Site W07, the drilling core site used in this study, is located in the uppermost part of the research area, with a water depth of 914.27 m and a borehole depth of 159 m below the seafloor (mbsf). In this study, we focused on the gas hydrate-bearing layers (121–152 mbsf) (Bai et al., 2024) and the continuously cored adjacent sedimentary layers, covering depths of 102–156 mbsf ( Figure 3 ). The gas hydrates in the study site mainly consisted of methane gas and belong to type I gas hydrates (Song et al., 2022), with approximately 152m thickness of the hydrate stable zone (HSZ) ( Figure 2 ). The maximum age of the studied layer is estimated to be 3.95 Ma based on biostratigraphic analysis of calcareous nannofossils from the GMGS4 cruise report ( Figure 3 ). A total of 132 samples were collected for grain-size measurements to reconstruct the deep-water sedimentary processes. A subset of 31 samples was analyzed using X-ray diffraction (XRD) to determine the clay mineral composition and infer the sedimentary provenance. Additionally, the major and trace element concentrations of 39 samples were measured, and the foraminiferal abundances of 37 samples were analyzed.

The grain-size parameters of all 132 samples were measured at the GMGS Test Center using a Marvin Mastersizer 2000 laser diffraction size analyzer. The samples were pretreated with deionized water and 0.5 mol/L sodium hexametaphosphate (Na₆P₆O₁₈) as a dispersant. To prevent fine particle aggregation, the suspension was left to stand for ~24 hours and gently stirred every 8 hours. The grain-size data, including the mean grain size, sorting, kurtosis, and skewness, were analyzed using the logarithmic (original) Folk and Ward (1957) graphical measures. The calculation Equations 1–4 are as follows:

where φ is the phi-scale grain size (φ = −log₂d, with d being the grain diameter in μm), which is plotted on the X-axis. The subscripts indicate the specific cumulative percentages on the Y-axis, corresponding to their intersections with the cumulative frequency curve (Wentworth, 1929).

A D8 Advance instrument (Bruker Corp., Germany) was used to acquire XRD data for the bulk and clay minerals. The methodology adhered to the protocols outlined by Moore and Reynolds (1989), Sakharov et al. (1999), and Hillier (1999). The ClayQuan software was used to process the results (Yang et al., 2024; Zhou et al., 2023). The XRD analysis was conducted collaboratively by the Key Laboratory of Polar Geology and Marine Mineral Resources, Ministry of Education, and the Micro Structure Analytical Laboratory, Peking University.

The major and trace elements and rare earth elements (REE) measurements were performed using a PerkinElmer Optima 8300 ICP-OES instrument at the Laboratory of the GMGS.

The foraminifera abundance measurements conducted in this study were performed by the Laboratory of the GMGS. The dried sediment samples (approximately 6–20 g) were soaked in distilled water to facilitate effective separation. Subsequently, the samples were wet-sieved using a copper sieve with a 0.063 mm aperture, and the >0.063 mm fraction was dried. Finally, the >0.15 mm fraction was used to quantify the foraminiferal abundance. The benthic and planktonic foraminifera were analysed separately. Additionally, the identification and relative abundance statistics of the calcareous nannofossils were conducted by the Laboratory of the GMGS.

In this study, the sediment grain-size distribution at Site W07 was analyzed using several parameters, including the mean grain size, sorting, skewness, kurtosis, and median grain size.

For the 132 samples, the mean grain size ranged from 5.66 to 7.87 φ (average of 6.91 φ); the sorting values ranged from 1.40 to 3.00 (average of 1.83); the skewness and kurtosis ranges were from −0.27 to 0.42 and from 0.86 to 1.26 (averages of 0.15 and 1.02), respectively; and the median grain size ranged from 5.66 to 7.84 φ (average of 6.78 φ) ( Figure 3 ).

In the sediment samples, the main minerals included quartz, feldspar, carbonates, and clay minerals (over 95%). The quartz content ranged from 23% to 46% (average of 36.8%), the feldspar content ranged from 8% to 14% (average of 11.0%), the carbonate content ranged from 14% to 42% (average of 28.2%), and the clay mineral content ranged from 11% to 30% (average of 18.2%) ( Figures 4a–d ).

The XRD analysis of the clay minerals revealed that the principal clay minerals at Site W07 included the illite-smectite (I/S) mixed layer, illite, kaolinite, and chlorite. The I/S mixed layer and illite accounted for 46–75% (average of 57.3%) and 13–28% (average of 21.7%) of the clay minerals, respectively, while the kaolinite and chlorite contents ranged from 6% to 12% (average of 9.3%) and from 6% to15% (average of 11.7%), respectively ( Figures 4e–h ).

