|Home | About | Journals | Submit | Contact Us | Français|
The abundance, diversity, activity, and composition of microbial communities in sulfide structures both of active and inactive vents were investigated by culture-independent methods. These sulfide structures were collected at four hydrothermal fields, both on- and off-axis of the back-arc spreading center of the Southern Mariana Trough. The microbial abundance and activity in the samples were determined by analyzing total organic content, enzymatic activity, and copy number of the 16S rRNA gene. To assess the diversity and composition of the microbial communities, 16S rRNA gene clone libraries including bacterial and archaeal phylotypes were constructed from the sulfide structures. Despite the differences in the geological settings among the sampling points, phylotypes related to the Epsilonproteobacteria and cultured hyperthermophilic archaea were abundant in the libraries from the samples of active vents. In contrast, the relative abundance of these phylotypes was extremely low in the libraries from the samples of inactive vents. These results suggest that the composition of microbial communities within sulfide structures dramatically changes depending on the degree of hydrothermal activity, which was supported by statistical analyses. Comparative analyses suggest that the abundance, activity and diversity of microbial communities within sulfide structures of inactive vents are likely to be comparable to or higher than those in active vent structures, even though the microbial community composition is different between these two types of vents. The microbial community compositions in the sulfide structures of inactive vents were similar to those in seafloor basaltic rocks rather than those in marine sediments or the sulfide structures of active vents, suggesting that the microbial community compositions on the seafloor may be constrained by the available energy sources. Our findings provide helpful information for understanding the biogeography, biodiversity and microbial ecosystems in marine environments.
Hydrothermal sulfide deposits, termed chimneys or mounds, are found on the deep seafloor. These deposits mainly consist of mineral sulfides, such as chalcopyrite (CuFeS2), pyrite (FeS2), and sphalerite [(Zn, Fe)S]. In sulfide structures of active vents, chemical disequilibria and steep physical gradients are created by mixing of reduced hot fluids with oxygenated cold seawater (78, 79). Metabolically and phylogenetically diverse chemolithoautotrophs thrive within the sulfide structures of active vents using chemical energy provided from redox reactions between reductants (e.g., H2, H2S, CH4, and Fe2+) in the vent fluids and oxidants (e.g., O2, NO3−, and SO42−) in the surrounding seawater (53). The diversity and abundance of microbial communities in active vent structures have been reported (36, 52, 54, 56, 63, 67, 68, 70, 72). However, little is known about the microbial communities in sulfide structures of inactive vents with the exception of one report (65): whether and how the abundance, diversity, and composition of microbial communities within the sulfide structures change after hydrothermal activity ceases in deep-sea vent fields is currently unclear.
Recent accumulation of 16S rRNA gene sequences from various habitats and the development of useful bioinformatics tools has helped us to get a big picture of prokaryotic biogeography and to address a fundamental question: what are the factors that control the diversity and distribution pattern of microbial communities in natural environments (49, 58). The geography and geology of active hydrothermal fields are different among mid-ocean ridges (e.g., the Mid-Atlantic Ridge, East Pacific Rise, Juan de Fuca Ridge, and Central Indian Ridge), island arcs (e.g., the Mariana Arc and Izu-Bonin Arc), and back-arc basins (e.g., the Manus Basin and Lau Basin). For a better understanding of the biogeography and biodiversity of microbial communities in hydrothermal environments, it is important to investigate the microbial communities in various deep-sea hydrothermal systems, in particular, back-arc hydrothermal systems because of the unique geochemical characteristic of the hydrothermal fluids, such as the abundance of volatile components and the very low pH (19). However, there are few studies of microbial communities in back-arc hydrothermal fields, i.e., the Manus and Lau Basins (70, 72), except the Okinawa Trough where the unique sediment-hosted system is greatly influenced by terrestrial organic inputs (54).
The Southern Mariana Trough (SMT) is an active spreading back-arc basin (17, 27). In this area, active hydrothermal fields have been found on the spreading ridge and on the off-ridge seamount (6, 26, 83). Microbiological analysis of warm crustal fluids and microbial mats of the SMT have been carried out, suggesting the presence of novel and diverse microorganisms in these habitats (33, 34). Geochemical characterizations of the sulfide structures of the SMT have been completed (26, 31); however, the microbial communities within the structures are still unknown. The goal of the present study is to characterize the microbial communities within the sulfide structures of active and inactive vents in four hydrothermal fields of the SMT. In these fields, the geological setting and the degree of hydrothermal activity are different. The comparative analysis among the active and inactive vents and other seafloor environments provides insights into the biogeography and biodiversity of the microbial communities not only in deep-sea hydrothermal fields but also on the seafloor.
Hydrothermal vents were found during cruises on- and off-axis of the back-arc spreading ridge in the SMT (26, 80, 83). The Fryer site (also called the Snail site, 12°57.22′N, 143°37.16′E; depth, 2,850 m) and Y site (12°56.60′N, 143°36.80′E; depth, 2,826 m) are located on the ridge-axis, the Pika site (12°55.15′N, 143°36.96′E; depth, 2,773 m) is on the off-axis seamount, and the Archaean site (12°56.35′N, 143°38.0′E; depth, 2,990 m) is located between the ridge-axis and the Pika seamount (see Fig. S1 in the supplemental material). The host rock is tholeiitic andesite at both on- and off-ridge areas (31).
