Heavy Metals in Suspended Particulate Matter of the Zhujiang River, Southwest China: Contents, Sources, and Health Risks

To investigate the abundance, water/particle interaction behavior, sources, and potential risk of heavy metals in suspended particulate matter (SPM), a total of 22 SPM samples were collected from the Zhujiang River, Southwest China, in July 2014 (wet season). Nine heavy metal(loid)s (V, Cr, Mn, Ni, Cu, Zn, As, Cd and Pb) in SPM were detected. The results show that the selected heavy metal(loid)s in SPM appear in the following order: Mn (982.4 mg kg−1) > Zn (186.8 mg kg−1) > V (143.6 mg kg−1) > Cr (129.1 mg kg−1) > As (116.8 mg kg−1) > Cu (44.1 mg kg−1) > Ni (39.9 mg kg−1) > Pb (38.1 mg kg−1) > Cd (3.8 mg kg−1). Furthermore, both the enrichment factor (EF) and geo-accumulation index (Igeo) indicate that SPM is extremely enriched in metal(loid)s of Cd and As, while SPM is slightly enriched, or not enriched, in other heavy metals. According to the toxic risk index (TRI) and hazard index (HI), arsenic accounts for the majority of the SPM toxicity (TRI = 8, 48.3 ± 10.4%) and causes the primary health risk (HI > 1), and the potential risks of V and Cr are also not negligible. By applying a correlation matrix and principal component analysis (PCA), three principal components (PC) were identified and accounted for 79.19% of the total variance. PC 1 (V, Cr, Mn, Ni, Cu, and Pb) is controlled by natural origins. PC 2 (As and Cd) is mainly contributed by anthropogenic origins in the basin. PC 3 (Zn) can be attributed to mixed sources of natural and anthropogenic origins. Moreover, all the partition coefficients (lgKd) exceeded 2.9 (arithmetical mean value order: Mn > Pb > Cd > V ≈ Cu > Cr ≈ Ni), indicating the powerful adsorptive ability of SPM for these heavy metal(loid)s during water/particle interaction.


Introduction
Heavy metals are one of the most significant pollutants in the environment, particularly in the aquatic environment, that may cause severe deterioration of water quality and do harm to living organisms due to their toxicity, persistence, non-biodegradability, and bio-accumulation [1][2][3]. Generally, heavy metals in the aquatic system can be divided into three phases: dissolved load, suspended particulate matter (SPM), and sedimentary [4][5][6][7][8]. Although the dissolved phase is believed to be more toxic to aquatic organisms and humans, its content is usually lower than the suspended particle phase [9]. Because of the high surface area and reactivity of suspended particulate matter, the dissolved heavy metals are also easily absorbed by SPM [8,10]. Therefore, more attention has been the sources of heavy metal(loid)s in SPM; and (iv) to assess the potential risk of heavy metal(loid)s in SPM, particularly to evaluate the health risk of human exposure firstly by referring to the U.S. EPA (Environmental Protection Agency) method. The results can be applied to increase prevention-control efficiency of heavy metal(loid) pollution as well as to prevent hazardous heavy metal(loid) pollution affecting the local people in the whole basin.

Study Area
The Zhujiang River (21°31′-26°49′ N, 102°14′-115°53′ E) is the largest river flowing into the South China Sea and is the major water source for the population of more than 30 million in southern China [26,27]. As the elevation decreases from northwest to southeast, the Zhujiang River flows through Yunnan, Guizhou, Guangxi and Guangdong provinces with a coverage area of 4.5 × 10 5 km 2 ( Figure 1). The Zhujiang River basin is characterized by a tropical to subtropical monsoon climate, where the annual temperature and annual precipitation range from 14 to 22 °C and 1200 to 2200 mm [25]. Various rocks, including carbonate rocks, metamorphic rocks, detrital sedimentary rocks, and magmatic rocks, are widely distributed in the Zhujiang River basin [25,26] (Figure 1). There are 24 large dams and 212 medium reservoirs located in the mid-lower reaches of the Zhujiang River [25].

