Spatial Variation in Microbial Community in Response to As and Pb Contamination in Paddy Soils Near a Pb-Zn Mining Site

Mining activity is a growing environmental concern as it contributes to heavy metals (HMs) pollution in agricultural soils. Microbial communities play an important role in the biogeochemical cycling of HMs and have the potential to be used as bioindicators. Arsenic (As) and lead (Pb) are the most hazardous HMs and are mainly originated from mining activities. However, spatial variation in microbial community in response to As and Pb contamination in paddy soils remains overlooked. In this study, the biological and chemical properties of sixteen soil samples from four sites (N01, N02, N03, and N04) near a Pb-Zn mining site at different As and Pb levels were examined. The results showed that soil pH, total As and Pb, bioavailable As and Pb, nitrate-nitrogen (NO3 −-N) and ammonia-nitrogen (NH4 +-N) were the most important factors in shaping the bacterial community structure. In addition, significant correlations between various bacterial genera and As and Pb concentrations were observed, indicating their potential roles in As and Pb biogeochemical cycling. These findings provide insights into the variation of paddy soil bacterial community in soils co-contaminated with different levels of As and Pb.


INTRODUCTION
Mining mineral ore has both significant positive and negative impacts on mining areas. Mining resources have been endorsed as a mean to stimulate socioeconomic development and reduce poverty (Li et al., 2014). However, mining activities generate waste, such as mine tailings, acid mine drainages and fly ashes that contain harmful trace heavy metals (HMs) (Chen et al., 2018;Liu et al., 2019). HMs can be transported from mining sites to neighboring agricultural soils through surface runoff and atmospheric dust deposition (Mu et al., 2020). Consequently, most neighboring agricultural soils are contaminated with HMs, and the soil thereafter serves as a long-term sink for HMs (Hahn et al., 2019;Cheng et al., 2020).
In China, mining activities are prevalent and this has led to serious soil HMs contamination nationwide (Zhong et al., 2020). According to the Chinese National Soil Contamination Survey Report, approximately 33.4% of soil around mining sites were contaminated with HMs (Mee and Mnr, 2014). Among them, arsenic (As) and lead (Pd) commonly co-exist in high concentrations due to mining and smelting of mineral ores (Wang et al., 2020a;Fernandez-Macias et al., 2020), and threaten soil, plant and human health. In paddy soil, excess As and Pb can reduce soil quality, stunt plant growth, and ultimately affect human health through the food chain (Du et al., 2018;Rai et al., 2019). For example, Wang et al. (2018) reported that the concentrations of As and Pb (up to 2749 and 624 μg kg −1 , respectively) in rice grains grown in fields neighboring mining sites were significantly higher than the threshold value of 200 μg kg −1 (GB2762-2017). Therefore, the investigation of the HMs pollution in paddy soils surrounding mining areas can assess the risk of HMs to human health and provide the scientific basis for the remediation of HMs contaminated soils.
HMs accumulation in soils affects both the soil physiochemical properties and the microbial community structure over short-and long-term periods (Zhu et al., 2019). Zhang et al. (2015) reported that HMs in paddy soil may have adverse effects on the microbial activities and nutrients bioavailability. Pb contaminations have been shown to inhibit soil respiration and reduce microbial biomass . Similarly, dynamic changes were observed in the soil microbial community as a result of As contamination (Wang et al., 2016). Furthermore, in the soil-rice system, As contamination can markedly influence the iron (Fe) plaque-associated microbial community structure and metabolic potential (Hu et al., 2019). Although many attempts were tried to remediate As and Pb pollution in soils neighboring mining sites, few have reported the combined effect of As and Pb contamination on the soil microbial community She et al., 2021). Molecular techniques such as high-throughput sequencing techniques have been used to evaluate the changes in microbial community due to HMs pollution (Ros et al., 2009;Gołębiewski et al., 2014). Changes in soil microbial community structure and diversity could serve as biomarkers to study the effect of HMs on soil health (Vinhal-Freitas et al., 2017;Li et al., 2020b;Di Cesare et al., 2020). However, it should be noted that some studies have reported that HMs contamination had limited effect on the microbial community structure (Lorenz et al., 2006). Other soil factors such as pH, organic matter (OM), moisture, clay content and nutrient content can have a profound influence on the microbial activities, which might play more important role in reshaping microbial communities than the HMs concentrations (Jiang et al., 2019). However, studies on the difference between microbial communities in paddy soils with different distances from the mining site and the responses of microorganisms to HMs and soil properties are limited. HMs polluted paddy soil with various distances from the mining site may present different soil characteristics and HMs concentration gradients, then the most important factor in shaping the microbial communities remains unknown.
For this purpose, this present study was conducted with soil samples from four sites along a Pb-Zn mining site to 1) explore the influence of the different distance from the mining site on the physiochemical properties of As and Pb polluted paddy soils; 2) evaluate the combined effects of As and Pb pollution on soil bacterial community structure; 3) examine the correlativity between soil physiochemical properties and bacterial communities. Chemical analysis coupled with Miseq sequencing techniques were applied to achieve the goals. The observations and findings in this study could be used to develop more effective bioremediation strategies for the polluted paddy soil in mining area.

