Response of Bacterial Communities to Heavy Metal Contamination in an Abandoned Chromate Factory

Abstract The chromium slags left by industry emission has caused serious damage to the local ecological environment, and microorganisms were sensitive to the variation in the surrounding environmental factors. In this work, 14 soil samples around an abandoned chromate factory were collected. We utilized the 16S rRNA regions from soil DNA to explore the possible effects of environmental factors on soil bacterial communities. High-throughput DNA sequencing results suggested that bacterial communities varied greatly from different soil samples, but Actinobacteria and Proteobacteria were predominant in all samples in the phylum level. In addition, Pseudarthrobacter, Thiobacillus, Paenisporosarcina, Sphingomonas, and Bacillus were abundant at a more refined species level. Based on redundancy analysis (RDA) and variation partitioning analysis (VPA), the results revealed that pH value, soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), chromium (Cr) and lead (Pb) could obviously affect the bacterial community structure. Heavy metals, soil nutrients, and pH could explain 37.69%, 13.61%, and 31.41% of bacterial community variation, respectively. This study would provide a theoretical basis for future remediation of local Cr-contaminated soil.


Introduction
Chromium (Cr) is a vital chemical element, which is extensively used in leather-tanning, electroplating, metallurgy, and corrosion control (Uddin 2017;Rager et al. 2019). In the process of these production activities in chromate factories, Cr and a series of incidental heavy metals will gradually accumulate in the surrounding environment, posing a severe threat to the local ecological environment and human health (Carlson et al. 2019;Costa 1997). Nevertheless, most previous studies mainly used various pollution indices (e.g., Geoaccumulation Index (I geo ), Single Pollution Index (P I ), enrichment factor (EF), contamination factor (C f ), and potential ecological risk (R I )), while less attention was paid to microorganisms (Kowalska et al. 2018).
The pedosphere is the bond between the atmosphere, lithosphere, hydrosphere, and biosphere, and is the basis of maintaining the function and service of terrestrial ecosystems (Hector and Bagchi 2007). Soil organisms interact to form a complex food web, which can also affect the soil environment and constitute the soil ecosystem in nature. Microorganisms are the crucial drivers of soil nutrient cycling, involving in organic matter degradation and elements transformation (Chodak et al. 2013). Soil microbial community diversity is a sensitive indicator of soil quality, which could provide information for the functional evaluation of soil quality (Pei et al. 2018;Sun et al. 2022;Wang et al. 2020a). Microorganisms in soils could affect the bioavailability of heavy metals through dissolution/precipitation, adsorption/enrichment, and redox, and then mitigate the risk of heavy metal pollution (Cai et al. 2020;Tang et al. 2019). The responses of diverse soil biological groups to environmental variation may be differentiated. For example, native fungal communities are more sensitive to heavy metals than bacterial communities (Deng et al. 2020). Meanwhile, variations in soil environments reshape in microbial community abundance, diversity, and structure as well. Therefore, microbial community characteristics could be used as one of the potential environmental indicators in various ecosystems (Sharma et al. 2021). However, the researches on factors of dominant soil microorganisms affecting the ecosystem in a polluted environment and partners of the microbial community responding to environmental variation are still insufficient. Therefore, it is necessary to investigate deeply on the microbial community structure, clarify the influencing factors of microbial community clustering, and identify the key microbial types that can adapt to high concentrations of heavy metals.
In the present study, we collected 14 soil samples from an abandoned chromate factory in Qinghai province, China, and adopted the Nemerow integrated pollution index (Wang et al. 2021) and Illumina sequencing of 16s rRNA amplicons to investigate: (i) the horizontal and vertical distribution of heavy metals; (ii) bacterial community structures; (iii) the potential relationships between environmental factors and bacterial communities. This study would provide a theoretical basis for future remediation of local Cr-contaminated soil.

Sites description and samples collection
The sampling sites were located in a chromite factory which had been abandoned for more than two decades (Figure 1). The 2200 tons of sodium bichromate (Na 2 Cr 2 O 7 Á2H 2 O) were produced during this period and an open-air chromate slag heap mountain was formed. Although most of the chromium slag in the area has been cleaned up, the surrounding soil are still being polluted due to the dissolution of chromium. And this area is still in the remediation stage. The area has a typical plateau continental climate with an average annual temperature of 0.5 C, an average annual precipitation of 411.4 mm, and an average annual evaporation of 1432.88 mm.