We measured the major, trace, and REE contents of the sediment samples from Site W07. Some of the elements were used to calculate indicators such as the P/Al, A/Ti, P/Ti, Sr/Ba, and Zr/Ti ratios, which are vital for sedimentary environment reconstruction. The details are shown in Figure 5 . The P/Ti ratios ranged from 0.0924 to 0.1915 (average of 0.1502), the Al/Ti ratios ranged from 13.67 to 19.64 (average of 16.32), and the P/Al ratios ranged from 0.0127 to 0.053 (average of 0.0094). Additionally, the Sr/Ba ratios ranged from 0.430 to 2.147 (average of 1.045). The Zr/Ti ratios ranged from 0.031 to 0.0472 (average of 0.0375).

The foraminifera abundance, defined as the number of undistorted, intact foraminifera shells per gram (ind/g), was determined following the method of Azharuddin et al. (2017). For the samples from Site W07, the total foraminifera abundance ranged from 12 to 14,878 ind/g, exhibiting a general trend of decreasing with increasing depth. The planktonic foraminifera abundance ranged from 11 to 14,720 ind/g, whereas the benthic foraminifera abundance ranged from 0.2 to 158.4 ind/g. In addition, the planktic/benthic foraminifera (P/B) ratio increased with depth, ranging from 13 to 423. The relative abundance of the calcareous nannofossils is defined as the number of individuals per field of view (ind/F). For the samples from the research layer, the relative calcareous nannofossil abundance ranged from 57.6 to 292.2 ind/F, exhibiting a general trend of decreasing with increasing depth.

Figures 4e–h illustrates that the trend of the clay mineral component value changed at the transition boundary between the two distinct sedimentary processes. Figure 4 further reveals the occurrence of significant longitudinal differences in the mineral composition within the studied sediment interval, particularly between the upper and lower subsections. The upper subsection has a significantly higher concentration of felsic minerals (quartz and feldspar), whereas the lower subsection has a higher total clay mineral content ( Figures 4e, f ). Additionally, the clay mineral species in the upper and lower subsections differ markedly. The split violin plot shown in Figure 9 clearly demonstrates the variations in the clay mineral components in the upper and lower subsection sediments. Notably, the variations in the I/S mixed layer, illite, and chlorite contents are the most pronounced. The I/S mixed layer content is higher in the upper subsection (57–75%) and lower in the lower subsection (46–64%). Conversely, the illite content is lower in the upper subsection (13–23%) and higher in the lower subsection (17–28%). The kaolinite and chlorite content variations between the two subsections exhibit features similar to the variations in the illite content. Specifically, the lower subsection contains slightly more kaolinite and chlorite than the upper subsection.

These differences in the clay minerals between the upper and lower subsections can be attributed to distinct provenance sources. Several studies have investigated the source-to-sink transport processes between seafloor surface sediments in the SCS and adjacent riverbed sediments, based on the similarities of their clay mineral components (Liu et al., 2016; Hu et al., 2021). To determine the provenances of the two subsections, we employed a ternary diagram of the primary clay mineral groups (illite + chlorite, kaolinite, and smectite) to compare the clay mineral contents of the sediment samples from Site W07 ( Figure 10a ) and the seafloor surface samples (indicated by shaded areas in Figure 10b ). Figure 10a indicates that the data points of the samples from studied layer at Site W07 clustered in area A. In contrast, the clay mineral composition of the upper subsection was shifted toward area B. In the northern part of the SCS, three primary sediment sources have been identified: the Pearl River (dominated by kaolinite and illite), Taiwan (dominated by illite and chlorite), and Luzon (dominated by smectite) (Liu et al., 2016). Area A’s clay mineral assemblage was marked by high concentrations of illite and chlorite, along with minor amounts of smectite and kaolinite, reflecting a predominantly Taiwan-sourced signature with an admixture of Luzon-derived smectite. This suggests that the sediments were primarily transported by currents flowing through the Luzon and Taiwan straits, possibly including the SCS Branch of the Kuroshio Current (SCSBKC) ( Figure 1a ). Moreover, the variations in sea level and current direction and intensity since the Quaternary have likely altered the influence zones of these three provenance areas (Steinke et al., 2008; Liu et al., 2010).

