Microbial community structure across fluid gradients in the Juan de Fuca Ridge hydrothermal system


Correspondence: Rika E. Anderson, School of Oceanography and Astrobiology Program, University of Washington, Seattle, WA, USA. Tel.: +1 206 543 0546; fax: +1206 543 6073; e-mail: rikander@u.washington.edu


Physical and chemical gradients are dominant factors in shaping hydrothermal vent microbial ecology, where archaeal and bacterial habitats encompass a range between hot, reduced hydrothermal fluid and cold, oxidized seawater. To determine the impact of these fluid gradients on microbial communities inhabiting these systems, we surveyed bacterial and archaeal community structure among and between hydrothermal plumes, diffuse flow fluids, and background seawater in several hydrothermal vent sites on the Juan de Fuca Ridge using 16S rRNA gene diversity screening (clone libraries and terminal restriction length polymorphisms) and quantitative polymerase chain reaction methods. Community structure was similar between hydrothermal plumes and background seawater, where a number of taxa usually associated with low-oxygen zones were observed, whereas high-temperature diffuse fluids exhibited a distinct phylogenetic profile. SUP05 and Arctic96BD-19 sulfur-oxidizing bacteria were prevalent in all three mixing regimes where they exhibited overlapping but not identical abundance patterns. Taken together, these results indicate conserved patterns of redox-driven niche partitioning between hydrothermal mixing regimes and microbial communities associated with sinking particles and oxygen-deficient waters. Moreover, the prevalence of SUP05 and Arctic96BD-19 in plume and diffuse flow fluids indicates a more cosmopolitan role for these groups in the ecology and biogeochemistry of the dark ocean.


Hydrothermal vents are dynamic, gradient-dominated ecosystems supporting high levels of microbial production and consumption (McCollom & Shock, 1997; Orcutt et al., 2011). Water-rock reactions deep in the subsurface generate high-temperature fluids that emerge at the crust-water interface, causing the precipitation of minerals to form sulfide structures. These reduced hydrothermal fluids mix with background seawater, creating gradients of temperature, pH, and chemical composition. In basalt-hosted vent ecosystems, such as those found along mid-ocean ridges, diffuse flows often emerge from the seafloor at porous regions alongside sulfide chimneys and comprise a mixture of high-temperature fluid and deep seawater.

The microbial community structure of diffuse flow fluids is thought to comprise a diverse mixture of cells originating in the deep hot biosphere and the surrounding seawater interface. Although temperatures are usually below 50 °C, some portions of the archaeal and bacterial groups sampled in diffuse flow fluids are thought to represent the subsurface microbial community that has been flushed up by the vent fluid (Huber et al., 2002, 2003, 2006; Opatkiewicz et al., 2009). As a result, diffuse flow fluids generally harbor thermophilic and hyperthermophilic organisms, with particularly high diversity in the Thermococcales (Huber et al., 2006) and Epsilonproteobacteria (Nakagawa et al., 2005; Huber et al., 2007). In the deep subsurface, it is thought that a high-temperature environment dominated by hydrothermal fluid, hosting largely thermophilic organisms, gives way to a more seawater-dominated regime populated by more mesophilic organisms (Huber et al., 2003). Variations in temperature as well as the availability of different carbon sources and electron acceptors and donors result in variations in the composition of the microbial community, depending on its situation within the gradient. These mixing gradients result in high microbial diversity in diffuse fluids. Huber et al. (2007) used high-resolution pyrotag sequencing targeting the small subunit ribosomal RNA (SSU or 16S rRNA) gene to identify more than 3000 archaeal and 37 000 bacterial operational taxonomic units from diffuse flow fluids at Axial Seamount, located on the Juan de Fuca Ridge off the coast of Washington and Oregon, with different population structures at each vent reflecting different geochemical regimes.

Geochemical gradients extend beyond the subsurface seawater interface to encompass hydrothermal plumes, whose native microbial communities are not as well characterized. Hydrothermal plumes, which rise tens to hundreds of meters above the seafloor, are formed when high-temperature hydrothermal fluid is emitted from vent structures into background seawater. The reduced hydrothermal fluid entrains deeper seawater as it rises, emerging as a neutrally buoyant plume above the vent field. An estimated 70% of plume waters are entrained from deep seawater, 30% donated from seawater at the depth of the plume, and ≤ 0.01% retained from the hydrothermal source (Lupton et al., 1985). Despite the minor contribution of hydrothermal vent fluid to the rising plume, the presence of trace metals (iron, manganese), gases (hydrogen, hydrogen sulfide, methane), and other reduced substrates create new niche spaces that are distinct from the surrounding background seawater. These substrates are removed from the plume sequentially, with manganese remaining as one of the longest-lasting plume signatures in the water column (Kadko et al., 1990).

Several studies have attempted to assess the metabolism of the microbial communities based on the removal of specific compounds from hydrothermal plumes. High rates of methane oxidation measured in plumes above the Main Endeavour Segment on the Juan de Fuca Ridge were linked to methanotrophic bacteria that appeared to be entrained in the plume from deep seawater (De Angelis et al., 1993). Similarly, high rates of ammonia oxidation in plumes above the Main Endeavour Segment were influenced by the presence of organic particles and abundant particle-associated ammonia-oxidizing bacteria (Lam et al., 2008). High-temperature cultivation of manganese-oxidizing bacteria from a plume in the Guaymas Basin indicated hydrothermal fluid entrainment of subsurface bacteria into the water column (Dick & Tebo, 2010), and more recent time-series analysis of hydrothermal vent plumes in the East Pacific Rise based on automated rRNA intergenic spacer analysis identified contrasting microbial community structures associated with changes in the particle composition of the plume (Sylvan et al., 2012).

Nevertheless, no study has yet assessed hydrothermal plume and diffuse flow community structure within the context of mixing regimes and fluid gradients. Here, we charted the microbial community structure of three different mixing regimes within the hydrothermal fluid-background seawater gradient: diffuse flow fluids (with a high input of hydrothermal fluid), hydrothermal plumes (with a minimal input of hydrothermal fluid), and background seawater (with little to no hydrothermal input) using a combination of small subunit ribosomal RNA (SSU rRNA or 16S rRNA) gene clone library sequencing, terminal restriction fragment length polymorphism (T-RFLP) analysis, and quantitative polymerase chain reaction (qPCR). Our results show that while some groups are confined to very specific regimes within these mixing gradients, the sulfur-oxidizing Gammaproteobacteria of both the SUP05 and Arctic96BD-19 groups are predominant across all three sample types, suggesting that these groups are more cosmopolitan than most other phylotypes in these habitats.