Study Area and Sampling
The study area is located near a Pb-Zn mining site (29°59′35.00″ N, 120°45′55.20″ E) in Shangyu city, Zhejiang province, China ( Figure 1). The study area has a typical subtropical climate, densely surrounded by paddy field. A total of 16 top-soil samples (0-20 cm) were collected from the four sites referred to as N01, N02, N03, and N04. The four sites are all paddy fields and one growth cycle is performed over one year. Soil samples (four in total, approximately 300 g each) were collected from random points at each site and were mixed uniformly to form a composite sample. Four replicate composite samples were collected from each site (denoted as samples N01-1, N01-2, N01-3, N01-4, and so on). Soil samples were properly sealed and transported to the laboratory on ice. The collected soil samples were divided into two parts; one part was stored at −80°C for molecular analyses, and the other part was used for the physicochemical analysis.

Chemical Analysis
Soil samples from N01, N02, N03, and N04 site were analyzed for different physicochemical parameters including soil pH, OM, moisture content, dissolved organic carbon (DOC), available nitrate-nitrogen (NO 3 − -N) and ammonia-nitrogen (NH 4 + -N) concentrations, total As, Pb and other HMs concentrations, total sulfur (S) and SO 4 2-. Soil pH and OM were measured following methods described in our pervious study (Zou et al., 2018). Moisture content was measured by drying fresh soil samples overnight in an oven at 105°C (O'Kelly, 2004). DOC was extracted with ultrapure water from fresh soil at a ratio of 1:2 (soil: water) and the concentration was analyzed with an automated N/C analyzer (Multi N/C 2100, Analytik Jena, Germany) (Xu et al., 2015). Available NH 4 + -N and NO 3 − -N concentrations were determined in 1 M KCl extracts (1:5, w/v) using the colorimetric methods (Stock, 1983). Soil SO 4 2was extracted by shaking freeze-dried soils with ultrapure water at a ratio 1:25 (soil:water) and analyzed by ion chromatography (Dionex, United States) (Xu et al., 2015). In order to determine the total Fe, S and other HMs, 0.2 g soil was digested using a concentrated acid mixture (5 ml of HNO 3 , 3 ml of HClO 4 , and 5 ml of HF) and was analyzed by inductively coupled plasma-optical emission spectrometry (ICP-OES, iCAP 6300, Thermo, United States). For total As analysis, 0.2 g air-dried soil was digested with 10 ml diluted aqua regia (aqua regia: water 1:1, v/v) at 100°C for 2 h, and As concentration in digested samples were determined by atomic fluorescence spectrometry (AFS-9130, Beijing Jitian Instrument Company, China) (Zou et al., 2018). In order to determine the accuracy of the total metal analysis, a reference soil (GBW 07429, the National Research Center for standards in China) was used. The extract efficiencies ranged from 97.8 to 104.6% of the target values.

