Provenance Evolution Since the Middle Pleistocene in the Western Bohai Sea, North China: Integrated Rare Earth Element Geochemistry and Sedimentological Records

Abstract

Despite extensive research on sediment provenance in the Bohai Sea (BS), a significant knowledge gap persists concerning long-term provenance evolution, particularly in the western BS since the Middle Pleistocene. This shortcoming limits reconstructions of paleoenvironmental evolution and its interplay with climatic variability and sea-level fluctuations. This study presents integrated Rare Earth Element (REE) geochemical and sedimentological analyses of sediments from core DZQ01 in the western BS. The mean ΣREE concentration of 178.78 μg/g is characterized by pronounced light REE (LREE) enrichment relative to heavy REE (HREE). Chondrite- and upper continental crust (UCC)-normalized patterns exhibit distinct negative Eu anomalies, variable Ce anomalies, marked LREE enrichment, and pronounced LREE/HREE fractionation. Grain size exerts the dominant control on REE distribution, whereas the weak correlation between HREE fractionation parameter indices (e.g., Gd/Yb) and redox-sensitive proxies (e.g., δEu_UCC and δCe_UCC) confirms their fidelity as provenance indicators. When integrated with the δEu_UCC-δCe_UCC diagram, discriminant functions, and paleoenvironmental proxies (Rb/Sr and Mg/Ca ratios), the data indicate that, during interglacial highstands, the Yellow River (YR) was the principal source, delivering fine-grained terrigenous material from the Loess Plateau and thereby elevating REE concentrations. Conversely, glacial lowstands shifted the depositional environment to subaerial conditions, with the YR, Hai River, and Luan River supplying a coarse-fine admixture. Multi-river provenance and dilution by coarse detritus collectively lowered REE concentrations during these intervals.

1. Introduction

The Bohai Sea (BS) is a semi-enclosed marginal continental-shelf sea [1] that functions as the estuary for large rivers such as the Yellow River (YR), Hai River (HR), Luan River (LR), and Liao River (LiaoR). It is also a critical confluence zone wherein the Yellow Sea Warm Current (YSWC) and the BS Coastal Current (BSCC) converge [2]. Throughout the Quaternary, the BS has experienced recurrent glacial–interglacial cycles [3,4], inducing repeated alternations and reworking of the sediment source [5,6]. The complex interplay among diversified sedimentary sources, variable hydrodynamic regimes, and fluctuating climatic conditions has rendered investigations into sediment provenance and flux systems of the BS increasingly pivotal.

Rare Earth Elements (REEs) efficiently preserve source signatures, making them valuable tracers of provenance. Consequently, their utility in sediment-provenance studies has been widely recognized [7,8,9]. Sedimentological and provenance-tracing investigations reveal heterogeneous provenances across the BS: the YR dominates sediment supply to the southern, southeast, and central BS, as evidenced by the strong geochemical affinity of core samples to YR-derived material. The LR directly affects deposits proximal to its estuary and exerts a secondary influence on the western BS. Sediments in the northern and central BS represent a mixture of YR and HR contributions, and the southwestern area of the BS predominantly receives YR detritus admixed with minor contributions from the HR and LR [10,11,12]. Previous investigations have concentrated on surface sediments or shallow cores [13,14,15], leaving research into long-term sediment-source evolution in the western BS, particularly since the Middle Pleistocene, conspicuously sparse. This knowledge gap constrains a comprehensive understanding of sediment-source evolution in the region.

Extending the record to the Middle Pleistocene, this study couples multi-proxy REE with microfossil analyses to quantitatively constrain provenance–climate relationships. This study integrates newly acquired data from borehole DZQ01 in the western BS with previously published chronological constraints [16]. Employing a multi-proxy geochemical approach—including REE abundances, Rb/Sr, and Mg/Ca ratios—this study aims to (1) quantify downcore variations in REE systematics of core DZQ01; (2) distinguish sediment contributions from the YR, HR, and LR across Marine Isotope Stage (MIS) 8–1; and (3) relate provenance shifts to regional paleoenvironmental evolutions. Consequently, the study elucidates both paleoenvironmental evolution and temporal changes in sediment provenance across the BS, thereby advancing understanding of the long-term development of its depositional system and its forcing by climate change, sea-level fluctuations, and associated boundary conditions.

2. Regional Setting

The BS comprises five major regions: Liaodong Bay, Bohai Bay, Laizhou Bay, the Bohai Strait, and the Bohai Central Basin (Figure 1) [5,17,18]. As an inland sea extending landward across the continental plain, the BS seafloor is predominantly filled by terrestrial deposits, yielding a characteristic coastal depositional environment [19]. Episodic rifting of the Bohai Basion since the Eocene has driven progressive thickening of the sedimentary succession, culminating in substantial Quaternary accumulations. These strata constitute a high-resolution archive of complex provenance and depositional dynamics [20], offering critical insight into eustatic fluctuations and paleoclimatic evolution, as well as recording the dynamic land–ocean transfer processes.