Materials and methods

Sample collection

Samples were collected aboard the R/V Atlantis in June 2009 at both the Main Endeavour Field and at Axial Seamount on the Juan de Fuca Ridge. Four samples each of diffuse flow fluid and hydrothermal vent plume water as well as one background seawater sample were used in this study. Maps of these regions are shown in Fig. 1. Diffuse flow samples were collected with a hydrothermal fluid and particle sampler (HFPS) (Butterfield et al., 2004) aboard DSV Alvin. At each sample site, 2.4–3.8 L of fluid was pumped through 0.22-μm Sterivex filters (Millipore) mounted on the HFPS on the submarine. Upon shipboard recovery, filters were placed in sterile 50-mL Falcon tubes (BD Sciences Labware) and frozen at −80 °C. Diffuse flow temperatures were measured on the HFPS as the fluids from Grotto, Easter Island, and Lobo vents in the Main Endeavour Field were sampled; in all cases, the average temperature from the entire sampling process is reported. The Hulk diffuse flow sample was collected using a 200-L barrel sampler that was lowered to the seafloor on an elevator. The barrel sampler setup and sample processing is described in the study by Anderson et al. (2011). Hulk diffuse flow fluids were filtered through four Steripaks, also with a 0.22-μm pore size and frozen at −80 °C.

Figure 1.

Schematic map of the Juan de Fuca plate, Main Endeavour Field, and Axial Seamount. Adapted from figures in Huber et al. (2002) and V Robigou (unpublished data).

Hydrothermal vent plume samples were detected on the basis of temperature or transmissivity anomalies and collected using a Niskin bottle rosette mounted on a conductivity-temperature-depth profiler (Seabird). In a 4 °C cold room on board the ship, 2 L of fluid from each Niskin bottle containing plume fluids was filtered through a 0.22-μm Sterivex filter, then placed in a sterile 50-mL Falcon tube, and frozen at −80 °C until further processing onshore. Plume samples were collected in this manner above Needle vent in the Main Endeavour Field, and above Castle vent and the CASM (Canadian American Seamount) field at Axial Seamount. For the plume sample taken above Hulk vent, 50 L of plume fluid was filtered through a 0.22-μm Steripak (Millipore) and then placed in a sterile 50-mL Falcon tube and frozen at −80 °C for further analysis. A background sample (no detectable plume) was taken south of the Main Endeavour Field at 47°56.00′N, 129°04.30′ W. Sterivex filter samples were collected with a Niskin bottle rosette as described above.

Eighteen milliliters fluid subsamples were taken from each sample site for cell counting. Formaldehyde (3.7% final concentration) was added to each fluid sample and placed in a 20-mL scintillation vial, which was placed at 4 °C while on shipboard. Onshore, cells were enumerated on an epifluorescence microscope (Zeiss) using 4′,6-diamidino-2-phenylindole (DAPI) (Sigma). At least 200 cells and 20 fields of view were counted for cell quantification.

DNA extraction and purification

DNA was extracted from Sterivex filters using a modified procedure from Huber et al. (2002). Briefly, DNA extraction buffer (0.1 M Tris-HCl, 0.2 M Na-EDTA, 0.1 M NaH2PO4, 1.5 M NaCl, and 1% cetyltrimethylammonium bromide) was added to each filter. Filters were capped with Medex caps (MedEx Supply) and sealed with parafilm. Sterivexes were freeze–thawed five times by alternating between a slurry of ethanol and dry ice and a 65 °C water bath. Thirty-six microliters of 50 mg mL−1 lysozyme was then added, and the filter incubated at 37 °C for 30 min. Forty-five microliters proteinase K (1%) and 90 μL SDS solution (20%) were then added, and the filter incubated at 65 °C on a shaker for 1.5 h. Lysate was removed from Sterivex filters and centrifuged for 5 min at 6000 g. DNA was extracted from the supernatant using phenol/chloroform/isoamyl alcohol and chloroform/isoamyl alcohol as described in the study by Huber et al. (2002). For DNA extraction from Steripaks, the filter units were freeze–thawed three times by alternating between a −80 °C freezer and a 60 °C oven. DNA was extracted as described above, but scaled up to accommodate for larger volumes.

Clone library construction

Clone libraries were constructed from the Hulk diffuse flow, Hulk plume, and background samples. Bacterial and archaeal 16S rRNA genes were amplified for clone library construction with GoTaq DNA polymerase (Promega) using universal bacterial primers 8Fb (Edwards et al., 1989) (5′-AGAGTTTGATCCTGGCTCAG-3′) and BAC1492R (Stackebrandt & Liesack, 1993) (5′-RGYTACCTTGTTACGACTT-3′) and universal archaeal primers ARC21F (DeLong, 1992) (5′-TTCYGGTTGATCCYGCCRGA-3′) and ARC922R (Opatkiewicz et al., 2009) (5′-YCCGGCGTTGANTCCAATT-3′). For PCR amplification, an initial denaturing step of 94 °C for 5 min was followed by 30 cycles of 94 °C for 30 s, 45 °C for 30 s, and 72 °C for 2 min for bacteria, followed by a 72 °C extension step for 10 min. Attempts at amplification of the target region at annealing temperatures above 45 °C were unsuccessful, and so this temperature was used for all samples. For archaea, the annealing temperature was 55 °C, and only 24 cycles were used. To minimize the formation of heteroduplex molecules, PCR products were reconditioned prior to cloning by using PCR product as template in a new PCR cocktail and repeating the thermocycling protocol for 5–10 cycles (Thompson et al., 2002).