Soil samples were collected in September 2019 and December 2020 from six representative sites (S1, W1-W5) in the abandoned chromite factory (Figure 1 and Table S1), of which S1 was the site of the first remediation (FR) and W1-W5 were the sites of the second remediation (SR). The samples were collected from three different areas: the legacy site of chromium slag (S1, W1), the auxiliary workshop (W2, W5), and the machine repair shop (W3, W4). Soil samples in S1 and W3 were collected mechanically from five vertical of 0 to 4 m layer (interval of 1 m), other soil samples were collected from topsoil, which was sampled by pooling surface loose soil after removal of weeds or litter within 0.5 m 2 . S1 and W1 are high Cr-contaminated sites. To sum up, a total of 14 soil samples were harvested in this investigation. Each sample was sealed by PVC bags, which was divided into two parts. One was stored at 4 C for soil properties analyses, the other was frozen in À80 C for further DNA extraction and Illumina MiSeq sequencing of the microbiome.

Samples properties analysis
The soil samples were air-dried, homogenized, and sieved (100-mesh) in the laboratory before properties analysis. The soil pH was measured by pH meter (METTLER TOLEDO, Switzerland) with suspensions (soil: deionized water ¼ 1: 2.5, w/v) which were shaken about 30 min. Soil organic carbon (SOC) and total nitrogen (TN) were analyzed using Elemental Analyzer (Vario MICRO CUBE, Germany). Total phosphorus (TP) and heavy metals including chromium (Cr), manganese (Mn), nickel (Ni), lead (Pb), vanadium (V), and copper (Cu) were investigated by Inductively Coupled Plasma Mass Spectrometer (ICPMS, Thermo Scientific iCAP RQ, America). Before measurement by ICPMS, soil samples were digested with an acid mixture (HNO 3 , HCl, and HF) using EPA Method 3051. GBW07405 was used as the soil reference material to evaluate the uncertainty of the measurement. The recoveries of all tested elements were between 90 and 105%, and the relative difference was not exceeding 5% for replicate samples. The Nemerow integrated pollution index (P N ) was calculated to reflect the comprehensive heavy metal pollution states and pollution level following the formulae: Where P i represents the single pollution index of heavy meal element in the soil sample; C i represents the content of heavy metal we measured (mg kg À1 ); S i represents the soil background value of A-layer soil in Qinghai province; P i and P imax were the average and maximum value of the single pollution index (P i ), respectively.

DNA extraction, PCR amplification and Illumina MiSeq sequencing
Soil DNA was extracted using PowerSoil DNA Isolation Kit (MoBio Laboratories, Carlsbad, CA) according to Omega Stool DNA Kit protocol. Bacterial 16S rRNA genes were amplificated using primer 338 F (5 0 -ACT CCT ACG GGA GGC AGC AG-3 0 ) and 806 F (5 0 -GGA CTA CHV GGG TWT CTA AT-3 0 ). 8 bp barcode sequences were added to the 5 0 ends of upstream and downstream primers to distinguish different samples. The PCR amplification solutions, including 5 mL ($30 ng) DNA, 1 mL forward primers (5 mM), 1 mL reverse primer (5 mM), 12.5 mL KAPA 2 G Robust Hot Start Ready Mix, and 5.5 mL ddH 2 O, was denatured at 94 C for 5 min followed by 30 cycles of denaturation at 94 C for 45 s, annealing at 50 C for 30 s, extension at 72 C for 60 s and final extension at 72 C for 7 min. 1% agarose gel electrophoresis (voltage: 170 V; time: 30 min) was used to detect the fragment length and concentration of PCR products. The PCR products were used to construct the sequencing library, and paired-end sequencing was performed using the Illumina Miseq PE300 high-throughput sequencing platform. The original sequence was uploaded to the SRA database of NCBI (Accession Number: PRJNA822704).

Statistical analyses
QIIME1 (v1.8.0) software was used to split the samples according to the Barcode sequence. Pear software (v0.9.6) was used for data filtering and splicing. After data splicing, Vesearch software (v2.7.1) was used to remove sequences with length less than 230 bp, and the chimeric sequences were compared and removed by the uchime method according to Gold Database. The Uparse algorithm was used to perform Operational Taxonomic Units (OTUs) clustering for high-quality sequences, and the similarity threshold was 97%. Alpha diversity index including Simpson, Shannon, Chao1, and ACE was calculated by Past software (V4.08).
