Introduction

Studies examining bacterial diversity associated with marine invertebrates have demonstrated that these bacterial communities are highly diverse and distinct from the surrounding environment. The composition of invertebrate-associated bacterial communities has been observed to vary according to both host species and collection location, suggesting that the host invertebrate and environmental factors influence the establishment and maintenance of invertebrate–microbe associations [1–3]. The biological role of invertebrate-associated bacteria is largely unknown; however, they may play important roles in nutrient cycling, response and adaptation to environmental changes and defense against predators, pathogenic microbes, and fouling agents [4–8]. While the majority of studies regarding invertebrate microbial ecology have been conducted on scleractinian corals and sponges, octocorals have also been shown to host diverse bacterial communities [9–12]. Among octocorals, Pseudopterogorgia elisabethae is unique due to its importance as the sole commercial source of the pseudopterosins, a family of diterpene natural products (NPs) used in a variety of topical cosmetic products due to their potent anti-inflammatory properties [13]. Despite the economic importance of P. elisabethae, the microbial community associated with this invertebrate has yet to be studied. Exploring the microbiome of P. elisabethae will provide important information regarding the microbial community of healthy specimens, which may be invaluable for the future population management of this commercially harvested octocoral species.

Marine invertebrates are a rich source of bioactive metabolites, with nearly 10,000 NPs identified between 1990 and 2009 [14, 15]. Sponges (Porifera) and corals (Cnidaria) are the primary sources of invertebrate-derived marine NPs, accounting for 49 and 29 % of discovered metabolites, respectively. Within Cnidaria, approximately 94 % of metabolites have been isolated from organisms belonging to the subclass Octocorallia [15]. Despite the reported diversity of octocoral bacterial communities and the wealth of NPs isolated from octocorals [9–12, 14, 15], few studies have investigated octocoral-associated bacteria for the ability to produce bioactive NPs [16–19]. In contrast, bioprospecting from other marine ecotopes, such as sponges and sediments, has shown marine bacteria to be a rich source of bioactive metabolites [20–24]. Consequently, octocoral-associated microorganisms may represent an untapped resource for the discovery of novel marine NPs.

To gain insight into the bacterial community associated with Colombian P. elisabethae specimens collected at Providencia Island, we have characterized the associated bacterial community using both culture-dependent and culture-independent approaches. Culture-independent bacterial community analysis was accomplished by bacterial tag-encoded FLX amplicon pyrosequencing of small subunit ribosomal gene (16S) amplicons generated from genomic DNA purified from P. elisabethae. Culture-dependent microbial diversity was investigated by dilution plating of octocoral samples on agar media. Taxonomic affiliations of purified isolates were determined by partial small subunit ribosomal (16S rRNA) gene sequencing.

Materials and Methods

Sample Collection

Samples from three P. elisabethae colonies (PeA, PeB, and PeC) were collected within a 20 m radius by SCUBA (ca. 20–30 m depth) at El Planchon (13°25′0″N and 81°23′0″W), Providencia Island, Colombia. Previous research determined that NP content is greatest in the tips of P. elisabethae branches [25]; thus, samples were collected from this region. Six samples, each possessing 8–12 branchlets, were obtained from each colony. Voucher specimens (INV CNI 1612–1614) were deposited at the invertebrate collection of Museo de Historia Natural Marina Colombiana at Instituto de Investigaciones Marinas de Punta Betín. Specimens were transported to the laboratory within 2 h of collection and either immediately processed aseptically for culture-based analysis or preserved for DNA extraction. To preserve tissue for DNA extraction ca. 2 g of tissue from each sample was rinsed briefly with sterile Instant Ocean® (IO) to remove loosely associated bacteria, aseptically transferred to sterile tubes containing biological grade phenol (pH 8.0), and stored at −80 °C. A seawater sample (100 ml) adjacent to PeB (depth, 26 m) was collected in a sterile plastic bag to serve as a seawater control for culture-dependent analyses.