At the Fryer site, black smokers were absent and only white to gray smokers were found during the sampling cruises. The maximum temperature of fluids measured at the Fryer site decreased from 270°C in 2003 to 116°C in 2005. At the Y site, many sulfide chimneys of inactive vents were found. Only low temperature shimmering was observed at this site. At the Pika site, there were black smokers discharging high temperature fluids (more than 300°C). In addition, diffusive fluids were discharging from the sulfide mounds adjacent to the black smoker chimneys. Many sulfide chimneys of inactive vents were observed at the eastern side of the forest of black smoker chimneys. At the Archaean site, we found active black smokers venting hot fluids (maximum, 341°C). The geochemical compositions of hydrothermal fluids at the Fryer, Pika, and Archaean sites were previously reported (26, 28). High concentrations of H2S (~8 mM) were detected in the high temperature hydrothermal fluids from the Archaean and Pika sites. In contrast, moderate H2S concentrations (~0.3 mM) were detected in the fluids from the Fryer site, where the hydrothermal fluids might mix with seawater below the seafloor. Samples of the sulfide structures were collected from four sites (the Fryer, Y, Pika and Archaean sites) during the YK03-09 cruise (October to November 2003) of the R/V Yokosuka (JAMSTAC [Japan Agency for Marine-Earth Science and Technology]) with the manned research submersible Shinkai 6500 (JAMSTEC), the TN167A cruise (May 2003) of the R/V Thomas G. Thompson (University of Washington) with the remotely operated vehicle ROPOS (Remotely Operated Platform for Ocean Science, Canadian Scientific Submersible Facility, Canada), and the YK05-09 cruise (July to August 2005) of the R/V Yokosuka with the Shinkai 6500. Geochemical and mineralogical characteristics of sulfide structures of the four sites have been determined by previous investigators (26, 31).
The samples contained sphalerite, pyrite, and chalcopyrite (26, 31). Iron was the most abundant metal element (4.45 to 26.9 weight percent [wt%]), and copper and zinc were also detected (1.92 to 9.08 wt% and 0.03 to 3.57 wt%, respectively) in the sulfide structures from the Fryer, Pika, and Y sites (31). The samples used in the present study are listed in Table Table1.1. A part of the sulfide mound of an active vent (maximum fluid temperature was 107°C) was collected at the Fryer site (sample AFhm). The thickness of the AFhm sample was ~5 cm. A part of a black smoker chimney (sample AAbs; maximum, 341°C) and a part (sample AAcs) of a chimney discharging clear fluids (maximum, 117°C), which was located a few meter below the AAbs, were collected at the Archaean site. A part of a black smoker chimney (sample APbsc) venting hot fluids (270°C) and a part of a clear smoker chimney (sample APcsc) discharging low-temperature fluids (19°C) were collected at the Pika site. Parts of some sulfide chimneys from inactive vents were also collected at the Pika site (samples IPltc and IPdc). The sulfide mineralogy of IPltc and IPdc was essentially identical to that of APbsc, but IPltc and IPdc were more silicified by later low-temperature hydrothermal processes (31). A part of a sulfide chimney of an inactive vent, approximately 50 cm in diameter, was collected at the Y site (sample IYdc). Parts of the same chimney of the IYdc sample were collected during both the YK03-09 and TN167A cruises. The former IYdc sample from the inactive vent chimney from the YK03-09 cruise was used for organic and enzymatic analyses, and the latter from the TN167A cruise was used for DNA analysis. The samples were carefully stored in a closable septum bio-box on the remotely operated vehicle and transferred to the mother ship. The samples were quickly transferred into a sealable plastic bag with a pack of deoxygenation reagent (Ageless; Mitsubishi Gas Chemicals Co.). Parts of the collected samples were stored at −80°C for DNA analysis and at −20°C for the other analyses. The first letter of the sample names (A or I) indicates that the samples were collected from active or inactive vents, respectively. The second letter of the sample name (F, A, P, or Y) indicates the sampling site, i.e., Fryer, Archaean, Pika, or Y site, respectively. The last two or three letters indicate the type of sample: hm, hydrothermal mound; bs and bsc, black smoker chimney; cs and csc, clear smoker chimney; and dc and ltc, inactive dead and low temperature chimney.
Each sulfide structure sample was separated into exterior, middle and interior parts using a clean chisel and hammer if the size of the sample was large enough. Parts of the sample were freeze-dried and then gently pulverized with a clean mortar in the on-shore laboratory. Aliquots of the pulverized sample were used for total organic carbon (TOC) analysis, bulk carbon isotope analysis, and enzymatic activity measurements. A part of the separated samples was placed in a glass vessel with 6 M HCl at 140°C for 1 h. Then, the residue was rinsed with distilled deionized water until the pH was neutral. After drying, approximately 3 mg of each sample was loaded into clean Sn capsules for combustion and analysis using an Elemental Analyzer-Isotope Ratio Mass Spectrometer (Costech 4010 Elemental Analyzer-Finnigan MAT DELTA Plus Mass Spectrometers [EA-IRMS]). The samples were introduced into a furnace with purified oxygen gas at 1,000°C. Combustion products were entrained in an ultrapure-grade He carrier gas, and the concentrations of carbon and nitrogen were measured with the elemental analyzer. The combusted gas samples were directly introduced into the mass spectrometer through the interface for carbon isotope measurements. The abundances of stable isotopes are expressed using the δ notation within ±0.2‰ in this system. These data indicate the difference (in ‰) between the relevant ratio in the sample and the same ratio in the Pee Dee belemnite standard (PDB), as follows: ‰ = (Rsample/Rstandard − 1) × 1,000, where R = 13C/12C.
