Introduction

Submarine mud volcanoes (MVs) are characterized by seepage-related geomorphological features and act as windows for emission of deep subsurface hydrocarbon gases (mainly methane), fluids, and a complex mixture of sediments1,2,3. They are considered as significant geological sources of global atmospheric methane emissions4,5. In the atmosphere, methane is a strong greenhouse gas that is >30 times more potent than carbon dioxide6. Recently, a number of active MVs featuring gas hydrate deposits were discovered in the Canadian Beaufort Sea7,8. This marine province is expected to experience an increases of 1 to 2 °C in bottom water temperature in the near future9. Similar to other Arctic continental margin sediments10,11,12, bottom water temperature increase may lead to the dissociation of gas hydrate deposits13,14 and elevated methane emissions to the water column and, potentially, to the atmosphere where it further contributes to global warming8,15,16.

Submarine MVs often host dense and diverse microbial and faunal communities fueled by gases and fluids from the subsurface17,18,19,20. The richness of such chemosynthetic communities depends on long-term availability of electron acceptors (e.g. oxygen and sulfate) and donors (e.g. methane) in the sediments21,22. In this regard, microbial processes are typically dominated by aerobic and anaerobic oxidation of methane (MOx and AOM, respectively), which are performed by different clades of aerobic methanotrophic bacteria (MOB)23, or anaerobic methanotrophic archaea (ANMEs)24. ANMEs are often accompanied by sulfate-reducing bacteria (SRB)25,26,27. At highly dynamic MVs, the capacity for methane removal is often reduced because upwelling geofluids, devoid of electron acceptors, hinder influx of electron acceptors (e.g., oxygen and sulfate) from the water column15,19. Physical disturbances by upwelling mud and fluids at active MVs also can give rise to heterogeneous distributions of methanotrophic communities at small scales <1 m of a single MV28,29.

Since the discovery of active MVs across the slope of the Canadian Beaufort Sea, the biogeochemical signatures of archaeal communities involved in AOM have shown spatially discriminative distributions among three MVs (MV282, MV420, and MV740), at water depths of 282 m, 420 m and 740 m, respectively8,30. In this study, we collected sediment push cores from visually discriminative chemosynthetic fields at an active MV (MV420): (i) devoid of megafauna and/microbial mats (DM) to the naked eye, (ii) covered with bacterial mats (BM), or (iii) colonized by siboglinid tubeworms (ST), using a small remotely operated vehicle (ROV) (Fig. 1). In addition, a sediment core was retrieved in a reference site, using a box corer (Fig. 1). Gas, porewater, bulk element, lipid, and nucleic acid analyses were performed to obtain a comprehensive picture of spatial variation in bacterial and archaeal methanotrophic communities. Our study of the spatial and vertical distributions of methanotrophic communities at MV420 sheds light on a key biogeochemical factor shaping discriminative communities.

Figure 1
figure 1

(a) Map showing the study area (red box) with an inset of the regional map of Alaska and northwestern Canada modified from Paull et al. (2015). (b) Map showing the mud volcano at a water depth of 420 m (MV420) and the reference site on the upper slope of the Canadian Beaufort Sea. The map was generated in Generic mapping tools version 5.4.5 (http://gmt.soest.hawaii.edu/projects/gmt). The high-resolution (1 m) bathymetry data was acquired using AUV in MBARI, processed in MB System (https://www3.mbari.org/data/mbsystem/index_ldeo.html). (c) Video photographs showing discriminative microbial habitats at MV420, i.e., devoid of megafauna and/microbial mats (DM), bacterial mats (BM), and siboglinid tubeworms (ST). The video photographs were taken by the MBARI ROV team during the Araon cruise in 2017.