Bacterial PCR products were cloned using the TOPO TA cloning kit (Invitrogen), and archaeal PCR products with the StrataClone PCR cloning kit (Agilent Technologies) according to manufacturer's instructions. Clones were amplified with the M13F (5′-GTAAAACGACGGCCAG-3′) and M13R (5′-CAGGAAACAGCTATGAC-3′) primers. Sequencing was conducted through the University of Washington High Throughput Genomics Center on an ABI 3730xl sequencing unit (Applied Biosystems). Primers used for sequencing included T3 (5′-ATTAACCCTCACTAAAGGGA-3′) and T7 (5′-TAATACGACTCACTATAGGG-3′), and either 515Fb (5′-GTGCCAAGCMGCCGCGGTAA-3′), 907Rb (5′-CCGTCAATTCMTTTRAGTTT-3′) or 110R (5′-GGGTTGCGCTCGTTG-3′) for bacterial clones, and 515Fa (5′-GTGGCASCMGCCGCGGTAA-3′) for archaeal clones. Contigs were assembled using sequencher 4.9 (Gene Codes Corporation). Sequences were aligned and checked for chimeras using greengenes (DeSantis et al., 2006). Taxon assignment was performed based on blastn queries against the greengenes (DeSantis et al., 2006) and ARB (Ludwig et al., 2004) 16S rRNA databases. Resulting outputs were summarized in table format and visualized as a dot plot using the custom perl script, bubble.prl (http://www.cmde.science.ubc.ca/hallam/bubble.php). Clone library diversity indices were calculated using mothur (Schloss et al., 2009) at a clustering distance of 0.03.

Phylogenetic analysis

Trees were constructed by aligning clone sequences and representative sequences from the NCBI database using the greengenes pipeline (DeSantis et al., 2006), then importing into ARB for comparison with reference sequences. Evolutionary history was inferred using the maximum likelihood method based on the Tamura-Nei model (Tamura & Nei, 1993). A total of 1000 replicates were used in the bootstrap test (Felsenstein, 1985). Initial trees for the heuristic search were obtained automatically as follows: when the number of common sites was < 100 or less than one-fourth of the total number of sites, the maximum parsimony was used; otherwise the BIONJ method with MCL distance matrix was used. All phylogenetic analyses were implemented using mega5 (Tamura et al., 2011).

T-RFLP community profiling

Bacterial and archaeal 16S rRNA genes were amplified for T-RFLP using universal primers ARC21F (5′-[6-FAM]TTCYGGTTGATCCYGCCRGA-3′), ARC922R (5′-YCCGGCGTTGANTCCAATT-3′), BAC68F (5′-[6-FAM]TNANACATGCAAGTCGRRCG-3′) and BAC1492R (5′-RGYTACCTTGTTACGACTT-3′). Each PCR (25μL) contained 1× GoTaq buffer (Promega), 1 U GoTaq polymerase (Promega) 2 mM MgCl2, 2 mM dNTPs, and 0.4 μM each primer. An initial denaturation step of 94 °C for 5 min was followed by 34 cycles of 94 °C for 30 s, 55 °C for 45 s, and 72 °C for 2 min for bacteria, with a final extension of 72 °C for 10 min. For archaea, the annealing step was 55 °C for 30 s, and only 23 cycles were used. To minimize PCR drift (Polz & Cavanaugh, 1998), between 5 and 10 replicate reactions were pooled and then cleaned and concentrated with QiaQuick PCR purification columns (Qiagen) according to the manufacturer's instructions.

Cleaned PCR products were digested with restriction endonucleases HaeIII or BstUI for all samples, plus MspI for bacterial PCR products and RsaI for archaeal PCR products. All digests were incubated overnight at 37 °C, except for BstUI, which was incubated at 60 °C. Digests were inactivated by freezing the solution at −20 °C. Samples were ethanol precipitated, dried, and resuspended in 0.25 μL ET900-R MegaBACE size standard, 4.75 μL 70% formamide/1 mM EDTA loading buffer, and 5 μL water. T-RFLP profiling runs were conducted on a MegaBACE 1000 (GE LifeSciences).

T-RFLP profiles were processed using DAx (2006 Van Mierlo Software Consultancy, the Netherlands). Peaks were standardized using the variable percentage threshold method (Osborne et al., 2006) and then normalized between each sample according to peak height. Peaks were binned into eight different bin shifts according to the method outlined by Hewson & Fuhrman (2006). Bins were 4 bp wide and shifted by 0.5 bp for each bin shift. For diversity indices, the average of the eight bin shifts was used, and indices were calculated using estimates (Version 8.2; R.K. Colwell). To create resemblance matrices, the maximum similarity of each of the eight bin shifts was calculated using estimates and hierarchical clustering dendrograms were created using the group average method in primer-e v.6.1.6 (Clarke & Gorley, 2006). Cophenetic correlation coefficients were determined by calculating the correlation coefficient between the resemblance matrix and the cophenetic matrix created by clustering. To identify T-RFLP peaks, clone sequences were trimmed using bioedit 7.0.9 (Ibis Biosciences) and digested in silico using the program REPK (Collins & Rocap, 2007). Resulting fragments were compared with fragments obtained from restriction digests; peaks were positively identified if they fell within 2 bp of an in silico clone fragment.


Relative percentages of SUP05 and Arctic96BD-19 compared to total bacteria were determined using qPCR. Total bacteria were quantified using a bacteria-specific forward primer (27F, 5′-AGAGTTTGATCCTGGCTCAG-3′) and a universal reverse primer (DW519R, 5′-GNTTTACCGCGGCKGCTG-3′) (Zaikova et al., 2010). SUP05 was quantified using a bacteria-specific forward primer (Ba519F, 5′-CAGCMGCCGCGGTAANWC-3′) and a group-specific reverse primer (1048R_SUP05, 5′-CCATCTCTGGAAAGTTCCGTCT-3′) (Zaikova et al., 2010). Arctic96BD-19 was quantified using Ba519F and a group-specific primer, 1048R_Arctic (5′-CTATTTCTAGAAAGTTCGCAGG-3′) (Walsh & Hallam, 2011). Each 20-μL reaction contained 2 μL sterile DNase free water, 2 μL each of 5 μM forward and reverse primers, 4 μL template, and 10 μL SsoFast EvaGreen Supermix (Bio-Rad Laboratories, CA). Reactions were carried out in 48-well white plates with optical caps (Bio-Rad). Reactions were run on a MiniOpticon Real-Time PCR System (Bio-Rad). Universal bacterial primers were run with the following protocol: initial denaturation at 95 °C for 3 min, followed by 45 cycles of 95 °C for 20 s, primer annealing for 30 s at 55 °C for total bacteria, 63 °C for SUP05, and 59 °C for Arctic96BD-19, and a plate read. The melt curve extended from 55 to 95 °C, increasing by 0.5 °C s−1. Data were analyzed with the CFX Manager for the MiniOpticon system (BioRad). A standard curve was created for each of the primer sets and run in parallel with each of the samples. A 10-fold dilution series of standards ranging from 4.3 × 102 to 4.3 × 105 (Arctic96BD-19) or 8.5 × 102 to 8.5 × 105 (SUP05) was prepared for each run. These standards were also used for quantification of total bacteria. To mitigate the impact of inhibitors (Lloyd et al., 2010), samples were run at either 1/10 or 1/100 dilutions, and the dilution level was kept consistent for each sample across each of the three primer sets. All samples were run in duplicate, and ratios of SUP05 or Arctic96BD-19 to total bacteria were carried out by averaging all four ratio combinations from each set of duplicates. Standard error of the percent abundances were calculated from the standard error of all four ratio combinations.

