Stochastic Environmental Research and Risk Assessment

https://doi.org/10.1007/s00477-019-01741-8 (0123456789().,-volV)(0123456789().
,- volV)

ORIGINAL PAPER

Risk assessment and source identification of heavy metals

in agricultural soil: a case study in the coastal city of Zhejiang
Province, China
Zhouqiao Ren1,2 • Rui Xiao3 • Zhonghao Zhang4 • Xiaonan Lv1,2 • Xufeng Fei1,2

 Springer-Verlag GmbH Germany, part of Springer Nature 2019

Abstract
Heavy metal contamination is a serious environmental problem, especially in developing countries such as China. In this
study, we collected 1928 soil samples from the southeastern coastal area of China and analyzed the pollution concentration
and potential ecological risk from heavy metals including arsenic (As), cadmium (Cd), chromium (Cr), lead (Pb), and
mercury (Hg). The mean concentrations of Cr, Hg, and Pb were lower than their corresponding background values,
whereas As and Cd were 1.31 and 1.59 times their background values, respectively. The calculation of the mean Pollution
Index (PI) for these heavy metals were, in decreasing order Cd (1.59), As (1.31), Cr (0.94), Pb (0.89), and Hg (0.78) and the
Nemerow Integrated Pollution Index revealed that almost one-fifth of the soil in the study area was moderately polluted.
According to the ecological risk index, about 12% of the soil was at a moderate or high ecological risk, and Cd and Hg
presented the highest ecological risk. The GeogDetector software was used to quantitatively assess the potential sources of
these metals. The GeogDetector results showed that the soil heavy metals have various sources, including: natural
processes had significant impacts on all heavy metals analyzed in this study; farmland types influenced the concentrations
of As and Cr significantly; industrial activities significantly increased As, Cr, and Hg; transportation-related activities
increased As, Cd, and Hg; and agricultural application of fertilizer and pesticides, had significant impacts on As, Cd, and
Pb levels. Based on the results of the interaction detector, natural processes and agricultural activities were determined to
be the main sources of heavy metals in the study area.

Keywords Heavy metal  Source analysis  Pollution assessment  GeogDetector model

1 Introduction

Heavy metal contamination in soil has become an impor-

tant global environmental concern in recent decades and
has thus received significant attention (Chen et al. 2009;
Islam et al. 2016; Jiang et al. 2017a). As a developing
& Xufeng Fei
[email protected] country with rapid urbanization and industrialization,
heavy metal pollution is a serious environmental hazard in
1
Zhejiang Academy of Agricultural Sciences, No. 198 Shiqiao China, especially in areas around industrial zones (Yu et al.
Road, Hangzhou 310021, Zhejiang, China
2012; Yang et al. 2017; Liu et al. 2019). Heavy metals
2
Key Laboratory of Information Traceability of Agriculture usually have long residence times and persistent bioavail-
Products, Minstry of Agriculture and Rural Affairs, Beijing,
ability, and can accumulate in agricultural soils, leading to
China
3
gradual decreases in soil fertility, degeneration of soil
School of Remote Sensing and Information Engineering,
biology, and reduction in crop productivity (Nanos et al.
Wuhan University, Wuhan, China
4
2015; Xu et al. 2017; Zhao et al. 2014). Moreover, they
Institute of Urban Studies, School of Environmental and
easily enter plants, animals, and humans by inhalation,
Geographical Sciences, Shanghai Normal University,
Shanghai 200234, China dermal absorption, or ingestion, and are thus biomagnified

