Introduction

Geochemical studies of sediments in rivers and estuaries have been extended in the last few decades due to the growing awareness of riverine and estuarine pollution and its effects on the ecosystem (Varol 2011; Attia and Ghrefat 2013; Ho et al. 2013). Most elements entering the aquatic environment are ultimately fixed to sediments. However, these trapped elements may be recycled back to water column where they pose a threat to aquatic ecosystems via the variation of the physicochemical characteristics of the overlying water and sediment (Kalnejais et al. 2010). Therefore, sediment cores provide an archive of historical input and output of various elements within both the river and estuary ecosystems. The collected geochemical information of various elements not only provides an insight into the past and present environmental quality of the system but also serves as a baseline for future investigations (Badr et al. 2009). In recent years, many studies have achieved success in using sediment cores to establish the effects of anthropogenic and natural processes on depositional environments in different water areas (Lavilla et al. 2006; Chatterjee et al. 2007; Harikumar and Nasir 2010; Tue et al. 2012).

Being the second largest river system in southern China and a major waterway, the Pearl River Estuary (PRE) is known as one of the most industrialized and urbanized regions in China and has been, and remains, subject to significant anthropogenic influences, including vast industrial, agricultural, and municipal activities around the estuary and along the river (Pan and Wang 2012). The PRE has been a hot spot for research in recent decades (Li et al. 2000; Liu et al. 2003; Ip et al. 2007; Zhou et al. 2009; Yu et al. 2010, 2012). Because of the heavy metal toxicity to aquatic and benthonic organisms and the subsequent effects on humans, the geochemical behavior and pollution assessment of heavy metals account for a large fraction of these studies. Experiments have been demonstrated that ascending salinity promotes Cd desorption from sediments and hence increases total Cd concentrations in the water column and Cd uptake by organisms (Greger et al. 1995; Du Laing et al. 2009). However, the geochemistry of major and trace elements in sediment cores combining the upstream and downstream Pearl River and estuary with a salinity gradient is unclear. Areas around rivers and estuary environments represent zones of interaction and transition between the two systems where geochemical and sedimentological processes are highly variable both spatially and temporally. These processes determine the destination of various elements in these aquatic environments. Therefore, a comparison study of the geochemistry of various elements in sediments from the river and the adjacent estuary is important not only to understand their natural and anthropogenic origin but also to evaluate their pollution status.

The objectives of this study are the following: (1) to determine locational and vertical variations of major and trace elements along the estuary of the Pearl River, (2) to identify the source of elements based on cluster analysis and principal component analysis, and (3) to evaluate trace element contamination using enrichment factors.

Materials and methods

Study area and sampling procedures

The Pearl River is the 13th largest river in the world in terms of freshwater discharge with an annual average of 3.26 × 1011 m3 year−1. At the end of the water course, the river divides into three main tributaries, namely East River, West River, and North River (Fig. 1). The Lingdingyang estuary, also known as the PRE, receives approximately 53 % of the freshwater flow from four outlets: Humen, Jiaomen, Hongqilimen, and Hengmen (He et al. 2010). A main stream of the Pearl River, named as Major Pearl River, and its estuary were chosen for sampling (Fig. 1). The Major Pearl River has a total length of 128 km and is divided into two waterways, the Guangzhou channel (the upper reach), with a length of 78 km, an average width of 431 m, and an average water depth of 4.78 m, and the Shiziyang channel (the lower reach), which is 50 km long, about 2,200 m wide, and 6.64 m in water depth (Mai et al. 2001). The Shiziyang channel receives inflows from the Guangzhou channel and the East River and runs across an area, Panyu and Dongguan, where agriculture and industry are well developed, draining into the Lingdingyang via the Humen outlet (Fig. 1).

