Heavy metal contaminations in a soil–rice system: Identification of spatial dependence in relation to soil properties of paddy fields

Journal of Hazardous Materials 181 (2010) 778–787

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Heavy metal contaminations in a soil–rice system: Identification of spatial dependence in relation to soil properties of paddy fields Keli Zhao a,b , Xingmei Liu a,∗ , Jianming Xu a,∗ , H.M. Selim b a Institute of Soil and Water Resources and Environmental Science, Zhejiang Provincial Key Laboratory of Subtropical Soil and Plant Nutrition, Zhejiang University, Hangzhou 310029, China b School of Plant, Environmental and Soil Sciences, Louisiana State University AgCenter, Baton Rouge 70803, USA

a r t i c l e

i n f o

Article history: Received 9 March 2010 Received in revised form 16 May 2010 Accepted 17 May 2010 Available online 24 May 2010 Keywords: Contamination Geostatistics Heavy metals Soil–rice system Spatial relationship

a b s t r a c t In order to identify spatial relationship of heavy metals in soil–rice system at a regional scale, 96 pairs of rice and soil samples were collected from Wenling in Zhejiang province, China, which is one of the well-known electronic and electric waste recycling centers. The results indicated some studied areas had potential contaminations by heavy metals, especially by Cd. The spatial distribution of Cd, Cu, Pb and Zn illustrated that the highest concentrations were located in the northwest areas and the accumulation of these metals may be due to the industrialization, agricultural chemicals and other human activities. In contrast, the concentration of Ni decreased from east to west and the mean concentration was below the background value, indicating the distribution of Ni may be naturally controlled. Enrichment index (EI) was used to describe the availability of soil heavy metals to rice. The spatial distribution of EIs for Cd, Ni and Zn exhibited a west-east structure, which was similar with the spatial structures of pH, OM, sand and clay. Cross-correlograms further quantitatively illustrated the EIs were significantly correlated with most soil properties, among which; soil pH and OM had the strongest correlations with EIs. However, EI of Cu showed relative weak correlations with soil properties, especially soil pH and OM had no correlations with EI of Cu, indicating the availability of Cu may be influenced by other factors. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Heavy metal contaminations of land resources continue to be the focus of numerous environmental studies and attract a great deal of attention worldwide. This is attributed to non- biodegradability and persistence of heavy metals in soils [1–3]. Agricultural soil has been widely contaminated as a result of uncontrolled discharge and emissions from rapidly expanding industrial areas, mining, the misuse of chemical fertilizers and land application of industrial effluents, and sewage waste [4,5]. Heavy metal contamination of agriculture lands poses a potential threat to the safe crop production in China and worldwide [6]. Rice is the dominant agricultural crop in China and ranks second by quantity in the world. The quality of rice, thus, affects greatly human health. Heavy metals in rice grown in soils may accumulate in the human body through the food chain, lead to psychotic disorders and cause many debilitating diseases [7]. Therefore heavy metals contamination in agriculture soils and their transfer in a soil–rice system have been of increasing concern.

∗ Corresponding authors. Tel.: +86 571 86971955; fax: +86 571 86971955. E-mail addresses: [email protected] (X. Liu), [email protected] (J. Xu). 0304-3894/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2010.05.081

Several studies on sustainable farming reported the accumulation and transfer of heavy metals in soil–rice systems, including the distribution in plant organs, influence of soil properties and rice genotypic on heavy metals uptake. The distribution and accumulation of heavy metals in plants vary among the various organs. In rice, grains contain heavy metals significantly lower than other organs (root, straw and sheet) [8–10]. Previous studies have investigated the effect of soil properties on the transfer of heavy metals, i.e. soil pH [11,12], organic matter [8,13,14], redox potential [15], cation exchange capacity [16], and phosphorous content [17]. Among these properties, soil pH was considered as the most important factor. In addition, increased attention has been focused on the interaction of heavy metal soil characteristics and genotypic. The positive interaction between soil type and genotype has a significant effect on heavy metal availability, followed by genotype effect [5]. Examples of heavy metal bioavailability which have been studied include cadmium (Cd), copper (Cu), chromium (Cr), lead (Pb), nickel (Ni), zinc (Zn), mercury (Hg), and others [5,6,16,18–20]. Among these heavy metals, Cd accumulation in rice was easier relative to other metals [8,9]. This suggests that Cd may post a potential risk. However, most previous studies on accumulation of heavy metals were carried out using pot or field experiments, and some of these works focused on special field areas (i.e. industry regions) and

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

779

Fig. 1. Location of samples and distribution of different soil types in the study area.

related risk assessment. Liu et al. [9], Jung and Thornton [16], Rogan et al. [21] investigated the trace element concentrations in rice from paddy fields that was irrigated with riverine water affected by acid mine drainage and other mining activities of ore districts. To our knowledge, little information is available on heavy metal accumulation in paddy fields on regional scales, and spatial correlation character between heavy metals in soil–rice system. Geostatistics as methods focus on spatial objects and spatial correlation and have been widely applied to describe spatial structures, provide input parameters for spatial interpolation, and assess the uncertainty at unsampled locations [22,23]. A number of other statistical procedures have been used for studying relationships among crop and soil properties, including simple linear regression, multiple regression, principal components, and other multivariate techniques [24]. However, Cross-correlogram, one of the geostatistical methods, is considered a useful method of characterizing relationship among variables, because it accounts for data location and spatial structure of data distribution. Such information is often ignored in traditional statistics. Because of these advantages, crosscorrelogram has become a useful tool to describe spatial aspects of the relationship between crops yield and soil properties [24–26]. Wenling city is one of the most developed regions in China, especially in southeast China. Many industries are distributed in this region, including machine and electric production, leather and plastic production, dye, among others. Particularly, the study area is one of the well-known electronic and electric waste (E-waste) recycling centers in southeast China. Since 1990s, small and open specialized E-waste recycling shelters or yards appeared, particularly to the center of the northwest area (Fig. 1). E-wastes commonly contain many heavy metals such as Cd, Cu, Pb, Hg and Zn. The unregulated disposal and processing of E-waste were used for the recovery of gold and other valuable metals. Rapid industrializations and simple techniques in E-waste recycling resulted in that heavy metals most entered the environment and then accumulated in soils and crops. The overall goals of this study were to quantify heavy metal concentrations and spatial patterns in surface soils as well as their bioavailability to rice plants grown on these soils in the Wenling city region. The specific objectives were (1) to study the spatial variability of heavy metals in surface soil and rice grains; (2) to

