproteobacterial clusters included a large grouping with the Rhizobiales, and a cluster with the Rickettsiaceae Wolbachia and Neorickettsia.

Figure 4.

 Neighbor-joining tree of clone library sequences grouping with the Proteobacteria, created in mega using the maximum comprehensive likelihood model. Percentages of 1000 bootstrap replicates are shown if they are >50%. Sequences in bold are from the present study; clones representing more than one sequence indicated by number of clones in parentheses. Clone sequences from other studies are labeled with their clone designation and accession number (in parentheses). The scale bar indicates 0.02 substitutions per nucleotide position. Alpha- (α) and Gammaproteobacterial (γ) groups are indicated by the bars to the right.

Outside of the three dominant groups (Tenericutes, Alphaproteobacteria, and Gammaproteobacteria), there was a diverse array of clones from the seven remaining taxa (Fig. 5). These sequences grouped with clones and isolates from deep-sea and shallow-water corals, marine sediments, soils, and chloroplasts.

Figure 5.

 Neighbor-joining tree of clone library sequences grouping outside the Proteobacteria, created in mega using the maximum comprehensive likelihood model. Percentages of 1000 bootstrap replicates are shown if they are >50%. Sequences in bold are from the present study; clones representing more than one sequence indicated by number of clones in parentheses. Clone sequences from other studies are labeled with their clone designation and accession number (in parentheses). The scale bar indicates 0.02 substitutions per nucleotide position. Bacterial groups are indicated by the bars on the right, with the following key: A, Bacteroidetes; B, Chloroplasts or Cyanobacteria; C, Verrucomicrobia; D, Spirochaeta; E, Acidobacteria; F, Actinobacteria.

There were a few clones (6205-B19, 6205-B12, and 6203-B78) not shown in Fig. 5 due to lack of sequence resolution. This group affiliated with the Chloroflexi, including a clone from tundra wet meadow soil (DQ450737, Costello & Schmidt, 2006).

Cultured isolates

All cultured isolates were isolated on GASWA (at ∼21 or 4 °C). The TCBS agar yielded no isolates. The cultured isolates from the samples showed a species distribution different from that of the clone libraries, with Gammaproteobacteria making up a large number of the isolates. Table 2 contains the distribution of phylogenetic data for the cultured isolates (note 6205 yielded no isolates). Many of the groups found in the clone libraries are not represented within the cultured isolates, notably the Rhizobiales and Wolbachia groups and most of the nonproteobacterial groups. Conversely, there were no Vibrionaceae in the clone libraries, yet they make up a sizeable portion of the cultured isolates (34 out of 136, 25%). Nearly twice as many isolates were recovered when the cultures were incubated at 4 °C, near their in situ temperature, rather than at shipboard room temperature of approximately 21 °C (88 cold-grown isolates vs. 48 room-temperature-grown isolates). Figure 6 shows the phylogenetic affiliations of the cultured bacteria. Out of the 136 isolates, only four had <97% sequence similarity to sequences in the GenBank database: 6201-C21 (Colwellia sp., 96%), 6203-C18 (Colwellia sp., 96%), 6203-C15 (Octadecabacter sp., 96%), and 6229-C8 (Colwellia sp., 96%). The Colwellia-like isolates 6201-C21 and 6203-C18 were most similar to each other with 95% sequence identity, whereas they were less similar to 6229-C8 (87% and 85%, respectively).

Table 2.   Taxonomic distribution of bacterial isolates
SampleCoral sp.GammaproteobacteriaVibrionaceaeAlphaproteobacteriaBacteroidetesActinobacteria
  1. Number of isolates falling within each taxonomic group.

6201P. arborea2413202
6203P. superba26121701
6206P. superba22000
6215C. koolsae152000
6229P. arborea464010
Figure 6.

 Neighbor-joining tree of cultured isolate sequences, created in mega using the maximum comprehensive likelihood model. Percentages of 1000 bootstrap replicates are shown if they are >50%. Sequences in bold are from the present study, with the ‘C’ designator indicating isolates grown at 4°C and the ‘R’ designator indicating isolates grown at ambient temperature (approximately 21°C); isolates represented by more than one sequence are indicated with the number of isolates in parentheses. The scale bar indicates 0.02 substitutions per nucleotide position. Bacterial groups are indicated by the bars on the right.