Pearson's correlation plot analysis was performed between soil physicochemical parameters using the Origin Apps 'Correlation plot'. The spearman correlations analysis between bacterial community diversity and environmental factors was performed by R packages 'psych'. Co-occurrence analysis was performed by calculating Spearman's correlation using the R packages 'Hmisc', and visual analysis was conducted using Gephi. Circos diagram (relative abundance top 26) was generated using Circos online (http://circos.ca/), based on the richness and evenness of OTUs among samples. Principal component analysis (PCA) was performed using Origin App 'Principal component analysis'. Redundancy analysis (RDA) was performed based on abundant bacteria genera (relative abundance top 10) and selected environmental factors by Canoco 5. Variation partitioning analysis (VPA) was determined the relative contributions of different environmental variables to changes in the bacterial community structure by using packages 'Vegan' in R software.

Physicochemical characteristics of soil in the sampling sites
Element contents and pH values of soil samples in the study sites were shown in Table 1, which indicated that soil physicochemical properties differed greatly among various sampling sites. The pH value of soil samples tended to be alkaline, in the range of 7.43-10.19, especially the surface soil in the legacy site of chromium slag exhibited strong alkaline (S_1_0: 9.69 ± 0.03, W_1_0: 10.19 ± 0.03).
Subsequently, various elements (SOC, TN, TP, Cr, Mn, Co, Ni, Cu, Pb, V) was measured in sampling sites, and was assessed pollution level by Nemerow integrated index (P N ) ( Table 2) (Nemerow 1991). The total Cr concentrations varied obviously, ranging from 43.9 to 1123.97 mg kg À1 . Based on the extent of Cr contamination, the sampling sites were divided into three different categories: (i) uncontaminated samples, including sampling site W_5 with a depth of 2-4 m, where the Cr concentration was less than Qinghai background soil; (ii) slight contaminated samples, including site S_1 with a depth of 1 to 4 m and W_2, where the Cr concentration slightly exceeded Qinghai background value, but less than Chinese soil risk screening value; (iii) heavy contaminated area, including topsoil at site W_1, W_3, W_ 4, W_5 and S_1, where the Cr concentration much higher than the China soil background soil. Nevertheless, Mn, Co, Ni, Cu, Pb, and V in all studied soils only exhibited inapparent enrichment comparing with local soil background values. The Nemerow integrated index for each heavy metal (P Nmetal ) was arranged in descending order as follows: Cr > V>Pb > Co > Mn > Ni > Cu.

Microbial community diversity and structure
The 16S rRNA gene sequences in 14 samples generated a total of 510,101 valid sequence, which could be clustered into 5115 OTUs respectively. According to the core OTUs analysis, the adaption characteristics and habitats of local bacteria surround soil in the abandoned chemical plant were identified. For this study, OTUs were classified into 46 phyla, 145 classes, 227 orders, 463 families, and 769 genera. With the increase of the number of sequences, the rarefaction curve tended to be gentle, indicating that the amount of data sequenced could reflect the microbial community composition of samples extensively ( Figure S1). Similarly, the Shannon diversity index verged on stabilization with the increase of sequences' number as well ( Figure S2). Alphadiversity indices (Shannon, Simpson, Chao1 and ACE) were determined for the samples, which were listed in Table S2. In general, Sample W_5_0 had the highest effective number of species or Shannon index, despite the relative high Cr concentration in this sample soil.
Judging from the relative abundance of the sequences, the bacterial community structures differed obviously among the various sampling sites. Soil bacterial community compositions across different samples with relative abundance above 1% at the phylum levels were shown in Figure 2(a). All OTUs could be clustered and classified into 46 phyla, including Actinobacteria (31.67%), Proteobacteria (23.97%), Bacteroidetes (9.53%), Chloroflexi (8.41%), Gemmatimonadetes (7.38%), Firmicutes (6.27%), Acidobacteria (4.92%), Saccharibacteria (2.62%), Cyanobacteria (1.02%), etc. The 6 dominant phyla (relative abundance >5%) made up 87.24% of the total OTUs and the top 10 phyla accounted for 96.55% of the bacterial community compositions. As shown in the Circos diagram ( Figure  2(b)), Actinobacteria, Proteobacteria, and Bacteroidetes were widespread at all sample sites. In addition, comparing the microbial composition of various samples, it was found that the difference of the top 10 bacterial phyla were mainly in the abundance rather than the category of the phylum.
