Main

Chemosynthesis-based communities represent a remarkable example of life’s ability to adapt and thrive in some of the most extreme conditions on Earth5,6,7,8. Since the initial discovery of these communities at hydrothermal vents9, the detection of chemosynthesis-based communities at cold seeps, supporting highly diverse and abundant chemosymbiotic biota1,3,4,5,6,7,8, has increased and expanded our understanding of deep-sea ecological systems and the biogeochemical processes that underpin them. The communities, typically dominated by Bivalvia and Siboglinidae, and sustained by microbial chemosynthesis, are confined to areas where fluids rich in hydrogen sulfide and/or methane are released through geological fractures6,7.

Despite being found on passive and active margins across a wide range of ocean depths5,6,7,8, cold-seep communities have remained largely unexplored in hadal trenches with depths exceeding 6,000 m. Only two small seep communities at hadal depths have been discovered by submersible and remotely operated vehicle, one dominated by vesicomyid clams at nearly 6,437 m depth and another dominated by thyasirid clams at 7,326–7,434 m in the Japan Trench3,4,10. In addition, probable chemosynthetic mats have been observed at a depth of 10,677 m at the bottom of the Mariana Trench11; however, they are not associated with any seep animal communities. The biogeographical distribution of chemosynthesis-based communities in the deepest parts of the world’s oceans, their potential roles in shaping the deep-sea ecosystem and the mechanisms behind the formation of methane seeps at such great depths remain elusive.

The Kuril–Kamchatka Trench and the Aleutian Trench are formed by the tectonic interactions between the Pacific Plate and the North American Plate, and meet at the Kamchatka Aleutian Transition connection12. This area is geologically highly active with many seismically and volcanically active sites. The Kuril–Kamchatka Trench is formed by the northwestward subduction of the Pacific Plate beneath the Okhotsk Plate and extends around 2,100 km from Hokkaido in the south to Kamchatka Peninsula in the north13. The maximum depth of the trench bottom reaches 9,578 m, based on our submersible conductivity–temperature–depth data. Previous research conducted by expeditions aboard research vessel (RV) Vityaz and RV Sonne in the Kuril–Kamchatka Trench shows that the trench is characterized by predominantly heterotrophic benthic fauna14,15. However, some specimens of chemosymbiotrophic taxa15,16, in particular frenulate siboglinids, were also recovered using a trawl, indicating that methane seeps may exist on the trench bottom. The Aleutian Trench is about 2,900 km long, extends from the Alaska and Kenai peninsulas in North America to Kamchatka13 and marks a boundary where the Pacific Plate has been subducting northwestwards beneath the Bering Sea Plate. Both trenches underlie a highly productive boreal region of the North Pacific Ocean and are characterized by distinct spring blooms and high annual primary production17.

Chemosynthetic community distribution and diversity

Investigations of the trench bottom of the Kuril–Kamchatka Trench and the western Aleutian Trench from 8 July to 17 August 2024, using RV Tansuoyihao with the full-ocean-depth manned submersible Fendouzhe capable of reaching the deepest part of the ocean at nearly 11,000 m, resulted in the discovery of widespread cold-seep chemosynthesis-based communities in both trenches (Fig. 1). During dive FDZ 271 in the Kuril–Kamchatka Trench, we first encountered dense chemosynthetic communities dominated by frenulate siboglinids at a depth of 9,533 m, situated atop black muds at the boundary between the trench bottom and the basement of the accretionary prism. This boundary corresponds to the outcrop of a geological normal fault, a feature formed by bending of the incoming plate, as indicated by regional seismic data18. We have designated this seep site as The Deepest because of its unparalleled depth, marking it as the deepest known seepage location discovered so far (Fig. 2a). Subsequent to this initial discovery, we executed a comprehensive series of 23 dives in geologically analogous settings to determine the spatial distribution, scale and biodiversity of the chemosynthetic communities inhabiting the trench floors. During 19 of these dives (Supplementary Table 1), we observed, recorded and sampled chemosynthesis-based communities, revealing their proliferation in a distinct zone along the basement of the accretionary prism. This zone stretches 2,500 km along the bottom of the trenches, representing a notable ecological feature of these two hadal trenches that has not been recognized or reported despite previous sediment and fauna investigations in these areas.

