Hadal trenches form due to downward bending of subducting ocean crust along subduction zones and represent the least-known ultra-deep oceanic environment1. Earthquakes, the frequent tectonic events at convergent plate boundaries, trigger redistribution of sediments that contain organic carbon (OC)2,3,4,5,6,7 from the continental and trench slopes8 into the trench axis. Sediments are efficiently transported into and focused at the trenches due to their V-shaped morphology, resulting in rapid accumulation of sedimentary organic carbon (SOC) in the trenches that is ~70 times higher than the global deep sea average2,6,9.

Active OC cycling in hadal trench sediments may be triggered by the high-volume material inputs and intensive microbial activities10,11,12. SOC is remineralized to dissolved organic and inorganic carbon (DOC, DIC) in the sediment interstitial water (IW). In addition, the sulfate-methane transition zone (SMTZ) constrains the preservation and production of biogenic methane13,14,15, serving as an OC diagenetic front in marine sediments16. The production and consumption of dissolved carbon in the hadal trench sediments are key processes in the deep subseafloor carbon cycle, supporting deep biosphere metabolisms. As carbon in trench sediments eventually enters the subduction zone system during plate convergence1, the transformations of SOC, DOC, DIC, and methane may influence the carbon exchange among the Earth’s surface biosphere, the lithosphere, and even the mantle17.

In the Japan Trench, pioneering studies documented SOC delivery to the trench by historical subduction zone mega-earthquakes2,6,18 with various magnitudes and recurrences on centennial-millennial time scales4, however, their impacts on the cycling of the dissolved carbon in trench sediments remain largely unexplored. Here we present the IW DOC and DIC concentrations, carbon isotopic profiles (δ13C and 14C ages) and headspace methane concentrations in combination of typical IW geochemical parameters from 12 sites along the Japan Trench axis (water depth >7.5 km, Fig. 1a) collected during the International Ocean Discovery Program (IODP) Expedition 38619. Using this deep subsurface dataset from a hadal trench, we attempt to elucidate the accumulation and transformation of dissolved carbon in the hadal trenches, and to decipher the roles of tectonic events in the carbon cycles of these ultra-deep-water environments connected to subduction zones.

Fig. 1: Site map in the Japan Trench and IW geochemical parameters.
figure 1

a Location of twelve IODP 386 sites; b Alkalinity (blue), sulfate (red) and methane concentrations (black), and the ratios of methane to ethane and propane (C1/(C2 + C3), gray) at the studied sites. The SMTZ are inferred by depths with minimum sulfate and methane concentrations and marked in gray. Source data are provided as a Source Data file.

Results and discussion

Labile dissolved carbon accumulation in trench sediments

The Japan Trench receives pelagic background SOC descended from the water column and earthquake-remobilized SOC with various provenances (land, continental shelf, slope, and trench slope, etc.) and isotopic signatures2,8. Previous studies found the pervasive existence of earthquake-generated turbidite deposits along the Japan Trench3,6,8. At Sites M0081, M0083, and M0084, the 14C and 13C discrepancies between different carbon pools (Supplementary Fig. 1) exclude the possibility that the majority of the dissolved carbon in Japan Trench subsurface sediments are sourced from the exchange with bottom water or remineralization of pelagic background SOC. Instead, the dissolved carbon is mainly produced from remineralization of earthquake-introduced, likely marine-sourced SOC with various 14C ages and chemical recalcitrance (Supplementary Text 1). Additionally, the high IW alkalinities along the trench axis (Fig. 1b) suggest that earthquake-triggered SOC accumulation2 is followed by intensive SOC remineralization. The surficial (above 0.35 meter below the seafloor, mbsf) IW DOC concentrations (Fig. 2) in the Japan Trench are ~6 times higher than those in other oceanic trenches20 and increase with depth, reaching maximum concentrations (Fig. 2) that are comparable to the coastal21 and continental shelf22 areas, highlighting a dynamic carbon cycle and huge dissolved carbon storage in the hadal trenches enhanced by earthquakes.

