Source-to-sink process and risk assessment of heavy metals for the surface sediment in the northern South China Sea

Heavy metal input from the coastal urbanization and industrialization and their potential ecological risks have been a great concern in the northern South China Sea (NSCS). Our results showed that the surface sediments in the NSCS mainly include sand, silt, and sandy silt. CaO and Sr are fixed in the fine-grained biogenic calcareous debris. Sc, V, Cr, Co, Ni, Ba, and REE are the least contaminated and mainly of felsic crustal origin, with the supply largely from the Han and Pearl River estuaries in the Eastern and Western NSCS, respectively. Enrichment in Cu-Pb–Zn might be from both natural and anthropogenic inputs, and their contamination is generally at a low-risk level. As-Cd accumulation is mainly from anthropogenic provenance related to the aerosol precipitation. The anthropogenic As-Cd contamination resulted both from the industrial/urbanized discharge along the Guangdong Eastern Coastal and the agricultural/aquafarming activities along the Moyang River Estuary-Hainan Eastern Coastal.


Introduction
Heavy metal elements (e.g., Pb, Cd, Ni, Cr, Cu, Zn, As) are important contaminants in the environment due to their traits of multiple provenance supply, long residual time, difficult degradation, and easy accumulation (e.g., Buat-Menard and Chesselet 1979;Balkis et al. 2010;Apeti et al. 2012;Kelepertzis 2014;Aiman et al. 2016;Wang et al. 2013a;Liu et al. 2018;Ding et al. 2019). The heavy metal contamination might inhibit the growth of organisms and be detrimental to human health by food chain transmission. As the input of heavy metals from coastal areas might continuously increase due to rapid urbanization and industrialization (e.g., Xu and Han 2009;Wang et al. 2013a), heavy metal contamination in marine environments has been of great concern, and their enrichment effect is becoming a key indicator for assessing marine environmental quality and potential ecological risks Ding et al. 2019). For example, the coastal areas (e.g., Hong Kong and the Pearl River Estuary) of the northern South China Sea (NSCS) have been referred as "hot spots" areas for potential heavy metal pollution (e.g., Wang et al. 2013a).
The South China Sea is the largest marginal sea in the western Pacific (inset in Fig. 1). With an area of 3.3 million km 2 , its coastwise countries have mostly become densely populated and vibrant economic entities. Numerous suspended sediments and dissolved substances have been released into the SCS, with some contaminants (e.g., heavy metals) potentially threatening the marine ecological environment and human health. Available data show that weathering materials from the Asia continent into the SCS via rivers are up to 768 × 10 6 ton/a. About 30% of the Yangtze River sediment discharge is southerly transported into the SCS along the East China offshore (Liu et al. 2016;Ding et al. 2019). For the NSCS, the suspended materials are mainly from soil erosion of the South China continent, and rapid urbanization and industrialization of the South China Coastal Provinces, and they are transported and discharged via the Pearl, Moyang, Han and Jiulong rivers ( Fig. 1; e.g., Wehausen and Brumsack 2002;Wang et al. 2003;Liu et al. 2016;Xu et al. 2016).
In response to the discharge process, heavy metals and the related contaminants are commonly locked in the marine surface sediments (e.g., Wehausen and Brumsack 2002). The surface sediments are naturally the key carriers for monitoring heavy metal contamination and evaluating marine ecological risks of the NSCS (Marchand et al. 2006;Zhu et al. 2011;Wang et al. 2013a;Liu et al. 2018;Ding et al. 2019). Over the last three decades, numerous researches focused on the relationship of sediment supply from the Pearl River with Quaternary East Asian monsoon activity, and the heavy metal investigation at the Pearl River Estuary and its vicinity (e.g., Wehausen and Brumsack 2002;Tamburini et al. 2003;Wang et al. 2003;Yang et al. 2008;Zhu et al. 2011;Chen et al. 2012;Wang et al. 2013a;Zhao et al. 2017). However, relatively less attention has been paid to spatial patterns of heavy metal content for the surface sediments in the NSCS (e.g., Xu et al. 2016). It remains to be poorly known for their source-to-sink pathways and potential contamination

Sampling background and analytical methods
The northern SCS shelf and its south have been accepted as a passive continental margin (e.g., Yim et al. 2006). It is located to the south of Guangdong-Fujian Province, the east of Hainan, and the southwestern end of the Taiwan Strait (inset in Fig. 1). The shelf southerly extends to the shelf break near to 200 m isobath, which is usually subdivided into two hydrographic units of inner and outer shelves across the water depth of 80 m. The hydrologic characteristics of the inner shelf result in low salinity and high-suspended particle contents due to the input of terrestrial materials (e.g., Pan et al. 2015). However, the involvement of open oligotrophic materials inducing the outer shelf or its south is dominated by high salinity and low suspended particle contents (e.g., Su and Pohlmann 2009;Pan et al. 2015;Xu et al. 2016). Available data show that numerous terrestrial and urbanized materials at the riparian zones are annually fed into the NSCS, at which sediment discharge amounts are up to ~ 84.3 Mt/year, ~ 10 Mt/year, and 1 Mt/year via the Pearl, Han, and Moyang River estuaries, respectively (e.g., Yim et al. 2006;Zhang et al. 2012;Milliman and Farnsworth 2011;Liu et al. 2016).