123
 Stochastic Environmental Research and Risk Assessment

through the food chain, posing a grave risk to food safety 2 Materials and methods
and human health (Sun et al. 2010; Zang et al. 2017).
Hence, there is an urgent need to quantitatively assess the 2.1 Study area
characteristics of soil heavy metal pollution and identify
the sources of heavy metals, to address the threats to soil The area of study, located in the southeastern coastal area
quality, food safety, and human health. of China (29110 –30330 N, 118210 –120300 E) is one of the
In recent years, several studies on soil heavy metal most important cities in the Yangtze River Delta city group
pollution have been conducted in China, especially on and the Ninghang ecological economic zone. It has a
pollution assessment and source analysis (Jiang et al. subtropical monsoon climate, with hot and humid sum-
2017a; Marrugo-Negrete et al. 2017; Zhao et al. 2014). mers, and cold and dry winters. According to statistical
Correspondingly, various indices have been widely used yearbook of the study area (http://tjj.hangzhou.gov.cn), the
for heavy metal pollution assessment in soil, such as Pol- annual average temperature, average humidity, precipita-
lution Index, Geoaccumulation Index, Contamination tion, and sunshine hours are 17.8 C, 70.3%, 1454 mm,
Factor, Enrichment Factor, Nemerow Integrated Pollution and 1765 h. respectively. The southwestern region of the
Index and Sediment Pollution Index (Hu et al. 2017; Loska study area is part of the western hilly areas of the Zhejiang
et al. 2004; Nemerow 1985). Furthermore, other indices Province, while the northeastern region is part of the
have been implemented to assess the impact of heavy metal northern plain area of Zhejiang. The main soil types are red
pollution on environment or human health, such as the soil and paddy soil derived from eluvium and alluvial soil
Potential Ecological Risk Index (Hakanson 1980), Hazard parent materials respectively. The area of study covers
Index and Carcinogens Risk Factors (USEPA 1997). 16,596 km2, and its total population was about 9.19 million
Generally, heavy metals are present in soil due to natural at the end of 2015 (http://tjj.hangzhou.gov.cn). Over the
processes and human activities. Natural sources of heavy past three decades, rapid urbanization and industrialization
metals are mainly controlled by their parent materials, have caused serious environmental problems in this city,
while anthropogenic sources include atmospheric deposi- including heavy metal pollution from anthropogenic sour-
tion, fertilizers and pesticides, mining, coal combustion, ces (Zhang and Wang 2009). Chemicals, machinery, non-
transportation, metalliferous industries, etc. (Liu et al. metallic minerals, metallurgy, and papermaking are the top
2015; Zhang 2006). Various multivariate statistical meth- five industries (Fei et al. 2018a). As one of the most
ods, such as principal component analysis, cluster detec- important food production bases in Hang-jia-hu Plain,
tion, multiple linear regression, and positive matrix heavy metal pollution in agricultural soil has profound
factorization have been widely used for complex dataset impacts on local food safety and human health. However,
interpretation and source identification (Luo et al. 2015; few studies have been able to conduct high density inves-
Qiao et al. 2011; Ming-Kai et al. 2013). However, most tigate of soil pollution in agricultural soil and evaluate the
studies focus on qualitative source identification, and little pollution sources in the context of spatial heterogeneity.
is known about quantitative source apportionment. More-
over, existing studies have generally classified heavy 2.2 Data sources
metals according to a single source (natural or anthro-
pogenic), and a multi-source analysis of heavy metals is A total of 1928 samples of topsoil (0–15 cm) were taken
thus lacking. Therefore, in this study, we conduct an from the centers of 1 km grid squares in 2015 (Fig. 1a). All
intensive survey to quantitatively assess the pollution risk the in situ samples were from rice, fruit, vegetable, and tea
of heavy metals, and identify the multi-sources of heavy farms. Since this study focuses on agricultural soil, the
metals in agricultural soils using a novel geographical downtown area was not sampled due to a lack of agricul-
detector model (Wang et al. 2010). tural soil. Five sub-samples around each sampling point
The aims of this paper are as follows: (a) determine the were collected and mixed thoroughly to get a representa-
concentrations of heavy metals (As, Cd, Cr, Hg, and Pb) in tive sample. With the help of the Global Positioning Sys-
agricultural soils, (b) assess the pollution levels and tem (GPS) the location of each sample site was recorded.
potential ecological risks of these heavy metals, and All soil samples were packed into polyethylene bags and
(c) quantitatively calculate the contribution of each source brought back to the lab, then air-dried at room temperature
to heavy metal pollution. The results of this study can and ground through 100 meshes for chemical analysis. To
potentially provide valuable information for soil quality analyze arsenic (As), cadmium (Cd), chromium (Cr), and
control and management, food safety insurance, and human lead (Pb), the soil samples (0.5 g) were acid-digested by a
health protection. HCl–HNO3–HClO4 mixture and their concentrations were
determined by plasma mass spectrometry (ICP-MS, TMO,

123
Stochastic Environmental Research and Risk Assessment

Fig. 1 The distribution of soil samples and contribution factors (sources) (a sampling sites and farmland type, b soil parent materials, c industrial
production, d Number of cars per 1000 people, e pesticide use, f fertilizer use)

USA) (Yang et al. 2017). For mercury (Hg), the soil 2.3 Heavy metal pollution assessment
samples were digested by a mixture of nitric acid (HNO3)
and hydrogen peroxide (H2O2) in a microwave-accelerated The pollution index was used to assess the degree of
reaction system, and its concentration was determined by contamination of each heavy metal, and is calculated as
atomic fluorescence spectrometry. Blind duplicates and follows (Zang et al. 2017):
standard reference materials (GSS-3, China National Ci
Center for Standard Materials) were used for quality PI ¼ ð1Þ
Cb;i
assurance and control. Standard sample recovery ranged
between 90 and 110%, and the relative standard deviations where PI is the pollution index, Ci is the concentration of
of the duplicate samples were between 3 and 8% (Fei et al. the ith heavy metal in the soil, and Cb;i is its corresponding
2018b). background value in the study area (Xu et al. 2012). Based
Previous studies have proven that natural sources also on the PI values, five categories were defined: unpolluted
contribute to the concentrations of heavy metals in soils (PI B 1), slightly polluted (1 \ PI B 2), mildly polluted
(Wang et al. 2015; Liu et al. 2017). To quantitatively assess (2 \ PI B 3), moderately polluted (3 \ PI B 5), and
their influence, the distribution of soil parent materials highly polluted (PI [ 5) (Zang et al. 2017). Since PI only
(Fig. 1b) was obtained from the soil database of Zhejiang represents the contamination of single heavy metal, the
Province (Wu et al. 2013). Furthermore, to quantitatively Nemerow Integrated Pollution Index (NIPI) (Nemerow
assess the impact of various anthropogenic sources on the 1985) was used to assess the overall heavy metal pollution
distribution of heavy metals (Liu et al. 2015; Zhang 2006), status of the soil. NIPI is calculated as follows:
data of industrial production in terms of ten thousand yuan sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
(proxy for industrial activities), number of cars per thou- ðPmax Þ2 þðPi Þ
NIPI ¼ ð2Þ
sand people (proxy for traffic), and pesticide and fertilizer 2
use in tons (proxy for agricultural activities) were obtained
where Pmax is the maximum value of all pollution indices
from the Hangzhou Statistical Yearbook in 2015, and their
(PIs as calculated above) of the soil heavy metals, and Pi is
distribution is shown in Fig. 1c–f. As previously stated, the
the average value of the pollution indices. Five categories
downtown area was not sampled and consequently, data of
were defined according to the NIPI: no pollution (NIPI
the anthropogenic sources in the downtown area was not
B 0.7), pollution warning threshold (0.7 \ NIPI B 1), low
collected.
pollution (1 \ NIPI B 2), moderate pollution (2 \ NIPI
B 3), and severe pollution (NIPI [ 3) (Nemerow 1985).