investigate the spatial dependence and correlation of heavy metals in soil–rice system; (3) to determine soil factors influencing the availability of heavy metal to rice. We utilized spatial interpolation with geostatistics methods including cross-correlogram in order to delineate spatial relationships on a regional scale in paddy fields. Our findings should provide guidelines beneficial to agriculture management and future strategic sustainable agriculture in China and other developing regions. 2. Materials and methods 2.1. Study area and sampling This research was carried out in Wenling, which is located in the southeast Zhejiang province, China (Fig. 1). It covers 121◦ 10 –121◦ 44 E longitude by 28◦ 13 –28◦ 32 N latitude, an area of 926 km2 with a population of some 1.2 million. Wenling is close to the China East Sea with a warm and humid subtropical climate. The annual mean temperature and rainfall are 17.3 ◦ C and 1693 mm, respectively. The major soil groups in the study area are acrisols, anthrosols, cambisols, lixisols, regosols and solonchaks, based on the soil classification from World Reference Base for Soil Resources (FAO/ISRIC/ISSS, 1998). A paddy soil refers to any type of soil for aquatic rice production and it is used to signify low land rice soils. So the term “paddy soil” in this paper is related to land use and not to any strict definition of soil in the pedological sense. As shown in Fig. 1, different soil types were used for growing aquatic rice, however, it mainly focused on anthrosols. The agriculture in Wenling was well developed and rice was the main grain crop. The growing period of rice plant (Oryza Sativa L.) is from June to October in Wenling city. The germinated seeds are grown in a shallow field and raised under flooding for 20–30 days. Then the seedings are transplanted into paddy fields. The soil is kept in a distinct cycle of flooded and non-flooded conditions during the entire growing season. The main fertilizers including urea, calcium superphosphate and potassium chloride are applied; pesticides are also used if needed. At rice harvest time (October 2006), 96 pairs of rice and soil samples with same location were collected from Wenling, by means of

780

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

a completely randomized design, on the basis of a land use map at 1:50,000 scale. Each sample was the composite of at least 5 sub-samples within a distance of 10 m surrounding a specific sampling location. The rice and their rooted soil samples (at 0–15 cm in depth) were collected from each site. The coordinates of sampling locations were recorded with a differential Global Position System (GPS). Sampling was focused in plain rice-production regions; mountainous and downtown areas were avoided (Fig. 1). Rice in this study was focused on the rice grain. 2.2. Chemical analysis Soil samples were air-dried in the laboratory for several days at ambient temperature. They were passed through a 2 mm nylon sieve for chemical and physical analysis of soil properties. A partial of the samples were ground in an agate mortar to pass through 100 meshes and stored in closed polyethylene bags for heavy metal concentrations and soil organic matter analysis. Soil samples were analyzed according to the basic methods in China [27]. Soil samples were digested for all the heavy metals using HF, HNO3 and HClO4 . Soil Cd concentration was determined by graphite furnace atomic absorption spectroscopy (GFAAS, PerkinElmer AA800, USA); Cu, Ni, Pb and Zn concentrations were determined by flameatomic absorption spectroscopy (FAAS, PerkinElmer AA800, USA). Soil pH and electrical conductivity (EC) were measured in an aqueous suspension (1:2.5 and 1:5 soil water ratio, respectively). Soil organic matter was determined by the potassium dichromate wet combustion procedure. Soil amorphous Fe oxide was extracted by ammonium oxalate (at pH 3.0–3.5). Soil particle size distribution (sand, silt and clay content) was analyzed by using the hydrometer method. Rice grain samples were oven-dried at 105 ◦ C for 1 h, then at ◦ 70 C to constant weight. Hull was removed from rice. Then rice samples were comminuted using a pulverizer, ground to pass through 100 meshes using a nylon sieve and stored in closed polyethylene bags for further heavy metal concentration analysis. Rice samples were digested using HNO3 and H2 O2 . Cd, Cu, Ni and Pb concentrations were determined by graphite furnace atomic absorption spectroscopy (GFAAS, PerkinElmer AA800, USA), while Zn concentration was determined by flame-atomic absorption spectroscopy (FAAS, PerkinElmer AA800, USA). The accuracy of determinations was verified using the Chinese standardized reference materials (GSS-4 and GSS-15 for soil samples; GBW(E)080684 for rice samples). The samples were measured in duplicate. For these procedures, analytical quality control showed good precision throughout. The analyzed results were calculated on the dry-weight basis. 2.3. Geostatistics analysis Semivariogram was used to quantify the spatial variability of a regionalized variable, which relates dissimilarity between paired data values to the distance between each sample pair [23,28,29]. There are some commonly used theorical models to fit experimental semivariograms, e.g., spherical, exponential, Gaussian, and linear models [29]. Fitted models to experimental variograms provide input parameters for spatial interpolation. Semivariogram can be expressed as: 1  2 [Z(xi ) − Z(xi + h)] 2N(h)