Fig. 1
figure 1

Map showing the location of the study area and sediment-core collection sites on the Pearl River and estuary

Full size image

Site 1 is located in Huangpu District and is the locus where the Pearl River flows out to the main urban area of Guangzhou (Fig. 1). Site 2 is located adjacent to Nizhou Village and near the East River south tributary. Site 3 is located at the Humen outlet and is the locus where the Pearl River flows into the estuary. Site 4, lying in the middle shoal of the PRE and near Neilingding Island, represents a moderate pollution location influenced by the four eastern discharge outlets. Four sediment cores from sites 1, 2, 3, and 4 were collected in 2007 (Fig. 1). Sediment cores 1 (22 cm in length), 2 (48 cm in length), and 3 (39 cm in length) were sliced every 3 cm from the top to the bottom. Sediment core 4 (55 cm in length) was sliced every 2 cm from the top to 40 cm depth, and then only two 2-cm subsamples were sliced namely 46–47- and 53–54-cm subsamples. After slicing, the subsamples were stored in sealed polyethylene bags and frozen at −20 °C until analysis.

Sample preparation, analyses, and quality control

The temperature, salinity, pH, dissolved oxygen (DO), and turbidity of the overlying water were determined in situ using the YSI 6600 V2 Songde water quality monitoring system (YSI Incorporated, USA) (Table 1).

Table 1 Sampling locations, water depth, and the environmental water parameters
Full size table

Sediment samples were freeze-dried and ground to <74 μm for the subsequent analysis. The loss on ignition (LOI) following combustion at 550 °C for 4 h was used as a proxy for organic carbon (OC) content in the sediments. Powder X-ray diffraction (XRD) was performed using an automatic Bruker D8 ADVANCE, with a Cu target tube. Diffractograms were recorded at 0.01° of the 2θ step, with an acquisition time of 3° min−1 at 40 kV and 40 mA.

A microwave-assisted digestion method was used to achieve total dissolution of the sediment. The following two-stage digestion procedure was employed. Approximately 150 mg of the sediment was accurately weighed in microwave vessels, and then 6 ml HNO3 and 1.5 ml HF were sequentially added. The samples remained in contact with the acid solution for 30 min. Then, the vessels were sealed, placed in the microwave oven, and heated at rated output power (900 W) for 15 min. After cooling the vessels, 8 ml of a saturated boric acid solution was added in the second stage of the digestion. The vessels were again placed in the microwave oven and heated for 10 min at 900 W. The obtained clear solutions were poured into 120-ml HDPE bottles and diluted to 100 g. For each set of measures, digested blanks were prepared in the same manner.

A certified reference standard sediment (MESS-3) was used to test the analytical and instrument accuracy of the method. The results indicated good agreement between certified and analytical values (Table 2). Moreover, reagent blank and replicate samples were randomly inserted into the analysis procedure. The relative standard deviations of all replicate samples were less than 10 %.

Table 2 Reference and determined element concentrations in the analyzed marine sediment reference materials (MESS-3)
Full size table

Enrichment factor

Enrichment factor (EF) can be used to discriminate if the observed metal concentration in marine sediments is influenced by anthropogenic sources (Celis-Hernandez et al. 2013). The EF was calculated using the following equation:

$$ EF={\left( C/ B\right)}_{sample}/{\left( C/ B\right)}_{background}, $$
(1)

where (C/B)sample is the ratio of the examined element concentration to the reference element concentration for the sample and (C/B)background is the respective baseline ratio. According to Sutherland (Sutherland 2000), EF < 2 indicates no or minimal pollution, 2 ≤ EF < 5 indicates moderate pollution, 5 ≤ EF < 20 indicates significant pollution, 20 ≤ EF < 40 indicates very strong pollution, and EF ≥ 40 indicates extreme pollution.

Statistical analysis

Multivariate statistical analysis can help simplify and organize large datasets to provide meaningful insight (Chen et al. 2007). Cluster analysis (CA), which can reveal specific linkages between sampling sites because it provides an indication of similarities or dissimilarities between their trace metal contaminations (Simeonov et al. 2000), was performed on the normalized dataset (z-scores) using Ward’s method with squared Euclidean distances. Principal component analysis (PCA) is a statistical technique used to verify and quantify interrelationships among the original variables in a dataset. This procedure reduces large datasets to summary variables which account for a majority of the variability in the original dataset. All statistical data processing in this work were carried out using SPSS software (version 15.0 for Windows).

