The Kermadec Trench, a major hadal feature (>6,000 m depth) in the southwestern Pacific Ocean, extends approximately 1,200 km along a north-northeast axis with a maximum width of 120 km (Figure 1). This convergent margin results from the subduction of the Pacific Plate beneath the Australian Plate, generating complex seafloor geomorphology. Its southern terminus connects with the continental margin of New Zealand via the Hikurangi-Rapuhia transition zone, while its northern continuity with the Tonga Trench is disrupted by the Louisville Seamount Chain collision zone (Jamieson et al., 2020). The trench exhibits pronounced flank asymmetry, characterized by a steep Kermadec forearc slope along its western margin contrasting with the gentler gradient of the eastern Pacific Abyssal Plain (Lemenkova, 2019). A distinctive ~2,800-m-high basaltic escarpment on the northern ridge acts as a topographic barrier to Lower Circumpolar Deep Water, forcing northwestward flow to bifurcate and generating a trench-parallel current along the hadal axis (Germineaud et al., 2021). This tectonic-hydrodynamic coupling drives an efficient sediment routing system, with persistent axial currents advecting particulate material toward the trench floor. As a highly active seismic zone, the Kermadec Trench experiences frequent large-magnitude earthquakes that significantly alter sediment dynamics. Notable recent events include three M≥7 earthquakes on March 4, 2021, and 15 seismic events in 2022, culminating in a M7.9 earthquake in November 2022 (Peng et al., 2023; Ye et al., 2025). These earthquakes act as dominant geological drivers, triggering rapid morphological adjustments through mass wasting and generating cascading sedimentary disturbances.

During the China-New Zealand Joint Expedition (October–December 2022), six sediment push cores were retrieved from the Kermadec Trench using the manned submersible Fendouzhe and its support vessel Tansuoyihao. Sampling targeted six stations along bathymetric gradients (6,000–9,232 m water depth), encompassing three distinct geomorphological units: the continental slope (FDZ126), the abyssal plain (FDZ128), and four stations along the trench axis (FDZ127, FDZ132, FDZ143, FDZ147) (Table 1). Stations were strategically organized into two orthogonal transects (Figure 1): Transect 1 (W–E orientation, perpendicular to trench axis) spanning FDZ126 (slope) → FDZ127 (axis) → FDZ128 (plain), and Transect 2 (N–S orientation, parallel to trench axis) spanning FDZ132 (north axis) → FDZ147 (central axis) → FDZ127 (central axis) → FDZ143 (south axis). Geospatial analysis confirmed the linear alignment of both transects, with inter-station distances calculated using the Haversine formula based on GPS coordinates. To facilitate subsequent location identification, station codes are appended with positional identifiers: FDZ126-LS (landward slope), FDZ127-MA (middle axis), FDZ128-AP (abyssal plain), FDZ132-NA (north axis), FDZ147-NA (north axis), and FDZ143-SA (south axis).

Upon retrieval, sediment cores were immediately transferred to the onboard laboratory for dissolved oxygen (DO) measurements and subsequent porewater extraction. Pre-drilled access holes at 2 cm vertical intervals, sealed with tape prior to deployment to preserve core integrity, maintain core integrity, facilitated both processes. Although autonomous benthic landers provide the highest accuracy for porewater DO measurements by minimizing artifacts associated with core retrieval (e.g., pressure changes, temperature shifts, and oxygen consumption during ascent) (Glud, 2008), our shipboard protocol prioritized rapid processing to minimize atmospheric exposure. DO measurements were performed first using a calibrated Unisensor (Denmark) microelectrode. Following immediate removal of the tape seal, the electrode was inserted through the access holes, and measurements were recorded upon signal stabilization. This approach provided a preliminary characterization of the oxygen gradient within the inherent constraints of ex situ shipboard analysis.