Fig. 1: Map showing the Kuril–Kamchatka Trench and western Aleutian Trench.
figure 1

The study area, situated in the northwest Pacific, is demarcated by a white rectangle in the inset. Orange dots represent dive sites where chemosynthesis-based communities were observed and sampled and crosses indicate dive sites lacking such communities. Open orange circles delineate potential seep sites characterized by black sediments. White arrows illustrate the direction of subduction for the Pacific Plate beneath the Okhotsk Plate and the Bering Sea Plate. The dashed white lines indicate the transitional connection zones between the Kuril–Kamchatka Trench and the Aleutian Trench. Bathymetric data were acquired using the KM-EM122 multi-beam bathymetric system during the research expedition. Scale bar, 200 km. Credit: map created using Global Mapper 14 software, with background data sourced from GeoMapApp (http://www.geomapapp.org), under a CC BY 4.0 licence.

Full size image
Fig. 2: Representative fauna of cold-seep sites in the Kuril–Kamchatka Trench and western Aleutian Trench.
figure 2

a, Free-moving polychaetes Macellicephaloides grandicirra (white; reaching 6.5 cm in size) navigate among dense colonies of frenulate siboglinids, with tubes 20–30 cm in length and approximately 1 mm in diameter, at 9,532 m at The Deepest. b, Clusters of frenulate siboglinids extending red haemoglobin-filled tentacles, with small Gastropoda (white spots) on tops of the tubes near the tentacles, at 9,320 m at Wintersweet Valley. c, Tightly packed frenulate siboglinids are home to abundant free-moving polychaetes M. grandicirra (white) at 9,332 m at Cotton Field. d, Dense aggregation of vesicomyid bivalves A. phaseoliformis (reaching 23 cm in size) in the sediment, with approximately 6–8 cm of valves exposed and often hosting Actiniaria, at 5,743 m at Clam Bed. e, Tube-dwelling polychaetes Anobothrus sp. and Actiniaria are dominant at 6,870 m at Aleutian Deepest, with spots of white microbial mats. f, Dense aggregation of vesicomyid bivalves I. fossajaponicum (reaching 3 cm in size) associated with black sediments and accompanied by tube-dwelling polychaetes Anobothrus sp. at 6,928 m at Aleutian Deepest. g, Dark blue muds surrounded by clusters of frenulate siboglinids, mark methane seeps at 6,800 m at Blue Marsh. h, Large patches of white, snow-like microbial mats stretch tens of metres, accompanied by frenulate siboglinid tubeworms at 6,700 m at Icy River. The images were taken by the manned submersible equipped with a high-definition camera system. The name of each cold seep indicated in the lower left corner. The distance between laser beams is 10 cm. An expanded showcase of cold-seep fauna is given in Supplementary Video 1.

Full size image

The species composition and structure of chemosynthesis-based communities in the Kuril–Kamchatka Trench, with depths ranging from nearly 7,000 m to 9,533 m, is distinct from those observed in the Kamchatka Aleutian Transition and the western Aleutian Trench where depths are predominantly less than 7,000 m. In the Kuril–Kamchatka Trench, most communities are dominated by frenulate siboglinids (Extended Data Fig. 1). The composition of siboglinid species among communities in the Kuril–Kamchatka Trench seems to differ considerably. The most abundant species are representatives of Lamellisabella, Polybrachia, Spirobrachia and Zenkevitchiana. A notable discovery in the central Kuril–Kamchatka Trench at a depth of 9,120 m (dive FDZ 274) is the Wintersweet Valley site, which is predominantly inhabited by two species of frenulate siboglinids (Lamellisabella and Polybrachia) (Fig. 2b). These frenulate siboglinids were found in great abundance along a substantial portion of a seep field, around 2 km in length, which was explored during a single dive. The colonies consist of thousands of individuals with tubes extending out of the sediment. Associated with the frenulate siboglinids is a suite of species obligate to this particular habitat, including tube-dwelling polychaetes Terebelliformia and numerous Gastropoda that settled on the siboglinid tubes, as well as a variety of heterotrophic benthic fauna including free-moving polychaetes Macellicephaloides, crinoids Bathycrinus, holothurians Elpidia hanseni and amphipods сf. Princaxelia.

