Holocene Organic Carbon Source−Sink Dynamics in the North Yellow Sea: Influences of East Asian Summer Monsoon and Sea-Level Change

Abstract

The Holocene evolution of organic carbon (OC) sources in the North Yellow Sea remains poorly constrained. In this study, a sedimentary dataset from core WHD01 retrieved from the Shandong Peninsula Mud Wedge is presented. After correcting grain-size, diagenetic and provenance biases in geochemical proxies and removing diagenetic offsets of sedimentary δ¹³C signals, the MixSIAR Bayesian mixing model was used to quantify three OC endmembers. The results reveal three distinct evolutionary stages of OC composition: pre-10 cal ka B.P. dominated by terrestrial C₃ OC; 10–4.2 cal ka B.P. dominated by marine OC during rapid sea-level rise; post-4.2 cal ka B.P. marine-derived OC remained dominant while C₃ terrestrial and estuarine OM increased alongside enhanced OC burial, driven by anthropogenic catchment erosion and improved organic preservation. Temporal shifts in OC source apportionment and burial flux are tightly coupled to East Asian Summer Monsoon variability and sea-level change, with conspicuous OM compositional reorganizations coinciding with the 8.2 ka and 4.2 ka climatic cold events.

1. Introduction

Carbon cycling is a core biogeochemical process regulating atmospheric CO₂ levels, global climatic stability, and marine ecosystem functioning. As a critical link connecting atmosphere–ocean–land–biosphere material and energy exchange, it sustains Earth’s habitability [1,2]. Marginal seas represent a key nexus of land–sea interactions and major transport conduits for terrigenous materials to the open ocean. Although continental shelves occupy only ~20% of the global ocean area, they receive over 90% of global fluvial sediment and sequester more than 80% of sedimentary organic carbon (OC), constituting Earth’s most efficient and largest sedimentary carbon sinks with enormous carbon storage potential [3,4]. Shelf sedimentary organic matter (OM) comprises both allochthonous terrigenous/estuarine inputs and autochthonous marine phytoplankton production. Such dual land–sea sourcing leads to complex OM composition and dynamic cycling, making shelf systems critical for understanding and predicting global carbon cycle feedbacks [5,6].

Notably, OC burial on East China continental shelves is predominantly concentrated in muddy depositional zones, which therefore serve as ideal archives for investigating shelf carbon sink processes. The East China marginal shelves were largely shaped by Holocene sea-level fluctuations and modulated by the East Asian Monsoon (EAM). Supplied primarily by massive fluvial inputs from the Yellow Rivers (YR), Yangtze River, and their major drainage system, these shelves contain a series of patchy muddy depositional belts governed by sea-level fluctuation, warm current and coastal current dynamics [7]. Seven typical muddy zones have been identified in the eastern China marginal seas: the Bohai Central Mud, Shandong Peninsula Mud Wedge (SPMW), Eastern Yellow Sea Mud, Southern Yellow Sea (SYS) Central Mud, East China Sea Inner Shelf Mud, Southwest Jeju Island Mud, and Beibu Gulf Mud [8,9,10,11] (Figure 1a). As a typical muddy deposit on the eastern China shelf, the SPMW lies at the convergence of YR inputs, coastal currents, and the Yellow Sea Warm Current (YSWC) [12,13,14]. It preserves continuous high-resolution Holocene sediments and records clear signals of terrestrial, estuarine, and marine OC inputs. Investigating OC evolution in the SPMW is crucial for revealing land–sea interaction processes, distinguishing the coupled impacts of the EAM and sea-level change, and understanding the formation mechanism of shelf sedimentary carbon sinks.

Previous investigations have primarily focused on SPMW, reconstructing Holocene OC burial processes and identifying potential controls such as sea-level change, the East Asian Winter Monsoon (EAWM), ENSO variability, and modern aquaculture impacts [18,19,20,21]. However, current studies remain limited by uneven spatial coverage concentrated in the SPMW [7,13], biased OC source estimates caused by inconsistent endmember (EM) values [14,19], unquantified early diagenetic effects on carbon isotopes, and poorly constrained coupling mechanisms between monsoon evolution and sea-level change [22,23]. As such, critical knowledge gaps remain regarding the quantitative evolution of terrigenous, estuarine, and marine OC EMs across the Holocene and the associated driving mechanisms of OC source–sink dynamics within the SPMW. To address these scientific questions, this study targets the WHD01 sediment core from the SPMW, with three main objectives: (1) to reconstruct Holocene OC source evolution by applying δ¹³C–δ¹⁵N isotopes and MixSIAR Bayesian mixing model for quantitative EM apportionment; (2) to infer changes in catchment weathering intensity, terrigenous input flux, and marine productivity using Rb/Sr, Sr/Ca, Ti/Al geochemical proxies; (3) to establish the linkage between East Asian Summer Monsoon (EASM) evolution, sea-level changes, and OC burial dynamics, and clarify the coupled controlling mechanisms governing Holocene OC source−sink processes in the SPMW.

2. Geological Setting

The northern Yellow Sea (NYS) is a semi-enclosed shallow shelf sea bounded by the Liaodong Peninsula to the north, the Korean Peninsula to the east, the Bohai Sea to the northwest via the Bohai Strait, the Shandong Peninsula to the west, and the SYS to the south by the Sulu Orogenic Belt-Qianliyan Uplift. As a typical low-relief dustpan-shaped depression in the western Pacific margin [24,25,26,27,28,29,30,31,32], it has an average water depth of ~38 m and a maximum depth of ~80 m [33]. The NYS receives abundant terrigenous materials from the YR, Yalu River, and rivers around the Shandong and Korean Peninsulas. Meanwhile, the YSWC can transport Yangtze River-derived sediments northward, forming a complex multi-source sedimentary system.

The regional circulation is dominated by the YSWC and shelf coastal currents. The YSWC, a branch of the Kuroshio current (KC), intrudes northwestward into the SYS and further propagates into the NYS along the Yellow Sea Trough, with pronounced seasonal variation (strong in winter and weak in summer) [12,34,35] (Figure 1a). The Shandong Coastal Current (SCC) flows eastward along the northern Shandong coast and turns southwestward, eventually converging with the YSWC to form a cyclonic circulation pattern that dominates sediment transport and redistribution [24] (Figure 1b).

