A sediment core 4.8 m long (Core C2) was collected from a water depth of ~25 m in the YRD (31°00′N, 122°44.4′E; Figure 1B ) in 2018. The core comprises mainly gray or grayish yellow silty clay and clayey silt ( Figure 2 ) interbedded with occasional shells and shell fragments. The core sediment was stored at room temperature (20°C) pending analysis, with working halves being photographed, lithologically described, and continuously sampled at 2 cm intervals.

Grain-size analysis involved a Malvern Mastersizer 2000 Laser Particle Analyzer with a measurement range of 0.02−2000 μm. Before measurement, samples were dispersed and homogenized by ultrasonic agitation for 30 s. Grain-size parameters (mean grain size (Mz), standard deviation (sorting), skewness, and kurtosis) were analyzed using the GRADISTAT program (Blott and Pye, 2001) by the method of Folk and Ward (1957).

Core chronology was based on OSL dating and ²¹⁰Pb and ¹³⁷Cs dating methods. OSL dating of seven samples was undertaken at the Luminescence Dating Laboratory, East China Normal University (ECNU), Shanghai, China, using a Risø TL/OSL-DA-20 system. Individual OSL dose values were determined by single aliquot regenerative protocols (Murray and Wintle, 2003). Details of sample pretreatment procedures and fine-grained quartz (4–11 μm) OSL age calibration methods have been provided by Zhou et al. (2021). OSL dating yielded ages of 80 ± 10 yr. BP (years prior to CE 2017) at 0.23 m depth, 210 ± 20 yr. BP at 0.8 m, 500 ± 40 yr. BP at 1.6 m, 765 ± 60 yr. BP at 2.2 m, 1240 ± 90 yr. BP at 3.5 m, 2070 ± 150 yr. BP at 4.1 m, and 2150 ± 150 yr. BP at 4.7 m (Zhou et al., 2021; Figure 2 ).

A high-resolution gamma spectroscopy system with a low-energy HpGe detector (Canberra Be3830) was used to determine ²²⁶Ra, ¹³⁷Cs, and total ²¹⁰Pb (²¹⁰Pb_tot) activities at 2–10 cm intervals for the top 130 cm of the core. Samples were sealed in a plastic box for one month and analyzed following the methodology of Wang et al. (2016). ²¹⁰Pb_ex activity was obtained by subtracting ²²⁶Ra activity from ²¹⁰Pb_tot activity. These analyses were undertaken at the State Key Laboratory of Estuarine and Coastal Research, ECNU. A constant initial concentration model (CIC; Appleby and Oldfield, 1978) was applied in calculating the average sedimentation rate. Using the ²¹⁰Pb, ¹³⁷Cs, and OSL ages, an age–depth model was constructed using the R package “Bacon” (Blaauw and Christen, 2011). Activities are reported in Becquerel (Bq or Bq kg^–1).

Previous studies have indicated that the YRD is an important source of organic matter (OM) for the YRE (Zhang et al., 2007; Li et al., 2012). A three-endmember mixing model was used to estimate temporal OM source variations in Core C2 and to elucidate historical sedimentary environmental changes in the YRD. This mixing model (Li et al., 2012; Hu et al., 2014) uses TOC/TN ratios and δ¹³C values as source markers to track the relative contributions of three different sources of OM in YRD sediments: riverine (OC_riverine), deltaic (OC_deltaic), and marine (OC_marine) sources. The following equations were applied:

where f_Riverine , f_Deltatic , and f_Marine are the fractions of OC_river, OC_deltaic, and OC_marine, respectively. Based on the results of Zhang et al. (2007), the endmember δ¹³C values for OC_river, OC_deltaic, and OC_marine are −28.7‰, −22.10‰, and −20.00‰, with TOC/TN ratios of 12.50, 17.54, and 6.49, respectively.

In the upper part of Core C2 (0–56 cm depth), the ²¹⁰Pb_ex activity is exponentially and negatively correlated with depth (R² = 0.84; Figure 2B ). The average sediment accumulation rate based on the CIC model is 0.73 cm yr⁻¹ ( Figure S1 ; Zhou et al., 2021). The ²¹⁰Pb deposition rate can be affected by many factors (e.g. the absence of physical and chemical calibrations, grain size, and the uncertainty of ²¹⁰Pb measured data) in the subaqueous YR delta (Liu and Fan, 2011), so sedimentation rates here are based on ¹³⁷Cs activity. There are ¹³⁷Cs peaks at 19 cm (1986 CE) and 36 cm (1963 CE) depth in the profile ( Figure 2B ), with ¹³⁷Cs activity decreasing to zero at 51–53 cm depth at 1953–1954 CE ( Figure 2B ; Leslie and Hancock, 2008). The core sedimentation rate over 0–50 cm depth is thus estimated to be 0.74 cm yr⁻¹ after 1963 CE, and 1.74 cm yr⁻¹ during 1953–1963 CE.