Fig. 2: Lithologies, IW DOC (red) and DIC (black) concentrations, δ13C values and 14C ages of sediment cores at three IODP 386 Sites.
figure 2

Sites a M0081, b M0083 and c M0084 are collected from the southern, central and northern Japan Trench, respectively. Source data are provided as a Source Data file.

In many marine environments, most of the labile OC is selectively remineralized to DIC while the refractory fraction remains within the DOC pool20,23,24, potentially leading to older DO14C ages than DI14C ages23,25,26. However, our data in the Japan Trench exhibit similar DO14C and DI14C ages (Fig. 3) that increase almost linearly with depth (Fig. 2). Essentially, this suggests a relatively fast equilibration of dissolved carbon pools supported by rapid IW diffusion27 before their burial into the deeper subsurface. Additionally, no significant 14C discrepancy between the DOC and DIC pools in the hadal sediments means that at least part of the labile DOC is preserved, and go through aging in the trench sediments rather than being intensively oxidized or diffused overwhelmingly into the overlying seawater, in sharp contrast to other oceanic settings where mostly refractory DOC is preserved23,24,28. While the labile DOC in shallow subsurface sediments typically acts as a carbon source to the atmosphere due to microbial carbon turnover, those in the deeper-subsurface anoxic sediments tend to be chemically recalcitrant and elude aerobic oxidation29. In the Japan Trench, however, the instantaneous deposition of up to several meter thick event layers4,30 driven by earthquakes prevents the exposure to oxidation and exchange with ocean water for most of the sediment9,12, resulting in a chemically-active dissolved carbon sink in trench sediments.

Fig. 3: DI14C ages vs. DO14C ages of IW samples from Sites M0081 (square), M0083 (circle), and M0084 (triangle).
figure 3

14C ages exhibit a linear relationship between DOC and DIC (R2 = 0.98, p < 0.01). The pink shading illustrates 95% confidence band of the linear regression. IW depth is indicated by color bar. Source data are provided as a Source Data file.

Although the IW DOC is derived from earthquake-induced SOC remineralization, our results show that the relatively constant production of DOC may not be completely regulated by SOC content (Supplementary Fig. 2). It is possible that only the most labile fraction of the SOC is remineralized into DOC, supported by the positive δ13CDOC evidence (Fig. 2). Despite processes such as remineralization and methanogenesis that consume IW DOC, we still find sufficient residual DOC that accumulates and ages with depth (Fig. 2), which remains available for biogeochemical processes in the deeper subsurface sediments. In addition, the linearly-increasing 14C ages of the residual DOC indicate the time since sediment deposition and may constrain the chronological models for trench sediments devoid of pelagic carbonates. This implies a steady state DOC turnover in a non-steady state oceanic sedimentary setting that is irregularly disturbed by turbidite deposition. The dissolved carbon pools in the trench sediments thus provide potential information about the hadal sedimentation and diagenesis processes.

Enhanced methanogenesis in hadal zones

The large amount of dissolved carbon may fuel microbial metabolisms in the deep subsurface sediments31,32. Below the SMTZ, the methane concentrations along the Japan Trench (up to 7.35 mM, Fig. 1b) and the high ratios of methane to ethane and propane (C1/(C2 + C3) > 1000, Fig. 1b) imply intensive microbial methanogenesis33. The abundant DIC in the deep subsurface sediments accumulated from intensive OC remineralization has the potential to promote methanogenesis through CO2 reduction. In turn, methane production and oxidation exert significant control on the DIC pools below and above the SMTZ, respectively (Supplementary Fig. 3 and Supplementary Text 1). The positive δ13CDIC values in the deep subsurface sediments (Fig. 2) indicate a carbon isotope fractionation during microbial methanogenesis, where the residual DIC becomes more 13C-enriched33,34. Up to 24%–38% of the DIC is estimated to be utilized during methane production in the Japan Trench (Supplementary Fig. 4, see Method for calculation), highlighting intensive methanogenesis processes fueled by the accumulated DIC in the trench sediments.