Sample Collection and Analysis
Based on the lithology distribution, population distribution and reservoir/dam distribution of the Zhujiang River basin, 22 sampling sites were selected ( Figure 1). Ten sites were located at the Nanpanjiang River (NPR, M1 to M6) and Beipanjiang River (BPR, B1 to B4) in the upper reaches of the Zhujiang River with widely distributed carbonates and a small population. Twelve sites were located at Xunjiang (XUR, M7 to M13) and Xijiang (XJR, M14 to M18) in the mid-lower reaches of the Zhujiang River, where there are large populations and many reservoirs/dams with metamorphic rock and magmatic rock development. Accordingly, a total of 22 river SPM samples were collected from the selected sites during July 2014 (wet season). The SPM samples in river water were firstly filtered through millipore nitrocellulose membrane filters, and the SPMs on the filter membranes were then removed by milli-Q water and dried at 55 °C in the laboratory. The digestion method of SPMs was modified from previous studies [21,33]. Briefly, 100 mg of SPM sample powder was digested with 1 mL pure HF and 3 ml pure HNO3 in a pre-cleaned PFA (Perfluoroalkoxy) sample jar

Sample Collection and Analysis
Based on the lithology distribution, population distribution and reservoir/dam distribution of the Zhujiang River basin, 22 sampling sites were selected ( Figure 1). Ten sites were located at the Nanpanjiang River (NPR, M1 to M6) and Beipanjiang River (BPR, B1 to B4) in the upper reaches of the Zhujiang River with widely distributed carbonates and a small population. Twelve sites were located at Xunjiang (XUR, M7 to M13) and Xijiang (XJR, M14 to M18) in the mid-lower reaches of the Zhujiang River, where there are large populations and many reservoirs/dams with metamorphic rock and magmatic rock development. Accordingly, a total of 22 river SPM samples were collected from the selected sites during July 2014 (wet season). The SPM samples in river water were firstly filtered through millipore nitrocellulose membrane filters, and the SPMs on the filter membranes were then removed by milli-Q water and dried at 55 • C in the laboratory. The digestion method of SPMs was modified from previous studies [21,33]. Briefly, 100 mg of SPM sample powder was digested with 1 mL pure HF and 3 ml pure HNO 3 in a pre-cleaned PFA (Perfluoroalkoxy) sample jar (Savillex, Eden Prairie, MN, USA) at 140 • C. After the samples were completely digested, 2 mL pure HNO 3 was added twice to break up residual fluorine compounds until evaporation to dryness. Finally, the remaining digest was re-dissolved using 100 mL 3% HNO 3 for the heavy metal(loid) analyses. The heavy metal(loid)s (V, Cr, Mn, Ni, Cu, Zn, As, Cd, and Pb) of the digested solutions were determined by ICP-MS (Elan DRC-e, Perkin Elmer, Waltham, Massachusetts, USA), and the aluminum for the enrichment factor calculation was also detected by ICP-OES (Optima 5300DV, Perkin Elmer, Waltham, Massachusetts, USA). All the samples and standards were analyzed in batches with a procedural blank. Relative standard deviations (RSD) for heavy metal(loid)s were ±5%.

Enrichment Factor (EF)
The EF normalizes the content of a heavy metal(loid)s to a conservative element, and has been extensively used to assess the enrichment of heavy metal(loid)s quantitatively [20,24,33,34]. Here, Al was approved as a reference element due to its extensive distribution in continental rocks and scarcity in various pollution sources [35], and can be used to calculate the EF as follows [20,24]: where C i is the concentration of the heavy metal(loid)s (mg kg −1 ), and C ref is the concentration of the reference heavy metal(loid)s (mg kg −1 ). The (C i /C ref ) ratio is calculated based on the local soil background values. The soil background values of the Yunnan and Guizhou provinces were used for NPR (M1 to M6), and BPR (B1 to B4) river reach samples, and the mean soil background values of Guangdong and Guangxi provinces were used for the downstream samples (M7 to M18) [36]. The corresponding enrichment level categorizations of the EF value [24] are listed in Table 1. The geo-accumulation index (I geo ) is also applied to assess the heavy metal(loid) contamination in SPM. This approach has been widely used in previous studies [8,20,37]. The I geo is calculated as follows [38,39]: where C i is the concentration of heavy metal i in the SPM (mg kg −1 ), B i is the local soil background concentration of metal i (mg kg −1 ), and the coefficient 1.5 in the equation is used to minimize the effect of possible variations in the background values. The I geo for each metal is classified using seven (0-6 grades) enrichment classes [38] (Table 1).