Bioavailable As and Pb Fraction in Soil
As and Pb were fractionated according to a 4-step modified European Community Bureau of Reference (BCR) sequential extraction procedure including F1: extracted with 0.11 M acetic acid (exchangeable fraction); F2: extracted with 0.5 M hydroxylammonium chloride (reducible fraction); F3: extracted with 8.8 M hydrogen peroxide followed by 1.0 M ammonium acetate (oxidizable fraction); F4: extracted with hydrofluoric acid followed by Aqua Regia (nitric acid + hydrochloric acid; 1:3 v/v) (residual fraction) (Rauret et al., 1999). The sum of exchangeable and reducible fractions was defined as bioavailable fractions according to a previous study (Roosa et al., 2014). The extracted solutions were analyzed by ICP-OES for different As and Pb fractions. The difference between the sum of each phase concentration and the total concentrations of As and Pb were compared to verify the accuracy of the sequential extraction procedure.

Microbial Biomass Carbon
Microbial biomass carbon (MBC) in soils was estimated by chloroform fumigation-extraction method (Vance et al., 1987;Brookes, 1995). Briefly, 6 subsamples of 5 g moist soil were weighed into different glass centrifuge tubes and three of them were fumigated with ethanol-free chloroform. Both fumigated and non-fumigated samples were incubated for 24 h at 25°C in the dark. After incubation, all the samples were extracted with 25 ml of 0.5 M K 2 SO 4 and filtered. Then, the fumigated samples were heated in a 100°C water bath for 60 min and lost moisture was replenished. Total dissolved carbon in the extracts was determined by an automated N/C analyzer (Multi N/C 2100, Analytik Jena, Germany). MBC was calculated according to Eq. 1 (Jaenicke et al., 2011): where C f total dissolved carbon extracted from fumigated soil, C nf total dissolved carbon extracted from nonfumigated soil and k EC 0.45.

Statistical Analysis
Statistical analyses for the experimental data were performed using the SPSS ® 19.0 (SPSS, United States) software. The significance of the treatment effects was assessed by analysis of variance (ANOVA), followed by comparisons between treatment means using the least significant difference (LSD) at p < 0.05. Alpha diversity index including Chao1, Shannon, Simpson, ACE and Coverage were calculated using QIIME software (V1.9.1) (Caporaso et al., 2010). The similarity among microbial communities in different soil samples was determined using UniFrac analysis by QIIME software. Principal coordinate analysis (PCoA) plots of the OTU-based weighed UniFrac distances between all samples and heatmap of Spearman's rank correlations coefficients between geochemical parameters and the top 35 bacterial genera were generated by R software (V3.5.1) (http://www.r-project.org/). Redundancy analysis (RDA) was performed based on abundant bacteria genera (relative abundance top 35) and selected physicochemical parameters by Canoco 5 (Biometrics Wageningen, The Netherlands) (Aazami et al., 2015).

