3.1 Descriptive statistics for the concentrations of major elements and REEs
The descriptive statistics for the concentrations of major elements and REEs in sediments collected from the CSHI have been presented in Supplementary Table 1. The average concentrations (as a percentage) of SiO2, Al2O3, Fe2O3, CaO, MgO, K2O, Na2O, MnO, P2O5, and TiO2 were 60.48%, 11.80%, 4.68%, 5.51%, 2.02%, 2.16%, 1.62%, 0.06%, 0.10%, and 0.62%, respectively. In addition, the average concentration of TOC was 0.58%. The average concentrations (as µg/g) of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu were 35.82 µg/g, 70.83 µg/g, 8.25 µg/g, 28.85 µg/g, 5.79 µg/g, 1.09 µg/g, 4.12 µg/g, 0.78 µg/g, 3.66 µg/g, 0.79 µg/g, 2.65 µg/g, 0.40 µg/g, 2.19 µg/g, and 0.36 µg/g, respectively. The total concentration of REEs (ΣREE) was between 44.21 µg/g and 221.68 µg/g (mean of 165.57 µg/g). The concentration of light REEs (LREEs, mean of 150.62 µg/g) was higher than for the heavy REEs (HREE, mean of 14.95 µg/g) in the sediment samples. Wilding (1985) divided coefficient of variation (CV) into three degrees of variation, namely low variation (CV < 0.16), moderate variation (0.16 < CV < 0.36), and high variation (CV > 0.36). The variation coefficients of La, Ce, Pr, Nd, Sm, Eu, Tb, Dy, Ho, Er, Tm, Yb, and Lu displayed a moderate degree of variation, namely 0.22, 0.22, 0.22, 0.24, 0.20, 0.24, 0.33, 0.33, 0.29, 0.32, 0.31, respectively (Supplementary Table 1). The result showed that Gd had a high degree of variation, with a variation coefficient of 0.44. In conclusion, all the REEs of the CSHI show moderate to high variability, indicating that they may be affected by human activities.
3.2 The distribution of REEs across the CSHI
REE concentrations in surface sediment samples of the CSHI are shown in Supplementary Table 1 and Fig. 2. Concentrations of the REEs in samples ranged from 44.21 µg/g to 221.68 µg/g. In the measured REEs, the content of Ce (16.00 ~ 94.21, mean 70.83) was the highest, while the contents of Tm (0.1 ~ 0.73, mean 0.40) and Lu (0.08 ~ 0.65, mean 0.36) were the lowest. All samples showed that LREE (La, Ce, Pr, Nd, Sm, and Eu) were more enriched than HREE (Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu). The LREE/HREE ratio ranged from 7.81 to 16.87. The distributions of REEs are shown in the Fig. 3. The concentrations of REEs in the north of study area were higher than in the south. The LREEs in the south are more abundant than in the north, which is shown by the higher LREE/HREE values in south than in the north.
3.3 Comparison of REE concentrations from the CSHI with the results from other areas
The mean concentrations of REEs in marine sediments collected from other continental shelves have been presented in Supplementary Table 2. The mean concentration of REEs observed in the CSHI were significantly higher than those reported for the Rea Sea (El-Tahera et al., 2019), Western Gulf of Thailand (Liu et al., 2019), Western Sunda Shelf (Wu et al., 2020), Chukchi Sea (Astakhov et al., 2019), and Eastern Gulf of Tigullio (Consani et al., 2020). However, the mean concentrations of REEs in the CSHI were similar to those reported for the East Siberian Sea (Astakhov et al., 2019). Compared with other areas in China, the mean concentrations of REEs in the sediments collected from the CSHI were similar to those reported for the Yellow Sea (Mi et al., 2020), Bohai Bay (Zhang and Gao et al., 2015), and East China Sea (Mi et al., 2020). It is worth noting that the continental shelf of China has a higher background value of REEs, and is more likely to reach the critical value of pollution because of human activities.
3.4 Enrichment factor (EF) of REEs
The enrichment factor (EF) of REEs in surface sediments can be used to evaluate the degree of anthropogenic pollution (N'Guessan et al., 2009). Al, Fe, and Ti are often used as reference elements due to their chemical stability (N'Guessan et al., 2009). In this study, the concentration of Ti vs. REEs showed a low coefficient of variation (0.27) and a high correlation (0.73), so Ti was used as a reference value to calculate the EF. The EF values of REEs in sediments were calculated by using formula (1).
EF = (Ci/Ti)sample/(Ci/ Ti)background (1)
(Ci/Ti)sample was the ratio of REEs and Ti concentrations in the sample, while (Ci/ Ti)background was the ratio of REEs and Ti concentrations in the upper continental crust. Sutherland (2000) assigned levels of pollution to one of five classes according to the EF value as follows: (i) EF < 2 (no or minimal pollution); (ii) 2 ≤ EF < 5 (moderate level of pollution); (iii) 5 ≤ EF < 20 (significant level of pollution); (iv) 20 ≤ EF < 40 (very high level of pollution); v) EF > 40 (extremely high level of pollution).
