Introduction

The worldwide water and sediment quality deterioration resulted primarily from growing human populations and economic development, particularly elevating nutrients leading to eutrophication and heavy metal pollution in the aquatic environment (Krishna et al. 2009; Nriagu and Pacyna 1988; Peierls et al. 1998; Holloway et al. 1998; Soylak and Yilmaz 2006; Li et al. 2008; Mendil et al. 2010; Venkatramanan et al. 2012, 2015). The natural sources of heavy metals consist of volcanism, bedrock erosion, atmospheric pollution, and the release from plants and anthropogenic activities. Particularly, mining and mineral processing have dominant influences on the biogeochemical cycles of heavy metals (Nriagu 1989, 1996; Hongyi et al. 2009). Heavy metal pollution leads to serious human diseases through the food chain and the loss of biodiversity, and it also degrades the environmental quality. In recent researches of heavy metals, the spatial variability reflects geological parent materials and anthropogenic sources in a geographically heterogeneous area (Imperato et al. 2003).

Industrialization and unplanned urbanization have also greatly changed the natural environment. In recent times, the environment has become aggressive, posing a threat to the health and welfare of the people due to the release of pollutants from industries, agricultural fields and urban waste. Their waste waters find their way into surface water bodies via rivers, canals and surface runoff. Thus, these water bodies receive pollutants, and surface sediments become contaminated. The impacts of the heavy metals in sediments were assessed in terms of contamination factor (CFs), geo-accumulation index (I geo) and pollution load index (PLI), which is reported based on the chemical analysis of the bulk sediments. These pollution indices methods have been widely utilized in river and marine sediments (Muller 1969; Bryan and Langston 1992; Buccolieri et al. 2006).

Principal components analysis (PCA) and cluster analysis (CA) have been considered as the most trustworthy approaches for data mining of matrices from environmental quality assessment (Astel et al. 2007, 2008; Simeonova and Simeonov 2007). Thus, PCA and CA are widely used in sediment quality assessment of natural fresh water system. Recently, human population and economic activities have greatly increased in the study area, and sediment and water quality has been continuously threatened from various contamination sources.

Rapid industrialization, urbanization and population increases in the last few decades caused a dramatic increase in the demand for river water, and significant deterioration in water and sediment quality, especially near large industrial complexes and in the lower basin. These trends are expected to be continued, unless appropriate management strategies are carried out to ensure adequate water supplies and to restore water and sediment qualities to appropriate standards for the intended uses (Chun et al. 2001; Venkatramanan et al. 2014). This research evaluates the physico-chemical parameters of river water and the selected heavy metals in surface sediments of Nakdong River in Korea to identify the sources of contaminants, using comparative evaluations of pollution load index (PLI), geo-accumulation index (I geo) and contamination factor (CFs). PCA and CA were also used for this purpose.

Materials and methods

Study area

Nakdong River is the longest river in Korea, which is 525 km in length with the total watershed area of 24,000 km2 (Fig. 1). Nakdong River originates in the vicinity of Taebaek city in Kangwon Province, and two dams of Andong and Imha Dam were constructed at the upstream of Nakdong River. Those dams are the multipurpose types of water supply, electricity production and flood control for the surrounding area, and also serve the improvement of water quality in the downstream area. The river finally flows into Busan City, and discharges to East China Sea. A barrage was constructed at Nakdong River mouth in 1987 to prevent seawater intrusion and supply fresh water for domestic usage. Nakdong River forms a natural border in the study area, and it supplies a large amount of water to water supply facilities of the surrounding areas.

Fig. 1
figure 1

Map showing Nakdong River Basin and sample stations

Full size image

Heavy flooding was a frequent phenomenon in this river before the construction of 9 small dams. In monsoon season, organic matter/sediments were heavily deposited at the delta area in the downstream of the river. The delta has sedimentary deposits of 60–90 m thickness around the river banks and is composed of backfill, sandy clay, stiff clay, sand and gravel. Most of sediments were deposited from the late Pleistocene Epoch, i.e., the end of 4th glacial period (Oh 1994; Ryu et al. 2011). Basal gravel bed indicates an unconformity between delta sediments and granites of the Cretaceous Period. The thickness of sandy clay ranges from 5 to 10 m, stiff clay from 10 to 30 m, lower sand from 10 to 40 m, and gravel from 5 to 40 m, respectively. Sandy clay is relatively soft and loose, but lower clay is stiff and dense.

