^{1}

^{3}

^{1}

^{2}

^{1}

^{*}

^{1}

^{2}

This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).

This study explored the spatial pattern of heavy metals in Beijing agricultural soils using Moran’s I statistic of spatial autocorrelation. The global Moran’s I result showed that the spatial dependence of Cr, Ni, Zn, and Hg changed with different spatial weight matrixes, and they had significant and positive global spatial correlations based on distance weight. The spatial dependence of the four metals was scale-dependent on distance, but these scale effects existed within a threshold distance of 13 km, 32 km, 50 km, and 29 km, respectively for Cr, Ni, Zn, and Hg. The maximal spatial positive correlation range was 57 km, 70 km, 57 km, and 55 km for Cr, Ni, Zn, and Hg, respectively and these were not affected by sampling density. Local spatial autocorrelation analysis detected the locations of spatial clusters and spatial outliers and revealed that the pollution of these four metals occurred in significant High-high spatial clusters, Low-high, or even High-low spatial outliers. Thus, three major areas were identified and should be receiving more attention: the first was the northeast region of Beijing, where Cr, Zn, Ni, and Hg had significant increases. The second was the southeast region of Beijing where wastewater irrigation had strongly changed the content of metals, particularly of Cr and Zn, in soils. The third area was the urban fringe around city, where Hg showed a significant increase.

Heavy metals often accumulate to excess in agricultural soils due to natural processes and anthropogenic activities. Since these can cause adverse effects on the environment, heavy metal soil pollution is an urgent problem. Successful assessment and remediation of heavy metal pollution in soils will depend on the understanding of their spatial variability and the relationships between heavy metals and the factors leading to pollution.

Both geostatistics and spatial autocorrelation statistics (Moran’s I) are the methods used for exploring the spatial pattern of heavy metals. Generally, geostatistics is widely used [

For heavy metals in Beijing soils, there are some related studies. However, the emphasis of such studies just covered the urban-rural transition zone, periurban, and rural zones, and/or had a relatively few sampling points [

Therefore, the main objectives of this research were: (1) to explore the spatial variability of heavy metals in Beijing agricultural soils using the spatial autocorrelation statistic, including the influence of different weight matrixes and sampling density on spatial autocorrelation; and (2) to reveal the local spatial patterns of heavy metals in Beijing agricultural soils in order to identify the potential risk areas for heavy metal pollution and its possible causes.

Beijing municipality is located in the northwest region of the north China plain, between longitude 115°25′–117°30′E and latitude 39°28′–41°05′N, and covers an estimated area of 1.6 × 10^{4} km^{2}. The elevation of Beijing ranges from 2,250 m in the northwest to 9 m in the southeast. Mountains cover approximately 62% of the entire area and plains cover the remaining 38% (^{9} m^{3} but has decreased to 1.3 × 10^{9} m^{3} since the end of the last century. The Yongding River provides water mainly for industry, the Chaobai River mainly for resident living, and the Beiyun River plays a dual role in wastewater drainage from industry and human living activities.

The primary types of agricultural soil in the area include Cab Ustic Luvisols, Hap Ustic Luvisols and Och Aquic Cambisols. According to the second agricultural census in 2006, cultivated land occupied 2.32 × 10^{5} hm^{2}, and orchard land 1.22 × 10^{5} hm^{2}, together covering 21% the total area of Beijing. The fertilizer input was 517.5 kg/hm^{2}, about 1.44 times of the national average level, and the pesticide inputs were 8.11 kg/hm^{2}, or about 2.33 times of the country level. The overuse of fertilizer and pesticides has imposed heavy environmental pressures on the area. The metal mineral resources were primarily Fe, Ag, Zn, and Pb elements, and ironstone was exploited on a large-scale, mainly distributed in the Huairou district and Miyun county (