The OTUs was used to further classify to the genus level. The heatmap of the relative abundance of bacteria in the genus level showed that the top 20 bacterial genera in all sample sites (Figure 2(c)), which belonged to dominant genera (relative Table 1. Soil physicochemical parameters and heavy metal contents of samples. Sample ID pH SOC (%)  and Lysobacter (1.11%). Among them, Thiobacillus, Pseudarthrobacter, and Pedobacter were the most abundant genera in sample W_1_0, W_4_0, and W_5_1, and the relative abundance accounts for more than 15% of the total abundance of their respective samples.
Principal component analysis (PCA) illustrated that the similarities and differences of bacterial community composition among soil samples. The PC1 and PC2 could explained 20.7% and 12.4%, respectively (Figure 3(a)). The site S_1 exhibited short distances among the sample S_1_0 to S_1_4, while it was relatively far from other samples. In addition, all the samples were clustered into a common area. The results suggested the clustering of samples was the result of the joint influence of the sampling period, sampling zone, and sampling depth.
The co-occurrence network could not only give a visual representation of the relationships among microorganisms but also be used to identify potential keystone taxa. In this study, an associative network of the bacterial community in the phylum level was established, which consisted of 40 nodes and 52 edges. (Figure 4). From the results, it was apparently observed that Spirochaetae were most likely to interact with other bacteria, followed by Acidobacteria and Deferribacteres. Therefore, the community at the study site had a wide range of mutually beneficial or competitive combinations of bacteria, which indicated that microorganisms did not aggregate irregularly in soil.

Relationships between the environmental factors and microbial communities
Multiple environmental factors played an important role in the composition and structure of the bacterial community. Redundancy analysis (RDA) could reveal the magnitude of the influence of environmental factors on bacterial communities. To explore the key driving forces shaping the microbial community in soils, eight geochemical variables were selected for analysis, including Cr, Pb, Mn, pH, SOC, H, TN, and TP. As illustrated from the results of RDA, the two first constrained axes explained 43.89% and 14.35% of the bacterial community differentiation at the phylum level, respectively (Figure 3(b)). The RDA results showed that soil nutrients such as TP (contribution ¼ 25.2%, pseudo-F ¼ 1.9), TN (contribution ¼ 16.8%, pseudo-F ¼ 1.5) and SOC (contribution ¼ 11.7%, pseudo-F ¼ 1.0) were largely shaped to bacterial community correspondingly to the length of the vector. Moreover, among the measured heavy metals, Cr and Pb was significant contributors to the structure of bacterial communities. Similar results also were obtained by bacterial alpha diversity, which Ace and Chao1 index were significantly correlated with TP, indicating the bacterial community richness index was mainly related to phosphorus in soil nutrients ( Figure S3). In addition, based on the results of VPA, changes of the bacterial community were explained by the relative proportion of specific environmental variables, which consisted of soil pH, soil nutrient (SOC, TN, TP). The results of VPA showed that heavy metals, soil nutrients, and pH could explain 37.69%, 13.61%, and 31.41% of bacterial community variation, respectively (Figure 3(c)). In general, 41.74% of the bacterial communities could be explained by these three groups of external variables.
In order to further study the correlation between bacterial community and environmental parameters, the spearman correlation heatmap analysis was conducted on the top 15 phylum-level bacterial community in relative abundance and multiple environmental factors ( Figure S4). The results showed that Proteobacteria presented the highest positive correlation coefficients with Cr, whereas Firmicutes and Latescibacteria displayed negative correlations (p < 0.05). TP was significantly contributed to the abundances of Fusobacteria, but negatively correlated with the proportions of Saccharibacteria and Deinococcus-Thermus. Unexpectedly, Cyanobacteria, Chloroflexi, and Actinobacteria had a great positive correlation with Pb. Furthermore, few phyla in the top 15 showed significant correlation with Mn, Co, and Cu, indicating that these environmental factors have a minor influence on the abundances of dominant bacterial communities. All these results indicated that there were differences in the responses of various bacterial compositions around the study site to environmental variables.