For depths below 50 cm, a chronology sequence based on ¹³⁷Cs and OSL ages was established using the Bacon program (Blaauw and Christen, 2011; Figure S1 ). The calculated average sedimentation rates varied little from 70 BCE to ca. 1000 CE (0.18–0.19 cm yr⁻¹), and increased to 0.25 cm yr⁻¹ from 1000 CE to 1950 CE ( Figure S1 ).

Sediment clay, silt, and sand contents are in the ranges of 8.4%–11.19%, 83.16%–90.62%, and 0.01%–7.30%, with averages of 11.82%, 87.08%, and 1.09%, respectively ( Figure 2C ). Mean varies over the range of 6.20–10.58 μm (average 7.77 ± 0.79 μm), fluctuating more over depths of 480 cm (70 BC) to 50 cm (1950 CE) and increasing to the present.

TOC and TN contents exhibit an upwardly decreasing trend from the bottom to a depth of 355 cm (ca. 640 CE) with a range of 1.09%–0.55% for TOC and 0.09%–0.06% for TN, with a relative steady trend during the period of 640-1765 CE, and a more marked decrease from 97 (1765 CE) to 25cm (1970 CE), and abnormal increase from 1970 CE to the present ( Figures 2 and 3 ). TOC/TN ratios vary in the range of 6.6–14.08 (mean 9.40 ± 1.38), increasing from 6.67 to 14.08 (mean 9.91 ± 1.82) during the period of 70 BC-640 CE, fluctuating slightly around 9.00 during 640-1765 CE, then increasing to 10.55 after 1765 CE (mean 9.57 ± 1.23; Figure 3 ). δ¹³C and δ¹⁵N values are in the ranges of −23.26‰ to −25.13‰ (mean −24.10 ± 0.42‰) and 3.46‰ to 5.93‰ (mean 4.95 ± 0.52‰), respectively ( Figure 3 ). The δ¹³C values decrease up to 640 CE (mean −24.29 ± 0.42‰), are relatively stable during the period 640-1385 CE, increase through 1385-1765 CE (mean −24.23 ± 0.36‰), then show relative stable since 1765 CE(−23.98 ± 0.34‰). δ¹⁵N values increase slightly up to 1385 CE before decreasing toward to the present (mean 4.49 ± 0.38‰; Figure 3 ).

Riverine OC made the greatest contribution to sedimentary OC in Core C2, at 36%–52% (mean 44% ± 5%; Figure 4A ), and the contributions of marine and deltaic OC were in the ranges of 4%–84% (mean 43% ± 15%) and <1%–56% (mean 17% ± 13%), respectively ( Figure 4C ). Before 640 CE, the relative contribution of riverine OC had an increasing trend (mean 45.62 ± 4.78%), while the contribution from marine OC decreased (mean 37 ± 14.19%) and the deltaic OC contribution fluctuated widely (mean 20 ± 15.70%; Figure 4B ). During the period 640–1385 CE, riverine and marine OC contributions were relatively stable, whereas the deltaic contribution fluctuated widely. During 1385–1765 CE, both riverine and marine OC contributions decreased sharply, with minimum values of 32% and 27%, respectively, whereas deltaic OC increased. After 1765 CE, riverine and deltaic OC contributions increased upward (means of 41± 3.57% and 19± 11.78%, respectively), whereas marine OC decreased (mean 41 ± 12.75%).

Temporal variations of major-element (Ca/Al) ratios and trace-metal (Li, Ti, Ni, Zr, Ho, Mo, Th, Sr, and Mn) contents in Core C2 are shown in Figure S2 . Before 1385 CE, there was a slight decrease in Li, Ti, Ni, Zr, and Ho contents, followed by a significant decrease during 1385–1765 CE, although Zr and Ho contents increased upward. Mo, Th, Sr, and Mn contents were relatively stable before 1385, increasing during 1385–1765 CE, right to the top of the core for Sr and Mn. The trend in Ca/Al ratio was consistent with those of Sr and Th ( Figure S2 ).

To understand the deposition pattern of these four evolutionary stages, the rare earth element (REE) patterns normalized for the upper continental crust were drawn separately ( Figure S3 ; McLennan, 2001). The results show similar variation trends, indicating the relative the relative consistency of the sediment source of the Core C2 in the past 2000 years. However, the sediments in 1385-1675 CE showed the lower pattern in most REE elements ( Figure S3 ).

