Diversity and biogeochemical structuring of bacterial communities across the Porangahau ridge accretionary prism, New Zealand


  • Editor: Gary King

Correspondence: Leila J. Hamdan, Marine Biogeochemistry Section, Code 6114, US Naval Research Laboratory, Overlook Ave. SW, Washington, DC 22375, USA. Tel.: +1 202 767 3364; fax: +1 202 404 8515; e-mail: leila.hamdan@nrl.navy.mil


Sediments from the Porangahau ridge, located off the northeastern coast of New Zealand, were studied to describe bacterial community structure in conjunction with differing biogeochemical regimes across the ridge. Low diversity was observed in sediments from an eroded basin seaward of the ridge and the community was dominated by uncultured members of the Burkholderiales. Chloroflexi/GNS and Deltaproteobacteria were abundant in sediments from a methane seep located landward of the ridge. Gas-charged and organic-rich sediments further landward had the highest overall diversity. Surface sediments, with the exception of those from the basin, were dominated by Rhodobacterales sequences associated with organic matter deposition. Taxa related to the Desulfosarcina/Desulfococcus and the JS1 candidates were highly abundant at the sulfate–methane transition zone (SMTZ) at three sites. To determine how community structure was influenced by terrestrial, pelagic and in situ substrates, sequence data were statistically analyzed against geochemical data (e.g. sulfate, chloride, nitrogen, phosphorous, methane, bulk inorganic and organic carbon pools) using the Biota-Environmental matching procedure. Landward of the ridge, sulfate was among the most significant structuring factors. Seaward of the ridge, silica and ammonium were important structuring factors. Regardless of the transect location, methane was the principal structuring factor on SMTZ communities.


Coastal and deep marine sediments are among the most microbially diverse environments on Earth (Inagaki et al., 2006; Sogin et al., 2006; Biddle et al., 2008). Microorganisms are involved in all major oceanic biogeochemical cycles (Galand et al., 2010) and mediate the flux of carbon in aquatic systems (Azam, 1998). Culture-independent molecular studies have provided a wealth of information on the identity of prokaryotic communities in sediments, shed light on communities in methane-rich marine sediments and provided evidence for the involvement of bacteria and archaea in the anaerobic oxidation of methane (AOM) (Boetius et al., 2000; Lanoil et al., 2001; Orphan et al., 2001; Knittel et al., 2005; Inagaki et al., 2006; Hamdan et al., 2008; Pernthaler et al., 2008).

Redox conditions in methane-rich sediments tend to be well characterized (Borowski et al., 1996; Burns, 1998; Kastner et al., 1998; Hoehler et al., 2000; Dickens, 2001; Haese et al., 2003; Borowski, 2004; Joye et al., 2004; Milkov et al., 2004), but by comparison, far fewer studies directly relate whole community structure and composition to local abiotic conditions (Edlund et al., 2008; Galand et al., 2010). As a result, the environmental factors that structure microbial communities remain unclear in many environments (Nunoura et al., 2009). Microbial communities are structured by physical (e.g. depth, grain size, hydrodynamics), geochemical (e.g. redox conditions, organic matter), spatial (e.g. latitude, nutrient loading) and temporal factors (e.g. season). As a result, microbial community composition in methane-rich sediments may vary in different geographic regions (Inagaki et al., 2006). In the same way that methane formation is locally influenced by geological, geophysical and geochemical factors, the communities involved in methane production and oxidation are likewise regionally impacted. In contrast, microbial communities at specific redox interfaces including the sulfate–methane transition zone (SMTZ) tend to be similar across geographic distances (Inagaki et al., 2006; Hamdan et al., 2008; Harrison et al., 2009). While similarities occur under narrowly defined conditions such as the SMTZ, the composition may be variable depending on the local impact of terrestrial or pelagic organic matter deposition, natural and anthropogenic nutrient loading, and methane advection or diffusion. Thus, the question emerges as to how local heterogeneity in environmental conditions impacts microbial communities throughout the sediment column in methane-influenced locations. The goals for this study were to investigate bacterial community composition in methane-charged and nearby methane-poor sediments on the southern Hikurangi margin, and to correlate community composition to local abiotic factors.

In the last two decades, numerous cold seeps have been discovered on the Hikurangi margin (Henrys et al., 2003; Faure et al., 2006; Crutchley et al., 2010; Naudts et al., 2010; Schwalenberg et al., 2010). Methane is prevalent in this region due to the accelerated sediment accretion of up to 12 mm year−1 (Martin et al., 2010). Fluids are expelled from subducting sediments in focused locations (Jones et al., 2010; Klaucke et al., 2010; Naudts et al., 2010; Pecher et al., 2010). These fluids often contain methane (Barnes et al., 2010) and support diverse benthic communities (Baco et al., 2010; Sommer et al., 2010; Thurber et al., 2010). Evaluations of seeps along the Hikurangi margin have focused on their geological framework (Thomsen et al., 2001; Barnes et al., 2010; Pecher et al., 2010), while others have evaluated the foraminiferal (Barnette et al., 1993; Martin et al., 2010; Thurber et al., 2010) and invertebrate diversity of chemosynthetic communities (Baco et al., 2010). Before the present study, however, bacterial communities had not been described. Because of its geological heterogeneity and the availability of methane, pelagic and accretionary material (Barnes et al., 2010; Pecher et al., 2010; Schwalenberg et al., 2010), this area provides an ideal backdrop for assessing the impact of local variability on bacterial community structure.