Kriging is a commonly used spatial interpolation method because it provides the best linear unbiased estimate for spatial variables relative to other methods [22,30]. Among numerous kriging techniques, ordinary kriging (OK) is probably the most familiar interpolation method, and has been widely applied in environmental science [31,32]. The cross-correlogram is a useful geostatistical tool to determine spatial correlation between two variables and a tool to compare spatial variability patterns of different variables [26]. Cross-correlogram describes existing correlation between variable 1 and variable 2 separated by a distance h. At zero distance, the cross-correlogram 12 (0) is equal to the Pearson correlation coefficient [28,33]; it can be used to describe the similarity of spatial patterns, which completely similar with 12 (0) equals to 1 and opposite similar with 12 (0) equals to −1 [26]. A large crosscorrelogram value indicates high correlation. It is calculated as: n 

1/n 12 (h) =

Z1 (xi )Z2 (xi + h) − m1−h m2+h

i=1



(2) 2 2 1−h 2+h

where Z1 (xi ) is the value of variable 1 at location xi ; Z2 (xi + h) is the value of variable 2 at a location separated by distance h from 2 location xi ; m1−h and m2+h are the head and tail means and 1−h

2 are the head and tail variances of Variable1 and Variable and 2+h 2, respectively; and n is the number of data pairs used to calculate the cross-correlogram at each distance h. Three parameters (r, range, shape) in the cross-correlogram are used to describe the spatial correlation between two variables. The cross-correlogram value at zero distance means the strength of the relationship; the spatial correlation range means the distance over which two variables are correlated; the shape of the crosscorrelogram means how fast the correlation between two variables deteriorates with distance.

2.4. Data analysis In linear geostatistics method, a normal distribution for the variable is desired in order to avoid distortions of data and low levels of significance [23,34]. In numerous data transformation methods, logarithmic transformation is widely applied [23]. In this study, the distribution of the data was tested for normality by the Kolmogorov–Smirnov (K–S) test. The logarithm transformation was performed on soil Cd, Cu, Pb, Zn concentration and rice Cd, Ni concentration for further analysis since these raw data sets did not follow a normal distribution pattern. Overlay analysis was performed between the distribution maps of each heavy metal in soil and rice to estimate the relation values within map units and quantify the spatial differences between heavy metal in soil and rice [35]. A specific dataset was extracted from the interpolated distribution maps by a completely randomized sampling. The density was about 1 km2 per sample so that 1000 samples were collected for further cross-correlogram analysis. Data analysis was carried out using geostatistical and crosscorrelogram methods, which were described by Goovaerts [28] and Isaaks and Srivastava [29]. All analyses in this study were carried out by using SPSS 13.0 for windows, GS+ for win 7.0, and ArcGIS 9.2 version.

N(h)

(h) =

(1)

i=1

where (h) is the semivariance at a given distance h; Z(xi ) is the value of the variable Z at the xi location, and N(h) is the number of pairs of sample points separated by the lag distance h.

3. Results 3.1. Heavy metal concentrations Descriptive statistics for Heavy metal concentrations in soil and rice was presented in Tables 1 and 2, respectively. Among these

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

781

Table 1 Descriptive statistics of heavy metal concentration in paddy soil. Element

Cd Cu Ni Pb Zn a b c d

Sample number

TDa

LogNb LogN Normal LogN LogN

96 96 96 96 96

Mean ± SD (mg/kg)

0.31 41.13 33.89 48.30 137.03

± ± ± ± ±

0.38 19.74 12.69 15.99 33.83

Range (mg/kg)

0.11–3.45 15.78–160.11 9.21–68.16 27.07–140.49 64.97–275.97

C.V. (%)

121.7 48.0 37.4 33.1 24.7

Background value

Second grade standardized value

Number

Percent

Thresholdc (mg/kg)

Number

Percent

Thresholdd (mg/kg)

91 77 47 94 81

94.79 80.21 48.96 97.92 84.38

0.129 30.54 36.48 30.46 107.79

26 14 6 0 5

27.08 14.58 6.25 0 5.21

0.3 50 40 250 200

TD: type of distribution. LogN: lognormal distribution. The background values of soil in Zhejiang province estimated by China National Environmental Monitoring Centre [37]. Recommended by Ministry of Environmental Protection of China [38].

Table 2 Descriptive statistics of heavy metal concentration in rice. Element

Sample number

Cd Cu Ni Pb Zn a b c d e

Mean ± SD (mg/kg)

TDa

b

96 96 96 96 96

LogN Normal LogN NDc Normal

Range (mg/kg)

0.072 ± 0.105 3.09 ± 0.96 0.221 ± 0.234 ND 20.69 ± 4.71

0.002–0.467 0.71–5.79 0.045–1.717 ND 11.45–35.39

C.V. (%)

Standardized value

146.0 31.1 105.9 ND 22.7

Number

Percent

Thresholde (mg/kg)

9 0 ndd 0 0

9.38 0 nd 0 0

0.2 10 nd 0.2 50

TD: type of distribution. LogN: lognormal distribution. ND: not detected. nd: means the threshold value is not defined by Ministry of Health of China [39]. Used as the benchmark of rice contamination recommended by Ministry of Health of China [39].