The BS is surrounded by four major rivers: the YR, HR, LR, and LiaoR (Figure 1). Consequently, sediment provenance has remained a central research theme in the BS [22,23,24]. The basin receives an enormous volume of terrigenous clastics, with the YR supplying the largest fraction and thus representing the principal sediment source [25,26]. The YR drains an area of 7.52 × 10⁵ km² across northern China and extends ~5465 km in length. Between 1950 and 2013, it discharged 29.65 km³/yr of water and 7.09 × 10⁸ t/yr of sediment. This fluvial influx, modulated by upstream climate variability and intensifying anthropogenic impacts, has profoundly influenced both provenance and depositional dynamics within the BS [27,28,29]. The resulting spatiotemporal heterogeneity in sediment properties not only yields critical insights into paleoenvironmental evolution and geological processes, but also offers valuable research opportunities in marine sedimentology and geochemistry.

3. Materials and Methods

Core DZQ01 was recovered in July 2022 by the Yantai Center of Coastal Zone Geological Survey (YCGS) at 38.76565° N, 118.262905° E in the Caofeidian area, western BS, in 23.1 m water depth (Figure 1). Drilling was conducted from the coastal research vessel Haiyang 66, penetrating to 203.15 m below seafloor (mbsf) and recovering 187.63 m of sediment (recovery = 92.36%). The core was subsampled at 2 cm intervals, major- and trace-element concentrations, including REE, were determined every 60 cm. To validate the provenance interpretations, 17 fluvial reference samples were collected proximal to the estuaries of the YR, HR, and LR for concurrent REE analyses (Figure 1).

3.1. Geochemical Elements Testing

All analyses were performed at the accredited testing facility of YCGS Testing Center. Sediment samples were oven-dried at 110 °C for 5 h and subsequently cooled to room temperature. Major and trace element concentrations—including Mg, Ca, Rb, and Sr—were determined by Axios X-ray fluorescence (XRF) spectrometry (Axios MAX4.0, PANalytical B.V., Alemlo, The Netherlands).

REE concentrations were obtained by iCAP RQ inductively coupled plasma mass spectrometry (ICP-MS, Thermo Fisher Scientific, Waltham, MA, USA) after a validated acid-digestion protocol. Each 50 mg aliquot was first treated with 4 mL concentrated HNO₃ and 1 mL concentrated HClO₄, followed by 4 mL concentrated HF plus 1 mL HClO₄, and finally brought to complete dissolution with 10 mL concentrated HNO₃. Blanks showed REE concentrations < 0.01 ppm; duplicates (n = 12) had relative standard deviations < 3%; detection limits ranged from 0.002 to 0.005 ppm. Fifteen REE (La–Lu plus Y) were quantified: La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, and Y. Total REE abundance (ΣREE) is defined as the sum of La–Lu concentrations, excluding Y. Light REEs (LREEs) comprise La–Eu, whereas heavy REEs (HREEs) comprise Gd–Lu [30,31].

δEu_UCC and δCe_UCC anomalies are employed as critical proxies for environmental fractionation processes: Ce anomalies reflect redox conditions, whereas Eu anomalies are associated with plagioclase dissolution. These two parameters are calculated as follows:

$${\mathsf{\delta }\mathrm{E}\mathrm{u}}_{UCC}={\mathrm{E}\mathrm{u}}_{UCC}/\sqrt{({Sm}_{UCC}·{Gd}_{UCC})}$$

(1)

$${\mathsf{\delta }\mathrm{C}\mathrm{e}}_{UCC}={\mathrm{C}\mathrm{e}}_{UCC}/\sqrt{({La}_{UCC}·{Pr}_{UCC})}$$

(2)

where ${\mathrm{E}\mathrm{u}}_{UCC}$, ${Sm}_{UCC}$, ${\mathrm{G}\mathrm{d}}_{UCC}$, ${\mathrm{C}\mathrm{e}}_{UCC}$, ${\mathrm{L}\mathrm{a}}_{UCC}$, and ${\mathrm{P}\mathrm{r}}_{UCC}$ denote upper continental crust (UCC) concentrations.

The discriminant function (DF) is applied to quantify the compositional similarity between the investigated samples and potential sources materials. Generally, |DF| < 0.5 indicates a strong provenance affinity and source relationship between the sample and the source [32,33], and lower absolute DF values denote greater provenance similarity. DF values were calculated following the algorithm presented in reference [34].

$$\mathrm{D}\mathrm{F}=\left|{(n}_{1\mathrm{M}}/n_{2\mathrm{M}})/({(n}_{1\mathrm{N}}/n_{2\mathrm{N}})-1\right|$$

(3)

In this context, $n_{1\mathrm{M}}/n_{2\mathrm{M}}$ denotes the concentration ratio of element 1 to element 2 in the studied sediment, whereas $n_{1\mathrm{N}}/n_{2\mathrm{N}}$ denotes the corresponding ratio in the candidate source material. To quantify the relative contribution of each river to the sediment archive at discrete stratigraphic intervals, DF values were converted into source contribution fractions via reciprocal transformation and subsequent normalization as follows:

Actual Proportion = (1/DF)/Σ(1/DF)

(4)

This procedure yields independent of the relative contributions of each river to the sedimentary sequence. The resulting source proportions are illustrated as a stacked chart versus depth.