Porewater was subsequently extracted using Rhizon soil moisture samplers (5 cm length, 0.1 µm pore size inert polymer tubes). Vacuum pressure applied via disposable 20 mL syringes, yielding approximately 15 mL of porewater per 2 cm interval over a 3-hour extraction period. Extracted porewater were immediately processed for preservation. Aliquots for nutrient analysis were stored in centrifuge tubes at −20 °C. Subsamples for dissolved inorganic carbon (DIC) analysis were preserved in 2 mL glass vials with 10 μL of saturated HgCl₂ solution, sealed, and stored at 4 °C to inhibit microbial activity. Although HgCl₂ preservation may affect DIC measurements in porewater with high H₂S concentration (Argentino et al., 2023), its impact is negligible here due to the shallow (<30 cm) sediment depth of our hadal samples, which typically exhibit very low H₂S levels. After porewater collection, sediment cores were sectioned at 2 cm intervals and stored at -20 °C for subsequent solid geochemical analysis.

Sediments within the Kermadec Trench are predominantly silty (77.2–86.4% silt), with minor clay (6.4–15.8%) and sand (0–10.3%) fractions (Figure 2). Median grain size (D₅₀) ranges from 6.7 to 11.1 μm (mean ±1σ: 8.9 ± 1.4 μm). Station FDZ147-NA exhibits anomalous granulometry, characterized by highly variable sand content (8.4–82.5%) and a maximum D₅₀ of 69.8 μm, accompanied by reduced silt-clay fractions (17.5–91.6%). Our preliminary comparison of adjacent sedimentary settings reveals distinct regimes: the abyssal plain sediments contain 11.7–47.2% sand (D₅₀: 12.03 ± 2.1 μm), slightly coarser than those on the landward slope (0–48.5% sand; D₅₀: 11.21 ± 3.4 μm). Two obvious sand-rich layers were identified: one containing 48.5% sand at 16 cm depth in FDZ126-LS (D₅₀ = 15.8 μm), and the other with 47.2% sand at 24 cm depth in FDZ128-AP (D₅₀ = 14.3 μm). These horizons exhibit >300% sand enrichment relative to adjacent strata, suggesting episodic high-energy deposition. The SSA values display significant spatial heterogeneity (12.49–84.95 m²/g; Supplementary Table S1). Northern trench axis stations exhibit systematically higher SSA than southern counterparts, peaking at 84.95 m²/g (FDZ147-NA). Cross-trench transects reveal a consistent SSA pattern: abyssal plain > trench axis > landward slope.

TOC decreases from 0.51% (surface, FDZ143-SA) to 0.16% (24 cm, FDZ128-AP), while TN declines from 0.08% to 0.02% (Figure 3). Notably, FDZ147-NA exhibits coupled extremes: minimum TOC (0.10% at 14 cm) coincides with maximum SSA (84.95 m²/g) and sand content (82.5%), suggesting grain-size mediated organic matter preservation. Along the trench axis, TOC decreases latitudinally from 0.43 ± 0.08% (south) to 0.21 ± 0.05% (north), coinciding with increasing water depth (9,240 m at FDZ143-SA to 10,050 m at FDZ147-NA). Perpendicular to the axis, surface sediment TOC shows distinct zonation: landward slope deposits contain 31% more TOC than trench axis sediments, which exceeds abyssal plain values by 21%. The steepest gradient occurs between landward slope and trench axis (ΔTOC = 0.09%/km), compared to 0.05%/km from trench axis to abyssal plain.

XRD analysis of selected layers (based on grain size distribution and porewater geochemistry; Supplementary Figures S1, S2, Supplementary Table S2) reveals hadal sediments are dominated by muscovite (avg. 28%), albite (avg. 34%), quartz (avg. 19%), and chlorite (avg. 11%), with albite as the primary phase. Volcanic material (clinoferrosilite and enstatite) was identified at FDZ147-NA (northern trench; 2 cm and 20 cm depths) and FDZ128-AP (abyssal plain; 24 cm depth). These volcanic-rich horizons correlate strongly (R² = 0.59) with sand content. Clinoferrosilite content peaked at 17.7% (20 cm layer, FDZ147-NA) but remained below 8% at FDZ132-NA. No volcanic ash was detected at FDZ126-LS or FDZ143-SA; however, the sand-rich layer at 16 cm depth in FDZ126-LS corresponded to elevated albite (40%), while FDZ127-MA contained trace ferrosilite (1%). Kaolinite was consistently detected at four axial trench stations (avg. 13.88%; Supplementary Figures S1 S2, Supplementary Table S2).