123
 Stochastic Environmental Research and Risk Assessment

Considering the toxicology of heavy metals, the eco- factor (source) to heavy metal concentration, the q value
logical risk index (RI) was also used to assess the eco- was estimated as follows (Wang et al. 2010):
logical risk posed by heavy metals in soils (Hakanson
1 X n
1980), and was estimated as follows: q¼1 Ni r2i ð5Þ
Nr2 i¼1
Xn
RI ¼ ERi ð3Þ where N is the total area of the study region, r2 is the
i¼1
  overall variance of heavy metal concentration, Ni and r2i
Ci are the area and heavy metal variance of the sub-region i,
ERi ¼ Ti  ð4Þ
Cb;i respectively, and n is the total number of sub-regions. The
q value measures the impact strength (contribution), such
where ERi is the ecological risk index for the heavy metal i,
that a q value ranging from 0 to 1 represents the impact
and Ti is the toxicity response coefficient for the metal i.
(from the weakest to the strongest) of a given contribution
The toxic-response factors for As, Cd, Cr, Hg, and Pb are
factor on the heavy metal concentration (Wang et al. 2010).
10, 30, 2, 40, and 5 respectively (Wang et al. 2015). Ci is
If the heavy metal (I) is completely controlled by a given
the concentration of the heavy metal i in the soil, and Cb;i is
contribution factor (S), the variance of its concentration in
its corresponding background value (Xu et al. 2012). ER
every sub-region should be 0, thus, q = 1. Whereas if the
indicating the single ecological risk of each heavy metal is
heavy metal (I) is totally uncorrelated to the contribution
defined as five categories: low risk (ER \ 40), moderate
factor (S), we get q = 0. Additionally, using this model, the
risk (40 B ER \ 80), considerable risk (80 B ER \ 160),
significance of the contribution of each source, the differ-
high risk (160 B ER \ 320), and very high risk (ER
ence in average heavy metal concentrations between sub-
C 320). RI representing the total ecological risk of evalu-
regions, and the interaction between the contributions of
ated heavy metals is defined as four categories: low risk
two sources can be quantitatively calculated (Li et al. 2013,
(RI \ 150), moderate risk (150 B RI \ 300), considerable
Fei et al. 2016). The relevant calculations were made using
risk (300 B RI \ 600), and high risk (RI C 600) (Hakan-
the GeogDetector software (Wang and Hu 2012) (www.
son 1980).
sssampling.org/geogdetector).
2.4 Statistical analysis
3 Results and discussions
In this study, GeogDetector—a novel relative spatial
variance analysis tool that works for both numerical and
3.1 Descriptive statistics of heavy metal
categorical variables—was used to quantitatively assess the
concentrations in agricultural soils
contribution factors (sources) of heavy metal contamina-
tion (Wang et al. 2010). The basic assumption of
The descriptive statistics of heavy metal concentrations in
GeogDetector is that when a heavy metal contaminant is
the agricultural soils of the study area, the background
present and a contribution factor (source) can be deter-
concentrations in Zhejiang province, and the risk screening
mined for that particular heavy metal, its concentration in
values for each heavy metal as defined in the Soil envi-
the area exhibits a spatial distribution similar to that of the
ronmental quality in China (GB 15618-2018), are shown in
contribution factor (Wang and Hu 2012). The process using
Table 1. If the concentration of a heavy metal in soil
GeogDetector is shown as a flowchart in Fig. 2. First, the
exceeds its corresponding risk value, it is considered
locations of heavy metals were converted to grid points i1,
harmful to human health (Liang et al. 2017). The concen-
i2, …, in though ordinary kriging technique (Fei et al. 2019,
trations of As, Cd, Cr, Hg, and Pb ranged from 2.63 to
Olea 2006) in Arcmap 9.3 software. Then, contribution
43.20, 0.05 to 1.42, 10.50 to 104.00, 0.03 to 0.50, and 17.00
factors such as pesticide or fertilizer use (layers S, E) were
to 62.60 mg/kg with the median values of 7.07, 0.20,
classified into different sub-regions (s1, s2, s3… and e1, e2,
54.40, 0.11, and 30.90 mg/kg, respectively. The mean
e3…) according to the principle of minimizing the disper-
concentrations of As, Cd, Cr, Hg, and Pb were 8.99, 0.27,
sion variance within each sub-region and maximizing the
52.90, 0.13, and 31.66 mg/kg, respectively, all of which are
dispersion variance between each sub-region. Subse-
lower than the Environmental Quality Standard risk
quently, the heavy metal layer I was overlaid with the
screening values. However, compared to their local back-
contribution factor layers (S or E) and the area of each sub-
ground values, the mean concentrations of Cr, Hg, and Pb
region and the corresponding variance of heavy metal
were lower, whereas the mean concentrations of As and Cd
concentration in each sub-region was then calculated.
were higher. The mean concentrations of As and Cd were
Finally, to quantitatively assess the contribution of each
1.31 and 1.59 times their respective background values.