3.2. Microfossil Analysis

Microfossil analyses were performed at Qingdao Sibada Analysis and Testing Co., Ltd. (Qingdao, China). Foraminifera and ostracod residues were oven-dried at 60 °C. A 25 g aliquot of dried sediment was transferred to a beaker, soaked in deionized water for 24 h to achieve complete disaggregation, and then repeatedly rinsed through a 63 μm brass sieve. The retained fraction was re-dried at 60 °C, weighed, and examined by taxonomic identification and quantitative census. Taxonomic identification and enumeration were performed using a transmitted-light microscope (Carl Zeiss Axio Scope A1, Oberkochen, Baden-Württemberg, Germany), targeting a census of ≥200 specimens. Samples yielding < 200 specimens were examined exhaustively. Identifications followed regional taxonomic keys [35,36]. Samples were classified as “no reliable data” and excluded from further analysis when any one of the following conditions was met: (i) fewer than 50 individuals remain after exhaustive picking; (ii) post-mortem abrasion affects ≥ 50% of specimens, indicating severe reworking; (iii) authigenic overgrowth, such as framboidal pyrite, obscures > 30% of diagnostic features under microscope; (iv) contamination by drilling mud or caving debris is evidenced by exotic mineral grains or modern microfossil contaminants; or (v) stratigraphic inconsistencies.

4. Results

4.1. Lithology and Sedimentary Facies

The chronostratigraphic framework for the core was established by Wu et al. [16], integrating nine accelerator mass spectrometry (AMS) ¹⁴C ages and nine optically stimulated luminescence (OSL) ages. A piece-wise linear regression model based on these age–depth tie points was applied to derive the final age model. Linear sedimentation rates (LSRs) ranged from 0.28 to 0.71 m/ka, consistent with regional datasets [37]. Integrating AMS ¹⁴C and OSL ages with lithological and microfossil data, the DZQ01 succession is subdivided into six depositional units (DUs: DU₁~DU₆) in ascending stratigraphic order as follows:

DU₁ (23.1–26.74 m, MIS 1): The lower interval comprises pale-yellow, bioturbated silty fine sand exhibiting limited ichnofabric development. Up-section, this facies grades into dark-grey, shell-bearing silty clay with disseminated black carbonaceous specks and moderate bioturbation. Foraminiferal abundance peaks at 52 specimens g⁻¹, and tests are notably larger. Ammonia annectens dominates the assemblage (Figure 2).

DU₂ (26.74–38.1 m, MIS 4~2): This unit is dominated by grey and grey-brown silt and sand displaying intense bioturbation, such as worm burrows. The absence of foraminifera and ostracods (Figure 2) indicates deposition within a fluvial regime. The lower sub-unit (32.27–38.1 m) comprises grey to dark-grey silty fine sand and fine sand interbedded with centimeter-scale (1–2 cm) silt laminae. Poor sorting, cross-bedding, and sparse millimeter-sized shell fragments are present, yet the lack of benthic foraminifera and ostracods confirms a meandering-river setting dominated by channel-fill sands.

DU₃ (38.1–44.4 m, MIS 5): The lower interval consists of grey-yellow clayey silt intercalated with silt lenses and millimeter-scale bands, exhibiting intense bioturbation. Sparse mud nodules and carbonaceous specks occur locally. Foraminiferal diversity increases markedly: species richness exceeds 15 and the complex diversity index surpasses 1.5, with abundances reaching 65 specimens g⁻¹ (Figure 2). These data indicate episodic shifts to a normal-marine shallow-shelf environment.

DU₄ (44.4–94 m, MIS 6): This unit grades upward from grey-yellow silty clay into grey, cross-bedded fine sand. Bioturbation is weak to moderate, and iron-stained sandy laminae are common. Microfossils are sparse in the lower part, whereas the upper interval contains non-marine ostracods such as Candona extima nom. nov. and Ilyocypris bradyi (Figure 2). DU₄ is interpreted as a braided-river system comprising channel, floodplain, and delta-front facies.

DU₅ (94–104.4 m, MIS 7): The lower interval comprises interbedded grey-yellow silt and silty clay with local wave-rippled grey beds, weak bioturbation, and abundant ferruginous mottles. The upper interval consists of dark-grey silty fine sand containing millimeter-scale shell fragments, abundant brown-yellow ferruginous mottles, and sparse bioturbation. Between 94.1 and 101.1 m, foraminiferal abundance and diversity peak. Ammonia beccarii dominates (maximum relative abundance ≈ 79%), accompanied by Astrorotalia subtrispinosa (≈22%) and Protelphidium granosum (Figure 2). This assemblage typifies a marginal-marine coastal setting.