parameters, coefficient of variability (CV) is the most discriminating factors for describing variability. When CV is less than 10%, it shows low variability; while CV is more than 90%, it shows extensive variability [36]. The result in Tables 1 and 2 showed that the CVs of soil Cd, rice Cd and Ni were 121.7%, 146.0% and 105.9%, respectively; indicating soil Cd, rice Cd and Ni had extensive variability in the study area. The CVs of other heavy metals in this study ranged from 22.7% to 48.0%, indicating moderate variability. In this study, the background values of soils in Zhejiang province [37] and the Environmental Quality Standard for Soils in China [38] were used as the basis for the threshold values for heavy metal pollution in the soil; and Maximum Levels of Contaminants in Foods in China [39] was used for rice. In Table 1, the mean values of heavy metal concentration in soil were higher than the background values, except Ni; however, they were below the second grade standardized critical values, except Cd. The two percentages of samples for each heavy metal exceeding the threshold values indicate that most of the study areas were contaminated with Cd, Cu, Ni, Pb and Zn to different degrees, and some areas posted potential Cu, Ni and Zn risk, especially Cd. It can be observed (Table 2) that all the mean values of heavy metal concentrations in rice were below the threshold values. Thus, the quality of the rice in the study area

was acceptable, except that rice Cd was at slightly elevated levels in some areas. 3.2. Spatial characteristics of heavy metals Semivariance models and their key parameters were given in Table 3. The best-fit theoretical model for the experimental semivariogram was chosen based on the highest decision coefficient value (r2 ) of all theoretical models. Cd in soil, Cd and Zn in rice were best fit with a Gaussian model; Ni in soil, Cu and Ni in rice were best fit with an exponential model; other metals were best fit with a spherical model. Among the parameters, the Nugget/Sill ratio can be regarded as criterion to classify the spatial dependence of soil properties. The relative degree of <25%, 25–75%, and >75% can describe the spatial structure that shows strong, moderate, and weak spatial autocorrelation, respectively [40]. The results (Table 3) showed that Ni concentration in soil had strong spatial dependence, while Cd, Cu, Pb and Zn in soil had moderate spatial dependence. Ordinary Kriging was used to map the spatial distribution of heavy metal concentrations both in soil and rice, and the results were presented clearly in Fig. 2. The accuracy of the interpolation was validated by means of root-mean-square-standardized-error

Table 3 Semivariance models and their parameters for each heavy metal both in soil and rice. Type

Element

Data transformation

Model

Nugget (C0 )

Soil

Cd Cu Ni Pb Zn

Log Cd Log Cu None Log Pb Log Zn

Gaussian Spherical Exponential Spherical Spherical

0.182 0.066 37 0.022 0.030

Rice

Cd Cu Ni Zn

Log Cd None Log Ni None

Gaussian Exponential Exponential Gaussian

1.15 0.668 0.067 15

Nugget/sill (%)

Range (km)

r2

RMSSE

0.442 0.135 256 0.065 0.068

41.2 48.9 14.5 33.4 43.4

39.8 18.6 45.9 13.2 21.7

0.92 0.93 0.98 0.97 0.97

1.000 0.994 1.027 0.993 1.045

4.51 1.337 0.145 60

25.5 50.0 46.2 24.3

42.5 71.9 56.4 61.2

0.99 0.89 0.97 0.98

1.039 0.993 1.006 1.004

Sill (C0 + C)

782

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

Fig. 2. Spatial distribution maps of heavy metal concentration in soil and rice.

(RMSSE). The RMSSE values (Table 3) were all close to 1, suggesting the interpolation in this study was acceptable. In Fig. 2, Soil Cd and Pb showed a similar spatial distribution with high concentrations in the northwest and low concentrations in the southeast (northwest-southeast trend) in the study area. Soil Cu and Zn also showed similar spatial distributions with high concentrations in the north and low concentration in the south areas (north-south trend). Soil Ni had a particular spatial pattern which is distinctly different from other metals. The spatial pattern showed an east-west distribution trend.

Comparing the spatial distribution patterns of heavy metals in soil and rice (Fig. 2), rice Cd had the most similar spatial pattern to soil Cd, rice Cu showed similarity with soil Cu to some degree, and the spatial patterns of Ni and Zn in soil and rice seemed different. 3.3. Transfer of heavy metals in soil–rice system In order to understand the relationship of heavy metals in soil–rice system, the enrichment index (EI) was determined in this study, which was defined as the metal concentration in rice divided

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

783

relation and transfer of heavy metals between soil and rice. Fig. 4 showed that the EI for Cd, Ni and Zn exhibited similar spatial pattern with high values decreasing from west to east; for Cu, the high value was concentrated in the center and northwest part of the study area.

3.4. Influence of soil properties on the availability of heavy metals

Fig. 3. Enrichment index of heavy metal concentration between soil and rice in paddy field. Different capital letters mean significantly different at the 0.05 level.

by that in soil. Enrichment index provides a useful indication of the metal availability from soil to plants [41]. The result of EI for each heavy metal was shown in Fig. 3. The average values for Cd, Cu, Ni and Zn were 0.269, 0.083, 0.007 and 0.160, respectively. The EI varied significantly (P < 0.05) with heavy metals in paddy fields; the availability of heavy metal to rice was generally in the order of Cd > Zn > Cu > Ni. In order to well explain the spatial relationship of heavy metals in soil–rice system, the distributions of EIs were presented in Fig. 4. These distributions are based on the spatial distribution maps of metal concentrations in soil and rice (Fig. 2). The EI distribution maps provide a visual presentation and spatial details of the cor-