Materials and methods

Regional setting

Sampling occurred aboard the R/V Tangaroa (National Institute of Water and Atmospheric Research) during the TAN0607 June–July 2006 cruise on the southern Hikurangi margin. Samples were collected along a 7 km cross-Porangahau ridge transect described by Schwalenberg et al. (2010) and Pecher et al. (2010) (Fig. 1). These previous studies incorporated seismic, thermal conductivity, controlled source electromagnetic analysis and preliminary geochemical results, and revealed a narrowly (meter scale) constrained area of methane-enriched fluid advection on the landward (western) flank of the ridge. An upward bend in the bottom simulating reflector (BSR) was evident on the seaward (eastern) flank of the ridge. Across the ridge, high-amplitude reflections starting at BSRs are attributed to free gas and gas hydrates in deep sediments (Crutchley et al., 2010; Pecher et al., 2010).

Figure 1.

 Study location along the Porangahau ridge.

Sample collection and handling

Four cores were selected for molecular biological analysis from the 12 cores collected across the ridge. Piston cores (PC) up to 4.8 m long were collected in 2.75 in. ID polycarbonate core liners and cut into 10 cm whole round sections at 10–43-cm intervals. Sample collection is described in detail in Schwalenberg et al. (2010). Sediment plugs (3 mL) were collected using polyethylene syringes with the ends cut off and transferred to 20-mL serum vials to measure headspace gas. Sterile polypropylene tubes were used to collect ∼20 g of sediment for molecular analysis. These were frozen at −20 °C during transport and held at −70 °C until analysis. Approximately 5 g of wet sediment was used for the gravimetric determination of porosity. Approximately 25 mL of pore-water was extracted from whole round sections using Reeburgh-style presses (Hamdan et al., 2008). Pore-water was frozen at −20 °C until analysis.

Geochemical analysis

Methane and ethane concentrations were determined according to Hoehler et al. (2000) using a Shimadzu 14-A gas chromatograph (GC) and a flame ionization detector (Hamdan et al., 2008; Schwalenberg et al., 2010). Methane stable carbon isotope ratios (δ13C) were obtained using a Trace GC interfaced via a GC-C III combustion unit to a Delta-Plus XP Isotope Ratio Mass Spectrometer (IRMS) (ThermoElectron). The measured values were plotted against reported δ13C values to generate a normalization equation that was applied to reference data to the Vienna Peedee Belemnite scale. Data are expressed in the standard δ-notation as ‰.

Dissolved inorganic carbon (DIC) concentration was measured using a UIC coulometer (Hamdan et al., 2008). A phosphoric acid–CuSO4 solution was used to convert DIC to carbon dioxide and precipitate dissolved sulfides. For DIC δ13C analysis, DIC was converted to carbon dioxide as above without CuSO4. Separation occurred on the trace GC equipped with a Porapak-Q column and DIC δ13C were measured and reported as described for methane.

Dissolved organic carbon (DOC) concentration was determined by wet chemical oxidation on an OI Analytical 1010 as described in Osburn & St-Jean (2007). DOC δ13C were determined in-line on the Delta-Plus XP IRMS.

Total inorganic (TIC) and total organic carbon (TOC) were analyzed in sediments using a Fisons EA 1108 interfaced with the Delta-Plus XP IRMS. Samples were analyzed first for total carbon (TC) and then for TOC after the addition of 10% HCl. Mass balance approaches were used to calculate TIC weight percent (wt%) and stable carbon isotopes (δ13CTIC) using the equations:




Sulfate and chloride concentrations were measured using a Dionex DX-120 ion chromatograph equipped with a 4-mm AS-9HC column (Paull et al., 2005). Dissolved ammonia, phosphate and silica were analyzed using the spectrophotometric methods of Grasshoff et al. (1983). Total dissolved sulfides (TDS) were determined according to Cline (1969).

Multitag pyrosequencing (MTPS) and phylogenetic analysis

Genomic DNA was extracted from ∼500 mg of wet sediment using the Bio 101 FastDNA® Spin kit for soil. DNA was quantified on a 1% agarose gel with ethidium bromide and diluted with DEPC water such that ∼10 ng of DNA was used as a template for PCR.

Before MTPS, triplicate samples were analyzed by length heterogeneity-PCR (LH-PCR) as a quality control for linear amplification of the community and to normalize PCR yield for samples before pooling. Hamdan et al. (2008) provide a detailed description of LH-PCR. Briefly, amplification of variable regions V1 and V2 of the small subunit rRNA gene was performed using the primers 6-FAM-27F and 355R. Controls accompanied reactions to determine PCR efficiency and calibrate the yield. PCR mixtures consisted of 1 × Gold buffer, 2.5 mM MgCl2, 0.2 mM dNTPs, 0.5 U AmpliTaq Gold, 0.5 μM primers, 0.01% bovine serum albumin and DEPC water. PCR was performed on a GeneAmp 9700 (Applied Biosystems) programmed as follows: initial denaturation at 95 °C for 11 min, 35 cycles of 95 °C (30 s), 48 °C (30 s), 72 °C (2 min plus 5 s per cycle) and a final extension at 72 °C (30 min). Product was visualized on an agarose gel, diluted based on product intensity, mixed with ILS-600 (Promega) and HiDi formamide, and analyzed on a SpectruMedix SCE9610 capillary sequencer.