Culture-Independent Characterization of P. elisabethae-Associated Bacterial Diversity

DNA Extraction and Purification

DNA was extracted from preserved octocoral tissue using a modified phenol–chloroform procedure [26]. Thawed tissue samples (0.5 g) were removed from phenol, pulverized in liquid nitrogen, and suspended in 5.0 ml of lysis buffer. Large tissue fragments were removed by centrifugation (100×g, 2 min), and the supernatant was amended with 1 μl RNase One (Promega, Madison, WI), 6 mg lysozyme (Thermo Scientific, Hampton, NH), 1 ml polyvinylpyrrolidone (9.0 mg ml−1), and 0.3 ml sodium dodecyl sulfate (10 % w/v) and incubated for 20 min at 37 °C. Proteinase K (6 mg; Promega, Madison, WI) was added, and the solution was incubated at 55 °C for 2.5 h. Insoluble material was removed by centrifugation at 4,500×g for 15 min. Sodium acetate was added to the supernatant (0.3 M), and the supernatant was extracted twice with an equal volume of phenol/chloroform/isoamyl alcohol (25:24:1) and once with an equal volume of chloroform/isoamyl alcohol (24:1). DNA was precipitated from the aqueous phase by the addition of 0.7 vol of isopropanol followed by centrifugation (10,000×g, 20 min). To remove co-extracted PCR inhibitors, DNA samples were further purified using the PowerClean® DNA Clean-Up Kit (MO BIO Laboratories, Inc., Carlsbad, CA). DNA was eluted with 50 μl of 10 mM Tris–HCl, and samples were diluted to 200 ng μl−1 prior to pyrosequencing.

Pyrosequencing of 16S rDNA Amplicons and Data Processing

Bacterial tag-encoded FLX amplicon pyrosequencing (bTEFAP) was performed by Research and Testing Laboratories (Lubbock, TX) (www.researchandtesting.com) as described previously [27]. To reduce the effect of sequencing artifacts on downstream data analysis, sequence data were processed using mothur version 1.28.0 [28] according to published recommendations [29, 30]. Briefly, sequence data were denoised using the mothur implementation of PyroNoise (shhh.flows), and sequences were removed if they contained homopolymers greater than 8 bp in length, ambiguous bases, more than one mismatch to barcode sequences and more than two mismatches to the forward primer sequence, or were shorter than 250 bp. Sequences passing these quality control criteria were aligned in mothur using the Silva reference alignment obtained from the mothur website (http://www.mothur.org/wiki/Silva_reference_alignment). Alignments were screened to ensure the sequences overlapped in the same alignment space. Alignments were filtered to remove gaps, and then sequences were pre-clustered using a “diffs” setting of 2. Chimeras were identified using UCHIME and removed from the analysis. Sequences were classified using the mothur Bayesian classifier (80 % confidence) utilizing the mothur-formatted version of the Ribosomal Database Project (RDP) training set (v. 9) [31]. Sequences classified as chloroplasts, mitochondria, or “unknown” (sequences not classified at the kingdom level) were removed from the analysis. Analysis of the taxonomic affiliations of the remaining sequences by nucleotide basic local alignment search tool (BLASTN) [32] searching of GenBank revealed the presence of several chloroplast sequences which were identified as “unclassified bacteria” in mothur. These sequences were manually identified as chloroplast sequences and removed from the analysis. Sequence data have been archived in the NCBI Short Read Archive under accession number SRA043548.

Bacterial Diversity Analysis

Operational taxonomic units (OTUs) were identified using mothur as described previously [30]. OTUs were classified as described above, but using the quality filtered dataset. Graphs of order level taxonomic composition of P. elisabethae microbial communities were prepared using Microsoft Excel (2007). Observed richness (S obs), estimated richness (Chao1), and the Shannon diversity (H′) and equitability (E) indices were calculated using mothur. To enable more accurate comparisons between coral colonies, calculations were performed on datasets normalized to 1,947 sequences (equivalent to the smallest sample size). To assess richness using another commonly used analysis program, OTUs were also identified using the RDP Pyrosequencing Pipeline (http://pyro.cme.msu.edu/index.jsp) [33]. Processed sequences generated via mothur were aligned using the INFERNAL Aligner program in the RDP Pyrosequencing Pipeline and then clustered using the RDP Complete Linkage Clustering application.