In addition, two other large cold-seep fields, the Dead Valley and Cotton Field sites, were observed at depths of 9,522 and 9,566 m during dive FDZ 275. These fields are located roughly 120 km southwest of Wintersweet Field and extend at least 2.2 km along the 9,520 m contour line. In Dead Valley, dense clusters of frenulate siboglinids occur, with tubes covered by white flocculent material and lying almost horizontally (Extended Data Fig. 2). Siboglinids seem to be dead, suggesting cessation of fluid activity in this part of the field. By contrast, Cotton Field (Fig. 2c), situated 50 m southwest of Dead Valley, supports dense populations of living frenulate siboglinids in association with numerous free-moving polychaetes (Macellicephalinae) and Gastropoda. It is noteworthy that frenulate siboglinids were encountered in nine of eleven dives conducted at the base of the accretionary prisms in the Kuril–Kamchatka Trench, albeit with varying densities of occurrence. During dive FDZ 278, frenulate siboglinids with a relatively high abundance of thyasirid bivalves (Tartarothyasira cf. hadalis) and tube-dwelling polychaetes (Anobothrus sp.) were discovered at a depth of 8,764 m. Tartarothyasira hadalis (initially described as Maorithyas hadalis10) has been documented in the hadal zone of the Japan Trench at a depth of 7,326–7,434 m (refs. 3,11). The presence of these chemosymbiotrophic bivalves in the Kuril–Kamchatka Trench marks the deepest occurrence of such organisms ever discovered in hadal zones.

In contrast to the Kuril–Kamchatka Trench, the chemosynthesis-based communities located in the Kamchatka Aleutian Transition are characterized by a high abundance of two species of chemosymbiotrophic vesicomyid clams Abyssogena phaseoliformis and Isorropodon fossajaponicum (Extended Data Table 1 and Extended Data Fig. 3). Fields of siboglinids and clusters of bivalves in the Kamchatka Aleutian Transition were located next to each other, but did not overlap. During dive FDZ 284, dense aggregations of bivalves were initially identified in muddy sediments at a depth of 5,988 m. The communities in this field are largely dominated by I. fossajaponicum and tube-dwelling ampharetid polychaetes (Anobothrus sp.). On the subsequent dive, FDZ 297, an expansive seep field termed Clam Bed (Fig. 2d) was discovered, extending roughly 2 km along a steep fault at a depth of 5,800 m. The dominant species in this community is A. phaseoliformis. In addition to clams, the seep field supports a diverse array of fauna, including gastropods Peltospiridae and Provanidae, tube-dwelling polychaetes Ampharetidae and Maldanidae, free-moving polychaetes Polynoidae and Dorvilleidae, amphipods cf. Abyssododecas and actiniarians Sagartiidae. Similar A. phaseoliformis dominated communities have previously been described for the Japan Trench and the eastern Aleutian Trench19,20.