An improved method for determination of phosphatase enzymatic activity (66) was previously described (74, 75). Briefly, acid phosphatase activities in the rock samples were measured spectrometrically by using 25 mM p-nitrophenyl phosphate as a substrate dissolved in modified universal buffer (MUB; pH 6.5) as follows. The pulverized sample was incubated with the substrate solution in a water bath at 37°C for an hour, and then the reaction was stopped by adding 0.2 ml of 0.5 M CaCl2 and 0.8 ml of 0.5 M NaCl. The solution was then separated from the solid sample by centrifugation, and the supernatant was filtered through a membrane filter (pore size, 0.2 μm). The rate of p-nitrophenol production (unit, nmol/min/g-rock) was calculated based on the absorbance change of the solution at 410 nm.
DNA was extracted from each subsample by using a Fast DNA kit for soil (Qbiogene, Inc., Irvine, CA) and a bead-beater FastPrep instrument (Qbiogene). Approximately 0.5 g of the sample was used for DNA extraction. Finally, DNA was eluted with 100 μl of ultra pure water and stored at −20°C. The extracted DNA was used for 16S rRNA gene clone analysis and quantitative PCR analysis.
The copy number of the prokaryotic 16S rRNA gene in the extracted DNA was determined by quantitative PCR as previously described (69). In brief, we performed quantitative PCR using TaqMan Universal PCR Master Mix (Applied Biosystems, California) and the primers Uni340F and Uni806R and the TaqMan probe Uni516F, which target most prokaryotes, on an ABI Prism 7700 sequence detection system (Applied Biosystems). The purified PCR product from the 16S rRNA gene of Escherichia coli was used as the standard DNA. A dilution series from 100 to 1010 copies of the amplicon was used to generate the standard curve. A no template control (NTC) was also performed in all assays. The data at and below the background level of the NTC were excluded from the standard curve. All assays were performed in triplicate.
The 16S rRNA gene analysis was performed as previously described (33). Briefly, partial 16S rRNA genes were amplified by PCR using the following oligonucleotide primers: Arc8F as an Archaea-specific primer and Uni515F and Uni1406R as universal primers. The PCR products were cloned with a TOPO TA cloning kit (Invitrogen, California). The nucleotide sequences of randomly selected clones were determined with a BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems) using M13 forward and reverse primers (Invitrogen) on an ABI Prism 3130xl genetic analyzer (Applied Biosystems). In addition, the internal primers 519r, 530f, 907r, and 926f (37) were used for archaeal 16S rRNA gene sequencing.
We checked all sequences to remove chimeric sequences with (i) BLAST analysis of sequence segments (2), (ii) the Chimera check program at the Ribosomal Database Project (46), and (iii) the Mallard program (4). The nucleotide sequences of the clones without chimeric sequences were aligned by using CLUSTAL W version 1.83 (77). Clones having 97% sequence similarity or higher were treated as the same phylotype using DOTUR (61). The sequences of the phylotypes were realigned with other published sequences including the closest one determined by BLASTN (2). Maximum-likelihood trees were constructed with the general-time-reversible model of nucleotide substitution incorporating invariable sites and discrete gamma distribution (GTR+I+G) using PHYML (21). Bootstrap values were estimated using 100 replicates.
Rarefaction analysis, Shannon diversity index, and Chao1 species richness estimators were performed using DOTUR based on the distance matrices that were generated from the alignment datasets of the clones from each library using ARB (45). Chao1 species richness estimates of shared phylotypes were calculated using SONS (62). To assess the difference among the microbial communities, an online tool, UniFrac (43), was used for the phylogenetic (P) test (48), the UniFrac significance test (44), the principal coordinates analysis (PCoA), and Jackknife environment cluster analysis (44). These tests and analyses of UniFrac were based on the phylogenetic tree that was reconstructed from the clones in the libraries using the neighbor-joining method by ARB. Jackknife values were estimated using 100 permutations and shown in the nodes of the unweighted pair group method using average linkages (UPGMA) tree. To compare the microbial communities in other habitats of the SMT, the clones detected in the microbial mat, hydrothermal fluid, and ambient seawater samples of the SMT previously reported (33, 34) were also included in the UniFrac analysis. Furthermore, for the comparative analysis of the microbial communities in the SMT and other areas, the 16S rRNA gene sequences, the range of which covers our clones (from positions 515 to 1406 in the E. coli numbering), were collected from the NCBI GenBank, compiled, and analyzed by using UniFrac.
These sequence data have been submitted to the DDBJ database under accession numbers AB424684 to AB425028 and AB292856 to AB293246.