Identification of Ten Most Abundant OTUs

OTUs identified by mothur (without normalization of the number of sequences in each sample) at the species level (distance of ≤0.03) were sorted based on OTU abundance. Representative sequences of each cluster were obtained using mothur, and BLASTN searching of the GenBank database was used to infer the taxonomic affiliations of representative sequences [32].

Characterization of the Culturable Bacterial Community Associated with P. elisabethae

Isolation

A 10 cm length of tissue from each specimen was rinsed with sterile IO to remove loosely associated bacteria. Tissue was cut into thin sections using a sterile scalpel and was homogenized under sterile conditions for 3 min (20 °C) in 100 ml of sterile phosphate-buffered saline (pH 7.2) using a commercial Universal blender. To isolate potential epiphytes, the surface of each colony (~10 cm2) was swabbed with a sterile cotton bud, and the swabbed material was dispersed in 1 ml of sterile IO. Serial dilutions of homogenized octocoral tissue samples, surface swab suspensions, and the seawater sample were prepared using sterile IO, and 100 μl of the 10−4 and 10−6 dilutions were spread on triplicate plates of Difco Marine Agar 2216 (MA) and Difco Sabouraud Dextrose Agar (SD) prepared with deionized water according to the manufacturer's recommendations and amended with cycloheximide (250 μg ml−1). The Parafilm®-sealed plates were incubated at room temperature and monitored for the appearance of colonies for 3 months. Representatives of unique colony morphotypes were serially subcultured until axenic on MA. Purified bacteria were examined for gross cell morphology and motility via light microscopy in combination with standard Gram-staining procedures. For long-term storage, purified bacterial isolates were grown in Difco Marine Broth 2216 and preserved in 50 % (v/v) glycerol at −80 °C.

Identification of Bacteria

Small subunit rRNA genes were amplified using a EconoTaq® Plus Green 2X Master Mix (Lucigen, Middleton, WI), the primers pA (5′-agagtttgatcctggctcag; 0.5 μM) and pH (5′-aaggaggtgatccagccgca; 0.5 μM), and template DNA (5 % v/v) [34]. Template DNA was prepared by dispersing pure colonies in 50 μl of PCR-grade DMSO (Sigma-Aldrich, Canada). PCR cycling conditions consisted of 95 °C for 5 min, 35 cycles of 95 °C for 1 min, 54 °C for 1.5 min, 72 °C for 2 min, and a final cycle at 72 °C for 7 min. Duplicate bacterial isolates were identified by restriction fragment length polymorphism analysis of 16S rDNA amplicons. Amplicons were digested with HhaI and HaeIII individually, and duplicate cultures were identified as those cultures having identical restriction patterns with both enzymes. Partial 16S rDNA sequences were obtained using the 16SR530 primer (5′-ccgcggctgctggcacgta), while nearly full-length sequences were obtained for selected isolates by sequencing with the primers pA, pH, 16SR530, and 16SF514 (5′-gtgccagcagccgcggtaa) [35]. All sequencing was performed using an ABI 3730xl DNA Analyzer (Applied Biosystems) at the McGill University and Genome Quebec Innovation Centre DNA sequencing facility. Sequences were analyzed using Vector NTI Advance 10 (Invitrogen, Carlsbad, CA) and aligned using ClustalX v.2.0.10 [36]. To determine taxonomic affiliation of cultured isolates, 16S rDNA sequences were compared to those of related strains represented in the list of prokaryotic names with standing in nomenclature [37]. Estimates of divergence between sequences were determined by calculating the number of base differences per site (p distance). Ambiguous bases were excluded from these calculations.