Full size image

Results

Geochemical properties

Methane concentration varied at three MV420 sites: DM, BM, and ST (Fig. 2). The highest methane concentration was observed at the DM site (6.6 ± 4.0 mM), followed by the BM (4.0 ± 3.6 mM) and ST (2.4 ± 1.2 mM) sites (Fig. 2). Methane concentration was below the detection limit at the reference site (Fig. 2). The carbon isotopic composition of methane (δ13CCH4) was −64.8‰ to −45.5‰, and generally showed 13C-enrichment toward the surface sediment layers, in accordance with the decrease of methane concentration (Fig. 2). Sulfate concentration ranged from 27.4 mM (sediment surface) to 1.1 mM (bottom section of the core), but penetration depths ranged from 1.5 cm to 7 cm, while the reference site showed a value of 27.8 ± 0.2 mM throughout the entire core (Table 1 and Fig. 2). Porewater chloride concentration varied between 495 mM and 550 mM, co-varying with sulfate concentration along the sediment depth (Fig. 2). Porewater dissolved inorganic carbon (DIC) concentration increased from surface sediments (1.8 mM) toward the bottom of the core (24.7 mM; Fig. 2). δ13CDIC values were −11.6 ± 3.2‰ at the DM site, while those of the BM and ST site were more 13C-depleted (Fig. 2). In contrast, at the reference site, porewater DIC concentration was relatively constant along the sediment depth at 2.7 ± 0.3 mM, and the δ13CDIC values were 13C-enriched (−4.8 ± 1.6‰). Diffusive methane flux was highest at the DM site (–648.5 mmol m−2 y−1), followed by the BM (−427.5 mmol m−2 y−1) and ST (−146.6 mmol m−2 y−1) sites (Table 1). The corresponding sulfate fluxes were 671.0 mmol m−2 y−1, 449.1 mmol m−2 y−1, and 191.4 mmol m−2 y−1, respectively.

Figure 2
figure 2

Depth profiles of gas (methane), porewater (sulfate, chloride, and dissolved inorganic carbon [DIC]), and bulk elements (total organic carbon and total sulfur) from sediment cores of (a) devoid of megafauna and/microbial mats (DM), (b) bacterial mats (BM), (c) siboglinid tubeworms (ST), and (d) reference sites. “ND” denotes “not determined.”

Full size image
Table 1 Temperature of the surface seafloor and methane and sulfate flux (calculated from concentration profiles and penetration depths of sulfate.
Full size table

Total organic carbon (TOC) content ranged from 0.8 wt.% to 1.9 wt.% at all MV420 sites and from 1.4 wt.% to 1.9 wt.% at the reference site (Fig. 2). δ13CTOC values (−28.0 ± 0.6‰) were lower in the upper 4 cm depth at the BM site, while other MV420 and reference sites showed relatively 13C-enriched values (−25.5 ± 0.1‰) (Fig. 2). Total sulfur (TS) content varied between 0.1 wt.% and 1.0 wt.%, with the highest value at the ST site (Fig. 2). δ34STS values were in the range of −18.6‰ to 9.2‰, with the largest sulfur isotopic variation (∆δ34S; ~20) at the ST site (Fig. 2).

Lipid signatures of methanotrophic communities

The 4α-methyl sterols that are putatively specific for bacterial groups involved in MOx31 were abundant at the surface of all MV420 sites but not the reference site. Concentrations were highest at the surface of the BM site, with a range of 0.03–0.37 μg g−1 (Fig. 3 and Supplementary Information Table S1). Specific fatty acids (FAs) such as C16:1ω8 and C16:1ω5, which were inferred as markers for aerobic methanotrophic bacteria and SRB, respectively,23,32 were detected in a range of 0.01 to 0.14 μg g−1 and 0.01 to 0.09 μg g−1, respectively (Fig. 3 and Supplementary Information Table S1). These FAs predominated near the surface of each MV420 site and decreased along the sediment depth.

Figure 3
figure 3

Vertical profiles of selected bacterial and archaeal lipid biomarker concentrations obtained from sediment cores of (a) devoid of megafauna and/microbial mats (DM), (b) bacterial mats (BM), (c) siboglinid tubeworms (ST), and (d) reference sites. “ND” denotes “not determined.”

Full size image

In contrast to bacterial lipids, isoprenoidal dialkyl glycerol diethers (DGDs) (e.g., archaeol and sn-2-hydroxyarchaeol), which are specific AOM-related biomarkers33, were identified in deeper sediment depths (4 to 16 cm) at each MV420 site but not at the reference site (Fig. 3 and Supplementary Information Table S1). Isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs) containing zero to three cyclopentane moieties (GDGT-0 to GDGT-3) and crenarchaeol were detected in all samples investigated (Fig. 3). Isoprenoid GDGTs were dominated by GDGT-0 and crenarchaeol, at concentrations of 0.05 to 0.72 μg g−1 and 0.03 to 0.77 μg g−1, respectively, while GDGT-1, GDGT-2, and GDGT-3 were much less abundant (<0.03 μg g−1) in all sediments analyzed (Fig. 3 and Supplementary Information Table S1).