Environmental setting

The characteristics of the sampling stations are given in Table 1. The Nakdong River Basin occupies about 24 % of Korean land, and 18.4 % of the land in the basin is used for farming (Ministry of Construction and Transportation 1998). Rice cultivation covers nearly 62 % of the farm land, and dry fields are 38 % of farm land. Building areas and industrial complexes occupy about 1.8 % (430 Km2) of the total basin area. A total of 149 industrial complexes including 7 multinationals, 51 regional and 91 rural industries inhabit the region. National and regional industrial complexes are located within or near the major urban centers, while rural industrial complexes are characterized by a wide spectrum of small scale industries scattered throughout the basin. The upper Nakdong Basin was developed for electronic industries, the middle basin for textile/dyeing industries and the lower basin for heavy industries and shoemaking. The use of fertilizer was rapidly increased by agricultural growth during the 1970s. However, it has decreased since 1991, because many regions were urbanized and industrialized. Airport and many residential houses are also located in the delta area of the lower Nakdong Basin (Chun et al. 2001).

Table 1 Characteristics of sampling stations at Nakdong river basin
Full size table

Sample collection

A total of 14 water samples were collected during July 2011 (summer season) using a 2-L Van Dorn plastic water sampler. Sample bottles were rinsed with the same sample water before sampling. Samples were preserved by acidifying to pH 2 with the addition of HNO3 and kept at 4 °C until further analysis. pH and dissolved oxygen (DO) were checked in situ with a portable meter Horiba U-51, Japan. Biological oxygen demand (BOD), chemical oxygen demand (COD), total organic carbon (TOC) were measured by complying standard procedures (APHA 1995). Fourteen sediment samples were collected using a Van Veen grab sampler on board with a hired fishing trawler, and the top 5 cm of the sediments were collected from the grab with a plastic spatula. The sediment samples were mixed with HNO3 + HCl and were digested for 40 min. The final solution was analyzed by atomic absorption spectrometer (AAS, Unicam 989, USA). The accuracy of the analytical method was analyzed by the standard reference material MAG-1 (marine mud from the United States Geological Survey, USA) and National Research Council of Canada Marine sediment reference material BCSS-1. The average recoveries ± standard deviations for each metal were 73 ± 15, 71 ± 14, 79 ± 24, 64 ± 22 and 68 ± 16 (MAG-1) and 62 ± 18, 60 ± 24, 83 ± 16, 65 ± 23 and 58 ± 26 (BCSS-1) for Cu, Cd, Pb, Zn and As, respectively.

Evaluation of pollution indices

The pollution indices formulas were developed for evaluating the sediment quality, and many calculation methods have been put forward to quantify the degree of metal enrichments. Various authors (Salomons and Forstner 1984; Muller 1969; Hokanson 1980) have proposed the pollution impact scales or ranges to convert the calculated numerical results into broad descriptive bands of pollution ranging from low to high intensity. Sediment quality guidelines provide values that allow quantification of sediment contamination, and that make an overall assessment of the metal contamination degree in river and marine sediments. Contamination factor (CFs), pollution load index (PLI) and geo-accumulation index (I geo) are very useful for the assessment of metal contamination in sediments. PLI and I geo were developed to evaluate the background values of upper continental crust (Wedepohl 1995), and the methods were used for all sampling stations for this study. CF was calculated for an uncontaminated site in the study area.

Geo-accumulation index (I geo) which is a common approach for the estimation of metal enrichment proposed by Müller (1969) is calculated as follows:

$$I_{\text{geo}} = { \log }_{2} C_{n} /1.5 \times B_{n}$$
(1)

where C n is the concentration of a metal in an enriched sample, and B n is the background value of a metal. The factor 1.5 is introduced to minimize the effect of possible variations in the background values, which may be attributed to lithological variations in the sediments (Müller 1969). Muller’s classification suggests that if the value is less than 0, a sediment is unpolluted (Class 0). In case of 0–1, it is unpolluted to moderately polluted (Class 1); 1–2, moderately polluted (Class 2); 2–3 moderately to strongly polluted (Class 3); 3–4 strongly polluted (Class 4); 4–5 strongly to extremely polluted (Class 5) and >5 extremely polluted (Class 6). CF is expressed as

$${\text{CFs}} = M_{x} /M_{b}$$
(2)

where M x is the concentration of metal and M b is the concentration of metal in the selected reference background value. If CFs > 1 for a particular metal, a sediment is contaminated by a metal. If CFs < 1, a sediment is uncontaminated by natural or anthropogenic inputs. The CFs is classified into four groups based on the calculated values (Pekey et al. 2004; Hokanson 1980; Savvides et al. 1995): CFs < 1, low contamination; 1–3, moderate contamination; 3–6, considerable contamination; and >6, very high contamination.