In total, eight samples of the sulfide structures from the active and inactive vents were collected in four hydrothermal fields of the SMT. The sampling sites are shown in Table Table11 and see Fig. S1 in the supplemental material (see also Materials and methods for details). Using these samples, we examined total nitrogen (TN) and total organic carbon (TOC) concentrations, enzymatic activity and 16S rRNA gene copy numbers. We constructed 16S rRNA gene clone libraries from these samples using PCR with prokaryote-universal and archaea-specific primers and performed several statistical analyses.
The TN concentrations correlated with the TOC concentrations (regression coefficient, r2 = 0.936), indicating that these values detected in the samples reflect organic compounds derived from microbial cells (Fig. (Fig.1A).1A). The concentrations of TN and TOC were higher in the exterior part than in the middle and interior parts of these inactive vent samples (IPltc, IPdc, and IYdc). The TN and TOC concentrations of the samples (AFhm and APbsc) of active vents were relatively lower than those of the inactive samples.
In the case of the samples of inactive vents, the acid phosphatase enzymatic activity showed the same tendency as the TN and TOC concentrations (Fig. (Fig.1).1). The enzymatic activity was higher in the exterior part of these inactive vent samples than in the middle and interior parts. Acid phosphatase activity shows a direct correlation with biomass and can be used as a biomarker of microbial activity (75). The exterior part of the IPdc sample of the inactive vent yielded the highest enzymatic activity. In contrast, a positive correlation between the acid phosphatase activity and organic content was not clear in the APbsc sample of the active vent. The results may reflect the temperature used for enzymatic activity measurement and the sampling operation: enzymes from thermophiles could not be detected at 37°C.
The maximum copy number of 16S rRNA genes in each sample ranged from 4.1 × 107 to 1.5 × 1010 copies/g (wet weight) for the active vent samples (except the AAbs sample) and from 3.3 × 108 to 3.7 × 109 copies/g (wet weight) for the inactive vent samples (Table (Table2).2). For the AAcs and APcsc samples from the active vents and the IPltc sample of the inactive vent, the 16S rRNA gene copy numbers were 100 times higher in the exterior parts than in the interior parts. In the IYdc sample of another inactive vent, the 16S rRNA gene copy number was similar for the exterior and interior parts.
The δ13C (‰, versus PDB) values ranged from −18.7 to −25.7‰, yielding a difference of 7‰. The δ13C values differed slightly among the different portions of the samples (Fig. (Fig.1B):1B): −19.4 to −22.8‰ (exterior), −18.7 to −23.2‰ (middle), and −20.3 to −25.7‰ (interior). In general, 12C is used preferentially in chemolithoautotrophic carbon fixation, which results in lower δ13C values. In the present study, the tendency of the portion with low δ13C values was different among the samples from inactive vents. Low δ13C values were noted in the interior part of the IPltc and IYdc samples of the inactive vents. In contrast, a low δ13C value was noted in the exterior part of the IPdc sample from the other inactive vent. The δ13C values were similar between the exterior and interior parts of the APbsc sample of the active vent.
To assess the diversity of the microbial communities, Shannon diversity index and Chao1 species richness estimates were examined for the universal clone libraries (Table (Table2).2). The Chao1 value indicates the estimated number of unique phylotypes in the microbial community. The combined clone libraries from the subsamples of each sample were used for rarefaction curve analysis, in addition to the above analyses; for example, clone libraries from the AFhm1 and AFhm2 subsamples were merged into a clone library of the AFhm sample. Rarefaction curves for three samples of the active vents (AAcs, APbsc, and APcsc) were congruent (see Fig. S2 in the supplemental material), corresponding to the similar values of the Shannon diversity index (Table (Table2).2). The AFhm sample of the active vent had the lowest diversity. Although the influence of methodological bias on estimation of microbial community diversity must be considered (25), the observed diversity in the samples was not influenced by the total clone number. For example, although there were more AFhm2 clones than APbsc4 clones, the diversity of the former was less than the latter (Table (Table2).2). The phylotypes in the two subsamples (AFhm1 and AFhm2) were mostly shared (the Chao1 richness estimator of shared phylotypes is 20), which is one of the reasons for the lower diversity of the combined AFhm compared to the combined libraries of the other samples, i.e., AAcs, APbsc, APcsc, IPltc, and IYdc (Table (Table22 and see Fig. S2 in the supplemental material). The diversity of the microbial communities in the IPltc sample of the inactive vent was highest among all samples, supported by both the rarefaction curve and Shannon diversity index. Furthermore, these analyses indicated the diversity of the microbial community in the integrated library from the samples of the inactive vents was similar to that from the samples from the active vents (Table (Table22 and see Fig. S2 in the supplemental material). The Shannon diversity index values and the Chao1 richness estimates for the integrated clone library from the active vent samples were 4.95 and 532 (95% confidence interval [CI] = 443 to 669), and those from the inactive vent samples were 4.75 and 457 (95% CI = 355 to 624), respectively.
For the archaeal clone libraries, we could not perform statistical analyses because only partial sequences (i.e., either 5′ or 3′ termini of the 16S rRNA genes) for most clones could be determined. However, the distribution pattern of the archaeal clones (see Fig. S3 in the supplemental material) clearly indicates that the archaeal communities in the active vent samples had higher diversity than those in the inactive vent samples.