Physicochemical Characterization of Soil Samples
The long term and continuous use of As and Pb contaminated paddy soil poses a serious health risks, since rice have a high tendency to accumulate As and Pb in its grains (Biswas et al., 2020). Paddy soil contaminated by As and Pb in mining areas has been widely documented (Kashyap et al., 2019;Zheng et al., 2020). However, the impact of mining activities on paddy soil physicochemical and microbial properties at various distances away from the mine is not well understood. HMs contamination and physicochemical properties of paddy soil play an important role in microbial metabolism (Zhao et al., 2013). Physicochemical characteristics of collected soils from the four sites were presented in Table 1. For all four sites, the soil pH was close to neutral (6.56-7.33), and the maximum pH was observed in site N01. The NO 3 − -N and NH 4 + -N concentrations were significantly higher in site N01 as compared to the other studied sites (p < 0.05).
As and Pb were found to be the dominant HMs in the selected sites ( Table 1 and Table 2). The high As and Pb concentrations can be ascribed to intensive mining activities in the area. Previous studies have shown that As and Pb can easily be translocated from the mining site to neighboring farmlands via surface runoff and/or dust particles deposition (Wang et al., 2011;Guo et al., 2017;Li et al., 2020c). The total As and Pb concentrations in the studied area ranged between 44.4-388 mg kg −1 and 0.20-2.12 g kg −1 , respectively. The results also showed that the As and Pb concentrations in all the sites exceeded the soil environmental quality risk control standard for soil contamination of agricultural land in China (GB15618-2018) (6.5 < pH ≤ 7.5, As 25 mg kg −1 , Pb 120 mg kg −1 ) (Mee, 2018). For example, the As and Pb concentrations in N01 were 15 and 17 times of the soil environmental quality risk control standard (GB15618-2018), respectively. These results were consistent with the previous studies that suggested agricultural soils near mining sites were seriously polluted (Leung et al., 2017;Zhen et al., 2019).
A clear As and Pb gradient was observed at all four sites. The As and Pb concentrations decreased with increasing distance from the mining site and vice versa ( Table 2). Previous studies have reported that the HMs concentrations in the topsoil usually decrease with the increase in the distance away from the mine (Li et al., 2015;Shen et al., 2017;Li et al., 2020c). A significant correlation was observed between total As and total Pb (r 0.970, p < 0.001) (Supplementary Table S1), which indicated that these metal (loid)s had similar pollution sources (Sun et al., 2010). Furthermore, HMs bioavailable concentrations in soils have been reported to have better correlation with their accumulation in plants, which would directly threaten human health via food chain . The bioavailable As concentration in N01 was significantly higher than other sites (p < 0.05). However, bioavailable As concentrations in N02, N03, and N04 were much lower compared to their total As concentrations. On the contrary, the bioavailable Pb concentrations were about half of the total concentrations for all four sites. Bioavailable As (r 0.715, p < 0.01) and Pb (r 0.667, p < 0.01) were positively corrected with soil pH, while bioavailable As (r −0.656, p < 0.01) and Pb (r −0.585, p < 0.05) were negatively corrected with soil OM (Supplementary Table S1). This indicates that pH and OM might be the main factors influencing As and Pb bioavailability in our study. pH may affect the bioavailability of As and Pb by influencing the adsorption of As and Pb on soil minerals. For example, the decreasing positive surface charge of soil minerals with increasing pH facilitated the desorption of As (Masscheleyn et al., 1991). With the increase of OM, the toxicity of HMs could be reduced through chelation and sequestration of metal cations onto OM (Macoustra et al., 2019).

Changes in Soil Microbial Biomass Carbon
Soil microbial biomass carbon (MBC) has been widely used as a bioindicator to evaluate soil quality since the soil microbial community is more sensitive to changes in soil composition compared to soil physiochemical properties (Feyzi et al., 2020). The MBC concentrations in N03 (1.39 mg g −1 ) and N04 (1.15 mg g −1 ) were significantly (p < 0.05) higher compared to those in N01 (0.52 mg g −1 ) and N02 (0.47 mg g −1 ) ( Figure 2). The MBC concentrations increased with decreasing As and Pb concentrations. Moreover, significant negative correlations between MBC and As and Pb concentrations were observed (Supplementary Table S1). Our results are consistent with those of a previous study (Zhen et al., 2019) reporting that MBC was significantly reduced in soils collected near a mining site in Shaoguan city, China, particularly due to the presence of high HMs concentration. Other studies also noticed that MBC decreased significantly with the increasing metal concentrations (Khan et al., 2010;Xu et al., 2019). This indicated that due to the presence of excess metals in the contaminated soils, the MBC was suppressed and the microbial population was impeded to multiply (Xu et al., 2019). This may have occurred because soil MBC is sensitive to HMs, since microorganisms require additional energy to deal with stress caused by HMs (Yang et al., 2006). This additional energy requirement may result in a reduction in the amount of substrate used for microbial growth .