PLI is the summation of several heavy metals and is defined as the nth root of the multiplication of concentration factors (CFHMk) (Tomlinson et al. 1980) and is calculated like

$${\text{PLI}} = n\sqrt {\sum\limits_{k = 1}^{n} {{\text{CF}}_{\text{HMk}} } }$$
(3)

where CFHMk is the ratio between the concentration of each heavy metal (CHM) and a background value (Taylor and McLennan 1995). When PLI > 1, it means that pollution exists. If PLI < 1, there is no metal pollution. PLI gives an assessment of overall toxicity status for a sediment (Lu et al. 2009). In the present study, PLI is a result of the contribution from five heavy metals of Cu, Cd, Pb, Zn and As.

Statistical techniques

The experimental data were subjected to statistical analyses using STATISTICA (Ver. 7) software. PCA was performed with a view to assess compositional differences among physical parameters in water and the variations of heavy metal concentrations in sediment sources. Factor analysis was performed by a varimax orthogonal rotation, which minimized the number of variables with a high loading on each component, and it facilitates the interpretation of PCA results (Howitt and Cramer 2005). Likewise, Pearson’s correlation matrix was also used to identify the relationship between different elements.

Cluster analysis was applied to identify groups of samples with similar heavy metal behavior (Panda et al. 2006). It was formulated according to the Ward-algorithmic method, and the rescaled linkage distance was employed for measuring the distance between clusters of similar heavy metal contents. Cluster analysis was used to determine the association of different sediment and water samples.

Results and discussion

Physico-chemical parameters in river water

The distribution of water parameters is shown in Fig. 2, and basic statistical analysis are given in Table 2. Levels of pH, DO, BOD, COD and TOC varies from 6.9 to 7.4, 7.7 to 13.2 mg/l, 0.5 to 2.6 mg/l, 2.8 to 6.2 mg/l and 1.69 to 5.04 mg/l, respectively. pH values exhibit slight alkalinity in the water samples, which is due to the variable inputs from industrial effluents (Santhiya et al. 2011). BOD, COD and TOC were slightly higher than irrigation water quality. DO plays a vital role in supporting aquatic life in surface waters, which is susceptible even to slight environmental changes (Iticescu et al. 2013). DO in water shows lower values as they are consumed for the oxidation of organic matters. Contributions of BOD (0.5–2.6 mg/l), COD (2.8–6.2 mg/l) and TOC (1.69–5.04 mg/l) were derived from agricultural activities, domestic and industrial effluents (Chun et al. 2001).

Fig. 2
figure 2

Variations of primary water quality values according to sample stations in Nakdong River Basin

Full size image
Table 2 Basic statistics of the primary parameters of water quality and metals of sediments
Full size table

Heavy metals in sediments

The general grain size characteristics of the sediments from Nakdong Basin exhibit 70–80 % of sand and gravel, and 20–30 % of silt and clay in the upper basin, while the middle and lower part of the basin consists of 30–40 % of sand and gravel, and 60–70 % of silt and clay. Concentrations of Cu, Cd, Pb, Zn and As (all values in mg/kg) in sediments ranged from 0.39 to 54.2, 0.03 to 0.28, 0.65 to 18.9, 1.97 to 76.6 and 0.02 to 0.45, respectively. The order of heavy metal levels with reference to the average values in surface sediment is as follows: Zn > Cu > Pb > As > Cd (Fig. 3). The higher values of Zn (76.6 mg/kg) and Pb (18.9 mg/kg) in river sediments are due to the impacts from textile and heavy industries (e.g., Adamo et al. 2006). Cu, Cd and As present a slightly higher concentration only in sediments of station 8. This is evident due to the presence of a number of heavy and chemical industries in the region as the runoff is lixiviated, precipitated and brought through various unregulated minor channels that drain into the aquatic region (Tuzen et al. 2004; Anithamary et al. 2012, 2013; Jonathan et al. 2013).

