Ecological Indicators 82 (2017) 32–42

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Contamination, ecological risk and source apportionment of heavy metals in MARK

sediments and water of a contaminated river in Taiwan
⁎
Chi Thanh Vu, Chitsan Lin , Chien-Chuan Shern, Gavin Yeh, Van Giang Le, Huu Tuan Tran
Department of Marine Environmental Engineering, National Kaohsiung Marine University, Kaohsiung 81157, Taiwan, ROC

A R T I C L E I N F O A B S T R A C T

Keywords: Houjing River, located in Kaohsiung City, is one of the most seriously contaminated rivers in Taiwan. This study
Heavy metals analyzed the concentrations of heavy metals (As, Cd, Cr, Cu, Pb, Ni, Zn and Hg) in sediments and water samples
Sediment quality guidelines collected from this river. Analysis of contamination factor (Cfi ) and potential ecological risk factor (E fi ) of heavy
Multi-element indices metals in water showed that there were low grades of contamination and potential ecological risk for all heavy
Positive matrix factorization
metals, suggesting that heavy metals in water were less likely to pose risks to the ecosystem. However, sentiment
samples were found to have severe contamination levels based on ranges found in sediment quality guidelines
(SQG). The average Cu concentration was almost twice as high as the upper standard values in all the guidelines.
Multi-element indices were used to evaluate the synergistic effects of different metals at the sampling sites. The
calculated results of different indices, the modified degree of contamination (mCd), Nemerow pollution index
(PN) and potential ecological risk index (RI), were in good accordance. ‘Heavy’ contamination and ‘severe’
ecological risk were found at three sites, Demin, Zhuwai and Renwu. Cadmium and copper contributed the
highest to the ecological risk there. Results of positive matrix factorization modeling identified four sources of
heavy metal pollution in both sediments and water. Heavy metal contamination in the Houjing River is at-
tributed to companies carrying out various industrial processes along the riverbank, including traditional metal-
plating, plastic manufacture and semi-conductor packaging. Therefore, future pollution control and management
plans should emphasize the strict regulation of discharge from these industrial activities.

1. Introduction of these heavy metals in the environment.

Heavy metal pollution in aquatic environments has attracted
widespread attention due to its persistence, accumulation in the food
chain and negative effects on the ecological and human health (Lin
et al., 2016; Zhang et al., 2016c). Industrialization creates a large
amounts of residual anthropogenic metals (e.g. iron oxides, manganese,
organic compounds, etc.), which are then rapidly deposited and
strongly attached to different types of sediment fine grains (Tessier and
Campbell, 1987; Simpson and Spadaro, 2016). The metals can also be
detached and released into the water column, negatively affecting
water quality (Simpson and Spadaro, 2016). This occurs more often for
surface sediments because there are large environmental variations that
can rapidly alter different factors such as pH, temperature, and bio-
turbation (or resuspension) (Simpson and Batley, 2007). Anthropogenic
metals, different from those that are lithogenic, are highly mobile and
bioavailable, thus are more likely to adversely affect aquatic species
(Tessier and Campbell, 1987). Understanding the distribution, potential
ecological risk, and sources of emission are crucial for the management
Although sequential extraction procedures can be used to determine
chemical speciation of heavy metals and provide detailed information
about their mobility, bioavailability and toxicity (Tessier and Campbell,
1987; Simpson and Spadaro, 2016), complications involved in labora-
tory work performing these procedures hinder their widespread appli-
cation. Total metal concentration has been commonly used to assess
pollution status as well as potential ecological risk (Duodu et al., 2016;
Lin et al., 2016; Villanueva and Ibarra, 2016). Many approaches have
been developed and optimized for the assessment of potential ecolo-
gical risk (Brady et al., 2015). The most common method of evaluating
heavy metal pollution and ecological risk in sediments is by calculating
contamination factor (Cf) of single elements as well as their enrichment
factor (Ef) and geo-accumulation (Igeo) (Islam et al., 2015b; El Nemr
et al., 2016). Yet, because heavy metals are more likely to have sy-
nergistic effects in the environment, these single element indices could
be insufficient in their assessment of contamination and risk (Duodu
et al., 2016). Multi-element indices, e.g. modified degree of con-
tamination (mCd) and Nemerow pollution index (PN), can take into

⁎
Corresponding author.
E-mail address: ctlin@webmail.nkmu.edu.tw (C. Lin).