Results and discussion

Geochemistry of major elements in sediments

Major element concentrations in the investigated cores varied within the range 6.75–13.6 % for Al and 3.00–6.12, 0.35–0.76, and 0.043–0.26 % for Fe, Ti, and P, respectively (Table 3). Al and Fe had slightly elevated values compared with those of Pearl River system sediments and the average upper crust. Moreover, Al and Fe concentrations were higher at site 1 than at the other sites.

Table 3 Summary of major elements in the four sediment cores from the Pearl River and estuary
Full size table

The mean values of Ca were 0.63, 0.24, 0.59, and 0.35 % at sites 1, 2, 3, and 4, respectively, far less than the average upper crust (3.00 %). This indicates that Ca is strongly depleted when compared with the average upper crust and other study areas (Table 3). Moreover, Ca was lower at site 2 than at the other sites. The Pearl River consists of three main rivers: the West River, the East River, and the North River (Fig. 1). The average concentrations of Ca were 1.12, 0.76, and 0.43 % in the West River, North River, and East River sediments, respectively (Zhang and Wang 2001). The West River basin is dominated by limestone, whereas the East River drains the South China granite area (Zhang and Wang 2001). Fe, Ca, and Mg were relatively depleted, and Al was relatively enriched in the East River sediments compared to those in the West River sediments. Site 2 is primarily controlled by the East River tributary. Moreover, site 3 is influenced by all the three River tributaries, which potentially resulted in the lowest Ca concentration at site 2. In comparison, the Ca vertical profile has large fluctuations; the highest value (1.13 %) was found at site 3 (Fig. 2), suggesting that some Ca in the core sediments was of biogenic origin, apart from the detrital contribution.

Fig. 2
figure 2

Vertical profiles of major elements in the sediments from the Pearl River and estuary

Full size image

Na concentrations increased from upstream (0.31–0.47, 0.44–0.75, and 0.68–1.01 % at sites 1, 2, and 3, respectively) to the estuary (0.90–1.35 % at site 4), which were consistent with the salinity results of the overlying water (Table 1). The river discharge in the Pearl River Delta is greatly affected by the monsoonal climate and characterized by distinct changes between dry (from November to March) and wet (from May to October) seasons, with approximately 80 % water discharging in the wet season (Yin et al. 2004; Gong et al. 2013). Salt intrusion in the wet season is largely suppressed by a large amount of river inflow, while the 5 ppt isohaline can extend 80 km landward from the estuary’s mouth during the dry season (Gong et al. 2013). Mg and K exhibited relatively constant values in the four cores. Na and Mg were more depleted when compared with the average upper crust. K concentrations were comparable to those in Ria de Vigo sediments (2.40–3.08 %) and the average upper crust (Table 3).

Mineralogical composition in sediments

The XRD spectra are recorded for the selected subsamples to determine the mineralogical composition. Semiquantitative calculations of the mineral percentages are made by peak surface measurements after background substraction. It should be noted that the percentages of kaolinite are actually the mixture of kaolinite and chlorite because the further methods, such as thermal and acid procedures or treatment with dimethyl-sulphoxide, are not conducted in this study (Gonzalez-Garcia and Sanchez-Camazano 1968). As shown in Fig. 3 and Table 4, the minerals of subsamples were dominated by quartz (25.6–50.4 %) with an average of 33.0 %, followed by kaolinite (15.2–34.3 %) and illite (11.7–25.5 %) with averages of 24.9 and 17.5 %, respectively. Montmorillonite and feldspars were observed in most subsamples with averages of 14.8 and 8.5 %, while small amounts of calcite, halite, and dolomite were only found in a few subsamples. Moreover, heavy minerals such as pyrite, goethite, and hematite cannot be identified in the four cores, which is probably due to the low level of these minerals (Seddique et al. 2011).