Cross-trench porewater DO profiles reveal redox zonation correlated with geomorphological position (Figure 4). Station FDZ131, located on the continental slope adjacent to FDZ132-NA (Supplementary Figure S3), exhibits complete oxygen depletion below 5 cm depth (DO < 4 μmol/L). Stations along Transect 1 (FDZ127-MA–FDZ128-AP) show a transitional redox regime, maintaining oxic conditions down to 20 cm depth. Porewater DIC concentrations generally increase with sediment depth, ranging from 2.48 ± 0.15 mM at 0–15 cm to 3.82 ± 0.23 mM at 30–40 cm (Figure 5). However, DIC concentrations lack a clear spatial distribution pattern. Notably, DIC-δ¹³C values revealed anomalies: at the southernmost station FDZ143-SA, values were enriched within 0–10 cm but sharply depleted below 12 cm. All three Transect 1 stations exhibit a DIC-δ¹³C inflection point at 6–8 cm depth, potentially indicating intensified mineralization within this layer. Porewater NO₃⁻ and NH₄⁺ distributions show pronounced spatial heterogeneity (Figure 6). Both landward slope and abyssal plain environments exhibit peak NH₄⁺ accumulation (>100 μM below 5 cm depth). In contrast, trench-axis stations maintain low NH₄⁺ concentrations (<10 μM below 2 cm), with only FDZ127-MA showing surface enrichment (49.3 μM). Nitrate (NO₃⁻) concentration correlates with oxygen availability. At the anoxic site FDZ132-NA, NO₃⁻ decreases with depth from 25.5 μM at the surface. Conversely, at the oxic Transect 1 stations, NO₃⁻ increases with depth, following the order FDZ128-AP > FDZ127-MA > FDZ126-LS. Phosphate (PO₄³⁻) concentrations show a slight north-to-south increasing trend, while silicate (SiO₃²⁻) concentrations exhibit a southward decrease (Figure 6). Dissolved iron (Fe) concentrations were generally low across the study area, with detectable levels found in only six sediment layers. Three layers in core FDZ147-NA showed Fe concentrations increasing with depth, reaching a maximum of 3.1 mg/L (Supplementary Table S3). Detectable Fe was also present in single layers at sites FDZ126-LS, FDZ127-MA, and FDZ143-SA.

To investigate spatial variations in OC and its transport within Kermadec Trench sediments, we grouped sampling stations into two transects based on distinct geomorphological settings. Transect 1 traverses the landward slope, trench axis, and abyssal plain, capturing a cross-sectional profile. Transect 2 follows the central trench axis along a north-south orientation, representing its longitudinal axis. This strategic grouping captures contrasting sedimentary depositional regimes, enabling comparative analysis of OC sources and transport pathways between the cross-section of trench and longitudinal axis. The transect-based approach thus facilitates systematic evaluation of how topographic position and sediment transport dynamics influence OC distribution patterns in these hadal environments.

The extreme pressure characteristic of trench environments may promote the breakdown of recalcitrant organic materials (Luo et al., 2018). Theoretically, sedimentary organic matter mineralization is fundamentally governed by Gibbs free energy, dictating the sequence of electron acceptor utilization (e.g., O₂, NO₃⁻, Mn(IV), Fe(III), SO₄²⁻) (Arndt et al., 2013). Measurements in the Kermadec Trench reveal localized oxygen penetration depths exceeding 20 cm (Figure 4; Xiao et al., 2024), coupled with generally higher oxygen consumption rates at the trench axis compared to adjacent regions (Glud et al., 2013). These observations confirm that aerobic respiration dominates OC mineralization in these sediments. Consistent porewater oxygen profiles across the continental slope-trench axis-abyssal plain transect (perpendicular to the trench axis) further support this primary aerobic pathway. Beyond aerobic respiration, key hypoxic processes contribute to OC cycling: denitrification consumes NO₃⁻ and organic matter to produce N₂, while ammonification during early diagenesis generates NH₄⁺ (Cai et al., 2010; Chen et al., 2023). Nitrification, evidenced by relatively high NO₃⁻ content in Transect 1, which is also significant in trench surface sediments. Elevated PO₄³⁻ and SiO₃^2- concentrations throughout the system provide additional evidence for active mineralization.