http://dx.doi.org/10.1016/j.ecolind.2017.06.008
Received 26 January 2017; Received in revised form 4 June 2017; Accepted 5 June 2017
1470-160X/ © 2017 Elsevier Ltd. All rights reserved.
C.T. Vu et al.
account the synergistic effects of different heavy metals (Hakanson
1980; Yan et al., 2016). Similarly, synergistic ecological risk can also be
quantitatively illustrated by the potential ecological risk index (RI).
Southern Taiwan, the most industrialized area of the whole island,
has been negatively affected by the industries there (Wang et al., 2015).
Kaohsiung City has been the center of a number of different heavy in-
dustries, which have ceaselessly grown for the last 30 years. Many in-
dustrial plants producing different metals have seriously polluted water
bodies around the city. The Houjing River, located in the northern part
of Kaohsiung City, receives a large amount of polluted water and se-
diment discharge from the industrial plants operating along its banks on
a daily basis. Historically, there has been continual illegal discharge of
huge amounts of different pollutants from the various manufacturing
companies there (Lin et al., 2009, 2010; Jiang et al., 2015). The quality
of the Houjing River has attracted wide public attention leading to the
need for assessment of risk and frequent monitoring to guarantee the
river’s support of the health of the environment there as well as the
health of its residents (Lin et al., 2009; Jiang et al., 2015). Despite this
great need, no study has investigated the heavy metal contamination in
the Houjing River in detail, assessed its impact or identified the possible
sources of pollution there.
Principal component analysis (PCA) and cluster analysis (CA) are
widely preferred when determining sources of contamination in sedi-
ments and water because they are easy to do and available in most
statistical software packages (Comero et al., 2011; Duodu et al., 2017).
However, since they are data-sensitive, their results may come with
skewed dataset distributions, requiring subsequent statistical normal-
ization (Praipipat et al., 2013). Furthermore, it is rare that much at-
tention is paid to the ability of these analytical methods to handle
outliers and below-detection-limit (BDL) and/or missing values (Pekey
and Dogan 2013). However, positive matrix factorization (PMF), an
advanced multivariate statistical method using least square approach
developed by Paatero and colleagues (Anttila et al., 1995), can address
those uncertainties by incorporating non-negative constraints into the
configuration process and point-to-point estimating each data value’s
error (Brown et al., 2015). Another advantage of PMF is that its asso-
ciated receptor-oriented model is effective even with analyzing limited
datasets (Bhuiyan et al., 2015). While the application of PMF has been
specifically preferred in the field of atmospheric chemistry for years
(Wang et al., 2016), it is uncommon to find it applied to understanding
contamination in sediment and water samples (Bhuiyan et al., 2015).
This study assessed the contamination and potential ecological risk
posed by heavy metals to the Houjing River employing PMF in source
apportionment of the contamination in sediments and water. The re-
sults of this study can help provide decision-makers with an insightful
view of the current contamination status of that river, essential to im-
pending changes in land-use there, the area’s economic development
and environmental protection strategies.
2. Methodology
2.1. Location
The Houjing River is situated in north-west Kaohsiung City in
southern Taiwan (Latitude 22°69′21.63″N − 22°73′17.74″N and
Longitude 120°25′74.36″E − 120°33′74.70″E). As can be seen in Fig. 1,
the river has two upstream reaches − Dashe to the north-east in
Ciaotou District and Renwu to the south-east direction in Renwu Dis-
trict. The two reaches then meet at Si-Chingpu Landfill before flowing
directly into the Taiwan Strait. The length of the Houjing River is about
13 km and its basin has long winters and summers, and short springs
and autumns. The main source of water for the river is discharge from
industrial areas, wastewater treatment plants, municipal discharge
from activities along the riverside and rainfall. In the past, the Houjing
River was a reliable resource of water for irrigation and various agri-
cultural activities in the area from where Kaohsiung City obtains much
33
Ecological Indicators 82 (2017) 32–42
of its food. Urbanization and industrialization have led to serious pol-
lution of the river and altered its nature and chemistry (Lin et al.,
2010).
2.2. Sampling
In order to sample waters from all potential sources of discharge
along the Houjing River, we established five sampling sites, Jingjian,
Renwu, Demin, Zhuwai and Dehuei. Jingjian, located at the Dashe
upstream reach, receives discharge from the Dashe Industrial Park, an
industrial park mostly populated with mid-stream (resin) petrochemical
factories. Renwu, situated at the Renwu reach, receives discharge from
the Renwu and Zhuhou Industrial Parks. These parks are home to many
large petrochemical (resin production for down-stream plastic manu-
facture) and traditional metal-surface-coating processing centers.
Demin and Zhuwai are situated where they receive discharge from two
points from the Nanzih Export Processing Zone (NEPZ), where many
metal-processing activities (for computer chip packaging) operates
daily along with some electroplating, surface-coating and plastic pro-
duction activities. Companies in the NEPZ have been reported by the
media to discharge illegally large amounts of heavy metal-polluted
acidic wastewater into the Houjing River. Dehuei, located close to Si-
Cing Bu Landfill, is encircled by a residential area with a number of
markets and condominiums, which are sources of pollution introduced
by leachate and illegal discharge. Waters from originating from all
these upstream sources collect at and mix together at Dehuei, making
this site the most prone to heavy metal contamination in the river.
This study used standard methods of water sampling (USEPA, 2013)
and sediment sampling (USEPA, 2014) to collect samples in four
durations, including November of 2015, and January, April and June of
2016. A total of twenty river sediment (10–15 cm from the surface) and
twenty water samples were collected by Ekman Dredge (for sediments)
and bucket (for water) from the middle of the river at an average dis-
tance of 5 m from the river banks. The samples were stored in poly-
ethylene bottles and bags and kept in dark condition below 4 °C until
analysis.
2.3. Instrumental analysis
Heavy metal analysis followed the instruction of US Environmental
Protection Agency (USEPA, 2000). Water samples were first acidized
with HNO3 (Zhang et al., 2016a,b). Subsequently, water samples were
filtered through Whatman glass filter paper (pore size 110 mm) prior to
analysis. Sediment samples (approx. 0.2 g) were lyophilized in a va-
cuum freeze dryer (Eyela FDU-1200, Tokyo Rikakikai, Japan, − 50 °C,
10 Pa, 24 h) and then mixed with acids (3 mL HCl and 1 mL HNO3).
The mixtures were next processed through a microwave digestion
system (Topex, Preekem, Shanghai, China) and subsequently filtered
through Whatman glass filter paper (pore size 90 mm) before being
injected into an Inductively Coupled Plasma − Optical Emission
Spectrometer or ICP-OES (Optima 2100 DV, PerkinElmer, USA) for the
measurement of As, Cd, Cr, Cu, Pb, Ni and Zn (USEPA, 2000). Analysis
for mercury, which only requires the sample be lyophilized first, was
performed in a mercury analyzer (NIC MA-2, Systematic, Taiwan).
Glassware and plastic sample containers for ICP operations were
soaked in 5% HNO3 and dried before use. Likewise, prior to mercury
analysis, ceramic sample containers (ship-shaped) were soaked in 5%
HNO3 and baked in an oven at 500 °C for 18 h. Mercury and ICP
standard solutions were obtained from J.T.Baker (Avantor Performance
Materials, USA) and High-Purity Standards (Charleston, USA), respec-
tively. All reagents were of analytical grade, and deionized water was
used in all analyses. Minimum detection limits (MDLs) were established
using triple standard deviation from the analysis of seven samples with
the same concentration. The MDLs for sediment samples (mg/kg dry
wt.) were as follows: As (0.1), Cd (0.09), Cr (0.1), Cu (0.1), Pb (0.1), Ni
(0.09), Zn (0.3), Hg (0.01). Similarly, the MDLs for water samples (mg/
C.T. Vu et al. Ecological Indicators 82 (2017) 32–42

Fig. 1. Spatial distribution of the sampling sites and major pollution sources for sediment and water heavy metal contamination in the Houjing River, Taiwan (sampled between
November 2015 and June 2016).

n
L) were as follows: As (0.003), Cd (0.005), Cr (0.005), Cu (0.0009), Pb 1
mCd = ∑ Cfi , where Cfi = C i/ Cref
i
(0.002), Ni (0.001), Zn (0.003), Hg (0.0005). Calibration curve was n i=1
ensured with correlation coefficient (R2), which fell into a range of
0.99860–0.99999. Blanks were always less than MDLs. Mid-point where Ci is the heavy metal concentration in sediment and water
i
checks for all metals ranged from 0.3 to 6.8%, except for Hg which was samples; Cref the reference value of the element (Turekian and
11%. Spike recoveries ranged from 96 to 110%. Wedepohl, 1961; Hakanson, 1980) and Cfi the contamination factor of
each element. The modified degree of contamination (mCd) index has
an advantage over single element indices because it takes into account
2.4. Risk assessment methods the synergistic effects of the contaminants at a study site (Brady et al.,
2015).
2.4.1. Modified degree of contamination
Modified degree of contamination (mCd) and Nemerow pollution
index (PN) are two important indicators which can be used to quickly 2.4.2. Nemerow pollution index
and efficiently assess the severity of the contamination of a site. The Nemerow pollution index also reflects the comprehensive ef-
Modified degree of contamination (mCd) is calculated as follows: fects of heavy metals (Yan et al., 2016) and can be used to interpret
heavy metal pollution at a particular site (Duodu et al., 2016). The
equation for calculating Nemerow pollution index (PN) is as follows:

34
C.T. Vu et al. Ecological Indicators 82 (2017) 32–42

Cf 2 + Cfmax
2 profile matrix (p × m) and E the residual matrix (n × m). p is the
PN = number of resulting sources (or factors).
2
The component form of the model is:
where PN is the Nemerow pollution index, Cf the arithmetic mean of
p
contamination factor of all heavy metals, and Cfmax the maximum
contamination factor among the heavy metals.
x ij = ∑ gik fkj + eij i = 1…n; j = 1…m ; k = 1…p
k=1

