In 2016, core C1624 was drilled from the western slope of the middle OT (125.685°E, 26.755°N) by the Qingdao Institute of Marine Geology (Figure 1, red circle). The core was collected at a water depth of 1000 m and had a total length of 570 cm. The sediments of core C1624 are primarily composed of clayey silt (Figure 2A). A distinct sand layer is observed at the bottom of the core, representing the coarsest grain-size interval within the entire core (Figures 2A, B). Upcore, the sand content gradually decreases while the clay content increases, leading to an overall fining upward trend in grain size (Figures 2A, B). However, significant fluctuations in sand and clay contents occur at approximately 170 cm depth, marked by an increase in sand and a decrease in clay (Figure 2A). The mean grain size (Mz) of the core sediments ranges from 5.3 to 29.5 μm, with an average of 10 μm.

Based on the lithological characteristics of core C1624, clean and undamaged monospecific planktonic foraminifera Neogloboquadrina dutertrei were selected at various depths (specific depths indicated in Figure 2C). These samples were sent to Beta Analytic Incorporated in the USA for AMS¹⁴C dating. The conventional ¹⁴C ages were calibrated to calendar ages, with the bottom age of core C1624 subsequently estimated to be 25.3 ka based on a linear age-depth model (Wang et al., 2025). Sedimentation rates vary between 8.2 and 370.4 cm/ka. For most of the core depth, the sedimentation rate remains below 25 cm/ka (Figure 2D). An anomalously high sedimentation rate phase occurred between 12.3 and 13.7 ka, during which rates sharply increased to 114.1–370.4 cm/ka (Figure 2D).

The elemental analysis of sediments was completed at the Qingdao Institute of Marine Geology, China Geological Survey. A total of 56 samples were collected at 10 cm intervals for subsequent analysis. Firstly, sediment samples were dried in an oven at 60 °C for 24 hours and then ground into 200 mesh powders using an agate mortar. Approximately 4 g of the prepared sample was weighed, placed into a mold, and pressed into pellets for analysis. The analysis was performed using a PANalytical Axios X-ray fluorescence (XRF) spectrometer from the Netherlands. The operating conditions were set at a high power of 4 kW, with a maximum excitation voltage of 60 kV, a maximum current of 120 mA, and high transmittance, utilizing an SST ultra-sharp long-life ceramic end-window (75 µm) Rhodium target X-ray tube. The SuperQ software was used to scan the vicinity of elemental spectral line peaks, carefully selecting background positions and identifying interfering elements. Quality control during the analytical process was maintained using national standard reference materials and replicate samples, with the relative standard deviation of the analytical results being less than 5%.

The major element composition of marine sediments serves as a key geochemical indicator for interpreting their provenance and depositional processes (Jung et al., 2023; Liu et al., 2017). Sediments in the OT are typically a mixture of terrestrial detrital input, marine biogenic components, and potential volcanic/hydrothermal contributions (Qin et al., 1996). Accurate identification and quantification of these end-members are fundamental for reconstructing paleoenvironments and depositional patterns. Principal component analysis (PCA), an effective multivariate statistical method, can reduce the dimensionality of complex geochemical datasets (Nesrstová et al., 2025), thereby identifying the main common factors controlling element distributions and providing a reliable basis for tracing material sources.

To reveal the characteristics of sediment sources in the southern middle OT, we performed PCA on the contents of SiO₂, Al₂O₃, CaO, MgO, K₂O, Na₂O, MnO, TiO₂, and TFe₂O₃ in core samples using SPSS 27.0 software (KMO = 0.789, p < 0.001). The analysis extracted two principal components (PC1 and PC2), which together accounted for 96.1% of the total variance (PC1: 78.6%; PC2: 17.5%) (Figure 4A).