The diversity of corals in the Aleutian archipelago is exceptional, and the region may harbor the highest abundance and diversity of cold-water corals in the world (Heifetz et al., 2005). Their bacterial communities, however, are almost entirely unknown and have not been sampled by either cultivation or cultivation-independent means. Compared with the cold-water scleractinian L. pertusa, the gorgonians of the Aleutian Islands have received very little attention.

Cultured isolates

The Gammaproteobacteria (the Vibrionaceae in particular) were dominant under all cultured conditions compared with the clone libraries. Cultured isolates from the samples grown at both 21 °C and at 4 °C indicated the importance of replicating the original growth conditions of the isolates, with half as many isolates growing at room temperature and slightly skewing selection even more toward Vibrionaceae (21% in the 4 °C isolates, 27% in the 22 °C isolates). The different bacteria recovered from cultured vs. clone library and the dominance of the Vibrionaceae within the cultured isolates indicate a notable selection bias arising from the GASWA medium. The presence of Vibrionaceae has been previously demonstrated from cold-water corals as disease-causing agents in temperature-stressed Eunicella verrucosa (Hall-Spencer et al., 2007), as has their association with disease in shallow-water corals (Kushmaro et al., 2001; Ben-Haim et al., 2003). However, Vibrios have also been shown to be a normal component of healthy tropical corals (Bourne & Munn, 2005).

There were few recovered isolates that were present in both of the samples from a given coral species. Among the cultured bacteria, all of the corals (with the exception of 6205, which failed to yield cultivable bacteria) had Vibrionaceae isolates. Both of the P. arborea samples had isolates that grouped closely with Colwellia psychrerythaea, a motile, psychrophilic heterotroph with a large repertoire of carbon-cycling abilities, as well as other nutrient-cycling functions (Methe et al., 2005) that may be important components of the biochemistry of the coral holobiont. All three species hosted close relatives to Photobacterium phosphoreum, the bacteria responsible for light production in many deep-sea fishes and other animals. Photobacterium species (also isolated on GASWA) have been determined to be common residents of the shallow-water, tropical coral Acropora palmata (Ritchie, 2006).

Clone libraries

The clone libraries were dramatically different from each other, with samples 6203 and 6205 showing far more diversity than samples 6206 and 6215 (Fig. 2). Rarefaction curves (Fig. 1) generated from the clone libraries from samples 6206 and 6215 tapered off quickly, reaching a plateau at approximately 25 operational taxonomic units. This is a pattern very similar to that seen in the white L. pertusa sampled by Neulinger et al. (2008), whereas samples 6205 and 6203 showed more diversity, generating a curve more similar to that of red L. pertusa. The rarefaction curve generated by Penn et al. (2006) for their libraries from deep-water corals (both bamboo and black corals) reaches a higher plateau. However, it is difficult to make direct comparisons because the libraries in that study were combined. The reason for the differences between these samples is not immediately clear. Samples 6203 and 6206 are both from P. superba, whereas samples 6205 and 6215 are from C. koolsae. It is striking that the differences between the corals could be so drastic, regardless of the coral species sampled.

The sites where samples 6205 and 6206 were collected are very close to each other, separated by <300 m, whereas the 6203 collection site is 790 m to the northwest, and the 6215 collection site is approximately 122 km to the southwest and all were within 3.5 km of land. Additionally, there is no temporal explanation for the differences between the clone libraries because 6203, 6205, and 6206 were all collected the same day, and 6215 was collected 5 days later. The Delta does not have the equipment to measure current speed, and so we can only note that currents were relatively strong at all locations. A notable difference between the grouping of 6203–6205 and 6206–6215 is depth. Samples 6203 and 6205 were collected deeper (138 and 110 m, respectively) than 6206 and 6215 (96 and 86 m, respectively). This forces the question of whether there is a significant difference in oceanographic environments that are separated by only 14 m depth (6205 vs. 6206, 14 m depth and 300 m horizontal distance), that may account for the observed differences. Historically, the area has a strong seasonal halocline, found at a depth of approximately 100–140 m (e.g. Coyle, 1998; Miura et al., 2002), that may affect microbial communities at various depths in the Aleutian Islands. Oceanographic measurements made during the time of collection did reveal the presence of a moderately strong thermocline at 100–110 m depth, however, that might be responsible for the observed differences in the microbial communities.