The δ13C values of 4α-methyl sterols were −104.9‰ to −30. 5‰, and these compounds were strongly 13C-depleted near the surface layer of each site. The δ13C values of FAs ranged from −63.8‰ to −41.2‰ in all sediments investigated. The δ13C values of isoprenoid DGDs varied between −109.6‰ and −38.6‰, with higher depletion in sn-2-hydroxyarchaeol (−91.4 ± 16.4‰) than in archaeol (−47.6 ± 6.5‰) (Fig. 4 and Supplementary Information Table S1). The δ13C values of biphytanes (BPs) derived from isoprenoid GDGTs were in the range of −79.5‰ to −19.0‰, showing more 13C-depleted BP-1 values at all MV420 sites.

Figure 4
figure 4

Vertical profiles of selected bacterial and archaeal lipid biomarker isotopic compositions from sediment cores of (a) devoid of megafauna and/microbial mats (DM), (b) bacterial mats (BM), (c) siboglinid tubeworms (ST), and (d) reference sites. “ND” denotes “not determined.”

Full size image

Dominant methanotrophic taxa

The number of 16S rRNA gene reads of bacteria and archaea ranged from 8711 to 122,357 and from 8070 to 63,065 sequences per sample, respectively (Supplementary Information Table S2). We identified 25 operational taxonomic units (OTUs) belonging to the order Methylococcales of Gammaproteobacteria, which are aerobic methanotrophic bacteria23,34 (Supplementary Information Table S3 and Fig. S1). Among them, B_OTU004 dominated near the surface of the DM site (8.0 ± 6.6%), followed by the BM site (3.3 ± 4.2%) and the ST site (1.1 ± 1.2%), while it was not detected at the reference site (Fig. 5 and Supplementary Information Tables S3 and S4 and Fig. S2). With the exception of B_OTU004, the proportions of other OTUs of Methylococcales were less than 0.2 ± 0.3% in the samples investigated (Supplementary Information Table S3).

Figure 5
figure 5

Vertical profiles of relative abundance of major OTUs (Methylococcales; B_OTU004 and Methanomicrobia; A_OTU001, A_OTU008, A_OTU014, and A_OTU015) obtained from sediment cores of (a) devoid of megafauna and/microbial mats (DM), (b) bacteria mats (BM), (c) siboglinid tubeworms (ST) sites, and (d) reference sites. “ND” denotes “not determined.”

Full size image

The relative abundance of A_OTU001, which belongs to ANME-3, was relatively higher (up to 97.1% of archaeal sequences) along sediment depths of MV420 sites, while it was less than 0.08% in the reference site (Fig. 5 and Supplementary Information Table S4 and Fig. S2). Another archaeal OTU (A_OTU008) that belongs to ANME-2 comprised 1.4% to 8.7% of archaeal sequences in the deeper sediment depths (4 to 11 cm) of the ST site. But this OTU was rare or not detected in the sediments of the DM and BM sites (Fig. 5 and Supplementary Information Table S4 and Fig. S2). A_OTU014 and A_OTU015 predominated near the surface of the DM site (up to 13.3% and 6.6% of archaeal sequences, respectively), while the proportion of these OTUs was less than 0.4% at the BM and ST sites (Fig. 5 and Supplementary Information Table S4 and Fig. S2).