DU₆ (104.4–111 m, MIS 8): This unit is dominated by brown silty fine sand and yellow-brown sand displaying negligible bioturbation. Foraminifera and ostracods are absent except for rare non-marine Candona extima nom. nov. at the top (Figure 2). DU₆ represents inter-channel facies within a braided-river system.

4.2. Mz, Rb/Sr and Mg/Ca Content

Mean grain size (Mz) attains maxima in DU₁ (6.21) and DU₃ (5.58), indicating a fine-grained, silt-dominated lithology. Intermediate Mz values occur in DU₅ (5.08), DU₆ (4.71), and DU₄ (4.22), whereas the minimum is recorded in DU₂ (3.56), reflecting a coarse-grained, sand-dominated lithology (Figure 3d and Figure 4a).

The Rb/Sr ratio, a widely applied paleoclimatic proxy, is greatest in DU₁ (0.65) and DU₃ (0.64), followed by DU₂ (0.43) and DU₆ (0.42), whereas minima are observed in DU₄ (0.35) and DU₅ (0.34) (Figure 3e and Figure 4b).

The Mg/Ca ratio, widely employed as a paleosalinity indicator, peaks in DU₃ (0.69) and DU₅ (0.59), and decreases in the order DU₄ (0.43) > DU₁ (0.42) > DU₆ (0.39) > DU₂ (0.34) (Figure 2f and Figure 4c).

4.3. REE Content Characteristics

4.3.1. General Characteristics of REEs

ΣREE concentration in core DZQ01 ranges markedly from 83.73 to 292.86 μg/g (mean = 178.78 μg/g). ΣLREE ranges from 74.26 to 230.58 μg/g (mean = 139.36 μg/g), indicating pronounced LREE enrichment, whereas ΣHREE varies from 9.46 to 68.02 μg/g (mean = 39.42 μg/g) (

Table 1. Abundance of Rare Earth Elements in core DZQ01 samples from the study area, Chinese continental shelf, and river sediments.

	Statistic	ΣREE	ΣLREE	ΣHREE	ΣLREE/ΣHREE	(La/Yb)UCC	(La/Sm)UCC	(Gd/Yb)UCC	δEuUCC	δCeUCC	References
DZQ01	Average value (N = 145)	178.78	139.36	39.42	3.95	1.27	1.26	1.41	0.85	1.02	This study
	Distribution range	83.73~292.86	74.26~230.58	9.46~68.02	2.62~5.61	0.93~2.21	0.97~1.67	1.17~2.02	0.71~1.14	0.05~1.24
YR	Average value (N = 5)	176.44	157.68	18.76	8.31	0.92	0.80	1.27	0.90	0.94
	Distribution range	120.79~262.55	107.17~238.39	13.62~24.16	7.57~9.87	0.77~1.17	0.68~0.91	1.15~1.42	0.76~1.03	0.75~1.03
HR	Average value (N = 3)	150.49	133.02	17.47	7.66	0.75	0.69	1.18	1.02	0.95
	Distribution range	98.15~167.67	86.79~148.81	11.35~20.36	6.99~8.61	0.69~0.9	0.66~0.76	1.1~1.29	0.96~1.12	0.85~1.05
LR	Average value (N = 3)	93.91	83.79	10.12	8.24	0.95	0.78	1.22	1.22	0.76
	Distribution range	78.65~111.23	69.19~100.59	9.46~10.64	7.31~9.45	0.91~1.01	0.73~0.88	1.17~1.24	1.11~1.35	0.6~1.02
Liao river	Average value	129.08	115.06	14.02	8.21	0.93	0.88	1.11	1.02	0.83	[24]
World rivers	Average value	205.00	191.50	13.50	14.19	0.94	0.96	0.83	1.19	1.04	[7]
BS	Average value (N = 3)	209.69	182.20	27.49	6.63	1.51	0.92	2.01	1.17	0.73	[38]
Yellow sea	Average value (N = 6)	123.01	111.59	11.42	9.77	1.11	0.95	1.51	1.07	0.99
East China Sea	Average value (N = 11)	120.13	107.25	12.88	8.33	0.98	0.74	1.56	0.92	1.00
South China Sea	Average value (N = 3)	166.26	152.14	14.12	10.77	1.87	1.07	2.12	1.10	0.65
China shelf sea	Average value (N = 23)	154.77	138.30	16.47	8.88	13.44	4.02	2.50	0.70	0.94

Note: REE statistics comprise 14 elements (La–Lu, excluding Y). UCC-normalized differentiation coefficients of core DZQ01 are following [39].

).

The mean ΣREE concentration lies within the global sediment range (150–300 μg/g) and exceeds reported averages for the Chinese continental shelf seas (154.77 μg/g), South China Sea (166.26 μg/g), East China Sea (120.13 μg/g), and Yellow Sea (123.01 μg/g), yet remains below the average for surface sediments in the BS (209.69 μg/g) [38,40]. Relative to fluvial sources, it closely matches the YR (176.44 μg/g) and surpasses the HR (150.49 μg/g), LR (93.91 μg/g), and LiaoR (129.08 μg/g) [41] (Table 1).