Low concentration and short residence time of dissolved substrates may result in relatively less significant methanogenesis through fermentation in marine environments33,35. However, below 30 mbsf, the DOC is younger than DIC in the Japan Trench (Fig. 3), likely suggesting that a newly-added DOC pool is more bioavailable to heterotrophic organisms36. The labile DOC accumulation implies the possibility of intensified methanogenesis through fermentation37 in hadal trenches along subduction zones. The thick earthquake-triggered turbidite deposit results in less permeability38 and prevents oxygen and sulfate supplies to the sediments below it, leading to intensive OC remineralization12 and elevated SMTZ, thereby enabling substrate availability for in-situ microbial fermentation. Enhanced sediment compaction by repeated deposition of thick turbidites30 further contributes to these processes. The earthquake-triggered labile DOC production and preservation thus fodder microbial methanogenesis in the deep subsurface sediments.

With increasing time after deposition and repeated occurrences of large earthquakes, the IW volume is reduced due to sediment compaction and earthquake-induced dewatering and consolidation39, resulting in relatively constant DOC concentration below 20 mbsf (Fig. 2). The maximum concentration at 15 ~ 20 mbsf (Fig. 2) may indicate a compensation depth of the labile DOC, below which the DOC is consumed faster than it is produced and condensed, and may be continuously converted to methane by the microbial communities in deeper subsurface sediments40. The cascading effects of earthquake-induced processes including voluminous SOC deposition, rapid sediment burial, and sediment compaction consequently stimulate microbial methanogenesis via multiple metabolic pathways in the trench sediments (Fig. 4a).

Fig. 4: Conceptual models of the earthquake-enhanced carbon cycle in trench sediments and the shallow subduction zone.
figure 4

a Dissolved carbon dynamics in trench sediments are enhanced by earthquakes. (i) Sufficient dissolved carbon production facilitated by the combined effects of earthquake-triggered sediment deposition and compaction leads to (ii) larger DOC and DIC pools that are aging with depth, which result in (iii) enhanced microbial methanogenesis via fermentation and CO2 reduction, and (iv) elevated SMTZ that enables methanogenesis at shallower depth. b The impact of earthquakes on the carbon cycle in the subduction zone. The purple star indicates megathrust earthquakes along the plate boundary, which trigger seismic remobilization of SOC to the trench. Carbon in the trench sediments enters the subduction zone and undergoes dehydration, forming carbon reservoirs during processes in the deep Earth.

Linking between shallow and deep carbon cycles

The OC preservation regulated by the penetration of electron acceptors (oxygen, sulfate, etc.) is usually correlated with water depth in deep-sea environments41. The OC preservation and turnover in the hadal trenches, instead, are regulated by earthquake disturbances. The earthquake-triggered process cascade prolongs the residence time of labile carbon, which would have been otherwise consumed and released into the bottom water on the time scale of centuries to millennia, and thereby promotes carbon aging and subduction in the trench sediments (Fig. 4a).

The labile DOC pools serve as food sources for microorganisms in the trench sediments and have the potential to fuel the microbial-mediated carbon transformation in the deep biosphere42,43 (Fig. 4b). Volcanic ashes, biosiliceous-rich sediments, detrital silicate minerals and the abundant alkalinity along the Japan Trench (Fig. 1b) are suggestive of a contribution from marine silicate weathering, which provides the alkalinity and cations needed for authigenic carbonate precipitation44 (Supplementary Fig. 5). This deeper-subsurface carbonate formation serves as a ‘carbon sink swap’ from OC to methane, carbonates and eventually to graphite as more stable forms in the subduction zone system45,46. Although substantial storage of methane has been reported in the continental shelf and cold seep sediments47, its presence in trench sediments is not thoroughly evaluated. A comparison of the methane data from the IODP database (Supplementary Fig. 6) suggests that the methane storage in the Japan Trench may be comparable to typical methane-rich seafloor areas. More methane supply from the deep subsurface would support stronger methane oxidation in the SMTZ and reduce the sulfate available for OC remineralization, thereby promoting deep carbon burial.