Forty-eight forward fusion primers (27F) tagged on the 5′ end with a seven base barcode along with the emulsion PCR A adapter were used with a reverse fusion primer (355R) and emulsion PCR B adapter for MTPS. PCR was performed using these barcoded primer sets. Samples were amplified for 30 cycles as above with a 10-min final extension step. Tagged samples were normalized by yield, pooled and used in the emulsion step of 454-pyrosequencing. Sequencing was performed on a GS-FLX instrument in a single slot of a four-well gasket. The length of sequences averaged 262 bp. The sequence analysis pipeline is described by Naqvi et al. (2010). Briefly, sequences shorter than 75 bp, with quality scores <25 and those with multiple Ns were removed before analysis. Sequences were assigned to samples based on barcodes. Operational taxonomic units (OTUs) were clustered using cd-hit and phylotypes were assigned to OTUs using megablast searches against GenBank. The abundance of each phylotype (taxa) was annotated using the number of reads in each OTU and the relative abundance was calculated using the total number of reads from each barcode. megablast searches were conducted using a minimum cutoff of 96%, an e-value of 10−9, a bit size of 60 and a word size of 50. This resulted in 95% of sequences producing significant hits. An additional layer of filtering was used to sort results by % identity, bit-score and coverage of alignment. A custom perl script parsed out two hits per sequence: the one with the highest match and the one with the highest match containing phylogenetic information. The hit with the highest match is reported; however, this approach provides helpful information for environmental studies that generate a large number of sequences related to unidentified environmental phylotypes.

Statistical analysis

primer v6 (Plymouth Marine Laboratories) was used for all statistical analyses. Phylotype diversity was estimated using the Simpson's index (D). A Bray–Curtis similarity matrix was constructed for MTPS data and evaluated at ρ=0.05. Nonmetric multidimensional scaling (MDS) and hierarchal cluster analysis were conducted on MTPS similarity data to graphically display sample assemblages. The minimum stress level for the analysis was 0.01 with 25 restarts. MTPS similarity data were applied to the Biota and Environment (Bio-Env) matching procedure to determine which abiotic factors could be attributed with structuring communities (Clarke & Ainsworth, 1993; Clarke & Warwick, 2006; Hamdan et al., 2008). The analysis generates a Spearman rank correlation coefficient (ρ) to explain the covariance between data sets and attributes covariance to geochemical variables. A ρ near zero indicates limited covariance and ρ near one indicates strong covariance.


Sediment geochemistry

PC4 and PC17 were landward (west) of the ridge (Fig. 1). In PC4, methane and ethane were near the limits of detection (LOD), except at depth (Fig. 2a and b). TOC and TIC wt% decreased moderately (Fig. 2c and e) and the DOC concentration increased (Fig. 2d) with depth in PC4. The DIC concentration increased from <5 mM to a maximum of 20 mM between 200 and 430 cmbsf (Fig. 2f). The sulfate concentration decreased from 26 mM, to the LOD in the same depth range (Fig. 3a). The TDS maximum in PC4 was concomitant with the DIC maximum (Fig. 3b). The chloride concentration showed no depth-dependent trends (Fig. 3c). Silica, ammonium and phosphate concentrations increased (Fig. 3d–f) and porosity decreased (Fig. 3g) with depth in PC4. The SMTZ was determined as the depth where sulfate and methane approached the LOD. The SMTZ in PC4 was at 405 cmbsf. At the SMTZ, methane and DIC were 13C-depleted compared with depths above and below (Table 1).

Figure 2.

 Depth profiles of methane (a), ethane (b), TOC (c), DOC (d), TIC (e) and DIC (f) concentration in PC4, PC7, PC14 and PC17.

Figure 3.

 Depth profiles of sulfate (a), TDS (b), chloride (c), silica (d), ammonium (e) and phosphate (f) concentration and porosity (water content) (g).

Table 1.   Summary of stable carbon isotope ratio data for methane, DIC, TIC, DOC and TOC
CoreDepth interval (cmbsf)Depth zone (relative to SMTZ) n δ13Cmethaneδ13CDICδ13CTICδ13CDOCδ13CTOC
  1. Data are displayed for each core individually. Depth zones reported are based on the empirically determined depth of the SMTZ. Values reported are the average for each depth zone. No SMTZ was observed in PC14; thus, data for only one depth zone are presented. Cores are listed in the table in order of position (west to east) on the Porangahau ridge transect.


Although PC17 was the shallowest core, methane and ethane concentrations were the highest (overall) at its base (Fig. 2a). There were no depth trends for TOC or DOC in the core (Fig. 2c and d). In PC17, a peak in TIC wt% was observed at 156 cmbsf (Fig. 2e). The DIC concentration increased linearly from 80 to 156 cmbsf. Minimum sulfate and maximum TDS and silica concentrations were at 180 cmbsf in PC17 (Fig. 3a, b and d). In PC17, no trend in ammonium concentration vs. depth was found; the phosphate concentration increased linearly with depth (Fig. 3e and f). Porosity in PC17 was reduced compared with PC4 (Fig. 3g). The depth of the SMTZ in PC17 was 156 cmbsf. At the SMTZ methane, TOC, DOC and TIC were 13C-depleted compared with the other depths (Table 1) and carbonate deposits were evident.