Risk Assessment
The toxic risk index (TRI) is applied to assess the integrated toxic risk (mainly the potential ecological risk to aquatic organisms) based on both the threshold effect level (TEL) and the probable effect level (PEL) of heavy metal(loid)s. Here, we selected consensus-based TEL and PEL values [40], which have been successfully used to assess the potential ecological risks of aquatic system trace metal(loid)s in previous studies [5,24]. The TRI of the SPM is calculated by the following equation [5]: where C i S is the concentration of metal i (mg kg −1 ) in the SPM, C i TEL and C i PEL are the TEL and PEL of metal i (mg kg −1 ), respectively. The toxic risks are classified into five categories (Table 1) based on the TRI calculation [5].
The health risk of human exposure to SPM of the Zhujiang River was evaluated by referring to the U.S. EPA method [41], which considers the amount of metal(loid)s entering the body and the relationship between the undesirable health effects and reference dose. The non-carcinogenic risk is calculated and assessed by the hazard quotient (HQ) and hazard index (HI, the potential hazard to the human health). In general, direct ingestion and dermal absorption are the two main exposure pathways to heavy metal(loid)s in the aquatic system for human beings [42,43]. Since humans rarely drink water with SPM directly (direct ingestion), here we considered that dermal absorption is the only exposure pathway for heavy metal(loid)s in the SPM. The HQ is the ratio between exposure via each pathway and the reference dose (RfD). HI is the sum of the HQs for each heavy metal from all the pathways (in this study, HI is equal to HQ because there is only one pathway). If the HQ or HI exceeds 1, non-carcinogenic risk effects on human health are a concern, and further study is necessary. In contrast, there are no deleterious effects when HQ or HI is less than 1 [37,43]. The HQ and HI are calculated as follows [37,44]: where ADD dermal is the average daily doses by dermal absorption (mg kg −1 day −1 ); RfD is the reference dose (mg kg −1 day −1 ) [37,45], and the other parameters and values in these Equations (4)-(6) are shown in Table 2. Statistical approaches, including a correlation matrix and principal component analysis (PCA), were applied to analyze the dataset to obtain descriptive statistics and to explore the possible sources of the heavy metal(loid)s. PCA is the most common multivariate statistical method used to explore the associations and origins of heavy metal(loid)s [46], which could reduce the dimensionality of the dataset to several influencing factors while trying to preserve the relationships presented in the original data [43,47]. The factor contribution or variables with minor significance attained from PCA are further reduced by the varimax rotation method [43]. The results of PCA, including the factor loadings, eigenvalues, variance, and communalities, constitute the component matrix. The result of PCA is acceptable if the communalities value is close to 1. The factor loadings (the correlation coefficients between each principal component and initial variable) are classified as "strong", "moderate", and "weak" according to the absolute loading values of >0.75, 0.75-0.50, and 0.50-0.30, respectively [48]. In this study, PCA is performed for heavy metal(loid)s of SPM in the Zhujiang River to distinguish their possible origins. The suitability of the dataset for PCA was checked using the Kaiser-Meyer-Olkin (KMO) test and Bartlett's sphericity test (p < 0.001) [47]. To avoid the numerical ranges of the original variables, the dataset was first standardized by a z-scale transformation.

Abundance of Heavy Metal(loid)s in SPM
The concentrations of heavy metal(loid)s in SPM of the Zhujiang River are shown in Table 3. The Kolmogorov-Smirnov (K-S) test, which is a non-parametric test, was used to test the normal distribution of our data. The test results show that all parameters are normally distributed in the Zhujiang River (K-S test significance >0.1), and the arithmetical mean values of all parameters are suitable for comparison [43]. Therefore, the nine selected heavy metal(loid)s in this study can be ranked by abundance as follows: . Mn and Zn are the most abundant metals, with maximums of 1487.1 and 732.8 mg kg −1 , respectively, compared to the soil background values of the Zhujiang River basin [36]. The concentrations of five metal(loid)s, including Cr, Mn, Zn, As, and Cd, in SPM are much higher than all soil background values, while the contents of the remaining metals are between the soil background values of several provinces. Cd concentration is 5.8-23.7 times higher than the soil background concentration values of the whole basin which can be considered the strongest enriched metal in SPM relative to the soil. Cr, Mn, Zn, and As concentrations are elevated 1.2-7.9 times the soil background concentration values. On a global scale (Table 4), V, Cr, and Zn are generally close to the world average, Mn, Ni, Cu, and Pb are lower than the world average, while As and Cd are much higher than the world average [11]. Compared with the rivers in Asia (China), the contents of V, Cr, and Mn in SPM of the Zhujiang River are similar, Ni, Cu, and Pb are slightly lower, while Zn is slightly higher. Additionally, the metals (Cr, Ni, Cu, Zn, and Pb) easily affected by human activities in SPM of the Zhujiang River are much lower than those in Europe (with many developed countries), which partly reflects the impact of economic development on heavy metal pollution in the fluvial environment.  Note: The data for global rivers are from Viers et al. [11]; -, no data.