2.4.3. Sediment quality guidelines and potential ecological risk index
This study compared our finding with different sediment quality
guidelines to more objectively reflect the heavy metal contamination of
the Houjing River. These guidelines were (1) the domestic standard
guidelines of TaiwanEPA (2010) as well as those of (2) the Canadian
Council of Ministers of the Environment (CCME, 1999), (3) the Aus-
tralian and New Zealand Environment and Conservation Council and
Agriculture and Resource Management Council of Australia and New
Zealand (Anzecc and Armcanz, 2000), and (4) the National Oceanic and
Atmospheric Administration (NOAA) (Long et al., 1995). The four
standard guidelines all have lower and upper limits for different heavy
metals. The ecological effects of an individual species can be considered
to be “rarely observed” if its concentration falls below the lower limit
(Lin et al., 2016). Those at upper-limit concentrations are likely to have
more apparent ecological effects, making their effects more “frequently
observed”. It would follow that when concentrations of these metals are
equal to or greater than the lower limit but below the upper limit, their
effects will be “occasionally observed” (Lin et al., 2016).
The potential ecological risk index (RI), proposed by (Hakanson,
1980), was used to evaluate the potential risk of one or multiple ele-
ments to the ecology. The sensitivity of the biological community can
be reflected by RI when it takes the potential ecological risk factor and
the toxicity response coefficient into account. The potential ecological
risk index (RI) is calculated as:
n for the number of factors (or sources) (Comero et al., 2014). The most
RI = ∑ E fi , where E fi = Cfi × T fi
i=1
where Cfi is the contamination factor, T fi the toxicity response coeffi-
cient of each element (As = 10, Cd = 30, Cr = 2, Cu = Pb = Ni = 5,
Zn = 1 and Hg = 40) (Hakanson, 1980; Islam et al., 2015a; Lu et al.,
2015) and E fi the potential ecological risk factor of each element (Yi
et al., 2011).
Classifications of mCd, PN and RI are presented in Table 1.
2.5. Positive matrix factorization (PMF)
PMF uses the weighted least squares method to extract “factors”
from different sources. PMF source apportionment is reliable because it
forces the values to be positive, making them particularly suitable for
environmental applications and making the physical meanings of factor
loading and scores more easily-interpreted (Chen et al., 2013). A dif-
ferentiated dataset of measured concentrations can be described as a
matrix X (n × m) with n samples and m measured species. The two-way
model can be simplified as follows:
X=G×F + E
G is the matrix that represents factor contribution (n × p), F the factor
where xij are the elements of the input data matrix, with each element
showing the measured concentration of the jth species for the ith
sample. gik is the contribution of the kth source to the ith sample, fkj the
portion of the jth species from the kth source, and eij the elements of the
residual concentration of the jth species for the ith sample.
PMF, different from PCA, creates purely constrained non-negative
factor loading and factor scores resulting in less rotational ambiguity.
Weighted least squares are used to fit with the indicated error estimates
of the data matrices’ elements involved in the derivation of the weights.
PMF is designed using an iterative method (conjugate gradient algo-
rithm) to minimize the so-called object function Q:
p 2
m
⎡ x ij − ∑k =1 gik fkj ⎤
Q= ∑ ⎢
sij
⎥
j=1 ⎢
⎣
⎥
⎦
where sij denotes the elements of the uncertainty matrix, each element
being the uncertainty value of the jth species for the ith sample. This
study employs the latest PMF version (PMF 5.0) available on the
website of the US Environmental Protection Agency (USEPA) (https://
www.epa.gov/air-research/positive-matrix-factorization-model-
environmental-data-analyses).
There is a need to test different numbers of factors until the most
reasonable result is reached in order to determine the optimal solution
reasonable result should be obtained when the Q value approximates
the degrees of freedom. If the Q value deviates largely from the theo-
retical value, the number of sources must be readjusted. Input needs
two datasets − the concentrations and the uncertainties of different
MDL
species. The uncertainty is denoted as (0.1 × concentration + 3 ). BDL
data are replaced with MDL , uncertainty was assumed to be 5 * MDL
2 6
(Wang et al., 2016). In this study, two sets of sediment and water heavy
metal measurements (random seed numbers) were run 20 times using
different factors from two to five in order to obtain the most reasonable
result. The element As was excluded from PMF model because it was
found at BDL concentrations at all sites (Comero et al., 2014). The se-
lected number of factors was four for both sediment and water samples
and will be discussed below in results and discussion Section 3.4.
2.6. Statistical analysis
SPSS for Windows, Version 22, was employed for statistical analysis.
Analysis of variance (ANOVA) was used to analyze the difference of
heavy metal content between the sites. Differences were deemed as
significant with 95% confidence interval.

Table 1
Classifications of heavy metal modified degree of contamination (mCd), Nemerow pollution index (PN) and potential ecological risk index (RI).

Class mCd Contamination degree PN Contamination degree RI Ecological risk

0 < 1.5 Unpolluted <1 Unpolluted < 110 Low risk

1 1.5 ≤ Cd < 2 Slightly polluted 1≤ PN < 2.5 Slightly polluted 110 ≤ RI < 200 Moderate risk
2 2 ≤ Cd < 4 Moderately polluted 2.5≤ PN < 7 Moderately polluted 200 ≤ RI < 400 Considerable risk
3 4 ≤ Cd < 8 Moderately-heavily polluted ≥7 Heavily polluted ≥400 Severe risk
4 8 ≤ Cd < 16 Heavily polluted
5 16 ≤ Cd < 32 Severely polluted
6 ≥32 Extremely polluted