PC7 and PC14 were seaward (west) of the ridge (Fig. 1). In PC7, the maximum methane concentration was at the core base. Ethane was not detected in either seaward core. The lowest average TOC (Fig. 2c) and the highest average TIC were in PC7 (Fig. 2e). A peak in TIC was observed from 190 to 240 cmbsf in PC7. DOC and DIC concentrations increased below 100 cmbsf in PC7 (Fig. 2d and f). The sulfate concentration remained constant in the upper 100 cm of PC7 and a linear decline in concentration was observed between 100 and 200 cmbsf (Fig. 3a). The TDS concentration was elevated between 115 and 140 cmbsf and peaked at the core base (Fig. 3b). The chloride concentration in PC7 was reduced at 215 cmbsf, the SMTZ depth (Fig. 3c). Because of low pore-water volume, no data for silica, phosphate or ammonium were available below 115 for PC7. In near-surface samples, the silica concentration and porosity were significantly lower than elsewhere (Fig. 3d and g). At the top of PC7, the ammonium concentration was equal to other cores; at greater depth, it was consistently higher (Fig. 3e). At the SMTZ in PC7 methane, TIC and DOC were 13C-depleted (Table 1).

Methane was at the LOD in PC14 (Fig. 2a) and the highest average TOC was observed (Fig. 2c). DOC and DIC concentrations increased gradually with depth between ∼200 and 322 cmbsf in PC14 (Fig. 2d and f). In PC14, sulfate concentrations were near 28 mM in the upper ∼200 cm of the core and declined gradually towards the base. The TDS concentration increased moderately in the same depth range (Fig. 3b). In PC14, the silica, ammonium and phosphate concentrations increased (Fig. 3d–f) and porosity declined moderately with depth. The SMTZ was not encountered in PC14.

MTPS results

In total, 40 144 sequences and ∼10 Mb of data were obtained. On average, 836 sequences per sample were analyzed with ∼220 phylotypes per sample. The Simpson's diversity index did not reveal statistically significant trends in numeric diversity with depth or among cores. However, the taxonomic composition varied among cores and with depth. More than 50% of the sequences were associated with taxa representing <1% of the total population. Approximately 4000 singletons were excluded from the discussion in order to simplify data presentation and because they could not be statistically correlated with environmental factors. Twelve percent of sequences could not be assigned a taxonomic identification because the best hit results were to uncultured taxa that contained no information for phylogenetic assignment. Such sequences with ‘no clear affiliation’ were related to 587 environmental phylotypes (Table 2).

Table 2.   Phylogenetic community structure based on 16S rRNA gene bacterial MTPS analysis
Phylogenetic class% of total
  1. All samples are included in these results.

Acidobacteria 0.25%
Actinobacteria 3.60%
Alphaproteobacteria 17.55%
Bacilli 1.11%
Bacteroidetes 1.07%
Betaproteobacteria 9.32%
Clostridia 2.38%
Cyanobacteria 0.17%
Deferribacteres 0.09%
Dehalococcoidetes 0.13%
Deltaproteobacteria 3.80%
Epsilonproteobacteria 0.06%
Erysipelotrichi 0.05%
Firmicutes 1.87%
Fusobacteria 0.17%
Gammaproteobacteria 2.78%
Gemmatimonadetes 0.09%
JS1 Candidate Division17.23%
Nitrospirae 0.03%
No clear affiliation12.42%
OD1 Candidate Division0.01%
OP1 Candidate Division0.15%
OP11 Candidate Division0.49%
OP3 Candidate Division0.05%
OP8 Candidate Division1.64%
Planctomycetacia 0.21%
Spirochaetes 1.40%
Tenericutes 0.02%
Tracheophyta chloroplast0.29%
Unclassified Proteobacteria0.10%
WS3 Candidate Division0.04%

Over 20% of the sequences were related to uncultured Chloroflexi/GNS (Table 2). The Chloroflexi/GNS were associated with 236 phylotypes. The majority of these were related (≥98%) to uncultured isolates from methane-rich sediments from the Gulf of Mexico, Chilean margin, Peru margin, Santa Barbara Basin and Sea of Okhotsk. Chloroflexi/GNS affiliated with Anaerolinea spp., Caldilineae spp. and Dehalococcoides spp. accounted for <4% of the total sequences. Proteobacteria were highly represented in the MTPS library. Of the Proteobacteria, Alphaproteobacteria were most abundant. Diversity within the Alphaproteobacteria was high, and sequences corresponded to ∼350 phylotypes. The majority of these had no cultured relatives; 25% were related to the order Rhodobacterales. Betaproteobacteria were ∼9% of sequences. More than 50% of Betaproteobacteria were affiliated with the order Burkholderiales. Epsilonproteobacteria were not well represented. Gammaproteobacteria and Deltaproteobacteria accounted for ∼3% of the sequences. Sequences related to the orders Altermonadales, Legionellales, Pseudomonadales and Xanthomonadales accounted for most Gammaproteobacteria. The majority of Deltaproteobacteria were related (>96%) to uncultured environmental isolates. Sequences related to the JS1 candidate division accounted for 17% of the library. Such sequences corresponded to 59 phylotypes, the majority of which originated from seeps and mud volcanoes in the Mediterranean Sea, Japan Trench, Gulf of Mexico, Chilean margin, Peru margin, Santa Barbara Basin and South China Sea. Bacilli, Bacteroidetes, Clostridia, Firmicutes, Fusobacteria, OP8 candidates, Planctomycetes and Spirochaetes were in low abundance, but accounted for up to 14% of the sequences in individual samples (Fig. 4).