Discussion
Similar to many previous reports (Hu et al. 2018;Pei et al. 2018), these sampling soil in the Cr-contaminated environment tended to be alkaline. At present, soil pH value was considered as the important factors affecting the charge characteristics, adsorption/desorption, precipitation/dissolution, and coordination balance processes of heavy metals in soil. We found that the pH value of soils at this study area are more alkaline than that of local uncontaminated soil (pH value is about 8.0). As the pH value at the legacy site of chromium slag (S_1 and W_1) was significantly higher than those of the other sites (p < 0.05) (Table 1), we speculated that the redox reaction of chromium might consume a large number of anions in the soil in the process of stacking chromium residue, leading to the increase of soil pH (Ashraf et al. 2017). As shown in Figure 5, There is a significant positive correlation between pH value and chromium concentration (p < 0.05). This is due to the process to produce chromate salts in the world. The main step in the production of chromium salts is the oxidation of chromite with air or other oxidizing agents in an alkaline medium (Huang et al. 2009;Sauve et al. 2000). Therefore, most of the Cr-contaminated soils due to the production of chromium salts have an alkaline pH value (Wang et al. 2020a;Xu et al. 2022). Besides, soil organic carbon, nitrogen, and phosphorus contents were generally considered to be indispensable and crucial indicators for shaping microbial communities (Wang et al. 2021;Li et al. 2022a). As the monitoring report showed, the content of these essential soil nutrients was much lower than other areas (Li et al. 2022b), reflecting the soil nutrient depletion surrounding the abandoned chemical plant. From the perspective of the local ecological situation, only a few psammophytes were sparsely distributed in the study site, which was also well consistent with the previous studies at the chromium-containing slag site (Huang et al. 2009).
In order to quantify the extents to which the soil samples were contaminated by heavy metals from anthropogenic sources, we compared the heavy metal contents of soils in this study with Qinghai and Chinese background soils surveyed by China National Environmental Monitoring Center. As compared to the local background values (CNEMC 1990), topsoil Cr in all sites exhibited significant enrichment. Cr contents were still slightly higher than the background value from the layers below 1 m depth. In addition, it was worthwhile noticing that the spatial and profile soil showed enrichment of Cr toward the chrome slag legacy site of and the uppermost layers, indicating that despite anthropogenic remediation efforts, there is still a serious risk of Cr contamination in surface soils at all sites in the study area. On a time scale, the Cr contents of soil samples collected for the second time (W1-W6) was significantly lower than that of the first time (S1), indicating that recent soil remediation processes were practicable and effective. Furthermore, analyzed heavy metals expect Cr in soils showed no clear spatial and vertical variations with lower concentrations, implying their sources was mainly attributed to nature influence rather than anthropogenic actions. The correlation matrix of pH and analyzed elements showed that there was a close relationship between various environmental factors ( Figure  5). Metals including Co, Ni, Cu was significantly correlated with each other (r Ni-Co ¼ 0.698, p < 0.01; r Cu-Co ¼ 0.575, p < 0.01; r Cu-Ni ¼ 0.702, p < 0.01), which suggested that Ni, Co, and Cu elements were probably derived from natural soil composition. Nevertheless, the element Cr had no correlations with other heavy metal elements, indicating Cr was the main pollutant released by the chemical plant and no new kinds of heavy metal pollutions were added during the soil remediation process. In this study, the Nemerow integrated pollution index (P N ) was also calculated to reflect the comprehensive pollution states and pollution level of various heavy metals in this area. By comparing the classification standard of the P N , Cr has reached the serious pollution level, and other heavy metals we analyzed was mild or warning line pollution level (Table 2).