Nucleotide sequence accession numbers

The GenBank nucleotide sequence accession numbers for the sequences in this study are JQ678046 through JQ678591.


A total of four diffuse flow, four plume, and one background seawater samples were collected from the Main Endeavour Field and at Axial Seamount on the Juan de Fuca Ridge in June 2009. Sample number, location, temperature, and cell count data are listed in Table 1. The concentration of Mg (33.3 mmol kg−1) and dissolved silica (6.38 mmol kg−1) of the diffuse flow samples from Hulk vent indicate an average temperature of 125 °C (Anderson et al., 2011). This fluid sample also had the highest cell counts (Table 1). The unique characteristics of this sample can be partially attributed to the nature of the sampling method: a funnel was attached to the sample hose on a barrel sampler, which was placed atop a region of diffuse flow covered in tube worms. The strong suction of the barrel sampler may have sealed the funnel onto the surface of the vent and drawn out higher temperature water from within the sulfide structure. The nozzle on the HFPS, in contrast, collected samples consistent with the temperatures measured within the animal communities on the surface of the vents.

Table 1. Summary of sample locations, depth, temperature/temperature anomaly, oxygen, and cell counts
SampleLatitude/longitudeDepth (m)Temperature (°C)Temperature anomaly (°C)Oxygen (uM)Cell counts (cells mL−1)
  1. For plume samples, temperature is reported as the temperature anomaly, which is calculated as the difference in temperature between the plume temperature spike and the background seawater temperature.

  2. a

    Temperature of the Hulk diffuse flow sampled was calculated from magnesium and silica concentrations in the fluids. This calculation is explained in greater detail in Anderson et al., 2011.

Background47°56.00′N 129°04.30′ W23001.8385.654.94E+04
Needle plume47°56.875′N 129°5.940′W21351.870.0686279.993.93E+04
Castle plume45°55.5690′N 129°55.8242′W14742.420.0138341.725.09E+04
CASM plume45°59.326′N 130°1.634′W15252.380.00144.151.21E+05
Hulk plume47°57.024′N 129°5.762′W20601.900.112982.457.37E+04
Hulk DF47°57.00′ N, 129°5.81′ W2198125a 1.69E+07
Lobo DF47°56.952′N, 129°5.910′W21887.1 5.71E+04
Grotto DF47°56.95′N, 129°5.898′W218718.1 3.62E+04
Easter Island DF47°56.880′N, 129°5.940′W21979.3 2.82E+05

Clone libraries

16S rRNA gene clone libraries were constructed to compare microbial community composition in diffuse flow fluid (with a high input of hydrothermal fluid), hydrothermal plume (with a minimal input of hydrothermal fluid), and background seawater (with little to no hydrothermal input). Both the diffuse flow and plume clone libraries were constructed from samples taken at Hulk vent to compare fluids from the same vent structure; it should be noted that the diffuse flow sample from Hulk vent was at an extremely high temperature and therefore represents the extreme end of the spectrum within the mixing regime of hydrothermal fluid and seawater.

Proteobacteria dominated 16S rRNA gene clone libraries across all three mixing regimes (Fig. 2). However, various subdivisions within the Proteobacteria exhibited distinct distribution patterns between samples. In both vent plume and background seawater, Alpha, Gamma, and Deltaproteobacteria, including the SAR11 cluster, Agg47, Hyd24-01, SUP05, Arctic96BD-19 and ZD0417, Myxococcales and SAR324, respectively, were prevalent. SUP05 bacteria were most abundant in the vent plume sample, contributing 39% of total bacterial clones (Supporting Information, Fig. S1). ZA3420c, Arctic96B-1, Geobacter and NB1-I were recovered solely from background seawater. Within the diffuse flow sample, Alpha, Beta, Gamma, and Epsilonproteobacteria, including the SAR11 cluster, Sphingomonas, Comamonas, Alteromonas, Pseudomonas, and Caminibacter, were prevalent. In addition to Proteobacteria, Microthrix, Chloroflexi, and Marine Group A were also recovered from vent plume and background seawater (Fig. 2). Groups affiliated with the Bacteroidetes were recovered from background seawater and diffuse flow samples, and groups affiliated with Planctomycetes and Verrucomicrobia were recovered from vent plume, diffuse flow, and background seawater samples (Fig. 2).

Figure 2.

Dot plot of (a) bacterial and (b) archaeal diversity from diffuse flow fluid and hydrothermal plume associated with Hulk vent, as well as background seawater, based on 16S rRNA gene sequence profiles (see methods). The size of each dot indicates the percentage identified 16S rRNA gene sequences falling within a particular taxonomic group. The number of bacterial clones sequenced per sample is background seawater = 78, vent plume = 102 and diffuse flow = 81. The number of archaeal clones sequenced per sample is background seawater = 96, vent plume = 95 and diffuse flow = 94.

In the case of the archaeal domain, two major lineages affiliated with Marine Group I and II archaea were recovered from vent plume and background seawater samples. The proportion of Marine Group I and Marine Group II archaea was similar in both samples, with Marine Group I contributing 69.5% and 66.7% and Marine Group II 18.9% and 20.8% of total archaeal clones to vent plume and background seawater, respectively (Fig. 2). In contrast, Methanococci and Thermococci were exclusively identified in diffuse flow fluids, where they comprised 8.5% and 85.1% of total archaeal clones, respectively (Fig. 2). Groups affiliated with Marine Group I, Marine Group II, and Thermoprotei were also recovered from the diffuse flow sample, ranging between 1% and 2% of total archaeal clones.