Twelve OTUs of Deltaproteobacteria with a greater than 5% relative abundance of bacterial sequences were separately affiliated with the family Desulfobacteraceae or Desulfobulbaceae (Supplementary Information Table S4 and Fig. S2). Among eight OTUs belonging to the family Desulfobacteraceae, most (B_OTU002, B_OTU009, B_OTU011, B_OTU014, B_OTU024, B_OTU029, and B_OTU076) showed relatively higher abundances at deeper sediment depths at all MV420 sites but not the reference site. B_OTU014 showed the highest abundance (57.5%) at the 16 cm sediment depth of the DM site but did not exceed 2.4% at the BM and ST sites. In contrast, B_OTU018 showed the highest abundance (7.4%) near the sediment surface of the ST site. Among four OTUs belonging to the family Desulfobulbaceae, B_OTU010 showed relatively high abundance (2.8 to 14.4%) in the upper 3 cm of all the MV420 sites but not the reference site. In contrast, B_OTU064 showed high relative abundance only in the upper 4 cm of the ST site (2.1 to 5.5%), and the proportions of B_OTU119 was higher at the 0 to 1 cm depth of the BM site.

Co-occurrence of ANMEs with sulfate-reducing Deltaproteobacteria

To infer the putative sulfate-reducing partners of ANMEs, co-occurrence between two archaeal OTUs (ANMEs) and 12 bacterial OTUs (Deltaproteobacteria) was investigated by calculating Pearson correlation coefficients (Supplementary Information Table S5). At the DM and BM sites, no OTUs showed significant correlation (p < 0.01) between ANMEs and SRB. At the ST site, both archaeal OTUs showed significantly positive correlation (r = 0.71 to 0.92, p < 0.01) with a few bacterial OTUs (B_OTU002, B_OTU011, B_OTU018, and B_OTU029).

Discussion

Biogeochemical characteristics along chemosynthetic habitats

Gas bubbles were observed in the ROV video rising within the water column during sampling in the chemosynthetic habitats of MV420. Previous studies using an autonomous underwater vehicle (AUV) and an ROV reported that the fresh and warm subsurface muds exposed during eruptions were mostly saturated with methane (δ13CCH4 = −64‰ and δDCH4 = −222‰), indicating primarily microbial sources8,35. Consistent with those studies, the overall geochemical signatures observed at the MV420 sites were clearly different in comparison with those of the reference site (Fig. 2).

Methane and sulfate fluxes calculated based on concentration gradient varied among the three MV420 sites (Table 1). The DM site, which was devoid of both sessile and mobile taxa such as polychaeta and bivalves, had the highest methane and sulfate fluxes (648.5 mmol m−2 y−1 and 671.0 mmol m−2 y−1, respectively). At the DM site, strong methane fluxes may cause an imbalance in hydraulic pressure, especially in shallow layers where sediments are not yet consolidated, resulting in active invasion of bottom seawater36. Moreover, the DM sites have been interpreted to be on mud flows which are too young to have yet recruited sessile organisms and in place not even been invaded by mobile taxa8,37,38. However, in accordance with decreasing fluid flux (<450 mmol m−2 y−1), free-living bacteria communities and ST patches had developed differently on the surface of MV420. Thiotrophic bacterial communities were commonly found in seeps with moderate fluid flow rates (~1 m yr−1), while STs dominated at seeping systems with a fluid flow rate less than 0.5 m yr−1 15,22. Such biological succession has been reported at various seeping sites, which are differentiated by water depths and methane origins — for instance, at the Håkon Mosby mud volcano (HMMV)19, the Congo deep-sea fan39, the Gulf of Mexico40, and the deep Eastern Mediterranean Sea41. Varying fluid fluxes therefore appear to be a plausible control factor for distinguishing the near seafloor chemosynthetic habitats within MV420.

Corresponding with the depleted chloride concentrations along sediment depth profiles, the penetration depth of sulfate supplied from seawater varied among the three MV420 sites, possibly leading to isotopic variations of the total sulfur pool (Table 1 and Fig. 2). The largest isotopic variation (>20) investigated at the ST site may be attributed to the higher contributions of δ34S values (ca. + 20‰) from the bottom seawater sources42. The sulfate-methane transition zone was formed at the deeper sediment horizons of the ST site compared with other sites. In this regard, δ13CCH4 isotopic signatures were more enriched at the ST site, indicating that the remaining methane pool became progressively enriched in 13C35,43,44. These results are in close agreement with 13C-depleted DIC values, as a result of isotopic fractionation through active AOM under lower methane fluxes43,44,45,46. Considering that endosymbiotic bacteria inhabiting STs exploit some substrates (e.g., methane and by-products involved in AOM) in subsurface sediments47,48, bioirrigation activities by STs may enhance the availability of electron acceptors in deeper sediments17,49,50,51. These activities allow niche separation between chemosynthetic organisms competing for common energy sources over time29,52,53. Under varying methane fluxes investigated at MV420, both chemosynthetic organisms (e.g., free-living bacteria and STs) inhabiting seeping sediments appear to have been distinctly separated based on energy source. These results imply that different habitat types mirroring the discriminative intensities of ascending gas-fluids may influence the spatial distribution of methanotrophic communities by altering the availability of electron acceptors.