Fig. 3
figure 3

X-ray diffraction patterns showing the main minerals found in selected bulk samples. Sample numbers (4–38): the 4 and 38 represent the sampling sites and the sample depth, respectively

Full size image
Table 4 Semiquantitative XRD mineralogy of selected bulk samples in four cores from the Pearl River and estuary
Full size table

The estuarine subsamples contained more quartz and less illite compared with the subsamples from the Pearl River. Meanwhile, the subsamples at site 1 and subsamples 2–39 possessed high contents of clay minerals (illite, montmorillonite, and kaolinite) but low contents of quartz. It might be the reason that the concentrations of Al, Fe, Ti, Cr, Ni, and V were higher than other elements in these subsamples. Previous study has clearly indicated that the sediment-associated clay minerals do play a significant role in the adsorption of heavy metals, although this role was less significant than Fe/Mn oxides and organic materials (Wand and Li 2011). Feldspars appeared in varying percentages in all studied subsamples, and the maximum was found at site 2. As discussed above, the East River drains the south China granite area, and this rock consists mainly of feldspar and quartz, which may contribute to the high feldspars in sediments from site 2. Halite was identified only at site 4, which was contributed to the high salinity in the estuary. The maximum Ca (1.13 %) was found in subsamples 3–9, the only subsample containing calcite. The largest sum of quartz and feldspar (65.9 %) was found in subsamples 2–15, which may act to dilute the contents of other elements (e.g., Fe, Al, Na, Cr, and Zn), leading to very low concentrations of these elements (Figs. 2 and 4). In contrast, K had a higher concentration in this subsample, which was much different from other elements primarily because it is the main constituent of K-feldspar.

Fig. 4
figure 4

Vertical profiles of trace elements and Mn in the sediments from the Pearl River and estuary

Full size image

Geochemistry of trace elements and Mn in sediments

Mn was the most abundant metal at sites 2, 3, and 4, while Zn had the highest concentration at site 1 (Table 5). Similar to Ca, Mn showed minimum content in the surface layer of core 2 and maximum concentration in core 3, which could be due to the distribution of Mn in the East River and West River, respectively. Other research also has shown that Mn concentrations are highest in the area between Zhuhai and the Hongqimen Gate, which is mainly controlled by the West River tributary (Woods et al. 2012). The average trace element concentrations were in the following sequential order at site 1: Zn > Mn > Ba > Cu > Cr > Pb > V > Ni > Sr > Li > Co > Mo > U > Cd > Ag > Tl, which was much different than the other cores.

Table 5 Summary of trace elements and Mn in the four sediment cores from the Pearl River and estuary
Full size table

The average concentrations of Ba, Ni, Cr, Mo, Pb, Cu, Zn, U, Cd, Ag, and Tl were much higher at site 1; the mean concentrations of Mn, V, Sr, Co, and Li were comparable to those at other sites (Fig. 4). Moreover, nine of the trace elements (Ba, Ni, Cu, Pb, Zn, U, Cd, Ag, and Tl) clearly decreased from upstream to the estuary, which was contrary to the Na distribution. The elevated concentrations of Zn, Cu, Pb, and Cd at site 1 were higher by a factor of two to three compared to adjacent sites (Wang et al. 2011), much greater than those in the Pearl River Estuary (Chen et al. 2012). However, they were comparable to those in the municipal sewage sludges (Fig. 5). Furthermore, it is noteworthy that the maximum concentrations of Zn (1,450 μg g−1), Cu (758 μg g−1), Pb (221 μg g−1), and Cr (232 μg g−1) at site 1 have never been reported in this area.

