1 Introduction

The East Asian monsoon (EAM) is one of the most important components of the global atmospheric circulation. It deeply influences the climate in China and adjacent area (Chen et al., 1991). As a result, it attracts the attention of many climatologists (e.g., Wang et al., 1995; Wang et al., 1999; An et al., 2000; Wang et al., 2005; Steinke et al., 2011; Liu et al., 2017; Li et al., 2018). The EAM is formed as a result of thermal differences between the Asian landmass and the Pacific Ocean, and is further enhanced by the thermal and dynamic effect of the Tibetan Plateau. The EAM is characterized by prominent seasonal changes in wind direction, precipitation and air temperature between summer (hot and humid) and winter (cold and dry) (An, 2000).

The South China Sea (SCS) is one of the largest marginal seas along the Asian continent (Fig. 1). It is surrounded by the southern China and Vietnam to the north and west, and the Philippine Islands and Borneo to the east and south. The SCS is a semi-closed basin during the glacial periods, especially during the Last Glacial Maximum (LGM), when the gateway to the Sulu Sea through the Balabac and Mindoro Straits was closed, and the Sunda shelf emerged, leaving the Luzon Strait being the only gateway to the Pacific Ocean (Wang and Wang, 1990). Because of the high sedimentation rate and carbonate preserving efficiency, sediments from the SCS are ideal samples for paleoceanographic studies (Qian, 1999). The northern SCS provides an ideal site for paleo-climate and sedimentology study for the following reasons: 1) the complexity of detrital sediment provenances, 2) the EAM- dominated upper layer circulation, 3) the intrusions of the subsurface Kuroshio, and 4) the deep waters from the Western Pacific through the Luzon Strait (e.g., Webster, 1994; Fang et al., 1998; Qu et al., 2006; Liu et al., 2010; Fig. 1). Previous SCS paleoceanographic studies mainly focused on the long term trends at glacial-interglacial scales (e.g., Wang et al., 1995; Li et al., 2004; Wan et al., 2007). Here we presented the records from a sediment core recovered from the northern SCS continental slope, aiming to reconstruct the history of the EAM during the post-LGM period.

2 Materials and Methods

The gravity core ZHS-8-1 (115°10'E, 19°07'N, see Fig. 1; water depth, 1950 m; recovery length, 183 cm) was recovered from the northern SCS continental slope in 2005. The grain size analysis was conducted on 92 subsamples with a 2 cm interval. For sediment pre-treatment, excess H₂O₂ (30%) and HCl (1 mol L⁻¹) were added to remove the organic matter and carbonate in bulk samples, respectively. Then 5–10 mL sodium hexametaphosphate (0.5 mol L⁻¹) was added to disperse the bulk sediment samples. Grain size analysis was performed using a Mastersizer- 2000 laser particle size analyzer (range: 0.01–2000 μm) at Key Laboratory of Submarine Geosciences, State Oceanic Administration, China.

Organic carbon analysis was also performed on the 92 subsamples. After treatment with 1 mol L⁻¹ HCl and freezedried, the decarbonated samples were analyzed for total organic carbon (TOC) contents and stable carbon isotopic compositions (δ¹³C) using elemental analyzer (Thermo EA1112) connected to a Thermo Finnigan Delta plus AD mass spectrometer by a Conflo Ⅲ interface. The average standard deviations of these measurements were ±0.01% for TOC, and ±0.2‰ for δ¹³C. Values of δ¹³C are expressed in standard delta notation relative to the Pee Dee Belemnite. The measurement of organic carbon contents and carbon isotopic compositions were performed at the Key Laboratory of Submarine Geosciences, State Oceanic Administration, China.

Planktonic foraminifera G. sacculifer (10–15 mg) were picked for AMS ¹⁴C dating at four depths along the core (Table 1). The measurement was performed at Beta Analytic Inc., USA. Age model of core ZHS-8-1 was constructed based on the linear interpolation and extrapolation of these four dating results, which were converted to calendar ages (Table 1) using the software CALIB 7.10 (Stuiver and Reimer, 1993). A regional deviation from the global reservoir effect (ΔR) of 140 ± 45 yr was considered (Wang et al., 1999).