Phylogenetic Analysis

Phylogenetic analysis was conducted using the neighbor-joining algorithm [38] based on distances estimated by Kimura's two-parameter model [39] using MEGA 4.0 [40]. All gaps and missing data were eliminated from the dataset, and the robustness of the inferred tree topology was evaluated with 1,000 bootstrap replicates [41].

Nucleotide Accession Numbers

DNA sequences were deposited in GenBank under accession numbers JQ282802 to JQ282841 (cultured isolates) and KC561023 to KC561051 (ten most abundant OTUs from each sequence library).

Results

Culture-Independent Bacterial Diversity

After denoising, quality filtering, and removal of nonbacterial 16S rDNA sequences, sequence libraries from three P. elisabethae samples (PeA, PeB, and PeC) contained between 1,947 and 5,730 sequences with an average length of 253–260 bp. The diversity of P. elisabethae bacterial communities (S, Chao1, H′, and E indices) at the species and class level distances (D = 0.03 and 0.15, respectively) is summarized in Table 1. Non-normalized S obs values varied from 87 to 327 OTUs at the species level and from 30 to 134 at the class level. Richness (D = 0.03) calculated using the RDP and mothur differed by 9–19 %, indicating the range of richness obtained using different computational methods. As diversity measures can be strongly influenced by sampling depth [29], we calculated the observed and predicted richness as well H′ and E from normalized datasets. Comparison of these values revealed that PeA and PeB possessed similar richness, while PeC exhibited more than double the observed (2.2–2.7-fold more) and estimated (Chao1, 2.4-fold more) richness than the other coral colonies. The H′ and E index values were similar across all three coral samples, indicating similar levels of overall diversity.

Table 1 Composition, richness and diversity of bacterial communities associated with three colonies (A, B and C) of P. elisabethae determined via analysis of 16S rDNA sequence libraries
Full size table

Eight phyla (Lentisphaerae, Planctomycetes, Verrucomicrobia, Proteobacteria, Bacteroidetes, Firmicutes, Actinobacteria, and Fusobacteria) were identified in P. elisabethae sequence libraries. Five of these (Verrucomicrobia, Proteobacteria, Bacteroidetes, Firmicutes, and Actinobacteria) were found in all three sequence libraries. Proteobacteria were the dominant phylum, accounting for 87.1–94.8 % of sequences, followed by unclassified bacteria (sequences which could not be assigned to a phylum; 4.1–10.0 %), Actinobacteria (0.7–1.2 %), Bacteroidetes (0.2–0.9 %), and Firmicutes (0.1–0.2 %). At the class level, P. elisabethae bacterial communities were dominated by Gammaproteobacteria (82.2–87.1 %), unclassified bacteria (4.1–10.0 %), and Alphaproteobacteria (4.0–7.45 %). To more precisely represent similarities and differences between P. elisabethae bacterial communities, community composition was further examined at the order level and is summarized in Fig. 1. The P. elisabethae bacterial communities were dominated by Pseudomonadales (37.0–51.1 %), Oceanospirillales (18.3–29.02.3 %), unclassified bacteria (4.1–10.0 %), unclassified Gammaproteobacteria (5.4–13.7 %), Xanthomonadales (3.1–6.6 %), and Caulobacterales (1.6–6.7 %). Collectively, these six groups accounted for 90.7–91.8 % of sequences in P. elisabethae sequence libraries. All other individual orders represented less than 2.4 % of each community.