With the northeastward SCS warm current separating from the branch of the Kuroshio and entering the NSCS from the Luzon Strait, the East Asian Monsoon is the prevailing climate in the NSCS and exerts distinct seasonal effects on the ocean circulation. It is signed by Guangdong Coastal Currents (GDCC), surface currents, warm currents, upwelling currents, and deep-water currents in the NSCS (Fig. 1;e.g., Fang et al. 2015;Liu et al. 2016). Driven by seasonal monsoons, the overall surface circulation in the SCS is dominated by cyclones (counterclockwise) in winter and anticyclones (clockwise) in summer, respectively. The GDCC flows northwesterly in winter but southeasterly in summer. In addition, the deep-water current turns southwest along the southeastern continental margin. Such currents eventually influence the transportation and discharge of the suspended sediments (e.g., Webster 1994;Fang et al. 2015;Liu et al. 2016). However, typhoons or tropical storms might have a minor impact on the source-to-sink process of the suspended materials in the NSCS (e.g., Liu et al. 2011;Wang et al. 2013a).
Thirty surface sediments were herein collected via the 2016-voyage at June-August 2016 (Fig. 1). The grab apparatus and sampling methods follow Cai et al. (2013) (Xu et al. (2016). Nineteen samples, with thickness of 5 cm, were collected using Van Veen Grab Samplers. The remaining 11 samples were separated from 1-cm sediment cores at the surface of the gravity column taken by a gravity corer with an automatic clutch and reverse catcher. Referred from Cai et al. (2013) and Xu et al. (2016), all the samples were sealed in clean polyethene zipper bags, refrigerated at 4-6 °C until analysis, packaged and transported according to the China Oceanographic Investigation Criteria (GB/T13909-92). Our sampling sites, along with those in Cai et al. (2013) and Xu et al. (2016), spatially constituted a relatively homogeneous distribution in the NSCS, as shown in Fig. 1.
Vacuum freeze-dried 150-mg sediment samples were pretreated for particle-size analysis. They were soaked in 10 ml 10% H 2 O 2 and 10 ml 10% HCl in a 50-ml clean centrifuge tube for 1 day to remove the organic matter and calcium carbonate. The suspension was centrifuged three times at 3500 rpm for 6 min in distilled water, and the supernatant was discarded each time. Adding 5 ml 0.05 mol·L −1 sodium hexametaphosphate solution, the sample was subsequently dispersed and homogenized by ultrasonic oscillation for 3 min. The particle size was measured using a Mastersizer 3000 laser particle analyzer at the Guangdong Provincial Key Laboratory of Geodynamics and Geohazards, Sun Yatsen University (GPKLGG, SYSU). The measurement repeatability error was less than 1% (Blott and Pye 2001).
Whole-rock surface sediments with organic matter and calcium carbonate removed were dried at 60 °C and then crushed in a steel mortar and pulverized to 200-mesh. Major oxide analyses were performed by ARL-Perform' X4200 X-ray fluorescence spectrometry at GPKLGG, SYSU. The relative standard derivations of the replicate samples were less than 5%. About 50 mg powdered sample was digested with the HNO 3 + HF mixture in a Teflon vessel for 32 h at 185 °C. The digested solutions were steamed on an electric heating plate, diluted to 4000 times with internal standard Rh and Re, and then analyzed using Thermo ICP-MS at GPKLGG, SYSU. Standard samples (i.e., GSR-1, -3 and -18, GSD-9, BCR-2 and BHVO-2) were used to monitor the instrument performance and data quality. The analytical results of particle sizes, major oxides, and trace elements for the 30 surface sediment samples are listed in Table 1.