3 Results 3.1 Chronology

Because there is turbulence in the sediments at the bottom of the core, we choose the segment at top 140 cm to discuss its chronological features in this study. The age of the earliest sediment determined in this study is about 17.0 kyr BP. The sedimentation rate is relatively steady with an average of 8.2 cm kyr⁻¹. Accordingly, the 2-cm sampling interval results in an average time resolution of about 240 years.

3.2 Grain Size Analysis

Core ZHS-8-1 mainly consisted of gray homogeneous sandy mud and mud (Fig. 2). The fluctuation trends of grain size parameters, including mean grain size, sorting coefficient, skewness, and kurtosis are shown in Fig. 2. The mean grain size varies in the range of 3.5–25.0 μm, with an average of 6.6 μm. The sorting coefficient indicates poor sorting, with an average of 2.3 φ. The mean value of the skewness is −0.2, implying negative skewness. The kurtosis oscillates between 0.7 and 1.6 (Fig. 2).

3.3 Organic Carbon Analysis

The content of TOC ranges from 0.70% to 1.78% with a mean of 1.12% in core ZHS-8-1 (Fig. 3). The averaged TOC value during the last glacial period is 1.49%, while that value during the Holocene is 0.92%. The organic δ¹³C ranges from −21.1‰ to −19.6‰ with a mean of −20.4‰ (Fig. 3). The averaged δ¹³C value is −20.3‰ during the last glacial period and −20.6‰ during the Holocene.

4 Discussion 4.1 Sources of Organic Matters

The δ¹³C value is a common index used to identify the sources of TOC. The organic matters in the sediments from the continental margin are mainly from terrestrial and marine sources. The organic matter produced from atmospheric CO₂ on land has an averaged δ¹³C_terrestrial values of −27.0‰ and −14.0‰ for C₃ and C₄ plants, respectively, and organic matter produced from aquatic bicarbonate by algae has an averaged δ¹³C_marine value of −19.0‰ (Fry and Sherr, 1984; Meyers, 1997). In general, the input of organic matters from C₄ land plants would influence the δ¹³C value of the sediments. But C₄ plants have a very limited distribution in the Pearl River catchment and surrounding drainage area, where the natural ecosystem is subtropical forest and the dominant cultivated plant is rice (C₃ plant) (Jia and Peng, 2003). Therefore, the contribution from C₄ plant will not be discussed in this work. The δ¹³C values of sediments in core ZHS- 8-1 show a mixing marine and terrestrial origin (Fig. 3). To estimate the proportions of terrestrial organic matter in the sediments, we set a two-end member mixing model as follows (Schultz and Calder, 1976; Minoura et al., 1997):

$f\% = ({{\rm{ \mathit{ δ} }}^{13}}{{\rm{C}}_{{\rm{marine}}}} - {{\rm{ \mathit{ δ} }}^{13}}{\rm{C}})/({{\rm{ \mathit{ δ} }}^{13}}{{\rm{C}}_{{\rm{marine}}}} - {{\rm{ \mathit{ δ} }}^{13}}{{\rm{C}}_{{\rm{terrestrial}}}}) \times 100\%, $

where f % is the content ratio of terrestrial organic matter to the total organic matter. The δ¹³C values of organic matters in core ZHS-8-1 varied from −21.1‰ to −19.6‰, and terrestrial organic matter accounts for 8.0% to 25.7% (averaged value of 17.1%) of the total organic matter (Fig. 3). The averaged content of terrestrial organic matter is 15.7% during the last glacial period, and 19.4% during the Holocene. These results exhibit mixing feature of terrestrial and marine organic matters, and the marine organic matter is dominated in core ZHS-8-1.