The Core WHD01 is strategically located in the SPMW (Figure 1b). This site is uniquely positioned to differentiate terrestrial input from the YR, estuarine phytoplankton signals, and marine autochthonous OC. It lies within the main deposition convergence zone controlled by both riverine supply and YSWC-SCC dynamic circulation, with relatively stable sedimentary conditions. Therefore, the WHD01 sequence can provide an ideal archive for reconstructing evolutionary changes in sedimentary environments and OC sources.

3. Materials and Methods

Sediment core WHD01 (37.963187° N, 122.715566° E) was acquired by the Yantai Coastal Geological Survey, China Geological Survey (YCGS, CGS) aboard the research vessel Geology 17 during 2024–2025 (Figure 1b). The core was drilled at a water depth of 49.5 m, with a total length of 210 m, representing the longest sediment core ever retrieved in the NYS to date. For the present study, we focused on the core depth interval of 0–14 m (corresponding to elevation 49.5–64 m), which constitutes the Holocene sedimentary sequence.

Upon recovery, the core was split longitudinally, visually described, and subsampled at 2 cm intervals for geochemical analyses. All subsamples were stored at 4 °C to preserve OM integrity. Differentiated subsampling intervals were adopted for specific analytical targets, considering research funding constraints and variations in experimental testing costs: grain size measurements at 8–12 cm intervals, total organic carbon (TOC), total nitrogen (TN) and their isotope analyses at 10–15 cm intervals, and major/trace element analyses at 50–60 cm intervals. For chronological framework establishment, benthic foraminifera and plant detritus were hand-picked from selected sediment layers for accelerator mass spectrometry (AMS) ¹⁴C dating.

3.1. AMS¹⁴C Dating

AMS ¹⁴C dating was performed at Beta Analytic Inc. (Miami, FL, USA). Prior to dating, benthic foraminifera and plant detritus samples underwent rigorous preprocessing: sediment samples were ultrasonically cleaned with methanol to remove contaminants, dried at 50 °C, and reacted with concentrated phosphoric acid at 70 °C for 20 min to release CO₂. The CO₂ was purified and collected in a vacuum line, then graphitized using the Zn-TiH₂-Fe method at 500 °C for 3 h, followed by 550 °C for 4 h.

AMS ¹⁴C measurements were conducted using a compact 200 kV MICADAS (MIni CArbon DAting System) accelerator mass spectrometer (Ionplus AG, Zurich, Switzerland), operating in cycle mode with a single cycle duration of ~3 min. Measurements were terminated when the net ¹⁴C count reached 160,000 (statistical precision: 2.5‰). The long-term ¹⁴C/¹²C measurement precision of the instrument is better than 3‰, with a background Fm value of 0.0004 (corresponding to ~63,764 a). For ¹⁴C dating of samples from the NYS, a regional marine reservoir correction (ΔR = −139 ± 59 a) [36] was adopted, with this regional average value derived from published Yellow Sea and Bohai Sea datasets [12,19]. All ¹⁴C ages were calibrated to calendar ages (cal. a B.P., referenced to 1950 CE) via CALIB 7.0.2 [37] using the updated MARINE 20 calibration curve, and corresponding 1σ uncertainties were calculated simultaneously.

3.2. Grain Size Analysis

Grain size analysis was conducted at the YCGS. Raw sediment samples underwent a rigorous preprocessing procedure prior to analysis: (1) a homogenized subsample of known weight was taken; (2) a 6% hydrogen peroxide (H₂O₂) solution was added to remove OM; (3) a 0.2 N hydrochloric acid (HCl) solution was used to dissolve carbonate cements and biogenic shells; (4) the sample was repeatedly rinsed with deionized water until the rinse solution reached neutral pH; (5) a 0.5 N sodium hexametaphosphate (Na₆P₆O₁₈) solution was added, and the sample was soaked for 24 h to achieve complete dispersion.

Grain size was measured using a Mastersizer 2000 laser particle size analyzer (Malvern Panalytical, Worcestershire, UK). The instrument covers a measurement range of 0.5–2000 µm, with a relative error of repeated measurements less than 1%. Grain size parameters, including mean grain size (Mz), sorting, skewness, kurtosis, were calculated according to the method of Folk and Ward [38].

3.3. Geochemical Analyses

Major and trace element analyses were carried out at the YCGS using an Axios X-ray fluorescence (XRF) spectrometer (PANalytical, Almelo, The Netherlands). Prior to analysis, samples were subjected to drying and pelletizing processes. The acid digestion procedure was performed as follows: (1) 4 mL nitric acid (HNO₃) and 1 mL perchloric acid (HClO₄) mixed acid were added to the sample; (2) subsequently, 4 mL hydrofluoric acid (HF) and 1 mL perchloric acid (HClO₄) mixed acid were added; (3) finally, 10 mL nitric acid (HNO₃) was supplemented to complete the digestion.

These elemental ratios (Rb/Sr, Sr/Ca, Ti/Al) serve as reliable indicators of catchment chemical weathering, terrestrial input, and marine influence, which are consistent with the results of OC isotopes and EM apportionments. The Rb/Sr ratio reflects catchment chemical weathering intensity: during silicate weathering, Sr²⁺ is leached and mobilized, while Rb⁺ is retained in clay minerals. Thus, elevated Rb/Sr ratios indicate intensified chemical weathering and can further record variations in paleoclimate and fluvial runoff [39]. The Sr/Ca ratio serves as an indicator of marine productivity: since phytoplankton preferentially utilize Ca over Sr, lower Sr/Ca values correspond to higher biological activity [40]. The Ti/Al ratio represents terrigenous clastic input: higher values indicate enhanced terrestrial flux, whereas lower values suggest reduced terrestrial supply and a greater contribution from marine authigenic and chemical components [41].