123
Stochastic Environmental Research and Risk Assessment

Fig. 2 The flowchart of GeogDetector

Table 1 Descriptive statistics of

Min Max Median Mean CV Backgrounda Standardb Skewness Kurtosis
soil heavy metal concentration
in soil (mg/kg) As 2.63 43.20 7.07 8.99 0.69 6.88 30/25/20 2.309 6.222
Cd 0.05 1.42 0.20 0.27 0.81 0.17 0.3/0.3/0.6 2.431 6.846
Cr 10.50 104.00 54.40 52.90 0.38 55.99 150/200/250 - 0.024 - 0.572
Hg 0.03 0.50 0.11 0.13 0.62 0.17 0.3/0.5/1.0 1.602 3.073
Pb 17.00 62.60 30.90 31.66 0.38 35.70 300/300/300 0.720 0.454
CV coefficient of variation
a
Data from Xu et al. (2012)
b
Based on guidelines in the Environmental Quality Standard for Soils in China (GB 15618-1995) and the
permissible limits for soils with pH less than 6.5, 6.5–7.5, and more than 7.5

This may be due to the combined effect of natural pro- 3.2 Pollution assessment
cesses and a long history of industrial activities and
urbanization (Fei et al. 2018a, Zang et al. 2017). The The PI values of each heavy metal in the agricultural soil
coefficient of variation (CV) of the heavy metals decreased samples are summarized in Table 2. The PI values of As,
in the order of Cd (81%) [ As (69%) [ Hg (62%) [ Pb Cd, Cr, Hg, and Pb ranged from 0.38 to 6.28, 0.32 to 8.35,
(38%) = Cr (38%). Additionally, according to the skew- 0.19 to 1.86, 0.20 to 2.94, and 0.48 to 1.75 mg/kg,
ness and kurtosis values, and the Kolmogorov–Smirnov respectively. The mean PI values for each heavy metal in
tests of normality (P \ 0.01 for all heavy metals), the decreasing order were Cd (1.59) [ As (1.31) [ Cr
heavy metals had a non-normal distribution. High CVs and (0.94) [ Pb (0.89) [ Hg (0.78). The PI values indicated
skewed distributions caused by a high degree of geo- that the study area was either unpolluted or slightly pol-
chemical variation (differed greatly with respect to differ- luted by Pb and Cr (PI B 2). Although Hg had the lowest
ent sites), indicate that the heavy metal concentrations were PI value, 2.88% of the soil samples exhibited mild pollu-
strongly influenced by human activities (Shao et al. 2016; tion (2 \ PI B 3). The level of As and Cd pollution was
Mamut et al. 2018). considerable; the percentages of mild pollution, moderate
pollution (3 \ PI B 5), and high pollution (PI [ 5) were
8.05%, 4.88%, and 1.12%, respectively, for As, and
10.09%, 6.79%, and 3.40%, respectively, for Cd.

123
 Stochastic Environmental Research and Risk Assessment

Table 2 The pollution index

Mean Min Max Percentage of unpolluted
value of each heavy metal in
and slightly polluted (%)
soil
As 1.31 0.38 6.28 85.95
Cd 1.59 0.32 8.35 79.72
Cr 0.94 0.19 1.86 100
Hg 0.78 0.20 2.94 97.12
Pb 0.89 0.48 1.75 100