Fig. 5
figure 5

Comparison of heavy metal concentrations in different samples. TS this study (site 1), GS1 (Cai et al. 2007) and GS2 (C.S. Zhang et al. 2011) sewage sludge of Guangzhou, FS sewage sludge of Foshan (Cai et al. 2007), BS sewage sludges of Beijing (Dai et al. 2007), CS sewage sludge of Changsha (Yuan et al. 2011), MS sewage sludge of China (T. B. Chen et al. 2003), GC and GHP sediment from the Pearl River (Wang et al. 2011), ER and WR sediments from East River and West River (G. H. Zhou et al. 2013), ZE1 (Woods et al. 2012) and ZE2 (Ip et al. 2007) sediments from Pearl River Estuary. There were no data for Cd in ZE2

Full size image

The Dashadi wastewater treatment plant is situated west of the Huangpu Wenchong Shipyard, south of Huangpu East Road, with a proposed daily treatment capacity of 0.2 million tons. Site 1 is located between Dajisha Island and the Huangpu Bridge, within approximately 2 km of the Dashadi wastewater treatment plant. Therefore, the elevated trace metals in core 1 were considered as the anthropogenic contaminations primarily related to the discharge from the Dashadi wastewater treatment plant. For example, the median concentrations of Cd, Cu, Zn, and Pb were higher by a factor of 6–24 in sediment cores from the site nearest to the sewage treatment plant outlet than from the site located in the central part of Lake Geneva, which is as far as possible from direct human influences (Loizeau et al. 2004). A previous study confirmed that the highest Cu, Pb, and Zn concentrations (204, 332, and 2,460 μg g−1, respectively) in fluvial sediment were associated with discharge from a sewage treatment plant (Birch et al. 2001). In addition, research has shown that the average trace metal contents in sewage particulate matter samples from the Malabar effluent in Sydney were 24.9, 176, 819, 97, and 1,190 μg g−1 for Ag, Cr, Cu, Pb, and Zn (Matthai et al. 2002). Therefore, sewage-derived heavy metals are the main contaminants in sediment from the site adjacent to the sewage treatment plant. These contaminants rapidly decrease as the distance to the outlet increases. A lot of researches have also reported high concentration of heavy metals in sewage sludge from different wastewater treatment plants (Yuan et al. 2011; Filipovic et al. 2013). Another possible reason for elevated trace metals in core 1 is the surreptitiously dumping of sludge into the Pearl River as reported by a local newspaper (NFDaily 2012), even though it is forbidden by Chinese law.

The maximum value for Li (97.4 μg g−1) and the minimum values for V, Sr, and Co (64.4, 53.3, and 10.8 μg g−1, respectively) were recorded in core 2, mainly attributed to the distribution of these metals in the three tributaries (the East, West, and North Rivers). Similar to Ca, the average concentrations of V (89 μg g−1), Sr (51.7 μg g−1), and Co (14.0 μg g−1) were the lowest in the East River sediments, while the average Li content (43.1 μg g−1) was the highest in the sediment of this tributary (Zhang and Wang 2001). In addition, the average values of Li were much higher in the four cores than in the three tributaries (40.1 μg g−1) (Zhang and Wang 2001) and the average upper crust (20 μg g−1) (McLennan 2001) but were comparable to those in the Pearl River Estuary (55.50–73.55 μg g−1) (Qi et al. 2010; Woods et al. 2012). Apart from core 1, the Mo concentrations were comparable to the average upper crust values (1.5 μg g−1), slightly higher than those in the Pearl River Estuary sediments (1.0 μg g−1) (Woods et al. 2012) and lower than those in the Pearl River system sediments (2.47 μg g−1) (Zhang and Wang 2001). The concentrations of Cu, Pb, Zn, Cd, and Ag were much higher in core 2 than in cores 3 and 4. Meanwhile, the lower value of DO and higher value of turbidity were also recorded at site 2 (Table 1), which was consistent with the previous study (Dai et al. 2006). The persistent hypoxia in upstream of the channel in Guangzhou caused by organic pollution and nitrification was demonstrated (Luo 2002). Therefore, the higher values of trace metal in core 2 might be attributed to the water pollution.