The TOC/SSA ratio reflects sediment exposure time and the influence of sedimentary processes on OC accumulation (Mayer, 1994; Li et al., 2017, 2020). TOC/SSA values < 0.4 typically indicate prolonged oxygen exposure, reducing sediment mineralization capacity (Blair and Aller, 2012). Data comparisons with the East China Sea and South China Sea (Figure 9a) show that OC loading values (0.1–0.2) in parts of the South China Sea shelf and abyssal plain influenced by riverine suspended sediments indicate moderate mineralization potential. Surface sediments at stations FDZ126-LS, FDZ127-MA, and FDZ143-SA (proximal to land) fall within this range, suggesting retained mineralization potential. In contrast, TOC/SSA values < 0.1 at stations FDZ132-NA and FDZ147-NA (northern study area) are comparable to that in some areas of the abyssal plain of the South China Sea, indicating limited mineralization potential (Figure 9b).

As previously noted, station FDZ147-NA is influenced by volcanic ash input. The overall increase in TOC content with depth suggests volcanic ash facilitates TOC preservation. However, a distinct TOC minimum observed at the 14–16 cm depth interval likely results from dilution by a concentrated layer of volcanic ash accumulation at this horizon. Furthermore, porewater nitrate concentrations at FDZ147-NA are relatively low compared to central and southern stations (FDZ126-LS, FDZ127-MA, FDZ128-AP, FDZ143-SA). This observation aligns with the previously described pattern of suppressed mineralization. Additionally, enriched silicate concentrations in the porewater of the ash-rich sediment layer suggest active dissolution and release of silica from the volcanic material. Collectively, the observed low NO₃⁻ concentrations, coupled with elevated porewater silicate levels indicative of active ash dissolution, point to a dual role for volcanic input: it acts as a source of reactive silica while simultaneously enhancing OC preservation. This enhancement likely arises from the formation of stabilizing mineral-organic associations (Zhao et al., 2024; Huang et al., 2023). Consequently, mineralization processes in trench surface sediments are controlled not only by water depth but also significantly by volcaniclastic input, with episodic ash deposition exerting a primary influence on OC mineralization and preservation.

Earthquake-triggered disturbances significantly enhance sediment transport from landward slopes into trenches via lateral transport (Kioka et al., 2019). The V-shaped trench morphology, combined with earthquake-induced turbidity currents, facilitates efficient particulate matter transport along the trench axis (Talling et al., 2024; Turnewitsch et al., 2014). Such events typically deposit sediment layers 30 cm to 1 m thick within trenches. For instance, the 2011 Tohoku earthquake deposited a 31 cm-thick sandy layer on the Japan Trench seafloor within a short period (Oguri et al., 2013). Similarly, sediments exhibiting positive grading and anomalously high sand content were observed at depths of 5–16 cm in Transect 1 of this study, suggesting deposition by earthquake-triggered turbidity currents. Given frequent seismic activity of the Kermadec Trench, including a magnitude 6.6 shallow-focus earthquake (intensity VI) west of the FDZ126-LS observatory on August 14, 2022 (USGS), we speculate that this event caused the anomalously high sand content via turbidity current deposition.

Earthquake-generated turbidity currents enable rapid OC deposition. The OC deposition rate is quantified as:

where OC deposition rate is in g m⁻² yr⁻¹, LSR is the linear sedimentation rate (cm yr⁻¹), TOC is the total OC content (wt%, g OC g⁻¹ dry sediment), and DBD is the dry bulk density (g cm⁻³).