Boxplot statistics indicate elevated ΣREE in DU₁ (mean = 195 μg/g), DU₃ (mean = 226 μg/g), and DU₅ (mean = 172 μg/g), whereas DU₂ (mean = 126 μg/g), DU₄ (mean= 142 μg/g), and DU₆ (mean = 167 μg/g) exhibit lower concentrations (Figure 4d). The ΣREE profile reveals two prominent peaks in DU₆ (191 μg/g and 219 μg/g) and a trough at 97.42 μg/g. DU₅ displays a modest increase, peaking at 219 μg/g between 95.92 and 98.92 m. DU₄ contains two high-ΣREE intervals (92.93–77.97 m and 66.43–55.04 m), peaking at 281 μg/g and 290 μg/g, respectively. DU₃ averages 223 μg/g, with the sole low value at 44.81 m (101.644 μg/g). DU₂ remains consistently low (≈121 μg/g), whereas DU₁ exhibits an upward increase from 92 μg/g to 239 μg/g (Figure 3).

4.3.2. Fractionation Characteristics of REE

REE concentrations were normalized to chondrite and UCC compositions to recognize variable provenance and geological background, ensuring inter-sample comparability. Following chondrite normalization [42], δEu_N ranges from 0.47 to 0.76 (mean = 0.56), indicating a pronounced negative Eu anomaly, whereas δCe_N spans 0.91–1.38 (mean = 1.12), reflecting a negligible Ce anomaly relative to chondrite. Fractionation indices yield mean ΣLREE/ΣHREE = 3.95, (La/Yb)_N = 12.44, (La/Sm)_N = 5.44, and (Gd/Yb)_N = 2.01, denoting a pronounced LREE enrichment and marked inter-REE fractionation.

After UCC-normalization, δEu_UCC ranges from 0.71 to 1.14 (mean = 0.85), signifying a distinct negative Eu anomaly, while δCe_UCC spans 0.05–1.24 (mean = 1.02), implying subdued Ce anomalies. Corresponding fractionation parameters yield (La/Yb)_UCC = 0.93–2.21 (mean = 1.27), indicating weak LREE–HREE fractionation. (La/Sm)_UCC = 0.97–1.67 (mean = 1.26), reflecting weak intra-LREE fractionation, and (Gd/Yb)_UCC = 1.17–2.02 (mean = 1.41), indicating marked intra-HREE differentiation (Table 1). Downcore profiles of these indices are broadly coherent, fluctuating around 1.2–1.4. However, between 76 and 67 m, they exhibit enhanced variability, suggesting intensified LREE–HREE fractionation within this interval (Figure 3j–l).

4.3.3. REE Distribution Patterns

Chondrite-normalized REE patterns exhibit steep LREE (La–Eu) segments (Figure 5a), reflecting pronounced LREE enrichment, whereas HREEs (Eu–Yb) are comparatively flat and uniform. The resultant left-high, right-low V-shape profile illustrates marked LREE–HREE fractionation, accompanied by a negative Eu anomaly indicative of moderate depletion.

UCC-normalized REE signatures of core DZQ01 and fluvial end-members (YR, HR, LR, and LiaoR) are compared in Figure 5b. Riverine profiles reveal negative Ce anomalies in the YR, LR, and LiaoR, but not in the HR, whereas Tm displays a subtle positive anomaly. Eu is enriched in the LR, and Gd is enriched in the HR. UCC-normalized REE patterns for individual DUs are generally flat, each exhibiting a pronounced positive Gd anomaly. Other units exhibit subtle positive Dy anomalies, DU₁ shows a negative Ce anomaly, whereas DU₂ and DU₅ display positive Yb anomalies.

5. Discussion

5.1. Controls on REE Composition in Core Sediments

REE abundances are primarily governed by provenance, chemical weathering, grain size, mineralogy, and climate [45,46,47,48]. Provenance exerts the dominant control [49,50,51], whereas grain size is a secondary determinant: fine-grained fractions (silts and clay) concentrate REEs via high adsorption capacities, whereas quartz- and feldspar-rich sands promote REE fractionation and depletion [46].

ΣREE exhibits a moderate-stronge positive correlation with Mz and CIA (R² = 0.56, 0.78; Figure 6a,b), implying a partial control exerted by grain-size and chemical weathering. Al₂O₃, geochemically stable during weathering, correlates strongly with fine-grained fractions. Al₂O₃-normalization effectively mitigates grain-size bias [52,53]: REE/Al₂O₃ ratios exhibit no significant relationship with Mz (R² = 0.069; Figure 6b). Internal (e.g., Gd/Yb) and environmental (e.g., δEu_UCC and δCe_UCC) fractionation indices likewise display negligible to weak correlations with Mz: (Gd/Yb)_UCC vs. Mz (R² = 0.003, Figure 6d), δEu_UCC vs. Mz (R² = 0.306, Figure 6e), and δCe_UCC vs. Mz (R² = 0.002, Figure 6f). The absence of significant linear relationships confirms that these fractionation indices are robust provenance proxies.