The taxonomic affiliation of the recovered phylotypes in the universal clone libraries from the subsamples is summarized in Fig. Fig.2.2. Most phylotypes in the universal clone libraries were affiliated with the domain Bacteria. Phylogenetic trees for these clones are shown in Fig. S4 to S6 in the supplemental material. We were unable to obtain PCR products from the AAbs sample using either the universal primer set or the archaeal primer set or from the AAcs sample using the archaeal primer set (Table (Table22).
The bacterial phylotypes recovered were affiliated with the following phyla: the Aquificae, Acidobacteria, Actinobacteria, Bacteroidetes, Chlamydiae, Chlorobi, Chloroflexi, Deferribacteres, Deinococcus-Thermus, Firmicutes, Fusobacteria, Nitrospirae, Planctomycetes, Proteobacteria, Spirochaetes, Thermodesulfobacteria, and Verrucomicrobia and the candidate divisions OP1, OP11, OP3, and TM6. Some phylotypes were not clearly classified in the known bacterial phyla (see Fig. S6F in the supplemental material). The phylotypes related to the Epsilonproteobacteria or to thermophilic phyla (i.e., Aquificae, Deinococcus-Thermus, and Thermodesulfobacteria) were abundant in clone libraries from the active vent samples but not from the inactive vent samples (Fig. (Fig.2).2). Members of the Aquificae include thermophilic hydrogen oxidizers, and those of the Epsilonproteobacteria include thermophilic and mesophilic hydrogen and/or sulfide oxidizers, which have been detected in sulfide structures of active vents in other deep-sea hydrothermal fields (1, 13, 41, 54-56, 67, 72, 82).
The archaeal phylotypes recovered were classified in the following clusters: the marine hydrothermal vent group 1 (MHVG-1, also called SHVA1 or SHVAG) (73), Korarchaeota (7, 8), marine benthic group E (MBGE) (81), deep-sea hydrothermal vent euryarchaeota 2 and 3 (DHVE2 and DHVE3) (68), terrestrial hot spring crenarchaeota (THSC) (8, 68), marine group I (MGI) (14, 18), the classes Thermoprotei, Archaeoglobi, and Thermococci. The phylotypes related to cultured hyperthermophilic archaea, the Thermoprotei, Archaeoglobi, and Thermococci, were abundant in the clone libraries from the active vent samples (see Fig. S3 in the supplemental material). Unclassified archaeal phylotypes were also recovered (see Fig. S7 in the supplemental material). The detection of these phylotypes was consistent with the measured high temperature of the venting fluids. Such hyperthermophilic archaeal phylotypes were also detected in the APcsc sample associated with low temperature fluids (19°C). The presence of phylotypes related to hyperthermophilic archaea in low temperature environments (<12°C) has also been reported by other researchers (16, 35).
In contrast to active vent samples, few or no phylotypes related to the thermophilic phyla and Epsilonproteobacteria and to cultured hyperthermophilic archaea were recovered from the inactive vent samples (Fig. (Fig.2).2). Instead, bacterial phylotypes related to Bacteroidetes, Actinobacteria, or Alphaproteobacteria and Gammaproteobacteria and the archaeal phylotypes related to MGI and DHVE3 were relatively abundant in the clone libraries in the samples of inactive vents. In the previous report (65), the phylotypes related to the marine benthic group E and to Magnetobacterium bavaricum of the Nitrospirae were predominant in the clone libraries constructed from samples of inactive vents; however, such phylotypes were not abundant in our libraries.
To evaluate the difference in the microbial communities among the sulfide structures, and between these structures and other environments (microbial mats, vent fluid, sub-seafloor crustal fluids, and ambient seawater) of the SMT, statistical analyses, i.e., the P and UniFrac significance tests, PCoA and Jackknife environment cluster analysis, were performed for the universal clone libraries using UniFrac (43). These tests and analyses with UniFrac are based on the distance between communities as the fraction of branch length in a phylogenetic tree that was reconstructed from the clones in the libraries. The PCoA shows the relative difference among the communities in a two-dimensional display. Jackknife environment cluster analysis is an unweighted pair group method using average linkages (UPGMA) clustering. The UPGMA tree shows the similarities among the communities with branch lengths. The communities in the subsamples of each sample clustered with each other (Fig. (Fig.3),3), except in the case of the APbsc sample from the active vent, as supported by P and/or UniFrac significance tests (see Table S1 in the supplemental material). The communities in the inactive vent samples (IPltc and IYdc) clustered with each other, while the communities in the active vent samples (AFhm, APbsc, APcsc, and AAcs) scattered in a wider field in the two-dimensional principal coordinate (Fig. (Fig.3B).3B). The communities in the samples from inactive vents were significantly different from those in the samples of active vents, as supported by all statistical analyses (Fig. (Fig.33 and see Table S1 in the supplemental material). The communities in the sulfide structures did not cluster with those in other habitats (i.e., microbial mats, hydrothermal fluids, and ambient seawater) (Fig. (Fig.3),3), indicating that the observed communities from the sulfide structures represent the indigenous communities as supported by the P and UniFrac values (<0.001; see Table S1 in the supplemental material).
