Evolution of the mud patch in the north Yellow Sea and its response to climate change in the past 160 years

As the most important sediment source to the north Yellow Sea (NYS), the sediment discharge from the Yellow River has substantially decreased for at least half a century. The response of the existing mud patches to this decline is not well understood. Here, we present high-resolution grain-size parameters and geochemical composition of one gravity core (B23) collected in the NYS mud patch to investigate the behavior of the mud patch variation in correspondence to the Yellow River’s sediment discharge reduction. The B23 age model is derived by 210Pb dating method, and the average sedimentation rate is estimated to be 0.49 cm/year. Results of a geochemical discrimination analysis suggest that the sediments of B23 were mainly transported from the Yellow River by the strong coastal current along the Shandong Peninsula, whereas the contributions of sediments from Yangtze and Yalu rivers are also measurable. In around 1980, the ratio of sediments from the Yellow River underwent an apparent decrease in the NYS mud. Simultaneous changes include coarsening of grain size and disaccord between East Asia Winter Monsoon intensity and content of sensitive grain size. These are all directly or indirectly due to the decline of the Yellow River’s sediment discharge.


Introduction
There are a number of mud deposits in the Yellow Sea, a semi-closed sea between the Chinese mainland and the Korean peninsula. These mud patches, from north to south, are the north Yellow Sea (NYS) mud patch, the Shandong mud wedge, the central Yellow Sea mud patch, and the southeastern Yellow Sea mud belt (Milliman et al. 1987;Park and Khim 1992). Previous studies have suggested that these mud deposits accumulated during the later stages of eustatic sea-level rise, at approximately 9-7.5 ka (Yang 1985;Shi et al. 1986;Milliman et al. 1987Milliman et al. , 1989. One common property of these mud patches is their thick deposit of mud from continuous, high-rate sedimentation that normally carries rich information on sediment source and sedimentary environment (Xiang et al. 2005). There have been numerous studies on the stratigraphic sequence (Jung et al. 1998), mineral composition (Park and Khim 1992;Yang et al. 2009), grain size characteristics , geochemical characteristics (Liu et al. 2009;Lan et al. 2015), biological characteristics (Kim and Kucera 2000), and its corresponding environmental significance of the Yellow Sea. However, the provenance and formation mechanism of these mud patches still remain controversial.
The mud area of the NYS is located in the southwest of the NYS, north of the Shandong Peninsula. After the present circulation framework became established in the Yellow Sea and its adjacent areas in the middle Holocene, sediment originating from the rivers of the Korean Peninsula (mainly the Yalu, Han, Keum, and Yeongsan rivers), the Yellow River, and the Yangtze River could all be transported by oceanic circulation to the mud areas (Liu et al. 2009). Over the past 2000 years, the Yellow River has annually delivered ∼1.1 × 10 9 t of sediment into the Bohai Sea (Milliman et al. 1987;Saito et al. 2001), with 1%-15% of total discharge being transported into the Yellow Sea directly (Alexander et al. 1991;Martin et al. 1993). The average annual sediment load of the Yalu River is 1.1 × 10 6 t and has decreased in recent years . In addition, the Yellow Sea warm current (YSWC) is believed to transport annually about 1 × 10 6 t of sediments originating from the Yangtze River into the NYS; about 0.2% of the total sediment discharge of the Yangtze River (Gao et al. 1996). Sediments from the main rivers of the Korean Peninsula can also be transported in part into the NYS during summer, but they only account for a small percentage of the total sediment load (Qi et al. 2004). Indeed Cheng and Gao (2000) showed that the NYS mud is a multi-source deposit, even though many other researchers argued that the Yellow River is the primary source of sediment in the NYS mud patch (Liu and Gao 2005;Li et al. 2014;Lan et al. 2015).
The sediment discharge of the Yellow River has been substantially declining for at least half a century (Yang et al. 2013), primarily because of increased aridity and the divergence of water for human consumption (Wang et al. 2006). The response of sediment deposit in the mud patches to this decadal-scale decline, however, has hardly been studied. Our aim here is, by analyzing a gravity core from the NYS mud patch that is nearest to the river mouth, to investigate the behavior of the mud patch variation in correspondence to the Yellow River's sediment discharge reduction. The sediment core (B23, see Fig. 1) is analyzed for the high-resolution variation of geochemistry, clay mineral, and grain-size properties that reflect the evolution of the sedimentary environment of the area. The impact of other forces, such as monsoons, other than sediment sources is also discussed.