Fig. 1
figure 1

Order level bacterial community composition determined from 16S rRNA gene sequence libraries of three P. elisabethae colonies (A, B, and C). Abundance (in percent) of each order is indicated on the y-axis for orders representing ≥1 % of the total community. Orders comprising less than 1 % of the total community were grouped into an artificial “other orders” category. GPB Gammaproteobacteria, PB Proteobacteria

Full size image

To examine the contribution of species level phylotypes (D = 0.03) to the bacterial communities of P. elisabethae, the ten most abundant OTUs from each individual bTEFAP sequence library were identified (Table 2). These OTUs accounted for the majority of sequences in each library (85.0–93.1 %). Conversely, rare phylotypes (<1 % abundance) accounted for the majority of bacterial diversity present in P. elisabethae sequence libraries as the majority of OTUs were present at abundances below 1 % (PeA, 92.0 %; PeB, 89.7 %; PeC, 97.6 %). Interestingly, Pseudomonas-related OTUs were the most abundant phylotypes in each library (36.9–51.0 %). Phylotypes related to Endozoicomonas spp. sequences originating from octocorals, scleractinian corals, and sponges were also a major component (24.3–31.8 %) of each sequence library (Table 2; PeA-OTUs 2, 4, 8, and 9; PeB-OTUs 2, 5, and 8; PeC-OTUs 2, 3, 6, and 10). Brevundimonas- and Stenotrophomonas-related phylotypes were found in all sequence libraries and were the major alphaproteobacterial component of the P. elisabethae microbiome, accounting for 3.1–6.5 and 1.0–6.6 % of sequences, respectively. Four of the abundant phylotypes listed in Table 2 exhibited limited sequence identity to characterized bacteria. Two of these OTUs (PeA-OTU6 and PeB-OTU3) were identical to each other and closely related to an uncultured gammaproteobacterium identified from the octocoral Muricea elongata (DQ917904.1; Ranzer et al., direct submission); a related OTU was also present in PeC and accounted for 0.4 % of sequences in this sample (data not shown). PeB-OTU3 and PeC-OTU4 were only observed in a single sequence library. PeA-OTU3 was closely related to an unclassified bacterium identified from the octocoral Gorgonia ventalina [12] but exhibited limited sequence identity (<85 %) to cultured Spiroplasma and Mycoplasma strains. PeC-OTU4 exhibited the greatest sequence identity (97.3 %) to an uncultured cyanobacterium reported from Montipora hispida (JX022138.1; Sato et al., direct submission). The mothur classifier did not classify PeC-OTU4 as Cyanobacteria, but rather as an unclassified bacterium; thus, this OTU is represented in the unclassified bacteria category in Fig. 1. A high degree of congruence was observed between the ten most abundant species level OTUs in each P. elisabethae library. Six of the ten most abundant OTUs present in each library shared >99.2 % identity with an OTU from each of the other libraries (e.g., OTU1 from all three libraries was identical). Collectively, these six groups accounted for 77–81.2 % of each sequence library, indicating an overall high degree of similarity between the microbial communities of the three P. elisabethae colonies. The phylogenetic relationship of OTUs reported in Table 2 and observed in two or more sequence libraries is shown in Fig. 2.

Table 2 Top ten most numerous OTUs (D = 0.03) in three P. elisabethae 16S rDNA sequence libraries (PeA, PeB, and PeC).
Full size table
Fig. 2
figure 2

Phylogenetic relationships of dominant OTUs (Pe-prefix) present in two or more P. elisabethae sequence libraries. The cultured isolate, RKHC-42, is included in the tree to illustrates its relationship to dominant OTUs. Clades composed of OTUs present in all three sequence libraries are indicated with an asterisk. GPB Gammaproteobacteria. Bootstrap values greater than 50 % are shown at the nodes. The scale bar represents the number of base substitutions per site. A total of 212 positions were used in the final dataset. Pyrococcus furious was used to root the tree. GenBank accession numbers are in brackets