The distribution trends of the data points of the samples from the two subsections facilitated interpretation of the variations in the sediment provenances. For the lower subsection, the data point distribution resembled that of area A. In contrast, the data points of the upper subsection had higher smectite contents, and the data points clustered near area B, implying a greater contribution from Luzon. The lower subsection had a lower smectite content, which implies a weak contribution from Luzon and a predominant influence from Taiwan. Furthermore, the lower subsection had higher kaolinite and chlorite contents than the upper subsection, indicating some influence from the Pearl River provenance.

Differences in sedimentary processes are closely linked to evolving paleoenvironments (Scheffler et al., 2006; Bjørlykke, 2014). The SCS, which includes our study area, is widely recognized as a natural laboratory for deciphering paleoenvironmental changes based on sedimentary records (Xu et al., 2017; Wang et al., 2020). In this study, we employed geochemical and foraminifera evidence from sediments as proxies to reconstruct changes in the sea level, primary productivity, and paleoenvironments in the research area.

Sea level fluctuations significantly influence various sedimentary processes that respond to paleoenvironment changes. The Al/Ti ratio is an effective proxy for tracing sea level changes by accurately reflecting the input of terrestrial sediments (Steinke et al., 2006; Chen et al., 2013). Al and Ti predominantly originate from terrestrial debris material (TDM), which is characterized by stable chemical properties and a high resistance to chemical weathering (Murray & Leinen, 1996). Additionally, the distributions of these elements vary between mineral types: Ti is primarily associated with heavier minerals, whereas Al is more abundant in light minerals. In summary, the Al/Ti ratio tends to be lower during lower sea levels and increases as the sea level rises (Chen et al., 2013). The Sr/Ba ratio is a widely adopted proxy for paleo-salinity and is closely associated with sea level changes (Wei & Algeo, 2020). Sea level fluctuations can modify the mixing ratios of freshwater and seawater, thereby influencing the Sr/Ba ratio. Generally, the Sr/Ba ratio exhibits a proportional increase with increasing water depth.

Potential factors affecting the Al and Ti concentrations must be eliminated to ensure the reliability of the Al/Ti ratio, particularly the potential increase in Ti related to volcanic activity (Smith and Fritz, 1989; Li et al., 2020). Figure 5e shows that within the studied interval, the Zr/Ti ratio remained low and exhibited a relatively smooth trend throughout the interval. Because sedimentary Zr is primarily derived from volcanic activity, the consistently low and stable Zr/Ti ratios in the studied interval may indicate that terrestrial detritus was the predominant source (Boës et al., 2011). The PAAS-normalized REE distribution pattern ( Figure 11a ) showed that both the upper and lower subsections exhibited essentially identical REE patterns and lacked significant positive Eu anomalies, thus further supporting the absence of active volcanism during the research interval (Green, 2006). Accordingly, the Al/Ti ratio ( Figure 5b ) is a robust proxy indicator for sea level changes in the study area.

To use the Sr/Ba ratio as a proxy for paleo-salinity, which reflects sea level changes, it is crucial to eliminate potential interfering factors. Carbonate minerals and calcareous nannofossils, including coccolithophores, primarily affect the Sr concentration. We performed correlation analysis between the Sr concentrations and the calcareous nannofossil abundances and the carbonate mineral contents of the upper and lower subsections. The results are shown in Figure 11b and c. In the upper subsection, the Sr concentration exhibited a weak positive correlation with the calcareous nannofossil abundance (R2 = 0.2928; Figure 11b ) and a stronger positive correlation with the carbonate mineral content (R2 = 0.6039; Figure 11c ). In contrast, the low subsection did not exhibit significant correlations between the Sr concentration and these two indicators ( Figures 11b, c ). These results suggest that the Sr/Ba ratio of the upper subsection may not reliably indicate the paleo-salinity. Because the studied sediments are unconsolidated, potential terrestrial influences, especially from feldspar, on the Ba concentration must be excluded. Moreover, there was a weak correlation between the Ba and feldspar contents ( Figure 11d ) in the upper subsection (R2 = 0.5484), while there was no significant correlation in the lower subsection. Thus, the Ba concentration in the upper subsection may be overestimated due to terrestrial inputs, whereas estimates for the lower subsection are likely more reliable.

Accordingly, the Al/Ti ratio indicated that the sea level was higher when the upper subsection was deposited and lower when the lower subsection was deposited ( Figure 5b ). The Sr/Ba ratio ( Figure 5d ) exhibited a trend similar to that of the Al/Ti ratio ( Figure 5b ), with fluctuations in the upper subsection attributed to the effects of feldspar and carbonate minerals on the Sr and Ba concentrations. However, the Sr/Ba ratio ( Figure 5d ) was higher in the upper subsection than in the lower subsection.

