During the 15^th cruise of the Tansuoyihao launched by the Institute of Deep-Sea Science and Engineering of the Chinese Academy of Sciences in 2019, shallow sediment samples were collected at the "Challenger" Deep Site MT02 at 10,954 m water depth (11.327°N, 142.188°E), and its adjacent abyssal plain Site MT04 at 5,800 m water depth (10.761°N, 142.274°E) in the Mariana Trench using a lander ( Figure 1 ). Short cores were subsampled using a transparent PVC tube on board. The core lengths were 18 cm for core MT02 and 24 cm for core MT04. Sediments in both cores were composed of silt and clay, and the "Challenger" Deep core MT02 contained some diatomaceous debris (Luo et al., 2022; Song et al., 2022). The sediment core was sub-sampled at 1 cm intervals for Site MT02 and at 2 cm intervals for Site MT04. Porewater was collected by using a pre-cleaned Rhizon sampler with a pore size of 0.22 μm attached to a 20 ml syringe. Porewater samples for FT-ICR MS analysis were stored in pre-combusted brown glass bottles and then kept at - 20°C.

Due to the small volume of pore water samples, all pore water samples collected at Site MT02 and Site MT04 were combined into two samples each to meet the requirement of at least 100 μg C of sample for FT-ICR MS analysis. Given that the retrieved sediment cores were within the oxic zone (Glud et al., 2013), the change in redox condition had negligible impact on the molecular signature of porewater DOM despite of the mixture of porewater sample at each core. The solid-phase extraction of DOM was carried out by a modified procedure according to Dittmar et al. (2008). Firstly, the pore water samples were acidified to pH = 2 with HCl. Then, the solid-phase extraction columns (Bond Elut-PPL, 100 mg, 3 ml; Agilent Technologies, Germany) of the modified styrene-divinylbenzene polymer were pretreated with ultrapure methanol (Mass Spectrometry grade), ultrapure water, and hydrochloric acid. The entire pore water sample was slowly passed through the extraction columns and then rinsed with acidified water to remove inorganic salt. Finally, the solid-phase extraction columns were eluted with 1 ml of methanol and stored at -20°C in the dark for subsequent FT-ICR MS analysis.

The FT-ICR MS analysis of pore water DOM was conducted on a Bruker Solarix equipped with a 7.0 T superconducting magnet at Tianjin University. Before analysis, all SPE-DOM samples were diluted with a mixture of methanol and ultrapure water (1:1) to achieve a final SPE-DOC concentration of 50~100 μg/g. The diluted SPE-DOM samples were ionized using an electrospray ionization source in negative ion mode at a flow rate of 150 μL/h. The capillary voltage was set at 5 kV, and the mass-to-charge (m/z) spectra in the range of 150-800 Da were obtained by scanning each spectrum approximately 256 times, with the data acquisition transient lasting for 4 M. The acquired mass spectra were analyzed using Data Analysis software version 5.0 (Bruker Daltonics). Compounds repeatedly identified in pore water samples within the mass-to-charge ratio range underwent internal calibration, with a root mean square error of internal calibration<1 ppm, ensuring reliable molecular formula assignment: ¹²C (1-60), ¹H (2-120), ¹⁶O (0-30), ¹⁴N (0-3), and ³²S (0-2). Furthermore, the molecular formulas were further validated by limiting them to H/C ≤ 2.25 and O/C ≤ 1.2 based on elemental ratios.

To reduce the complexity of the FT-ICR MS spectral analysis, the molecular formulas were categorized based on their elemental composition into four groups: 1) molecular formulas containing C, H, and O atoms (CHO); 2) molecular formulas containing C, H, O, and N atoms (CHON); 3) molecular formulas containing S (CHOS); and 4) molecular formulas containing N and S (CHONS).