Figure 4.

 Phylogenetic community composition in four cores collected across the ridge. Cores are arranged according to position on the ridge. Landward cores collected west of the ridge are PC4 (a) and PC17 (b). Seaward cores collected east of the ridge are PC14 (c) and PC7 (d).

Chloroflexi/GNS were the most abundant group in PC4 and accounted for >20% of the sequences. However, at depths <80 cmbsf and at the SMTZ, the abundance of Chloroflexi/GNS was reduced (Fig. 4a). Alphaproteobacteria sequences were also abundant in PC4. Near the surface of the core, Alphaproteobacteria accounted for >50% of the total sequences, and abundance declined with depth. By contrast, the abundance of JS1 candidates and Betaproteobacteria increased with depth in this core.

In PC17, Chloroflexi/GNS and Alphaproteobacteria sequences were most abundant near the top of the core and declined with depth. Like PC4, JS1 candidates and Betaproteobacteria were abundant at the core base (Fig. 4b). PC17 had the highest concentration of Deltaproteobacteria overall. Deltaproteobacteria were ∼7% of the sequences in PC17 and were concentrated at the SMTZ and at the top of the core. Deltaproteobacteria in PC17 were largely related to the order Desulfobacterales. Desulfosarcina/Desulfococcus isolates from the SMTZ in seeps on the Chilean margin (Hamdan et al., 2008) and Santa Barbara Basin (Harrison et al., 2009) were similar (≥98%) to sequences from this study.

Chloroflexi/GNS were ∼43% of the sequences in shallow sections of PC7 (Fig. 4c); below 90 cmbsf, their abundance was reduced. In contrast to other cores, Alphaproteobacteria sequences were minimal in PC7. Betaproteobacteria were concentrated at mid-depths in PC7. At 140 cmbsf in PC7, Betaproteobacteria accounted for >60% of the sequences. The largest accumulation of Gammaproteobacteria was in the upper 200 cm of PC7. A peak in JS1 candidate and Deltaproteobacteria sequences was observed at the SMTZ in PC7.

Chloroflexi/GNS sequence abundance was reduced in PC14 compared with the other cores (Fig. 4d). Alphaproteobacteria and Betaproteobacteria accounted for ≥10% of the sequences at all depths. As was the case at PC4, PC7 and PC17, JS1 candidate abundance generally increased with depth. Unclassified Deltaproteobacteria accounted for ∼10% of the sequences at the top of PC14, but were <2% of the sequences at other depths.

Bio-Env matching analysis of MTPS and geochemical data

The MDS plot revealed three clusters in the data set (Fig. 5). One cluster was composed of samples from PC7 (PC7 cluster). Samples 7_65 to 7_66 obtained near the top of PC7 clustered with most samples from PC4, PC14 and PC17 (Main cluster). The third cluster contained the SMTZ samples from PC4 and PC7 (SMTZ cluster). The SMTZ sample from PC17 (17_137) was a near outlier of this group. The Bio-Env matching analysis was used to describe the abiotic factors responsible for the clustering of samples (Table 3). In landward cores PC4 and PC17, sulfate was an important structuring factor. In PC4, DIC also yielded a ρ value>0.63. In PC17, sulfate, TDS, methane and silica all yielded ρ values>0.70. In the seaward core PC7, sulfate was the only variable that yielded a ρ>0.50. In PC14 also seaward of the ridge, silica and ammonium yielded ρ values of 0.55 and 0.44, respectively. The Bio-Env analysis was conducted on the sample clusters identified in the MDS plot (Fig. 5; Table 3). In order to improve the statistical relevance of the analysis, two additional samples were included in the SMTZ that were ordinates to the cluster (17_137 and 17_138). In the Main cluster, silica and TOC yielded the highest ρ values. TIC provided the highest ρ value for the PC7 cluster. For the SMTZ cluster, the variable with the greatest covariance with MTPS data was methane (ρ=0.68).

Figure 5.

 MDS plot of the assemblages of MTPS sequences in each sample. Contours were generated by a hierarchal cluster analysis conducted on a Bray–Curtis analysis of similarity for MTPS data.

Table 3.   Bio-Env procedure results
Sample groupBest variable (ρ)Best variable combinations (ρ)
  • Best variable results and best combination of variables yielding the largest Spearman rank correlation ρ coefficient between similarity matrices of bacterial community data and abiotic variables.

  • *

    For PC7, silica, phosphate and ammonium were excluded from this analysis because of incomplete data sets.

  • Note: abiotic variables included in the Bio-Env analysis: chloride, sulfate, TDS, DIC, methane, ethane, silica, phosphate, ammonium, TOC, TIC and DOC concentration and porosity. Stable isotope data were excluded because they are not entirely independent of concentration data.