To compare the members among the communities, the Chao1 species richness estimates of shared phylotypes were calculated using SONS (Fig. (Fig.4).4). The Chao1 value indicates the estimated number of unique phylotypes in the microbial community. The shared phylotype number between the active and inactive vent communities was estimated to be 65 (Fig. (Fig.4A).4A). More specifically, the shared phylotype number between the AAcs and IPltc samples was estimated to be 54 (Fig. (Fig.4B);4B); the Shannon diversity index value and the Chao1 species richness estimator are shown in Table Table2.2. The shared phylotype number among the subsamples of the AAcs or IPltc sample and the Shannon diversity index value and the Chao1 species richness estimator of these subsamples are also shown in Fig. 4C and D and Table Table22.
To date, microbial communities in nascent and mature hydrothermal deposits of active vents have been well studied (1, 36, 52, 56, 57, 63, 68, 70, 72). However, microbial communities in sulfide structures of inactive vents are poorly understood (65). Furthermore, compared to mid-ocean ridges, there have been few studies of microbial communities within hydrothermal deposits in back-arc hydrothermal systems (70, 72). This is the first report of comparing the microbial communities in the active and inactive sulfide structures, and also other habitats (microbial mats, vent fluids, sub-seafloor crustal fluids, and ambient seawater), which were investigated using the same experimental procedure so that methodological bias could be minimized. Our comprehensive analysis, which includes data from various habitats in other seafloor areas, provides insight into biodiversity and biogeography of microbial communities not only in deep-sea hydrothermal fields but also on the seafloor.
The analyses of TOC and TN concentrations and enzymatic activity suggest that biomass and microbial activity in inactive vent structures is likely to be comparable to or higher than those in active vent structures. Furthermore, our results also suggest that, in sulfide structures of inactive vents, microbial activity is likely to be higher in the exterior parts that are exposed to surrounding deep seawater than in interior parts. The microbial activity may reflect the availability of organic carbon and O2 as carbon and energy sources within inactive vent structures. In fact, the TOC was higher in the exterior parts than the interior parts of the inactive vent samples.
The maximum cell abundance in sulfide structures may not be significantly different between active and inactive vents, which are likely to yield 108 to 109 cells/g (wet weight). Many prokaryotes have multiple copies of the 16S rRNA gene in their genomes; the average copy number per cell is 4.06 and 1.77, for Bacteria and Archaea, respectively (38). Given that the average copy number is roughly four copies/cell for the whole prokaryotic community, maximum cell numbers in each sample are estimated to range from 1.0 × 108 to 3.8 × 109 cells/g for the active vent samples and from 8.2 × 107 to 9.2 × 108 cells/g for the inactive vent samples. These values are consistent with the previous reports: ~109 cells/g (56) and 108 cells/g (65) for maximum cell numbers in the sulfide structures for active and inactive vents, respectively. Cell abundance is higher in exterior parts of the active vent structures than interior parts in other deep-sea hydrothermal fields, such as the Mid-Atlantic Ridge (22), the Juan de Fuca Ridge (63), and the Okinawa Trough (54). However, our results could not be simply compared to these previous data because the tendency of 16S rRNA gene copy numbers observed in the present study could reflect potential methodological bias resulting from the efficiency of DNA preparation from fragile (i.e., the exterior parts) or solid (i.e., the interior parts) samples for quantitative PCR. Nevertheless, the maximum cell abundances in the sulfide structures are likely to be similar between active and inactive vents.
In seafloor sulfide structures of inactive vents, unique chemolithoautotrophs using iron sulfides as energy sources play a potential role in the microbial ecosystem as primary producers. The difference in the observed δ13C values among the subsamples of each inactive vent is expected to represent chemolithoautotrophic activity in these habitats. Relatively low δ13C values were noted in the interior part of the IPltc and IYdc samples where biomass and microbial activity were low (Fig. (Fig.1).1). These results suggest that chemolithoautotrophic activity within the inactive vent structures may be relatively high in the interior parts. On the contrary, in another inactive vent sample (IPdc), a relatively low δ13C value was noted in the exterior parts (Fig. (Fig.1B),1B), suggesting that chemolithoautotrophic activity may be relatively high in the exterior part in accordance with the high biomass and microbial activity. Unique phylogenetic clusters consisting of uncultured phylotypes within the Alphaproteobacteria, Gammaproteobacteria, and Actinobacteria detected in the inactive sulfide structures were observed in the phylogenetic trees (the cluster A in Fig. S4A, the cluster B in Fig. S4B, and the cluster C in Fig. S6B in the supplemental material). These clusters contain environmental clones recovered from the seafloor basaltic rocks (50) but do not correspond to the ocean crust clades, which are a unique phylotype group for oceanic basalts defined by Mason et al. (50, 51). Iron-oxidizing chemolithoautotrophs belonging to the Alphaproteobacteria and Gammaproteobacteria have been reported in inactive sulfide structures at neutral pH using metal sulfides (such as pyrite) as electron donors (15). Members of the Actinobacteria include chemolithoautotrophic and heterotrophic iron (including pyrite) oxidizers (11, 12, 29, 30), although they are acidophiles (optimum growth, pH <3). Our phylotypes in cluster A within the Actinobacteria have 91% similarity or less to the closest cultured species, the iron-oxidizing bacterium Ferrimicrobium acidiphilum (29). These unique phylogenetic clusters potentially contain iron-oxidizing chemolithoautotrophs that are capable of using iron sulfides such as pyrite in the sulfide structures for growth.