3.4. TOC, TN, δ¹³C, and δ¹⁵N Analyses

Analyses of TOC, TN contents, and stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopes were conducted at the Yantai Institute of Coastal Zone Research, Chinese Academy of Sciences (YICZR, CAS). Prior to analysis, sediment samples were freeze-dried using a Genesis G35EL pilot-scale vacuum freeze dryer (SP Scientific VirTis, Gardiner, NY, USA), then ground to a 100-mesh sieve. The mass of an empty centrifuge tube was first recorded as m1. Approximately 1–2 g of the sieved freeze-dried sample was transferred into a centrifuge tube, and the combined mass of the sample and centrifuge tube was recorded as m2. To remove inorganic carbon, 6 mol/L HCl was added dropwise to the tube while gently shaking to ensure sufficient reaction between the acid and carbonate minerals. The tube was then left to stand for 1–2 days to complete the reaction, followed by centrifugation to separate the solid sample from the supernatant. After removal of inorganic carbon, the residue was rinsed with ultra-pure water via centrifugation and supernatant discard, repeated 1–2 times to eliminate residual acid. The treated samples were then dried either by freeze-drying for 24 h or in an oven at a temperature of <60 °C. The combined mass of the acid-treated sample and centrifuge tube was recorded as m3. The OC content measured directly by the instrument was denoted as C₂, and the OC content (C₁) of the original sediment sample was calculated using the following Formula (1):

C₁ = C₂ × (m3 − m1)/(m2 − m1)

(1)

subsequently, the pretreated samples were subjected to TOC and TN content analysis using a TOC-VCPH analyzer (Shimadzu Corporation, Kyoto, Japan), which was coupled with a MAT 253 gas stable isotope ratio mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) for the determination of δ¹³C and δ¹⁵N. The analytical precision, determined by duplicate analyses of samples, was ±0.02‰ for δ¹³C and δ¹⁵N, and ±0.01% for TOC and TN, respectively, calibrated against international standards (Pee Dee Belemnite for carbon, atmospheric nitrogen for nitrogen).

3.5. OC Source Apportionment Using the MixSIAR Model

To quantitatively assess the relative contributions of different OM EM to sedimentary OC, the Bayesian isotope mixing model MixSIAR (v3.1.12) was employed. MixSIAR is an open-source R package (v4.5.3) designed for multi-source quantitative apportionment using isotope tracers, which provides results in the form of probability distributions and quantifies uncertainty—addressing the limitations of traditional linear mixing models.

3.5.1. Model Principle

A core assumption of the model is that the stable isotope composition (δ_S) of sedimentary OC is a mixture of multiple OM sources, following the linear mixing equation:

$${\delta }_S=f_1\left({\delta }_1+{\gamma }_1\right)+f_2\left({\delta }_2+{\gamma }_2\right)+\cdots +f_i({\delta }_i+{\gamma }_i)$$

(2)

where $f_i$ is the relative contribution of the i-th OM EM to sedimentary OC (with $\sum f_i=1$), δ_i is the characteristic isotope value of the i-th EM, and ${\gamma }_i$ is the corresponding fractionation factor.

Based on Bayesian theorem, the probability of a specific contribution f_q is calculated as:

$$\mathrm{P}\left(f_q|x\right)={\displaystyle \frac{L\left(x|f_q\right)·p\left(f_q\right)}{\sum L\left(x|f_q\right)·p\left(f_q\right)}}$$

(3)

where x is the measured isotope value (δ¹³C or δ¹⁵N) of sedimentary OC, f_q is the contribution of the $q$-th OM source, L(x∣f_q) is the likelihood function of x given f_q, p (f_q) is the prior probability distribution of f_q, and ∑L(x∣f_q)·p(f_q) represents the marginal likelihood of x.

3.5.2. Diagenetic Fractionation Correction of δ¹³C

Diagenetic fractionation parameters were calculated using the least-altered layer as the baseline, representing the least-modified depositional environment. The fractionation strength Δ(‰) was defined as the difference between the mean δ¹³C of each stratigraphic unit and the baseline values:

Δδ¹³C = δ¹³C_unit − δ¹³C_baseline

(4)

where Δδ¹³C is the diagenetic fractionation strength (deviation from baseline, ‰); δ¹³C_(unit) is the mean δ¹³C value of a given stratigraphic unit; δ¹³C_(baseline) is the mean δ¹³C value of the unaltered baseline layer. The fractionation coefficient α was calculated as:

α = (1 + δ¹³C_baseline/1000)/(1 + δ¹³C_unit/1000)

(5)

where α is the diagenetic fractionation coefficient (isotopic exchange coefficient). The weighted contribution of each unit was obtained by multiplying Δδ¹³C by the number of samples, and the fractionation contribution rate was expressed as a percentage of the total weighted contribution.

These parameters were used to correct the δ¹³C values prior to MixSIAR modeling, removing diagenetic biases to ensure accurate OM source apportionment.

3.5.3. Model Setup and Validation

The δ¹³C and δ¹⁵N values of sediment samples and identified OM EMs were imported into the MixSIAR model for OM source apportionment. To avoid artificial constraints and unstable partitioning caused by overlapping isotopic signatures, uninformative Dirichlet priors (α = 1) were adopted instead of fixed zero priors, which may cause unrealistic unbounded solutions.

The model was configured with three Markov chains and run with a very long MCMC simulation: Chain Length = 1,000,000; Burn-in = 500,000; Thin = 500. Isotopic uncertainties of EMs and samples were fully incorporated into the model simulation to account for analytical errors and natural variability in source signatures. Model convergence was validated via z-score diagnostic (within ±2) [42] and Gelman-Rubin potential scale reduction factor (PSRF < 1), with effective sample size (ESS > 1000) [43], confirming stable and reliable outputs [44]. The mean of the posterior distribution was taken as the final relative contribution of each OM EM.

3.6. Calculation of OC Burial Flux

The water content (W) and dry bulk density (${\mathsf{\rho }}_d$) of sediment samples were calculated using Equations (6) and (7):

$$\mathrm{W}=({\mathrm{m}}_{\mathrm{w}}/{\mathrm{m}}_{\mathrm{d}})-1$$

(6)

$${\mathsf{{\rm P}}}_d={\displaystyle \frac{{\mathsf{\rho }}_w{\mathsf{\rho }}_s}{W{\mathsf{\rho }}_s+{\mathsf{\rho }}_w}}$$

(7)

where m_w is the mass of wet sediment, m_d is the mass of dried sediment, ρ_s= 2.65 g/cm³ is the sediment grain density, and ρ_w = 1.03 g/cm³ is the density of seawater.

The OC burial flux (${\mathrm{B}}_{\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{x}}$) was then determined according to Gao et al., [7]:

$${\mathrm{B}}_{\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{x}}={\mathrm{T}\mathrm{O}\mathrm{C}}_{\mathrm{sediment}}\times \mathsf{\rho }\times \mathsf{\nu }$$

(8)

where ${\mathrm{B}}_{\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{x}}$ is the OC burial flux (g/cm²/a), TOC_sediment is the sedimentary TOC content (%), $\mathsf{\rho }$ is the dry bulk density (g/cm³), and ν is the sedimentation rate (cm/a).