Materials and method
The 80 cm long gravity core B23 was recovered in 2016, onboard R/V DongFangHong 2 from 52 m of water depth inside the NYS mud area (Fig. 1, 122°44′20″E, 38°13′12″N). In the laboratory, it was split, described, and subsampled in 1 cm intervals for grain size and geochemistry analysis.
A solution of 10% H 2 O 2 was added to samples to decompose organic matter, and 1 mol/L HCl was also added to remove carbonate. After dispersion and homogenization by ultrasonic vibration, the grain sizes of sediment samples were measured with a Malvern Mastersizer 3000 laser particle size analyzer (Malvern Instruments Ltd., Malvern, UK) in the range of 0.01-3000 μm (see Table S1 in the Supplementary Material 2 ). The relative deviation of reproducibility was 2%. Size parameters were calculated in accordance with the moment's formula (McManus 1988).
For rare earth element (REE) (La, Ce, Pr, Nd, Sm, Eu, Gd, Dy, Ho, Er, Yb, Lu, and Y) analyses, the bulk sediments were first oven-dried at 60°C overnight, then ground to powder and homogenized with an automated agate mortar and pestle. In each case, about 0.05 g of powdered sample was weighed into a graphite crucible and admixed with 1.0 g LiBO 2 . These mixtures were then fused at 900°C for 20 min, and the resultant cake was cooled and dissolved in 200 mL of cold, 5% nitric acid (HNO 3 ). Each solution was analyzed for REE contents using inductively coupled plasma mass spectrometry (ICPMS; Perkin-Elmer ELAN 5000) at the Ocean University of China (see Table S2 in the Supplementary Material 2 ). The detection limits ranged from 0.1-0.5 μg/g for La, Sm, Eu, Gd, Dy, Ho, Er, Yb, and Lu to 1 μg/g for Ce, Pr, Nd, and Y.
An age model of the core was obtained by 210 Pb dating at the Nanjing Institute of Geography and Limnology, Chinese Academy of Science. The 210 Pb activity of sediment was measured by alpha counting its daughter, 210 Pb. About 0.5 g of 208 Po spike was added to 1-3 g of dried sediment prior to leaching with HCl, ammonium citrate tribasic, and hydrogen peroxide. The acid solution was separated from the particles by centrifugation, dried, and then pickled in 1 mol/L HCl. An Ag planchet (1 cm × 1 cm) was added to the solution to remove the polonium isotopes. Supported levels of 210 Pb activity were determined by measuring 210 Pb below the region of excess activity.

Core lithology and sedimentation rate
The entire sediment core consists of muddy sediments of the generally consistent composition of clay, silt, and sand (Fig. 2). The compositions of silt and clay range from 55% to  Yang and Liu (2007). Arrows show the general trend of current systems. BSCC, Bohai Sea coastal current; LCC, Liaodong coastal current; KCC, Korean coastal current; YSWC, Yellow Sea warm current; YSCC, Yellow Sea coastal current. 73% (average value of 69%) and from 16% to 26% (average value of 23%), respectively. The median grain size (Mz) of the sediments mostly varies between 9 and 14 μm except for an enormous spike between 67 and 72 cm below the surface that reaches as high as 24 μm. The variabilities of standard deviation (σ), skewness (Sk), and kurtosis (Kg) are small, with good sorting, positive skewness, and high kurtosis. Similarly, anomalies also coincide with the 67-72 cm section. The B23 age model was conducted by the 210 Pb dating method (Fig. 3), and the excess 210 Pb data are shown in Table 1. Generally, a steady decline due to the decay of 210 Pb indicates that the sediment accumulated at a relatively constant rate in a stable sedimentary environment. In an unstable sedimentary environment, the excess 210 Pb profiles exhibit segmented patterns as a result of the changes in sediment supply, grain size, hydrodynamic-induced mixing, and biological perturbation (Li 1993). A best-fit linear regression with a determination coefficient R 2 = 0.74 was found in the normal decreasing trend within 80 cm, which yields an estimate of average sedimentation rate of 0.49 cm/year. Sedimentation rates in the mud areas around the Shandong Peninsula range from 0 cm/year to 0.96 cm/year with an average of 0.37 cm/year (Qiao et al. 2017), suggesting the sedimentation rate calculated from core B23 is valid.