Water/Particle Interaction
The partition coefficient (K d ) is the ratio of the element content in solid form (SPM in this study) to dissolved content in water (ppm/ppm) [21], which provides empirical information about the water/particle interaction for trace metals [8,49] and is usually expressed as lgK d . A high lgK d value signifies a powerful affinity of the metals with SPM [15]. In combination with the dissolved heavy metal concentration in the same water samples of Zhujiang River reported in our early work [32], the lgK d values of the seven metals are calculated and summarized in Table 5. The lgK d values of V, Cr, Mn, Ni, Cu, Cd, and Pb ranged from 3.6 to 5.0, 3.3 to 4.5, 4.7 to 7.0, 3.7 to 4.5, 2.9 to 5.3, 4.6 to 5.5, and 5.4 to 6.2, respectively. All the lgK d values exceeded 2.9, indicating the powerful adsorptive ability of heavy metals for the SPM. The mean partition coefficients of seven metals decreased in the order of Mn > Pb > Cd > V ≈ Cu > Cr ≈ Ni (Table 5); mainly controlled by the ionic radius and particle reactivity of these metals and the particle size of the SPM [8,23,49]. Compared to some rivers in the world, the lgK d values of seven metals are within the range of world river values [6,8,[50][51][52][53] (Table 5). The partition coefficients of Cr, Cu, and Cd are comparable to some rivers in China [52], particularly the Beijiang River [8], a significant tributary of the lower reaches of the Zhujiang River. However, the lgK d values of Mn, Ni, and Pb are relatively higher than those of rivers in China [8,52]. It is noteworthy that all the mean lgK d values (except Pb) in the present study are lower than the monthly mean values of the upper Zhujiang River [21], which indicates the possible seasonal variations in water/particle interaction.

Enrichment Factor
The abundance of heavy metal(loid)s in SPM is normalized by the corresponding soil background values [36] in this study ( Figure 2). Most metal(loid)s had a soil-normalized value which approached one and ranged from 0.1 to 4.1, with the exception of Zn, As and Cd. Soil-normalized values of As and Cd were 1.7 to 15.9 and 3.3 to 39.7, respectively, and indicate that all the SPM samples are enriched in metal(loid)s of As and Cd. Zn shows the soil-normalized value of varying degrees (0.8 to 7.4), which is more obvious in the headstream reach (M1 to M6, B1 to B4) and the XJR reach (M14 to M18). Note: Min, minimum; Max, maximum; AM, arithmetical mean; -, no data; Rivers in US [53]; Tigris River [6]; Day River [51]; Sava River [50]; Yangtze River and Jialingjiang River [52]; Beijiang River [8]; Upper Zhujiang River [21].