Fig. 3
figure 3

Variations of heavy metal concentrations in surface sediments according to sample stations in Nakdong River Basin

Full size image

The classification of total heavy metal contents (mg/kg) of Cu + Cd + Zn + Pb + As and pH indicates the near-neutral field of high heavy metal sector (Fig. 4). The above inference reveals that metals are anthropogenic in nature mainly due to the presence of industrial and agricultural fields (Ficklin et al. 1992; Caboi et al. 1999; Jonathan et al. 2013).

Fig. 4
figure 4

Classification of samples based on total heavy metals vs. pH

Full size image

Evaluation of pollution indices

Geo-accumulation index of heavy metals was used for the better understanding of the pollution indices (Fig. 5). I geo values ranged from −33.8 to 0.19, and it suggests that the heavy metals are from natural weathering processes (e.g., Zhang and Liu 2002). This method assesses the degree of metal pollution in terms of seven enrichment classes based on the increasing numerical values of the index.

Fig. 5
figure 5

Geo-accumulation index values of heavy metals in the study area

Full size image

CF values ranged from 1.41 to 44.32 and it is classified into four categories (e.g., Pekey et al. 2004; Hokanson 1980; Savvides et al. 1995): low (CFs < 1), medium (1–3), considerable (3–6) and high (CFs > 6). Most of samples in the study area are moderately polluted, while few samples fall at very high polluted category (Fig. 6a).

Fig. 6
figure 6

Variations of heavy metal pollution indices: a CF; b PLI

Full size image

PLI was used to determine the synthetic pollution effect at different stations by the metals. In case PLI = 1, heavy metal loads are close to background, and samples are considered as nontoxic. PLI > 1 represents a progressive pollution (Tomlinson et al. 1980). PLI values of all samples in the present study ranged from 1.36 to 2.73 and exhibit a progressive pollution (Fig. 6b).

All samples belong to medium to very high pollution in CFs. CFs and PLI values show almost quite similar trends at various sampling points, but I geo represents a different variation from CFs and PLI. In this study, most of the samples fall in medium to high pollution category, which is attributed to the local point sources associated with industrial and agricultural fields.

Identification of pollution sources

Correlation matrix

The Pearson’s correlation coefficient matrices for the analyzed parameters are presented in Table 3. Correlation coefficients less than 0.5 are not included in the table, because they are not at significant levels. The inter-relationship of different parameters is useful in revealing some association of heavy metals with the physico-chemical conditions in the river water. Association of BOD shows significant correlations with As (r 2 = 0.66) indicating that it could be directly related to the absorbing capacity of organisms in the study area (Yang et al. 2009). Likewise, COD exhibits good correlation with TOC (r 2 = 0.81), suggesting that considerable amount of organic matter is brought by tributaries and industries (Venkatramanan et al. 2014). The association of heavy metals indicates that Cu correlates significantly with Cd (r 2 = 0.70), Pb (r 2 = 0.94), Zn (r 2 = 0.93) and As (r 2 = 0.65), suggesting that they have various assimilated contaminants from industrial wastewater, agricultural effluents and municipal sewage (Yalcin et al. 2008; Tariq et al. 2010). The significant correlations among Cu, Zn, and Pb indicate that they may have originated from common sources, preferably from anthropogenic activities. Moreover, it is also due to the textile/dyeing and paint industries, or pesticides (Fukushima et al. 1992).

Table 3 Correlation matrices for primary water quality parameters and heavy metals
Full size table

Principal component analysis (PCA)

PCA varimax orthogonal rotation (Gotelli and Ellison 2004) was used to maximize the sum of the variance of the factor coefficients which explained the possible groups/sources in the Nakdong River Basin. Four factors were extracted from the chemical data set based on eigenvalues. The calculated factor loadings, together with percentages of variance, and cumulative percentage of variance by each factors, are listed in Table 4.