The calculated EF values for REEs for this study have been presented in Supplementary Table 3. The ranges (mean in parentheses) of the EF of La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu in the study area were 0.76–4.58 (1.24), 0.74–3.36 (1.21), 0.73–7.80 (1.27), 0.68–6.26 (1.16), 0.73–4.17 (1.33), 0.60–3.75 (1.16), 0.22–3.58 (1.06), 0.22–3.32 (1.19), 0.66–2.59 (0.98), 0.5–2.01 (1.00), 0.59–2.43 (1.24), 0.75–2.53 (1.41), 0.77–3.20 (1.15), and 0.67–4.69 (1.23), respectively. The mean EF values could arranged from highest to lowest as follows, Tm > Sm > Pr > Er > La > Lu > Ce > Tb > Eu > Nd > Yb > Gd > Ho > Dy. The mean EF values of REEs in surface sediments ranged from 0.98 to 1.41 (Fig. 4), which indicates all samples could be assigned to the no or minimal pollution class (Sutherland, 2000). However, the maximum values for individual elements indicate different levels of pollution. The maximum EF values of Pr and Nd were 7.80 and 6.26 respectively, consistent with the significant pollution class described above. The maximum EF values of the other rare earth elements are between 2 to 5, which would indicate moderate levels of pollution. The distribution of EF for REEs in the CSHI have been showed in the Fig. 5. There is no pollution by REEs in the northern reaches of the CSHI. However, the EF values of REEs in the southeast of the shelf near Hainan Island are the highest, which could be attributed to human activity. This may be related to the close proximity of this area to Wenchang City, a city with a range of industries and port facility.
3.5 The correlation analysis between major elements and REEs in the sediment
The correlations between major elements and REEs are shown in Fig. 6. The depth of color in the graph is based on the strength of the correlational relationship between the variables. REEs are correlated with TiO2 (always ≥ 0.66), indicating that REEs are hosted in rutile. There is no correlation between TOC and REEs, indicating that the content of organic matter has no effect on the distribution of REEs in sediments. It is well known that acidic mine drainage can form small-scale (colloidal) "iron hydroxide" deposits (Ayora et al., 2016). These hydroxides can enrich REEs by precipitation and adsorption (Barcelos et al., 2018), and then enter the sediments through the so-called "salting out" process (Sholkovitz and Szymczak, 2000; Kulaksiz and Bau, 2007). This also explains the high correlation between Fe2O3 and REEs (always ≥ 0.44). There is also a high correlation between Al2O3 and REEs (always ≥ 0.41), while REEs may also occur in clay minerals. There is a significant correlation among the REEs, which indicates that the factors controlling the spatial variation in the concentrations in the sediments should be the same (Wang et al., 2019). In general, the main factors affecting the distribution of REEs in this area are the composition of naturally occurring minerals and industrial pollution.
3.6 Sediment source discrimination
The upper continental crust (UCC)-normalized REE concentrations are often used to identify the source of sediments (Hossain et al., 2010; Wu et al., 2020). The results of UCC-normalized REE concentrations are showed in Fig. 7. The sediments of the CSHI can be divided into three provinces.
Province I is the northeastern area of the CSHI near the mouth of the Pearl River. The Pearl River, the largest river in southern China, discharges 84.30 Mt of suspended particulate matter into the SCS each year (Cao et al., 2019). The UCC-normalized REE patterns for sediments from province I are similar to those for the Pearl River and Red River (Fig. 7a). This indicates that the northern part of the region is affected by the Pearl River, while the Red River sediments enter the region through the Qiongzhou Strait. This would also explain the high REEs concentrations in the north and the Qiongzhou Strait (Fig. 3). Province II is in the northwestern part of the Hainan Island shelf, with the Red River in the north. The spatial variation in UCC-normalized REE concentrations for province II are similar to those for the Red River and the Pearl River, but also slightly affected by sediments from Hainan Island (Fig. 7b). It also explains the higher REE concentrations in this area, which is located close to the mouth of the Red River, Qiongzhou Strait and Hainan Island (Fig. 3). The spatial variations of UCC-normalized REE concentrations in province III are different from the other two provinces, which indicates the importance of Hainan Island as a sediment source for province III (Fig. 7c). This is consistent with the distribution of REE from high to low in the south of CHSI (Fig. 3). In summary, the spatial variation of REEs in province I was determined by close proximity to the mouths of the Red and Pearl rivers, whereas the concentrations in sediments in province II and III could be attributed to a mixed contribution from three sources (Fig. 7d).