To investigate the pollution status of heavy metals in Beijing agricultural areas, a large–scale soil sampling project was conducted after the crop harvest in the Autumn of 2006. According to the agricultural land distribution and land use type maps of Beijing, a non-uniform distribution of the stratified sampling technique was adopted to collect samples and ensure the representativeness of sample. The sampling process was divided into three steps to collect a total of 1,018 samples. First, 231 soil samples were collected from the agricultural soils in the entire study area, with uniform sampling being the low sampling density (C). Secondly, another 360 soil samples were added from areas with more agricultural soils to create the medium sampling density (M). Third, 427 soil samples were further collected on the basis of the two previous samplings and the agricultural soils to make a high sampling density (F). General ideal is that more areas of agricultural land more sampling points.

For each sample, five surface soil (0∼20 cm) sites were sampled within 10 × 10 m square areas and then mixed. Global Positioning System was used to precisely locate each sampling position (latitude and longitude); and a total of 1 kg of mixed soil per sample was collected. All soil samples were collected using a stainless steel spade and a scoop made from bamboo and then stored in polyethylene bags.

The soil samples were air-dried, crushed in an agate mortar, and then passed through a 100-mesh nylon sieve. The concentrations of eight heavy metals, including Cr, Ni, Cu, Zn, As, Cd, Pb, and Hg, were analyzed in the soil samples following the Chinese Environmental Quality Standard for Soils (GB15618-1995).

Spatial autocorrelation is an assessment of the correlation of a variable with reference to its spatial location and it deals with the attributes and the locations of the spatial features [

Global Moran’s I is a global test statistic for spatial autocorrelation, which is based on cross-products for measuring attribute association. It is calculated for _{i}_{j}_{ij}

The weight matrix depicts the relationship between an element and its surrounding elements. The weights are based on contiguity relations or distance. In a weight matrix based on contiguity, a value unequal to zero in the matrix represents pairs of elements with a certain contiguity relation and a zero represents pairs without contiguity relation. The rook case and queen case are the typical examples for contiguity relation. The first takes only four neighbours into account with common boundaries, and the latter takes into account all eight surrounding cells, including common boundaries and common corners. In a distance-based weight matrix, a threshold distance is specified such that all locations within the given distance are considered to be “neighbors”. Alternatively, the k-nearest neighbor weight matrix is also based on distance, which is computed as the distance between a point and the number (k) of nearest neighbor points.

The value of Moran’s I will vary between 1 and −1. A higher positive Moran’s I implies that values in neighboring positions tend to cluster, while a low negative Moran’s I indicates that high and low values are interspersed. When Moran’s I is near zero, there is no spatial autocorrelation, meaning that the data are randomly distributed [

The Moran scatter plot describes the distribution of all observation values (x-axis) in relation to their neighbors (y-axis). Observations in the lower left (Low–low) and upper right (High–high) quadrant represent potential spatial clusters (values surrounded by similar neighbors), whereas observations in the upper left (Low–high) and lower right (High–low) suggest potential spatial outliers (values surrounded by dissimilar neighbors) [

Local Moran’s I is a local test statistic for spatial autocorrelation, and identifies the autocorrelation between a single location and its neighbors. It is computed as follows:

The notations in

A map showing locations with significant Local Moran statistics, classified by types of spatial correlation, is defined as a Local Indicators of Spatial Autocorrelation (LISA) cluster map [

Soil samples were stored using the ArcView 3.2 software to create a spatial database. The Thiessen polygons of samples were created and all spatial analysis including global and local spatial autocorrelation was carried out using Geoda095i software [

Eight different spatial weight matrixes were used to impose a neighborhood structure on the 1,018 samples and assess the spatial autocorrelation of eight heavy metals. The results of global Moran’s I values are given in

Generally, the numbers of neighbours using the queen criterion will be equal to or greater than that using the rook criterion. However,

Because the selection of spatial weigh was empirical, as well as the same weight matrix under a certain distance limit was assigned to all points, there may be had a certain impact on the spatial autocorrelation of the heavy metals. If the spatial weights based on decay distance were designed, the results of the influence of spatial weight on spatial autocorrelation of heavy metals may be more reasonable.