Table 4 Q and R modes principal component analysis of primary water quality parameters and heavy metals
Full size table

PC1, PC2, PC3, and PC4 show 40, 22, 20, and 15 % of total variance, respectively. PC1 is heavily loaded on Cd and Pb, which are mostly distributed in sample stations of 6, 9, 10, 11 and 12. PC1 components are derived from mixed sources of industrial wastewater, agricultural effluents and municipal sewage. PC2 is loaded on Cu and Zn, which may be derived from the pesticides and industries (e.g., Jonathan et al. 2004; Jayaprakash et al. 2007). These parameters are importantly distributed in sample stations from 11 to 14. PC3 contributed by pH and As represents direct anthropogenic source from the industries and agriculture, which are significantly distributed in sample stations of 1, 2, 4, 8 and 11. PC4 is loaded on DO only, which occurs as an important parameter in sample station 1, 9 and 12, which directly reflects the influence of irrigation runoff. When sampling sites were plotted on the plane of the first two principal components (PCA 1 vs. PCA 2) of the Q-mode PCA, PCA 1 separated 3, 4, 5, 13 and 14 from the other stations (Fig. 7a), indicating that heavy metals are derived from external sources such as agricultural and industrial region of the basin.

Fig. 7
figure 7

Multivariate statistical analysis of water and sediment samples: a PCA; b dendrogram of R-mode; c dendrogram of Q-mode

Full size image

Cluster analysis (CA)

Cluster analysis was also performed to understand the chemical components and heavy metals groupings in the data set (Fig. 7b, c). Parameters in the same cluster are likely to originate from a common source. The R-mode CA performed on the samples produced four clusters. Cluster 1 includes pH and DO; cluster 2 consists of BOD, Cd and As; cluster 3 contains COD and TOC; cluster 4 includes Pb, Cu and Zn. Cluster 1 is related with PC3 and PC4 of R-mode PCA, and Cluster 2 and 3 are related with PC 1 of R-mode PCA. Cluster 4 is related with PC1 and PC2 of R-mode PCA, reflecting the influence of anthropogenic sources and minor natural input.

In case of Q-mode analysis, the 14 sampling stations in Nakdong River Basin are classified into four major clusters. Cluster 1 consists of four sampling points (1, 2, 5 and 7). These sites receive inputs from industries (1, 2) and agricultural contaminants via irrigation (5, 7). Cluster 2 consists of five sampling points (4, 6, 12–14), whereas cluster 3 consists of four stations (3, 9, 10, 11). This indicates the accumulation of metal loads from industrial effluents, metal-laden agricultural runoff and domestic sewage. Cluster 4 consists of only one station (8) because this station lies in mouth of the river, where the contaminants accumulate through flocculation and coagulation process (Venkatramanan et al. 2014).

Comparison of heavy metals with world regions

Heavy metal concentrations of present study were compared with other regions around the world, indicating higher values for Cu (54.2 mg/kg) than other regions expect Pearl River, China. In the case of Cd (0.28 mg/kg), it is on the lower side than the other regions. Pb (18.9 mg/kg) and Zn (76.6 mg/kg) values in this study were lower than other regions expect Uppanar River and Thirumalairajan River, India. According to ecotoxicological values, Cu concentration exhibits LEL and ERL and the rest of metals are below the toxic condition (Table 5).

Table 5 Comparison of heavy metal concentrations in surface sediments of Nakdong River with those of other rivers in the world
Full size table

Conclusion

Heavy metal pollution indices and multivariate statistical analysis were used to assess the intensity and sources of pollution in the Nakdong River Basin. The sediment quality of the River Basin at Busan City (St.no 8) had the highest concentration of metals. The order of primary water quality parameters and heavy metal levels in descending order is as follows: DO > COD > TOC > BOD (water) and Zn > Cu > Pb > As > Cd (sediment), respectively. The pH values and metals concentrations of this river basin exhibited near-neutral high metal sector. Heavy metal pollution indices seemed to be promising and beneficial in assessing the enrichment/contamination status of sediments in terms of heavy metals. CFs suggested that all samples fell in medium to very high polluted zone. Otherwise, PLI showed that all samples belonged to progressive pollution. I geo revealed that all samples fell in the unpolluted to moderately polluted sector. Multivariate statistical analysis indicated that anthropogenic impact was responsible for controlling the variability of primary water quality parameters and total heavy metal contents in this study region. Sediment quality analysis clearly states that the heavy metals are released from industrial wastewater, irrigation effluents and domestic sewage. The contamination of surface sediments by heavy metal pollutants could generate serious threats to human health and ecological habitat in this river basin. In conclusion, this research represents that the combination of pollution index methods and multivariate statistical analysis is an important tool to identify contamination sources and origins.