Full size image

Culturable Bacterial Diversity

A total of 214 bacteria were isolated from P. elisabethae: 150 from surface swabs and 64 from homogenates. Only four isolates were obtained from the seawater sample. The low number of isolates obtained from the seawater is likely due to the low starting dilution (10−4) used for dilution plating. These isolates formed 40 sequence groups sharing <99 % 16S rDNA sequence identity. The isolation source and taxonomic affiliations of the 40 sequence groups are summarized in Table 3. Comparison of isolation media revealed that MA was the most effective isolation medium, with 82.5 % of the sequence groups obtained exclusively on MA. In contrast, only 7.5 % of the sequence groups were cultured only on SD (Table 3). SD was used to mimic conditions potentially encountered in P. elisabethae tissues, such as reduced salinity, slight acidity, and increased nutrient availability. However, the low salinity, high nutrient concentration, and acidity (pH 5.6) of SD clearly did not promote the growth of the majority of bacteria associated with P. elisabethae. Examination of the distribution of sequence groups between coral and water samples identified only two sequence groups consisting of coral- and water-derived isolates (groups 6 and 18). These data suggest that the majority (92.5 %) of sequence groups identified in this study have a specific association with P. elisabethae colonies PeA, PeB, and PeC; however, this comparison should be viewed with caution given the low number of isolates obtained from the seawater sample. The 39 distinct bacterial groups isolated from P. elisabethae were evenly distributed among the phyla Actinobacteria, Firmicutes, and Proteobacteria (Table 3). At the class level, Bacilli, Actinobacteria, and Gammaproteobacteria were most common. The greatest genus level diversity was observed among the Actinobacteria, which was comprised of nine genera, followed by Gammaproteobacteria and Bacilli with five and four genera, respectively. The majority of bacteria isolated from P. elisabethae were closely related to previously cultured bacteria as indicated by high 16S rRNA gene similarities to those of type strains (Table 3). Two isolates, RKHC-14 and RKHC-63B, exhibited the least amount of similarity to previously described bacterial species (97.9 and 95.6 %, respectively) and may represent new species of Ruegeria and Vibrio, respectively. Further polyphasic taxonomic characterization would be required to conclusively determine the taxonomic novelty of the bacteria isolated from P. elisabethae.

Table 3 Bacteria isolated from P. elisabethae and the seawater control
Full size table

Analysis of the origins of bacteria isolated from P. elisabethae revealed that the majority of isolates were associated with the octocoral surface (Table 3). Of the 214 isolates, 70 % were obtained from surface swabs, suggesting an epiphytic association with P. elisabethae. Surface-associated strains were also dominant (70 %) among the 39 P. elisabethae-associated sequence groups: 44 % were only isolated from surface swabs and 26 % were isolated from surface swabs and homogenates. Twenty-five percent of the sequence groups were composed of isolates exclusively obtained from octocoral homogenates, suggesting that these isolates may have an endophytic association with P. elisabethae.

To determine if the bacteria cultured from P. elisabethae were represented in bTEFAP sequence libraries, we compared representative 16S rDNA sequences of the cultured bacteria sequence groups to the bTEFAP sequence libraries. The majority of the cultured bacterial sequence groups were either not found (70 %) in bTEFAP sequence libraries or found at very low frequencies (<0.26 %) in one or two sequence libraries (based on a >97 % sequence identity cutoff). Only sequence group 35, represented by RKHC-42, shared significant sequence identity (97.2–97.4 %) with abundant OTUs present in all sequence libraries (PeA, PeB, PeC-OTU1). The phylogenetic relationship between RKHC-42 and abundant OTUs is shown in Fig. 2.

Discussion

Culture-Independent Bacterial Diversity

Symbioses between microorganisms and marine invertebrates are essential to the functioning of reef ecosystems, as they facilitate photosynthetic productivity, nutrient cycling, and NP production [4]. Elucidation of the nature of bacterial associations with scleractinian corals and sponges are relatively well studied [1, 2, 8, 12, 20]; however, few investigations have explored the bacterial diversity associated with octocorals using modern high-throughput sequencing approaches [12, 42]. This study elucidates, for the first time, the structure and diversity of bacterial communities associated with healthy specimens of P. elisabethae using a 16S rDNA amplicon pyrosequencing approach [32].