Changes in Microbial Community Structure
Soil microbial community structure and diversity have also been used as an index for soil HMs pollution due to their sensitivity to soil physiochemical changes (Song et al., 2018). A total of 882,322 valid reads were obtained from the sixteen paddy   Table S2). The alpha diversity indexes including ACE and Chao1 from N03 and N04 were significantly higher than the samples from N01 and N02 (p < 0.05) (Supplementary Table S2). Soil microbial diversity in moderately and heavily contaminated soil were observed to be reduced compared to those of healthy soil (Golebiewski et al., 2014). Similar to our study, Shen et al. (2019) also observed that As and Pb contamination could decrease microbial diversity. Furthermore, the PCoA [PC1 vs PC2 (explaining 86.74%)] analysis suggested an obvious separation of the bacterial communities in N01, N02 and N03 while the bacterial communities in N04 was close to N02 and N03 (Supplementary Figure S1).
The microbial community structures differed among the four sampled sites. The relative abundance of the sequences obtained from the different sites at the phylum level are shown in Figure 3. The top 10 dominant phyla in all samples were, Proteobacteria (28.9-39.8%), Acidobacteria (9.3-16.6%), Chloroflexi (6.2-18.7%), Nitrospirae (2.8-12.7%), Bacteroidetes (2.0-8.9%), Gemmatimonadetes (1.9-4.4%), Planctomycetes (2.5-4.0%), Verrucomicrobia (1.9-3.9%), Chlorobi (1.9-3.9%) and Actinobacteria (1.2-2.2%). These phyla have also been reported in previous studies as dominant bacterial groups in paddy soil Zhen et al., 2019;Li et al., 2020a). Pollution-related changes in the relative abundance of the phyla were observed. The relative abundances of Proteobacteria, Bacteroidetes, Planctomycetes, and Verrucomicrobia were significantly higher in N01 than those in N02, N03 and N04 (p < 0.05), while the relative abundances of Chloroflexi, Nitrospirae, and Chlorobi were significantly lower in N01 (p < 0.05). These results indicate that soil HMs pollution may suppress some microbial phyla while promote the growth of HMs resistant microbes . Thus, we can deduce from our results that the phyla Proteobacteria and Bacteroidetes are able to endure high As and Pb stresses. Previous studies have reported that the members of the phylum Proteobacteria are able to tolerate high HMs stress (Schneider et al., 2017;Li et al., 2020c), which was consistent with the results of our study. Moreover, other studies have shown that Proteobacteria dominate in the metal-tolerant cultures, because many species belonging to this phyla can reduce Fe and facilitate soil-adsorbed metals release which inadvertently increase metal stress (Francis and Dodge, 1990;Jiang et al., 2017). Proteobacteria and Bacteroidetes have also been reported to harbor the majority of HMs resistance genes base on  Frontiers in Environmental Science | www.frontiersin.org April 2021 | Volume 9 | Article 630668 metagenome sequencing (Yan et al., 2020). Furthermore, an increase in Bacteroidetes relative abundance have been observed in Pb spiked soils compared to the unspiked soils (Hou et al., 2020). In order to further understand the microbial community response to geochemical conditions and HMs, the OTUs were further classified to the genus level (Figure 4). The four sampled sites were clustered into two distinct groups (one group included N01 and the other group included N02, N03, and N04) which was consistent with the results shown in the PCoA plot (Figure 4 and Supplementary Figure S1). Flavobacterium (2.0%), Haliangium (1.8%), Nitrospira (1.5%), Sphingomonas (1.4%) and Pseudomonas (1.3%) were the most abundant genera in N01, while Geobacter (1.4-1.9%), Thiobacillus (0.8-3.0%), and Bryobacter (0.6-1.5%) were the most abundant genera in N02, N03 and N04. The relative abundance of Haliangium, Nitrospira, Flavobacterium, Opitutus, Sphingomonas, Pseudomonas, Aquicella, Acidibacter, Blastocatella, and Sorangium were significantly higher in N01 than those in N02, N03, and N04 (p < 0.05). Previous studied have shown that the most of the bacterial strains from these genera are commonly involved in As resistance Titah et al., 2018;Li et al., 2020c).
Haliangium was observed to flourish in HMs stressed environments (Marti et al., 2017). Similarly, the abundances of Flavobacterium were observed to increase under high Pb stress (Wang et al., 2020b). Flavobacterium was also isolated from As contaminated soil as an As-resistant genus and it was able to reduce arsenate (As(V)) under erobic conditions (Wang and Mulligan, 2006). Besides, Flavobacterium related groups are able to methylate As as a detoxification strategy (Honschopp et al., 1996). Nitrospira are nitrite-oxidizing bacteria that can provide nitrate which takes part in the oxidation of arsenite (As(III)) . Studies have found that in HMs polluted soil, Sphingomonas was the most abundant genus and had a potential to remove Pb from seawater (Chen, 2012;Guo et al., 2017). Pseudomonas has been reported to be involved in As redox processes (Cavalca et al., 2013). Moreover, the As(III) efflux pump protein gene arsB and As transporter protein gene acr3P was also detected within the genus of Pseudomonas (Sanyal et al., 2016). Acidibacter and Blastocatella increased in HMs polluted soil and their significant positive correlation with Pb or As was also observed by Guo et al. (2017). The relative abundance of Geobacter, Thiobacillus, Sideroxydans, Anaerolinea, Anaeromyxobacter, Leptolinea, Spirochaeta_2, Desulfobacca, Syntrophorhabdus, and Ignavibacterium were significantly higher in N02, N03, N04 than those in N01 (p < 0.05). These results indicated that these microorganisms were relatively sensitive to As and Pb toxicity and the abundances were abruptly declined with the increase of HMs concentrations (Giller et al., 1998). Geobacter populations were reported as Fe and sulfate reducing bacteria, while Anaeromyxobacter were Fe reducing bacteria and Desulfobacca were sulfate reducing bacteria (Sun et al., 2015a). Besides, Thiobacillus were found to be involved in Fe and S oxidation and Thiobacillus can oxidize elemental sulfur . Sideroxydans were involved in Fe oxidation (Blothe and Roden, 2009). Most of these sensitive microorganisms were Fe-and S-related bacteria indicating their important roles in lightly contaminated sites. We speculated that As and Pb contamination in paddy soil might have an inhibition effect on Fe and S cycling in paddy soil.