As shown in Fig. 2b, the interval pattern of sandy and silty sediments was presented from west to east of the NSCS, at which the sedimentary transportation along the rivers channel of the Moyang, Pearl, and Han River estuaries resulted in the deposition of coarse-grained materials. From north to south, sub-parallel to the GDCC, the surface sediments defined a pattern of coarse-fine-coarse-fine interval, geographically corresponding to the water depth of < 40 m , ~ 40-90 m, ~ 90-200 m, and > 200 m. Such a configuration suggests strong hydrodynamic condition and distinct underwater geomorphology in the NSCS, at which the finegrained fractions (i.e., clay) bypassed the shelf and transported into slope and deep-water basin (Fig. 2b).
End-member analysis (EMA) for the particle size of surface sediments, as documented by AnalySize Software Model (e.g., Paterson and Heslop 2015), might provide key constraints on the sedimentary dynamic environment. Our data yielded three end-members of EM1, EM2, and EM3 (Fig. 2c). EM1 was constituted by the poorly-sorting 75.4% silt + 18.8% clay + 5.9% sand, showing a wide normal pattern with the size peak at 10.2 ± 2.9 µm. EM2 and EM3 were composed of the 93.6-98.8% well-sorted sand and 1.2-6.4% silt, signed by the narrow normal distribution with the sizepeaks of 133 ± 2 µm and 355 ± 2 µm, respectively (Fig. 2c). As shown in Figs. 1 and 3a-c, the sites with the EM1-predominant component (i.e., B01, B07, D02-1, D03-1, E01, E02, E08-1, F01, F07-1, G04, H03, and I02) geographically corresponded to relatively low-energy sedimentary environment and dominated by wind wave and tidal current. EM3 component, indicative of strong hydrodynamic condition, is signed by Sites A02, A03, F03, S19, S20, H01, S06, and S07 and spatially equivalent to the Pearl, Moyang, and Jiulong River Estuaries or their river channels. EM2preponderant components, signed by Sites A05, C02, C03, D06-1, and H02, geographically correspond to transform areas of EM1-and EM3-components in the NSCS. Hierarchical cluster analysis (HCA) yields two main clusters. One is marked by Site F01 to west of the Pearl River Estuary, at which the highest contents are given for all heavy metal elements. The other is constituted by sub-cluster A (H02 to S12) and B (A02 to S23), justly coordinating with the sites with fine (EM1) and coarse-grained (EM2 and 3) surface sediments (Fig. 4), hydrodynamically corresponding to the low-and high-energy environments, respectively.
The average content of all heavy metal elements for the surface sediments at the sites (except for F01) in the NSCS were lower than or near to those in the upper continental crust (UCC) and the main bays (i.e., Daya, Quanzhou, Xiamen, and Bohai) and estuaries (i.e., Hong Kong, Pearl, and Yangtze) in China ( Fig. 5a; e.g.,, Rudnick and Gao 2003;Gao and Chen 2012;Hu et al. 2013;Liu et al. 2015;Wang et al. 2013a;Zhao et al. 2017). They are also far lower than those in India Bengal Bay and Chirica Lagoon, Spanish Cádiz Gulf, and Turkish Candarli Strait ( Fig. 5a; e.g., Rodriguez- Barroso et al. 2010;Pazi 2011;Raju et al. 2011), but higher than those in Turkey's Gökova Gulf and Puerto Jobos Gulf (e.g., Balkis et al. 2010;Apeti et al. 2012). Thus, the marine environment of the NSCS sites is generally better than that of the domestic and overseas sea domains. In the UCC-normalized diagrams (Fig. 5b-c), their Sc, V, Cr, and Ni contents from the sampling sites show the sub-parallel patterns, distinctive from those for Cu and Zn. For Pb, As, and Cd, the spiky patterns are observed, respectively. The Pearson correlation (PC) calculation demonstrated good    Major oxide content: wt%; trace elements contents: μg/g correlations among Sc, V, Cr, and Mz (Φ), with PC coefficients > 0.85, and moderate correlations among Cu, Zn, Cd, Ni, Pb, and Mz (Φ), with PC coefficients of ~ 0.65, respectively. However, poor correlations (PC coefficients of less than 0.30) were given between As and other heavy metals, as well as particle-size for the surface sediments in the NSCS (Table 3).