Heavy minerals (zircon, apatite, monazite, and garnet) occur in trace amounts in sandy sediments, yet can markedly elevate ΣREE concentrations because of their high REE contents [44,54,55,56]. Zircon (Zr) is almost exclusively hosted by zircon and serves as a proxy for its abundance [57]. Within core DZQ01, La versus Zr yields a negligible correlation (R² = 0.062, Figure 6g), implying negligible zircon control on REE distribution. Similarly, the weak-to-negligible correlation between La and Fe (R² = 0.052, Figure 6h) or Th (R² = 0.26, Figure 6i) suggests that Fe-Mn oxides and associated REE-bearing phases (e.g., monazite) exert limited influence on REE distributions.

5.2. Source Analysis of Sediments in Core DZQ01

The YR annually delivers ~1.08 × 10⁹ t of sediment to the BS, equivalent to ~6% of the global fluvial discharge to the oceans. Since the 1855 course shift into the BS, mean annual sediment discharge has averaged ~1.2 × 10⁹ t [25,26]. Adjacent rivers (LiaoR: 2–5 × 10⁷ t/yr; LR: 2.7 × 10⁷ t/y; HR: 6 × 10⁶ t/yr) also deliver significant loads [23]. Sediment provenance signatures reflect catchment lithology, regional climate, weathering intensity, and tectonic setting, collectively governing the compositional characteristics of BS sediments [58].

Chondrite-normalized REE patterns (Figure 5a) closely resemble chondritic compositions, indicating continental provenance and precluding extraterrestrial contributions. UCC-normalized curves (Figure 5b) reveal subdued negative Ce anomalies in the YR, LR, and LiaoR basins, consistent with intense chemical weathering of the Loess Plateau [59], reflecting Ce depletion under oxidizing conditions. Positive Tm anomalies imply contributions from variable hydrodynamic regimes, local lithological heterogeneities [60], and seawater solubility and rock-chemical interactions [61]. Positive Eu anomalies in the LR reflect bauxite or calcareous inputs [62,63], whereas elevated Gd in the HR implies calcareous enrichment or distinctive aqueous geochemistry [64]. UCC-normalized δEu_UCC values in core DZQ01 exhibit pronounced negative anomalies (Figure 5b), signifying variable degrees of differentiation relative to the UCC, attributable to source-area weathering or shifting depositional conditions.

The δEu_UCC–δCe_UCC diagram effectively discriminates sediment provenance and has been validated in numerous studies [52]. In Figure 7, DU₁ samples plot within the YR field, indicating predominant YR provenance. DU₂ samples yield in the YR fields, implying mixed YR–HR input. DU₃ samples cluster within the YR field, denoting dominant YR contribution. DU₄ samples occupy the triple overlap among YR, HR, and LR, indicating a composite three-end-member provenance. DU₅ samples reside exclusively within the YR field, signifying a near-pure YR derivation. DU₆ samples mirror DU₄, lying at the confluence of YR, HR, and LR, thus reflecting a comparable three-river mixture.

Subsequently, the REE-based DF was applied to quantify the relative contributions of four potential sources—YR, HR, LR, and LiaoR sediments—to the study area. DF values for each source were calculated from the (Gd/Yb)_UCC ratio (Figure 8a). Core DZQ01 exhibits distinctly lower DF values for the YR, LR, and HR relative to the LiaoR, signifying that the former three are the dominant sources, whereas the LiaoR contributes negligibly. This conclusion is reinforced in Figure 8b, where the stacked bar chart exclusively incorporates contributions from the YR, HR, and LR.

In DU₁, DF_LR and DF_HR are moderate (mean = 0.15), whereas DF_YR is persistently low (mean = 0.06, Figure 8a). Correspondingly, the YR contribution attains its maximum (Figure 8b). Consequently, detrital material in DU₁ is overwhelmingly YR-derived.

In DU₂, DF_YR remains stable (mean = 0.06), whereas DF_LR is elevated at the top and base but minimal in the middle (0.15 → 0.04 → 0.15, Figure 8a). Thus, sediment supply is predominantly YR at the onset and termination, shifting to LR in the mid-stage.

In DU₃, DF_HR is elevated at the top and base but reduced in the middle (0.16 → 0.08 → 0.15). DF_YR displays the inverse trend (0.05 → 0.15 → 0.1 Figure 8a). Proportional data indicate a dominant YR input at the top and base, with HR prevailing in the middle (Figure 8b). Sediment provenance evolves as follows: YR → HR → YR.

In DU₄, all three DF indices fluctuate markedly. DF_HR remains elevated (mean = 0.15), whereas DF_YR and DF_LR are subdued (0.06 and 0.12, respectively) in the upper and lower intervals. In the middle interval, DF_HR decreases to 0.08, while DF_YR and DF_LR rise sharply to 0.5 and 0.6, respectively (Figure 8a). The upper interval is dominated by YR and LR inputs, whereas HR dominates the middle interval. All three rivers contribute substantially throughout DU₄ (Figure 8b). Thus, DU4 sediments originate from YR + LR in the upper section, HR in the middle, and YR + LR again in the lower section.