Correlation Between Environmental Parameters and Microbial Community
In microbial ecology, understanding the relationship between environmental variables and microbial community structure is a key goal (Liang et al., 2011). Spearman's rank correlations coefficients between geochemical parameters and bacterial abundances were determined ( Figure 5). The relative abundance of Acidibacter positively correlated with total As, total Pb, bioavailable Pb (p < 0.05) and bioavailable As (p < 0.001) concentrations in soil. Similar results have also been reported by Guo et al. (2017). Nitrospira, Aquicella and FIGURE 5 | Heatmap of Spearman's rank correlations coefficients between geochemical parameters and the relative abundance of top 35 bacterial genera. The corresponding value of heatmap is Spearman correlation coefficient where "r" value is between −1 and 1, r < 0 is negative correlation, r > 0 is positive correlation. The asterisk indicates significant differences. *p < 0.05, **p < 0.01, ***p < 0.001. Abbreviations: As, Pb, Fe (total As, Pb, Fe in soil); Bio-As, Pb (bioavailability As, Pb); MBC (microbial biomass carbon); OM (organic matter); DOC (dissolved organic carbon).
Frontiers in Environmental Science | www.frontiersin.org April 2021 | Volume 9 | Article 630668 Sorangium were also positively correlated with total As, bioavailable As, total Pb and bioavailable Pb concentrations.
Other genera like Sphingomonas, Pseudomonas and Massilia also had positive correlations with total or bioavailable As or Pb. These results suggested that As and Pb contamination may facilitate the growth of these bacterial groups. Thioalkalispira (p < 0.001), Desulfobacca (p < 0.01) and Spirochaeta_2 (p < 0.01) were negatively correlated with total As, bioavailable As, total Pb and bioavailable Pb concentrations. Among them, Nitrospira, Aquicella and Sorangium were positively correlated with different forms of As and Pb, indicating their high HMs tolerance while Thioalkalispira, Desulfobacca and Spirochaeta_2 were sensitive to As and Pb pollution. pH, moisture and NH 4 + -N had significant correlations with most of the genera. Redundancy analysis (RDA) was used to reveal the possible linkages between the microbial communities and geochemical parameters ( Figure 6). Although the pollution source of the soil was from a Pb-Zn mining site, no significant correlation between the microbial communities and the Zn concentration could be found. It might be due to the low Zn concentrations in soil. Besides, it has been reported that high levels of Zn had no effect on the species diversity compared to unpolluted soil in mining areas (Xu et al., 2017). Total As, bioavailable As, total Pb and bioavailable Pb concentrations along with pH, NH 4 + -N, NO 3 − -N, moisture, SO 4 2and MBC were the main factors influencing bacterial community compositions ( Figure 6). Total As, bioavailable As, total Pb and bioavailable Pb concentrations along with pH, NH 4 + -N and NO 3 − -N positively correlated with microbial communities from N01 while negatively correlated with microbial communities from N02, N03, and N04. Soil moisture positively correlated with microbial communities from N02, N03, and N04. It is widely accepted that pH has a significant effect on the overall diversity and composition of microbial communities (Gu et al., 2017). Any significant deviation in pH could impose stress on single-celled organisms because the intracellular pH of most microorganisms are usually within 1 pH unit of neutral (Fierer and Jackson, 2006). Soil pH could also influence other environmental parameters (e.g. nutrient availability or cationic metal solubility), that plays an important role in microbial survival (Sun et al., 2015b). Previous studies also reported that pH had a greater impact on bacterial community composition in HMs contaminated agricultural soils when compared with HMs in soil (Cavani et al., 2016;Jiang et al., 2016). Two types of nitrogen (NH 4 + -N and NO 3 − -N) were significantly correlated with soil genera in this study, suggesting their important role in shaping the soil microbial community (Deng et al., 2018). Anderson et al. (2009) reported that NH 4 + -N and NO 3 − -N showed positive correlations with the shifts in microbial community structure, which was consistent with our results. Paddy soils could provide good habitats for nitrogen cycling, which in turn have the potential to shape the microbial communities (Sun et al., 2015a). Additionally, moisture is a major environmental factor that determines the redox conditions of soils and the metabolism of microbes (Honma et al., 2016). For example, wet-dry transition in soil would result in a microbial community shift of anaerobic to erobic microbes (Pett-Ridge and Firestone., 2005). This is especially true for paddy soil which are intermittently flooded during rice production.