Enrichment Factor
The abundance of heavy metal(loid)s in SPM is normalized by the corresponding soil background values [36] in this study ( Figure 2). Most metal(loid)s had a soil-normalized value which approached one and ranged from 0.1 to 4.1, with the exception of Zn, As and Cd. Soil-normalized values of As and Cd were 1.7 to 15.9 and 3.3 to 39.7, respectively, and indicate that all the SPM samples are enriched in metal(loid)s of As and Cd. Zn shows the soil-normalized value of varying degrees (0.8 to 7.4), which is more obvious in the headstream reach (M1 to M6, B1 to B4) and the XJR reach (M14 to M18). In order to quantitively evaluate the enrichment degree of heavy metal(loid)s in the Zhujiang River SPM, the enrichment factor (EF) was applied in the present study. As shown in Figure 3 Table 1), and a few sampling sites exceed 50 (M6, M8, and B1), which can be defined as extremely severe enrichment ( Table 1). The EF values of As mainly range from 5 to 10, which is a moderately severe enrichment. Cr, Mn, Ni, Cu, and Zn are slightly enriched, with mean EF values between 1.4 and 3.2, while the remaining metals (V and Pb) show no enrichment characteristics in most of the sites (EF < 1). It should be noted that the EF values of V (6.2), Cr (3.3), Cu (5.0), and As (79.8) are highest at B1, and the rest of the metals also have higher EF values, which illustrates that site B1 is the most strongly related to human activities [24]. Compared with the monthly SPM sampling of BPR (the mean EF values are 2.8, 3.1, 1.9, 2.7, 1.8, 2.4, 11.9, and 2.0 for V, Cr, Mn, Ni, Cu, Zn, Cd, and Pb, respectively) [21], most of the metals in our study have a relatively lower EF value, indicating that although the lgKd values in this study (wet season) reflect the powerful adsorption capacity of SPM for heavy metals, there may be stronger water/particle interaction at the monthly scale, particularly particle adsorption. Furthermore, compared with the mean EF values of 11.0, 12.5, In order to quantitively evaluate the enrichment degree of heavy metal(loid)s in the Zhujiang River SPM, the enrichment factor (EF) was applied in the present study. As shown in Figure 3 As. In the current study, the EF values of Cd in most sampling sites exceed 10 (severe enrichment, Table 1), and a few sampling sites exceed 50 (M6, M8, and B1), which can be defined as extremely severe enrichment ( Table 1). The EF values of As mainly range from 5 to 10, which is a moderately severe enrichment. Cr, Mn, Ni, Cu, and Zn are slightly enriched, with mean EF values between 1.4 and 3.2, while the remaining metals (V and Pb) show no enrichment characteristics in most of the sites (EF < 1). It should be noted that the EF values of V (6.2), Cr (3.3), Cu (5.0), and As (79.8) are highest at B1, and the rest of the metals also have higher EF values, which illustrates that site B1 is the most strongly related to human activities [24]. Compared with the monthly SPM sampling of BPR (the mean EF values are 2.8, 3.1, 1.9, 2.7, 1.8, 2.4, 11.9, and 2.0 for V, Cr, Mn, Ni, Cu, Zn, Cd, and Pb, respectively) [21], most of the metals in our study have a relatively lower EF value, indicating that although the lgK d values in this study (wet season) reflect the powerful adsorption capacity of SPM for heavy metals, there may be stronger water/particle interaction at the monthly scale, particularly particle adsorption. Furthermore, compared with the mean EF values of 11.0, 12.5, 10.0, 5.0, 19.6, and 19.6 for Cr, Ni, Cu, Zn, Cd, and Pb, respectively, in the polluted river (Soan River, Pakistan) [20], the enrichment degree of heavy metals in the Zhujiang River SPM is relatively slight.

Geo-Accumulation Index
Based on the local soil background values (Table 3), the contamination degrees of heavy metal(loid)s in SPM of the Zhujiang River are assessed by the geo-accumulation index method (Equation (2)). The mean value of Igeo shows a contamination level order similar to EF (Cd > As > Zn > Mn > Cr > Cu ≈ Ni > V ≈ Pb, Figure 4). The most contaminated heavy metal(loid)s are Cd and As, with mean Igeo values of 3.4 and 2.1, respectively (Figure 4), revealing heavily polluted and moderately to heavily polluted levels. The mean value of Igeo for Zn (0.5), Mn (0.3), and Cr (0.1) classifies these metals as lightly polluted. The remaining metals (Cu, Ni, V, and Pb) have mean Igeo values of less than 0, indicating the unpolluted level ( Figure 4). The mean Igeo values of the present study are consistently lower than those of the Beijiang River, an important tributary of the lower Zhujiang River, with several polymetallic mines and metal smelting enterprises (the mean values of Igeo are 2.1, 2.7, 3.1, 7.0, and 1.5 for Cu, Zn, As, Cd, and Pb, respectively) [8], revealing that the pollution intensity of heavy metal(loid)s in SPM is assuaged by the varying landscape setting of the whole Zhujiang River basin. This could be further confirmed by the comparison with polluted rivers [20].

Geo-Accumulation Index
Based on the local soil background values (Table 3), the contamination degrees of heavy metal(loid)s in SPM of the Zhujiang River are assessed by the geo-accumulation index method (Equation (2)). The mean value of I geo shows a contamination level order similar to EF (Cd > As > Zn > Mn > Cr > Cu ≈ Ni > V ≈ Pb, Figure 4). The most contaminated heavy metal(loid)s are Cd and As, with mean I geo values of 3.4 and 2.1, respectively (Figure 4), revealing heavily polluted and moderately to heavily polluted levels. The mean value of I geo for Zn (0.5), Mn (0.3), and Cr (0.1) classifies these metals as lightly polluted. The remaining metals (Cu, Ni, V, and Pb) have mean I geo values of less than 0, indicating the unpolluted level ( Figure 4). The mean I geo values of the present study are consistently lower than those of the Beijiang River, an important tributary of the lower Zhujiang River, with several polymetallic mines and metal smelting enterprises (the mean values of I geo are 2.1, 2.7, 3.1, 7.0, and 1.5 for Cu, Zn, As, Cd, and Pb, respectively) [8], revealing that the pollution intensity of heavy metal(loid)s in SPM is assuaged by the varying landscape setting of the whole Zhujiang River basin. This could be further confirmed by the comparison with polluted rivers [20].