There was also a certain part of the samples in the right lower and left upper quadrants, indicating that negative spatial autocorrelation could not be neglected. With the decrease in spatial autocorrelation coefficients, the scatter plot tended to became further disaggregated, and these samples were far from the Moran’s I regression line and strongly influenced the global spatial autocorrelation, particularly for Cr and Zn, indicating some local nonstationarity (

Moran’s I values can be plotted against distance classes, called a spatial correlogram [

Moran’s I for Cr had a distinct difference within 13 km, then the difference disappeared for the three density levels [

The spatial correlogram analysis revealed that Cr, Ni, and Zn had the similar sampling density effect. For Cr, Ni, and Zn, the higher sampling density enhanced the spatial dependence. In contrast, it may be the existence of extreme value, the spatial dependence of Hg at higher sampling density became weaker. These scale effects existed within a certain distance. Moreover, the spatial correlogram can help to find the maximal spatial positive correlation range and the suitable neighborhoods for spatial dependence analysis. Therefore, LISA was further used to identify the detail spatial variability for the four metals at the F density level.

The LISA indicates the spatial variability details. The distance where the spatial dependence of Cr, Ni, Zn, and Hg was the strongest at F density level is selected to reveal the local spatial pattern (High-high, Low-low, High-low, Low-high and no significance). The distance weight matrix was 6 km, 4 km, 4 km, and 4 km for Cr, Ni, Zn, and Hg, respectively.

Although the four heavy metals had a significant global spatial positive correlation, more than half of the samples of the four metals had no significant spatial pattern (

The samples with Cr pollution were only 0.69% and all of these occurred in the significant High-high spatial pattern. Zn pollution was less (only 0.1% samples) in the no significant spatial pattern type (

Compared with the spatial randomness, the significant spatial patterns of these heavy metals demonstrated that the underlying enrichment processes were more stable, and as such, would present more difficulties for their remediation. In addition, in soil heavy metal evaluation, the outliers may represent the potential pollution areas, such as Ni and Hg in this study. If further spatial interpolation will be produced, the outliers cannot be deleted arbitrarily and should adopt more complicated geostatistics approach.

The LISA map can further detect the locations of the interesting spatial patterns for heavy metals. The northeast region was strongly influenced by High-high pattern of Cr, Ni [

However, because the LISA map was generated based on the soil samples, the boundaries between different spatial pattern types are easily confused and unreadable. In future research, the method of zoning should be introduced for improving to distinguish the spatial pattern types of boundaries. Such as, LISA clusters map, with other possible driving factors maps in GIS software, can also quantify their spatial relationships to confirm and refine their effects. Their distributions could be used to delineate the potential monitoring and remediation zones. Moreover, these zones can assist in developing measures and policies that can be responsive to the spatial variations and pollution processes.

Compared with geostatistics, although spatial autocorrelation analysis cannot be used to estimate unobserved points, global and local Moran’s I analysis have their advantages. The strength and significance of spatial dependence could be easily compared and tested. Moreover, the spatial correlogram can describe the changes of spatial dependence with distance, which indicate the maximal spatial positive correlation range and the suitable neighborhoods for spatial dependence analysis. In addition, local Moran’s I analysis can help to detect spatial outliers or hot spots. Therefore, it is an effective exploratory spatial analysis technique for regional variables.

This study explored spatial pattern of heavy metals in the entire Beijing agricultural soils based on 1,018 samples collected in 2006 using Moran’s I analysis. The four elements Cr, Ni, Zn, and Hg had significant and positive spatial correlations with large Moran’s I values. The Moran’s I values were affected by spatial weight matrix. Cr, Ni, and Zn had similar sampling density effects. Higher sampling density enhance the spatial dependence, whereas, higher sampling density may reduce the spatial dependence for Hg. In addition, the maximal spatial positive correlation ranges of the four metals did not change at different sampling density. The global Moran’s I of the four metals was scale-dependent on distance, initially taking stronger spatial dependence with the increase of the distance, then becoming weaker with further expansion of the distance. It is worth noting that the scale effect existed in a certain distance.