The vertical distribution of trace metals in the four cores showed distinct differences. In core 1, most trace metal concentrations increased toward the surface with a peak at 6 cm and then decreased low contents at 3 cm, while Cd and Ag concentrations increased in the top 15 cm, followed by a period of lower levels (Fig. 4). Uranium exhibited a slight decreasing trend with depth and had the highest concentrations at 3 cm (10.7 μg g−1). In core 2, most metals generally increased downward from 0 to 9 cm, dropped significantly at depth of 15 cm, and then increased again to ~40 cm, where the highest contents of metals were recorded. Mn, V, Sr, Ni, Cr, Co, and Li in cores 3 and 4 showed relatively large fluctuations, while other trace metals, especially Cu, Zn, Cd, and Ag, exhibited almost constant trend.

Multivariate statistical analysis

Cluster analysis

In the current study, CA was applied for the bulk sediments to look for groups of subsamples in the four cores according to their similarities. The examined data matrix included the concentrations of major and trace elements in the bulk sediments. The result obtained is presented in a dendrogram (Fig. 6), where all 57 subsamples were classified into three statistically significant clusters. The subsamples in these clusters had similar pollution level and/or sources.

Fig. 6
figure 6

Dendrogram showing clustering of subsamples in the four cores. Label (Core 4–32): the 4 and 32 represent the sampling sites and the sample depth, respectively

Full size image

The character of sediment samples in cluster 1 (all belonging to core 1) was quite different from the other two clusters, because of their very high trace metal concentrations (Fig. 4). This cluster was obviously categorized as highly contaminated sediments. As illustrated above, core 1 was characterized as a wastewater- or sludge-contributed site. The second cluster might be regarded as less contamination level due to these subsamples that had a slightly lower level for major elements and significantly lower contents for trace elements. The other subsamples, which all belonged to the cluster 3, might be explained as the moderate pollution level.

Principal component analysis

Core 1 was characterized as a wastewater- or sludge-contributed site, which is very different than the other cores. Therefore, the data at sites 2, 3, and 4 were analyzed using PCA. Major and trace elements and LOI in cores 2, 3, and 4 were further investigated by applying PCA to obtain an overview of their behaviors and possible metal sources. Two factors of the principal components were extracted that described 77.3 % of the cumulative variance (Fig. 7). Factor 1 accounted for 48.0 % of the total variance and had strong positive loadings (>0.70) of Al, Fe, P, Mn, Sr, Li, Cr, Co, Ni, Cu, Zn, and U and moderate positive loadings of Mg, Ti, Ba, V, Pb, Tl, and Mo. Usually, Al, Fe, and Mg are the major constituents of common silicate minerals. Moreover, Al held in the lattice of aluminosilicate minerals is extremely immobile in the marine environment (Qi et al. 2010). Therefore, this factor represents the lithogenic origin from weathering and erosion of rocks and soil materials in the Pearl River catchment.

Fig. 7
figure 7

Principal component score plots of major and trace elements in sediment cores. Data are obtained from the subsamples of cores 2, 3, and 4

Full size image

Factor 2, which accounted for 29.3 % of the total variance, was distinguished by highly positive loadings of Na (0.845), Mg (0.763), Ti (0.686), V (0.682), Ca (0.542), and LOI (0.531). In contrast, K (−0.809), Cd (−0.774), Ag (−0.731), Ba (−0.687), Tl (−0.668), Zn (−0.618), U (−0.599), and Pb (−0.506) showed highly negative loadings for factor 2. The main source of Na to the estuarine environment was seawater; the marine source of Na also showed a strong relationship with Mg and K (Spencer 2002). Previous studies in the Pearl River Estuary had also showed that the distribution of Mg and Ca may reflect the combined effects of parent material influences and sedimentation processes in the interaction with marine water (Li et al. 2000; Liu et al. 2003). Na had a strong positive correlation with Mg and a negative correlation with K in this study (Table 6). An examination of the K data revealed an impoverishment of K in estuary sediments and a relative enrichment in river sediments, which was contrary to Na data (Fig. 2), suggesting that there may be an additional anthropogenic source of K in the study area.