Our comparative analysis suggests that the diversity of microbial communities in sulfide structures of inactive vents is likely to be comparable to or higher than in active vent structures. In the sulfide structures of active vents, thermal and geochemical gradients are generated by mixing reduced hot fluids with oxygenated cold seawater (78, 79), which could contribute to the increased microbial community diversity in these habitats. In fact, the comparative analysis for the subsamples of the AAcs sample showed that the communities were similar to each other (Fig. (Fig.22 and and3,3, and Table S1 in the supplemental material) and approximately half of the total number of phylotypes was shared (Fig. (Fig.4C);4C); however, all phylotypes were not shared. This could lead to the high diversity observed in the combined community of the AAcs sample (Table (Table22 and Fig. S2 in the supplemental material). In contrast, such physicochemical characteristics in habitats of inactive vent structures are not clear. The in situ temperature within the sulfide structures of inactive vents is expected to be close to that of ambient seawater (2 to 4°C). The concentration of organic compounds usable as energy and carbon sources decreased from the exterior to interior parts of the structures. The O2 concentration would decrease due to microbial activity in the same direction. Furthermore, the mineralogical characteristics were also different among the parts of the inactive sulfide structures; sphalerite is abundant in the exterior parts and chalcopyrite is distributed in the interior parts (31). The difference in these environmental conditions potentially contributes to the difference in the observed phylotypes among the locations of the inactive vent structures as shown in Fig. Fig.4D,4D, which could lead to the observed high diversity of the microbial communities within the inactive vent structures (Table (Table22 and Fig. S2 in the supplemental material), as in the case of the active vent structures.
Epsilonproteobacterial phylotype richness in the communities is likely to be a major factor for characterization of the biogeography of microbial communities in the SMT. We found that the divergence of communities along the first principal coordinate axis had a positive correlation with the fraction of the number of unique epsilonproteobacterial phylotypes in the total phylotype number of the libraries (Fig. (Fig.3B,3B, r2 = 0.79). The epsilonproteobacterial phylotype richness may be affected by the hydrothermal activity, i.e., variation of temperature and availability of reduced chemical species, e.g., H2S and H2 as energy sources for the growth of members of the Epsilonproteobacteria (10). However, the observed differences in the microbial communities would also involve other factors such as geology (on- and off-ridge), geographic distance, mineralogy, and habitat type (e.g., mats, fluids, and sulfide structures) because the first principal coordinate in the PCoA analysis described only 11.04% of the total variation among the communities. In fact, in the case of the communities in the active sulfide structures, the variation along the second principal coordinate (9.02% of the total variation) seems to be related to the differences in the sampling sites (i.e., Fryer site versus Pika site versus Archaean site) (Fig. (Fig.3B3B).
Although potential experimental bias associated with the methods used for DNA extraction, PCR and phylogenetic analyses must be considered when we assess the differences in the community compositions between the SMT studied by us and other areas reported by previous researchers, we are able to address the commonality in the community composition even though different methods were used. The results of our 16S rRNA gene phylotype analysis support the notion that bacterial members of the Epsilonproteobacteria and archaeal members of the Thermoprotei, Thermococci, and Archaeoglobi are likely to be common in microbial communities in active deep-sea hydrothermal fields of back-arc basins and mid-ocean ridges that are geographically and geologically different from each other (71); however, such phylotypes were a minor component in the clone libraries from inactive vent structures (65).
Regarding epsilonproteobacterial populations in marine environments, their biogeography is potentially constrained by water depth, as well as hydrothermal activity. The PCoA analysis using the compiled data of marine epsilonproteobacterial phylotypes (details in Table S2 in the supplemental material) showed that the epsilonproteobacterial populations in the shallow marine environments (0 to 200 m in depth) were lopsided along the first principal coordinate (describing 16.93% of the total variation) (Fig. (Fig.5).5). For the observed tendency in the comparative analysis of these communities (Fig. (Fig.5),5), methodological biases could be negligible because the experimental procedures (e.g., clone numbers) of these previous studies were randomized. The shallow marine environments included a hydrothermal vent field of the Taketomi Island, western Pacific Ocean (23). Although there are other shallow hydrothermal fields such as Vulcano in Italy (3), White Point in California (32), and Milos in Greece (64), no epsilonproteobacterial sequences are available for analysis. In contrast, epsilonproteobacterial populations in deep-sea environments (over 200 m in depth) were widely scattered in the two-dimensional display and the area overlaps the cluster of the shallow marine environments (Fig. (Fig.5).5). In the case of other prokaryotic populations, SAR11 bacteria and marine group I archaea also have higher diversity in deep sea regions compared to shallow regions (20), implying that biogeographic variation may be affected by water depth. Alternatively, the first principal coordinate may be related to the hydrothermal activity. The communities of the deep-sea hydrothermal fields plotted in the cluster of shallow environments (point D in Fig. Fig.5)5) may not be influenced by the hydrothermal activity. It should be noted that, in the present study, we observed no positive correlation between the epsilonproteobacterial biogeography and the difference in geological settings such as mid-ocean ridges and back-arc basins (Fig. (Fig.5),5), nor between the epsilonproteobacterial biogeography and the difference in the sample types, i.e., fluids, mats, sediments, whale corpses, chimneys, symbionts, and in situ growth chambers (data not shown).