4. Results

4.1. Chronostratigraphy and Sedimentary Sequence

Based on AMS ¹⁴C dating results of core WHD01 (

Table 1. AMS¹⁴C dating data of core WHD01.

Beta ID	Depth (m)	Material	Measured Radiocarbon Age (a)	Calibration (a B.P.)	1σ	Sedimentation Rate (cm/a)
763140	50.34	Foraminifera	1030 ± 30	584	514–650	14
763141	52.31	Foraminifera	1780 ± 30	1317	1228–1405	27
763142	54.33	Foraminifera	2260 ± 30	1858	1745–1957	37
756867	55.92	Foraminifera	2890 ± 30	2639	2535–2744	20
763125	56.7	Foraminifera	4440 ± 30	4598	4478–4719	4
756863	57.49	Foraminifera	5680 ± 30	6038	5930–6136	5
756864	58.32	Foraminifera	6110 ± 30	6488	6388–6595	18
763126	59.08	Foraminifera	6660 ± 30	7108	7009–7217	12
763127	59.9	Foraminifera	7640 ± 30	8063	8156–8971	9
756865	60.73	Foraminifera	8820 ± 30	9467	9371–9548	6
763129	62.3	Foraminifera	9050 ± 30	9753	9613–9880	55
756866	63.12	Foraminifera	9140 ± 30	9890	9749–10,034	60
763130	63.72	Foraminifera	10,280 ± 30	11,442	11,292–11,570	4
728359	64.36	Plant material	10,940 ± 40	12,446	12,336–12,583	6
728362	64.69	Plant material	11,410 ± 40	12,875	12,763–12,962	8

), a robust Holocene chronostratigraphic framework was established via piecewise linear age–depth interpolation between adjacent calibrated radiocarbon anchor points (Figure 2). According to this chronological calibration, the Holocene sedimentary succession is divided into three principal depositional stages, including the Early Holocene (12–8.2 ka cal B.P.), Middle Holocene (8.2–2.8 ka cal B.P.) and Late Holocene (2.8 ka cal B.P.—present). The Early Holocene interval can be further split into two sub-stages with a chronological boundary at 10 ka. A higher sedimentation rate prevailed during the Early Holocene, followed by an obvious decrease to relatively low values in the Middle Holocene. Sedimentation rates began to rise in the early Late Holocene and subsequently experienced a gradual decline in the Late Holocene. The depositional evolution of the core WHD01 reveals a continuous Holocene sequence that is broadly consistent with adjacent cores (NYS101 [12,15] and DLC70-2 [17,45]), reflecting regionally coherent changes in sediment supply, hydrodynamics, and relative sea level.

(1)

Early Early Holocene

WHD01 (64.27–63.1 m, 12–10 ka cal B.P.), the lowermost interval of this core consists of yellowish-green to grayish-green clay and bioturbated silty clay, with extremely high sand and silt contents, poor sorting, and sedimentation rate peaking at ~60 cm/a (Figure 2). These features are indicative of a fluvial-dominated continental–estuarine setting prior to Meltwater Pulse (MWP)-1B, marked by rapid terrigenous infilling and weak marine influence, representing a freshwater terrigenous phase before the main Holocene transgression. The NYS101 core (66.6–62.9 m) records a coeval fluvial-dominated interval with similarly high sand–silt contents and elevated sedimentation rate, while DLC70-2 (57.6–54.5 m) only begins continuous Holocene deposition at ~11 ka, indicating that this high-energy terrigenous phase was most strongly expressed in the proximal WHD01 and NYS101 locations (Figure 3).

(2)

Late Early Holocene

WHD01 (63.1–60 m, 10–8.2 ka cal B.P.) is dominated by dark gray silt and silty clay with abundant shell debris, stable silt content (~75%), improved sorting, and sharply reduced sedimentation rate (Figure 2). This transition reflects the establishment of a tide-dominated estuarine-bay environment following the early Holocene transgression. NYS101 records a similar lithologic transition at ~62.9 m, accompanied by a decline in sedimentation rates, and DLC70-2’s late Early Holocene deposits are dominated by silty clay and shell debris (Figure 3), consistent with regional expansion of the estuarine system.

(3)

Middle Holocene

WHD01 (60–56 m, 10–4.2 ka cal B.P.) consists of homogeneous dark gray silt, locally intercalated with thin beds of silty fine sand, well-developed bioturbation, stable shell debris, stable grain-size distributions, slightly decreased sorting coefficients, near-zero skewness (weakly positive), and stable kurtosis (~0.95), with steady sedimentation rates (10–20 cm/a) (Figure 2). These features reflect a fully marine inner shelf setting under highstand conditions. NYS101 (61.6–49 m) contains a comparable interval, where the lower section consists of ~8 m of gray clay overlain by ~5 m of clayey silt. DLC70-2 (53.8–53 m) records continuous Middle Holocene deposition with similar lithology and bioturbation (Figure 3), confirming regionally uniform tidal dynamics and sediment supply.

(4)

Late Holocene

WHD01 (56–50 m, 4.2 ka cal B.P. to present) is dominated by clayey silt, with reduced sand fractions, further improved sorting, and stable skewness and kurtosis. Sedimentation rates, which were high in the early stage (37 cm/a), gradually declined and stabilized at 15 cm/a (Figure 2). These features are indicative of a modern inner shelf environment under diminished fluvial influence. A correlative Late Holocene record is absent in NYS101, while DLC70-2 preserves a limited Late Holocene succession (53–52.4 m), consisting of clayey silt-dominated deposits under stable highstand conditions (Figure 3).

4.2. Geochemical Elements Results

4.2.1. Multi-Factor Correction of Geochemical Proxies

Sedimentary geochemical proxies are inherently susceptible to three major confounding factors: grain-size sorting, sediment provenance shifts, and post-depositional diagenesis. Grain-size fractionation can alter elemental ratios via mechanical mineral partitioning; provenance heterogeneity may shift baseline elemental compositions; and late diagenesis coupled with redox processes tends to induce element remobilization and recombination, thereby obscuring primary palaeoenvironmental signals. Without prior identification and elimination of these biases, the interpretations of Rb/Sr, Sr/Ca, and Ti/Al in terms of palaeoweathering, marine palaeoproductivity and terrigenous input would remain ambiguous and non-unique.