The earthquake-enhanced dissolved carbon storage in the trench sediments may eventually contribute to carbon subduction1,48. Through sediment deposition and compaction, earthquake-triggered turbidites instantaneously increase the depth of the carbon residing in the sediments, leading to intensified carbon cycling in deeper subsurface depth than expected. The dissolved and sedimentary carbon may enter the subduction zones without becoming chemically recalcitrant for utilization by the deep biosphere12. The pore fluid containing dissolved carbon enters the crustal and mantle parts of the subducting slab and is transformed into solid phases, forming various carbon reservoirs (carbonates49, sediment-derived fluid50, graphite46,51 etc., Fig. 4b) at forearc depth52,53. Earthquakes therefore accelerate and magnify the processes in the subduction zones (Fig. 4b). In the meantime, spatial heterogeneity of our IW geochemical results exists along the trench. Considering the global variations of sedimentation environments, SOC characteristics, and earthquake frequencies in the trench systems11,54,55, we suggest that the spatial and temporal dynamics of dissolved carbon reservoirs will be further amplified by irregularity of earthquake occurrence and difference in their magnitudes in trenches globally, and will therefore cause critical and variable impacts on the global carbon cycle.

Our presented data and results, acquired through advanced drilling and radiocarbon measurement techniques, offer a comprehensive perspective of the hadal trench carbon cycle in the deepest and least explored environments on our planet, where geological events such as great subduction zone earthquakes may serve as key modulators. The contribution from our newly-discovered carbon reservoirs in the Japan Trench sediments provides essential, detailed information regarding how intensive earthquake-induced SOC accumulation leads to enhanced microbial activities in the deep biosphere. The occurrence of the dissolved carbon compensation depth indicates intensive carbon maturation and alteration in the trench subsurface sediments. These processes occur at the inlet to the subduction zone systems, where fresh materials are irregularly introduced by geological events. The cascade of the earthquake-triggered physical and chemical processes accelerates the transformation between carbon in different forms (i.e., ‘carbon sink swap’) at the shallowest setting of the subduction zone systems. Meanwhile, open questions remain regarding whether, and how much of, the carbon in the trench sediments is subducted in its various forms. We suggest that the earthquake-enhanced dissolved carbon reservoir in the trenches will constitute a potential source for the deep carbon reservoir in the lithosphere or mantle.


Research materials

The Japan Trench is located between the Kuril-Kamchatka Trench and the Izu-Bonin Trench. It is formed by the subduction of the Pacific Plate beneath the Okhotsk Plate with a convergence rate of 8.0 – 8.6 cm/yr (ref. 56). It is 611 km long and has a surface area of 37,854 km2 (ref. 57). Normal fault grabens resulting from the flexural bending of the slightly obliquely subducting Pacific Plate oceanic crust form numerous isolated basins that act as terminal sinks for trench-fill sedimentation along the axis of the Japan Trench30,58. Sediment cores were collected with up to 40 m long giant piston corers (GPC) on R/V Kaimei during the International Ocean Discovery Program (IODP) expedition 386 in 2021 (Fig. 1a & Supplementary Table 1).