Occurrence of aerobic methanotrophs

MOx- and AOM-related microbial communities assessed according to lipid biomarkers and 16S rRNA gene sequences showed a clear down-core shift within each site and variance among the MV420 sites. For the first time in Beaufort MV, we found abundant and 13C-depleted 4α-methyl sterols in concert with high proportions of OTUs belonging to the order Methylococcales near the sediment surface, strongly indicating a high abundance of aerobic methanotrophs in these sediments (Figs. 3–5), as has been reported from various other seepage sites19,28,53,54. The predominance of aerobic methanotrophs at the surface appears to result from the intensities of oxidant-depleted subsurface fluids that limit downward diffusion of sulfate and oxygen from seawater to the uppermost centimeters of the sediment28,53,54,55. In such situations, which do not easily allow AOM reactions in deeper sediment layers, a preferential niche of MOx-related methanotrophs is formed in the surface sediment where oxygen is available. Notably, 13C-enriched 4α-methyl sterols in deeper sediment layers indicate that these compounds originate from processes other than methanotrophy (e.g. organisms using methanol)56. However, with the current data set, it is challenging to elucidate exact processes involved, and thus further research is needed.

With discriminative distributions of MOx-related methanotrophs among the three MV420 sites, 13C-depleted 4α-methyl sterols were much more prevalent at the BM site than at other sites (Figs. 3–5). Moreover, 13C-depleted TOC was abundant at the surface of the BM site, indicating that MOx-related biomass was incorporated and preserved in the organic carbon pool for this site57,58,59. Such distribution patterns investigated at MV420 appear to differ slightly from those of similar MV systems (i.e., HMMV) that showed predominant MOx-related methanotrophs under the highest methane flux19,28. One possible reason for this difference might be that the mud flows migrating to the tops of the DM site were too active and too young for colonization of MOx-related methanotrophs8, in contrast to the BM and ST sites. Moreover, previous studies of various seeping sites have suggested that seafloor temperature and oxygen availability are the key factors shaping seep microbial communities28,54. In this study, in situ bottom temperature in the surface did not vary across the MV420 sites (0.37 ± 0.1 °C), indicating that temperature is likely not the key factor explaining the discriminative distribution of MOx-related methanotrophic Methylococcales within a single MV.

Occurrence of anaerobic methanotrophs

The predominance of sn-2-hydroxyarchaeol over archaeol and OTUs belonging to ANME-2 and ANME-3 clades in sediment depths (4 to 16 cm) indicates the presence of AOM-related methanotrophic communities in these sediments, in contrast to the reference site (Figs. 3–5). The overall 13C-depleted sn-2-hydroxyarchaeol in all MV420 sites indicated persistent occurrence of AOM in sediments containing sulfate. Based on 16S rRNA results, ANME-3 was more dominant at the BM and ST sites than the DM site. On the other hand, ANME-2 was most frequent at the ST site, indicating a preference for a niche with lower methane flux. Notably, the low abundances of isoprenoid GDGTs (GDGT-1, GDGT-2, and GDGT-3) and 16S rRNA gene sequencing results indicate that the ANME-1 contribution was negligible for AOM in the sediments of MV42030,33,60,61,62. It is worthwhile to note that all surface sediments including that of the reference site showed a dominant crenarchaeol contribution to the total isoprenoid GDGTs pool, indicating a potential contribution of marine Thaumarchaeota63. The enriched δ13C values of archaeol and BPs suggest that archaeal communities are mixed due to processes other than AOM (e.g., assimilation of AOM-derived inorganic carbon)64. These δ13C signatures are similar to methane-related archaeal distributions investigated at other MVs (at water depths of 290 m and 740 m) in the Canadian Beaufort Sea30. In this regard, niche stratifications among ANMEs appear to be related to differences in environmental preferences within the sampled habitats with varying methane fluxes53,54.