T-RFLP community profiles

To determine whether community composition patterns recovered in clone libraries from the Hulk vent were representative for the Juan de Fuca Ridge system, T-RFLP profiles for archaeal and bacterial 16S rRNA genes were obtained across multiple diffuse flow and plume samples from different vent locations (Table 1). Peaks were identified based on in silico digestion of clone library sequences (see Methods).

Archaeal T-RFLP profiles indicated that the majority of plume and diffuse flow samples were strikingly similar, with Marine Groups I and II dominating the community structure (Fig. 3a). Plume T-RFLP traces, such as the Castle trace shown in Fig. 3a, were characteristic of all archaeal samples digested with RsaI, with the exception of the Hulk diffuse flow sample. This extremely high-temperature sample, shown in Fig. 3b, exhibited a unique community profile, with Thermococcus and Methanocaldococcus groups dominating the community. T-RFLP community profiling with other restriction enzymes indicated similar patterns. Clustering of T-RFLP profiles based on the Chao abundance-based Jaccard Index (Fig. 3c) indicates that the Hulk diffuse flow sample was < 20% similar to all other samples, with most of the other plume and cooler diffuse flow samples clustering together at over 70% similarity, although some heterogeneity is evident in the Hulk plume and background seawater samples.

Figure 3.

T-RFLP community profiling of samples amplified with universal archaeal primers. (a) Representative T-RFLP trace of plume samples digested with restriction enzyme RsaI. (b) T-RFLP trace of Hulk diffuse flow sample. Y-axes are relative fluorescence units (RFUs) and are not to scale. Peaks were identified through in silico digestion of clone library sequences using the online tool REPK (Collins & Rocap, 2007). (c) Cluster diagrams of sample similarity based on archaeal sample T-RFLP traces digested with BstUI. Distance matrices were produced using Chao's abundance-based Jaccard Index (Chao et al., 2005) and were calculated from the maximum similarity of eight different bin shifts. Samples were clustered using the group average method in primer-e. Cophenetic correlation coefficient for the dendrogram is shown.

Bacterial profiles exhibited a much higher degree of variation based primarily on differences in the presence or height of minor peaks. In samples digested with HaeIII, a 369-bp peak corresponding to sulfur-oxidizing Gammaproteobacteria groups SUP05 and Arctic96BD-19 was visible in all but two of the T-RFLP traces. Examples of this peak can be seen in the plume sample from Needle (Fig. 4a) and in the diffuse flow sample from Easter Island (Fig. 4b). Diffuse flow samples isolated from Hulk and Grotto vents, the two highest-temperature samples, were the only samples lacking this peak. Samples digested with restriction enzymes MspI and BstUI did not resolve a peak unique to SUP05 or Arctic96BD-19, but community similarity analyses from these samples did indicate trends corresponding to the type of environment from which samples were taken. A community similarity cluster dendrogram based on samples digested with BstUI (Fig. 4c) and showed that the very high-temperature Hulk diffuse flow sample clustered separately from all other samples at a very low level of similarity. Other samples clustered roughly according to the temperature at which they were sampled: diffuse flow samples Lobo and Grotto clustered together at about 70% similarity, while the plume samples and the Easter Island diffuse flow sample clustered together at about 60% similarity. Easter Island was one of the two lower-temperature diffuse flow samples taken for this study. Plume samples from CASM and Castle vents, with the lowest temperature anomalies of the plumes sampled here, clustered together as well, while the background seawater sample clustered at a low level of similarity with the cooler diffuse flow and plume samples.

Figure 4.

T-RFLP community profiling of samples amplified with universal bacterial primers. (a) Representative T-RFLP trace of plume samples digested with restriction enzyme HaeIII. Needle is shown here. (b) Representative T-RFLP trace of diffuse flow samples digested with restriction enzyme HaeIII, and Easter Island is shown here. Y-axes are relative fluorescence units (RFUs) and are not to scale. Peaks were identified through in silico digestion of clone library sequences using the online tool REPK (Collins & Rocap, 2007). (c) Cluster diagrams of sample similarity based on archaeal sample T-RFLP traces digested with BstUI. Distance matrices were produced using Chao's abundance-based Jaccard Index (Chao et al., 2005) and were calculated from the maximum similarity of eight different bin shifts. Samples were clustered using the group average method in primer-e. Cophenetic correlation coefficient for the dendrogram is shown.

Diversity indices calculated for both clone libraries and T-RFLP profiles indicated that the hydrothermal plume and diffuse flow samples, in general, had higher diversity than background seawater for archaea across all diversity indices calculated (Sobs, Chao1, and Jackknife) (Table 2). Within the hydrothermal plume samples, the two plume samples with a higher temperature anomaly (Hulk and Needle) had higher diversity than the plume samples with a lower-temperature anomaly (Castle and CASM), in both the archaeal and the bacterial domains, across all diversity indices reported. However, on the whole, no clear trend in terms of relative diversity emerged between plume and diffuse flow samples in either the bacterial or archaeal domains. Finally, while the diversity of the bacterial community in background seawater appeared to be lower according to T-RFLP analyses, clustering of clone libraries at 97% resulted in a higher number of observed OTUs in the background sample than in the other samples. This discrepancy could be the result of diversity in the 16S gene that could not be detected with the restriction enzymes used.

Table 2. Diversity indices calculated using various methods for plume, diffuse flow, and background samples used in this study
DomainSample locationSample typeMethodSobsaChao1bJackknifec
  1. T-RFLP diversity indices are listed as the average across all restriction enzymes used ± standard deviation of the average. Bacteria estimates are averaged across T-RFLP profiles for BstUI, HaeIII, and MspI; archeal estimates averaged across T-RFLP profiles for BstUI and HaeIII. T-RFLP diversity indices were calculated using estimates (Version 8.2; R.K. Colwell); clone library diversity estimates were calculated using mothur (Schloss et al., 2009). See Methods for details.

  2. a

    Number of OTUs observed. For clone libraries, this is the number of clusters at a distance of 0.04. For T-RFLP, this is the Mau Tau expected richness (Colwell et al., 2004).

  3. b

    Chao1 richness estimator (Chao, 1987).