To constrain the potential influences of grain size, diagenesis and provenance on geochemical proxies, systematic correlation analyses were performed for Rb/Sr, Sr/Ca and Ti/Al (Figure 4). Correlation with Mz shows that Sr/Ca is barely affected by particle size (R² = 0.10, Figure 4a), whereas Rb/Sr (R² = 0.4986, Figure 4b) and Ti/Al (R² = 0.7495, Figure 4c) exhibit moderate and strong grain-size dependence, respectively. Accordingly, these two proxies were corrected using the regression residual method. After correction, the correlation between the Rb/Sr_corrected and Mz was almost completely eliminated (R² ≈ 0.00000002, Figure 4d), while that of the Ti/Al_corrected was reduced to an extremely low level (R² = 0.1365, Figure 4e), demonstrating a robust grain-size correction effect.

Prior to grain-size correction, the original Rb/Sr, Ti/Al and Sr/Ca all exhibited weak correlations with the redox-sensitive Fe/Mn ratio (R² = 0.26, R² = 0.0087, and R² = 0.1111, respectively, Figure 4f–h), indicating inherently minimal diagenetic overprinting. Notably, the original Ti/Al also showed a weak correlation with the provenance-sensitive Ti/Zr ratio (R² = 0.1801, Figure 4k), reflecting stable sediment provenance independent of grain-size effects. After grain-size correction, the correlations between geochemical proxies and diagenetic/provenance indicators remained consistently weak (i.e., no significant enhancement or qualitative change): both Rb/Sr_corrected and Ti/Al_corrected still displayed negligible or weak correlations with Mn/Fe (R² = 0.3021, R² = 0.0123, respectively, Figure 4i,j), while the Ti/Al_corrected showed an even weaker correlation with Ti/Zr (R² = 0.0368, Figure 4l). These results collectively confirm that grain-size correction only eliminates grain-size-induced biases without altering the inherent signals of diagenetic stability and provenance consistency.

Overall, Rb/Sr_corrected and Ti/Al_corrected, together with the original Sr/Ca, have effectively mitigated interferences from grain size, diagenesis and provenance. These proxies can thus be reliably applied to reconstruct palaeoweathering intensity, terrigenous sediment supply, and marine palaeoproductivity.

4.2.2. Spatiotemporal Variations of Geochemical Proxies

During the early Early Holocene, the Rb/Sr_corrected ratio increased from −0.174 to −0.128 (Figure 5a). Sr/Ca values rose continuously from 36.64 to 44.64 (Figure 5b), while the Ti/Al_corrected ratio declined from 0.0037 to −0.0012 (Figure 5c).

In the late Early Holocene, the Rb/Sr_corrected ratio displays a fall-rise-pattern, evolving from 0.123 to −0.251 and subsequently to −0.011 (Figure 5a). The Sr/Ca ratio follows a three-step shift: decreasing from 54.25 to 50.3, peaking at 60.88, then falling to 48.85 (Figure 5b). Ti/Al_corrected remains nearly invariant at first, undergoes modest depletion, and finally rebounds, ranging from 0.0003 via −0.0002 to 0.0005 (Figure 5c).

Through the Middle Holocene, the Rb/Sr_corrected ratio hovers consistently near 0.02, punctuated by a distinct minimum of ~−0.057 at 6–7 ka cal B.P. (Figure 5a). The Sr/Ca ratio drops rapidly from 60 to 48.85 before settling around a baseline of ~50 for the remainder of the interval (Figure 5b). The Ti/Al_corrected ratio features a well-defined rise-fall-rise three-stage trend, varying sequentially from −0.0033 to 0.0005, −0.0031 and 0.003 (Figure 5c).

For the Late Holocene, the Rb/Sr_corrected ratio oscillates within a narrow envelope of −0.02 and 0.02 (Figure 5a). The Sr/Ca ratio remains stable at first and shifts to large-amplitude fluctuations between 42 and 55 after 2.1 cal ka B.P. (Figure 5b). The Ti/Al_corrected ratio is constrained between 0.129 and 0.13 prior to 2.6 ka cal B.P., before switching to pronounced variability spanning −0.0006 to 0.0033 in the later period (Figure 5c).

4.3. OC Results

4.3.1. Diagenetic Fractionation of δ¹³C

Significant depth-dependent δ¹³C and δ¹⁵N fractionation was observed in the core WHD01, with clear Holocene gradients (

Table 2. δ¹³C and δ¹⁵N characteristics and fractionation effects of OM in different Holocene stratigraphic stages.

Stratigraphic Stage	Number of Samples	δ13C Mean (‰)	δ15N Mean (‰)	α (δ13C)	α (δ15N)	Δδ13C (‰)	Δδ15N (‰)	Weighted Contribution	Fractionation Contribution (%)	Fractionation Grade
Late Holocene	49	−23.041	6.457	1	1	0	0	0	0	Baseline weak
Mid-Holocene	23	−23.054	7.972	1	1.002	−0.013	1.515	−0.299	1.049	Moderate
Late Early Holocene	13	−23.064	7.943	1	1.001	−0.023	1.486	−0.302	1.071	Moderate-strong
Early Early Holocene	16	−24.787	7.479	0.998	1.001	−1.746	1.022	−27.938	100	Strong

Note: Fractionation parameters are calculated relative to the Late Holocene baseline (0–7.24 m, δ¹³C = −23.041‰, δ¹⁵N = 6.457‰).

and Figure 5). Downcore, δ¹³C becomes progressively more negative, whereas δ¹⁵N becomes more positive (Figure 5d,e). These trends are consistent with prolonged microbial degradation and anoxic early diagenesis. Preferential microbial utilization of ¹²C and ¹⁴N enriches residual OM in ¹³C and ¹⁵N, leading to higher δ¹⁵N. The pronounced downcore depletion in δ¹³C, most evident in the early Early Holocene, is likely amplified by enhanced contribution from reworked, ¹³C-depleted aged terrestrial OC.

Diagenetic fractionation was quantified relative to a least-altered Late Holocene baseline (δ¹³C = −23.041‰, δ¹⁵N = 6.457‰). The late Holocene unit exhibits negligible fractionation (Δδ¹³C = 0‰, Δδ¹⁵N = 0‰), supporting its use as a primary EM. The middle Holocene and late Early Holocene show weak to moderate fractionation, reflecting moderate microbial activity under fluctuating redox conditions. The early Early Holocene exhibits the strongest diagenetic modification (Δδ¹³C = −1.746‰, Δδ¹⁵N = +1.022‰), corresponding to a “strong” alteration grade and 100% fractionation contribution.