Source-to-sink pathways for heavy metals and related elements
The source-to-sink pathways for chemical composition (i.e., major oxides, Co, Cu, Ni, Zn, Zr, V, and Sc) in the marine surface sediments might be controlled by many factors, such as carbonatization, provenance mixture, chemical weathering, sorting, hydrodynamic condition, and bottom currents (e.g., Cullers 2000). Our samples, along with those reported by Xu et al. (2016) and Cai et al. (2013), show that CaO contents range from 0.75 to 26.8 wt%, and linearly increase with increasing LOI and water depth at the depth of < 250 m (Fig. 6a). With the exception of several samples with veryhigh LOI content, LOI generally increases with increasing particle size (Fig. 6c). In addition, Sr positively correlates with CaO and LOI but poorly correlates with Sc or Al for the NSCS surface sediments (Fig. 5b). Available data show that the suspended carbonate particles are absent and carbonate contents are very low in the Pearl River Estuary and its fluvial sediments (e.g., Borges and Huh 2007;Xu and Han 2009). Such characteristics consistently indicate that CaO and Sr are mainly locked in the fine-grained calcareous debris, reflective of a prominent effect by biogenic carbonate rather than terrestrial fluvial input. The immobile trace elements (e.g., La, Sc, Co, Zr, Ba, Gd, Yb, Cr) might have effectively remained in the solid loads during erosion and sedimentation, and La/Sc, Th/Co, Co/Al, Cr/Al, and Zr/Sc are good discriminators for the provenance rocks (e.g., Bhatia and Taylor 1981;Wronkiewicz and Condie 1987). Th/Co and La/Sc ratios for the surface sediments in the NSCS are higher, but FeOt + TiO 2 are lower than those in the UCC. Co/Al 2 O 3 and Cr/Al 2 O 3 ratios are generally higher than those in the surface sediments of the Pearl River Estuary, but lower than those of MORB (Fig. 6d). Such signatures suggest their felsic rather than mafic provenance. In comparison with the surface sediments in the Western NSCS, Th/Co and La/Sc ratios in the Eastern NSCS are higher and FeOt + TiO 2 contents are lower. They have Co/Al 2 O 3 ratios ranging from 0.49 to 1.57 (peak at 0.9), 0.4 to 2.1 (peak at 1.0), and 1.25 to 3.92 (peak 1.6) and Cr/Al 2 O 3 ratios from 3.04 to 8.33 (peak at 4.7), 2.65 to 13.91 (cluster at 3.61-7.92), and 3.1 to 21.2 (peak at 5.3 and 12.6) in the Eastern, Western, and SW NSCS, respectively. The line of evidence shows that the Southern South China continent is dominated by granitoids with poor exposure of mafic rocks (e.g., Wang et al. 2013b). Thus, such a trend likely indicates a decreasing Al content from northeast to southwest in the NSCS. Taking into account the GDCC that flows southwestward (e.g., Wang et al. 1986), it is concluded that the Al-rich light minerals (i.e., illite and kaoline) with long-distance migration in the suspended materials might be controlled by the surface current (e.g., Liu et al. 2016;Cai et al. 2020).