All samplesSulfate (0.33)Sulfate, phosphate, TOC, TIC (0.460)
Ethane (0.32)
TIC (0.32)
TOC (0.31)
PC4DIC (0.65)DIC, silica, phosphate (0.67)
Sulfate (0.63)
Silica (0.61)
Ethane (0.57)
PC7*Sulfate (0.50)Sulfate (0.50)
Ethane (0.33)
DIC (0.31)
TIC (0.21)
PC14Silica (0.54)Silica (0.54)
Ammonium (0.44)
Phosphate (0.42)
TDS (0.41)
PC17Silica (0.84)Sulfate, TDS, silica, TOC, TIC (0.90)
TDS (0.74)
Methane (0.74)
Sulfate (0.72)
PC7 clusterTIC (0.59)Chloride, DIC, TIC (0.77)
Sulfate (0.44)
Ethane (0.40)
DIC (0.35)
SMTZ clusterMethane (0.68)Methane, TOC (0.72)
TIC (0.45)
TOC (0.33)
Phosphate (0.10)
Main clusterSilica (0.34)TDS, silica, phosphate, TOC, TIC (0.46)
TOC (0.32)
DIC (0.22)
Sulfate (0.21)


Factors structuring bacterial communities

Geological and geochemical conditions of the sedimentary environment play an important role in bacterial community structure and diversity (Kuehl et al., 1996; Edlund et al., 2008). Likewise, the availability of organic carbon and methane at each site likely drives bacterial diversity. With the exception of the PC7 cluster (Fig. 5), there were no distinct separations of communities based on location alone. However, different abiotic factors were attributed as influences on community structure in each core and in each MDS cluster.

PC7 cluster

The central question of this study was determining the influence of local environmental heterogeneity on bacterial community composition. Local influence was best demonstrated at PC7. The features that delineated the PC7 cluster were minimal Alphaproteobacteria and elevated Betaproteobacteria sequences (Fig. 4b). Samples in the PC7 cluster had a high concentration of Betaproteobacteria (Fig. 5) associated with the order Burkholderiales and genus Achromobacter. Such sequences were 99% similar to two cultured isolates of Achromobacter insolitus (GenBank accession EU221379 and EU520399). In laboratory experiments, these isolates exhibited diazotrophism (Sala et al., 2008) and synthesis of the enzyme 1-aminocyclopropane-1-carboxylate (ACC) deaminase, which is responsible for hydrolysis of the ethylene precursor ACC into ammonia and α-ketobutyrate. Common to these isolates is the capacity to generate ammonium. The presence of sequences related to the isolates described above may explain the elevated ammonium concentrations in PC7 relative to other cores (Fig. 3e) and suggest a biological pathway for ammonium generation. The geochemical data support a hypothesis of in situ ammonium production in PC7. In PC7, % nitrogen (not shown) averaged 0.05, approximately half that observed elsewhere. The average TOC wt% was reduced compared with the other locations, and C : N ratios were <6.2 compared with an average of 7.6 at other sites. Surface porosity (Fig. 3g) and dissolved silica concentrations were reduced in the upper 100 cm of PC7 relative to other cores (Fig. 3d). Dissolved silica may be used as an indicator of phytodetritus deposition because its concentration in sediment is controlled by the burial and dissolution of silicate (Schink et al., 1974, 1975; Demaster, 1981). The depth profile shape for silica in PC7 was similar to other cores, although the surface concentrations were reduced. This suggests that controls on silica dissolution are similar across transect, but that silicate is diminished at the surface of PC7. Reduced silica, TOC, bulk nitrogen and porosity, and low C : N ratios suggest that resupply of organic matter and nitrogen was reduced in surface sediments at PC7. This was unexpected because basins such as the one PC7 was positioned in often act as sediment traps, which could explain ammonium enrichment. However, seismic data do not suggest substantial sedimentation near PC7 (Pecher et al., 2010). Current flow is channeled down the basin (Carter & Manighetti, 2006) and causes erosion (I. Pecher, pers. commun.), which may remove surface deposition and establish a nitrogen requirement. Because some Achromobacter generate ammonium from nitrate under nitrate-limited conditions (King & Nedwell, 1985), the physical conditions at PC7 may have conferred an advantage to A. insolitus-related taxa and may explain their elevated sequence abundance.

Other studies have demonstrated in situ nitrogen production in methane-rich marine sediments. Dekas et al. (2009) documented diazotrophism in aggregates of ANME-2 and Desulfosarcina/Desulfococcus. Pernthaler et al. (2008) commented on the potential involvement of Burkholderiales in nitrogen biogeochemistry in methane charged marine sediments. Further study may reveal whether diazotrophism, ACC cleavage or other microbial nitrogen cycling partnerships occur in these sediments.

Main cluster

The Main cluster included most samples from PC4, PC17 and PC14. In these cores, Alphaproteobacteria were abundant near the surface and declined with depth (Fig. 4a, c and d). Others report similar depth trends for this group in marine sediments (Webster et al., 2006; Mills et al., 2008). The most abundant Alphaproteobacteria in the Main cluster were 99% related to uncultured isolates from organic-rich surface sediments near the Crozet Island archipelago (GenBank accession FM214379 and FM214399). These phylotypes accounted for 26% of Alphaproteobacteria in this study, were observed in nearly all samples from the Main cluster and were absent from PC7. Rhodobacterales accounted for most Alphaproteobacteria with known phylogeny. The most abundant Rhodobacterales was associated with a Loktanella sp. (GenBank accession FJ196061) isolated from coastal sediments off east Antarctica. The second most abundant Rhodobacterales phylotype was associated with a Rhodobacter sp. isolate (GenBank accession EU979473) from the Columbia River estuarine turbidity maximum. Rhodobacterales are the primary surface sediment colonizers in coastal waters (Dang & Lovell, 2000) and often play a role in the oxidation of newly deposited organic matter. The Bio-Env results implicate silica and TOC as important structuring factors on the Main cluster (Table 3). These results may explain why sequences associated with organic matter utilization were so abundant in surface samples in the Main cluster.