All stratigraphic δ¹³C values were systematically corrected (Figure 5d) following Equations (4) and (5) based on calculated Δδ¹³C and fractionation coefficients (α). This correction removes diagenetic overprinting and recovers near-primary isotopic compositions, supporting robust OM apportionment.

4.3.2. TOC, TN, δ¹³C, and δ¹⁵N Variation

The δ¹³C_corrected values of core WHD01 range from −25.8 ± 0.059‰ to −22.26 ± 0.029‰ (mean = −23.11 ± 0.02‰). Values increased from −25.8 ± 0.059‰ to −22.59 ± 0.109‰ between 12–10 cal ka B.P., and stabilized near ~−22.5‰ after 10 cal ka B.P (Figure 5d). The δ¹⁵N values vary from 5.37 ± 0.05‰ to 8.49 ± 0.058‰ (mean = 7.16 ± 0.02‰), remaining relatively high and stable (~8.33‰) between 8.78–12 cal ka B.P., showing a decreasing trend (8.32 to 6.23‰) from 8.78 cal ka B.P. to the present (Figure 5e).

TOC contents range from 0.45 ± 0.098 wt% to 1.45 ± 0.057 wt% (mean = 0.79 ± 0.01 wt%), with low values (0.47–0.70 wt%) during the early Early Holocene, and stabilizing near 0.8 wt% after 10 cal ka B.P. (Figure 5f). TN contents range from 0.003 ± 0.00005 wt% to 0.124 ± 0.002 wt% (mean = 0.79 ± 0.01 wt%), generally low (0.01–0.05 wt%) before 10 cal ka B.P., stabilizing at 0.1 wt% only between 10–8.4 cal ka B.P., and slightly decreasing from 0.12 to 0.09 wt% after 8.4 cal ka B.P. (Figure 5g). C/N molar ratios fluctuated markedly (6.97–24.45) during the early Early Holocene, and subsequently stabilized with a mean value of ~8.96 (Figure 5h).

5. Discussion

5.1. Sources of OC

Qualitative source apportionment of Holocene sedimentary OC in core WHD01 was performed using dual δ¹³C_corrected–δ¹⁵N isotope EM tracing (Figure 6a), displaying typical multi-source mixing patterns that reflect variable terrigenous–marine contributions under changing land–sea interaction regimes through time.

During the early Early Holocene, sediment δ¹³C_ corrected values were significantly negative (−25.8 ± 0.059‰ to −23.878 ± 0.055‰), while δ¹⁵N increased from 5.37 ± 0.05‰ to 8.52 ± 0.079‰ (Figure 5d,e). Synchronously, TOC remains relatively low (0.47–0.70 wt%), TN shows persistently low concentrations (0.01–0.05 wt%), and C/N molar ratios fluctuate intensely across a wide range of 6.97–24.45 (Figure 5f–h). Isotopic data plot mainly between C₃ terrestrial plants and estuarine phytoplankton (Figure 6a), indicating that OC was dominated by terrestrial C₃ plants under low sea-level, fluvial-dominated conditions.

From the late Early Holocene to the Late Holocene, sediment δ¹³C_corrected values shifted to higher values (~−24‰ to ~−22‰), whereas δ¹⁵N stabilized then decreased to ~6‰ (Figure 5d,e). Accompanying this isotopic transition, TOC rises and stabilizes around a mean of ∼0.80 wt%; TN rises distinctly to peak near 0.10–0.12 wt% in the late Early–Middle Holocene and gently falls to ~0.09 wt% toward the Late Holocene; meanwhile, C/N molar values settle down to a steady average of ∼8.96 with greatly subdued fluctuation (Figure 5f–h). Data clustered primarily between the marine algae and estuarine phytoplankton EMs (Figure 6a), indicating that OC sources became dominated by aquatic autochthonous production from marine algae and estuarine phytoplankton, with weak terrigenous inputs. This corresponds to stable highstand conditions and a fully marine-influenced coastal environment.

Overall, the spatiotemporal variations in δ¹³C_corrected–δ¹⁵N isotopic compositions qualitatively reveal a continuous Holocene evolution of OC sources at WHD01 (Figure 6a): from terrestrial-dominated inputs during the Early Holocene, when fluvial systems prevailed under lowstand conditions, to aquatic-dominated sources in the Late Holocene, when marine phytoplankton and estuarine production became dominant under stable highstand conditions.

Terrestrial OM in sediments of the NYS domain is predominantly supplied by the YR, and surface vegetation in North China is overwhelmingly dominated by C₃ plants. Based on this observation and previous studies, isotopic parameters for each EM were defined as follows: C₃ terrestrial plants: δ¹³C = −27‰ to −26‰ (mean = −26.5‰, SD = 0.5‰), δ¹⁵N = 2‰ to 6‰ (mean = 4‰, SD = 2‰) [46]; Estuarine phytoplankton: δ¹³C = −24.5‰ to −23.5‰ (mean = −24‰, SD = 0.5‰) [47,48], δ¹⁵N = 4‰ to 8‰ (mean = 5‰, SD = 2‰) [49,50]; Marine algae: δ¹³C = −22‰ to −20‰ (mean = −21‰, SD = 1.0‰) [51], δ¹⁵N = 6‰ to 12‰ (mean = 9‰, SD = 3.0‰) [52]. The EM parameters used in the MixSIAR model are illustrated in Figure 6b.

Based on the MixSIAR EM analysis (Figure 7), marine algae dominate bulk sedimentary OC with a mean contribution of 55.60% (95% CI: 52.92–58.25%, SD = 1.36%). C₃ terrestrial plants and estuarine phytoplankton contribute nearly equally: mean proportions reach 22.10% (95% CI: 19.62–24.68%, SD = 1.29%) and 22.30% (95% CI: 19.41–25.26%, SD = 1.49%), respectively.

Correlation analysis shows a highly significant negative interdependence among the three OC sources. C₃ terrestrial plant input correlates negatively with estuarine phytoplankton (r = −0.56, p < 0.001) and marine algae (r = −0.52, p < 0.001). Such inverse covariation implies that enhanced terrigenous supply elevates water turbidity, restricts light penetration, and suppresses aquatic primary production via a dilution-inhibition effect. Meanwhile, estuarine phytoplankton and marine algae display a moderate significant negative correlation (r = −0.41, p < 0.001), reflecting competitive alternation between brackish-estuarine and fully marine producers and highlighting frequent salinity fluctuation across the study marginal marine environment.