The comparative analysis among seafloor environments, including sediments (9, 39, 40, 42, 59, 76), basaltic rocks (50), and inactive sulfide structures (65) (see Table S3 in the supplemental material for details), showed the microbial communities of the inactive sulfide structures were relatively similar to those in nonhydrothermal seafloor environments (Fig. (Fig.66 and Table S4 in the supplemental material). This result suggests that microbial communities in sulfide structures of inactive vents are likely to be similar to those in nonhydrothermal seafloor environments, such as sediments and rocks, and different from those in sulfide structures of active vents. This difference may have originated from the presence/absence of hydrothermal fluids as discussed above. Except for the hydrothermal sediments collected from the Rainbow site on the Mid-Atlantic Ridge (point K in Fig. Fig.6)6) (42), the community is similar to the communities in other nonhydrothermal environments, which is supported by P and UniFrac significance values (see Table S4 in the supplemental material). The habitat at the sampling point presumably has not been significantly influenced by the hydrothermal fluids as suggested by the low detection frequency of phylotypes related to the Epsilonproteobacteria (1 of 82 total clones) and thermophiles (none detected) (42). The distinction of the communities between hydrothermal and nonhydrothermal seafloor environments as shown by the PCoA analyses (Fig. (Fig.33 and and6)6) may be strongly influenced by in situ temperature and/or the availability of reduced chemical species derived from venting hydrothermal fluids, even if the geographic distance between these habitats is close and the habitat type (i.e., sediment or sulfide structure) of the communities is the same.
Interestingly, among the nonhydrothermal environments, the communities in the basaltic rocks and the inactive sulfide structures were similar to each other and could be distinguished from those in sediments (Fig. (Fig.6).6). One of the factors that may have led to this biogeographic pattern (along the second axis in the Fig. Fig.6)6) is the difference in energy sources to support the microbial ecosystems, i.e., ferrous iron for the communities of basaltic rocks (5) and inactive sulfide structures versus others such as methane provided from the deeper anoxic region (24, 47) and/or organic compounds derived from terrestrial and/or sea surface environments for the communities in sediments. Iron oxidizers presumably thrive in basaltic rocks as suggested by other previous reports (5, 60), as well as in inactive sulfide structures, as discussed above. To date, phylotypes recovered from such nonhydrothermal seafloor environments have low similarity (<95%) to cultured species (9, 39, 40, 42, 50, 59, 60, 65), which is consistent with our 16S rRNA gene analysis, indicating that there are novel mesophilic or psychrophilic species at the genus level of taxonomic affiliation in these environments. Determination of physiological characteristics of these uncultured phylotypes will help to unveil the factors that contribute to the biogeographic variation in each habitat type shown in the first axis of the PCoA (Fig. (Fig.66).
In the present study, we examined the abundance, diversity, activity and composition of microbial communities in sulfide structures collected from distinct hydrothermal fields (on- and off-axis) of the back-arc spreading center in the SMT. The comparative analysis between the sulfide structures of inactive and active vents supports the hypothesis that the microbial community composition within sulfide structures dramatically changes after hydrothermal activity ceases (65). Our results suggest that the abundance, activity, and diversity of the microbial communities within sulfide structures of inactive vents are comparable to or higher than those of active vents. Epsilonproteobacterial phylotype richness in the communities may be a major factor for characterization of the prokaryotic biogeography in the SMT, and the epsilonproteobacterial biogeography is potentially constrained by water depth in addition to hydrothermal activity. Furthermore, our comprehensive analysis of the communities in nonhydrothermal seafloor environments suggests that their biogeography may be selected by the energy sources associated with habitat types. Further studies, such as cultivation, in situ measurement of metabolic activity (e.g., carbon fixation, iron, and sulfide oxidation), and also geochemical and mineralogical characterization in fine scale, will provide helpful information for elucidating microbial ecosystems in seafloor massive sulfide structures and for understanding the biogeography of microbial communities on the quiet seafloor.
We thank the crews of the R/V Thomas G. Thompson and R/V Yokosuka and the operation teams of ROV ROPOS and Shinkai 6500 for their cooperation in sample collection. We are also grateful to the scientists that joined the YK03-09 cruise, TN167A cruise, and YK05-09 cruise and to the members of the Archaean Park Project for providing valuable samples and for helpful discussions. We thank three anonymous reviewers for their helpful comments.
This research was funded by the Ministry of Education, Culture, Science, and Technology (MEXT), Japan, through a special coordination fund (Archaean Park Project: International Research Project on Interaction between Sub-Vent Biosphere and Geo-Environments).
Published ahead of print on 12 March 2010.
†Supplemental material for this article may be found at http://aem.asm.org/.