The downcore variations of δ¹³C and terrestrial organic matter contents are considerably different during the Holocene and last glacial period (Fig. 3). During the last glacial period, enhanced winter monsoon strengthens the upwelling, which intensifies the mixing of water and nutrients (Huang et al., 1997; Jian et al., 1999; Higginson et al., 2003). The increased primary productivity increases the supply of TOC to the sediments. Meanwhile, the coastline is at about the present −100 m isobaths during that period (Hanebuth et al., 2000, 2009; Liu et al., 2004). The influence of the terrestrial organic matter increases due to the seaward proceeding of the paleo-estuary. However, the amplitude of the increase for terrestrial organic matter is lower than that of marine organic matter. Therefore, the percentage of terrestrial organic matter is relatively low, and δ¹³C values of organic matters are high during the last glacial period (Fig. 3). The content of marine organic matter can reach about 92%. The content of terrestrial organic matter reaches a peak value at around 14.8 kyr BP, which corresponds with the meltwater pulse (MWP)-1a (Fig. 3). During that period, the sea level rises from present −95 to −80 m quickly, the Taiwan Strait may re-open (Zhou et al., 2008), and the summer monsoon strengthens, which was reflected by the planktonic foraminiferal oxygen isotopic compositions (e.g., Wang et al., 1999; Ge et al., 2010). All of these features mean that more terrestrial organic carbon is transported into the study area.

After entering the Holocene, the reduced content of marine organic matter indicates a lower primary productivity. The content of terrestrial organic matter has generally been increasing since the early Holocene (Fig. 3), which may indicate the increasing of the riverine derived materials, corresponding to the enhanced summer monsoon. Meanwhile, lots of inorganic clastics are also transported into the study area by rivers, whose dilute effect can also reduce the content of TOC (Ding et al., 2015). This climate transition is also recorded by the δ¹⁸O value of stalagmite D4 from Dongge Cave (Fig. 3; Dykoski et al., 2005). The variation trend of the terrestrial organic matter content in the sediments follows the general variation of incoming solar radiation (Fig. 3). However in the Holocene, peak monsoon intensities do not exactly coincide with the peak insolation values as they do in the last interglacial and glacial periods. The comparison (Fig. 3) between the content of terrestrial organic matter and the average insolation at 25°N (Dykoski et al., 2005) exhibited a lag of the former, which is also observed in other Holocene monsoon records (Overpeck et al., 1996; Fleitmann et al., 2003; Dykoski et al., 2005). A correlation between monsoon strength and North Atlantic warmth in the early-mid Holocene suggests that glacial climate boundary conditions, in addition to solar insolation, influenced the climate in this period (Overpeck et al., 1996). It was only after the ice sheets had retreated that insolation forcing began to dominate the monsoon. The sediment record in core ZHS-8-1 provides an evidence to support this as a possible explanation.

4.2 Implication of Sensitive Grain Size Component Since 17 kyr BP

It has long been recognized that grain size distributions of most hydraulic and aeolian sediments are polymodal and represent different transport or deposition processes. Due to the different sources of terrestrial materials and driving forces, the sediment stratigraphy and grain size compositions are different and complex even in the same sedimentary environment (Gao and Collins, 1998). Therefore, grain size parameters of the bulk samples cannot be simply used as environmental indicators in sedimentary studies. It is desirable to separate specific grain size components in bulk samples. Although it is impossible to physically isolate a specific component from the bulk samples, theoretical partitioning can provide useful information for understanding the transportation and deposition processes affecting a specific grain size component (Sun et al., 2002).

There are several mathematical methods to partition grain size components of sediments, which mainly include the fitting function method based on end member modeling of grain size (Weltje, 1997), the Weibull distribution method (Sun et al., 2002), and grain size class vs. standard deviation method (Boulay et al., 2003; Sun et al., 2003). In this study, we define the most sensitive grain size class using the method of the highest standard deviation (Boulay et al., 2003; Sun et al., 2003), which provides direct identification of the grain size intervals with the highest variability along a sedimentary sequence. We plot the grain size classes against their corresponding standard deviation to extract the sensitive grain size component of core sediments. Two peaks (high standard deviation) are identified at 4 μm and 250 μm, respectively. In addition, we found a low standard deviation value at 32 μm (Fig. 4). High standard deviation values indicate high variability in the sample group at the corresponding grain size classes. So there are two environmental sensitive grain size components for sediments in core ZHS-8-1. As the grain size classes of fine-grained sediments from the Pearl River are usually less than 31 μm (Jia et al., 2005), we choose 4 μm as the sensitive grain size to reconstruct the paleoclimatic records.