The local spatial autocorrelation analysis revealed that the four metals all had important High-high pattern, and Low-high and High-low spatial outliers, indicating strong enrichment processes for the four heavy metals in Beijing agricultural soils. Thus, these areas should be receiving more attention: the northeast and southeast region of Beijing, where significant increases in Cr, Zn, Ni, and Hg occurred, as well as the urban fringe around city where Hg showed a significant increase. The global Moran’s I was proved to be a useful measure of overall clustering, while the local Moran’s I was an important tool for detecting local spatial patterns for possible polluting areas or interesting patterns for further monitoring. Therefore, spatial autocorrelation analysis can be a useful method to explore the spatial pattern of heavy metals.

The research was supported by the National Key Technology R&D Program (Grant Nos. 2006BAD10A06 and 2004BA617B04) from the China Ministry of Sciences and Technology and Beijing agricultural scientific programs support. The authors are grateful to the Agricultural Environmental Monitoring Station of Beijing for their soil sampling and analysis.

The study area.

Distribution of soil samples at the three densities.

Moran scatter plot for Cr, Ni, Zn, and Hg.

The spatial correlograms of the metals at three sampling density.

LISA cluster maps for heavy metals.

Global spatial autocorrelation coefficient (global Moran’s I value) based on different spatial weighs for heavy metals.

First order rook contiguity | 0.495 |
0.317 |
0.026 | 0.282 |
0.077 |
0.119 |
0.051 |
0.290 |

First order queen contiguity | 0.495 |
0.317 |
0.026 | 0.282 |
0.077 |
0.119 |
0.051 |
0.290 |

4-nearest neighbors | 0.541 |
0.327 |
0.006 | 0.287 |
0.071 |
0.124 |
0.036 | 0.283 |

5-nearest neighbors | 0.521 |
0.320 |
0.018 | 0.277 |
0.080 |
0.125 |
0.047 |
0.275 |

6-nearest neighbors | 0.513 |
0.321 |
0.018 | 0.265 |
0.084 |
0.122 |
0.049 |
0.271 |

7-nearest neighbors | 0.498 |
0.315 |
0.018 | 0.253 |
0.078 |
0.118 |
0.045 |
0.263 |

8-nearest neighbors | 0.486 |
0.309 |
0.018 | 0.246 |
0.073 |
0.119 |
0.047 |
0.258 |

4km distance band | 0.472 |
0.318 |
0.024 | 0.293 |
0.090 |
0.127 |
0.056 |
0.272 |

Significant at the 0.05 level.

Sample percent of local spatial pattern types of LISA analysis (%).

No significance | 56.09 | 69.94 | 66.70 | 67.78 |

High-high | 14.34 | 7.07 | 8.74 | 9.63 |

Low-low | 22.20 | 12.48 | 13.46 | 11.30 |

Low-high | 3.05 | 7.96 | 7.56 | 8.35 |

High-low | 4.32 | 2.55 | 3.54 | 2.95 |

Sample pollution status distribution in local spatial pattern types (%).

Cr | Polluted | 0.69 | ||||

Unpolluted | 56.09 | 13.65 | 22.20 | 3.05 | 4.32 | |

Ni | Polluted | 2.65 | 0.79 | 0.49 | ||

Unpolluted | 67.29 | 6.29 | 12.48 | 7.47 | 2.55 | |

Zn | Polluted | 0.10 | ||||

Unpolluted | 66.60 | 8.74 | 13.46 | 7.56 | 3.54 | |

Hg | Polluted | 3.44 | 2.26 | 0.49 | 0.10 | |

Unpolluted | 64.34 | 7.37 | 11.30 | 7.86 | 2.85 |