35
C.T. Vu et al.
3. Results and discussion
3.1. Heavy metal contamination in sediments and water
Fig. 2 shows the heavy metal content in sediment and water samples
collected from different sites in the Houjing River. In sediments, except
for As (below detection limit or BDL at all sites), the concentrations of
heavy metals were as follows: Ni (BDL) 326 mg/kg dry wt., Zn 620 mg/
kg dry wt., Hg 0.01–0.328 mg/kg dry wt., Pb 15.6–258 mg/kg dry
wt.,Cd 22.2 mg/kg dry wt.,Cr 99.9 mg/kg dry wt., and Cu 2156 mg/kg
dry wt. Most heavy metal content was concentrated at Demin and
Zhuwai, especially Cu and Zn. Renwu was the least heavy metal con-
taminated site, significantly lower than other sites (P < 0.05). Com-
pared with findings of other studies around the world, Cu in our sedi-
ment samples were five to a hundred times higher than those reported
for the Brisbane River (Duodu et al., 2016), the Elbe and Wiser Rivers
(Pache et al., 2008), the Seine River (Le Cloarec et al., 2011) and the To
Lich River (Thuong et al., 2013), indicating serious copper con-
tamination in the Houjing River. Copper commonly exists in the dis-
charge of metal-plating industry, which occupies the majority of NEPZ
and some portions of Dashe and Renwu Industrial Parks (Wang et al.,
2015). Zinc concentrations were high at all Houjing River sites, parti-
cularly at Demin, Zhuwai and Dehuei. Standard deviations from the
mean Cu concentration at Demin and Zhuwai were high probably due
to high temporal variation (Fig. S1). Although Zn concentrations were
comparable to those of Elbe and Wiser Rivers (Pache et al., 2008), they
were two to twenty times higher than those reported by recent studies
of the region, including those of the Nakdong River (Chung et al.,
2016), Buriganga River (Bhuiyan et al., 2015), Bortala River (Zhang
et al., 2016c), Yellow River (Yan et al., 2016) and Vaigai River
36
Ecological Indicators 82 (2017) 32–42
Fig. 2. Heavy metal concentrations in (a) sediments
(mg/kg dry wt.) and (b) water (mg/L) (mean ± SD)
in the Houjing River, Taiwan (sampled between
November 2015 and June 2016).
(Paramasivam et al., 2015). Zinc, in our study, could be discharged
from a variety of sources such as commercial activities, traffic, atmo-
spheric deposition, etc., which will later be discussed in the PMF
modeling results. Sediment concentrations of Hg in the Houjing River
were one order of magnitude higher than those found in the Bortala
River (Zhang et al., 2016c) and the Jinjiang River (Chen et al., 2013).
Similar to Zn, Hg was also released from the commercial activities and
traffic along the riverbanks and will be discussed in PMF modeling
result part.
There have not been many studies of heavy metal content in water
samples, possibly because heavy metals precipitate quickly and bind
strongly to sediments (Simpson and Batley, 2007; Simpson and
Spadaro, 2016). It is possible for metals to detach from sediments and
be released back to the water column, negatively impacting water
quality (Simpson and Batley, 2007). The element As in the Houjing
River water samples, like the sediment samples, was below detection
limit at all sites. Other heavy metal concentrations in water were as
follows: Ni (BDL) 0.082 mg/L, Zn 0.011 mg/ L, Hg 0.018 mg/ L, Pb
0.569 mg/ L, Cd 0.008 mg/ L, Cr 0.096 mg/ L, and Cu 0.99 mg/ L.
Demin, Zhuwai and Dehuei were the most severely contaminated sites,
significantly higher than the other two sites (P < 0.05). Cu and Pb
concentrations, which were highest at all sites, also had high standard
deviations due to temporal variation. Compared with reports from
other studies, the concentration of Cu in the Houjing River water was
similar to that of the Buriganga River (Bhuiyan et al., 2015) but two
orders of magnitude higher than that of the Hindon River (Suthar et al.,
2009). Pb concentrations were two times higher than those of the
Hindon River (Suthar et al., 2009). Cadmium, chromium, zinc and
mercury in the Houjing were found at one to two orders of magnitude
lower concentrations than those of the Buriganga (Bhuiyan et al., 2015)
C.T. Vu et al. Ecological Indicators 82 (2017) 32–42

Table 2
Contamination factor and potential ecological risk factor of heavy metals in water categorized for aquatic life and irrigation in the Houjing River, Taiwan.

Elements Average water heavy metal content AWPLa Aquatic life Irrigation
(mg/L)
C if A Grade of contamination E if A Grade of ecological IWPLb C if I Grade of contamination
risk

As ND 50 – – – – 100 – –
Cd 0.002 1.8 0.00003 Low 0.0003 Low 10 0.00002 Low
Cr 0.035 2 0.00070 Low 0.0070 Low 100 0.00035 Low
Cu 0.401 4 0.00802 Low 0.0802 Low 200 0.00401 Low
Pb 0.475 7 0.00951 Low 0.0951 Low 200 0.00475 Low
Ni 0.128 150 0.00256 Low 0.0256 Low 200 0.00128 Low
Zn 0.002 30 0.00004 Low 0.0004 Low 1000 0.00002 Low
Hg 0.022 0.1 0.00044 Low 0.0044 Low 2 0.00022 Low

a
Aquatic life water permissible limits (values in mg/L) (CCME, 2007).
b
Irrigation life water permissible limits (CCME, 2007), except Ni (NRMMCA, 2011) and Hg (values in mg/L) (Anzecc and Armcanz, 2000).

and Hindon Rivers (Suthar et al., 2009).
3.2. Ecological risk assessment for heavy metals in water
While heavy metal ecological risk assessment has been fairly well
established for sediments, it has not been so for water, the most
common method used being target hazard quotient and target cancer
risk (El Nemr et al., 2016). Recently, however, Sharifi et al. (2016)
calculated the contamination factor and potential ecological risk factor
i
for heavy metals in water. In this study, Cref were water quality indices
of heavy metals proposed by different guidelines, including those of
Canadian Council of Ministers of the Environment (CCME, 2007), Na-
tional Resource Management Ministerial Council, Commonwealth of
Australia (NRMMCA 2011), and Australian and New Zealand En-
vironment and Conservation Council & Agriculture and Resource Man-
agement Council of Australia and New Zealand (Anzecc and Armcanz,
2000). Our study used water quality reference values for aquatic life
and irrigation because the Houjing River was previously the main water
supply for a large area involved in agriculture and aquaculture and is
still used at such in some areas.
Table 2 summarizes our heavy metal contamination and ecological
risk calculation findings for water in the Houjing. Our findings were one
to four orders of magnitude lower than the already low reference values
in the guidelines, suggesting that heavy metal contamination and eco-
logical risk needed not be a major concern for future pollution control
and management plans for the river, though there is still a need for
heavy metal monitoring to prevent unnecessary risk. Changes in hy-
draulic conditions can at times lead to the release of heavy metals from
sediments to the water column.
3.3. Heavy metal contamination and its ecological risk in sediments
3.3.1. Comparison with sediment quality guidelines (SQG)
According to El Nemr et al., the most classical approach to assessing
heavy metal contamination is to compare present findings with re-
ference values found in sediment quality guidelines (El Nemr et al.,
2016). This approach, still used today, is very effective when used to
assess heavy metal contamination in sediments (Duodu et al., 2016; Lin
et al., 2016). The periodically updating of the reference values not only
facilitates the continued use of the SQG approach but also shows that
the field considers this approach reliable. Generally, the SQGs present
two limits, one marking the frequency and the other the severity of
ecological impacts.
Table 3 compares our findings with the reference values established
in different SQGs. The reference values established by the Australian
and New Zealand Environment and Conservation Council & Agriculture
and Resource Management Council of Australia and New Zealand
(Anzecc and Armcanz, 2000) are similar to those of the National
Oceanic and Atmospheric Administration (Long et al., 1995). Those
established by the Taiwan EPA (TaiwanEPA, 2010) and the Canadian
Council of Ministers of the Environment (CCME, 1999) are similar to
each other. In general, the values of the former two are somewhat
higher than the latter two, indicating that the latter two represent more
stringent values for SQG approach. The average concentration of Cd
was around 1.5 times higher in our study than the reference risk values
found in Taiwan EPA’s upper limit and CCME’s probable effect level
(PEL) but lower than ANZECC & ARMCANZ’s high level and NOAA’s
effects range-median (ERM). Similarly, chromium in our study ex-
ceeded CCME’s ISQG, but it was lower than the Taiwan EPA’s lower
limit, ANZECC & ARMCANZ’s low level and NOAA’s effects range-low
(ERL). Copper represented the worst heavy metal contamination in the
Houjing River, as its concentrations in sediments far exceeded (about
two times) any upper risk values of the referenced SQGs, which is un-
derstandable since its concentrations were severe, as seen in 3.1 above.
Lead concentrations in this study fell between the lower and upper
values of all the guidelines. Nickel greatly exceeded ANZEC-
C & ARMCANZ’s high level and NOAA’s ERM, but was lower than the
Taiwan EPA’s upper limit. Zinc fell between the lower and upper values
of the Taiwan EPA, ANZECC & ARMCANZ and NOAA, but slightly ex-
ceeded CCME’s PEL. Finally, Hg was lower in our study than any of the
values listed in the other guidelines. In summary, except for As and Hg,
average concentrations of the heavy metals in this study exceeded the
reference limits in the different SQGs, suggesting that heavy metal
contamination is severe in sediments of the Houjing River.
3.3.2. Modified degree of contamination and nemerow pollution index
In the environment, a number of heavy metals can co-exist and their
negative impacts often result from the combined effects between them
(Brady et al., 2015). Single-element indices do not adequately reflect
their synergistic effect (Duodu et al., 2016). Because the current study is
mainly concerned with the combined effects of heavy metals, we also
wanted to analyze our data using the multi-element indices known as
the modified degree of contamination (mCd) and Nemerow pollution
index (PN). These two indices describe the synergistic effect of different
contaminants at a site (Duodu et al., 2016; Yan et al., 2016).
Fig. 3 shows the results of the mCd index and PN index calculations
at the sampling sites. Although the calculation and classifications of the
two indices differ, their results are uncannily similar. The sampling sites
at Jingjian and Dehuei, which were be equally impacted by heavy
metals, were found to be much less severely contaminated than Renwu,
Demin and Zhuwai, which explains the reason the two sites were
classified as ‘Class 2′ in both mCd and PN calculations. While the Jing-
jian site was mostly affected by Cu, Cd and Zn, the heavy contamination
at the Dehuei site came mainly from Cu and Zn (Table S1). Demin and
Zhuwai were the most impacted by heavy metals, classified as ‘3′ in PN
and ‘4′ in mCd. The element contributing most to mCd at Demin and
Zhuwai was Cu, followed by Zn and Ni.
Renwu, polluted mainly by Cd and Pb, was categorized as ‘Class 3′