The order Rhizobiales was well represented in samples in the Main cluster and accounted for 9% of Alphaproteobacteria. Rhizobiales related to the Methylocystaceae family were in highest abundance in PC4, limited in PC17 and PC14 and not observed in PC7. The elevated abundance of Methylocystaceae in PC4 compared with the other cores suggests that although Methylocystaceae are methanotrophic, phylotypes in this study may proliferate under conditions of low methane concentration. In addition, anoxic conditions in PC17 and PC7 (indicated by the presence of TDS –Fig. 3b) would inhibit obligately aerobic Methylocystaceae and may explain their reduced sequence abundance in the cores.

The Chloroflexi/GNS featured largely in samples from the Main cluster (Fig. 5). This and other studies report that the location of Chloroflexi/GNS within the sediment column is variable (Inagaki et al., 2003, 2006; Kormas et al., 2003; Heijs et al., 2007; Biddle et al., 2008; Hamdan et al., 2008; Harrison et al., 2009). Chloroflexi/GNS are ubiquitous in methane-rich sediments (Webster et al., 2006). In such sediments from the Chilean margin, Santa Barbara Basin, Mediterranean Sea and Gulf of Mexico, Chloroflexi/GNS are most prevalent below the SMTZ (Reed et al., 2006; Heijs et al., 2007; Hamdan et al., 2008; Harrison et al., 2009). Others have discovered Chloroflexi/GNS to be uniformly distributed in the deep subsurface (>50 m) sediments on the Peru margin (Biddle et al., 2008). Inagaki et al. (2006) noted that on the Peru margin, Chloroflexi/GNS dominate organic-rich methane-bearing sediments that lacked gas hydrates and that this was a key biological factor distinguishing nearby sites that contained gas hydrates from those that did not.

Frequent observations of Chloroflexi/GNS in methane-rich sediments have prompted speculation on their role in methanogenesis (Sekiguchi, 2006). The semi-ubiquitous appearance of Chloroflexi/GNS in studies of marine sediments (Kormas et al., 2003; Biddle et al., 2008) indicates that factors other than methane are correlated with their appearance. In this study, no direct relationship between Chloroflexi/GNS and methane was observed. Methane concentration was generally low in samples with high Chloroflexi/GNS abundance, and the Bio-Env analysis (Table 3) indicated that TOC, not methane, was a structuring factor on the Main cluster.

Chloroflexi/GNS were abundant in sediments from the top of PC7. In PC7 samples that grouped in the Main cluster (Fig. 5), the majority of Chloroflexi/GNS were affiliated with isolates from the Brazos-Trinity Basin of the Gulf of Mexico (Nunoura et al., 2009). The Brazos-Trinity Basin like PC7 is relatively TOC poor, having C : N ratios that are non-Redfield (Gilhooly et al., 2008). The physicochemical and biological parity between these sites suggest bacterial composition to be a highly specific indicator of abiotic conditions in marine sediments. The location where PC7 was obtained was likely eroded by hydraulic forces. Because of this, newly deposited, labile TOC would be limited at the surface. However, erosion would expose buried recalcitrant TOC. Studies demonstrate that recalcitrant TOC is responsible for the long-term survival of microorganisms in the deep subsurface (Fredrickson & Balkwill, 2006) and may explain the observations of Chloroflexi/GNS in this and similar environments (Inagaki et al., 2006; Biddle et al., 2008; Nunoura et al., 2009).

Although most of PC17 was nested in the Main cluster, the robust Spearman Rank correlation coefficients for the core deserve some individual discussion. Silica was the main structuring factor in PC17 (Table 3). A significant increase in the silica concentration was observed below the SMTZ, concomitant with the TDS maximum (Fig. 3d). Studies demonstrate silica dissolution in conjunction with sulfate reduction (SR), AOM and carbonate precipitation (Birnbaum & Wireman, 1984; Jørgensen & Boetius, 2007; Pierre & Fouquet, 2007). Elevated silica concentration near the SMTZ in PC17 was likely a secondary result of metabolic activity. Hence, the high ρ for silica may be the result of covariance of factors rather than direct cause and effect on community structure. This is reflected in the Bio-Env results, which indicate that sulfate, TDS and methane are also important correlates on community structure in PC17 due to their coupled relationships in sediment diagenesis.

SMTZ cluster

The biological features common to the SMTZ were elevated concentration of Deltaproteobacteria and JS1 candidate sequences relative to other depths (Fig. 4). The most abundant JS1 phylotype in this study was ≥98% similar to an isolate from hydrate-bearing sediments from the Chilean margin (GenBank accession EF093942). The Bio-Env analysis indicated that methane was the main abiotic driver on the SMTZ cluster. Although the biogeochemical role of JS1 candidates is unknown, it has been suggested that their abundance is controlled by the presence of methane (Webster et al., 2006) and low sulfate concentration (Parkes et al., 2007). In this study, the influence of methane on JS1 candidates cannot be ruled out because they were abundant in methane-rich sections of PC4, PC7 and PC17 (Fig. 4). However, because they were observed in PC14 (Fig. 4c), where methane was below the LOD, other factors must influence their presence. Similarly, if methane principally governed JS1 candidate abundance, it is expected that they would be most abundant in PC17, which had the highest methane concentrations; however, this was not the case. JS1 candidates are often observed in gas hydrate-containing sediments (Inagaki et al., 2006), although none were observed in this study. The distribution of JS1 candidates in this study suggests metabolic diversity in this group, which involves more than methane metabolism.

