Changes in rhizosphere bacterial communities during remediation of heavy metal-accumulating plants around the Xikuangshan mine in southern China

Mining and smelting activities are the major sources of antimony (Sb) contamination. The soil around Xikuangshan (XKS), one of the largest Sb mines in the world, has been contaminated with high concentrations of Sb and other associated metals, and has attracted extensive scholarly attention. Phytoremediation is considered a promising method for removing heavy metals, and the diversity and structure of rhizosphere microorganisms may change during the phytoremediation process. The rhizosphere microbiome is involved in soil energy transfer, nutrient cycling, and resistance and detoxification of metal elements. Thus, changes in this microbiome are worthy of investigation using high-throughput sequencing techniques. Our study in Changlongjie and Lianmeng around XKS revealed that microbial diversity indices in the rhizospheres of Broussonetia papyrifera and Ligustrum lucidum were significantly higher than in bulk soil, indicating that plants affect microbial communities. Additionally, most of the bacteria that were enriched in the rhizosphere belonged to the Proteobacteria, Acidobacteria, Actinobacteria, and Bacteroidetes. In Changlongjie and Lianmeng, the diversity and abundance of the microbial community in the B. papyrifera rhizosphere were higher than in L. lucidum. In parallel, the soil pH of the B. papyrifera rhizosphere increased significantly in acidic soil and decreased significantly in near-neutral soil. Redundancy analyses indicated that pH was likely the main factor affecting the overall bacterial community compositions, followed by moisture content, Sb, arsenic (As), and chromium (Cr).