The 4 μm grain size component belongs to fine silt, and the contents of this component display an in-phase correlation with the sea level curve during the last glacial period (Fig. 3). Core ZHS-8-1 is located on the northern SCS continental slope with a distance of 500 km away from the Pearl River Estuary (Fig. 1). The continental shelf seaward off the estuary is shallow and wide. This implies that local deposition environment is extremely sensitive to sea level changes. When the coastline retreated to around −100 m isobaths during the last glacial period (17.0 kyr to 14.8 kyr BP) (Hanebuth et al., 2000, 2009; Liu et al., 2004), the shelf was extensively exposed, and the position of core ZHS-8-1 was closer to the paleo-Pearl River Estuary, thus received more sediments from the exposed continental shelf. Most of the Pearl River sediments are coarser than clay fraction (He, 1992). Therefore, the physical reworking coarse detrital sediments and riverderived coarse sediments diluted the content of fine silt during the low sea level stand. As the sea level rise, the content of 4 μm grain size component increases gradually (Fig. 3). With the acceleration of sea level rise during the last deglaciation period (14.8 to 11.5 kyr BP), the shelf was submerged gradually, and the Pearl River Estuary gradually retreated from the location of core ZHS-8-1. As a result, the content of 4 μm grain size component increases quickly (Fig. 3). The content of 4 μm grain size component has a sudden decrease at around 12.5 kyr BP (Fig. 3), which implied a dramatic change of the terrestrial input. This situation is also found by Huang et al. (2011). Xu et al. (2009) proposed that the Zhejiang-Fujian Coast Current began to form around 12.3 kyr BP. The terrestrial sediments are transported by the southwestward Chinese Coastal Current, Kuroshio Current and deep water current. Therefore, the SCS modern current system has been formed since 12.5 kyr BP, and become the major controlling factor of terrestrial input. The influence of sea level rise on terrestrial input has been weakened since then.

As mentioned above, from the start of the Holocene, the influence of sea level change weakens and the roles of current system and provenance strengthen. During the early Holocene (11.5 to 7.0 kyr BP), the sea level also rose gradually. However, the main source of the sediment shifts from Pearl River catchment to Taiwan in this period (Ge et al., 2010). As the continental shelf off SW Taiwan is narrow, the rise in sea level may not significantly affect the estuary position. Meanwhile, the summer monsoon, which is recorded by oxygen isotopes of stalagmite D4 from Dongge Cave in southern China, is strong (Figs. 1 and 3; Dykoski et al., 2005). The strengthened summer monsoon resulted in the intensified precipitation, and more fluvial sediments are able to be transported into the sea. Therefore, the strengthened current, intensified supply and the gradually rising sea level together slowly decreased the content of 4 μm grain size component during the early Holocene (Fig. 3). The sea level has been relatively stable since 7.0 kyr BP (Fig. 3; Liu et al., 2004), and the ocean dynamics and sedimentary environment may be same as those in modern SCS. The fluctuations of 4μm grain size component content are relatively small during this period (Fig. 3). The small variations may be interrelated with the strength changes of winter monsoon. The strong winter monsoon will strengthen the Guangdong Coastal Current, which can re-suspend and transport those previously deposited coarse riverine sediments on the continental shelf to the slope (Yang et al., 1992). Therefore, the contents of 4 μm grain size component are low when the winter monsoon picked up its strength after 7.0 kyr BP.

5 Conclusions

In this paper, we combined the organic carbon contents and the grain size analysis to determine the sources of organic carbon in the core sediments and rebuild the history of East Asian monsoon since 17.0 kyr BP. The results indicated that the marine organic matter was dominated during the last glacial period. And the contents of terrestrial organic matters increase gradually after entering the Holocene. The contents of environmental sensitive grain size component are mainly controlled by the sea level changes during the last glacial period. With the increase of summer monsoon intensity, the influence of sea level changes weakens and the roles of current system and provenance strengthen during the Holocene.

Acknowledgements

This work is supported by the National Programme on Global Change and Air-Sea Interaction (Nos. GASI-GEO GE-03 and GASI-04-01-02), and the National Natural Science Foundation of China (Nos. 41476047, 41106045, 41506064 and 41427803). We gratefully thank Profs. X. G., Yu, and W. Y., Zhang, from the Second Institute of Oceanography, Ministry of Natural Resources for the assistance in the laboratory.