At the SMTZ Deltaproteobacteria were largely affiliated with the Desulfosarcina/Desulfococcus subgroup. Other studies report enrichment of this subgroup at the SMTZ (Mills et al., 2003; Inagaki et al., 2006; Hamdan et al., 2008) and on their suspected role in AOM. The appearance of these sequences outside of the SMTZ was limited, indicating that conditions at the SMTZ select for these phylotypes. Below the SMTZ, Deltaproteobacteria generally accounted for <1% of the sequences. Above, phylotypes related 98% to a Desulfobacter isolate (GenBank accession U85476) obtained from salt marsh sediments were abundant. The high concentration of Desulfobacterales– the order that encompasses most marine sulfate reducers, above the SMTZ is indicative of the SR zone (Arakawa et al., 2006).

The data for Deltaproteobacteria indicate that sulfate-dependent AOM and organoclastic SR likely occurred in cores from this study. Without a conservative tracer to differentiate physical mixing from biogeochemical processes, it is difficult to delineate SR processes (i.e. organoclastic SR vs. AOM) from the nonlinear sulfate profiles (Pohlman et al., 2008). However, sulfate and methane profiles, particularly from PC7 and PC17, the shallow depth of the SMTZ in these cores and the concentration of Desulfosarcina/Desulfococcus-related sequences suggest that SR associated with AOM was focused at the SMTZ.

Methane, DIC, TOC and DOC δ13C (Table 1) and methane and sulfate concentration data support the idea that both AOM and organoclastic SR impact sediment biogeochemistry in PC4, PC17 and PC7. TOC δ13C above the SMTZ in these cores fall within the range of pelagic organic matter (Claypool & Kaplan, 1974; Peterson et al., 1994; Niggemann & Schubert, 2006). Thus, above the SMTZ 13C-enriched DIC may have resulted from the metabolism of pelagic organic matter. However, at the SMTZ of PC4 and PC17 13C-depleted DIC could only result from oxidation of 13C-depleted methane because δ13C for TOC and DOC in these cores cannot explain the substantially 13C-depleted DIC. At the SMTZ in PC7 and PC17, the TIC concentration was elevated relative to PC4 and PC14 (Fig. 2e) and carbonate deposits were observed in PC17. Reduced porosity at the SMTZ was observed in both PC17 and PC7. These observations suggest carbonate precipitation in conjunction with AOM. In PC17, 13C-depleted TIC at the SMTZ can only be explained by AOM (Table 1). DOC δ13C indicate that 13C-depleted methane was incorporated into the organic carbon pool at the SMTZ of PC17 (Table 1). Both TIC and DOC were 13C-depleted at the SMTZ in PC7 (Table 1). Neither TIC accumulation nor TIC 13C-depletion was observed in PC4; however, 13C-depleted DIC was (Table 1). Because the abundance of Desulfosarcina/Desulfococcus at the SMTZ in PC4 was reduced compared with PC17 and PC7, this may be indicative of lower rates of AOM, and hence, a less distinct geochemical signature. However, the reduced supply of methane at PC4 relative to PC17 and PC7 may have driven the reduction in Desulfosarcina/Desulfococcus in the first place. As a whole, these data suggest that the local influence of methane on bacterial communities is significant in PC17 and PC7, reduced in PC4 and negligible in PC14.


Geomicrobial studies that incorporate large geochemical datasets in conjunction with next-generation sequencing assist in understanding the biogeochemical role of uncultured bacterial populations. In methane seep sediments, much uncertainly remains as to the role of Chloroflexi/GNS and JS1 candidates. Their frequent appearance in such environments has resulted in the expectation that methane is correlated with their appearance. However, statistical analysis indicates that TOC may be an important structuring factor on Chloroflexi/GNS. Methane likely influences the presence of many JS1 candidates; however, because of their appearance in methane-poor locations, other substrates must be used as metabolites by this group. Heterogeneity in the supply of nutrients and the composition of carbon pools across the ridge is associated with geophysical, hydrodynamic and geological features. Such heterogeneity is reflected in the composition of microbial communities. In some cases, heterogeneity may be the result of metabolic activity by different communities. However, at the SMTZ, a somewhat homogeneous bacterial population was evident and community structure had characteristics similar to those of SMTZ communities observed in methane seeps throughout the ocean.


This work was supported by the Naval Research Laboratory Chemistry Division Young Investigator Program and the Office of Naval Research platform support program. We thank the captain and crew of the R/V Tangaroa for field assistance, Roswell Downer and Layton Bryant for sample recovery, Rebecca Plummer and Dillon Gustafson for laboratory assistance and Ingo Pecher and Suzannah Toulmin for helpful discussions. We thank Co-Chief Scientists Ingo Pecher and Stewart Henrys and the CHARMNZ science party for their support of this study.