According to the formula in Section 2.2, the OC burial flux (Fc) was calculated from the TOC content, bulk density, water content, and sedimentation rate, with a range in Core C2 of 11–70 g m² yr⁻¹ (mean 21 g m² yr⁻¹; Figure 5 ). Before 640 CE, the burial rate had an increasing but widely fluctuating trend (mean 16 ± 2.19 g m² yr⁻¹); during 640–1385 CE, it was higher but more variable (mean 17 ± 3.13 g m² yr⁻¹); during 1385–1765 CE, it decreased from 18 to 13 g m² yr⁻¹ (mean 16 ± 1.60 g m² yr⁻¹); and after 1785 CE (particularly after 1940 CE), the burial flux increased from 14 g m² yr⁻¹ to 70 g m² yr⁻¹.

δ¹³C values and TOC/TN ratios have been widely used to identify sediment OM sources in estuarine and coastal marine environments. Terrigenous OM usually has higher TOC/TN ratios (>12) and lower δ¹³C values (~−27‰) than OM of marine origin, which has TOC/TN ratios of 6–8 and δ¹³C values of ~−20‰ (Meyers, 1997; Lamb et al., 2006). Previous studies in the YRD area (Zhan et al., 2012) have indicated that values of OM proxies (TOC and TN contents, δ¹³C and δ¹⁵N values) for surficial sediments, soil, and suspended particulate matter were not significantly different in riverine settings with higher TOC/TN ratios (12.0–18.9) and more negative δ¹³C values (−28.7‰ to −24.4‰), whereas surficial sediments and suspended particle matter have lower TOC/TN ratios (3.8–7.6) and higher δ¹³C values (−22.7‰ to −20.0‰) in shallow marine settings.

Sedimentary OM is affected by many factors in estuarine–coastal zones, including inorganic nitrogen and grain-size effects. There is a no obviously linear relationship between TOC and TN in Core C2 (R = 0.38; Figure 6A ), so the effect of inorganic nitrogen on the sediments is negligible (Hu et al., 2014). Furthermore, the organic proxies are not linearly correlated with mean grain size ( Figures 6C–E ), with sedimentary OM in the study area likely being affected predominantly by OM provenance. Here, TOC/TN ratios range from 14.1 to 6.7 (mean 9.4 ± 1.4; Figure 3 ), suggesting a predominantly terrestrial OM contribution. The δ¹³C value of Core C2 ranges from −25.13‰ to −23.26‰ (mean 24.10‰ ± 0.42‰), with OM in the core sediment thus being derived from a mixture of terrestrial and marine sources, as indicated by previous organic geochemistry studies of the YRD (Deng et al., 2006; Gao et al., 2008; Zhan et al., 2012; Li et al., 2015).

However, the correlation between the δ¹³C values and TOC/TN ratios is weak (R = 0.52; Figure 6B ), implying that the OM source of the core cannot be distinguished by a simple binary mixing model (Zhang et al., 2007; Gao et al., 2008; Hu et al., 2014). There is a possible OC contribution from the YRD (Zhang et al., 2007; Li et al., 2012), with its unique hydrodynamic conditions related to preservation and circulation of sediment OM in the YRD (Zhang et al., 2007; Li et al., 2012; Hu et al., 2014). Therefore, a three-endmember-mixing model (riverine, deltaic, and marine) is likely more suitable for estimating temporal OM source variations for the core.

Previous sedimentological studies have shown that there are various natural (e.g., EASM, ENSO, and PDO) and anthropogenic factors that have altered variations in terrestrially derived OC in the YRE on decadal, centennial, and millennial scales (Yang et al., 2011b; Hu et al., 2014; Xu et al., 2021). EASM plays the dominate role in controlling the transport and burial of YR organic carbon (Wang et al., 2005), and ENSO and PDO are important climate factors which affected on the variation of EASM system and OC input into the Sea (Meng et al., 2015; Zhou et al., 2021). However, there is a lack of understanding of the continuous centennial scale variability of OC in the YRE and its adjacent shelf, and its response to these factors. Here, we aim to reconstruct historical OC variations, analyzing factors influencing carbon burial in the YRE over the last 2000 years.

During the period of 2000–640 CE, there was an increasing trend in f_Riverine and Fc ( Figures 7A, B ) and decreasing trend of f_Marine ( Figure 4C ), suggesting increased terrigenous OC input to the core site. The relative stronger EASM during this period ( Figure 7D ) likely strengthened precipitation in the YR basin, thus increasing YR basin OM input to the YRE.