37
C.T. Vu et al. Ecological Indicators 82 (2017) 32–42

Table 3
Comparison between data of this study (the Houjing River, Taiwan) and sediment quality guidelines (SQG) worldwide (values in mg/kg).

e f
Elements Average Taiwan EPA Taiwan EPA CCME CCME ANZECC & ARMCANZ low ANZECC & ARMCANZ high NOAA NOAA
current data lower limit b upper limit a ISQG c PEL d ERL g ERM h

As ND 11 33 5.9 17 20 70 8.2 70
Cd 4.39 0.65 2.5 0.6 3.5 1.5 10 1.2 9.6
Cr 53.11 76 233 37.3 90 80 370 81 370
Cu 432.29 50 157 35.7 197 65 270 34 270
Pb 57.23 48 161 35 91.3 50 220 46.7 218
Ni 71.17 24 80 – – 21 52 20.9 51.6
Zn 341.88 14 384 123 315 200 410 150 410
Hg 0.10 0.23 0.87 0.17 0.486 0.15 1 0.15 0.71

a,b
Taiwan Environmental Protection Agency (Taiwan EPA)’s sediment quality guideline upper and lower limits (TaiwanEPA, 2010).
c,d
Canadian Council of Ministers of the Environment (CCME)’s interim sediment quality guideline (ISQG) and probable effect level (PEL) (CCME, 1999).
e,f
Australian and New Zealand Environment and Conservation Council (ANZECC) & Agriculture and Resource Management Council of Australia and New Zealand (ARMCANZ)’s low and
high levels (Anzecc and Armcanz, 2000).
g,h
National Oceanic and Atmospheric Administration (NOAA)’s effects range-low (ERL) and effects range-median (ERM) (Long et al., 1995).

in the mCd calculation and different from Demin and Zhuwai. However,
all three sites were grouped into one class in PN calculations. This dif-
ference was due to the difference in specificity of the indices’ two
classification systems. Because mCd classification is more specific,
heavy metal contamination was better categorized. The categorization
of heavy metal contamination based on mCd and PN calculations agreed
relatively well with the intensity of heavy metal concentrations dis-
cussed in 3.1, where Demin and Zhuwai were found to have the highest
accumulation of heavy metals.
3.3.3. Ecological risk assessment
This study also employed the potential ecological risk index (RI),
another multi-element index, to assess the potential impact of heavy
metal contamination to the ecosystem of in the Houjing River (Fig. 4).
While Dehuei and Jingjian were less exposed to ecological risk of heavy
contamination (Class 1 and Class 2, respectively), the other three sites

Fig. 3. Multi-element indices of heavy metals in se-

diments: (a) modified degree of contamination (mCd)
and (b) Nemerow pollution index (PN) in the Houjing
River, Taiwan.