Additionally, a modified aromaticity index (AI_mod) and the ratio of H/C to O/C were used to classify different categories of molecular formulas into six compound groups (Koch and Dittmar, 2006, 2016): 1) saturated fatty acid compounds (H/C > 2, O/C < 0.9); 2) highly unsaturated compounds (AI_mod < 0.5, H/C < 1.5, O/C < 0.9); 3) aromatic compounds (0.67 > AI_mod ≥ 0.5); 4) peptide-like compounds (2 > H/C ≥ 1.5, O/C < 0.9, N > 0); 5) unsaturated aliphatic compounds (2 ≥ H/C > 1.5, O/C < 0.9, N = 0); and 6) the CHO-formulas matching the molecules were identified as carboxyl-rich alicyclic molecules (CRAM) (0.3 < DBE/C < 0.68, 0.2< DBE/H < 0.95, 0.77 < DBE/O < 1.75). Labile components generally exhibit low degree of unsaturation and aromaticity with elevated H/C, which are mostly bioavailable. Recalcitrant DOM molecules are characterized by high aromaticity and unsaturation and an increase in structural complexity, rendering them resistant to microbial degradation. Here, we classified a series of compounds as labile DOM compounds, including saturated fatty acids, peptides, and unsaturated aliphatic compounds that exhibited low aromaticity and unsaturation. In contrast, aromatic compounds, highly unsaturated compounds, and CRAM were classified as recalcitrant compounds (Hertkorn et al., 2006; Koch and Dittmar, 2006; Rossel et al., 2016). Additionally, the “island of stability” (IOS) represents a specific range of elemental combinations (H/C = 1.17 ± 0.13, O/C = 0.52 ± 0.10, MW = 360 ± 28 and 497 ± 51 Da) where compounds persist over geological timescales for 1.5 times the average residence time of DOC (Lechtenfeld et al., 2014). The AI_mod values were calculated as follows:

The overall stability of DOM was evaluated by calculating the molecular stability index (MLB_L, the number of formulas with H/C > 1.5 divided by the total number of formulas) (D’Andrilli et al., 2015). Each sample was normalized to the sum of FT-ICR MS peak intensities. Then, the weighted average values of molecular parameters and the proportions of the relative intensity of different classes of molecular formulas were calculated. The number proportion was obtained by dividing the number of each chemical formula type by the total number of chemical formulas.

The pore water DOM of Site MT02 and Site MT04 contained 4,906 common molecular formulas. After excluding the common formulas from the total formulas at each site, there were 2,499 unique formulas at Site MT02 and 1,311 unique formulas at Site MT04. The DOM from Site MT02 had 1,188 unique formulas more than Site MT04, indicating a greater diversity of compound types which were mainly refractory highly unsaturated and CRAM compounds. The relative intensities of these two refractory compounds types with lower H/C and higher O/C ratios ( Figure 4 ) at Site MT02 (70.4% and 51.9%, respectively) were significantly higher than those at Site MT04 (38.9% and 29.7%, respectively) ( Figure 5 ). In contrast to refractory compounds, the relative intensities of peptide and unsaturated aliphatic labile compounds were lower in the unique formulas of Site MT02 compared to those of Site MT04 ( Figures 5A , 5B ), with higher H/C and lower O/C ratios ( Figure 4 ). Additionally, the relative intensities of refractory compounds with CHO formulas in the unique formulas of the pore water DOM at Site MT02 was 17.2%, and the relative intensities of labile compounds were 4.9% ( Figure 5C ), which were significantly higher than those formulas composed of the other elements and unique formulas of Site MT04 ( Figures 5C , D ).

Sedimentary DOM is produced by microbial utilization of particulate organic matter (Komada et al., 2013; Burdige and Komada, 2015). Highly unsaturated compounds and CRAM are ubiquitous because they are resistant to be degraded (Medeiros et al., 2015; Martínez-Pérez et al., 2017). The axis of the Mariana Trench exhibited a higher mineralization rate and stronger microbial activity compared to the abyssal plain (Glud et al., 2013; Wenzhöfer et al., 2016; Luo et al., 2018). The elevated microbial carbon turnover led to a production of more highly degraded and persistent DOM fractions as indicated by the increase in humification indices and absorbance ratios at the trench axis cores (Hu et al., 2021).

The porewater SPE-DOM at both sites consisted mainly of highly unsaturated compounds and CRAM ( Figures 3A , B ). However, both the proportion of total molecular numbers and relative intensity of these two groups of recalcitrant compounds were significantly higher at the "Challenger" Deep Site MT02 than those in the abyssal plain Site MT04 ( Figures 3C , D ), especially the relative intensities of the extremely recalcitrant IOS compounds ( Figures 3A , B ). This suggests that the enhanced microbial activity at the trench axis may lead to increased production and accumulation of refractory DOM.