At the three investigated MV420 sites, specific archaeal lipids and 16S rRNA signatures clearly reveal the discriminative presence of ANME-2 and ANME-3 involved in AOM between the BM and ST sites (Figs. 3–5). In contrast to ANME-1, which was observed at a wide range of temperatures (4 to 70 °C) and typically lower methane fluxes65,66,67, ANME-2 and ANME-3 have been found in samples with temperatures below 20 °C, vigorously emitting MVs such as HMMV19,28,53,54 and other seepages52,66. Among them, ANME-2 (particularly ANME-2c) was most frequently found in sediments bioturbated by benthic communities such as vesicomyid clams inhabiting low-fluid flux regimes29,54,68,69. Considering that vesicomyid clams have so far not been found in the western Arctic, our results indicate a requirement of ANME-2 for lower methane flux. The overall dominance of ANME-3 at the BM site is consistent with previous findings, prevailing across settings in Arctic seeping sites19,28,53,54. Meanwhile, at the ST site, ANME-2 can be regarded as the additional methane oxidizer under environmental conditions such as bioturbation and lower methane seepage60,70,71. Given that the produced sulfide is quickly oxidized by thiotrophic bioturbators, the presence of ANME-2 below tubeworm patches may imply that they seem to form a syntrophic relationship with SRB during AOM19,54. To some degree, ANME-3 appears to tolerate high sulfide concentrations produced during active AOM processes72, although the sulfide sensitivity of ANMEs needs to be investigated further. AOM activities performed by both ANMEs seem to support the metabolism of sulfur-oxidizing endosymbionts inhabiting frenulata Oligobrachia haakonmosbiensis73,74. The presence of diverse methanotrophs at the ST site can therefore be explained by environmental conditions under steady replenishment of sulfate and oxygen levels into the sediments, in accordance with the development of ST patches.

Niche adaptation between ANMEs and SRBs

With respect to the consortium of ANMEs and SRBs, the presence of sulfate-reducing Deltaproteobacteria, including the partner SRB of ANMEs, was assessed by analyzing specific FAs and deltaproteobacterial OTUs (Desulfobacteraceae and Desulfobulbaceae) at the three MV420 sites. The 13C-depleted, SRB-specific lipid (C16:1ω5) probably derived from Desulfobacteraceae was relatively abundant near the sediment surface, while another specific FA (C17:1ω6) derived most likely from Desulfobulbaceae was not identified in any sediments60. Most deltaproteobacterial OTUs showed non-significant correlations with major OTUs of the ANME-2 and ANME-3 clade. Only a few deltaproteobacterial OTUs had a positive correlation with major ANMEs at the ST site. This suggests a weak coupling between ANMEs and Desulfobacteraceae at the MV420 sites, which is inconsistent with reported couplings of SRBs (Desulfosarcina/Desulfococcus and Desulfobulbus) and ANMEs (ANME-2 and ANME-3) at HMMV19,28. Desulfobacteraceae often occur as single cells in surface sediments of bacterial mat- or clam-covered seeps75,76 and decrease with increasing sediment depths75. Therefore, this SRB seems to have a metabolic advantage over other SRBs in oxygenated sediments. Accordingly, at MV420, the occurrence of overall ANMEs without partner SRBs may imply that ANMEs were either associated with other SRBs, occurred as loose aggregates or a single cell without direct contact with SRBs24, or performed AOM without a partner SRB77. More studies are necessary to further investigate ANME-SRB consortia by, for example, applying the technique of fluorescence in situ hybridization.

Summary

In this study, we provide the first evidence for MOx- and AOM-related methanotrophic communities in connection with chemosynthetic communities at an active MV in the Canadian Beaufort Sea. Our results revealed that niche diversification of methanotrophic communities (i.e., Methylococcales and ANME-2 and ANME-3 groups, respectively) may have been shaped by environmental factors such as availability of electron acceptors and bioirrigation activities by chemosynthetic organisms. In comparison with highly local diversification of methanotrophic communities at heterogenous seeping sites54, our finding confirms that the supply of electron acceptors is a key factor determining local diversification of methanotrophic communities. In particular, our study is one of the few that show both methanotrophs (i.e., Methylococcales and ANME-3) distributed separately under seeping fluids and high methane fluxes19,24,28,66. Our study of methanotrophic signals provides valuable information on the effective removal pathways involved in methane release from marine sediments.