Table 6 Pearson’s correlation coefficients (r) between major and trace elements in cores 2, 3, and 4 (n = 50)
Full size table

Factor 2, which is typically indicative of a marine source, was dominated by Na, Mg, Ti, V, and Ca. Generally, an opposite relationship between the anthropogenic metals (fluvially/terrestrially derived) and Na (marine-derived) would be expected (Spencer 2002). Therefore, K, Cd, Ag, Ba, Tl, Zn, U, and Pb may be derived from anthropogenic sources. It is noteworthy that Ba, Tl, Zn, U, and Pb had strong or moderate loadings both on factor 1 and/or factor 2, indicating that these trace elements were derived both from lithogenic and anthropogenic sources, while K, Cd, and Ag were primarily derived from anthropogenic sources.

Enrichment factor

The first but crucial step for EF calculations is to select the reference element and the background values. Typically, Al is considered as an appropriate element to normalize the metal contents because it is a conservative element and one of the most abundant elements on the earth, usually with no contamination. In this study, aluminum was better correlated with trace metals than Fe, suggesting that Al was the main geochemical carrier in the studied sediments. The world average shale and the upper continental crust are often used to provide background metal levels, which allow comparison with the values obtained in the studied areas, as well as with the EF reported in other estuarine areas (Celis-Hernandez et al. 2013). However, these metal levels tend to be very general and may be misleading in a specific coastal area (Gibbs 1993). The regional background values would be more appropriate and reliable, but it is difficult to ascertain the regional preindustrial sediment data. Therefore, EFs were calculated using Al and three background values (regional background values (Qi et al. 2010), the upper crust for Central East China (UCCEC) (Gao et al. 1998) and the upper crust for the world (UCW) (McLennan 2001)), with the purpose of comparing the effect of different background values. It should be noted that the regional background values used were lacking data for some trace metals (i.e., Mo, U, Cd, Ag, and Tl), which may often occur when using regional background values to calculate the EFs.

The EFs using regional background values were 0.5–1.3 for EF Cr , 0.3–0.6 for EF Co , 0.4–1.4 for EF Ni , 0.6–9.7 for EF Cu , 0.6–6.1 for EF Zn , 0.7–3.1 for EF Pb , and 0.3–0.5 for EF Li , suggesting that Cr, Co, Ni, and Li may be entirely from crustal materials or natural weathering processes (Table 7). In contrast, the higher average of EF Cu (7.2), EF Zn (4.8), and EF Pb (2.7) at site 1 suggests that these metals were moderately or significantly polluted. Comparing these metal values with effects range low and effects range median (Table 5), it can be concluded that site 1 was severely polluted by Cu, Zn, and Pb. Moreover, moderate pollution was found using the average EF Zn (4.8) and EF Pb (2.7) based on the regional background values. EF Co , EF Ni , EF Zn , EF Cd , and EF Li were comparable between the two crust-referenced EFs, whereas EF Cu , EF Pb , EF Mo , EF U , EF Ag , and EF Tl were relatively different. Specifically, EF Cu , EF Pb , and EF Ag were higher based on the UCW than those based on the UCCEC. However, EF Mo , EF U , and EF Tl were reversed. As illustrated by the PCA, Cu mostly originated from a natural source; U and Tl were partially derived from anthropogenic sources. In view of these conflicting features, selecting an appropriate background value is essential for obtaining reliable EFs. Other research has also shown that EF values will change considerably with different baseline values (Abrahim and Parker 2008; Celis-Hernandez et al. 2013). Consequently, it is more logical and appropriate to use the UCCEC for normalization than to use regional background values or the UCW.