The P/B ratio, defined as the ratio of planktonic foraminifera abundances to benthic foraminifera abundances, is strongly correlated with paleo-productivity and typically increases as the sea level rises (van der Zwaan et al., 1990; Azharuddin et al., 2017). Tyrrell (1999) proposed that P inputs govern oceanic primary productivity, as P is not regenerated once consumed. To mitigate the influence of the absolute P concentration, we employed the P/Ti and P/Al ratios as geochemical proxies for deciphering the paleo-productivity in the study area. The paleo-productivity is an essential factor in reconstructing paleoenvironments due to its close association with climate change. These parameters can be used to reconstruct the sedimentary environment and reveal longitudinal variations within the studied interval.

In the bottom layer (155–147 mbsf), the overall foraminifera abundance ( Figure 12a ) decreased with decreasing water depth (i.e., sea level drop). The planktonic foraminifera abundance ( Figure 12b ) exhibited a similar decreasing trend, while the benthic foraminifera abundance ( Figure 12c ) remained relatively stable. Notably, the significant decrease in the P/B ratio ( Figure 12d ) reflected a reduction in the surface primary productivity. Both the P/Al ( Figure 5c ) and P/Ti ( Figure 5a ) ratios exhibited decreasing trends, thereby supporting this inference. Furthermore, the Al/Ti ( Figure 5b ) and Sr/Ba ( Figure 5d ) ratios suggested that this layer corresponded to a persistently low sea level phase.In the middle-lower layer (147–133 mbsf), the foraminifera abundance ( Figure 12a ) initially increased, accompanied by increasing P/Al ( Figure 5c ) and P/Ti ( Figure 5a ) ratios, which indicates enhanced surface primary productivity. Simultaneously, both the Al/Ti ( Figure 5b ) and Sr/Ba ( Figure 5d ) ratios increased, suggesting a rise in sea level and a warmer climate. The unusually high P/B ratio ( Figure 12d ) may result from the reducing conditions of the marine subsurface environment. Subsequently, both the P/Al ( Figure 5c ) and the P/Ti ( Figure 5a ) ratios decreased, demonstrating a cooling climate, falling sea level, and reduced productivity. In the middle-upper layer (133–121 mbsf), the foraminifera abundance ( Figure 12a ) decreased sharply before gradually recovering. Simultaneously, both the planktonic ( Figure 12b ) and benthic foraminifera abundances ( Figure 12c ) increased, leading to a sharp increase in the P/Ti ratio ( Figure 5a ). Additionally, the P/Al ratio ( Figure 5c ) increased gradually, while both the Sr/Ba ( Figure 5d ) and Al/Ti ratios ( Figure 5b ) increased significantly. These observations indicate a rise in sea level and a warm climate, resulting in enhanced surface primary productivity, as evidenced by the marked increase in the foraminifera abundance ( Figure 12a ). During the deposition of the top layer (121-102mbsf), a higher sea level and warmer climate prevailed, as indicated by the higher Al/Ti ( Figure 5b ) and Sr/Ba ( Figure 5d ) ratios, along with a substantial increase in foraminiferal abundance (>2 orders of magnitude) ( Figure 12 ). Meanwhile, GMGS4 cruise report show that this layer exhibits a marked shift in planktonic foraminiferal assemblage, with G. ruber, a surface-dwelling warm-water species, rapidly becoming dominant, indicating a significant climatic warming event (Wang, 2000; Kubota et al., 2010). The observed decrease in the surface primary productivity, as reflected by the decreases in the P/Al ( Figure 5c ) and P/Ti ( Figure 5a ) ratios, may be attributed to nutrient consumption caused by the extensive foraminifera.

Overall, the studied layer indicated a transition from a glacial period characterized by lower sea levels to an interglacial period characterized by higher sea levels. Moreover, the transitions of the paleoenvironment and paleoclimate coincided with the boundary separating the upper and lower subsections.

The sedimentary processes at Site W07 are closely correlated with the distribution of the gas hydrate reservoirs and the non-gas hydrate layers. The lower subsection, dominated by gravity flows, corresponds to the gas hydrate reservoirs, whereas the upper subsection, influenced by traction (contour) currents, corresponds to the non-gas hydrate layers.