The cold-seep communities in the western Aleutian Trench are dominated by Bivalvia, Siboglinidae or Ampharetidae. During dives FDZ 285 and FDZ 296, we documented the presence of thyasirid (Tartarothyasira cf. hadalis and Axinus sp.) and vesicomyid (I. fossajaponicum) bivalve aggregations (Extended Data Table 1), as well as tube-dwelling polychaetes Anobothrus sp. (Fig. 2e,f), distributed discontinuously along a 5-km stretch of a subducting normal fault between depths of 6,900 and 6,930 m. In addition, during dive FDZ 286, a large field of Tartarothyasira cf. hadalis more than 500 m across was observed at a depth of 6,756 m (Extended Data Fig. 2). Frenulate siboglinids are primarily located around the periphery of areas with aggregations of bivalves. During dives FDZ 294 and FDZ 295, a seep field designated as Blue Marsh (Fig. 2g) extending over at least 2 km, was discovered at a depth of 6,635 m. The central regions of these seeps are commonly dominated by tube-dwelling Ampharetidae, whereas the margins are predominantly inhabited by frenulate siboglinids, Polybrachia, Spirobrachia and Zenkevitchiana. Furthermore, a seep field named Icy River (Fig. 2h) was noted east of the Blue Marsh field at a depth of 6,630 m. This field is characterized by white microbial mats surrounded by frenulate siboglonids. In comparison with the Kuril–Kamchatka Trench, the abundance of bivalves and tube-dwelling polychaetes in the western Aleutian Trench is notably higher.

The seep communities identified in this study, akin to those documented in the hadal zone and at shallower depths3,4,20,21,22, are distinguished by their exceptionally high abundance and density of specialized species. We found maximum densities of up to 5,813 ± 1,335 (mean ± s.d.) Siboglinidae and 293 ± 69 Bivalvia per square metre in two trenches (Supplementary Table 1). The composition and structure of the investigated communities exhibit a marked patchiness across various spatial scales. This variability is observed not only in individual trenches and across depth ranges, but also between different trenches. Furthermore, certain species are distributed across a broad latitudinal and bathymetric gradient. The consistent presence of the same chemosymbiotrophic species across the Japan, Kuril–Kamchatka and Aleutian trenches suggests the presence of a connected system of reducing habitats in the hadal zone of the northern Pacific. Connections among the trench-hosted seep communities may extend even farther south to the Japan and Mariana trenches, as suggested by the presence of similar looking clam taxa20,21,22.

Pore-water and carbonate geochemistry

The molecular and isotopic compositions of gases associated with cold seeps in the seep sediments provide compelling evidence for a microbial origin of methane (Extended Data Table 2). Analysis of hydrocarbon in head gases and sediments of pushcores showed that methane constitutes 100% of the total hydrocarbons, suggesting a microbial origin for the methane23,24. Analysis of carbon (C) and deuterium (D) stable isotope showed that methane samples exhibited δ13CVPDB (Vienna Pee Dee Belemnite) values ranging from −78.1‰ to −95.7‰ and δDVSMOW (Vienna Standard Mean Ocean Water) values ranging from −142.2‰ to −188.8‰. These findings further indicate that the methane is derived from microbial carbonate reduction rather than from methyl-based methanogenesis25,26,27,28 (Fig. 3a). These geochemical data collectively indicate that the methane present in the seep sediments is the result of microbial reduction of CO2 derived from sedimentary organic matter.

Fig. 3: Origins and phases of methane in hadal cold seeps.
figure 3

a, δ13CVPDB and δDVSMOW diagram classifying the source of methane. The base diagram is adapted from previous studies25,26,28, which used field measurements of stable carbon and hydrogen isotopes of methane (CH4) from diverse sedimentary environments to distinguish between different types of microbial and thermogenic methanogenesis. Yellow dots in the CO2 reduction (CR) category represent data from the methane samples analysed in this study. b, Schematic phase diagram of methane hydrate–sea water system. Solid black lines represent the phase boundaries between H+I – H+L and H+L – L+V ; the shaded (grey blue) H+L phase zone and grey lines therein indicate the methane hydrate solubilities in sea water (mCH4, in ppm). Red diamonds represent the temperature and pressure conditions at stations where seeps are present. H + I, methane hydrate + ice; H + L, methane hydrate + liquid; L + V, liquid + vapour; M, methyl-based methanogenesis.