38
C.T. Vu et al.
were determined to be likely to face ‘severe’ ecological risk (all Class 3),
all in accordance with mCd and PN calculations. Renwu, however, was
found most likely to be exposed to ecological risk, illustrating that
degree of contamination does not always represent potential for eco-
logical risk. Therefore, using both degree of contamination degree and
potential ecological risk in a single study is reasonable and encouraged.
Renwu faces ecological risk posed by Cd mostly and to some extent
Hg (Table S2). Demin faces major ecological risk posed by Cu and Hg,
while Zhuwai faces ecological risk from Cd. Jingjian mostly faces risk
from Cd and Cu, and Dehuei from Hg. Therefore, it seems that Cu, Hg,
and Cd pose the greatest ecological risk to the Houjing River. As seen in
Fig. 5, three elements account for 90.17% of the potential ecological
risk. Cd impacts largely Renwu and Jingjian, making up 45.84% of the
total risk followed by Cu. Cu makes up 30.06% of total risk at Demin
and Zhuwai. Hg poses the biggest threat to Demin and constitutes
14.27% of the total potential ecological risk. Zn, found in relatively
high concentrations at all sites, does not pose high risk to the eco-
system, while Cd and Hg, found at relatively low concentrations in
sediments, pose high risks to the ecosystem.
3.4. Source apportionment by positive matrix factorization (PMF)
PMF allows us to determine how much each source contributes to
study site. To find out whether the PMF model is well fit for different
elements, we compared observed values with their predicted heavy
metal values. The correlation coefficients (R2) between the observed
and predicted values in both sediments and water were all 0.99 (Fig.
S2). Good correlations were also found between their observed and
predicted values in sediments and water, 0.95–0.99 and 0.83–0.99,
respectively (Table S3). Together, these good positive correlations show
Fig. 5. Contribution of different species to ecological risk index (RI) of heavy metals in
sediments in the Houjing River, Taiwan.
39
Ecological Indicators 82 (2017) 32–42
Fig. 4. Ecological risk index (RI) of heavy metals in
sediments in the Houjing River, Taiwan.
the PMF model to be reliable.
Heavy metals in sediments and water have various negative effects
on the ecological and human wellbeing. Source apportionment can
provide sound results that can help guide future pollution control and
management plans. In our PMF models, four factors were identified for
both heavy metals in sediments and water in the Houjing River.
3.4.1. PMF results for heavy metals in sediments
Fig. 6a and Fig. S3a show source apportionment results and the
factor contribution percentage for heavy metals in sediments, respec-
tively. As can be seen in Fig. 6a, there were discrepancies in the con-
centrations and percentages of heavy metals occupying factor 1. For
example, a low concentration of Hg accounted for more than 60%,
while a high concentration of Pb made up a percentage of lower than
10%. This pattern was repeated for other metals in other factors.
Whether a particular metal shows a high concentration within a specific
factor is not as important as whether that metal occupies a large per-
centage within that factor. This means that the factor is more likely to
be signified by the metal’s percentage, though it cannot be denied that
the higher the concentration, the more likely it is to make up a much
higher percentage. Thus, a graph produced by PMF showing both heavy
metal concentrations and percentages helps greatly with explaining and
discussing source apportionment.
Factor 1, which explains only 3.7% of the contribution, has higher
loadings of Hg (67.5%) and Zn (53.8%) than that of Cr (42.4%). Factor
1 is likely to originate from commercial activities, waste tires, atmo-
spheric deposition, and from a scattering of small businesses, etc.
(Bhuiyan et al., 2015). With many condominiums and bustling com-
mercial and small business activities alongside the river, there are nu-
merous sources of municipal discharge into the river, meaning the need
for control of heavy metal discharge from this source should be great.
However, because the factor only constitutes a tiny amount of the total
heavy metal contamination, it should be given less weight in future
pollution control plans.
Factor 2, constituting a moderate proportion (24.0%) of the con-
tribution, correlated well with Cd (96.5%). This metal signifies che-
mical and petrochemical discharge (Cechinel et al., 2016). Cadmium is
well known to be used in material plastic stabilizers (cadmium oxide)
(Lu et al., 2007). In this study, the sampling sites near Dashe and Renwu
Industrial Parks, home to a variety of mid-stream plastic industries,
were found to be seriously contaminated with cadmium. As factor 2 is
filled mostly by Cd, it is reasonable to assume that in this factor the
source of cadmium discharge is the plastics industry.
Factor 3, which accounted for a very large portion (36.5%) of the
contribution, was composed mainly of relatively equal proportions of
Cr (42.7%), Zn (46.2%), and Pb (49.9%). These metals are typically
C.T. Vu et al.
Fig. 6. Results of PMF source apportionment modeling for heavy metals in (a) sediments and (b) water in the Houjing River, Taiwan.
found in traditional electroplating and metal-surface-processing in-
dustries, which have played an important role in Taiwan’s economic
development (Wang et al., 2015). Industrial plants all over the island
are home to metal-related industries at least in part. Along the Houjing
River, electroplating and metal-surface-processing firms occupy the
entire Zhuhou Industrial Park and a small percentage of the firms in the
Nanzih Export Processing Zone (NEPZ). Therefore, factor 3 should be
40
Ecological Indicators 82 (2017) 32–42
attributed to the heavy metal contamination caused by traditional
electroplating and metal-surface-processing industries, which should be
targeted in future environmental monitoring plans.
Factor 4, accounting for 35.8% of the contribution, consists mostly
of Cu (77.5%) and moderate portions of Ni (43.1%). Cu and Ni are used
at many different stages of metal-related processes employed in com-
puter chip packaging. This industry has thrived in recent decades in
C.T. Vu et al.
Taiwan where there has been a great increase in these factories and
companies in the large industrial plants around the island. A large
proportion of factories in the NEPZ are involved in high-tech semi-
conductor packaging activities. Moreover, historically, plants involved
in these activities have been reported by the media to discharge illeg-
ally untreated wastewater with heavy metals into the Houjing River.
Thus, factor 4 could be the source of heavy metals discharged from
NEPZ, which should also be targeted in future environmental mon-
itoring and pollution prevention plans for the river.
3.4.2. PMF results for heavy metals in water
Fig. 6b and Fig. S3b display the results of PMF modeling and the
percentage of contribution of each factor for heavy metals in water,
respectively. The three first factors of water heavy metals seem to be
distributed evenly. Factor 1, which made up 31.9% of the contribution,
contains high loadings of Cr (77.7%) and Ni (48%). These two elements
are likely to indicate discharge from traditional electroplating and other
metal-processing industries (Lee et al., 2017). The electroplating in-
dustry has brought about much of Kaohsiung City’s economic devel-
opment, and such metal-processing factories make up most of the fac-
tories in industrial plants there. Cr and Ni are used to create different
surface properties in the plating industry. Hence, a large amount of Cr
and Ni can be observed in water of a seriously industrially con-