Zr always trends to be concentrated in heavy minerals (i.e., zircon) that is highly resistant to chemical weathering, and preferentially sunk into the surface sediments. Therefore, Zr is a key element for tracing provenance composition and sedimentary sorting to some extent. Our data show that Zr/Sc ranges from 14.8 to 44.8 (average 29), 10.5 to 50.6 (average 24), and 6.6 to 21.7 (peak at 8.9) for the surface sediments from the Eastern, Western, and SW NSCS, respectively. Such signatures suggest that surface sediments of the Eastern and Western NSCS are mainly from the Han and Pearl River estuaries, respectively ( Fig. 6e; Gaillardet et al. 1999;. During summer, rainfall is always abundant in the Pearl drainage basin, inducing the production of numerous suspended loads discharged into the Western NSCS, and then southerly transported to the SW NSCS by waves, currents, and river channels. All these data synthetically indicate the felsic supply of the surface sediments in SW NSCS is mainly from the Pearl River rather than Han River. The marine surface sediments might be from natural/ crustal and/or anthropogenic/non-crustal provenance. Alumina and Sc are the structural elements of terrigenous aluminum silicates and the primary lithogenic component; thus, the correlations of Sc/Al with other elements are reliable indicators for evaluating terrestrial crustal or non-crustal provenance (e.g., Shumilin et al. 2002;Shevchenko et al. 2003;Carranza-Edwards et al. 2005;Zhang et al. 2009). K 2 O and Na 2 O, which are easily leached by chemical weathering (e.g., Nesbitt et al. 1980), are lower for our surface sediments (with average contents of 2.0 wt% and 1.43 wt%, respectively) than those in the UCC (2.80 wt% and 3.27 wt%, respectively), suggesting their intense chemical weathering in the drainage basin (e.g., Cai et al. 2013). Al 2 O 3 contents are lower, but SiO 2 are higher for the surface sediments from the NSCS than those from the Pearl River Estuary. SiO 2 /Al 2 O 3 ratios mostly range from 3.4 to 17.1 (average 7.8), greater than those from the UCC, but similar with those reported by Cai et al. (2013). Sc or Al (or Al 2 O 3 ) positively correlates with K 2 O, Na 2 O, FeOt, MgO, TiO 2 and P 2 O 5 , respectively. In addition, V, Cr, Co, Ni, Cu, Pb, Zn, Nb, Cs, Ba, REE, and Th linearly increase with increasing Sc, respectively. Such signatures indicate that the crustal provenance is related to the erosion of basement rocks and chemical weathering. Principal component analysis (PCA), conducted using SPSS® for Windows Release 16.0 (SPSS Inc. U.S.), shows that Cr, Ni, Cu, Zn, Pb, V, and Sc contribute 92-99% to the first principal component (PCA1) at 87% total variance. As PCA1 is highly correlated with Al 2 O 3 ; thus, such results argue either for their crustal origin or particle sorting (e.g., Windom et al. 2002;Liu et al. 2003).
The particle size is considered as one of the controlling factors affecting heavy metal contents of the surface sediments in response to the natural process (e.g., Horowitz and Elrick 1987;Shevchenko et al. 2003;Zhang et al. 2009;Zhu et al. 2011;Hu et al. 2013). Our surface sediments show that Al 2 O 3 , FeOt, MgO, TiO 2 , and P 2 O 5 positively but SiO 2 negatively correlated with particle size (Mz/Φ). For Sc, V, Cr, Co, Ni, Cu, Pb, Zn, Nb, Cs, Ba, REE, and Th, they positively correlated with particle size (Mz/Φ). Such signatures indicate that major oxides (except for CaO) and trace elements might be easily absorbed by the fine-grained aluminum silicate during hydrodynamic sorting and discharge, and they mainly are crustal provenance for the surface sediments of the NSCS. These samples exhibit the regularly decreased variation coefficient from Cd, Cu, As, and Pb, to V and Zn, and then to Sc, Ni, and Cr, indicating their increasing migration capacity. In combination with their high PC coefficients of > 0.85 and    Folk et al. (1970), b contour map, and c end-member analysis (EMA) frequency of the particle-size (Mz/Ф) for the surface sediments in the northern South China Sea. EMA was calculated on the basis of AnalySize Software Model (e.g., Paterson and Heslop 2015). Also shown for the surface circulation patterns (Liu et al. 2016;Fang 1998). Data sources in a are from this study and Xu et al. (2016) similar patterns in Fig. 5b-c, it is concluded that Sc, V, and Cr (or Ni) are mainly from terrigenous detrital components and share consistent source-to-sink pathways via river transportation. The estuarine areas or river channels with high-energy environment, characterized by the coarse sediments, are the main contributors of terrigenous detrital materials. Along two sides of the river channel, the hydrodynamic condition is relatively weak and the flow velocity rapidly decreases, inducing the deposition of fine-grained sediments and the discharge of terrigenous crustal-derived elements.  