  4. c

    First-order Jackknife richness estimator (Burnham & Overton, 1978, 1979; Smith & van Belle, 1984; Heltshe & Forrester, 1983).

BacteriaDeep seawaterDeep seawaterClone library51174.5226.94
Deep seawaterDeep seawaterT-RFLP14.68 ± 4.0614.6 ± 4.0116.52 ± 7.81
HulkPlumeClone library45107110.96
HulkPlumeT-RFLP46.375 ± 1.7346.31 ± 1.8365.91 ± 4.97
NeedlePlumeT-RFLP52.14 ± 3.0852.09 ± 3.0873.30 ± 6.31
CASMPlumeT-RFLP23.02 ± 3.1022.98 ± 3.0232.52 ± 4.33
CastlePlumeT-RFLP29.72 ± 2.8529.43 ± 2.8743.52 ± 4.61
HulkDiffuse flowClone library42117.6117.63
HulkDiffuse flowT-RFLP43.52 ± 1.2143.45 ± 1.2362.58 ± 3.92
Easter IslandDiffuse flowT-RFLP34.54 ± 3.3834.3 ± 3.0349.96 ± 4.78
LoboDiffuse flowT-RFLP48.91 ± 2.2148.71 ± 2.2568.71 ± 5.62
GrottoDiffuse flowT-RFLP39.03 ± 2.5338.92 ± 2.3156.59 ± 4.04
ArchaeaDeep seawaterDeep seawaterClone library121717
Deep seawaterDeep seawaterT-RFLP10.08 ± 6.7110.39 ± 6.5510.39 ± 12.28
HulkPlumeClone library141920
HulkPlumeT-RFLP88.81 ± 5.9188.96 ± 6.04124.425 ± 9.31
NeedlePlumeT-RFLP63.26 ± 5.7363.27 ± 5.7986.74 ± 9.47
CASMPlumeT-RFLP32.40 ± 2.9832.80 ± 2.9347.47 ± 4.27
CastlePlumeT-RFLP41.12 ± 3.4841.39 ± 3.5259.75 ± 4.44
HulkDiffuse flowClone library185148.00
HulkDiffuse flowT-RFLP55.26 ± 5.5655.41 ± 5.7277.89 ± 8.56
Easter IslandDiffuse flowT-RFLP45.81 ± 3.2446.02 ± 3.1665.98 ± 4.18
LoboDiffuse flowT-RFLP60.14 ± 6.2960.19 ± 4.6783.33 ± 10.32
GrottoDiffuse flowT-RFLP52.56 ± 3.6952.71 ± 3.7374.74 ± 5.37

SUP05 and Arctic96BD-19 diversity and abundance

Given the prevalence of SUP05 and Arctic96BD-19 among and between mixing regimes, we conducted a more in-depth phylogenetic analysis based on 16S rRNA gene sequences recovered from Hulk samples and other marine ecosystems to better constrain biogeographic or ecological type (ecotype) relationships. SUP05 and Arctic96BD-19 16S rRNA gene sequences recovered from Hulk plume and background seawater samples partitioned into previously defined clades (Fig. 5). Specifically, most of the plume clones in this cluster grouped with other SUP05 samples obtained from vent environments, such as the Suiyo Seamount (Sunamura et al., 2004), or with vent endosymbionts. In contrast, the majority of background 16S rRNA gene sequences fell into the Arctic96BD-19 group, along with clones recovered from the northeastern subarctic Pacific (Walsh et al., 2009) and the San Pedro Channel, CA (Brown et al., 2005), as well as from the Saanich Inlet (Walsh et al., 2009) and the Namibian shelf (Lavik et al., 2008).

Figure 5.

Evolutionary relationships of clones and reference sequences within the sulfur-oxidizing Gammaproteobacteria group. See Methods for techniques in tree construction. The evolutionary history was inferred using the maximum likelihood method; the bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed. Branches corresponding to partitions reproduced in less than 50% bootstrap replicates are collapsed. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. Clones from this study are shown in red, clones from other hydrothermal systems are shown in green, and symbionts are shown in blue. GenBank accession numbers are provided for clones not from this study.

We next used quantitative PCR to determine 16S rRNA gene copy numbers for SUP05 and Arctic96BD-19 in relation to total bacteria across the diffuse flow and plume samples used in T-RFLP analysis. The SUP05 group was abundant in the plume samples, reaching up to 27% of the total bacteria (Table 3). SUP05 16S rRNA gene copy number decreased in samples with a weaker plume signature such as CASM, at 4.1%. This was consistent with reduced recovery of SUP05 in the background sample, at 3.2%. SUP05 16S rRNA gene copy number was also high in most diffuse flow samples, reaching up to 18.7% in the Easter Island sample. However, the relative abundance decreased dramatically in the high-temperature Hulk diffuse flow sample, dropping to 0.4% of total bacteria. Arctic96BD-19 16S rRNA gene copy number was high across all sample types, reaching up to 64.7% of the bacterial community in the CASM plume sample (Table 3). Moreover, Arctic96BD-19 16S rRNA gene copy number tended to decrease as plume signatures became stronger. However, this trend did not continue for the diffuse flow samples or the background sample, where Arctic96BD-19 16S rRNA gene copy number reached up to 25.5% and 22.7%, respectively. Similar to SUP05, Arctic96BD-19 16S rRNA gene copy number decreased in the high-temperature Hulk diffuse flow sample, but was still greater than that of SUP05.

Table 3. Relative abundances of SUP05 and ARCTIC96BD-19 across all sample types
SampleTemperature/temp anomaly (°C)SUP05% abundance (relative to total bacteria)Standard error of SUP05% abundanceArctic96BD-19% abundance (relative to total bacteria)Standard error of Arctic96BD-19% abundance
  1. Abundances quantified through quantitative PCR. Quantities are expressed as percentages of each group relative to total bacteria.