The loading plot reveals that PC1 accounts for the dominant portion. Its positive loading is strongly associated with SiO₂, Al₂O₃, MgO, K₂O, MnO, TiO₂, and TFe₂O₃, while CaO shows a significant negative loading (Figure 4A). This pattern clearly indicates that PC1 represents the mixing relationship between terrestrial detrital input (positive end) and marine biogenic carbonate input (negative end) (Jiang et al., 2011), which is the predominant process controlling sediment composition in the study area. PC2 is primarily associated with the positive loading of Na₂O (Figure 4A). The score plot visually confirms this end-member mixing relationship. The sample points show a good continuous distribution along the PC1 axis, and their color (representing CaO content) systematically transitions from high values on the left to low values on the right (Figure 4B), confirming a strong negative correlation between carbonate content and PC1 scores. This implies that as the proportion of terrigenous detrital material increases (higher PC1 score), the carbonate component may be diluted.

Core C1624 is located on the western slope of the trough, whereas volcanic materials in the OT are mainly enriched on the eastern slope (Xu et al., 2025). Furthermore, no distinct tephra layers were identified in core C1624 (Figure 2). Additionally, elements typically considered sensitive indicators of hydrothermal activity, such as TFe₂O₃, MgO, and MnO (Li et al., 2021; Yang et al., 2015), are highly aggregated at the positive end of PC1 in this PCA result, showing strong positive correlations with stable terrigenous elements like Al₂O₃, K₂O, and TiO₂. Moreover, there is a lack of significant factors independent of PC1 and PC2 that are dominated by typical hydrothermal element associations (e.g., Fe-Mg-Mn-rich). While dissolved or colloidal hydrothermal iron could potentially adsorb onto fine-grained sediments (Li et al., 2021), its influence is likely very limited. Therefore, it is suggested that iron in sediments is primarily derived from terrestrial rock weathering and erosion, being transported and deposited in detrital form, with its distribution controlled by a unified terrestrial input process (Plewa et al., 2012), rather than by independent hydrothermal activity.

Although authigenic carbonate formed via microbial-mediated early diagenetic processes (e.g., anaerobic oxidation of methane) is a recognized carbonate factory in marine sediments (Dai et al., 2025), its influence on the bulk sediment CaO content in the study area is considered minor. A previous geochemical study of core CSHC-4 from the OT has quantified that such methane-seep-related authigenic carbonates contribute only a small fraction (averaging ~3.7%) to the total carbonate pool (Li et al., 2024). Therefore, the PCA results effectively support the conclusion that sediment provenance in the study area is primarily controlled by terrigenous detritus and biogenic carbonate.

Sediment particle characteristics significantly influence elemental concentrations (Yang et al., 2003). To accurately assess the terrestrial input signal and eliminate the potential interference of sediment grain size variations on geochemical elemental indicators, we conducted a systematic correlation analysis between key terrigenous (Fe, Al) and biogenic (Ca) elemental indicators and sediment grain-size parameters (Figure 5).

Firstly, there is no significant correlation between TFe₂O₃ content and the Mz, which reflects the overall grain-size level of the sediment (r = 0.32, p < 0.05; Figure 5A). This indicates that fluctuations in TFe₂O₃ content are not primarily driven by simple overall coarsening or fining of the sediment. However, TFe₂O₃ shows a significant positive correlation with the clay fraction content (Y = 4.43 × X + 5.15, r = 0.44, p < 0.01; Figure 5B). This aligns with previous studies in marginal seas, where fine-grained particles, especially clay minerals, exhibit a strong adsorption and carrying capacity for iron (hydr)oxides due to their larger specific surface area (Gu et al., 2025; Kong et al., 2024). Aluminum (Al) is typically regarded as a classic tracer for terrigenous clay minerals. It remains stable during depositional processes and is largely unaffected by marine biological processes and diagenesis (Dymond et al., 1997; Feng et al., 2021; Murray and Leinen, 1996). In recent years, the method of elemental normalization using Al to eliminate physical interferences such as grain size has been widely applied (Sun et al., 2008). As shown in Figure 5C, TFe₂O₃ and Al₂O₃ contents exhibit an extremely strong positive correlation (Y = 1.66 × X + 5.44, r = 0.96, p < 0.01). This strongly confirms that, in the study area, both are jointly and predominantly controlled by the input of terrigenous clay minerals, rather than being significantly affected by later diagenetic alteration.