Fig. 3 Contour maps of a-c particle-size end-members (EM1, EM2, and EM3 with diameter-peaks at 10.2 μm, 133 μm, and 355 μm, respectively), and d-m Ni, Cr, V, Sc, Cu, Zn, Pb, Cd, and As contents for the surface sediments in the northern South China Sea, respectively Cu, Zn, Pb, and Cd might be derived from crustal materials with anthropogenic inputs and show the coupling sourceto-sink pathways, which are signed by their relatively low migration capacity, and similar moderate PC coefficients (~ 0.65) among Mz (Φ), Cu, Zn, Cd, Ni, and Pb (Table 3). Poor correlations are observed between particle size (and Sc) and As. Available data show that As, Cd, Cu, and Pb are usually sensitive to human activities. They might originate from the alloying, electroplating, dyeing industries, coal burning, and agricultural activities (i.e., aquafarming) besides the crustal origin (e.g., Gao and Chen 2012;Hu et al. 2013;Xu et al. 2015). For example, high As and Cd contents might generally be ascribed to multifarious sources related to rechargeable batteries, coatings and photovoltaic and aquafarming. The exacerbating accumulation of Cu-Pb and As are closely related to fossil fuel burning, vehicle traffic, and intensive usage of phosphate fertilizers, mariculture and soil amendments (e.g., Lambert et al. 2007;Zhang and Shan 2008;Kelepertzis 2014;Chambers et al. 2016). Thus, the poor and weak correlations of Sc (Mz/Φ) and As-Cd and Cu-Pb (Table 3) suggest that their spatial distributions were additionally influenced by the aerosols' precipitation of anthropogenic materials. Zhao et al. (2015) also proposed that the Cd contents for river sediments in the eastern Hainan Island were mainly derived from anthropogenic source. Zhuang et al. (2018) monitored the Cd and Pb pollution for the surface sediments at the Pearl River Estuary and found their main sources from the floating suspended materials related to automobile exhaust emissions. For PCA1 with 86.7% total variance, As and Cd contributed up to 69% and 58%, respectively. However, they contributed 72% and 78% respectively to the second principal component (PCA2) at 8.83% total variance, also suggesting significant anthropogenic contamination (e.g., Liu et al. 2003;Ip et al. 2004;Wong et al. 2003), as shown in Sites F01 and B01. In addition, the contour patterns for Sc and Cd-As contents are slightly overlapped. Thus, As and Cd are mainly from anthropogenic provenances, and Cu-Pb-Zn for Hierarchical cluster analysis (HCA) Fig. 4 Hierarchical cluster analysis (HCA) atlas of the sampling sites for the particle size from the surface sediments in northern South China Sea  S31 S21 S27 S20 S29 S28 S22 S2 S12 S14 S26 S4 S7 S16 S24 S25 S32 S33 S30 S13 S23 S8 S9 S1 S6 S11 S17 S19 S15 S5 S18 S3 S10 the NSCS surface sediments might be shaped by natural and anthropogenic dual effects.

N o r t h e r n S o u t h C h i n a S e a P e a r l R iv
As listed in Table 2, Cr-Ni contents for the surface sediments are generally similar among the NSCS, Guangdong Coastal areas, Hong Kong and Pearl River estuaries and Daya, Hailing Zhanjiang, and Dongzhai bays. However, the average contents for Cu, Pb, and Zn from the surface sediments of the NSCS are far lower than those from the Guangdong Eastern Coastal areas and the "hot-spots" estuaries and bays (i.e., Daya, Pearl, and Hong Kong), but similar with those from the Guangdong Western Coastal and Hainan Eastern Coastal areas (i.e., Hailing, Zhanjiang and Dongzhai, Table 2).  Rudnick and Gao (2003), Cai et al. (2013), and Xu et al. For example, our surface sediments from the NSCS have Cu contents nine times lower than those from the Pearl River Estuary and Benggai Bay (Zhao et al. 2017). For As and Cd, the average contents for the surface sediment in the NSCS are far lower than those of the sediments in the Guangdong Eastern and Western Coastal areas (Table 2). Our data also show that As and Cd contents for the surface sediments at aquafarming areas are significantly higher than those at the non-aquafarming areas, as observed in the Hailing Bay. These data collectively indicate that the anthropogenic enrichment in Cu-Pb-Zn and As-Cd at the Eastern NSCS to the east of Pearl River Estuary might mainly source from rapid urbanization and industrialization, but As-Cd enrichment might be from the agricultural and aquafarming activities at the Moyang River Estuary-Hainan Eastern Coastal of the Western NSCS.