Diffuse flow
Easter Island9.318.70.9511.90.92


Geochemical gradients resulting from mixing between reduced, high-temperature hydrothermal fluid and cooler, oxidized seawater are a dominant feature of hydrothermal vent ecosystems. Temperature, considered a proxy for chemistry in these systems, is positively correlated with sulfide and negatively correlated with oxygen (Corliss et al., 1979; Johnson et al., 1986). The availability of different electron acceptors and donors changes with the degree of mixing between fluid types. As a result, diffuse flow samples are enriched in reduced compounds including hydrogen, sulfide, ammonia, and iron relative to plume or background seawater; and plume samples, in which only 0.01% of the fluid is derived from a hydrothermal source, still manifest elevated concentrations of reduced compounds and metals relative to background seawater (Kadko et al., 1990). By studying microbial community structure within these gradients, we can better understand patterns of redox-driven niche partitioning and adaptive radiation among and between microbial groups in the dark ocean. Samples in the current study ranged between 9 and 125 °C among diffuse flow samples, with temperature anomalies between 0.00 and 0.011 °C in plume samples indicating multiple different geochemical conditions. While more exhaustive geochemical analyses were not available for this particular study, some patterns were revealed in these analyses that are worthy of note and can act as a starting point for future analyses of community partitioning across geochemical gradients in these systems.

Archaeal community composition was relatively homogeneous between samples, with similar communities found across background seawater, plume, and diffuse flow samples. Marine Group I and II archaea from diffuse flow samples collected between 9 and 18 °C were similar to plume and background samples, suggesting that members of these groups are adapted to a wide range of temperatures and geochemical conditions. In contrast, in the high-temperature Hulk diffuse flow sample, at 125 °C, Marine Groups I and II were almost entirely replaced by Thermococcales and Methanococcales. Consistent with this observation, the optimal growth temperatures of Thermococcus and Methanocaldococcus strains range between 80 and 100 °C, confining them to a narrow range within the hydrothermal fluid-seawater gradient where the increased proportion of high-temperature hydrothermal fluids enrich for different taxa.

As with the archaeal communities, the high-temperature diffuse flow sample from Hulk was found to be unique among the bacterial communities. This sample was dominated by Epsilonproteobacteria in the Caminibacter and Nautiliales groups, which has been observed previously in diffuse flow fluids (Huber et al., 2003, 2007). These groups appear to flourish in the warmer, more reduced fluids characteristic of higher temperatures in these vent systems. The bacterial communities in general were quite heterogeneous between samples, a trait also observed previously in vent systems (Opatkiewicz et al., 2009), and may be the result of differences in vent chemistry between sites, as well as microbial endemism from one vent site to the next. Also interestingly, clustering of communities based on the relative abundance of different T-RFLP peaks indicated that samples from similar mixing regimes clustered together, providing evidence of partitioning across fluid gradients. Castle and CASM, the plume samples with the lowest temperature anomaly, tended to cluster together and also had the lowest species richness of the samples taken. The high-temperature Hulk sample, in contrast, did not necessarily exhibit higher species richness, yet clustered separately in cluster dendrograms for both the bacterial and archaeal domains. This was likely due to a compositional shift in community membership, from dominance of mesophiles such as Gammaproteobacteria and Marine Groups I and II in cooler plume and diffuse flow samples, to dominance of thermophiles in the Epsilonproteobacteria, Thermococcales, and Methanococcales groups in the high-temperature sample.

Despite the abundance of many different species, however, two groups of sulfur-oxidizing Gammaproteobacteria, SUP05 and Arctic96BD-19, were particularly abundant across all three sample types. The SUP05 and Arctic96BD-19 groups are related to sulfur-oxidizing gill symbionts of deep-sea clams and mussels (Cavanaugh, 1983; Newton et al., 2007). The SUP05 lineage, initially identified in a hydrothermal vent plume originating from the Suiyo Seamount (Sunamura et al., 2004), encompasses the clam and mussel symbionts, while Arctic96BD-19 forms a closely related sister clade to SUP05. Within marine oxygen minimum zones, SUP05 and Arctic96BD-19 exhibit overlapping but not identical distribution patterns consistent with redox-driven niche partitioning. Indeed, SUP05 appears to thrive in regions of sulfide and nitrate depletion, deriving energy from the oxidation of reduced sulfur compounds and using nitrate as terminal electron acceptor (Lavik et al., 2008; Walsh et al., 2009). In contrast, Arctic96BD-19 appears to thrive under more oxic water column conditions, deriving energy from reduced sulfur compounds and using oxygen as a terminal electron acceptor (Swan et al., 2011; Walsh & Hallam, 2011). Both SUP05 and Arctic96BD-19 have the potential to harness the energy obtained from sulfur-oxidation to fix inorganic carbon via 1,5-bisphosphate carboxylase/oxygenase (RubisCO) (Swan et al., 2011; Walsh & Hallam, 2011), implicating them as primary producers in the dark ocean. The extent to which they contribute to food web structures in hydrothermal vent ecosystems remains to be determined.

Phylogenetic placement of SSU rRNA gene sequences recovered from Hulk plume and diffuse flow fluids resolved into the previously recognized SUP05 and Arctic96BD-19 groups and partitioned roughly between plume and background seawater. The SUP05 group contained the majority of hydrothermal plume-derived sequences and some background seawater sequences that grouped most closely with sequences recovered from Saanich Inlet, a seasonally anoxic basin on the coast of Vancouver Island, British Columbia, and with sequences recovered from the Suiyo Seamount plume (Fig. 5). In the case of Arctic96BD-19, the majority of hydrothermal plume- and background-derived sequences grouped most closely with sequences recovered from the Line-P transect in the northeastern subarctic Pacific water column (Walsh et al., 2009) and clones recovered from the San Pedro Channel (Brown et al., 2005). The partitioning of Juan de Fuca sequences with sequences recovered from Saanich Inlet, San Pedro, and the northeastern subarctic Pacific is consistent with gene flow between the vent ecosystem and northeastern Pacific waters as a whole.