Of the clones recovered, approximately 23% have a high-ranking blast match to another deep-water coral clone, and an additional 13% have a match to a shallow-water coral clone. Very few of the sequences recovered from the clone libraries are shared between samples of the same coral species. This makes it especially difficult to determine whether there are specific bacterial symbionts for these coral species. A few sequences were found in both of the samples from C. koolsae: one group nearest to Steroidobacter denitrificans (clones 6205-B10 and 6215-B83) and an Alphaproteobacterium (clones 6205-B78 and 6215-B73) that was most closely related by blast to Anaplasma phagocytophilum, a pathogenic Rickettsiales. Rickettsiales have been described in close relationship with both diseased and healthy A. palmata and Acropora cervicornis (Casas et al., 2004). The only closely grouped clones recovered from P. superba were the Spirochaete-like sequences. These isolates and clones may make up some of the microfauna that are unique to the coral species, or at least common enough in the species to be captured by our sampling strategies.

This study is the first to report spirochete-like sequences from a gorgonian coral in the deep sea, with samples from both of the P. superba corals containing one spirochete-like clone each. The two Spirochaete-associating sequences that were recovered (6203-B80 and 6206-B24) grouped with a clone from Montastraea faveolata (Sunagawa et al., 2009) and an isolate from the Peruvian continental margin oligochaete worm Olavius crassitunicatus (Blazejak et al., 2005). Spirochaete-like sequences have been reported from a deep-sea scleractinian L. pertusa (Kellogg et al., 2009), but those sequences only share approximately 80% similarity with the clones in this study.

The large variability of bacterial assemblages between samples of even the same species of coral has been observed in previous deep-sea studies. Penn et al. (2006) noted that there were significant differences between bamboo corals sampled from the same seamount, with one coral having primarily Alphaproteobacteria and Acidobacteria, whereas the other had mostly Firmicutes (Penn et al., 2006), including Mycoplasma that were reclassified as Tenericutes at a later date. In deep-water scleractinian corals, bacterial composition has also been shown to be quite variable within the same species of coral. In L. pertusa, samples from different areas grouped together based more on collection area, with most of the samples from the deeper area (VK826) dominated by Tenericutes, whereas the shallower site was dominated by Gammaproteobacteria (Kellogg et al., 2009).

All of the clone libraries had clones that grouped into a collection of two clades within the Tenericutes (Fig. 3). One of the clades grouped together with clones from both a deep-sea octocoral from the Gulf of Alaska (DQ395563, Penn et al., 2006) and a shallow-water octocoral, Muricea elongata (DQ917875 and DQ917898; L.K. Ranzer, P.F. Restrepo, & R.G. Kerr, unpublished data). A second clade consists only of clone sequences generated in this study with representative sequences from both the C. koolsae 6215 clone library and the P. superba 6206. These two octocoral-associated clades grouped together with the clone sequences from the deep-sea scleractinian L. pertusa in the Gulf of Mexico (Kellogg et al., 2009) and ‘Candidatus Mycoplasma corallicola’ (AM911412.1) from L. pertusa in Norwegian fjords (Neulinger et al., 2009). None of the sequences recovered in this study grouped closely with the scleractinian-associated Mycoplasma, indicating that each coral group may have a group of Mycoplasma with which they coexist. Additional Tenericutes sequences were recovered from sample 6205 that grouped with Spiroplasma, Lactobacillus, and Shimazuella, and not with any other marine or coral clones.

Through the use of FISH, Neulinger et al. (2009) demonstrated that the ‘Candidatus Mycoplasma corallicola’ was located near the nematocysts within the tissue of the L. pertusa from Norway, possibly to absorb nutrients from the sloppy feeding of the L. pertusa polyps. Kellogg et al. (2009) showed that these mycoplasmas are conserved within L. pertusa, based on identical and similar Mycoplasma-like sequences in the Gulf of Mexico. Mycoplasmas have also been found in clone libraries from a shallow-water gorgonian (Fig. 3; L.K. Ranzer, P.F. Restrepo, & R.G. Kerr, unpublished data) and in metagenomic analyses of Porites compressa, a shallow-water scleractinian (Thurber et al., 2009). The prevalence of mycoplasmas in all of the clone libraries from the current study is of some note, because it is one of the few groups shared in all of the clone libraries as well as being commonly found in coral clone libraries from both deep-sea octocorals and scleractinians (Penn et al., 2006; Neulinger et al., 2008; Kellogg et al., 2009). This would indicate a possibly significant role may be played by the mycoplasmas in these corals. Mycoplasmas seem to be a key, yet understudied, part of the coral microbiome.