Table 7 Mean metal enrichment factor values in the four sediment cores from the Pearl River and estuary normalized with respect to the aluminum content in regional background, upper crust for Central East China, and upper continental crust, respectively
Full size table

The results obtained using the UCCEC demonstrate that high EF values (e.g., 25 for Cd and Ag, 14 for Cu, 12 for Zn and Mo, 7.0 for Pb, and 4.7 for U) were found at site 1 (Table 7), while EF values for these metals were significantly lower at sites 2, 3, and 4. In contrast, the average EF value for Co at site 1 was comparable with its values at other sites, suggesting that sewage discharges or sludge may not be the primary sources for Co compared with Cu, Zn, Pb, Cr, Ni, and Cd. Specifically, the average EF Cd (21) at site 1 was larger than 20, indicating that Cd contamination was very high. The average EF values for Ag, Cu, Zn, Mo, and Pb at site 1 and for Cd at site 2 were between 5 and 20, suggesting significant contamination of these metals. The majority of the remaining EF values were less than 5, and some were even less than 2, indicating that these metals were at most moderately contaminated.

Comparison of the PCA and EF

It is noteworthy that conflicts were observed for some trace elements when comparing the results of PCA, ERL-ERM with EF. For example, Li mainly originated from natural sources based on the PCA. However, it was moderately polluted based on the EF. On the contrary, Tl primarily originated from anthropogenic sources based on the PCA. However, nearly no pollution was found based on the EF. The concentrations of Cd and Ag were lower than the ERM value but exceeded the ERL value at site 1, while it was highly polluted based on the EF. These conflicts can be attributed to the background values used in the EFs. The average value (18 μg g−1) of Li in the UCCEC was much lower than the minimum value (47.5 μg g−1) in our study. It is also lower than the minimum value (26.8 μg g−1) in surface sediments from the Pearl River Estuary (Woods et al. 2012). However, other trace elements (i.e., Cr, Co, Ni, Cu, Zn, Pb, Cd, and Tl) were higher in the UCCEC compared with the relevant minimum values in the same literature (Woods et al. 2012). In the case of Tl, the minimum value (0.18 μg g−1) in surface sediments from the Pearl River Estuary (Woods et al. 2012) was approximately 43 % of the UCCEC content (0.42 μg g−1) (Gao et al. 1998). Therefore, the low background value of Li and the high background value of Tl may result in the high and the low EFs of Li and Tl, respectively. It is important to note that selecting the background values for normalization should consider the existing data in the study area (including the core samples and/or the unpolluted surface sediment samples) and the regional and worldwide average values in shales and/or continental crusts. Furthermore, carefully comparing these data to determine the combined background values, which may be more appropriate and logical than single-source background values, is necessary.

Conclusions

The present study showed preliminary but still relevant information regarding the mineralogical and geochemical compositions among sediments in four cores from the Pearl River and estuary. One core (Huangpu) with very high contents of trace elements was identified as heavily polluted based both on sediment quality guidelines (ERL-ERM) and EFs. Sewage disposal and/or the surreptitious dumping of sludge was likely the source for the trace elements in this site. Moreover, significant metal pollutions in sediment from the other three sites were not found. Very high pollution for Cd and significant contaminations for Ag, Cu, Zn, Mo, and Pb were identified based on EFs. Comparing the EF results with the PCA and ERL-ERM results, conflicts are observed for some trace metals, which were attributed to the background values used to compute the EFs. Therefore, combined background values may be more appropriate and logical than single-source background values. The multivariate analysis procedures were useful tools for the classification of the pollution levels. Subsamples (Huangpu) were excluded prior to performing the PCA based on cluster analysis, which separated the subsamples (Huangpu) with a distinct single source from others. Continuous monitoring and further studies of the area adjacent to the outlet of the wastewater treatment plant are needed to identify the pollution sources and assess the toxicity of the very high levels of trace metals to the benthic organisms. The authorities should also take forceful measures to prevent the dumping of sludge into the rivers.