For calcium, an indicator element for biogenic carbonate (Jiang et al., 2011), the content of its oxide (CaO) shows no clear correlation with the Mz (Figure 5D). Further analysis indicates that CaO has no significant association with any of the sand, silt, or clay fractions (Figure 5E), demonstrating that its distribution is not controlled by sediment grain size. However, the PCA results show that CaO is strongly negatively correlated with the PC1 factor, which represents the dominance of terrigenous detritus (Figure 4). This relationship is directly manifested in elemental contents as a high negative correlation between CaO and Al₂O₃ (Y = -0.29 × X + 17.349, r = -0.85, p < 0.01; Figure 5F).

As elemental ratios are insensitive to the closure effect, they are often more indicative of material input than individual elemental concentrations (Weltje and Tjallingii, 2008). Based on the above comprehensive analysis, this study adopts the TFe₂O₃/Al₂O₃ ratio to indicate the relative enrichment degree of terrestrial material input (Jung et al., 2023), thereby excluding the simple grain-size enrichment effect. The CaO/Al₂O₃ ratio is used to more reliably indicate changes in the contribution of biogenic carbonate relative to the total terrigenous detritus. These two ratio indicators effectively circumvent the influence of grain size on elemental concentrations (Xu et al., 2025), thus providing a reliable geochemical basis for subsequent discussion on the evolution of terrestrial input under the context of sea-level changes.

Given that much of the modern ECS shelf is shallower than 120 m, the distance between the OT and the coastline was highly sensitive to sea-level changes during the deglaciation. Consequently, sea-level changes since the LGM have likely exerted a significant influence on the delivery of terrestrial materials to the OT (Chen et al., 2023; Diekmann et al., 2008). To gain a deeper understanding of the macro-scale mechanisms controlling variations in terrestrial and biogenic input, we placed the terrestrial input index (TFe₂O₃/Al₂O₃) and the biogenic contribution index (CaO/Al₂O₃) from core C1624 within the framework of regional paleoenvironmental evolution for comparative analysis (Figure 6). The results reveal that since the LGM, sediment provenance and depositional processes in the study area exhibited a clear, stage-wise response to sea-level changes (Stages S1–S3).

Stage 1 (25.4–11.6 ka BP): This stage corresponds to the LGM and early deglaciation, characterized by low sea levels (Figure 6E). During the LGM, global sea level was 120–130 m lower than present (Kao et al., 2006b; Lambeck and Chappell, 2001). This caused the coastline to prograde eastward, exposing almost the entire Yellow Sea and most of the ECS continental shelf (Qin et al., 1996; Saito et al., 1998). High-resolution seismic profiles indicate that paleo-Changjiang channel systems were widely distributed on the exposed ECS shelf during the LGM and the deglaciation (Li et al., 2005; Xu et al., 2022), implying that lower sea level allowed the paleo-Changjiang river system to extend across the exposed shelf, providing a direct conduit for sediment delivery. Concurrently, the KC was weak or absent during this period, which also favored the transport and deposition of fluvial sediments (Li et al., 2018; Ujiié et al., 2003; Zheng et al., 2014). The TFe₂O₃/Al₂O₃ ratio remained consistently high throughout this period (Figure 6B). Similarly, the Fe₂O₃/TiO₂ ratio in core C01 from the western slope of the middle OT (location in Figure 1) also showed high values during this stage (Xu et al., 2025). In contrast, the East Asian summer monsoon (EASM) was weaker during the glacial and deglacial periods compared to the Holocene (Figure 6A) (North Greenland Ice Core Project Members, 2004), suggesting relatively lower riverine input, which stands in opposition to the observed high TFe₂O₃/Al₂O₃ ratios. Furthermore, the EASM exhibited strong fluctuations during the deglaciation (Figure 6A; Dykoski et al., 2005), but this signal is not reflected in the TFe₂O₃/Al₂O₃ record of core C1624, the Fe₂O₃/TiO₂ record of core C01, or the Sr-Nd isotopic record of core DGKS9604 (Figures 6A–C; Dou et al., 2012; Xu et al., 2025). This indicates that terrestrial input to the OT during the glacial and deglacial periods was primarily governed by sea level. Based on previous reconstructions of the paleo-channels of the Huanghe and Changjiang (Li et al., 2005; Liu et al., 2010; Ujiié and Ujiié, 1999; Xu et al., 2022), core C1624 is located closer to the paleo-Changjiang system. Sr-Nd isotopic records from the nearby core DGKS9604 on the western slope of the middle OT (Figure 6C, core location in Figure 1; Dou et al., 2012) further support the interpretation that sediments deposited during this period were likely predominantly sourced from the Changjiang. The lower CaO/Al₂O₃ ratio (Figure 6D) suggests that biogenic carbonate production and/or preservation was suppressed, likely due to enhanced terrestrial detrital input (dilution effect) and/or relatively cold glacial marine conditions.