In order to understand the effect of soil properties on the availability of heavy metals to rice, cross-correlograms were constructed to quantitatively determine the spatial relationship between the enrichment index (EI) of heavy metal and different soil properties. The results were presented in Fig. 5. The EIs of all heavy metals were positively spatially correlated with soil OM and sand contents, and negatively correlated with soil pH, EC, silt and clay contents. The EIs of Cd and Ni were positively correlated with amorphous Fe oxide, while those of Cu and Zn were negatively correlated with amorphous Fe oxide. The cross-correlograms for Cd and Ni exhibited similar structures, in which, the EIs were significantly (P < 0.01) spatial correlated with soil properties except amorphous Fe oxide; the soil pH, OM and EC had stronger cross-correlations with the EIs than soil texture (sand, silt and clay). In the cross-correlogram for Zn, the EI was significantly (P < 0.01) spatial correlated with all the soil properties. The soil texture, especially sand and silt, were stronger correlated with the EI than soil pH, OM and EC as soil texture showed higher correlation coefficients (r). Soil pH and OM showed relative higher ranges in the cross-correlograms for Cd, Ni and Zn. However, the cross-correlation for Cu was relatively weaker within the range of 10 km than those for Cd, Cu and Zn within the range up to 20 km. Soil texture, EC and Fe oxide were significantly

Fig. 4. Spatial distribution of enrichment index (EI) of heavy metals in soil–rice system.

784

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

Fig. 5. Cross-correlogram for each heavy metal between enrichment index and soil properties (including soil pH, OM, EC, Fe oxide, sand, silt and clay). The r values mean the cross-correlogram value at the separated distance zero, which are equal to the Pearson’s coefficients. The Pearson’s coefficients (r) were significant at the 0.01 level, while NS means not significant.

(P < 0.01) correlated with EI of Cu, whereas soil pH and OM were not spatial correlated with EI of Cu.

4. Discussion 4.1. Heavy metal contamination in paddy fields Compared to the threshold values for heavy metals pollution in soils, this investigation (Table 1) indicated that the studied paddy fields were contaminated to some degree, especially by Cd. The CV data confirmed that Cd was obviously enriched in some parts of the studied soils, and Cu, Ni, Pb and Zn were slightly enriched in some areas. The semivariance model parameters provide information on spatial variability, including intrinsic variability and extrinsic variability. The intrinsic variability is considered to be mainly from natural variation such as parent material, while extrinsic variability is mainly due to human activities, i.e. agricultural practices, industrial sources, road transport and other anthropic activities [22,42–44]. Thus, the nugget/sill ratio in Table 3 suggested that Cd, Cu, Pb and Zn variability with moderate spatial dependence may be determined by both intrinsic factors and extrinsic factors; while Ni variability with strong spatial dependence may controlled by intrinsic factors. Also, the mean value of Ni concentration was below the background value as well as a relative low CV value, indicating Ni in the studied area was less influenced by extrinsic factors. The results were consistent with previous investigations on the sources of heavy metals in agriculture soils [22,42,44], which stated that the variance of Cd, Cu, Pb and Zn were controlled by both intrinsic and extrinsic factors, whereas the variability of Ni were mainly naturally influenced.

Geostatistical methods are useful in visually describing the spatial distribution patterns of the variance which can only be roughly understood by descriptive statistics. As has been shown in Fig. 2, the high concentration for Cd, Cu, Pb and Zn were all located in the northwest of the study area. Based on a survey in the study area, it was clearly found that many industries are distributed in this region, which included machine and electric production, leather and plastic production, dye, among others. These industries are the likely sources for the extrinsic factors leading to these high concentrations of heavy metals in the soil. Particularly, the study area is one of the well-known electronic and electric waste (E-waste) recycling centers in southeast China [45]. Since 1990s, small and open specialized E-waste recycling shelters or yards appeared in the study area, particularly to center in the northwest area (Fig. 1). E-wastes commonly contain several heavy metals such as Cd, Cu, Pb, Hg and Zn. The unregulated disposal and processing of E-waste usually recovers gold and other valuable metals by applying some simple techniques such as burning, melting, by use of acid chemical bath [46]. As a result, heavy metals entered into the environment and accumulated in soil. Fig. 1 presented the main centers of E-waste recycling, and they encompassed some areas surrounding these sites. Fig. 2 illustrated that the high concentrations of heavy metals exhibited in these E-waste areas, indicating the Ewaste activities did affect the contamination of heavy metals in soils. Furthermore, the accumulation of heavy metals in paddy fields may partly be due to the application of agrochemicals. In the past three decades, agrichemicals have been widely applied in the studied paddy fields in order to improve rice production and quality. However, agrichemicals such as fertilizers contain Cd, Cu, Pb and Zn, which are 0.0005–0.5, 0.41–11.6, 0.0008–0.93

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

785

Fig. 6. Spatial distribution maps of soil properties (soil pH, organic matter, electrical conductivity, Fe oxide, soil sand, silt and clay content) in the paddy fields.

and 4.87–348.2 mg/kg, respectively [47]. Therefore, the long-term application of agrichemicals may result in the accumulation of heavy metals in paddy fields so that heavy metal concentrations in most soils exceeded the background values (Table 1). Particularly, the average concentration of Cd was more than twice the background value.

4.2. Spatial relationship of metals in soil–rice system and the influence of soil properties The availability of heavy metals to rice was significantly (P < 0.05) different among heavy metals studied and the accumulation of Cd in rice was highest (Fig. 3). These results are in agreement