taminated river like the Houjing River. This industry and its discharge
should be a target for pollution control and management plans in the
future.
Factor 2, which made up 31.6% of the contribution, had slightly
equal loadings of Cd (34.7%) and Zn (34.6%). As discussed above, Cd is
widely used in the plastics industry. Zinc, as a pigment, is also em-
ployed for plastic production (Gakwisiri et al., 2012). Plastic industries
make up much of the industry found in Dashe and Renwu Industrial
Parks. Since there is always wastewater from these industries dis-
charged into the river, it is reasonable to assume that they are the
source of much of the Cd and Zn in water in the Houjing River. Future
pollution control plans will also want to focus on these industries when
trying to mitigate the potential negative effects of the metals on the
ecology.
Factor 3, which makes up only 27.5% of the contribution, is com-
posed essentially of Cu (76.7%). Semi-conductor packaging produces
large amounts of Cu discharge in the form of copper-containing wastes
and drainage (Mirbagheri and Hosseini, 2005). Copper is also used in
plating baths as it has outstanding adhesion, corrosion and resistance
qualities (Algarra et al., 2005). Semi-conductor packaging plants make
up a large proportion of the companies in the NEPZ, so wastewater from
that industrial park would have high amounts of copper. As Cu fre-
quently appears in high amounts at the sampling sites near the NEPZ,
this element and the NEPZ should be the targets for future pollution
control plans.
Factor 4, which accounted for 9.0% of the contribution, mainly
consisted of slightly equal proportions of Ni (52%), Hg (47.9%) and Pb
(47.7%). This factor is attributed to municipal discharge for the same
reasons as mentioned above. However, the municipal sources are too
scattered to be identified. The Houjing River’s basin is 21 km long and
encompasses a total area of approximately 70 square km. Thousands of
daily commercial and small industrial activities along the riverbank
occur there. It is very difficult to pin point the main sources that con-
tribute to these metals in municipal discharge. However, certain types
of domestic wastewater from large condominiums, and the waste and
wastewater from night markets should be considered.
4. Conclusions
Based on our SQG, mCd, PN and RI findings, the heavy metal con-
centrations in the water of the Houjing River are less likely to present
high risk to the ecosystem, while those in the sediment do. Our com-
parisons with different sediment quality guidelines reveal that almost
41
Ecological Indicators 82 (2017) 32–42
all the elements we investigated exceeded standard values, indicating
‘moderately’ to ‘frequently observed’ threats from heavy metal con-
tamination of the Houjing River. The results of our calculations of mCd,
PN and RI, multi-element indices use to reflect synergistic effects,
agreed well with each other in the determination of ‘heavy’ con-
tamination and ‘severe’ ecological risk at Demin, Zhuwai and Renwu.
Cadmium and copper pose the greatest potential ecological risk to the
river. Positive matrix factorization modelling revealed four factors or
pollution sources for heavy metals in both sediments and water of the
Houjing River. These were industrial activities, especially traditional
metal-plating, plastics, and semi-conductor packaging, as well as mu-
nicipal discharge from densely residential and commercial areas along
the riverbank. Future pollution control and management plans should
focus on discharge from these industries situated alongside the river.
Acknowledgements
The authors acknowledge the enthusiastic assistance of the sam-
pling and analysis staff at the Center of Environmental Analysis Services
(CEAS), National Kaohsiung Marine University.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.ecolind.2017.06.008.
References
Algarra, M., Jimenez, M.V., Rodriguez-Castellon, E., Jimenez-Lopez, A., Jimenez-
Jimenez, J., 2005. Heavy metals removal from electroplating wastewater by ami-
nopropyl-Si MCM-41. Chemosphere 59 (6), 779–786.
Anttila, P., Paatero, P., Tapper, U., Jarvinen, O., 1995. Source identification of bulk wet
deposition in Finland by positive matrix factorization. Atmos. Environ. 29 (14),
1705–1718.
Anzecc, Armcanz, 2000. Australian and New Zealand Guidelines for Fresh and Marine
Water Quality. Australian and New Zealand Environment and Conservation
Council & Agriculture and Resource Management Council of Australia and New
Zealand, Australia. https://www.environment.gov.au/system/files/resources/
e080174c-b267-455e-a8db-d3f79e3b2142/files/nwqms-guidelines-4-vol3.pdf
(Accessed 26, March 2017).
Bhuiyan, M.A.H., Dampare, S.B., Suzuki, M.A.I.S., 2015. Source apportionment and
pollution evaluation of heavy metals in water and sediments of Buriganga River,
Bangladesh, using multivariate analysis and pollution evaluation indices. Environ.
Monit. Assess. 187 (1).
Brady, J.P., Ayoko, G.A., Martens, W.N., Goonetilleke, A., 2015. Development of a hybrid
pollution index for heavy metals in marine and estuarine sediments. Environ. Monit.
Assess. 187 (5).
Brown, S.G., Eberly, S., Paatero, P., Norris, G.A., 2015. Methods for estimating un-
certainty in PMF solutions: examples with ambient air and water quality data and
guidance on reporting PMF results. Sci. Total Environ. 518, 626–635.
CCME, 1999. Canadian Sediment Quality Guidelines for the Protection of Aquatic Life.
Canadian Council of Ministers of the Environment, Winnipeg, Canada. http://www.
ccme.ca/en/resources/canadian_environmental_quality_guidelines/ (Accessed 26,
March 2017).
CCME, 2007. Canadian Water Quality Guidelines for the Protection of Aquatic Life:
Summary Table. Canadian Council of Ministers of the Environment, Canada. http://
www.ccme.ca/en/resources/canadian_environmental_quality_guidelines/ (Accessed
26, March 2017).
Cechinel, M.A.P., Mayer, D.A., Pozdniakova, T.A., Mazur, L.P., Boaventura, R.A.R., De
Souza, A.A.U., De Souza, S.M.A.G.U., Vilar, V.J.P., 2016. Removal of metal ions from
a petrochemical wastewater using brown macro-algae as natural cation-exchangers.
Chem. Eng. J. 286, 1–15.
Chen, H.Y., Teng, Y.G., Wang, J.S., Song, L.T., Zuo, R., 2013. Source apportionment of
trace element pollution in surface sediments using positive matrix factorization
combined support vector machines: application to the jinjiang river, China. Biol.
Trace Elem. Res. 151 (3), 462–470.
Chung, S.Y., Venkatramanan, S., Park, N., Ramkumar, T., Sujitha, S.B., Jonathan, M.P.,
2016. Evaluation of physico-chemical parameters in water and total heavy metals in
sediments at Nakdong River Basin, Korea. Environ. Earth Sci. 75 (1).
Comero, S., Locoro, G., Free, G., Vaccaro, S., De Capitani, L., Gawlik, B.M., 2011.
Characterisation of Alpine lake sediments using multivariate statistical techniques.
Chemometr. Intell. Lab. 107 (1), 24–30.
Comero, S., Vaccaro, S., Locoro, G., De Capitani, L., Gawlik, B.M., 2014. Characterization
of the Danube River sediments using the PMF multivariate approach. Chemosphere
95, 329–335.
Duodu, G.O., Goonetilleke, A., Ayoko, G.A., 2016. Comparison of pollution indices for the
assessment of heavy metal in Brisbane River sediment. Environ. Pollut. 219,
C.T. Vu et al.
1077–1091.
Duodu, G.O., Ogogo, K.N., Mummullage, S., Harden, F., Goonetilleke, A., Ayoko, G.A.,
2017. Source apportionment and risk assessment of PAHs in Brisbane River sediment,
Australia. Ecol. Indic. 73, 784–799.
El Nemr, A., El-Said, G.F., Ragab, S., Khaled, A., El-Sikaily, A., 2016. The distribution,
contamination and risk assessment of heavy metals in sediment and shellfish from the
Red Sea coast, Egypt. Chemosphere 165, 369–380.
Gakwisiri, C., Raut, N., Al-Saadi, A., Al-Aisri, S., Al-Ajmi, A., 2012. A critical review of