These pairwise correlations support the interpretation that shifts in OC sources are driven primarily by changes in relative sea level, salinity regime, and terrigenous sediment supply, rather than independent productivity trends.

Temporal variations in OC EM proportions reveal stepwise shifts in sedimentary OC sources throughout the Holocene (Figure 8a–d).

During the early Early Holocene (before 10 cal ka B.P.), C₃ terrestrial plants dominated OC sources, with a mean contribution of 44.42% ± 16.01% (95% CI:12.63–78.36%). Estuarine algae (18.13% ± 14.90%, 95% CI:0.74–56.72%) and marine algae (37.45% ± 14.37%, 95% CI:7.84–63.90%) were subordinate (Figure 8a–c). High C₃ fractions, coupled with large SD values, are consistent with variable fluvial input under low post-glacial sea level [53], implying bulk OM was predominantly supplied by terrigenous C₃ higher plants. Limited marine inundation likely constrained aquatic primary production at this stage. Although there was a peak in terrigenous C₃ plant-derived OC during the early Early Holocene, in-situ B_flux remained relatively low (1–9 mg/cm²/a, Figure 8d).

From the Late Early Holocene (10–8.2 cal ka B.P.), terrigenous C₃ fractions declined sharply and remained persistently low. During 8.48–9.88 cal ka BP, C₃ terrestrial input decreased to an average of 18.27% (95% CI: 1.06–44.90%), while marine algae rapidly became the dominant EM with a mean proportion of 67.70% (95% CI: 41.75–89.65%); estuarine algae remained marginal at 14.03% (95% CI: 0.51–41.03%) (Figure 8a–c). Located far from the main YR depocenter, the study site mainly receives runoff from small Shandong Peninsula catchments, which may have limited terrestrial organic supply. Post-glacial sea-level rise was associated with widespread marine transgression after 6 cal ka BP [54]. B_flux initiated with relatively high burial values at the onset of this phase (35–41 mg/cm²/a), followed by a progressive decreasing trend toward a pronounced minimum near the 8.2 ka climatic event (~4–5 mg/cm²/a, Figure 8d).

In the Middle Holocene (4.2–8.2 cal ka BP), marine algae reached their Holocene maximum mean contribution of 68.32% (95% CI: 42.43–90.02%), with C₃ terrestrial plants (17.22%, 95% CI: 1.08–43.31%) and estuarine algae (14.46%, 95% CI: 0.61–41.14%) remaining consistently subdued (Figure 8a–c). Stabilized sea-level highstand was likely favourable for sustained marine algal proliferation and the formation of open-marine environments [55]. OC B_flux maintains persistently low magnitudes throughout this period (2–10 mg/cm²/a, Figure 8d). Although marine algal-derived OC dominated the source pool, well-oxygenated open-shelf hydrographic conditions rates could have restricted OC burial efficiency, consistent with the overall depressed B_flux values.

During the Late Holocene (post-4.2 cal ka B.P.), the OC budget remained marine-dominated, yet terrestrial and estuarine EM fractions exhibited moderate recovery. Marine algae contributions declined slightly to a mean value of 55.00% (95% CI: 28.67–79.45%), accompanied by modest increases in both C₃ terrestrial plants (21.44%, 95% CI: 3.21–48.76%) and estuarine phytoplankton (23.56%, 95% CI: 2.15–51.32%) (Figure 8a–c). Persistent sea-level highstand promoted extensive marine inundation across the study area [55], whilst fluctuating riverine freshwater and nutrient input possibly enhanced nearshore estuarine productivity and terrigenous organic delivery. B_flux increased sharply in the late Late Holocene, forming the second Holocene high-value peak (15–27 mg/cm²/a, Figure 8d). Intensified anthropogenic erosion and land-use modification are plausible contributors to elevated catchment terrigenous OC export, which synchronously increased C₃ terrestrial and estuarine phytoplankton OC inputs to enhance sedimentary OC loading. Meanwhile, moderately increased sediment accumulation rates may have facilitated organic preservation, with these factors collectively contributing to the prominent late-stage burial enhancement.

5.2. Mechanisms Controlling OC Burial

Holocene climate and sea-level evolution appear to have exerted substantial potential influences on OC cycling across global marginal seas. Here, downcore records from core WHD01 in the NYS are integrated with Dongge Cave δ¹⁸O (a widely used EASM proxy) [25] and global sea-level compilations [55] (Figure 8) for regional comparison. Dongge Cave δ¹⁸O variations generally track large-scale EASM evolution, whereas sea-level fluctuations provide the boundary for post-glacial shelf transgression. Multiple proxies from WHD01 show coherent variations with regional monsoon and sea-level trends (Figure 8). Accordingly, the co-variation between EASM evolution and sea-level change is plausibly an important factor modulating Holocene shifts in OC sources and burial efficiency at this site.

OC sources of core WHD01 consist of C₃ terrestrial plants, estuarine phytoplankton and marine algae, whose proportional changes show good synchronism with monsoon variability and transgression–regression cycles, as summarized in the conceptual carbon-cycle model (Figure 9).

EASM variability is regarded as a key potential climatic factor modulating terrestrial OC delivery, correlating closely with catchment weathering, fluvial runoff and terrigenous clastics export [56,57]. Variations in monsoon strength coincide with two contrasting OC assemblages: predominant C₃ terrestrial input before 10 cal ka B.P. and elevated marine algae fractions during 10–4.2 cal ka B.P.

(1)

Before 10 cal ka B.P.: Co-variation of EASM and Sea-Level and Associated OC Source Transition.

At the onset, the Younger Dryas event (YD, 12.9–11.7 cal ka B.P.) was characterized by massive North Atlantic freshwater input derived from Laurentide and Lake Agassiz outflows via the St. Lawrence and Mackenzie drainage systems [58,59], which correlates with weakened EASM signals preserved in the Dongge Cave δ¹⁸O record (Figure 8h) [25]. Under this monsoon background, suppressed chemical weathering and dominant physical erosion are implied across the catchment. Limited runoff likely transported terrestrial plant debris efficiently, consistent with high C₃ terrestrial contribution (Figure 8a). Lowstand configuration restricted seawater intrusion, which probably limited the development of large-scale marine productivity.