Geo-Accumulation Index
Based on the local soil background values (Table 3), the contamination degrees of heavy metal(loid)s in SPM of the Zhujiang River are assessed by the geo-accumulation index method (Equation (2)). The mean value of Igeo shows a contamination level order similar to EF (Cd > As > Zn > Mn > Cr > Cu ≈ Ni > V ≈ Pb, Figure 4). The most contaminated heavy metal(loid)s are Cd and As, with mean Igeo values of 3.4 and 2.1, respectively (Figure 4), revealing heavily polluted and moderately to heavily polluted levels. The mean value of Igeo for Zn (0.5), Mn (0.3), and Cr (0.1) classifies these metals as lightly polluted. The remaining metals (Cu, Ni, V, and Pb) have mean Igeo values of less than 0, indicating the unpolluted level ( Figure 4). The mean Igeo values of the present study are consistently lower than those of the Beijiang River, an important tributary of the lower Zhujiang River, with several polymetallic mines and metal smelting enterprises (the mean values of Igeo are 2.1, 2.7, 3.1, 7.0, and 1.5 for Cu, Zn, As, Cd, and Pb, respectively) [8], revealing that the pollution intensity of heavy metal(loid)s in SPM is assuaged by the varying landscape setting of the whole Zhujiang River basin. This could be further confirmed by the comparison with polluted rivers [20].

Correlation Analysis
A Pearson correlation matrix was employed to distinguish correlations between the nine heavy metal(loid)s of the SPM in the Zhujiang River (Table S1). The heavy metals with high correlation coefficients in the aquatic system could have similar sources, migration processes and chemical behavior [43,54]. In the current study, Cr, Mn, Ni, Cu, and Pb are remarkably positively correlated with each other (p < 0.01), indicating that these metals may be derived from the same source. Strong positive correlations are also observed between As and Cd (0.780), but these are poorly correlated with the remaining metals, suggesting that the sources of As and Cd are different from those metals. V is only significantly correlated with Cr (0.741), while Zn is not correlated with any metal (Table S1).

Principal Component Analysis
In this study, PCA with the varimax rotation method was performed for heavy metal(loid)s of SPM in the Zhujiang River. There are three principal components (PC, eigenvalues >1) that are extracted and summarized in Table 6. PC 1 explains 44.51% of the total variance and predominantly includes V, Cr, Mn, Ni, Cu, and Pb. PC 2 explains 22.36% of the total variance with significant loadings of As and Cd. PC 3 explains 12.33% of the variance which is only contributed by Zn, and most of the heavy metal(loid)s exhibit a strong loading in their PCs (loading values >0.75) [48,55]. In total, these three PCs account for 79.19% of the total variance and are presented in a three-dimensional space, as shown in Figure 5. For PC 1, V is from lithophile elements [56], and Mn, Ni, and Cr are from natural sources of rock weathering and subsequent pedogenesis [24,57]. Although urban and industrial activities such as mining, metal smelting, and automobile exhausts can be the primary source of Cu and Pb [58], the lower EF values of Cu (1.6) and Pb (0.9) (Figure 3) indicate that the contribution of anthropogenic sources is limited [7,20]; hence, we attribute PC 1 to the natural origins controlled by geology and lithology. There are two metal(loid)s (As and Cd) with positive loadings on PC 2, and the correlation analysis suggests that the sources of As and Cd are different from those metals in PC 1. Considering the extremely high EF values of As (11.0) and Cd (23.3), we conclude that PC 2 is mainly contributed by anthropogenic origins in the basin [20,59]. In addition, Zn is the sole contributor to PC 3 and is not correlated with any metal (Table S1). In combination with the moderate enrichment of Zn (EF = 3.2), PC 3 can be attributed to mixed sources of geologic and anthropogenic origins. Note: Extraction method, principal component analysis; Rotation method, Varimax with Kaiser normalization; the "bold" values mean the factor loadings (the correlation coefficients between PC and initial variable) are "strong" or "moderate".