Contamination assessment and ecological risks
As mentioned above, the heavy metal contents of the surface sediments in the NSCS might be from crustal and/or anthropogenic inputs. The anthropogenic impact would potentially result in ecological risks and environmental contamination (e.g., Müller 1971;Loska and Wiechula 2003;Shevchenko et al. 2003;Zhu et al. 2011;Wang et al. 2013a;Liu et al. 2018;Ding et al. 2019). The primary, secondary, and tertiary criteria (MSQ-1, -2, and -3) for marine sediments of GB 18668-2002 Standards are set by the China State Bureau of Quality and Technical Supervision. They are the references for assessing the ecological risks associated with metal toxicity to wildlife and human activities (MSQ-1), general industries and coastal tourism (MSQ-2), and harbors and offshore exploration (MSQ-3), respectively (Ding et al. 2019). The heavy metal contents for the sampling sites are mostly lower than or near to the MSQ-1 for the surface sediments in the NSCS except that Cu in two sites met with MSQ-2, suggesting a slightly-impacted contamination of the heavy metals for the NSCS surface sediments.
Threshold effect level (TEL) and probable effect level (PEL) are also important indicators for evaluating the potential ecological risks. Lower heavy metal contents are relative to TEL for the surface sediments, suggesting the rarely happening adverse biological effects. In contrast, the adverse biological effects happen frequently once the heavy metal contents are higher than PEL (e.g., Long et al. 1995;Xu et al. 2016). Our data, along with the reported data on the NSCS (e.g., Xu et al. 2016;Cao et al. 2019;Cai et al. 2020), show that the heavy metal contents were lower than PEL values at all sites. These sites with Cu, Zn, As, Pb, and Cd contents lower than TEL accounted for 97%, 97%, 87%, 77%, and 67%, respectively, suggesting a small likelihood of adverse biological effects on marine life. For Cr, Ni, and Cd, those sites accounting for 23%, 63%, and 33% are higher than TEL but lower than PEL values, respectively, suggesting occasional averse marine biological effects on the local aquatic ecosystems (Table 2).
Enrichment factor (EF) and geo-accumulation index (I geo ) methods have been considered as the key indicators in distinguishing anthropogenic from natural origins and quantifying the anthropogenic contamination degree for heavy metals of each environmental sample (e.g., Buat-Menard and Chesselet 1979;Zhang and Liu 2002;Shumilin et al. 2002;Shevchenko et al. 2003;Feng et al. 2004;Zhang et al. 2007Zhang et al. , 2009Wang et al. 2015;Aiman et al. 2016;Ke et al. 2017;Shirani et al. 2020). In order to rectify the particle size effects under different sedimentary environments, heavy metal contents for these samples should be firstly normalized by the contents of conservative elements, i.e., Al, Fe, Co, and Sc. We herein selected Sc as the normalized element in EF and I geo calculation (Zoller et al., 1974;Buat-Menard and Chesselet 1979). Enrichment factor (EF) is defined as (X sample /Sc sample )/(X baseline /Sc baseline ), where X sample , X baseline , Sc sample , and Sc baseline represent the heavy metals and Sc contents of samples, and background reference, respectively (Buat-Menard and Chesselet 1979;Rudnick and Gao 2003). EF values of 0.5-1.5 and > 1.5 are generally interpreted as the main contribution to crustal and anthropogenic provenances in the source-to-sink process, respectively (e.g., Müller 1971;Zhang and Liu 2002;Shevchenko et al. 2003, Yongming et al. 2006Cai et al. 2013;Diop et al. 2015). Zhang and Liu (2002) further proposed that EF values of 1.5-2.0 and > 2.0 reflect the minimal and significant heavy metal contamination in the environment, respectively. I geo is defined as log 2 (C n /1.5B n ), at which C n and B n are the measured and background contents of heavy metals, respectively (Müller 1971). Positive I geo values demonstrate significant heavy metal contamination, suggesting the level of human activities (e.g., Hu et al. 2013;Wang et al. 2015).