From 640 to 1385 CE, the proportions of f_Riverine OC and Fc in Core C2 were relatively high, and f_Marine was stable. A high f_Riverine OC supply to the YRE is also indicated by the frequent occurrence of severe hypoxia there during this period (Ren et al., 2019). However, the declining trend of EASM is likely reflected in reduced input of OC to the YRE ( Figure 7D ). This may be linked to frequent flood events caused by El Niño events ( Figure 7E ) and weakened EASMs ( Figure 7D ) (Zhou et al., 2021). During weak EASM years and El Niño phases, the subtropical high in the western Pacific locates more southern than in normal years, and the rainband stagnates over the middle–lower reaches of the YR basin for long periods, resulting in heavy precipitation and flood events in the YR basin (Jiang et al., 2006; Zhou et al., 2021) and leading to increased supply of OC to the sea. The increased coarseness and variability of the mean grain size in the core ( Figure 2D ) also indicate increased freshwater discharge and frequent flood events during 640–1385 CE. Furthermore, plant species from which pollen was derived changed markedly after ca. 700 CE in association with increased anthropogenic activity in the YR basin (Yi et al., 2003), which may also have increased soil erosion and resulting sediment load in the YRE.

During 1385–1765 CE, the proportions of f_Riverine OC, and Fc and the mean grain size were sharply reduced, likely in association with a cold and dry climate during the Little Ice Age (LIA, Figure 7C ). The North Atlantic cooling ( Figure 7B ; Kubota et al., 2015) during the LIA period can significantly weaken the EASM intensity (Wang et al., 2005; Zhang H et al., 2018; Figure 7D ). Porter et al. (2021) suggest that the LIA was primarily defined by a weak, negative IPO (Inter-decadal Pacific Oscillation). Climate simulations suggest that a huge cyclone occupies the North West Pacific in the negative period of the IPO. And La-Niña periods can strengthen the cyclone by weakening the trade winds, which leads to a weak Meiyu front (weak EASM) (Zhang X et al., 2018). Therefore, weak EASMs, negative IPO (Porter et al., 2021), and less frequent El Niño events ( Figure 7E ) may reduce riverine discharge and hence the supply of OC to the YRD. However, the abrupt increase in sedimentation rate indicated by Core C2, from 0.19 to 0.25 cm yr⁻¹ ( Figure S1 ), suggests a significant sediment supply to the study site during the period. This is supported by the abrupt increase in shoreline progradation of the YRD over the last 1000 years (Hori et al., 2001). An increased OM supply to the ESC is also indicated by organic geochemical records for the ECS inner shelf (Hu et al., 2014). Debate continues regarding the contribution of HR sediment (Section 4.3) and human activities during this period.

After 1765 CE, the f_Riverine of Core C2 remained low; however, Fc increased significantly in association with the high sedimentation rate ( Figure S1 ). The higher temperatures, enhanced EASM ( Figure 7D ), increased TOC content ( Figure 3A ), frequent flood events (Zhou et al., 2021), and intensive anthropogenic impacts ( Figure 7G ) in the YR basin after 1765 CE were all conducive to increased OM input to the YRE and the adjacent ECS inner shelf.

Channel geomorphic dynamics of the HR and its huge sediment discharge have significantly affected the coastal evolution of the Bohai and Yellow seas over the past 3000 years (Chen, 2019). During 1128–1855 CE, the lower HR was diverted southward to the Huaihe River, with its sediment discharging into the Yellow Sea ( Figure 1B ). In the winter of 1128 CE, the Song army broke the HR levee to ward off advancing Jurchen troops. This led to a major avulsion, with large amounts of HR sediment being transported southward through the Huaihe River (Zhang, 1984; Chen, 2019), especially after 1194 CE. During 1194–1494 CE, ~40% of the HR discharge flowed into the Huaihe River, with the sediment load being trapped mainly in HR channels and with the shoreline progradation rate being relatively low (~50 m yr⁻¹; Zhang, 1984). During 1494–1855 CE, massive amounts of sediment were transported to the Yellow Sea, resulting in rapid shoreline progradation of 100–500 m yr⁻¹. After the lower HR shifted northward in 1855 CE, the Abandoned Yellow Delta suffered rapid erosion of 20–300 m yr⁻¹ (Zhang, 1984). Southward transport of HR sediment by the SCC was enhanced, promoting YRD aggradation and progradation (Wang et al., 2019).