Toxic Risk Index (TRI)
According to MacDonald [40], when the negative effects are less than 10% within the minimal effect range, the TEL is considered reliable, while the PEL is considered reliable if the negative effects exceed 65% of the probable effect range [5,40]. Thus, the TRI integrating the TEL and PEL, does not consider only the acute toxicity but also the lasting chronic toxic effects of heavy metals [24]. Based on the consensus TEL and PEL values [40] in (Table 3) and Equation (3), the TRI of seven metal(loid)s were calculated to evaluate the total toxic risk of both the acute and chronic toxic effects of SPM heavy metal(loid)s; V and Mn were excluded from the TRI calculations due to the lack of TEL and PEL values. As shown in Figure 6, the TRI values of the 22 sites range from 9.5 (M6) to 32.9 (B1), with a mean value of 17.9, indicating considerable toxic risk for most of the sites (15 < TRI ≤ 20). Additionally, three sites (M7, M16, and B1, TRI > 20) present very high toxic risk, while low toxic risk is observed at M6 (5 < TRI ≤ 10) ( Figure 6). In contrast to the EF and Igeo values, the mean TRI of individual metal(loid)s follow a decreasing order of As (8.  13.0 ± 5.5%, 7.7 ± 3.0%, 6.3 ± 4.5%, 5.8 ± 3.0%, and 3.3 ± 2.1%, respectively, to the TRI, indicating that As accounts for the majority of the overall SPM toxicity. The considerable contributions of As and Cd to the TRI are attributed mainly to their relatively low TEL and high concentration in SPM. This highlights the potential toxicity of SPM in the Zhujiang River, with two metal(loid)s (As and Cd) deserving more concern.

Toxic Risk Index (TRI)
According to MacDonald [40], when the negative effects are less than 10% within the minimal effect range, the TEL is considered reliable, while the PEL is considered reliable if the negative effects exceed 65% of the probable effect range [5,40]. Thus, the TRI integrating the TEL and PEL, does not consider only the acute toxicity but also the lasting chronic toxic effects of heavy metals [24]. Based on the consensus TEL and PEL values [40] in (Table 3) and Equation (3), the TRI of seven metal(loid)s were calculated to evaluate the total toxic risk of both the acute and chronic toxic effects of SPM heavy metal(loid)s; V and Mn were excluded from the TRI calculations due to the lack of TEL and PEL values. As shown in Figure 6, the TRI values of the 22 sites range from 9.5 (M6) to 32.9 (B1), with a mean value of 17.9, indicating considerable toxic risk for most of the sites (15 < TRI ≤ 20). Additionally, three sites (M7, M16, and B1, TRI > 20) present very high toxic risk, while low toxic risk is observed at M6 (5 < TRI ≤ 10) ( Figure 6). In contrast to the EF and I geo values, the mean TRI of individual metal(loid)s follow a decreasing order of As (8.  13.0 ± 5.5%, 7.7 ± 3.0%, 6.3 ± 4.5%, 5.8 ± 3.0%, and 3.3 ± 2.1%, respectively, to the TRI, indicating that As accounts for the majority of the overall SPM toxicity. The considerable contributions of As and Cd to the TRI are attributed mainly to their relatively low TEL and high concentration in SPM. This highlights the potential toxicity of SPM in the Zhujiang River, with two metal(loid)s (As and Cd) deserving more concern.

Health Risk Assessment
To better assess the health risk of human exposure to SPM of the Zhujiang River, the hazard index (HI) for the selected heavy metal(loid)s is calculated based on the reference dose (RfD) of each metal [37,45,60] (Table S2). The mean HI values are shown in Figure 7, and the HI calculated results

Health Risk Assessment
To better assess the health risk of human exposure to SPM of the Zhujiang River, the hazard index (HI) for the selected heavy metal(loid)s is calculated based on the reference dose (RfD) of each metal [37,45,60] (Table S2). The mean HI values are shown in Figure 7, and the HI calculated results for each site are summarized in Table S2. It should be noted that mean HI values of As exceed 1 for both children (3.3) and adults (2.4), indicating that non-carcinogenic effects may occur. For both adults and children, the HI for all the metals (except As) are less than 1 (Figure 7, Table S2), indicating that for these metals, little hazard is presented through the only exposure pathway-dermal absorption-in the whole basin area. In general, children have a higher HI value than adults (Figure 7), indicating that children face greater serious health risks due to SPM heavy metals than adults. Additionally, the previous studies concluded that negative health effects may occur for HI values >0.1 in the child cohort [37,61]. Consequently, the V and Cr (with mean HI values of 0.24 and 0.25 for children, Table S2) exposure to the SPM is non-negligible in this study. Considering species-specific toxicity, arsenic (As) mainly afflicts the mucous membrane by directly damaging the capillaries [37,62]; chromium (Cr) can result in asphyxia via reducing oxygen demand of the biochemical process [63]; and vanadium (V) exhibits hepatotoxic, nephrotoxic properties and reproductive system toxicity [64]. Here, we conclude that As is the primary health risk and more attention should also be paid to V and Cr in the Zhujiang River.