786

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787

with previous investigations. Kashem and Singh [8] reported that the accumulation of Cd and Zn was higher than that of Ni in rice. Jung and Thornton [16] reported that heavy metal concentrations of Cd and Zn increased with increasing metal concentrations in paddy soil than that of Cu and Pb. Cd appears to prefer sorbed by rice, though it is not an essential element. As a result, the accumulation of Cd in rice may present relative high potential risk shown in Table 2. Thus, the transfer of Cd from soil to rice may exhibit health concerns. The spatial variability of heavy metals (Fig. 2) presented some similarities of structures for paddy soil and rice; especially for Cd. It illustrated that the heavy metals in rice are spatially correlated with that in soil to some degree. On the other hand, the results suggested that the transfer of heavy metals in soil and rice may be affected by other factors besides the concentrations of heavy metals in soils. The enrichment index provided information on the availability of heavy metals in soil–rice system. The maps of enrichment index (Fig. 4) showed high values were generally located in the west area, further indicating that other factors, i.e., soil properties, may also play some role on influencing the accumulation and availability of heavy metals to rice. It was presented (Fig. 1) that acrisols, lixisols and regosols were dominantly distributed in the west study area. Acrisols and lixisols usually have low pH, CEC content, and contain kaolinite as the main clay minerals. Particularly, regosols have not been well developed, and they are dominant by gravel with high sand content and low CEC content. Because of these soil properties, heavy metals may have stronger mobility in the soil, so that the availability of metals to rice plant was higher in the west study area. Numerous previous studies investigated the effect of soil properties on metals uptake by plants. Soil pH is a significant factor controlling uptake of heavy metals, and above all, is perhaps the most important factor [16,48]. In this study, soil pH increased from west to east; while soil OM and sand content decreased (Fig. 6). Compared to the distribution of soil properties, EIs of Cd and Ni (Fig. 4) showed similar spatial structures with soil pH and OM; EI of Zn was very similar with soil sand content in structure. The results indicated that soil properties did affect the availability of heavy metals in paddy fields. Cross-correlograms further quantified the spatial correlation between the availability of heavy metals (EIs) and soil properties, and provides informative on delineating the size of potential management zones. Fig. 5 revealed that high soil OM and sand would increase the accumulation and availability of heavy metals in rice. In contrast, high soil pH, EC, silt and clay contents decreased the accumulation and availability. Furthermore, soil pH and OM showed relatively higher spatial correlation with EIs for most heavy metals up to a range of 20 km; except Cu. It confirmed that soil pH and organic matter were the most important soil properties influencing the uptake of heavy metals by rice, which was consistent with previous reports [11,12,16]. Soil pH influenced the dissolution of heavy metals, particularly in acid paddy fields; low pH may result in increased solubility and high availability of heavy metals for rice. Soil metals occur in various chemical forms such as exchangeable fraction, Fe–Mn oxide bound fraction, organic bound fraction and residual fraction, among which, the Fe–Mn oxide and organic bound fractions are relatively stable under normal conditions; residual fraction is entrapped within the crystal structure of minerals and represents the least liable fraction [49]. Furthermore, some previous studies reported that the effect of OM on the availability of heavy metals should be due to lower solubility of heavy metals in soil, and the availability of heavy metals commonly decreases with increasing soil OM [4,8]. However, the reverse was found in the present study; i.e., soil OM was positively spatially correlated with EIs of heavy metals (Fig. 5). The paddy field is a complex system with a distinct cycle of flooded and non-flooded conditions, and the organic bound metals could be degraded under

oxidizing conditions [49], thus, the bioavailability of heavy metals in organic fraction may be controlled by binding of heavy metals to soluble organic matter [50]. Due to the redox changes in the paddy fields, the concentration of Fe oxides would be changed. The Fe oxide change may reduce the influence on the availability of heavy metals to rice as shown in Fig. 5 that there were no significant correlations between Fe oxide and EIs of Cd and Ni. However, it is reasonable that Fe oxide was significantly and negatively correlated with EI of Cu and Zn since heavy metals may form stable Fe oxide bound fraction which would decrease the availability of heavy metals to rice. Fig. 5 illustrated that soil EC and texture also played a role in the availability of heavy metals in soil–rice system. Negative correlation was observed between soil clay and EI for all heavy metals since the present of clay may reduce the solubility of heavy metals in soils. The observed cross-correlogram (Fig. 5) indicated that soil properties (pH, OM, EC, Fe oxide and texture) did influence the transfer of heavy metals from the soil to rice, which was not limited to individual sites but encompassed to some areas. The influence of soil properties on the availability of heavy metals in rice varied among heavy metals. Compared to other heavy metals, the accumulation of Cu in rice was weakly correlated with soil pH and OM. In contrast, it was significantly correlated with soil EC, Fe oxide and texture. The relative weak correlations for Cu suggested that the transfer of Cu in paddy soils and rice may be affected by factors other than soil total Cu and soil properties in the present study, and further research is needed. 5. Conclusions Based on the threshold values for heavy metal pollution, the paddy fields of Wenling city showed Cd, Cu, Pb, Zn and Ni contaminations to some degree. Particularly, high concentrations of Cd, Cu, Pb and Zn in the northwest areas may be due to the industrialization, agriculture development and other human activities. The rice grown in the study area showed potential Cd health risk which posed highest availability to rice plant among the studied metals. The higher availability of heavy metals to rice plant was generally located in the west area where the dominant soil types were acrisols, lixisols and regosols with low soil pH, low CEC content, high OM, high sand content. The cross-correlogram results showed strong spatial correlation between heavy metal availability and soil properties. Soil properties including pH, OM, EC, Fe oxide and soil texture, exhibited noticeable effect on the availability of most heavy metals to rice plant in the paddy field. Among these properties, soil pH and OM generally had the most significant effect. Soil properties studied had relatively weak effect on the Cu availability; soil pH and OM seem no effect. The cross-correlogram results also demonstrated that the optimization of management zone may be necessary. It may provide basic information for rational site-specific management based on the soil properties in paddy fields. Acknowledgements This research was sponsored by the National Basic Research Program of China (2005CB121104), the National Natural Science Foundation of China (40601051), the Fundamental Research Funds for the Central Universities (JD09020, KYJD09021), and the China Scholarship Council. The author would like to express our appreciation to the laboratory group for their assistance in some of the analysis of the samples. References [1] D.C. Adriano, Trace Elements in Terrestrial Environments: Biogeochemistry, Bioavailability, and Risks of Metals, Springer, New York, 2001.