The results obtained in this study revealed that P. elisabethae colonies from Providencia Island (Colombia) host moderately diverse bacterial assemblages as indicated by Shannon diversity index values ranging between 1.93 and 2.3 and observed (S obs) and predicted (S Chao1) richness levels ranging between 87–327 and 197–475 species level OTUs (D = 0.03), respectively. The structure of P. elisabethae bacterial communities is similar to that of other marine invertebrates in that the communities are dominated by a few abundant OTUs, while the majority of the diversity is comprised of numerous rare OTUs [12, 43, 44]. These rare phylotypes may serve as important reservoirs of metabolic diversity, which enable the coral to more rapidly adapt to abiotic and biotic changes in their environment [12]. According to the “probiotic hypothesis,” rare microbes increase in abundance when conditions favor their growth (e.g., increased temperature), thereby replacing the metabolic functions provided to the invertebrate by previously abundant microbes less adapted to the new environmental conditions [6]. In terms of composition, P. elisabethae bacterial communities were dominated by Gammaproteobacteria, a result which has been observed in several studies of octocorals and other marine invertebrates [1, 9, 44, 45]. Interestingly, the bacterial communities of P. elisabethae were dominated by two groups of gammaproteobacterial OTUs belonging to the genera Pseudomonas and Endozoicomonas. Collectively, OTUs belonging to these genera accounted for 68.7–75.3 % of each sequence library. Pseudomonas- and Endozoicomonas-related sequences are frequently detected in culture-independent studies of coral microbial diversity [46–48] and have been found to constitute a major proportion of the bacterial communities of other octocorals [12, 42]. Recent research has determined that an abundance of Gammaproteobacteria-related sequences (including Pseudomonas- and Endozoicomonas-related sequences) in invertebrate microbial communities correlates with the presence of photosymbionts in the invertebrate holobiont (invertebrate host and all associated microbes) [42]. This association is believed to be driven by the large quantities of dimethylsulfoniopropionate produced by invertebrate photosymbionts, which can serve as carbon and sulfur sources for marine bacteria [48]. P. elisabethae hosts photosymbiotic dinoflagellates (Symbiodinium) which it relies on for much of its nutritional requirements [49]. Consequently, it is likely that interactions between photosymbiotic Symbiodinium and heterotrophic bacteria may play an integral role in shaping the microbial communities of P. elisabethae. In addition to nutrient cycling, both Endozoicomonas- and Pseudomonas-related members of the P. elisabethae bacterial community may also play a role in regulating bacterial colonization of P. elisabethae via the production of bioactive secondary metabolites or probiotic mechanisms such as competitive exclusion of pathogenic bacteria [3, 46, 50, 51].

Numerous studies have demonstrated that invertebrates maintain specific microbial communities, which drastically differ from the surrounding ocean environment and are species specific and geographically stable [1, 2]. While we did not analyze the bacterial diversity present in seawater surrounding Providencia Island, the composition of P. elisabethae bacterial communities differs significantly from those reported for seawater collected from the South West Caribbean Sea (Panama) and Eastern Atlantic Ocean, which is typically dominated by phototropic Cyanobacteria (e.g., Synechococcus sp., and Prochlorococcus) and Alphaproteobacteria (e.g., Pelagibacter) [12, 52]. Additionally, a high degree of similarity was observed between bacterial communities of the three P. elisabethae colonies, with 77–81 % of each sequence library attributed to six OTUs common to all samples. These data suggest that P. elisabethae maintains a microbial consortium distinct from its environment. However, further studies are needed to elucidate how environmental factors influence the composition of P. elisabethae bacterial communities and to determine if they are species specific and geographically stable.