Turbidite processes, including both sandy and fine-grained types, are generally considered to be favourable for gas hydrate accumulation because their coarser and heterogeneous grain sizes enhance the pore space and provide migration pathways for gas (Ito et al., 2015; Su et al., 2021). At Site W07, the upper subsection has slightly finer sediment grain sizes than the lower subsection. The mean grain sizes were 5.66–7.87 φ (average of 6.94 φ) in the upper subsection and 6.05–7.32 φ (average of 6.74 φ) in the lower subsection. The median grain sizes were 5.66–7.84 φ (average of 6.90 φ) in the upper subsection and 5.69–7.24 φ (average of 6.49 φ) in the lower subsection. However, the difference in the sorting values between the two subsections was minimal. Some of the samples from the upper subsection even exhibited poorer sorting, possibly due to their higher foraminifera contents. This presents a paradox. Although some studies suggest that a high foraminifera content promotes gas hydrate accumulation due to their coarse grain size and chambered structures (Chen et al., 2013; Zhang et al., 2017; Bai et al., 2022), at Site W07, the non-gas hydrate layers have significantly higher foraminifera contents ( Figure 12a ) (Jin et al., 2020). This suggests that coarser components do not necessarily promote gas hydrate accumulation. Both the grain size and foraminifera abundance indicated that grain size variations had a limited influence on gas hydrate accumulation in fine-grained sediments.

Furthermore, the data for Site W07 indicated that the varying sedimentary processes led to distinct mineralogical variations. Previous studies have shown that the mineralogy significantly influences gas hydrate formation. Quartz, due to its hydrophilic surface, promotes hydrate nucleation (Yu et al., 2024), whereas clay minerals hinder accumulation due to their high strong-bound water contents and complex structures (Venaruzzo et al., 2002; Zhang et al., 2017; Bai et al., 2022). Among clay minerals, the I/S mixed layer plays a dominant role due to its complex internal structure and large specific surface area, which significantly reduce porosity and permeability, effectively hindering the migration of gases and fluids in sediments and thereby suppressing hydrate formation (Zhang et al., 2022; Bai et al., 2024; Yang et al., 2024). These findings are consistent with the mineralogical variations observed at Site W07 ( Figure 4 ).

Analysis of the sedimentary processes and depositional environments at Site W07 (Stage 1; Figure 13 ) indicated that the gravity flow-dominated lower subsection formed under cold climatic and low sea level conditions and that the sediments were primarily sourced from Taiwan. Under these conditions, the reduced inflow of the SCSBKC through the Luzon Strait limited the Luzon-derived sediment input, while the lower sea levels enhanced the sediment discharge from the paleo-Pearl River. The study site’s proximity to shelf edges, where slumping events are more frequent, also contributed to the sediment delivery. These conditions led to mineral assemblages enriched in felsic minerals and depleted in the I/S mixed layer, both of which are favourable for gas hydrate accumulation. In Stage 2 ( Figure 13 ), the contour current-dominated upper subsection developed under warmer climatic and higher sea level conditions. The increased inflow of the SCSBKC through the Luzon Strait enhanced the Luzon-derived sediment input, while the rising sea level reduced the influence of the paleo-Pearl River discharge and suppressed gravity flow activity. These changes resulted in mineral assemblages with lower felsic contents and higher I/S mixed layer contents, conditions unfavourable for gas hydrate accumulation. It is worth noting that during periods of relatively intensified contour current activity, the sedimentary environment may become more oxidising, which could hinder the preservation of organic matter by promoting its degradation (Stow and Faugères, 2008; Amsler et al., 2022). This degradation leads to a reduction in the total organic carbon content of the sediments, potentially suppressing the generation of in-situ microbial gas and thereby exerting a negative influence on gas hydrate formation (Zhang et al., 2017, 2019).

Due to the high sensitivity of HSZ to environmental factors, such as geothermal gradient and formation pressure, gas hydrate reservoirs may undergo repeated cycles of dissociation and reformation with changes in the thickness of HSZ, making hydrate accumulation a dynamic process. In site W07, the gas hydrates at 121~152 mbsf formed at relatively recent ages, possibly after 0.15 Ma (Song et al., 2022), much later than the sedimentary formation of gas hydrate-bearing layers. Thus, the spatiotemporal linkage between sedimentation and gas hydrate accumulation is clarified. The prior formation of fine-grained turbidites (the lower subsection) and overlying contourite sediments (the upper subsection) provided suitable reservoir space for the upward migration and accumulation of thermogenic gas or gas released from hydrate dissociation in the underlying layers. Given the minor contribution of microbial gas to hydrate formation in the study area, the different enrichment of gas hydrates in the two sedimentary subsections is primarily attributed to variations in their sediment properties.

Therefore, the link between sedimentary processes and gas hydrate reservoirs is governed by two key factors: (1) mineralogical variations driven by different sedimentary processes, and (2) the joint control of depositional environments on both sedimentation and gas hydrate accumulation.