REE parameters and patterns
Depth profiles of Mz, original REE concentration, and REE parameters in core B23 are shown in Fig. 4. Generally, fine-grained mud and silty mud have a higher content of REEs than that of coarse-grained sand and silt (Lan et al. 2009), but there is no obvious correlation between grain size and REEs in this core (in fact, the correlation coefficient between REEs and median grain size is below 0.2), suggesting a much more complex deposition process. The ratio of light REEs (from La to Sm; LREE) to heavy REEs (from Gd to Lu; HREE) varies between 7.2 and 9.0. The chondrite-normalized δEu value in all the samples shows negative Eu anomalies, in contrast to the slight positive anomalies prevailing in the δCe value, which means that Eu has been significantly different from the chondrite deposits, and the degree of differentiation is close to the continental crust. In addition, the content of La, Nd, and Sm has some changes around 22 cm in depth. Figure 5 compares the REE data from the core, normalized by the average REE mass fraction from the upper continental crust, with the published data for the rivers surrounding the study area (Yellow, Yangtze, and Yalu rivers of China, Han and Keum rivers of Korea). The patterns of core B23 show a middle REE enrichment from Sm to Tb with a slight Gd depression. Curves of different depths are basically linear, indicating low degrees of REE differentiation, which are similar to Yellow, Yangtze, and Yalu rivers and far from those of the Korean rivers (Mei et al. 2011).

Sources of sediments in the mud patch
To determine the source of the sediment in the NYS mud area, REE, the stable elements whose composition and distribution pattern in sediments mainly depend on the source rocks in the river catchment basin, are often used as a source tracer for sediments (Yang et al. 2003;Yan et al. 2013). The REE data from the gravity core (B23) provide qualitative indicators for the contributions from the three sediment sources. Similarities are clearly identifiable between the core and both Yellow River and Yangtze River when the mean value of ΣREE is plotted against the mean LREE/HREE ratio (Fig. 6a). In contrast, the REE indicator of the three Korean rivers is very different from those of the Chinese rivers and the core, perhaps due to the much higher values of ΣREE and LREE/HREE ratios in Korean rivers (Liu et al. 2009;Mei et al. 2011). Among the three Chinese rivers (Fig. 6b), REE clusters of the Yellow River in Gd/Yb versus La/Sm bear the highest similarity with the core. The several outlier points of core B23 fall into the cluster of the Yangtze River and even the Yalu River, suggesting a measurable contribution of sediment from these two rivers.