Culturable Bacterial Diversity

Forty distinct bacterial groups sharing <99 % 16S rDNA sequence identity were obtained from octocoral and seawater samples (Table 1). While the broadly accepted definition of a bacterial species on the basis of 16S rDNA sequence identity is >98.5 % [53], 99 % was chosen as the cutoff for this study, as many bacteria sharing this degree of sequence identity can possess significant genomic variation, which can manifest itself in the form of species- and strain-specific production of bioactive secondary metabolites [54]. A majority of the bacterial groups obtained from P. elisabethae samples were associated with the octocoral surface, while only 25 % appeared to be associated with octocoral tissue. Nutrient-rich coral mucous has previously been shown to host a diverse array of bacteria; thus, it is not surprising that the majority of isolates were surface associated [7, 55]. Additionally, some bacteria may be excluded from P. elisabethae tissues due to the presence of endogenous antimicrobial diterpenes [56]. For putative endosymbiotic isolates, further experimentation using in situ hybridization approaches would be required to prove their localization within octocoral tissues. The octocoral-associated bacteria isolated in this study were closely related to widely distributed, metabolically diverse genera commonly encountered in the marine environment. Several of the bacteria isolated from P. elisabethae are also commonly encountered in terrestrial habitats (e.g., Bacillus species), raising the possibility that these bacteria may merely be dormant terrestrial visitors to the marine environment. However, bacteria traditionally considered terrestrial in origin are frequently isolated from marine habitats, indicating that these bacteria are metabolically flexible and may be adapted to both marine and nonmarine habitats [17, 35, 57]. The diversity of bacterial groups cultured from P. elisabethae was similar to that reported in other studies of octocorals; however, the P. elisabethae culture library was less dominated by Proteobacteria and more evenly distributed between phyla [9, 10, 45]. Interestingly, comparison of the 16S rDNA sequences of cultured isolates to culture-independent 16S rDNA sequence libraries revealed a nearly total lack of overlap between cultured bacteria and bacteria detected in DNA isolated from the octocoral holobiont. Only one group of isolates (represented by RKHC-42) exhibited >97 % sequence identity to the dominant, Pseudomonas-related OTUs identified in 16S rDNA sequence libraries. Notably, isolates belonging to this group were only isolated from coral homogenates, suggesting that these bacteria may have an endophytic association with P. elisabethae. The phylogenetic relationship of this bacterium to the OTUs present in the P. elisabethae culture-independent sequence library is illustrated in Fig. 2. Although closely related to OTU1 from each sequence library, RKHC-42 forms a well-supported sister clade to these sequences, suggesting that RKHC-42 may be a different strain or species of Pseudomonas. However, given the limited nature of the dataset (212 positions), further research is needed to fully elucidate the relationship between RKHC-42 and the dominant members of the P. elisabethae bacterial community. Our inability to culture the vast majority of the bacterial diversity observed in P. elisabethae culture-independent analyses is consistent with the “great plate count anomaly” observed in many studies of culturable bacterial diversity, in which rare microbes capable of growing on complex media are preferentially cultured over abundant and potentially symbiotic bacteria [58–60]. The narrow range of isolation media and dilutions used in this study likely contributed, at least in part, to our inability to culture a more diverse and representative selection of the P. elisabethae microbiome. A greater diversity of marine bacteria may also have been cultured had marine salts been added to SD medium and the pH adjusted to that of the local seawater. The use of isolation media low in nutrients and/or containing specific host factors derived from P. elisabethae is a promising approach that could aid in the recovery of potential bacterial symbionts of P. elisabethae [61]. Additionally, the use of isolation techniques aimed at sequestering slow-growing microbes, to prevent their overgrowth by fast-growing bacteria, is another strategy that could be used to increase the diversity of bacteria isolated from P. elisabethae [61, 62].

Conclusions

This is the first report of the bacterial diversity associated with the octocoral P. elisabethae. The data presented here demonstrate that P. elisabethae hosts a diverse assemblage of bacteria comparable to that observed in other coral species. This study also represents a valuable starting point for further characterization of the microbiome of P. elisabethae to answer fundamental questions regarding the specificity and stability of P. elisabethae bacterial communities over larger geographic scales. Finally, we have established a diverse library of 39 distinct bacteria associated with P. elisabethae, which can serve as a future resource for the discovery of novel secondary metabolites.