Sample collection

Interstitial water (IW) was collected on board R/V Kaimei immediately on recovery using Rhizone samplers (CSS 19.21.23 F; Rhizosphere Research Products, Netherlands). Rhizone samplers consisted of the microporous tube (2.5 mm diameter, 5 cm length) that was supported by a nylon wire/PEEK wire and connected to PVC tubing with a female Luer lock connector, that can be attached to a syringe. Before use, Rhizone samplers were soaked in purified lab water for approximately 60 minutes. A 3.8 mm hole was drilled into the plastic core liner using the drill bit, and the Rhizone filament was inserted into the sediments. Negative pressure was applied by attaching a 24 mL plastic pulled-back syringe to the Rhizone sampler. The sediment cores remained capped throughout this process to minimize ambient gas penetration. First 0.1 mL (1 – 2 drops) were discarded because it would include purified lab water, and then Rhizone samplers were reassembled and remained in the cores for a maximum 6 h. The IW was split into sub-samples for alkalinity (3–4 mL) and sulfate (3 mL) measurements following IODP standards19, and for dissolved organic carbon (DOC, 8 mL) and inorganic carbon (DIC, 2 mL) measurements. Split samples for alkalinity and sulfate measurements were stored at 4°C. Sample splits were added with 50 µL 10% HCl and stored at –20°C for DOC measurement, or were added with 10 µL saturated HgCl2 and stored at 4°C for DIC measurement.

Bottom water (BW) was collected from the top of the pilot cores using a BW sampling system consisting of a Duran bottle and vacuum pump unit connected by tubes. Negative pressure was applied and BW samples were sucked into the Duran bottle through the tube placed in the overlying water of the trigger core. After filtering using a Rhizon sampler, BW samples were split for DOC (1 L) and DIC (5 mL) measurements. Sample splits were added with 1 mL 10% HCl and stored at –20°C for DOC measurement, or were added with 10 µL saturated HgCl2 and stored at 4°C for DIC measurement.

Headspace samples were taken directly from the base of each core section after core cutting. A total volume of 5 cc (using 2.5 cc cut tip syringe twice) was collected and each 2.5 cc sample was then added into 20 mL crimp vial with 5 ml of 1 M NaOH and stored upside down and cooled at 4°C.

Geochemical analysis

IW and BW geochemical analysis was carried out during the onshore and offshore phases of the IODP Expedition 386 following standard IODP procedure19.

Water splits were analyzed onboard for alkalinity immediately following extraction with an autotitrator (Metrohm 888 Titrando). An aliquot of 1–3 mL water split-adjusted to 3 mL with 0.7 M KCl solution was titrated with nominal 0.1 M HCl at 25 °C. A 100 mM Na2CO3 solution was used for the calibration of the acid each week. Quality checks were conducted using a 50 mM solution of NaHCO3 and 0.1 M HCl once per day.

Water splits were measured for sulfate on a Metrohm 882 Compact IC ion chromatograph at the MARUM. A 40-fold dilution of IAPSO seawater and standards prepared from commercial single anion standards was used for calibration. Analytical precisions were ± 0.95% for sulfate.

Gas concentrations in headspace samples were measured between September and November 2021 onboard of D/V Chikyu. The samples were vigorously shaken by hand for 2 minutes, then shaken by a lab shaker for 1 h and left for 23 h at room temperature. Each vial was placed in an Agilent 7697 A headspace sampler, and was heated to 70 °C for 30 minutes. An aliquot of the headspace gas was then automatically injected (split mode) into an Agilent 7890B gas chromatography equipped with a packed column (HP PLOT-Q) and linked to a flame ionization detector. The carrier gas helium flow rate was 10 cm³/min. The oven temperature was programmed to start at 60 °C and was ramped to 150 °C at a rate of 10 °C/min. The instrument was calibrated for the chromatographic response with commercial standards (Types VIII–XII, GL Sciences, Japan). Quantification of hydrocarbon gases was achieved by comparison of its chromatographic response with a three-point calibration curve.