Stage 2 (11.6–8.7 ka BP): This stage corresponds to a period of rapid global sea-level rise (Figure 6E) (Lambeck et al., 2014; Li et al., 2014). The TFe₂O₃/Al₂O₃ ratio decreased sharply from its high values at the end of S1 (Figure 6B), marking the rapid attenuation of the S1 source input and instability in the transport process. Sediment discharged from the Changjiang began forming its modern delta at this time (Li et al., 2000, 2002). Mineralogical characteristics of core C1624 indicate a mixed provenance from the Changjiang and western Taiwan rivers during the early Holocene, with decreasing contribution from the Changjiang and increasing influence from western Taiwan rivers (Wang et al., 2025). The Sr-Nd isotopic data from core DGKS9604 began a fundamental shift: ϵNd values started a significant positive excursion, accompanied by a synchronous decrease in ⁸⁷Sr/⁸⁶Sr ratios (Figure 6C), indicating a markedly increased contribution of materials from Taiwan (Dou et al., 2012). This shift is attributed to profound changes in sediment dispersal patterns, driven by rapid sea-level rise. The inundation of the previously exposed continental shelf displaced the Changjiang estuary farther from the study area. Concurrently, the KC strengthened and/or re-entered the OT in the later phase of the deglaciation, acting as an effective agent for transporting Taiwanese materials northward (Dou et al., 2010b; Wang et al., 2024). These dramatic changes in paleogeography and circulation patterns profoundly reconfigured sediment transport pathways. The CaO/Al₂O₃ ratio showed a rapid increasing trend during this stage (Figure 6D). This may reflect the recovery of marine productivity under a warming climate (Figure 6A) (Dykoski et al., 2005; North Greenland Ice Core Project Members, 2004), and is also attributable to the sharp decrease in terrestrial detrital input flux, which reduced the dilution of biogenic components.

Stage 3 (8.7–0 ka BP): Entering the mid Holocene, sea level stabilized at a high level (Figure 6E) (Lambeck et al., 2014; Li et al., 2014), and the modern circulation pattern became established (Kim and Kucera, 2000; Li et al., 2009), with the main axis of the KC stably flowing through the OT (Kao et al., 2006b; Lee et al., 2013). Under this circulation regime, sediments from the Changjiang are largely deposited near the estuary or transported southwestward to form the mud belt of the ECS inner shelf (Bian et al., 2013; Liu et al., 2007b; Zhang et al., 2023). Conversely, sediments from western Taiwan can reach the OT region via the Taiwan warm current (Dou et al., 2016; Hsiung and Saito, 2017; Wang et al., 2024). Studies of drifter trajectories in the Taiwan Strait and ocean circulation models around Taiwan show that sediments from western Taiwan flow northeastward through the Taiwan Strait, then enter the area north of Taiwan, eventually reaching the southern and middle OT via the main Kuroshio stream (Jan et al., 2002; Tseng and Shen, 2003). The TFe₂O₃/Al₂O₃ ratio remained generally low during S3 (Figure 6B). The Fe₂O₃/TiO₂ ratio in core C01 was also in a low-value range during this stage compared to the last glacial and deglacial periods (Xu et al., 2025). The stable Sr-Nd isotopic signature observed in core DGKS9604, with ⁸⁷Sr/⁸⁶Sr ratios confined to a narrow range (0.7107–0.7112) and ϵNd values remaining high (averaging -10.7), clearly indicates an established and predominant contribution from Taiwanese sources (Dou et al., 2012).

In summary, the record from core C1624 reveals a multi-stage control pattern on sediment input in the southern middle OT since the LGM. On glacial-interglacial scales, sea-level change emerges as the paramount factor reshaping the regional depositional pattern and driving provenance transitions.