Health Risk Assessment
To better assess the health risk of human exposure to SPM of the Zhujiang River, the hazard index (HI) for the selected heavy metal(loid)s is calculated based on the reference dose (RfD) of each metal [37,45,60] (Table S2). The mean HI values are shown in Figure 7, and the HI calculated results for each site are summarized in Table S2. It should be noted that mean HI values of As exceed 1 for both children (3.3) and adults (2.4), indicating that non-carcinogenic effects may occur. For both adults and children, the HI for all the metals (except As) are less than 1 (Figure 7, Table S2), indicating that for these metals, little hazard is presented through the only exposure pathway-dermal absorption-in the whole basin area. In general, children have a higher HI value than adults (Figure 7), indicating that children face greater serious health risks due to SPM heavy metals than adults. Additionally, the previous studies concluded that negative health effects may occur for HI values >0.1 in the child cohort [37,61]. Consequently, the V and Cr (with mean HI values of 0.24 and 0.25 for children, Table S2) exposure to the SPM is non-negligible in this study. Considering species-specific toxicity, arsenic (As) mainly afflicts the mucous membrane by directly damaging the capillaries [37,62]; chromium (Cr) can result in asphyxia via reducing oxygen demand of the biochemical process [63]; and vanadium (V) exhibits hepatotoxic, nephrotoxic properties and reproductive system toxicity [64]. Here, we conclude that As is the primary health risk and more attention should also be paid to V and Cr in the Zhujiang River.

Heavy Metal Export Budget Estimation
Based on the concentrations of the heavy metals in SPM and the discharge of the wet season (April to September) at the last site (M18) of the Zhujiang River (River and Sediment Bulletin of China, http://www.mwr.gov.cn/sj/tjgb/zgstbcgb/), river fluxes of each heavy metal in SPM are estimated that range from 38.6 (Cd) to 16,171 (Mn) tons (Table 7). Here, we only calculate the budget of the wet season, and the results may be overestimated due to sampling only once. However, considering that we do not have any samples in the dry season, the overestimated part could approximately equal the export flux of the dry season. Therefore, our results can represent the annual export budget of SPM heavy metal to a certain extent. In combination with the data for dissolved heavy metals [32], the total export budget of each heavy metal was evaluated and decreased in the order of Mn > V > Cr > Ni > Cu > Pb > Cd (Table 7). To eliminate the large uncertainty in evaluation, high-frequency samplings and observations are needed to quantify the annual heavy metal budget, especially in the wet flow season, when the heavy metal concentrations could vary significantly after a storm event. Table 7. Export fluxes of heavy metals (t yr −1 ) and proportions (%) of SPM and the dissolved flux to the total flux in the Zhujiang River.

Conclusions
In conclusion, this study indicates that systematic analyses of data on nine heavy metal(loid)s in SPM samples of the Zhujiang River using multi-indicators/statistical techniques-including partition coefficient, enrichment factor (EF), geo-accumulation index (I geo ), toxic risk index, hazard index, correlation analysis and principal component analysis-can provide important support regarding the prevention-control of heavy metal pollution, and health risk control in the whole basin. Our results show that the SPM samples contained high concentrations of several heavy metal(loid)s, including Cr, Mn, Zn, As, and Cd (higher than all soil background values), and the investigated heavy metal(loid)s are powerfully adsorbed by the SPM during water/particle interaction. In particular, the enrichments of As and Cd are noticeable in the SPM, with the consistently high EF and I geo values. Anthropogenic emissions are the main source of the SPM extremely enriched elements (As and Cd), while natural origins are the source responsible for V, Cr, Mn, Ni, Cu, and Pb, and the sources of the remaining heavy metals are controlled by mixed anthropogenic and geologic origins. Moreover, our systematic risk assessment concluded that As could pose potential non-carcinogenic effects on human health and accounted for the majority of the SPM toxicity in the entire catchment. The potential risks of V and Cr with their relatively higher hazard index, is also not negligible. In order to incorporate the possible uncertainty of the single sampling and the variations of geochemical fractions of heavy metal(loid)s in SPM, and to estimate the potential risk clearly, there is a need for further research including high-frequency sampling and heavy metal(loid)s speciation analysis that would help understand the geochemical cycle of heavy metal(loid)s and its environmental effect in the Zhujiang River basin.
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/16/10/1843/s1, Table S1: Pearson correlation matrix of heavy metal(loid)s of the SPM in the Zhujiang River. Table S2: Hazard index (HI) calculated results for each site, and reference dose for heavy metal(loid)s in the Zhujiang River.