DOC and DIC concentration and isotopic analysis

Holes M0081D, M0083D, and M0084D were chosen to represent the DOC and DIC signatures of the corresponding sites (Supplementary Table 1). All DOC and DIC samples were analyzed at the National Ocean Science Accelerator Mass Spectrometer (NOSAMS) facility at Woods Hole Oceanographic Institution for concentration, 13C and 14C analyses. Due to the pandemic, water splits were stored at the Kochi Institute for Core Sample Research of the Japan Agency for Marine-Earth Science and Technology for 6 months before being shipped to NOSAMS. The DIC concentration was likely to be affected by some carbonate precipitation during storage given the extreme IW alkalinities of up to 100 mM. The DOC and DIC measurement procedures were described in detail in refs. 59,60. Briefly, BW samples were filtered through pre-combusted 0.45 μm GF/F films before measurement. A quartz reactor prefilled with Milli-Q water was irradiated with UV for 1.5 h to reduce the potential contribution of extraneous carbon. Approximately 2 mL water sample and 0.4 g 33% ultra-pure hydrochloric acid were then added to the reactor. DOC samples were irradiated with UV for 3 hr. The generated CO2 was purged by ultra-high purity helium (UHP He) at 120 mL/min for 66 min, and was cryogenically trapped, purified, manometrically quantified, and collected by flame sealing in a 6.4 mm OD Pyrex glass tube. DIC was also extracted on the DOC line by replacing the quartz reactor with a gas-washing device equipped with an injection port. The device was added with Milli-Q water and H3PO4 and then cleaned with He before sample injection. The generated CO2 was purged by UHP He at 120 mL/min and collected by a CO2 trap.

δ13C were measured by a VG Optima or VG prism Isotope Ratio Mass Spectrometer at NOSAMS. δ13C values were reported in ‰ relative to the VPDB standard. The typical precision of δ13C measurement was between 0.1‰ – 0.23‰. 14C were measured by a continuous-flow accelerator mass spectrometer (CFAMS) system constructed around an accelerator (National Electrostatics Corporation, Middleton, WI, model 1.5SDH-1) at NOSAMS61,62. AMS uncertainty was calculated as the larger of either the statistical uncertainty using the total number of 14C counts measured for the target, or an uncertainty calculated from the reproducibility of multiple measurements of the target, both propagated with uncertainties from the normalizing standards and blank subtraction. Δ14C values were reported as 14C ages63. For the IW samples, the 14C measurement errors were within 0.34% – 2.03% and thus didn’t impact data robustness.

DIC conversion estimation

During methanogenesis via CO2 reduction in the marine environment, DIC is utilized by methanogens with a kinetic isotope effect that discriminates 13C. The fractionation factor of DIC to methane (εc) is between –50 to –90‰ in natural observations33. Due to the lack of δ13Cmethane data, we use both reported εc values33 as upper and lower limits to estimate the amount of DIC that is converted to methane. A Rayleigh distillation function can be used40 to describe the 13C fractionation during methanogenesis in sediment IW:

$${{{{{{\rm{\delta }}}}}}}^{13}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{DIC}}}}}},{{{{{\rm{t}}}}}}}{{{{{\rm{\hbox{-}}}}}}}{{{{{{\rm{\delta }}}}}}}^{13}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{DIC}}}}}},0}={{{{{{\rm{\hbox{-}}}}}}}{{{{{\rm{\varepsilon }}}}}}}_{{{{{{\rm{c}}}}}}}{{{{{\rm{ln}}}}}}{f}_{{residual}}$$

Where δ13CDIC,0 and δ13CDIC,t are the isotopic compositions of the initial and residual DIC pools at a certain time, respectively, and fresidual represents the amount of residual DIC expressed as a fraction (f ranges from 0 to 1). The fraction of DIC that has been utilized during methanogenesis, fconverted, can be calculated as:


We use the linearly-extrapolated δ13CDIC at 0 mbsf (–12.43‰, –14.00‰ and –13.58‰ for Sites M0081, M0083 and M0084, respectively) as δ13CDIC,0. The fraction of residual DIC can thus be calculated according to Eq. (1) by using the δ13CDIC at the given depth as illustrated in Supplementary Fig. 4. Calculation results of the deepest IW sample of each Hole are presented in Supplementary Table 2.

Reporting summary

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