Methods

Sample collection

Sediment samples were obtained during the 2017 ARA08C cruise of the Korean icebreaker RV Araon at the Beaufort MV (water depth of 420 m) from different chemosynthetic provinces (ARA08C-DIVE105-8 and -9: BO site, ARA08C-DIVE104-7 and-13: BM site, and ARA08C_DIVE105-12 and -14: ST site) using a push core operated by an ROV system and at the reference site (ARA08C-21BC, water depth of 420 m) using a box corer (Fig. 1). Upon recovery, push- and box-core samples were subsampled horizontally with 20 mL serum vials, which were then sealed immediately with butyl rubber stoppers to prevent gas exchange and stored at 4 °C. Porewater was extracted from holes drilled at intervals of 2 to 3 cm down the core liner using rhizon samplers connected to 20 mL syringes. Extracted porewater was filtered through a 0.20 μm disposable polytetrafluorethylene in-line filter and stored at 4 °C until analysis. After porewater extraction, core sediments were sliced into 1cm sections and subsampled for analysis of bulk elements, lipid biomarkers, and 16S rRNA gene sequences. Subsamples were stored at −80 °C until analysis.

Gas and porewater analyses

Headspace methane concentration was measured with a gas chromatograph (GC) (7890 A, Agilent Technologies, CA, USA) with flame ionization detector. The precision of repeated standard analyses exceeded 5%. Isotopic compositions of methane were obtained using an isotope ratio mass spectrometer (Finnigan MAT 252; Thermo Fisher Scientific, Waltham, MA, USA) with a Combustion III interface (Thermo Fisher Scientific) at Nagoya University. Isotopic values were expressed as δ-notation (per mil) relative to Vienna Pee-Dee Belemnite (VPDB). The δ13C values were calibrated using analyses of a methane standard with δ13C value of −35.2‰, as certified by the National Institute of Standards and Technology (NIST, USA). Precision for δ13CCH4 was greater than ±0.2‰.

Chloride (Cl–) and DIC were analyzed by titration using silver nitrate (AgNO3) and 0.1 N HCl, respectively. The precision determined by repeated titrations of International Association of Physical Sciences of the Oceans standard seawater was ±0.5% for Cl– and <2% for DIC. Sulfate was measured using ion chromatography (Metrohm 761). The δ13C values of DIC were obtained using an isotope ratio mass spectrometer (Finnigan DELTAplusXL; Thermo Fisher Scientific, Waltham, MA, USA) with a Finnigan GasBench-II headspace autosampler at Oregon State University (Corvallis, OR, USA). Isotopic values were expressed as δ-notation (per mil) relative to VPDB. The δ13C values were calibrated through analyses of a CaCO3 standard with δ13C value of –0.4‰, certified by Oregon State University. The precision for δ13CDIC was greater than ±0.02‰.

Calculation of methane and sulfate fluxes

Assuming that related flux over the linear range of the sulfate profile was proportional to methane flux at a particular location78, we determined an AOM depth where the linear sulfate and methane concentration gradients overlapped in each core sediment. Diffusive fluxes of solutes (methane and sulfate) into the AOM layers were determined using Fick’s first law, and the slope of methane and sulfate concentration was plotted against depth over a linear range of data79,80 using the equation

$$J=-{\rm{\phi }}{D}_{s}\frac{\partial {\rm{C}}}{\partial {\rm{z}}}$$

where J represents the diffusive methane and sulfate fluxes (mmol m–2 yr–1), φ is the sediment porosity, Ds is the sediment diffusion coefficient, C is the range of methane and sulfate concentrations, and z is the range of depth for the linear section of the methane and sulfate porewater profiles. Ds values were calculated from free diffusion coefficients in seawater (D0, cm2 yr–1) by correcting for sediment tortuosity81:

$${D}_{s}=\frac{{D}_{0}}{1-\,\mathrm{ln}({\phi }^{2})}$$

Downward sulfate diffusion into the sediment is expressed as negative, and upward methane flux out of the sediment is expressed as positive.