As for the NIPI, the values ranged from 0.59 to 6.39. respectively. Cd and Hg had the highest contribution to RI
Overall, 81.22% of the soils had a value indicating no (48.43% and 31.81%, respectively). The contributions of
pollution, pollution near the warning threshold, or low As, Pb, and Cr were 13.32%, 4.51%, and 1.93%, respec-
levels of pollution; 11.06% were moderately polluted and tively. ER and RI are effective indicators of the degree of
7.72% were severely polluted. The results of the PI and the individual and comprehensive ecological risk of soil
NIPI indicated that almost one-fifth of the soil in the study heavy metals (Jiang et al. 2017b). The results showed that
area exhibited moderate pollution, with As and Cd being there is a moderate ecological risk of soil heavy metal
the main pollutants. pollution in the study area. Approximately 12% of the soil
The ER values of each heavy metal in the agricultural exhibited a moderate or higher ecological risk, with Cd
soil samples are summarized in Table 3. The ER values of presenting the highest ecological risk due to its high con-
As, Cd, Cr, Hg, and Pb ranged from 3.82 to 62.79, 9.53 to centration and toxic-response factor (30). Moreover, Cd is
250.59, 0.38 to 3.71, 8 to 117.65, and 2.38 to 8.78, easily absorbed by crops and can harm human health due to
respectively. The mean ER values of each metal in when ingested (Hu et al. 2017). Thus, Cd pollution of the
decreasing order were Cd (47.52) [ Hg (31.21) [ As soil present in the study area requires further attention.
(13.07) [ Pb (4.43) [ Cr (1.89). According to the ER
values of Cr and Pb, all the soil samples were in the low 3.3 Identification of sources of heavy metals
risk category. Although only 2.88% of the soil samples
were classified as mildly polluted based on the PI value of According to the Spearman correlation analysis, the heavy
Hg, 19.63% and 2.88% of soils were classified as moderate metals were closely related to each other. The high
and considerable ecological risks, respectively, due to the homology between the heavy metals in agricultural soils
high toxic-response factor of Hg (40). In the case of Cd, indicated that they may have common sources such as
58.14% of the soil samples were in the low risk category, lithogenic components, soil parent materials and agricul-
while 28.79%, 10.14%, and 2.93% of the soils were in the tural activities (Gao et al. 2017). The contributions
moderate, considerable, and high risk categories respec- (q values) of the detected contribution factors (sources) on
tively. In the case of As, despite its relatively high PI the distribution of heavy metals are shown in Table 4. The
values, most of the soil samples (96.98%) were in the low GeogDetector model also revealed that the soil parent
risk category and 3.02% of the soils were in the moderate material had a significant impact on the distribution of all
risk category because of its relatively low toxic-response heavy metals. The q values increased in the order of Hg
factor (10). (0.113) \ Pb (0.125) \ Cd (0.127) \ As (0.164) \ Cr
The RI value ranged from 31.48 to 374.20. Overall, (0.214). Many studies have reported that Cr originates from
88.24%, 11.47%, and 0.29% of the soil samples were in the
low risk, moderate risk, and considerable risk categories, Table 4 The q values of the influence factors on heavy metals in soil
SPMs FT IP NC FU PU
Table 3 The ecological index value of each heavy metal in soil As 0.164** 0.112** 0.118** 0.135** 0.124** 0.122**
Mean Min Max Percentage of low risk (%) Cd 0.127** 0.002 0.004 0.121** 0.006 0.110**
Cr 0.214** 0.126** 0.125** 0.004 0.011 0.032
As 13.07 3.82 62.79 96.98
Hg 0.113** 0.015 0.110** 0.110** 0.032 0.032
Cd 47.52 9.53 250.59 58.14
Pb 0.125** 0.006 0.017 0.007 0.114** 0.110**
Cr 1.89 0.38 3.71 100
Hg 31.21 8 117.65 77.49 SPMs soil parent materials, FT farmland type, IP industrial produc-
tion, NC number of cars (1/1000), FU fertilizer use, PU pesticide use
Pb 4.43 2.38 8.78 100
**Statistically significant at 0.01 level