Previous studies on the Kermadec Trench have shown that under stable conditions, the linear sedimentation rate (LSR) measured on the southern side of FDZ126-LS is 2.01 × 10⁻² cm yr⁻¹, and the average dry bulk density (DBD) is 1.49 g cm⁻³ (Xu et al., 2021). Based on grain size distribution profile, the sediment thickness deposited during the interval from the earthquake occurrence to sampling (three months) was set at 16 cm. The TOC content in the bottom layer of FDZ126-LS is 0.22 wt%, which was used to estimate the OC buried in this single event. Calculations based on the above data indicate that the amount of event-driven OC transport and preservation is at least 20-fold higher than that under stable sedimentation conditions. High-frequency seismicity in subduction trenches drives sedimentary OC burial via earthquake-triggered turbidity currents and mass wasting. These events likely dominate OC sequestration in hadal systems, potentially constituting a major pathway in global carbon budgets. In contrast to its influence on sediment burial efficiency, seismic activity also exerts a strong control on porewater oxygen dynamics and sediment mineralization. Oxygen penetration depth (OPD) exceeds 20 cm at stations FDZ127-MA and FDZ128-AP, contrasting sharply with only 10 cm at FDZ132-NA, which experiences minor seismic influence. This disparity is attributed to seismically triggered turbidity currents in Transect 1, enhancing sediment oxygenation (Fernandes et al., 2016). The resulting increased OPD intensifies surface OC mineralization, evidenced by elevated NO₃⁻ and PO₄³⁻ concentrations (0–8 cm depth) across all Transect 1 stations (Figure 6).

Beyond seismic activity, volcanic eruptions represent another major geological force shaping sediment composition and biogeochemical cycling in the Kermadec Trench system. The January 5, 2022, eruption north of the trench (Tonga) significantly reshaped these processes. XRD analysis reveals a north-to-south decrease in volcanic ash content within trench sediments (Supplementary Table S2), indicating dominant ash transport in this direction. This volcanic input correlates with sediment properties: higher SSA corresponds to OC loading (Figure 8), with XRD confirming abundant volcanic ash in these high-SSA layers (Supplementary Figures S1-2, Supplementary Table S2). Sediment provenance varies spatially: at FDZ126-LS, a sand content peak at 10–15 cm depth coincides with minimum SSA, indicating turbidity current-transported material. In contrast, FDZ128-AP and FDZ147-NA sediments are predominantly volcaniclastic. This pattern suggests seismic disturbances facilitate transport of slope-derived OC into the trench interior, while rapid volcanic deposition dilutes OC on the abyssal plain at our site. Supporting this, TOC content increases with depth at FDZ147-NA. Critically, the absence of positive grading at FDZ147-NA confirms its volcanogenic sand provenance, independent of seismic activity, consistent with the 2022 eruption’s influence north of the trench. Although volcanic ash generally facilitates organic matter preservation, concurrent mineralization occurs during burial. This eruptive event further modulated mineralization processes, evidenced by a strong negative correlation (R² = 0.975, p < 0.01) between dissolved Fe and δ¹³C-DIC in porewater (Supplementary Figure S4). Heterotrophic Fe(II)-oxidizing bacteria utilize structural iron within igneous minerals, releasing bioavailable Fe and associated metallic elements. Strains isolated from volcanic ash exhibit heterotrophic Fe(II) oxidation coupled with nitrate reduction (Sudek et al., 2017), providing a plausible mechanism for the observed denitrification at 5–10 cm depth along Transect 1. Integrating these findings with the established depositional pattern (landward slope → trench axis → abyssal plain), we infer that material derived from the southern slope is transported northward along the trench axis by ocean currents, influenced by the New Zealand North Island. During transport, marine OC undergoes gradual mineralization, whereas terrestrial OC accumulates in the northern trench. Meanwhile, earthquake-triggered turbidity currents on the landward slope facilitate rapid burial of OC. Given the eruption’s location north of the trench, the northern trench axis receives substantial volcaniclastic input. In contrast, the abyssal plain is influenced by both volcaniclastic density currents and deposition of surface-derived marine plankton, as illustrated in Figure 10.