In DU₅, uniformly low DF_YR (mean = 0.05, Figure 8a) indicates an overwhelming YR source.

In DU₆, shallow intervals yield low DF_YR (0.03) and DF_HR (0.04), and deeper intervals exhibit low DF_YR (0.02) and DF_LR (0.03) (Figure 8a). Consequently, provenance evolves from YR + LR in the low DU₆, to YR + HR in the upper DU₆.

5.3. Paleoenvironmental Implications of Sediment Source Changes

Core DZQ01 preserves a complete Middle Pleistocene record punctuated by three transgressive–regressive cycles. This paper reconstructed the western BS depositional by coupling Rb/Sr ratios (paleoclimate), Mg/Ca ratios (paleo-salinity), global sea level, and temperature records.

DU₆ (300–272 cal. ka B.P., MIS 8): Miaodao-uplift-induced tectonism severed BS oceanic exchange [65,66], fostering extreme cold–arid conditions (Rb/Sr; Figure 9f) and resulting in an almost complete absence of microfossils [67] (Figure 2). Intensified East Asian Winter Monsoon (EAWM) lowered the sea level and enhanced evaporation [18,68] (Figure 9c), promoting continental-shelf fluvial–lacustrine deposition and paleochannel preservation [18,68]. Provenance shifted from YR + LR (pre 290 cal. ka B.P.; Figure 10a) to a dominant YR input with minor HR influence (290~280 cal. ka B.P.; Figure 10a). Multi-river supply and coarse-particle dilution reduced ΣREE (Figure 9g).

DU₅ (272–205 cal. ka B.P., MIS 7): Warming triggered polar-ice decay and global sea-level rise (Figure 9b,c), driving oceanic thermal expansion and altered hydrological cycling. Marine incursion transformed BS into a brackish, tide-dominated embayment (elevated Mg/Ca; Figure 9e) [69]. Benthic foraminiferal (dominant Ammonia beccarii and Protelphidium granosum) and marine ostracods (dominant Neomonoceratina delicata, Sinocytheridea latiovata, Pistocythereis bradyi, and Pistocythereis bradyiformis) indicate a warm, shallow, brackish setting (Figure 2). Rare non-marine ostracods (Candona extima nom. nov. and Ilyocypris bradyi) imply an episodic freshwater influx (Figure 2). The warm, humid YR-basin climate elevated fluvial discharge, delivering fine-grained sediments to the BS (Figure 10b). REE enrichment in fine-grained particles elevated marine REE (Figure 9g).

DU₄ (205–130 cal. ka B.P., MIS 6): Global climate cooled markedly (Figure 9b), causing polar-ice expansion and a pronounced sea-level fall (Figure 9c). Continental-shelf exposure intensified ocean-to-continent heat and moisture transfer [1,70], establishing a dominantly terrestrial depositional regime. The MIS 6 cooling is attributed to reduced tropical-Pacific temperatures and a southward Intertropical Convergence Zone (ITCZ) displacement [71]. Non-marine ostracods and sparse benthic foraminifera (Ilyocypris bradyi, Bicornucythere bisanensis, Sinocytherides latiovata, Rotalinoides compressiuscula, and Elphidium advenum) occur in DZQ01 (Figure 2). Before 180 cal. ka B.P., temperatures dropped to −10 °C and sediment derived primarily from YR, LR, and HR (Figure 9b). Between 160 and 140 cal. ka B.P., global sea level fell to ~−120 m, redirecting western BS provenance to the HR. The YR likely migrated southward into the South Yellow Sea (SYS), supplying most of its sediment [72], while Laizhou Bay records indicate smaller-river dominance [73]. Hydrologically stable and less climate-sensitive, the HR basin supplanted the YR as the dominant source. After ~140 cal. ka B. P., northward re-entry of the YR re-established its regional dominance, while the HR’s influence declined commensurately (Figure 10a). ΣREE exhibits a positively covariance with YR input; rising YR contribution elevates ΣREE (Figure 9g and Figure 10a).

DU₃ (130–71 cal. ka B.P., MIS 5): Global warming raised sea levels via the increased link of tropical-Pacific sea surface temperatures (SST) to orbital forcing and a probable northward ITCZ shift (Figure 9b), thereby altering El Niño–Southern Oscillation (ENSO)-like variability [74]. A rising Rb/Sr ratio indicates warm–humid conditions (Figure 9f), while Mg/Ca ratios denote a transitional marine–terrestrial setting (Figure 9e). Benthic foraminifera and ostracods flourished, including Buccella frigida, Cribrononion cf. Tsudai, Hemirotalia foraminulosa, Protelphidium granosum, Quinqueloculina lamarckiana, and Rotalinoides compressiuscula. Abundant marine ostracods (Bicornucythere bisanensis, Neomonoceratina delicata, and Sinocytheridea latiovata) proliferated, and sparse occurrences of Pistocythereis bradyi and Pistocythereis bradyiformis further signal a thriving marine ostracod assemblage (Figure 2). Enhanced precipitation in the YR Basin elevated discharge and sediment load, establishing the YR as the dominant BS source. Although HR discharge also increased (Figure 10b), its sediment flux remained subordinate to that of the YR [69]. Consequently, DU₃ deposits are predominantly fine-grained (Figure 9a) and display elevated ΣREE concentrations (Figure 9g). Positive Ce anomalies may reflect an early diagenetic reduction in anoxic sediments [75], whereas Eu anomalies remain consistent with source mineralogy, suggesting minimal post-depositional alteration [76] (Figure 3m,n).