Full size image

The sediments associated with cold seeps are typically dark grey to black, coloration attributed to the high iron sulfide content present in the sediment matrix. Upon retrieval, sediment cores from these seeps exhibited a distinct hydrogen sulfide odour. The vertical profiles of methane, ammonia \(({{\rm{NH}}}_{4}^{+})\), dissolved organic carbon, dissolved inorganic carbon (DIC), δ13C-DIC, sulfate \(({{\rm{SO}}}_{4}^{2-})\), hydrogen sulfide and salinity in two pushcores showed typical distribution patterns associated with cold-seep environments (Extended Data Fig. 4). These patterns indicate the processes of anaerobic methane oxidation coupled with sulfate reduction5,29,30, organic matter diagenesis5,30 and hydrate decomposition31. In addition, metastable hexahydrate calcium carbonate (ikaite, CaCO3·6H2O), with clusters of euhedral spear-like crystals (Extended Data Fig. 5), are found in association with the black sediments of cold seeps. Ikaite δ13C values range from −17.29‰ to −26.89‰ (Extended Data Table 3), suggesting that they are probably authigenic precipitates from \({{\rm{CO}}}_{3}^{2-}\) supplied from the early diagenetic decomposition of sedimentary organic matter32, preceding microbial methane oxidation.

The modelling of methane phases under hadal zone conditions (Fig. 3b) indicates that at all observed seep sites at depths ranging from 5,662 m to 9,533 m, methane occurs in a dissolved form in pore water and as hydrate, with the absence of a vapour phase. This finding aligns with in situ observations conducted by submersible, which have noted the absence of bubbling and any associated gas phase during dives. Methane concentrations measured in the headspace of sediment pushcore reached 118,882 ppm, substantially exceeding the methane solubility in equilibrium with hydrate at these depths (545 ppm at 9,500 m). Calculation of the hydrate stability zone (Extended Data Fig. 6) indicates the potential presence of hydrates at these profound depths in sedimentary layers. A focused coring and drilling campaign is needed to revisit these seeps to validate this hypothesis.

Formation of hadal cold seeps

The formation processes of cold seeps on trench bottoms differ markedly from those located on accretionary prisms at shallower depths. Cold seeps on accretionary prisms are characterized by the ejection of fluids through thrust faults33,34,35, with these fluids originating from subducted sediments and migrating upwards over distances often exceeding 1,000 m (ref. 36). By contrast, this study indicates that the fluids associated with trench-bottom seeps are sourced from trench sediments that have not been subducted. This hypothesis is bolstered by the composition of the fluids, primarily biogenic methane—typically produced through microbial carbonate reduction a few to several hundred metres beneath the seabed37. We propose that these fluids from the deep sediment layers of the trench migrate upwards along bending-related faults or other major fractures that occur at the leading edge of the accretionary prisms.

Figure 4 illustrates a geological model of cold-seep formation in the Kuril–Kamchatka and Aleutian trenches. In these trenches, the V-shaped topography serves as a natural trap, facilitating the rapid accumulation of substantial quantities of organic matter on the trench bottom38,39. This organic matter is introduced from the highly productive ocean surface where phytoplanktonic blooms frequently occur and from the trench slope by means of gravity flows triggered by earthquakes and landslides40,41. Under the anoxic conditions prevalent in the deep sediment layers, microbial reduction of carbon dioxide derived from sedimentary organic material can sustain methanogenesis25,26,27,28. We argue that, over time, this methane accumulates beneath impermeable sediment layers in the form of dissolved methane fluid and methane hydrate. The compression forces associated with incoming plate subduction42 then facilitate the lateral migration of methane fluids in the deep sediment layers at the trench bottom, directing them towards the accretionary prism. Ultimately, methane fluids migrate upwards through large fault zones (for instance, bending-related normal faults) that occur at the basement of the accretionary prism. Upon reaching the sea floor, methane fluids escape as a hadal cold seep and support the observed chemosynthesis-based ecosystems.