Bulk element analysis

Sediment samples were freeze-dried and homogenized using an agate mortar prior to bulk geochemical analyses. For TOC analysis, sediment samples (~1 g) were treated with 8 mL of 1 N HCl to remove carbonates. The TOC content and its isotopic composition were measured using an elemental analyzer (EuroEA3028, Eurovector, Milan, Italy) connected to an isotope ratio mass spectrometer (Isoprime, GV instruments, Manchester, UK) according to Lee et al. (2018). Isotopic values are expressed as δ13C values in per mil relative to VPDB. The δ13C values were calibrated through a CH3 standard with a δ13C value of −24.7‰, certified by the International Atomic Energy Agency (IAEA, Austria). The precision for TOC and δ13CTOC was better than ±0.2 wt.%, and ±0.1‰, respectively. The total sulfur (TS) content and its isotopic composition were measured using an elemental analyzer (EA1110, Thermo) connected to an isotope ratio mass spectrometer (Dual pumped 20–20 S, Secon). Isotopic values are expressed as δ34S values in per mil relative to Vienna Canyon Diablo Troilite (VCDT). The δ34S values were calibrated using S-2 and S-3 standards with δ34S values of 22.7‰ and −32.3‰, respectively, certified by the IAEA. The precision for TS and δ34S was better than ±0.4 wt.% and ±0.4‰, respectively.

Lipid biomarker analysis

Lipid extraction and purification for different compound classes and analyses by gas chromatography, gas chromatography–mass spectrometry, gas chromatography–combustion–isotope ratio mass spectrometry, and high-performance liquid chromatography–atmospheric pressure positive–ion chemical ionization–mass spectrometry were conducted according to previously reported methods30,61,82. Double-bond positions in fatty acid methyl esters were determined by analysis as their dimethyl disulfide adducts according to the method of Nichols et al. (1986). All isotopic values were reported using δ-notation (per mil) with respect to VPDB. The precision for δ13C values was greater than ±0.5‰, corrected for introduction of additional carbon atoms during sylilation and methylation.

Genomic DNA extraction, amplification, and sequencing

Genomic DNA was extracted using an Exgene Soil SV mini kit (Cambio, UK) at GeneAll (Seoul, South Korea). The V4-V5 regions of bacterial 16S rRNA genes and V6-V8 regions of archaeal 16S rRNA genes were amplified by polymerase chain reaction using primer pairs 515F/926R and A956F/A1401R, respectively, at the Integrated Microbiome Resource (IMR) at Dalhousie University, Canada (http://cgeb-imr.ca)83. Sequencing of the amplicons was carried out at IMR using the paired-end (2 × 300 bp) Illumina MiSeq system (Illumina, USA).

16S rRNA gene sequences processing

Pair end sequences were trimmed based on Sickle84 quality scores followed by BayesHammer error corrections85. The resulting quality-trimmed and error-corrected paired end sequences were assembled using PANDAseq.86. Further sequence-processing steps were performed according to the mothur Pipeline package87. The assembled bacterial sequences were aligned against a SILVA alignment (http://www.arb-silva.de), and archaeal sequences were aligned against a SINA alignment88 and subsequently denoised using the “pre.cluster” command. Chimeric sequences were removed using the “chimera.uchime” command in de novo mode89. The sequences were further clustered to OTUs at a 97% sequence similarity level using the opticlust clustering algorithm. Taxonomic assignments of each OTU were determined by sequence similarity searches against the EzBioCloud database90. All sequence data used in this study were deposited in the Sequence Read Archive at the National Center for Biotechnology Information under the accession number PRJNA559003.

Phylogenetic analysis

Phylogenetic trees of major archaeal OTUs of Methanomicrobia, bacterial OTUs of Deltaproteobacteria and Methylococcales of Gammaprotoebacteria with greater than 5% relative abundance (Supplementary Information Sequence 1–3) were constructed using the maximum-likelihood algorithm91 with the GTR evolutionary model using IQ-TREE92.