REE response to the decline of Yellow River discharge
Most of the sediment discharged from the modern Yellow River mouth is trapped in the subaqueous delta or within 30 km of the delta front by gravity-driven underflow (Wright et al. 1988). While in the winter season, the sediment re-suspension in the coastal region became extremely active due to the strengthening monsoonal climate . These re-suspended sediments could be transported into NYS by the monsoon-enhanced coastal currents (Wu et al. 2015). Therefore, the sediment of the Yellow River entering the NYS mud contains two parts: the fine fraction of the suspended sediment dispersed from the river mouth, and the coarse fraction of re-suspended sediment from the subaqueous delta. In fact, just the decrease in suspended sediment could cause the change of provenance in the NYS mud. The sediment load of the Yellow River declined substantially after 1970 due to the influence of human activities and climate changes (Wang et al. 2006), and this phenomenon is recorded in the REE parameters of core B23. The discriminant function (DF) (Li and Li 2001) is used to determine the closeness of the Yellow River's sediments to core B23. REE parameters, such as Ce/La, Sm/Nd, La/Yb, Sm/Lu, LREE/HREE, and δEu, can all be used as a basis for the discrimination Fig. 4. Vertical distributions of REE content and parameters with Mz in core B23. (Yang et al. 2003;Xu et al. 2009b;Mei et al. 2011). For two-endmember sedimentary system, DF can be defined by where i is the element or ratio of elements; c ix is the content of element i in sediments to be determined; c i1 and c i2 are the content of element i in the sediments of source areas 1 and 2, respectively. The smaller the DF value, the closer the sediment is to source area 1.  (1980). The Yellow River and Yangtze River data are from Yang et al. (2004), and other rivers' data are from Yang et al. (2003).  Qiao and Yang (2007), and Jiang et al. (2009), and data for the other rivers are from Yang et al. (2003) and Yan et al. (2010).
To distinguish the effects of sediment from the Yangtze River and the Yalu River on the NYS mud patch, whose main source is the Yellow River's sediment, DF hc and DF hy is calculated, respectively.
where h is the Yellow River, c is the Yangtze River, and y is the Yalu River. According to the REE patterns of Chinese rivers (Fig. 5), the concentration of medium rare earth element (MREE) in the sediments of the Yellow River and the Yangtze River is higher than that of the Yalu River, especially for Sm and Gd, while the concentration of HREE in the sediments of the Yellow River and the Yalu River is higher than that of the Yangtze River, especially for Yb and Lu. Therefore, Sm/Lu ratio and Sm/Nd ratio can be used as the basis to calculate DF hc and DF hy .
The values of DF hc and DF hy calculated for the past 160 years with REE from the core and river sediments (Fig. 7) are all below 0.5, suggesting the dominant status of the Yellow River's sediments. Besides, the contribution of sediments from the Yangtze River is higher and more stable to the NYS mud patch than the Yalu River, which varies a lot in 1855-1885 and 1975-2015. Combined with the change of river sediment transport, the source change of core B23 can be divided into three stages in the last 160 years. Stage 1 (80-64 cm), Fig. 7. Provenance evolution of core B23 and sediment discharge of surrounding rivers. REE data of rivers are the average values provided by Yang et al. (2003Yang et al. ( , 2004. Sediment discharge data of rivers are from Wang et al. (2006), Zhang et al. (2008), and Liu et al. (2015). corresponding to the period of 1855-1885, is characterized by relatively high DF hy values. Before 1855, the Yellow River entered the Yellow Sea from northern Jiangsu Province, whose sediments would first be transported southward under the influence of the Yellow Sea coastal current and then enter the NYS carried by the YSWC. Thus, the sediment source of the NYS mud patch was presumably dominated by the Yalu River (Fig. 8). After 1855, the Yellow River was diverted to its current location entering the Bohai Sea, the sediments of which began to become the main source of the NYS mud patch (Liao et al. 2015). The contribution of the Yalu River's sediment was relatively reduced, leading to this transitional period from 1855 to 1885. Stage 2 (64-22 cm) indicated the sedimentation under the main control of the Yellow River's sediments from 1885 to 1980 with relatively low DF hy values (0.2 on average). At stage 3 (at a depth of 22-1 cm), corresponding to the period after 1975, the sediment load of the Yellow River was significantly reduced, and consequentially the proportion of the Yalu River sediments in the sedimentary area increased, resulting in an increase in the DF hy value. It is worth noting that the DF hc value remains basically unchanged in the above three stages, indicating that the Yangtze River's contribution to the NYS mud patch is constant, in spite of the fact that the sediment load of the Yangtze River has also been decreasing.
However, there is a decadal lag between the change of REE indicators and the Yellow River sediment discharge, which may be explained by the following. First, it takes time for the sediment to be transported, in suspension or bedload, from the river mouth to the NYS mud patch. Second, the lag could be the result of age uncertainty. In establishing the age model of core B23, we assumed that the age of the top of the core is 2015, the time when the core was collected, and ages of other depths were calculated using the average sedimentation rate (0.49 cm/year). In addition, the decreasing sediment load of the Yangtze and Yellow rivers, together with the constant Yalu River's sediment discharge (Fig. 7), may have led to a decrease in sedimentation rate. If the core top had been disturbed (representing an unknown number of years before 2015), or the sedimentation rate was actually not a constant, a false lag could have been created. Therefore, the 10 year lag value should be taken with caution.