CONCLUSION
In this study, we used the high throughput sequencing to examine the effects of physicochemical parameters on the microbial community variation in HMs contaminated soils along an As and Pb gradient. Our results showed that high amounts of HMs were presented near the mining site, and their concentrations decreased with increasing distance from the mining site. As and Pb contamination have a profound impact on the microbial community structure of paddy soils. Different forms of As and Pb (total and bioavailable) and soil physicochemical components (pH, NH 4 + -N, NO 3 − -N, moisture, and MBC) had significant influences on the bacterial community compositions. Some genera including Haliangium, Nitrospira, and Flavobacterium were positively correlated with different forms of As and Pb, indicating their high tolerance and their important role in As and Pb cycling while other genera including Geobacter, Thiobacillus, and Sideroxydans were sensitive to HMs contamination. Our finding provided important information that should be taken into consideration when developing bioremediation strategies for the polluted paddy soil in mining area.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

AUTHOR CONTRIBUTIONS
XT and JN performed the study conceptualization and supervision. LZ performed the literature search and wrote the original draft. YL, YD, MK, WG, and LY carried out the writing, reviewing and JS and JX performed the study supervision and validation editing. All authors read and approved the final manuscript.

ACKNOWLEDGMENTS
The National Key Technology R&D Program (2018YFD0800202) supported this study.