K. Zhao et al. / Journal of Hazardous Materials 181 (2010) 778–787 [2] P.M. Huang, G.R. Gobran, Biogeochemistry of Trace Elements in the Rhizosphere, Elsevier, Amsterdam, The Netherlands, 2005. [3] R. Raghunath, R.M. Tripathi, A.V. Kumar, A.P. Sathe, R.N. Khandekar, K.S.V. Nambi, Assessment of Pb, Cd, Cu, and Zn exposures of 6 to 10-year-old children in Mumbai, Environ. Res. 80 (1999) 215–221. [4] Z.W. Li, L.Q. Li, G.X. Pan, J. Chen, Bioavailability of Cd in a soil–rice system in China: soil type versus genotype effects, Plant Soil 271 (2005) 165–173. [5] F.R. Zeng, Y. Mao, W.D. Cheng, F.B. Wu, G.P. Zhang, Genotypic and environmental variation in chromium, cadmium and lead concentrations in rice, Environ. Pollut. 153 (2008) 309–314. [6] W.D. Cheng, G.P. Zhang, H.G. Yao, P. Dominy, W. Wu, R.Y. Wang, Possibility of predicting heavy-metal contents in rice grains based on DTPA-extracted levels in soil, Commun. Soil Sci. Plan. 35 (2004) 2731–2745. [7] N. Coen, C. Mothersill, M. Kadhim, E.G. Wright, Heavy metals of relevance to human health induce genomic instability, J. Pathol. 195 (2001) 293–299. [8] M.A. Kashem, B.R. Singh, Metal availability in contaminated soils: II. Uptake of Cd, Ni and Zn in rice plants grown under flooded culture with organic matter addition, Nutr. Cycl. Agroecosys. 61 (2001) 257–266. [9] H.Y. Liu, A. Probst, B.H. Liao, Metal contamination of soils and crops affected by the Chenzhou lead/zinc mine spill (Hunan, China), Sci. Total Environ. 339 (2005) 153–166. [10] W.X. Liu, L.F. Shen, J.W. Liu, Y.W. Wang, S.R. Li, Uptake of toxic heavy metals by rice (Oryza sativa L.) cultivated in the agricultural soil near Zhengzhou City, People’s Republic of China, Bull. Environ. Contam. Toxicol. 79 (2007) 209–213. [11] M.B. Kirkham, Cadmium in plants on polluted soils: effects of soil factors, hyperaccumulation, and amendments, Geoderma 137 (2006) 19–32. [12] C.N. Reddy, W.H.J. Patrick, Effect of redox potential and pH on the uptake of cadmium and lead by rice plants, J. Environ. Qual. 6 (1977) 259–262. [13] M. Haldar, L.N. Mandal, Influence of soil moisture regimes and organic matter application on the extractable Zn and Cu content in rice soils, Plant Soil 53 (1979) 203–213. [14] M.A. Kashem, B.R. Singh, Metal availability in contaminated soils: I. Effects of flooding and organic matter on changes in Eh, pH and solubility of Cd, Ni and Zn, Nutr. Cycl. Agroecosys. 61 (2001) 247–255. [15] K.S. Sajwan, W.L. Lindsay, Effects of redox on zinc deficiency in paddy rice, Soil Sci. Soc. Am. J. 50 (1986) 1264–1269. [16] M.C. Jung, I. Thornton, Environmental contamination and seasonal variation of metals in soils, plants and waters in the paddy fields around a Pb–Zn mine in Korea, Sci. Total Environ. 198 (1997) 105–121. [17] M. Haldar, L.N. Mandal, Effect of phosphorus and zinc on the growth and phosphorus, zinc, copper, iron and manganese nutrition of rice, Plant Soil 59 (1981) 415–425. [18] J.X. Li, X.E. Yang, Z.L. He, G. Jilani, C.Y. Sun, S.M. Chen, Fractionation of lead in paddy soils and its bioavailability to rice plants, Geoderma 141 (2007) 174– 180. [19] M.I. Rubio, I. Escrig, C. Martínez-Cortina, F.J. López-Benet, A. Sanz, Cadmium and nickel accumulation in rice plants. Effects on mineral nutrition and possible interactions of abscisic and gibberellic acids, Plant Growth Regul. 14 (1994) 151–157. [20] Q.R. Wang, Y.S. Cui, X.M. Liu, Y.T. Dong, P. Christie, Soil contamination and plant uptake of heavy metals at polluted sites in China, J. Environ. Sci. Health Part A: Toxic/Hazard. Subst. Environ. Eng. 38 (2003) 823–838. ´ ´ M. Dolenec, Major and trace [21] N. Rogan, T. Dolenec, T. Serafimovski, R. Jacimovi c, elements in rice seeds from Koˇcani Field, Macedonia, Acta Chim. Slov. 54 (2007) 623–634. [22] T. Chen, X.M. Liu, X. Li, K.L. Zhao, J.B. Zhang, J.M. Xu, J.C. Shi, R.A. Dahlgren, Heavy metal sources identification and sampling uncertainty analysis in a field-scale vegetable soil of Hangzhou, China, Environ. Pollut. 157 (2009) 1003–1010. [23] R. Webster, M.A. Oliver, Geostatistics for Environmental Scientists, John Wiley & Sons Ltd., Chichester, 2001. [24] A.N. Kravchenko, K.D. Thelen, D.G. Bullock, N.R. Miller, Relationship among crop grain yield, topography, and soil electrical conductivity studied with crosscorrelograms, Agron. J. 95 (2003) 1132–1139. ˜ F.J. Pierce, H.M. Melakerhan, Spatial analysis of soybean yield in [25] F. Avendano, relation to soil texture, soil fertility and soybean cyst nematode, Nematology 6 (4) (2004) 527–545.