Influence of East Asia winter monsoon (EAWM)
In winter, the sediments in the southern Bohai Sea and along the Shandong Peninsula can be re-suspended by stronger wind-induced waves and currents and then transported eastward to the NYS (Qin and Li 1983;Alexander et al. 1991). Sensitive grain size (SGS) of core sediments is a common parameter that researchers use to reconstruct the history of EAWM of different time scales: the stronger the winter monsoon, the greater the SGS content (Xiang et al. 2006;Xu et al. 2009a;Zhou et al. 2014). In this study, SGS is extracted by calculating the standard deviation of different grain sizes (Boulay et al. 2003). One peak (37.2-88.4 μm) could be observed in the grain-size class versus standard deviation values (Fig. 9), and it represents a population of grains with the highest variability through time. In the entire core B23, the content of this size class is about 20%, while sediments coarser than 88.4 μm accounted for only 1%-4%, suggesting 37.2-88.4 μm sediments within SGS (37.2-88.4 μm) is the representative of a coarse particle of core B23. Figure 10 plots the EAWMII index (1948, He and Wang 2012 against the core's content of SGS. Prior to the 1980s, the two parameters appear to change in a correlated manner; the correlation coefficient of their linear fit can reach 0.73. But after the 1980s, even though the monsoon had weakened significantly, the content of SGS displayed an overall upward trend, with a correlation coefficient of nearly zero ( Fig. 11; Table 2). This suggests that the sedimentary environment of the NYS mud area underwent significant changes around 1980, and the EAWM's control of sediment grain size weakened substantially. This kind of change has also occurred in the south Yellow Sea, the reason for which is explained by the coarsening of sediment from the Yangtze River and the strengthening of the East Asian summer monsoon (Gao et al. 2016). As for the NYS, EAWM-forced coastal current carries the sediment mainly from the Yellow River. In the context of the reduction of sediment load in the Yellow River, the YSWC and the Liaodong coastal current continued to carry the sediment of the Yangtze River and the Yalu River into the NYS mud area, which could also lead to a decline in the EAWM control.
The sediment of the Yellow River that ended up in the NYS mud contains two parts: the suspended sediment dispersed from the river mouth, and the re-suspended sediment from the subaqueous delta and the coastal waters north of Shandong Peninsula. The median grain size of sediments deposited near the Yellow River mouth is about 10-15 μm and gradually coarsens to 40-50 μm toward the Bohai Strait (Yuan et al. 2016) suggesting that the re-suspended sediment should be coarser. The sediment load of the river decreased    10. Change of SGS content in core B23, EAWMII, and sediment discharge of the Yellow River. (a) EAWMII, a proxy of EAWM intensity, from He and Wang (2012); (b) the content of 37.20-88.40 μm in core B23; and (c) measured data of the Yellow River's sediment discharge, from Wang et al. (2006). dramatically around 1970, leading to the change of provenance in the NYS mud. The continuing decrease of Yellow River discharge after the 1980s is believed to be the major cause of coarsening of the mud patch that received presumably constant input of coarse sediments re-suspended from the subaqueous delta and the coastal waters north of Shandong Peninsula by the annual winter monsoon. Consequently, the content of SGS representing coarse particles gradually increased during 1980-2015 (Fig. 9b).

Conclusion
Sedimentary records from the gravity core B23 demonstrated the evolution and provenance change in the NYS mud patch. The results from REE indicate that the main source of the NYS mud patch is the sediments from the Yellow River, followed by sediments from the Yangtze River and the Yalu River with a smaller but considerable proportion, while the sediments from Korean Peninsula have little influence on this area. From 1855 to 1970, sediments from the Yellow River were dominant. After 1970, as the sediment discharge from Table 2. The year of EAWMII and the vertical position of B23 SGS in the linear fitting process (Fig. 11).
EAWMII (year) SGS (cm)   2007  5  2006  6  2005  7  2003  8  2002  9  2001  10  1999  11  1998  12  1997  13  1995  14  1993  15  1992  16  1990  17  1989  18  1988  19  1987  20  1986  21  1985  22  1984  23  1983  24  1975  25  1973  26  1969  27  1968  28  1967  29  1965  30  1962  31  1961  32  1960  33  1959  35  1958  37  1956  38  1955  39  1954  40  1953  42  1952  43  1951  44  1950  46  1948  47 the Yellow River dramatically decreased, the proportion of sediments in the Yellow River decreased accordingly. It is worth noting that there is a decadal lag between the changes displayed by the sedimentary core and the discharge of the Yellow River. Along with the shift of sources, some new changes have occurred since the 1980s, including the disaccord between EAWM and SGS and the coarsening of grain size. Because the EAWM-driven coastal current transports the Yellow River sediments to the NYS mud patch, its control over the content of SGS in the NYS mud patch naturally decreased when the sediment discharge from the Yellow River was reduced. The coarsening of grain size is mainly due to the decrease of the fine fraction of the suspended sediment dispersed from the Yellow River mouth, leading to the increased ratio of the coarse fraction of re-suspended sediment from the subaqueous delta and the coastal waters north of Shandong Peninsula.