DU₂ (71–14 cal. ka B.P., MIS 4~2): Global cooling and polar-ice expansion cause a pronounced sea-level fall (Figure 9a,b). Rare-gas measurements of the Taylor Glacier 0.24 m core from Antarctica indicate that MIS 4 Oceanic maxima were the coldest of the last glacial cycle [77], with full glacial extremes only achieved by MIS 2. This suggests that both global temperatures and the eustatic sea level declined progressively from MIS 4 to MIS 2 [78]. Microfossil assemblages were sparse, comprising euryhaline benthic foraminifera (e.g., Ammonia beccarii, Ammonia tepida, and Hemirotalia foraminulosa) in DU₂ upper layers (26–28 m depth), reflecting a warm, shallow, salinity-tolerant setting. Sparse marine ostracods (Bicornucythere bisanensis, Pistocythereis bradyi) were also present (Figure 2). Falling sea levels contracted the BS (Figure 9c), especially in the west, creating shallower, more enclosed conditions that altered hydrodynamic [1,68] and modulated sediment sourcing. After ~30 cal. ka B.P., continued regression attenuated tidal, wave, and ocean-current forcing, reducing open-ocean fines delivery [79]. During MIS 4, the YR and LR dominated provenance sediment source; in MIS 3, southward YR diversion [80] rendered the LR primary; in MIS 2, the renewed northward YR flow re-established its dominance (Figure 10a). LR input remained substantial (Figure 10a), yielding coarser facies (Figure 9a) and diminished ∑REE (Figure 9g).

DU₁ (14 cal. ka B.P.–Present, MIS 1): Post-glacial warming (Figure 9a) and anthropogenic forcing have driven global sea-level rise via oceanic thermal expansion, ice-sheet mass loss and land-water storage [81]. The elevated Rb/Sr ratio records a warm–humid climate (Figure 9f) that fostered abundant euryhaline benthic foraminifera (Ammonia annectens, Ammonia tepida, Buccella frigida, Cribrononion cf. tsudai, Hemirotalia foraminulosa, Protelphidium granosum, and Quinqueloculina Lamarckiana) and marine ostracods (Bicornucythere bisanensis, Neomonoceratina delicata, Pistocythereis bradyi, and Pistocythereis bradyiformis) (Figure 2). Enhanced precipitation in the YR basin intensified discharge and sediment transport load, consolidating its role as the dominant BS source [28]. Progressive sea-level rise (Figure 9b) elevated the basin water level, improved marine hydrodynamics, and expanded the YR mouth, thereby reinforcing its control on western-BS sedimentation [29]. Intensifying seawater infusion enlarged the YR mouth, augmenting its sediment-transport efficiency. DU₁ deposits were predominantly fine-grained (Figure 9c), mirroring YR basin hydrology and topography conducive to effective fines export. Western BS sediments therefore display a pronounced YR provenance signature, underscoring its regional dominance. Progressive YR dominance (Figure 9h) is accompanied by a concomitant ∑REE enrichment (Figure 9g). Our REE-based provenance shifts align with records from the YSS, where enhanced YR input during MIS 1 is linked to sea-level rise and increased monsoon intensity [82,83].

6. Conclusions

(1)

ΣREE concentration in DZQ01 core varies from 83.73 to 292.86 μg/g (mean = 178.78 μg/g), exhibiting pronounced LREE enrichment. Chondrite- and UCC-normalized patterns display marked negative Eu anomalies (δEuN = 0.56; δEu_UCC = 0.85). Fractionation indices (ΣLREE/ΣHREE = 3.95; (La/Yb)_N = 12.44) demonstrate a pronounced LREE–HREE fractionation. Resultant REE patterns exhibit left-high, right-low “V” shape profiles with ubiquitous positive Gd anomalies.

(2)

Geochemical and sedimentary archives from core DZQ01 demonstrate a systematic coupling between depositional environments, fluvial provenance, and REE inventories in the western BS since the Middle Pleistocene. In warm, humid marine intervals (e.g., DU₅, DU₃, DU₁), the YR dominated provenance, delivering fine-grained Loess Plateau detritus; REE enrichment within fine-grained fractions yielded elevated marine REE values. Conversely, during cold, arid terrestrial phases (e.g., DU₆, DU₄, DU₂), the YR, HR, and LR supplied a mixed coarse–fine sediment load. Multi-river sourcing and coarse-particle dilution accordingly lowered REE concentrations.