787

[26] A. Stein, J. Brouwer, J. Bouma, Methods for comparing spatial variability patterns of millet yield and soil data, Soil Sci. Soc. Am. J. 61 (1997) 861–870. [27] Agricultural Chemistry Committee of China, Conventional Methods of Soil and Agricultural Chemistry Analysis, Science Press, Beijing, 1983 (in Chinese). [28] P. Goovaerts, Geostatistics for Natural Resources Evaluation, Oxford University Press, New York, 1997. [29] E.H. Isaaks, R.M. Sivastava, An Introduction to Applied Geostatistics, Oxford University Press, New York, 1989. [30] B. Burgos, E. Madejón, A. Pérez-de-Mora, F. Cabrera, Spatial variability of the chemical characteristics of atrace-element-contaminated soil before and after remediation, Geoderma 130 (2006) 157–175. [31] T. Chen, X.M. Liu, M.Z. Zhu, K.L. Zhao, J.J. Wu, J.M. Xu, P.M. Huang, Identification of trace element sources and associated risk assessment in vegetable soils of the urban–rural transitional area of Hangzhou, China, Environ. Pollut. 151 (2008) 67–78. [32] X.M. Liu, K.L. Zhao, J.M. Xu, M.H. Zhang, B. Si, F. Wang, Spatial variability of soil organic matter and nutrients in paddy fields at various scales in southeast China, Environ. Geol. 53 (2008) 1138–1147. [33] C.V. Deutsch, A.G. Journel, Geostatistical software Library and User’s Guide, Oxford University Press, New York, 1998. [34] I. Clark, W.V. Harper, Practical Geostatistics, Ecosse North America, Columbus, Ohio, USA, 2000. [35] J. Wu, W.A. Norvell, D.G. Hopkins, R.M. Welch, Spatial variability of grain cadmium and soil characteristics in a durum wheat field, Soil Sci. Soc. Am. J. 66 (2002) 268–275. [36] X.Y. Zhang, Y.Y. Sui, X.D. Zhang, K. Meng, S.J. Herbert, Spatial variability of nutrient properties in black soil of northeast China, Pedosphere 17 (1) (2007) 19–29. [37] China National Environmental Monitoring Centre, Soil Element Background Values in China, China Environmental Science Press, Beijing, China, 1990. [38] Ministry of Environmental Protection of China, Environmental Quality Standard for Soils, 1995, http://www.chinaep.net/hjbiaozhun/hjbz/hjbz017.htm (GB15618-1995). [39] Ministry of Health of China, Maximum Level of Contaminants in Food, 2005, http://www.moh.gov.cn/moh/main?fid=open&fun=show dbquery pdf& nid=104 (GB2762-2005). [40] C.A. Cambardella, T.B. Moorman, J.M. Novak, T.B. Parkin, D.L. Karlen, R.F. Turco, A.E. Konopka, Field-scale variability of soil properties in Central Iowa soils, Soil Sci. Soc. Am. J. 58 (1994) 1501–1511. [41] C.G. Lee, H.T. Chon, M.C. Jung, Heavy metal contamination in the vicinity of the Daduk Au–Ag–Pb–Zn mine in Korea, Appl. Geochem. 16 (2001) 1377–1386. [42] A. Facchinelli, E. Sacchi, L. Mallen, Multivariate statistical and GIS-based approach to identify heavy metal sources in soil, Environ. Pollut. 114 (2001) 313–324. [43] S.S. Huang, Q.L. Liao, M. Hua, X.M. Wu, K.S. Bi, C.Y. Yan, B. Chen, X.Y. Zhang, Survey of heavy metal pollution and assessment of agricultural soil in Yangzhong district, Jiangsu Province, China, Chemosphere 67 (2007) 2148–2155. [44] J.A.R. Martín, M.L. Arias, J.M.G. Corbí, Heavy metals contents in agricultural topsoils in the Ebro basin (Spain): application of the multivariate geostatistical methods to study spatial variations, Environ. Pollut. 144 (2006) 1001–1012. [45] G. Zhao, Y. Xu, G. Han, B. Ling, Biotransfer of persistent organic pollutants from a large site in China used for the disassembly of electronic and electrical waste, Environ. Geochem. Health 28 (2006) 341–351. [46] J.J. Fu, Q.F. Zhou, J.M. Liu, W. Liu, T. Wang, Q.H. Zhang, G.B. Jiang, High levels of heavy metals in rice (Oryza sativa L.) from a typical E-waste recycling area in southeast China and its potential risk to human health, Chemosphere 71 (2008) 1269–1275. [47] Q.C. Wang, Z.W. Ma, Heavy metals in chemical fertilizer and environmental risks, Rural Eco-Environ. 20 (2) (2004) 62–64 (in Chinese). [48] N.T. Basta, J.A. Ryan, R.L. Chaney, Trace element chemistry in residual-treated soil: key concepts and metal bioavailability, J. Environ. Qual. 34 (2005) 49–63. [49] A. Tessier, P.G.C. Campbell, M. Bisson, Sequential extraction procedure for the speciation of particulate trace metals, Anal. Chem. 51 (1979) 844–851. [50] M. Jalali, Z.V. Khanlari, Environmental contamination of Zn, Cd, Ni, Cu and Pb from industrial areas in Hamadan Province, western Iran, Environ. Geol. 55 (2008) 1537–1543.