123
Stochastic Environmental Research and Risk Assessment

the soil parent material (Chen et al. 2016; Salonen and number of cars per thousand people, the concentrations of
Korkka-Niemi 2007; Xue et al. 2014) and thus, the soil As, Cd, and Hg increased from 6.92 to 13.87, 0.17 to 0.46,
parent material showed the highest q value for Cr con- and 0.14 to 0.20 mg/kg, respectively (Fig. 3b). Heavy
centrations. Table 5 shows the concentrations of heavy traffic causes air pollution, indicating that As, Cd, and Hg
metals in different soil parent materials. Generally, As, Cd, concentrations in soil are significantly influenced by
Hg, and Pb had the lowest concentrations in estuarine atmospheric deposition, which agrees with the findings
facies, whereas Cr had the lowest concentration in alluvial from previous studies (Engle et al. 2005; Liang et al. 2017;
facies; As and Cd had the highest concentrations in eluvial Micó et al. 2006).
facies, whereas Cr, Hg, and Pb had the highest concen- Fertilizer use significantly influenced the concentrations
trations in lacustrine facies. of As (q = 0.124) and Pb (q = 0.114). With the increase in
Farmland types had a significant influence on the con- the use of fertilizers, the concentrations of As and Pb
centrations of As (q = 0.112) and Cr (q = 0.126). Farmland increased from 4.81 to 12.94, and 26.65 to 37.30 mg/kg,
types with heavy metal contamination were, in increasing respectively (Fig. 3c). Pesticide use significantly influ-
order of concentrations, paddy fields (9.58 and 51.98), dry enced the concentrations of As (q = 0.122), Cd
land (10.24 and 56.19), orchards (13.91 and 58.53), and tea (q = 0.110), and Pb (q = 0.110). With the increase in the
gardens (20.75 and 63.57) of mg/kg As and Cr, respec- use of pesticides, the concentrations of As, Cd, and Pb
tively. Statistical analyses showed that there were signifi- increased from 4.81 to 14.15, 0.18 to 0.45, and 26.65 to
cant differences (P \ 0.05) in As and Cr concentrations in 38.08 mg/kg, respectively (Fig. 3d). Although fertilizers
paddy fields compared to tea gardens and orchards. This and pesticides are essential for successful harvests, long-
indicates that different types of agriculture have different term use of the same fertilizers and pesticides can accu-
influences on the concentrations of heavy metals in soil. mulate heavy metals in soils (Marrugo-Negrete et al.
Thus, by adjusting the type of agriculture, the pollution 2017). Previous studies confirmed that the long-standing
levels of As and Cr can be reduced, which can improve farming practices of fertilizer and pesticide application can
food safety and human health (Hu et al. 2017). lead to the accumulation of heavy metals such as As and Pb
Industrial production had a significant influence on the in soils (Jiang et al. 2017b; Liu et al. 2017; Qiao et al.
concentrations of As (q = 0.118), Cr (q = 0.125), and Hg 2011). This is especially true for Cd, which is closely
(q = 0.110). As a result of increased industrial production, related to the intensive use of chemical fertilizers and
the concentrations of As, Cr, and Hg increased from 4.81 to pesticides, and is usually seen as a marker of agricultural
12.53, 27.24 to 58.72, and 0.14 to 0.21 mg/kg, respectively activities (Wang et al. 2015).
(Fig. 3a). Many industrial activities, such as non-metallic Despite this, pesticides did not have a significant influ-
mineral smelting, metallurgy, and chemical manufacturing, ence on all heavy metals. When considering the impact of
are carried out in the study area (Fei et al. 2018a). Previous the interaction of two source factors, the joint impact of the
studies reported that these industries cause the enrichment soil parent materials and pesticides exhibited the highest
of As, Cr, and Hg in soil, through atmospheric deposition q values for all the heavy metals. The interaction q values
(Liang et al. 2017; Liu et al. 2017; Luo et al. 2015). were 0.316, 0.255, 0.279, 0.186, and 0.275 for As, Cd, Cr,
Industrial activities had a huge impact on heavy metals in Hg, and Pb, respectively (Table 6). Besides, the interaction
soil, particularly on As and Hg which demonstrated heavy q values were higher than the sum q values of soil parent
pollution and high ecological risk according to the results materials and pesticide use which indicating that these two
discussed in the previous section. Therefore, industrial sources had a nonlinear enhancement. The reasons behind
activities must be properly regulated and strictly limited to this phenomenon are very complicated that needed to be
protect this area. analyzed in further studies, for example, agricultural
The number of cars per thousand people had a signifi- activities can accelerate the release of heavy metals in soil
cant influence on the concentrations of As (q = 0.135), Cd parent materials, diverse heavy metal contamination level
(q = 0.121), and Hg (q = 0.110). With the increase in the is resulted by different types and strengths of agricultural

Table 5 The mean

Lacustrine facies Estuarine facies Alluvial facies Eluvial facies
concentration (ppm) of heavy
metals in various soil parent As 7.52 5.51 8.96 14.19
materials
Cd 0.19 0.17 0.39 0.41
Cr 72.87 58.13 46.83 53.55
Hg 0.21 0.14 0.15 0.14
Pb 35.23 27.27 35.1 34.82

123
 Stochastic Environmental Research and Risk Assessment

Fig. 3 The concentrations of heavy metals from stratification of different source factors (1–5 corresponded to low, mild, moderate, high, and
very high stratification respectively)

Table 6 The max interaction q values on various heavy metals in soil 4 Conclusions
Interaction factors q value
This study quantitatively assessed the pollution level,
As SPMs & PU 0.316 ecological risk, and sources of heavy metals in agricultural
Cd SPMs & PU 0.255 soils in a typical coastal industrial area, undergoing rapid
Cr SPMs & PU 0.279 industrialization and urbanization. According to the results,
Hg SPMs & PU 0.186 Cd was the most polluting element in the soil, and, due to
Pb SPMs & PU 0.275 its high toxic-response factor, it also poses the highest risk
SPMs soil parent materials, PU pesticides use, non-linear enhance- for ecological systems and human health. Another element
ment: interaction q values greater than the sum of their respective responsible for heavy pollution was As. However, due to its
q value relatively low toxic-response factor, most of the soils were
in the low ecological risk category. In contrast, although
Hg contamination had relatively small concentrations, one-
fifth of the soil had a moderate ecological risk due to its
activities on various soil parent materials, etc. These
high toxic-response factor. A novel geographical detector
interaction results indicated that natural processes and
model was employed to quantitatively identify the different
agricultural activities were the main sources of soil heavy
sources of heavy metals. Natural processes (soil parent
metal pollution in the study area.
material) had a significant impact on all the heavy metals