removal of zinc from wastewater. In: Proceedings of the World Congress on
Engineering. July 4–6. London, U.K..
Hakanson, L., 1980. An ecological risk index for aquatic pollution control.a sedimento-
logical approach. Water Res. 14 (8), 975–1001.
Islam, M.S., Ahmed, M.K., Habibullah-Al-Mamun, M., Hoque, M.F., 2015a. Preliminary
assessment of heavy metal contamination in surface sediments from a river in
Bangladesh. Environ Earth Sci. 73 (4), 1837–1848.
Islam, M.S., Ahmed, M.K., Raknuzzaman, M., -Al-, Habibullah, Mamun, M., Islam, M.K.,
2015b. Heavy metal pollution in surface water and sediment: a preliminary assess-
ment of an urban river in a developing country. Ecol. Indic. 48, 282–291.
Jiang, J.J., Lee, C.L., Fang, M.D., Boyd, K.G., Gibb, S.W., 2015. Source apportionment and
risk assessment of emerging contaminants: an approach of pharmaco-signature in
water systems. PLoS One 10 (4).
Le Cloarec, M.F., Bonte, P.H., Lestel, L., Lefevre, I., Ayrault, S., 2011. Sedimentary record
of metal contamination in the Seine River during the last century. Phys. Chem. Earth
36 (12), 515–529.
Lee, C.G., Lee, S., Park, J.A., Park, C., Lee, S.J., Kim, S.B., An, B., Yun, S.T., Lee, S.H., Choi,
J.W., 2017. Removal of copper, nickel and chromium mixtures from metal plating
wastewater by adsorption with modified carbon foam. Chemosphere 166, 203–211.
Lin, C., Lee, C.J., Mao, W.M., Nadim, F., 2009. Identifying the potential sources of di-(2-
ethylhexyl) phthalate contamination in the sediment of the Houjing River in southern
Taiwan. J. Hazard. Mater. 161 (1), 270–275.
Lin, C.E., Kao, C.M., Jou, C.J., Lai, Y.C., Wu, C.Y., Liang, S.H., 2010. Preliminary iden-
tification of watershed management strategies for the Houjing river in Taiwan. Water
Sci. Technol. 62 (7), 1667–1675.
Lin, Q., Liu, E., Zhang, E., Li, K., Shen, J., 2016. Spatial distribution, contamination and
ecological risk assessment of heavy metals in surface sediments of Erhai Lake, a large
eutrophic plateau lake in southwest China. Catena 145, 193–203.
Long, E.R., Macdonald, D.D., Smith, S.L., Calder, F.D., 1995. Incidence of adverse bio-
logical effects within ranges of chemical concentrations in marine and estuarine se-
diments. Environ. Manage. 19 (1), 81–97.
Lu, L.T., Chang, I.C., Hsiao, T.Y., Yu, Y.H., Ma, H.W., 2007. Identification of pollution
source of cadmium in soil − application of material flow analysis and a case study in
Taiwan. Environ. Sci. Pollut. Res. Int. 14 (1), 49–59.
Lu, S., Wang, Y., Teng, Y., Yu, X., 2015. Heavy metal pollution and ecological risk as-
sessment of the paddy soils near a zinc-lead mining area in Hunan. Environ. Monit.
Assess. 187 (10), 1–12.
Mirbagheri, S.A., Hosseini, S.N., 2005. Pilot plant investigation on petrochemical was-
tewater treatment for the removal of copper and chromium with the objective of
reuse. Desalination 171 (1), 85–93.
NRMMCA, 2011. National Resource Management Ministerial Council. National Resource
Management Ministerial Council, Commonwealth of Australia, Canberra, Australia.
https://www.nhmrc.gov.au/guidelines-publications/eh52 (Accessed 26, March
2017).
Pache, T., Brockamp, O., Clauer, N., 2008. Varied pathways of river-borne clay minerals
in a near-shore marine region: a case study of sediments from the Elbe- and Weser
rivers, and the SE North Sea. Estuar Coast Shelf Sci. 78 (3), 563–575.
Paramasivam, K., Ramasamy, V., Suresh, G., 2015. Impact of sediment characteristics on
the heavy metal concentration and their ecological risk level of surface sediments of
Vaigai river Tamilnadu, India. Spectrochim. Acta A 137, 397–407.
Pekey, H., Dogan, G., 2013. Application of positive matrix factorisation for the source
42
Ecological Indicators 82 (2017) 32–42
apportionment of heavy metals in sediments: a comparison with a previous factor
analysis study. Microchem. J. 106, 233–237.
Praipipat, P., Rodenburg, L.A., Cavallo, G.J., 2013. Source apportionment of poly-
chlorinated biphenyls in the sediments of the delaware river. Environ. Sci. Technol.
47 (9), 4277–4283.
Sharifi, Z., Hossaini, S.M.T., Renella, G., 2016. Risk assessment for sediment and stream
water polluted by heavy metals released by a municipal solid waste composting plant.
J. Geochem. Explor. 169, 202–210.
Simpson, S.L., Batley, G.E., 2007. Predicting metal toxicity in sediments: a critique of
current approaches. Integr. Environ. Assess. Manag. 3 (1), 18–31.
Simpson, S.L., Spadaro, D.A., 2016. Bioavailability and chronic toxicity of metal sulfide
minerals to benthic marine invertebrates: implications for deep sea exploration,
mining and tailings disposal. Environ. Sci. Technol. 50 (7), 4061–4070.
Suthar, S., Nema, A.K., Chabukdhara, M., Gupta, S.K., 2009. Assessment of metals in
water and sediments of Hindon River, India: impact of industrial and urban dis-
charges. J. Hazard. Mater. 171 (1–3), 1088–1095.
TaiwanEPA, 2010. Soil and Groundwater Pollution Remediation Act. Taiwan
Environmental Protection Administration, Taipei, Taiwan. https://sgw.epa.gov.tw/
(Accessed 26, March 2017).
Tessier, A., Campbell, P.G.C., 1987. Partitioning of trace metals in sediments: relation-
ships with bioavailability. Hydrobiologia 149 (1), 43–52.
Thuong, N.T., Yoneda, M., Ikegami, M., Takakura, M., 2013. Source discrimination of
heavy metals in sediment and water of To Lich River in Hanoi City using multivariate
statistical approaches. Environ. Monit. Assess. 185 (10), 8065–8075.
Turekian, K., Wedepohl, K., 1961. Distribution of the elements in some major units of the
earth's crust. Geol. Soc. Am. Bull. 72 (2), 175–192.
USEPA, 2000. Method 6010C − Inductively Coupled Plasma-Atomic Emission
Spectrometry. United States Environmental Protection Agency, Washington, DC,
USA. https://www.epa.gov/sites/production/files/2015-07/documents/epa-6010c.
pdf.
USEPA, 2013. Surface Water Sampling. United States Environmental Protection Agency,
Washington, DC, USA. https://www.epa.gov/sites/production/files/2015-06/
documents/Surfacewater-Sampling.pdf (Accessed 26, March 2017).
USEPA, 2014. Sediment Sampling. United States Environmental Protection Agency,
Washington, DC, USA. https://www.epa.gov/sites/production/files/2015-06/
documents/Sediment-Sampling.pdf (Accessed 26, March 2017).
Villanueva, M.C., Ibarra, A.A., 2016. Assessing the ecological stress in a Garonne River
stretch, southwest France. Ecol. Indic. 67, 466–473.
Wang, Y.B., Liu, C.W., Wang, S.W., 2015. Characterization of heavy-metal-contaminated
sediment by using unsupervised multivariate techniques and health risk assessment.
Ecotox Environ Safe. 113, 469–476.
Wang, G., Cheng, S.Y., Wei, W., Zhou, Y., Yao, S., Zhang, H.Y., 2016. Characteristics and
source apportionment of VOCs in the suburban area of Beijing, China. Atmos. Pollut.
Res. 7 (4), 711–724.
Yan, N., Liu, W.B., Xie, H.T., Gao, L.R., Han, Y., Wang, M.J., Li, H.F., 2016. Distribution
and assessment of heavy metals in the surface sediment of Yellow River, China. J.
Environ. Sci-China 39, 45–51.
Yi, Y.J., Yang, Z.F., Zhang, S.H., 2011. Ecological risk assessment of heavy metals in
sediment and human health risk assessment of heavy metals in fishes in the middle
and lower reaches of the Yangtze River basin. Environ. Pollut. 159 (10), 2575–2585.
Zhang, Z.Q., Wang, J., Ali, A., Delaune, R.D., 2016a. Heavy metal distribution and water
quality characterization of water bodies in Louisiana's Lake Pontchartrain Basin, USA.
Environ. Monit. Assess. 188 (11).
Zhang, Z.Q., Wang, J.J., Ali, A., Delaune, R.D., 2016b. Heavy metals and metalloid
contamination in Louisiana Lake Pontchartrain Eestuary along I-10 bridge. Transport.
Res. D Trans. Environ. 44, 66–77.
Zhang, Z.Y., Li, J.Y., Mamat, Z., Ye, Q.F., 2016c. Sources identification and pollution
evaluation of heavy metals in the surface sediments of Bortala River, Northwest
China. Ecotox. Environ. Safe. 126, 94–101.