Actinobacteria, Proteobacteria, Firmicutes, and Chloroflexi were the dominant phyla in all soil samples, consistent with previous studies in the Cr-contaminated sites (Gao et al. 2021;Haferburg and Kothe 2007;Tang et al. 2022). Proteobacteria (34.3%), Chloroflexi (19.0%), Acidobacteria (14.6%), and Bacteroidetes (10.5%) were identified as dominant phyla in the paddy soils surrounding a nonferrous smelter (Tipayno et al. 2018). A study on the bacterial community structure in soils of a chromate factory revealed Proteobacteria (33.0%-96.7%), Actinobacteria (0.3%-6.9%), Chloroflexi (0.0%-13.1%), and Firmicutes (0.0%-5.8%) as the most abundant bacterial phyla ). The differences in the abundance are mainly determined by the sensitivity of the microorganisms to the content of heavy metals (Fernandez et al. 2018). We speculated that the dominant phyla at the sampling sites participated in a large number of organic matter decomposition and metabolic process that they had a high survivability to the Cr-contaminated environment . The dominant bacteria in the highly Cr-contaminated soil samples (S1 and W1) were similar, and the relative abundances of the two most abundant phyla were Actinobacteria and Proteobacteria. It can be seen that the total contributions of Actinobacteria and Proteobacteria fluctuated little in all soil samples. Similar to those in previously reported Cr-contaminated soils, both of them could tolerate a variety of heavy metals, such as Mn, Pb, and Cu Yin et al. 2017). Actinobacteria are common in the soils surrounding many chromate plants (Pei et al. 2018). It has also been reported previously that Actinobacteria could reduce the mobility of metals by reducing metals or forming metal complexes through the activities of multiple genes encoding the heavy metal enzyme (Polti et al. 2014;Altimira et al. 2012). A higher relative abundance of Proteobacteria was found in sample W_1_0 than in S_1_0. The difference of abundance was possibly attributing to the high adaptability of Proteobacteria to the environment. Compared to other bacteria, the members of Proteobacteria are better able to survive in low-temperature conditions (Zhao et al. 2019). The properties of Proteobacteria, an environmentally adaptable and metabolically diverse species, can explain the high presence of these in soil samples (Rogiers et al. 2021). In addition, Proteobacteria presented a positive correlation coefficient with Cr, suggesting it may have the potential for Cr remediation under diverse Cr stresses and could be used as indicators of Cr contamination. This result is consistent with previous studies (Bier et al. 2015;Martins et al. 2009). Based on the metagenome sequencing, Proteobacteria contain the vast majority of heavy metal resistance genes, followed by Actinobacteria. Proteobacteria was considered to be an r-strategic bacterial phylum that was more competitive than other bacteria in a polluted environment (Garavaglia et al. 2010). Besides, Chloroflexi with high abundance were well adapted to extreme environments and were closely associated with sugar respiration and carbon dioxide fixation (Hug et al. 2013). Although there was a significant negative correlation between the Firmicutes and Cr content (p < 0.05), a few previous studies showed that some Firmicutes could produce resistance differentiation structures by forming endospores under the stress of heavy metals (Fajardo et al. 2019). Comparing the soil samples collected in December and September (W_1 and S_1), it was found that the relative abundance of Bacteroidetes and Chloroflexi was a significant difference.
At the genus level, major genera included Pseudarthrobacter (average abundance of 4.48%), Thiobacillus (2.27%), Paenisporosarcina (2.02%), Sphingomonas (1.85%), and Bacillus (1.82%). Heatmap (Figure 2(c)) visually showed the difference in the average abundance of the top 20 bacterial genera among soil samples. There were significant differences (p < 0.01) in the abundance of Sphingomonas between highly Cr-contaminated surface soils (S_1_0 and W_1_0) and other soils. Guo et al. (2017) found that Sphingomonas was the most abundant genus of Proteobacteria in heavy metal-polluted soils and was considered as a potential bioremediating inoculant. The members of Actinobacteria have been reported in cases of Cr(VI) reduction and biosorption of Cu 2þ , Ni 2þ , and Zn 2þ (Masood and Malik 2015;Rodr ıguez and Quesada 2006;Tabaraki et al. 2013;Zakaria et al. 2007).
The microorganism was an indispensable part of the ecosystem, and microbial flora were mainly affected by the physicochemical characteristics of the soils. In this study, we determined the environmental factors affecting the bacterial community structure surrounding the chromate factory and elucidated the contribution of each factor to the community diversity. Based on RDA and VPA, multiple environmental agents jointly shaped the population and structure of bacterial communities. Previous studies confirmed that soil pH often played a vital role in driving bacterial community structure, which could affect community composition by directly affecting enzyme activity or indirectly changing redox conditions (Kemmitt et al. 2006). Some researchers have suggested that at pH < 6.5, there is a strong negative correlation between soil microbial diversity and pH (Kikot et al. 2009;Montebello et al. 2013). However, it was found that the pH value of soils tended to be alkaline but pH had insignificant relationships with the diversities of bacterial communities, indicating that a single parameter might not be the key factor which affected the diversities. The soil characteristics of Cr-contaminated areas are more complicated and changeable, involving a variety of factors, such as weather, surface temperature, and artificial disturbance (Zhang et al. 2018).