123
Stochastic Environmental Research and Risk Assessment

analyzed in this study. Additionally, farmland types had a in the Yangtze river delta, China. Int J Environ Res Pub Health
significant influence on Cr and As, industrial activities had 14(9):1042
Islam MS, Ahmed MK, Habibullah-Al-Mamun M (2016) Apportion-
a significant influence on As, Cr, and Hg, traffic signifi- ment of heavy metals in soil and vegetables and associated
cantly influenced As, Cd, and Hg concentrations, and health risks assessment. Stoch Environ Res Risk Assess
agricultural applications of fertilizer and pesticide had a 30:365–377
significant impact on As, Cd, and Pb. Arsenic, with diverse Jiang Y, Chao S, Liu J, Yang Y, Chen Y, Zhang A, Cao H (2017a)
Source apportionment and health risk assessment of heavy
sources, requires further analysis in future studies. Natural metals in soil for a township in Jiangsu Province, China.
processes and agricultural activities were found to be the Chemosphere 168:1658–1668
main sources of soil heavy metal pollution in the study Jiang X, Xiong Z, Liu H, Liu G, Liu W (2017b) Distribution, source
area. identification, and ecological risk assessment of heavy metals in
wetland soils of a river–reservoir system. Environ Sci Pollut R
24(1):436–444
Acknowledgements This work was partially supported by the Li X, Xie Y, Wang J, Christakos G, Si J, Zhao H, Ding Y, Li J (2013)
National Key R&D Program (2018YFD0200500 and Influence of planting patterns on fluoroquinolone residues in the
2017YFD0200600) and the National Natural Science Foundation of soil of an intensive vegetable cultivation area in northern China.
China (No. 41801302). They have no role in study design; in the Sci Total Environ 458:63–69
collection, analysis and interpretation of data; in the writing of the Liang J, Feng C, Zeng G, Gao X, Zhong M, Li X, He X, Fang Y
report; and in the decision to submit the article for publication. (2017) Spatial distribution and source identification of heavy
metals in surface soils in a typical coal mine city, Lianyuan,
Compliance with ethical standards China. Environ Pollut 225:681–690
Liu J, Liang J, Yuan X, Zeng G, Yuan Y, Wu H, Huang X, Liu J, Hua
Conflict of interest The authors declare that they have no conflict of S, Li F, Li X (2015) An integrated model for assessing heavy
interest. metal exposure risk to migratory birds in wetland ecosystem: a
case study in Dongting Lake Wetland, China. Chemosphere
135:14–19
Liu H, Zhang Y, Zhou X, You X, Shi Y, Xu J (2017) Source
References identification and spatial distribution of heavy metals in tobacco-
growing soils in Shandong province of China with multivariate
Chen T, Liu X, Li X, Zhao K, Zhang J, Xu J, Shi J, Dahlgren RA and geostatistical analysis. Environ Sci Pollut R
(2009) Heavy metal sources identification and sampling uncer- 24(6):5964–5975
tainty analysis in a field-scale vegetable soil of Hangzhou, Liu S, Pan G, Zhang Y, Xu J, Ma R, Shen Z, Dong S (2019) Risk
China. Environ Pollut 2009(157):1003–1010 assessment of soil heavy metals associated with land use
Chen H, Teng Y, Lu S, Wang Y, Wu J, Wang J (2016) Source variations in the riparian zones of a typical urban river gradient.
apportionment and health risk assessment of trace metals in Ecotox Environ Safe 181:435–444
surface soils of Beijing metropolitan, China. Chemosphere Loska K, Wiechuła D, Korus I (2004) Metal contamination of
144:1002–1011 farming soils affected by industry. Environ Int 30(2):159–165
Engle MA, Gustin MS, Lindberg SE, Gertler AW, Ariya PA (2005) Luo XS, Xue Y, Wang YL, Cang L, Xu B, Ding J (2015) Source
The influence of ozone on atmospheric emissions of gaseous identification and apportionment of heavy metals in urban soil
elemental mercury and reactive gaseous mercury from sub- profiles. Chemosphere 127:152–157
strates. Atmos Environ 39(39):7506–7517 Mamut A, Eziz M, Mohammad A (2018) Pollution and ecological
Fei X, Wu J, Liu Q, Ren Y, Lou Z (2016) Spatiotemporal analysis and risk assessment of heavy metals in farmland soils in Yanqi
risk assessment of thyroid cancer in Hangzhou China. Stoch County, Xinjiang, Northwest China. Eurasian Soil Sci 51:985
Environ Res Risk Assess 30(8):2155–2168 Marrugo-Negrete J, Pinedo-Hernández J, Dı́ez S (2017) Assessment
Fei X, Lou Z, Christakos G, Liu Q, Ren Y, Wu J (2018a) Contribution of heavy metal pollution, spatial distribution and origin in
of industrial density and socioeconomic status to the spatial agricultural soils along the Sinú River Basin. Colomb Environ
distribution of thyroid cancer risk in Hangzhou, China. Sci Total Res 154:380–388
Environ 613:679–686 Micó C, Recatalá L, Peris M, Sánchez J (2006) Assessing heavy metal
Fei X, Lou Z, Christakos G, Ren Z, Liu Q, Lv X (2018b) The sources in agricultural soils of an European Mediterranean area
association between heavy metal soil pollution and stomach by multivariate analysis. Chemosphere 65(5):863–872
cancer: a case study in Hangzhou City, China. Environ Geochem Ming-Kai QU, Wei-Dong LI, Zhang CR, Shan-Qin WA, Yong YA,
Heal 40(6):2481–2490 Li-Yuan HE (2013) Source apportionment of heavy metals in
Fei X, Christakos G, Xiao R, Ren Z, Liu Y, Lv X (2019) Improved soils using multivariate statistics and geostatistics. Pedosphere
heavy metal mapping and pollution source apportionment in 23(4):437–444
Shanghai City soils using auxiliary information. Sci Total Nanos N, Grigoratos T, Rodrı́guez Martı́n JA, Samara C (2015)
Environ 661:168–177 Scale-dependent correlations between soil heavy metals and As
Gao Y, Liu H, Liu G (2017) The spatial distribution and accumulation around four coal-fired power plants of northern Greece. Stoch
characteristics of heavy metals in steppe soils around three Environ Res Risk Assess 29:1531–1543
mining areas in Xilinhot in Inner Mongolia, China. Environ Sci Nemerow NL (1985) Stream, lake, estuary, and ocean pollution. Van
Pollut R 24(32):25416–25430 Nostrand Reinhold Publishing Co., New York
Hakanson L (1980) An ecological risk index for aquatic pollution- Olea RA (2006) A six-step practical approach to semivariogram
control—a sedimentological approach. Water Res 14:975–1001 modeling. Stoch Stoch Environ Res Risk Assess 20(5):307–318
Hu B, Jia X, Hu J, Xu D, Xia F, Li Y (2017) Assessment of heavy Qiao M, Cai C, Huang Y, Liu Y, Lin A, Zheng Y (2011)
metal pollution and health risks in the soil–plant–human system Characterization of soil heavy metal contamination and potential