Fig. 4: Formation of hadal trench cold seeps.
figure 4

a, Schematic geological model presenting the cross-section view of the subducting plate and the overriding plate along the trench, as indicated by seismic survey data18 from these areas. The light green arrows depict the migration of organic matter into the trench, encompassing both downward and lateral movements. The white arrow denotes the direction of subduction, and the dashed line signifies the trench axis, which is nearly parallel to the striped zone of cold seeps. Black triangles point to the trench’s axis. Note the prevalence of normal faults developed in the subducting plate. b, Detailed view of the area outlined by the black rectangle in a showing the formation of gas hydrates in deep sediment layers. Methane-rich fluids migrate horizontally towards the accretionary wedge as a result of the compression forces associated with subduction. Normal faults at the leading edge of the accretionary wedge create a pathway that facilitates the upward migration of seep fluids. Figure created using previously published location data18 of subducting bending faults.

Full size image

Implications

Although hadal seeps have been documented before, the depths probed here, coupled with the thriving communities discovered and distribution ranges observed, significantly expand the known habitat, depth and biogeographical distributions for a great many species. Moreover, studies of organisms thriving in these communities can provide clues about physiological boundaries, adaptive strategies and previously unknown animal–microbe interactions shaped by high-pressure conditions1. Our findings also challenge the traditional perspective on the energy sources sustaining hadal fauna, which were predominantly believed to be derived from surface-derived particulate organic matter and carrion fall43. The widespread occurrence of chemosynthesis-based communities suggests an underappreciated role of chemical energy in shaping the hadal ecosystem. The co-existence of chemosynthesis-based organisms with a substantial number of heterotrophic benthic fauna, such as Actiniaria, Echiura, Holothuroidea and Amphipoda (Extended Data Fig. 7), suggests that methane seep production in the trench extends its influence to the broader benthic community44. If such a trophic subsidy for heterotrophic benthic fauna is demonstrated and quantified by follow-up research, this study would have served to highlight a previously unrecognized contribution of chemosynthetic processes to the overall functioning of hadal ecosystems.

The detection of anomalously high methane concentrations and the potential for gas hydrate formation in the hadal zone provide new insights into deep carbon cycling. The widespread methane-rich environments in two hadal trenches, where microbial reduction of CO2 from sedimentary organic matter presumably results in methane production, suggest a vibrant and active microbial community in the hadal sediments. This indicates that the deep-subsurface biosphere may exert a more important influence on biogeochemical processes in subduction zones, representing a previously unrecognized energy supply45,46. The accumulation of methane in sedimentary layers generated by this deep-subsurface biosphere could potentially sequester considerable amounts of sedimentary organic carbon, suggesting a portion of subducting organic carbon can be stored in the trench sediments in a form of methane for prolonged geological time, rather than being subducted to the deep lithosphere. It remains unknown whether the current findings can be extrapolated to other trench systems, but given the geological similarities hadal methane reservoirs may be more widespread, irrespective of the presence of fault zones that could serve as conduits for the release of methane-rich fluids. This hypothesis is supported by the recovery of gas hydrates from drilling sediments in the Middle America Trench47 and the Peru–Chile Trench at depths surpassing 5,000 m (ref. 48), and the presence of similar seep communities in the Japan Trench3,4. These findings underscore the complex nature of carbon cycling in the deep sea and highlight the critical need to integrate hadal processes into global carbon models49,50 to improve the accuracy of predictions about carbon dynamics and climate change responses on geological timescale. Furthermore, the potential presence of methane hydrates at great depths in hadal trenches may enhance global inventory of methane gas hydrate resources.

Methods

The investigation was carried out during TS42 cruise between 7 July and 18 August 2024 by RV Tan Suo Yi Hao with the full-ocean-depth human-occupied vehicle Fendouzhe, which was fitted with hydraulically powered manipulators on two swing arms. Under the guidance of operators in the human-occupied vehicle, the arms efficiently acquired the samples and stored them safely in a biological box and a geological box of the vehicle.

Processing of benthic fauna and sea floor video footage

Upon retrieval of the submersible, all collected specimens were promptly transferred from the biological collection box and slurp sampler to the shipboard laboratory. The specimens were then sorted into main taxonomic groups of different levels using visual inspection or under stereomicroscopes. Each organism was counted and preserved in pre-cooled, non-denatured 95% ethanol or in a 4% buffered formaldehyde solution depending on the taxon. Following initial preservation, certain taxonomic groups were further transferred to 70% ethanol for long-term storage.