The average EF values for the surface sediments in the NSCS gradually decrease with the order of Cd > Pb > Zn > As > Cu > Cr > Ni > V. V, Cr, Cu, and Ni at all sites were the least contaminated. The EF values for Cd at these sites range from 0.4 to 5.2 with an average value of 1.6, and I geo values are more than 0.0 for half the sites, suggesting the main anthropogenic origin with moderate Cd contamination in the NSCS. The moderate enrichment in Pb, Zn, and As are observed, with the EF values ranging from 1.5 to 8.1 for Pb at 29 sites, 1.5 to 3.0 for Zn at 28 sites, and 1.5 to 19.8 for As at 16 sites, respectively. Such data reveal the crustal and/or anthropogenic provenances for enrichment in Cu, Pb, Zn, and As of the NSCS surface sediments with "slightly to moderately contamination level" at several sites (Fig. 7c). As shown in Fig. 7a-b, at Site F01 to west of the Pearl River Estuary, the EF and I geo values range from 1.71 to 3.79, and 0.70 to 1.04 for Cu-Pb-Zn and As-Cd, respectively, arguing for significant anthropogenic contamination. At the Site, Cd, As and Pb contents have the point-like enrichment, with their contamination up to the moderate-high level (Fig. 7a-c). For other sites in the NSCS, the I geo and EF values in Zn, Cu, V, Cr, and Ni are less than zero and 1.5, respectively, reflective of crustal origin with little human influence. However, at A02, A03, F03, F06-1, H01, and I01, EF values for Pb range from 2.46 to 8.12 in spite that their I geo values are less than 0.0, suggesting potential heavy metal contamination related to human activities.
As shown in Fig. 3j-l, Pb, Cd, and As contents in the surface sediments are significantly enriched along the Han-Moyang coastline and the Shenhu shoal. They show the "high-low-high" beaded patterns from west to east along the Sanya-Dongsha areas and generally higher at the domains to the west of the Pearl River Estuary than those to its east (Fig. 3k-l). A bead-like enrichment pattern has been observed along the Hainan Eastern Coastal. For Pb contents, the pattern with high-value contour lines is defined along the Han-Moyang River Estuaries, together with ribbon-like pattern along the Hainan Eastern Coastal (Fig. 3j). Such patterns for Cd and As likely argue for the multiple impacts of estuary desilting, coastal currents and aerosols precipitation of the crustal and anthropogenic provenances (Yang et al. 2008). At the sites along the Hainan Eastern Coastal areas (i.e., S2, S3, S8, S7, S15, H02, and I01), As and Cd contaminations have been observed, at which surface sediments are signed by fine-to medium-grained silt and sandy silt (Figs. 2 and 7c). As and Pb contaminants in the coarse-grained sandy sediments also exist to some extent, i.e., S21, E08, F06, A2, A3 and B01 (Fig. 7c). Such characteristics further illuminate that the anthropogenic input of As and Cd, and likely Pb, are closely related to the atmospheric aerosol transportation. All these data synthetically indicate that the ecological risks have a decreasing order from Cd > As > Pb, > Cu (Zn) > Cr (V), accordant with the values of the bioavailability of heavy  H a n R iv e r contamination in As and Cd contamination in As and Pb contamination in As-Pb-Cd-Zn metals in marine surface sediments of the NSCS. At our sampling sites, the heavy metal levels in the surface sediments are generally at low risk, with the exception of Site F01 (Table 2). However, more attention should be paid to As-Cd and Pb contamination from the industrial and urbanized discharge at the Guangdong Eastern Coastal areas, and As-Cd contamination from the agricultural and aquafarming activities along the Guangdong Western Coastal areas (Moyang River Estuary) and Hainan Eastern Coastal areas.

Conclusion
A comprehensive study on the heavy metals of the surface sediments in the northern SCS allows us to draw the following conclusions.
(1) The surface sediments in the NSCS are signed by the interval pattern of sandy and silty grains, with particle sizes peaking at 10.2 µm, 133 µm, and 355 µm for EM1, EM2, and EM3 end-members, respectively. (2) CaO and Sr are locked in fine-grained biocalcareous debris, but Sc, V, Cr, Co, Ni, Ba, and REE are mainly from felsic crustal provenance transported from the Han and Pearl River estuaries. (3) Cu-Zn-Pb contents in the NSCS surface sediments show the coupling source-to-sink pathways, with the natural and anthropogenic origin. (4) The potential ecological risk in the northern SCS is generally at a low level. As-Cd contamination along the Guangdong Eastern Coastal might be mainly related to the urbanized discharge, but to the aquafarming activities along the Guangdong Western and Hainan Eastern Coastal areas.