123
 Stochastic Environmental Research and Risk Assessment

health risk in metropolitan region of northern China. Environ Xu DM, Yan B, Chen T, Lei C, Lin HZ, Xiao XM (2017)
Monit Assess 172(1–4):353–365 Contaminant characteristics and environmental risk assessment
Salonen VP, Korkka-Niemi K (2007) Influence of parent sediments of heavy metals in the paddy soils from lead (Pb)-zinc (Zn)
on the concentration of heavy metals in urban and suburban soils mining areas in Guangdong Province, South China. Environ Sci
in Turku, Finland. Appl Geochem 22(5):906–918 Pollut R 24(31):24387–24399
Shao D, Zhan Y, Zhou W, Zhu L (2016) Current status and temporal Xue JL, Zhi YY, Yang LP, Shi JC, Zeng LZ, Wu LS (2014) Positive
trend of heavy metals in farmland soil of the Yangtze River matrix factorization as source apportionment of soil lead and
Delta Region: field survey and meta-analysis. Environ Pollut cadmium around a battery plant (Changxing County, China).
219:329–336 Environ Sci Pollut R 21(12):7698–7707
Sun Y, Zhou Q, Xie X, Liu R (2010) Spatial, sources and risk Yang Y, Christakos G, Guo M, Xiao L, Huang W (2017) Space-time
assessment of heavy metal contamination of urban soils in quantitative source apportionment of soil heavy metal concen-
typical regions of Shenyang, China. J Hazard Mater tration increments. Environ Pollut 223:560–566
174(1–3):455–462 Yu S, Zhu YG, Li XD (2012) Trace metal contamination in urban
USEPA (United States Environmental Protection Agency) (1997) soils of China. Sci Total Environ 421:17–30
Exposure factors handbook; office of research and development. Zang F, Wang S, Nan Z, Ma J, Zhang Q, Chen Y, Li Y (2017)
National Center for Environmental Assessment, Washington, Accumulation, spatio-temporal distribution, and risk assessment
DC of heavy metals in the soil-corn system around a polymetallic
Wang JF, Hu Y (2012) Environmental health risk detection with mining area from the Loess Plateau, northwest China. Geoderma
GeogDetector. Environ Model Softw 33:114–115 305:188–196
Wang JF, Li XH, Christakos G, Liao YL, Zhang T, Gu X, Zheng XY Zhang C (2006) Using multivariate analyses and GIS to identify
(2010) Geographical detectors-based health risk assessment and pollutants and their spatial patterns in urban soils in Galway,
its application in the neural tube defects study of the Heshun Ireland. Environ Pollut 142(3):501–511
Region, China. Int J Geogr Inf Sci 24(1):107–127 Zhang M, Wang H (2009) Concentrations and chemical forms of
Wang Y, Yang L, Kong L, Liu E, Wang L, Zhu J (2015) Spatial potentially toxic metals in road-deposited sediments from
distribution, ecological risk assessment and source identification different zones of Hangzhou, China. J Environ Sci
for heavy metals in surface sediments from Dongping Lake, 21(5):625–631
Shandong, East China. CATENA 125:200–205 Zhao L, Xu Y, Hou H, Shangguan Y, Li F (2014) Source
Wu J, Hu Y, Zhi J, Jing C, Chen H, Xu J, Lin H, Li D, Zhang C, Xiao identification and health risk assessment of metals in urban soils
R, Huang H (2013) A 1: 50 000 scale soil database of Zhejiang around the Tanggu chemical industrial district, Tianjin, China.
Province, China. Acta Pedol Sin 50(1):30–40 Sci Total Environ 468:654–662
Xu YY, Shi J, Zhou LY (2012) Characteristics of heavy metals
distribution in agricultural soils of Hangzhou and its environ- Publisher’s Note Springer Nature remains neutral with regard to
ment significances. Environ Monit China 28(4):74–80 jurisdictional claims in published maps and institutional affiliations.

123