Visual assessment of species identification, density and spatial structure of the macro-epifauna and mega-epifauna of the seep communities was carried out on the basis of the analysis of video footage recorded by two  high-definition cameras mounted on the human-occupied vehicle. For each dive, between three and ten representative screenshots showing the densest cold-seep communities were selected from the video footage. The area of each image was estimated using the submersible’s laser scale, which projects two parallel laser points 10 cm apart onto the sea floor. This provided a reliable spatial reference for calculating the area of the sea floor captured in each image. A standardized quadrat (for example, 50 × 50 cm) was drawn near the laser dots by using this laser scale as a reference in the image. The calculated area was then converted into square metres for standardized density calculations. Animals visible in each quadrat were manually counted, and faunal density was expressed as the number of individuals per square metre. For each dive, density values from the selected images were used to calculate the mean faunal density, and the standard deviation was computed to quantify the variability among replicate images.

Phylogenetic analyses of the coxI gene sequences

Up to 0.5 cm3 of fauna tissue was cut into tiny pieces and subjected to DNA extraction using the PowerSoil DNA Isolation kit (MoBio Laboratories). The extracted DNA was quantified using a Qubit dsDNA HS Assay Kit with Qubit 2.0 fluorometer (Invitrogen). A metagenomic library was constructed using the VAHTS Universal DNA Library Prep Kit for Illumina v.4 and sequenced on the NovaSeq X Plus platform (Illumina) to generate 2 × 150 bp pair-ended reads. Raw sequencing reads were qualified using Fastp v.0.23.2 and assembly into contigs using MEGAHIT v.1.2.9. The coxI gene encoding cytochrome c oxidase subunit I of the fauna was retrieved from metagenomic contigs. The retrieved coxI was searched against the National Center for Biotechnology Information GenBank database for preliminary taxonomy identification.

Gas and pore-water sampling

During each dive, 6–12 sediment pushcores were collected using the manipulators of the submersible. Upon recovery, these pushcores were immediately transported to the ship’s cold room, which is kept at around 4 °C, to facilitate subsequent processing. Among them, one or two pushcores were allocated for gas concentration analysis onboard, with only subsamples exhibiting high methane concentrations being prepared for further carbon and deuterium isotope analysis. In addition, one or two pushcores were used to extract pore water samples for geochemical analysis.

Pore-water sampling was conducted using Rhizon samplers (as described in ref. 51). These samplers were inserted into the cores at 2-cm intervals and connected to 50-ml evacuated disposable syringes fitted with three-way Luer-lock stopcocks. The first approximately 1 ml of extracted pore water was discarded to remove any contaminants. Subsequently, around 15 ml of pore water was collected within a 2-h time frame. One portion of the pore water was preserved with a 20% zinc acetate solution for subsequent hydrogen sulfide analysis. The remaining pore water samples were transferred to 15-ml centrifuge tubes and frozen for subsequent ion analysis.

For the analysis of gas composition and isotopes, sediment samples were collected using a 2.5-ml cutoff plastic syringe, which was inserted through pre-drilled holes in the pushcore tube at depth intervals of 4 cm. A 2.5-ml sediment sample was then transferred to a 20-ml gas-tight glass vial, which was filled with 2 M NaOH solution and sealed with a crimp cap containing butyl rubber stoppers. The vials were vigorously shaken and stored in an upside-down position at 4 °C until analysis, which was conducted onboard in a single day. Before analysis, the vials were shaken again, and 2 ml of the NaOH solution was replaced with helium gas to create a headspace. The headspace gas from the push core was also directly extracted using a 20-ml syringe equipped with a three-way stopcock and was immediately transferred to a 12-ml vacuum Labco vial for further analysis.