Although no SUP05 or Arctic96BD-19 SSU rRNA gene sequences were recovered in the clone libraries from the high-temperature Hulk diffuse flow fluid sample, both groups were indicated in other diffuse flow samples using T-RFLP and qPCR. Given the limited environmental parameter data and taxonomic resolution of the current study, we can only begin to speculate on the forces driving ecotype selection along geochemical gradients in plume and diffuse flow fluids. While sequences matching the SUP05 group have been recovered in samples from several hydrothermal systems (i.e., Sunamura et al., 2004; Dick & Tebo, 2010; German et al., 2010; Bourbonnais et al., 2012), the relatively high abundance of Arctic96BD-19 in diffuse flow fluids (with a relatively high input of hydrothermal fluid) and hydrothermal plumes in the Juan de Fuca system had not been observed in previous studies. In some plume samples, Arctic96BD-19 dominated the bacterial community, reaching up to 64% of total bacterial SSU rRNA gene copies in the CASM plume. As previously identified members of this group appear to thrive under more oxic water column conditions than SUP05, the Arctic96BD-19 group may take advantage of elevated concentrations of reduced sulfur compounds present in attenuating plume fluids. Arctic96BD-19 was also prevalent in background seawater, where sulfide levels would be undetectable, reaching up to 22% of total bacterial SSU rRNA gene copies. It remains possible that even in such background waters, geochemical traces of the plume continue to fuel microzones of chemolithoautotrophic growth. Similar observations have been made for particles in the dark ocean with the potential to support anaerobic processes such as sulfate reduction and methanogenesis that in turn fuel chemolithoautotrophic growth in the surrounding water column (Karl et al., 1984; Alldredge & Cohen, 1987; Shanks & Reeder, 1993; Karl & Tilbrook, 1994; Woebken et al., 2007). In diffuse flow samples where the contribution of reduced hydrothermal fluid is greater, Arctic96BD-19 reached up to 25% of total bacterial SSU rRNA gene copies despite elevated temperature and lower oxygen conditions. Indeed, in most of these samples including the high-temperature Hulk sample, Arctic96BD-19 was more prevalent than SUP05, suggesting a potentially more versatile energy metabolism and temperature tolerance than previously recognized. Given the emerging role for Arctic96BD-19 in carbon and sulfur cycling in the dark ocean, this versatility warrants more in-depth exploration of ecotypes using cultivation and single-cell genomic approaches.

The extent to which SUP05 and Arctic96BD-19 are biogeochemically active members of hydrothermal plumes, diffuse flow fluids, and background seawater cannot be determined from the present study given our inability to distinguish between active and dormant cells. Thus, while we can place these groups in specific mixing regimes and comment on their potential ecological and biogeochemical roles, we are unable to link these groups with specific processes in the environment. A recent study by Bourbonnais et al. (2012) working at some of the same vent sites surveyed in this study implicated SUP05 in nitrogen loss processes in diffuse flow fluids. Curiously, SSU rRNA gene sequences affiliated with Arctic96BD-19 were not detected in a clone library recovered from 24.8 °C diffuse flow fluids at the Hulk site 1 year earlier. These observations highlight the challenges associated with dynamic hydrothermal vent ecosystems and point to the need for more statistical sampling approaches to more accurately identify ecological patterns under changing geochemical conditions.

Geochemical models of hydrothermal vent plumes have suggested that the oxidation of elemental sulfur and metal sulfides represents one of the largest potential sources of metabolic energy in these fluids (McCollom, 2000). However, similar models suggest that methanogenesis and reduction of sulfate or elemental sulfur are favored thermodynamically at temperatures above 38 °C (McCollom & Shock, 1997). While these models are based on sulfur and sulfide concentrations from hydrothermal vent fluids on the East Pacific Rise, the relative proportions of compounds at the Juan de Fuca Ridge are similar (Butterfield et al., 1997). Moreover, these geochemical models indicate that the activities of sulfur-oxidizer bacteria are likely to diminish in the later stages of the plume, as elemental sulfur and sulfide are depleted (McCollom, 2000). Our results are broadly consistent with temperature effects predicted in these models, with Methanococcales and Thermococcales dominating in the high-temperature samples. However, the prevalence of presumptive SUP05 and Arctic96BD-19 in all three mixing regimes deviates from the expectation that attenuated plumes are less hospitable to sulfur-oxidizing bacteria and points to the presence of cryptic elemental cycling in these fluids (Canfield et al., 2010). Moreover, the identification of Arctic96BD-19 in high-temperature diffuse flow fluids adds a new perspective to the microbial ecology and biogeochemistry of hydrothermal vent ecosystems. Under more stratified water column conditions associated with increasing sulfide concentrations, both SUP05 and Arctic96BD-19 are replaced by sulfur oxidizers affiliated with the Epsilonproteobacteria (Labrenz et al., 2007; Grote et al., 2008, 2012; Lin et al., 2008). It will be of interest to determine the effect of sulfide concentration on SUP05 and Arctic96BD-19 in relation to physiological succession and taxon replacement under the more dynamic geochemical conditions found in plume and diffuse flow fluids.

In conclusion, our results point to common and unique microbial community structures associated with geochemical gradients within three hydrothermal vent mixing regimes within the Juan de Fuca system. While compositional changes reflected known temperature constraints on microbial community structure, the prevalence of SUP05 and Arctic96BD-19 in most samples posits a conserved role for these groups in carbon, sulfur, and nitrogen cycling at different ecological scales throughout the dark ocean. It is possible that in marine ecosystems, SUP05 and Arctic96BD-19 are important constituents of the ‘rare biosphere’ (Sogin et al., 2006) that exist in low abundances under most environmental conditions, but ‘bloom’ in response to specific geochemical conditions. Under these circumstances, different SUP05 and Arctic96BD-19 ecotypes associated with geochemical gradients as diverse as mussel gills, sinking particles, stratified water columns, and hydrothermal fluids have the potential to serve as ecological indicators for a changing global ocean.


The authors wish to thank Chief Scientist Jim Holden as well as Sanjoy Som and Alden Denny for assistance with sample collection at sea. Dave Butterfield assisted with HFPS sample collection, provided chemistry data, and assisted with data interpretation. Billy Brazelton assisted with writing scripts for T-RFLP analyses, and Melissa Baird assisted with MegaBACE operation. Annie Bourbonnais assisted with data interpretation, and Alyse Hawley assisted with data visualization. RA was funded by an NSF Graduate Research Fellowship through NSF grant number DGE-0718124, an NSF IGERT grant, and the ARCS Foundation. MTB was funded by a Científicos y Tecnólogos Consejo Nacional de Ciencia y Tecnología (CONACyT) scholarship, Genome Canada and the Natural Sciences and Engineering Research Council (NSERC). Additional funding was provided through a NASA Astrobiology Institute grant through Cooperative Agreement NNA04CC09A to the Geophysical Laboratory at the Carnegie Institution for Science, NSERC, and the Canadian Institute for Advanced Research (CIFAR).