Many of the Gammaproteobacteria-affiliated clones (Fig. 4) grouped very closely with several isolates from various other deep-sea investigations. There was also a cluster of clones grouping with the anaerobic denitrifier S. denitrificans (Fahrbach et al., 2008), indicating that there may be many microenvironments within the coral colonies that allow for anaerobic or microaerophilic metabolism, further increasing the available niches and possibly increasing microbial diversity.

Clones clustering with the Verrucomicrobia grouped with a ridge-flank crustal-fluid clone (Huber et al., 2006), a Bering Sea sediment clone, and with another deep-sea octocoral clone (Penn et al., 2006). Verrucomicrobia have been described from both the white and the red L. pertusa from Norwegian fjords (Neulinger et al., 2008), although they did not make up nearly as high a percentage in this study (6203, 8%; 6205 10%) as in the red L. pertusa (28%). Verrucomicrobia have been isolated previously from seawater near coral (e.g. Yoon et al., 2007) and from black-band-diseased Siderastrea siderea (Sekar et al., 2006).

Actinobacteria are frequently found in corals, in both shallow (Lampert et al., 2006) and deep water (Neulinger et al., 2008), and are a target for novel natural-products research (e.g. Bull & Stach, 2007). The Actinobacteria represented 19% and 10% of the clone library sequences from the white and red (respectively) L. pertusa in a Norwegian fjord, whereas in the present study, they comprise approximately 10% of both 6203 and 6205. This indicates that, although they are not necessary to the survival of the corals (6206 and 6215 both lacked Actinobacteria), they may be common coral associates.

Several chloroplast sequences were recovered and clustered with a clone from a white-plague-diseased M. faveolata (Sunagawa et al., 2009) and were most closely related to diatom plastids. The recovery of sequences that grouped with diatom plastids in samples 6203 and 6205 is interesting, considering that the sample sites are well below the photic zone. Phototrophs such as cyanobacteria have been observed in the deep sea before, in L. pertusa clone libraries from the Gulf of Mexico (Kellogg et al., 2009), as well as shallow-water (40-m depth) L. pertusa from a Norwegian fjord (Neulinger et al., 2008). Both diatoms and cyanobacteria have been shown to be able to live heterotrophically (Lewin, 1953; Zehr et al., 2008). Although the sequences grouped with a pair of Cyanobacteria sequences, both those sequences (AM501739 and AY711425) and the sequences recovered in this study cluster more tightly in blast searches with diatom plastids than with Cyanobacteria. However, the sequences detected in this study may represent components of the corals' last meals, the remains of diatoms in the marine snow from the surface. Clones that grouped with the Chloroflexi may also be facultatively heterotrophic because Chloroflexi species have been described from deep-sea sediments (Inagaki et al., 2006), indicating more phenotypic diversity within the group than is generally assumed.

The variability between the clone libraries from the same species is large enough that we must consider that there is no ‘typical’ bacterial community associated with these species, although Tenericutes will likely be found to play a major role. It is possible that, like the ‘low-bacterial abundance’ sponges (Taylor et al., 2007), their bacterial assemblages are largely a function of the water column and their diet. However, it can be suggested that the association of bacteria such as Colwellia sp., with their nutrient- and carbon-cycling abilities, indicates that the corals may be dependent upon bacterial processes for a portion of their nutrition and that we simply need to sequence more thoroughly and attempt culture with a greater variety of media to more accurately capture bacterial diversity.

The North Pacific Fisheries Management Council and National Marine Fisheries Service recently closed vast areas of seafloor in the Aleutian archipelago to bottom trawling in order to protect sensitive deep-sea coral habitat (71 FR 36694). By understanding the microbial consortia associated with these corals, we gain a better understanding of their relationship with their environment and the ecosystem services they provide. Additional research is needed on these corals to further characterize the microbial consortia and to understand the expanded biochemical repertoire afforded to the corals by their associated bacteria. Repeated sampling would help us understand the differences we observed within the coral species and their microbial consortia.


The authors especially thank the crew of the RV Velero IV, Delta Oceanographics, and Helmut Lehnert and Dave Carlile for help in collecting the study animals. This project was partially funded by the North Pacific Research Board and National Marine Fisheries Service (Alaska Fisheries Science Center). Any use of trade names is for descriptive purposes only and does not imply endorsement by the US Government.