It was worth noting that heavy metals were one of the most important factors affecting the bacterial community around the study site (Figure 3(c)). Compared with previous studies, the results found that heavy metals in soil could inhibit the abundance of bacterial communities, but did not obviously inhibit the composition of the community. In addition, according to the heatmap of Spearman's rank correlations coefficients between geochemical parameters and a-diversity index. Phosphorus content presented significant positive correlations with bacterial abundance (p < 0.05). Our results were consistent with previous studies showed that P could affect the composition of the soil microbial community and alleviate the potential toxicity of heavy metals to microorganisms by means of immobilization (Tan et al. 2013). It was reported that Cr was toxic to a variety of microorganisms, inhibiting their growth, which in turn cause a reduction in microbial community richness and diversity. However, in this research, bacterial communities appeared to have high survivability to Cr, as no significant difference was found between Cr contents and a-diversity index (richness or diversities) in various soil samples, and even the sampling W_5_0 with high Cr concentration was found to have the highest richness and Shannon index of the bacterial community. The possible mechanism is that some bacteria are actively associated with Cr degradation and transformation, which play an important role in the soil ecological regulation process. Some researchers have reported that the predominant bacterial taxa in Crcontaminated soils included Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Gemmatimonadetes, and Proteobacteria, with a community structure similar to that of the present study. The metal resistance of these could be achieved by metal export using appropriate transporters, pumps, and ion channels, or by metal biotransformation, or by the antioxidant response. In addition, our results showed that there was no significant correlation between most heavy metals and bacterial communities, which might be due to the complexity between heavy metals and soil bacterial communities. We speculated that this result might be due to only bioavailable presented toxicity to the microbial organism (Kunito et al. 1999), or that long-term exposure of heavy metals to the environment lead to stable bacterial abundance and compositions.
Seasonal changes generally affect bacterial communities mainly through two parts: temperature and precipitation (soil moisture). However, the moisture in the soil around the study area remained constant, while the average temperature varied from À10 to 9 C. Thus, we deduced that although these two phyla harbor the majority of heavy metals resistance genes (Fan et al. 2016;Zhang et al. 2019), their growth was more susceptible to temperature. In addition, by comparing the bacterial community structure and composition in the various layers of soil, we found that the differences among bacterial communities mainly focused on the relative abundance of different phyla, and the diversity did not change obviously. This result was contrary to the previous study on soil microbial diversity around chromated factory , which might be due to the depth changes in the previous study only focus on the area above the illuvial horizon of soil.
Unfortunately, only the composition and structure of soil bacterial communities were analyzed in the present investigation. Further research was expected to focus on functional genes and related metabolic pathways of bacterial communities through genomics and metabonomics.
To sum up, heavy metals emission by chromate production can seriously affect bacterial richness and diversity and then shape the microbiota composition. In general, microorganisms resistant to heavy metals can dominate the relative abundance of the community because they can remove heavy metals through antioxidant reactions, metal complexation, or biotransformation, thus changing the composition of the microflora. A Chromate factory, a typical polluted area where the heavy metals contents are high, is a treasure for the isolation of high heavy metals resistant strains . Understanding the effects of chromium and other heavy metals on soil microbial communities facilitates the isolation and identification of local Cr(VI)-reducing strains, which can then be used to remediate locally contaminated soils.

Conclusions
In this present research, polluted areas were effectively assessed by exploring the interaction between environmental factors and microorganisms. Thereinto, Cr contamination significantly reduced soil bacterial abundance, whereas the diversity did not change obviously due to the high tolerance and adaptability of bacteria. Actinobacteria, Proteobacteria, Bacteroidetes, Chloroflexi, Gemmatimonadete, and Firmicutes were the dominant phyla in the study site. Heavy metals were the primary influence on the bacterial community structure and diversity, followed by pH and soil nutrients. pH, SOC, TN, TP, Cr, and Pb were the main factors affecting the diversity of the bacterial community. Our results will aid in the theoretical direction of bioremediation technology in Cr-contaminated areas.