Studies have shown that complex substrates and microbial diversity contributed to the molecular diversity of DOM (Chen et al., 2022). The number of pore water DOM unique formulas at the "Challenger" Deep Site MT02 is significantly higher than the abyssal plain Site MT04. Moreover, the unique formulas types of DOM in the trench axis sediments were mainly highly unsaturated compounds and CRAM ( Figures 4A - D ), while he abyssal sediments contained more relatively labile peptides and unsaturated aliphatic compounds ( Figures 4E - H ). The difference in the porewater DOM molecular compositions may be associated with the diverse microbial communities between the trench axis and abyssal plain sediments (Glud et al., 2013; Nunoura et al., 2015; Wu et al., 2023). The alternative explanation could be the presence of giant diatom debris in the trench axis sediments. It has been suggested that the hydrolysis and degradation of hydrocarbon compounds in the giant diatom-bearing sediments in the southern Mariana Trench sediments could lead to the release of CHO-containing DOM components into porewater (Hu et al., 2024). Therefore, the enhanced microbial activity in the "Challenger" Deep likely generate a higher abundance of refractory compounds, mainly CHO-containing highly unsaturated compounds and CRAM.

The molecular composition of sediment pore water DOM at different sites and water depths is highly diverse (Schmidt et al., 2017). Understanding the potential controlling factors of these molecular compositions is crucial as it can provide insights into the biogeochemical processes and environmental conditions in different marine settings. We performed principal component analysis (PCA) >of our DOM data with published data at water depths from 30 to 2,300 m (Schmidt et al., 2017; Hu et al., 2023) and those from the seaward slope of the Mariana Trench at ~7,000 m water depth (Hu et al., 2024) ( Table 2 ; Figures 6A , B ). Molecular parameters incorporated in PCA were all abundance-weighted averages. Considering the potential offset in molecular parameters due to instrument differences, prior to PCA analysis, we applied independent sample t-tests to identify parameters that may have been greatly affected by different instruments (see Supplementary materials ). To ensure that PCA analysis mainly reflects the real differences between samples, parameters significantly affected by instrument differences were not used in PCA analysis. Ultimately, H/C_wa, AI_modwa, and N_wa were included in PCA analysis.

The PCA results show that our data and those reported in Schmidt et al. (2017) do not exhibit obvious grouping ( Figure 6 ); both located within the 95% confidence interval ellipse of Group 2, likely suggesting a small influence of instrument differences on analytical results. In addition, K-means cluster analysis was used to verify that three parameters used for PCA analysis were not impacted by artefacts among instruments (see Supplementary materials ). The sample from the "Challenger" Deep core MT02 is distributed within the confidence interval ellipse of Group 1, while MT04 and MT03-1 are outside ( Figures 6A , B ). This implies that the "Challenger" Deep sample is more similar to samples from shallower water (<2,300 m) than to those from the abyssal plain (5,800 m) and seaward slope (~7,000 m). The correlation analysis shows that the TOC content in surface sediments at different water depths has a significantly negative correlation with the first principal component (PC1) ( Figure 6C ). This suggests that the TOC content may be a key factor leading to the differences in the molecular composition of the pore water DOM among different marine environments.

Correlation analysis of the nine datapoints located within the confidence interval ellipse of Group 1 showed that H/C_wa was significantly negatively correlated with TOC content, and that AI_modwa was significantly positively correlated with TOC content ( Figure 7 ; red lines). This further suggests that TOC exerts an important control on the molecular composition of DOM. However, when the two datapoints outside the confidence interval ellipse of Group 1 were included in the regression analysis, the overall correlations were significantly reduced ( Figure 7 ; black lines). This is mostly due to higher H/C_wa ratios and lower AI_modwa in the abyssal plain (MT04) and seaward slope (MT03) than those in the trench axis (MT02), which is likely associated enhanced microbial activity and transformation of organic matter (Glud et al., 2013, 2021; Wenzhöfer et al., 2016; Hu et al., 2021). Through comparisons with published data, we suggest that TOC content is likely a key controlling factor of pore water DOM characteristics among different environments and that microbial transformation can also play a role.