Gas concentration and isotope

The concentration of dissolved gases was determined using a gas chromatograph (Trace GC1310; Thermo Scientific) installed onboard the research vessel. Headspace volumes ranging from 100 to 500 μl were sampled and injected into a gas chromatograph equipped with a flame ionization detector. The analytical precision achieved in these measurements is consistently lower than 5%.

The δ13C and δD isotopic compositions of methane were analysed using a gas chromatography–isotope ratio mass spectrometer system, which consisted of a Trace GC1310 connected to a Delta V Advantage Isotope Ratio MS (Thermo Scientific). The analysis was conducted at the Institute of Deep-Sea Science and Engineering (IDSSE), Chinese Academy of Sciences.

Methane was selectively separated from the gas matrix using a gas chromatography column (27 m × 0.3  mm × 20.00 μm; PoraPLOT Q). The separated methane was then combusted in a combustion oven at a temperature of 1,000 °C, which was interfaced with the isotope ratio mass spectrometer (IRMS) for subsequent analysis. The resulting CO2 was directly introduced into the IRMS for measurement.

The precision of the repeated analyses, expressed as the standard deviation (1σ), was ±0.5‰ for δ13C and ±2‰ for δD. The isotopic compositions of individual carbon compounds were reported as δ-values (‰) relative to the international standards Vienna Peedee Belemnite (VPDB) for δ13C and Vienna Standard Mean Ocean Water (VSMOW) for δD.

Pore-water geochemistry

Hydrogen sulfide concentrations were determined colorimetrically using the methylene blue method (with a limit of detection of approximately 0.5 μM). The concentration of \({{\rm{SO}}}_{4}^{2-}\) in pore water was quantified by ion chromatography using a Dionex ICS-900 system, which was equipped with an AS50 AutoSampler. To ensure that the \({{\rm{SO}}}_{4}^{2-}\) concentrations fell in the optimal analytical range for the ion chromatograph, the anion samples were diluted 70-fold with Milli-Q water. The analytical precision for the determination of \({{\rm{SO}}}_{4}^{2-}\) was ±3.0%. \({{\rm{NH}}}_{4}^{+}\) concentrations were measured using a fluorescence spectrometer (LS55, PE) following the procedure reported in ref. 52, with a relative deviation of 0.5%. DIC concentrations and δ13C-DIC values were analysed using a Gas Bench II IRMS at IDSSE. The samples were pretreated with an acid solution on the Gas Bench II; the resulting carbon dioxide, carried by helium, was separated by a constant-temperature chromatographic column and subsequently analysed for isotope abundance using a MAT253 gas stable IRMS. The analytical precision for DIC was ±0.15% and that for δ13C-DIC was ±0.5‰. Dissolved organic carbon concentrations were measured by a high-temperature catalytic oxidation method using a Multi N/C 3100 (CLD) analyser at IDSSE. Samples were thawed at room temperature and acidified to pH 2 with 2 mol l−1 hydrochloric acid before injection. Salinity was measured in the laboratory using a handheld digital salinity meter (ATAGO PAL-SALT) after the samples were thawed to room temperature.

Methane phase modelling

The phase boundaries and methane hydrate solubility in the methane hydrate–sea water system were calculated using thermodynamic models53,54. The chemical potential of the hydrate phase was determined through the application of the Van der Waals–Platteeuw hydrate model, in conjunction with angle-dependent ab initio intermolecular potentials, as previously described in ref. 55. The Gibbs–Thomson equation, incorporating appropriately parameterized hydrate–water interface values, was used to account for the capillary effects of porous sediments on the hydrate–liquid–vapour equilibrium and the hydrate–liquid equilibrium. The influence of surface textures and mineral components was neglected in this analysis.

The activity coefficients for water and methane in the methane–sea water system were calculated using the Pitzer model. By applying the Poynting correction to the fugacity of methane dissolved in aqueous solution at the hydrate–liquid equilibrium, where methane gas is absent, the extended model for three-phase equilibrium was adapted to predict the solubility of methane in aqueous solution at the hydrate–liquid equilibrium.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.