Based on previous studies of mineralogical and geochemical proxies (Liu et al., 2010), environmental magnetism (Wang et al., 2020), and zircon U–Pb dating (Shang et al., 2021), YR sediment transport to the YRD increased greatly at ~600 yr BP. Surface sediment studies (Zhang et al., 2020; Qi et al., 2020) have indicated that the low OM content of AHD sediment contributed to the distribution and composition of OM in the YRE. In addition to organic geochemistry, elemental geochemistry has been widely applied in distinguishing between YR sediment and its form in the coastal zone (Yang et al., 2002; Yang et al., 2003; Xu et al., 2009), with HR sediments being relatively enriched in Ca and Sr but lower than that of YR sediments in most trace elements (e.g., Li, Ti, Ni, Ho; Yang et al., 2002; Yang et al., 2003). The sudden increase in Ca/Al ratios and Sr contents during 1385–1765 CE is illustrated in Figure S2 . Ti and Zr are naturally found in the earth crust and are widely used in the reconstruction of lithogenic flux records as indicators of terrestrial debris (Di Leo et al., 2002). The positive correlations ( Figure S4 ) between Ca and Ti contents (R = 0.54), Ca and Zr (R = 0.56), and Sr and Ca (R = 0.62) indicate that the Sr and Ca in the Core C2 sediments are sourced mainly from natural carbonate debris rather than bioclastic material. This is supported by variations of f_Riverine , which coinciding well with the trends of Zr. Most trace elements (e.g. Ti, Li, Ni) displayed a decreasing trend during 1385–1765 CE, with the f_Riverine of Core C2 also decreasing sharply. Additionally, the river sediments from the YR catchment generally have higher REE values, because of the wide occurrence of basic rocks in the basins. Relatively, the river sediments from the HR catchment have lower REE values because of the widely distribution of Quaternary loess in the HR middle reach (Yang et al., 2002; Mi et al., 2017). The relative lower REE values in the period of 1385–1765 CE, indicating that there are other sediment sources on the YRD during this period. However, terrigenous sediments from Minjiang, Taiwan, and Okinawa Trough have higher REE concentrations than the sediments from the YR (Mi et al., 2017). Therefore, we infer that the southward transport of HR sediment by the SCC to the YRD has increased since 1385 CE, changing the composition of sedimentary OM and reducing OC burial at the core site because of the relatively low contents of HR (0.11% ± 0.13%; Zhang and Wang, 2019) and AHD sediments (0.35% ± 0.13%; Qi et al., 2020) in the AHD. This is also supported by the abrupt increase Ca/Al values during this period ( Figure 7H ) from the core ECS-0702 near the core C2 ( Figure 1A ).

To further identify OM source changes in the C2 sediment, organic proxy results were compared with potential HR, YR, AHD, and marine OM sources ( Figure 8 ). Organic geochemical data for the period 1385–1765 CE are more similar to those of AHD ( Figure 8A ) and HR ( Figure 8B ) sediments than those of YR sediments, confirming that OM-deficient HR sediments contributed significantly to the Core C2 site, reducing OM burial during this period.

The f_Riverine values of Core C2 are not consistent with the fraction of riverine OM from core THB-2 in the southern part of Zhe-Min Mud zone (Hu et al., 2014) after 1385 CE ( Figures 7A, E ). The increased f_Riverine of Core THB-2 since 1260 CE may have resulted from the combined effects of intensive human activities and different sediment sources. Intensive human activities have enhanced sediment OM delivery to the sea since 1385 CE, increasing terrigenous OM deposition in the Zhe-Min Coastal Mud zone, with sedimentary OM there being sourced predominantly from the YR (Milliman et al., 1985; Liu et al., 2007). Although HR sediment may have affected the composition of sedimentary OM in the southern part of the Zhe-Min Coastal Mud zone, its overall influence may be neglected (Qi et al., 2020) because of the distance between the AHD and Core THB-2. In contrast, the significant contribution of AHD sediment (average 26%; Shang et al., 2021) with a low OM content reduced the OM content of the YRD, even though it involves mainly YR sediment. The timing of AHD sediment effects on sedimentary OM in the YRD is of interest. Coarse-grained-zircon U–Pb dating in the northern YRD indicates a significant influence of the AHD since ca. 500 yr BP (Shang et al., 2021). However, sedimentary records for the YRD indicate that AHD sediment has contributed greatly to the YRD over the last 600 years (Liu et al., 2010; this study). This inconsistency may have resulted from the OM tending to be enriched in fine-grained particles, with the clay–silt fraction being suspended more easily and transported further than that of the silt–sand fractions. The finer-grained sediment was thus carried earlier to the southern YRD, as suggested by Shang et al. (2021). Considering the chronological uncertainty, we conclude that the significant contribution of AHD sediments to the YRD began at least 600 years ago.