The subsequent MWP-1B event (11.5–11.2 cal ka B.P.) corresponds to abrupt global sea-level acceleration [60]. Rising Rb/Sr_corrected ratios (Figure 8e) document progressive EASM intensification. Enhanced monsoonal precipitation likely boosted catchment chemical weathering and fluvial discharge, leading to strong physical dilution of land-derived organic fractions. Consequently, C₃ plant-derived OC declined substantially (Figure 8a). Despite persistent sea-level rise, limited shelf-ocean water exchange maintained relatively high Sr/Ca ratios (Figure 8f). Such semi-enclosed hydrodynamic conditions favored in situ marine phytoplankton production, driving an elevated fractional contribution of marine algal OC (Figure 8c).

Collectively, shifts in EASM intensity are closely correlated with altered sediment transport and aquatic productivity, coinciding with the OC source transition from C₃ terrestrial plants toward marine algae. Terrestrial-derived OM constituted the dominant carbon pool for the early-Holocene regional cycle (Figure 9).

(2)

Mid-Holocene (10–4.2 cal ka B.P.): Monsoon–Sea-Level Interaction and Cold-Event-Related OC Modification.

The interval of 10–9.5 cal ka B.P. features remarkably elevated OC B_flux, forming one of Holocene burial peaks (Figure 8d). Two interrelated factors may explain this pattern: (1) Postglacial transgression after ~12 ka coincided with accelerated sediment accumulation (Figure 2j), elevated sedimentation rates likely shortened organic degradation duration and improved preservation, while coarser grain size (Figure 2a) implies enhanced fluvial clastic and terrestrial OM supply; (2) Early-Holocene terrestrial OC was largely retained on nearshore shelves before being reworked and redeposited during the early Late Early Holocene, producing a lagged burial maximum decoupled from the peak of initial C₃ input.

The 8.2 ka cold anomaly lasted 100–200 yr, and δ¹⁸O records from Dongge Cave document a notable EASM reduction (Figure 8h) [25]. Such climatic deterioration is potentially linked to Laurentide Ice Sheet meltwater perturbation slowing Atlantic Meridional Overturning Circulation (AMOC) [26,27,28], southward Intertropical Convergence Zone (ITCZ) migration and intensified EAWM [29,30]. Regional sea surface temperature (SST) dropped by ~0.5 °C [31], and intensified EAWM signals have been detected in neighbouring shelf sediment and magnetic datasets. Reduced EASM-driven runoff likely lowers catchment weathering and the transport capacity of Shandong Peninsula small rivers, consistent with falling C₃ terrestrial fractions (Figure 8a). Diminished riverine nutrient input is probably responsible for restrained estuarine algal growth (Figure 8b). The cold event is thus correlated with disrupted terrestrial–marine OC balance and reduced land-derived OM flux. Together with prolonged transgression lengthening terrestrial transport distance and expanded well-oxygenated open shelf favoring organic decomposition, these multiple factors are consistent with the occurrence of Holocene burial minimum near 8.2 cal ka B.P. (Figure 8d).

The 4.2 ka cold shift represents another prominent mid-Holocene climatic transition. Multiple proxies record weakened EASM and strengthened EAWM around this period [25], possibly associated with reduced solar output, southward ITCZ migration and amplified El Niño–Southern Oscillation (ENSO) variability. Regional SST fell by approximately 1.5 °C [31], and adjacent shelf sequences register reduced terrigenous input and strengthened aeolian signals. Decreasing Rb/Sr_corrected ratios alongside elevated Ti/Al_corrected ratios from WHD01 further record this environmental shift (Figure 8e,f).

(3)

Late Holocene (4.2 ka to present): Variable EASM, Stabilized Highstand and Dynamic OC Burial.

Late Holocene OC cycling is closely associated with fluctuating EASM and persistent sea-level highstand. Elevated and variable Rb/Sr_corrected ratios, together with unstable Ti/Al_corrected ratios, imply irregular monsoon activity (Figure 8e,g). Moderate Sr/Ca ratios under stable highstand conditions suggest slightly weakened but sustained marine productivity compared with the mid- Holocene (Figure 8f). Termination of rapid post-glacial sea-level rise altered the modulating effect of estuarine front and freshwater input on coastal production, coinciding with modest growth of estuarine phytoplankton contribution, whereas marine algae remained dominant (Figure 8b). Variable terrestrial supply drove fluctuating burial efficiency, resulting in more dynamic OC accumulation relative to the mid-Holocene. This shift possibly marks a transition from sea-level-dominated toward nutrient- and hydrodynamic-controlled carbon cycling under persistent highstand conditions.

The conceptual model (Figure 9) synthesizes the above evolutionary pattern: early-Holocene OC was mainly sourced from terrestrial C₃ plants and estuarine algae; progressive transgression increased marine algal proportion during the middle Holocene, and both estuarine- and terrigenous-derived OM recovered moderately in the Late Holocene. In-situ OM degradation and CO₂ exchange further modulate TOC inventory, indicating that the study area functions as a transitional sink coupling terrestrial and marine carbon pools.

6. Conclusions

(1)

This study corrected grain size, diagenesis and provenance interference on geochemical proxies to reconstruct sedimentary evolution, terrigenous supply, and marine productivity within the SPMW. After eliminating OC fractionation and diagenetic deviation of downcore δ¹³C signals, a Bayesian mixing model was applied to quantitatively distinguish three OC EMs (EMs: C₃ terrestrial plants, estuarine phytoplankton, and marine algae) in core WHD01.

(2)

OC EM assemblages exhibit three distinct evolutionary phases across the Holocene. C₃ terrestrial dominates OC inventory prior to 10 cal ka B.P.; rapid sea-level rise facilitates the development of marine OC composition from 10 to 4.2 cal ka B.P.; relatively stable highstand after 4.2 cal ka B.P., marine-derived OC remained dominant whereas C₃ and estuarine OM rebounded. Intensified anthropogenic erosion, together with enhanced organic preservation, collectively boosted OC burial flux.

(3)

Holocene variations in OC sources and burial at the site show good synergy with EASM and sea-level evolutions. EASM fluctuations correlate closely with catchment weathering intensity, river runoff and nutrient export, whereas sea-level change is tightly linked to the magnitude of marine transgression and land-ocean connectivity. The 8.2 ka and 4.2 ka climatic cold anomalies were accompanied by remarkable shifts in regional setting and OC constituent proportions.