Results
soil properties and concentrations of major heavy metals. Soils from different sites varied in their physiochemical properties and heavy metal concentrations, as listed in Tables 1 and 2. For both tree species, the rhizosphere soils generally contained higher moisture content than the bulk soils. The organic matter (OM) of all samples ranged from 1.06% to 5.29%, and the total nitrogen (TN), total phosphorous (TP), and total potassium (TK) were 0.48-1.12, 0.03-0.70, and 9.54-21.02 g/kg, respectively. The soil pH fluctuated widely, from 4.74 (CLR) to 7.49 (LLR). There was a clear difference in pH between the two sites, with acidity at Changlongjie (4.74-5.18 [CBR]) and near-neutral pH at Lianmeng (6.92-7.49 [LBR]). The heavy metal contents of rhizosphere and bulk soils differed. At both sites, lower concentrations of Sb and Cr were measured in the rhizospheres of both tree species compared to the bulk soil samples. The same phenomenon was also observed with Zn and Cd content, except in the bulk and rhizosphere soils of B. papyrifera at Lianmeng. The As concentration was lower in the rhizosphere  Table 1. Physicochemical parameters of the soil. The first letter in the sample name represents the sampling area; the second represents tree species; the third indicates rhizosphere or bulk soil; C, Changlongjie; L, Lianmeng; L, Ligustrum lucidum; B, Broussonetia papyrifera; R, rhizosphere; B, bulk soil. Data are reported as means ± standard deviations (n = 3). Different letters indicate significant differences between the four samples at same site (p < 0.05).
Scientific RepoRts | (2019) 9:1947 | https://doi.org/10.1038/s41598-018-38360-2 of B. papyrifera than in bulk soil, whereas that in the L. lucidum rhizosphere was greater than in bulk soil. The Pb concentration displayed no obvious patterns between rhizosphere and bulk soils. Compared to global soil background values, the levels of Sb, As, Cr, and Cd contamination were more severe at our study sites.
taxonomic distribution and comparisons between the two sites. The rarefaction curves for bacterial communities indicated that variation in operational taxonomic unit (OTU) density within the soil samples was sufficiently captured at the sequencing depth used (Fig. S1). Therefore, the data were sufficient to allow analyses of microbial communities. There were 609,268 valid reads identified for all soil samples collected (including repetition) through the Illumina MiSeq platform after filtering out the low-quality reads and chimeras, and trimming the adapters, barcodes, and primers (Table S2). These reads were clustered into 3,285 bacterial OTUs and 37 bacterial phyla, whereas only 16 phyla presented a sum abundance greater than 99% and approximately 1% of the sequences remained unidentified. The soil bacterial diversity and abundance were evaluated in the samples by estimating OTUs at a 3% sequence dissimilarity level. The Chao 1 index was used as an indicator of species richness ( Fig. 1 and Table S1). The PD_ whole-tree index was used as an indicator of phylogenetic diversity. The Shannon and observed_species indices were used to show the diversity of the microbial community in each sample. The results indicated that bacterial richness and diversity differed between the two sites. There were also differences between the rhizosphere and bulk soils of each tree species, except in the LBR and LBB samples. Notably, the differences between the rhizosphere and bulk soil samples of the tree species were greater at Changlongjie than at Lianmeng. The CBR sample showed the highest overall bacterial diversity and richness. The above results indicate that soil biodiversity differed between sampling locations, and different soil properties and tree species had an effect on bacterial diversity and richness.
The bacterial community composition of each sample at the phylum level is shown on the right side of Fig. 2. The dominant taxa across all soil samples were Acidobacteria, Proteobacteria, Chloroflexi, Actinobacteria, and Gemmatimonadetes; these groups accounted for more than 80% of the bacterial phyla sequenced. The relative abundance of each bacterial phylum indicates that each bacterial group responded to differences between the sites. The differences between the two sites can be seen in Figs 1 and S2. More Proteobacteria, Chloroflexi, Actinobacteria, and Nitrospirae were detected in Lianmeng samples than in Changlongjie samples. By comparison, Acidobacteria were relatively more abundant in Changlongjie samples than in Lianmeng soil samples. This may be closely related to the acidic environment in the Changlongjie soils 23 .
The cluster tree on the left side of Fig. 2 is based on the Bray-Curtis distances. The branches indicate the distances of the genetic relationships among all samples. The bacterial community structure was classified into two large groups with good similarity between parallel samples. The first group was composed of the samples from Lianmeng, with a pH range of 6.92-7.49. The second group was composed of samples from Changlongjie, with a pH range of 4.58-5.18. The bacterial communities at the two sites were clearly separated, likely due to the difference in pH.
Bulk soils vs. rhizospheres of the different plant species at each site. Principal component analysis (PCA) was used to visualize differences in community structure between rhizosphere and bulk soil samples (Fig. 3). Principal components 1 and 2 explained 45.61% and 12.44% of the total sample variability, respectively. The results indicated a relatively clear separation between the samples from the two sites. However, compared to those at Changlongjie, the rhizosphere and bulk soil samples of both tree species at Lianmeng were closer together on the PCA plot. This is may be due to the difference in pH values between the two sites. The analysis revealed clear differences between the bacterial community structures of rhizospheres and bulk soils.
For each site, significance analyses of the bacterial community were conducted to determine the selective/ enrichment effects of the rhizosphere in each tree stand (Fig. 4). These analyses were performed at the class level because the bacteria were not fully identified at other levels. We selected the bacterial taxa with an average abundance ≥1% for analyses using ANOVA (P > 0.05 for each taxon tested). At the Lianmeng site, the rhizosphere samples of L. lucidum were significantly enriched in Alphaproteobacteria, Thermoleophilia, Acidimicrobiia, Sphingobacteria, and OPB35_soil_group. The B. papyrifera rhizosphere samples were significantly enriched in

Relationships between bacterial communities and environmental variables.
Redundancy analysis (RDA) showed that the relative abundance of bacteria was affected by both growing position and soil properties (Fig. 5). The first two axes explained 66.50% and 12.88% of the total variance, respectively. The strongest determinant of bacterial communities was pH, followed by Cr, Sb, As, Zn, and moisture content. The samples at Lianmeng were positively correlated with pH, whereas the samples from Changlongjie, except for CBR, were negatively correlated with pH. The first component of the RDA (RDA1) separated Lianmeng soil samples from the four Changlongjie soil samples. These results imply a clear difference in bacterial abundance at the two sites. In addition, the bacterial abundance of the three rhizosphere samples (LLR, LBR, and CLR) was positively correlated with moisture content, whereas the distribution in bulk soils was relatively dispersed. Separately, the abundance in Lianmeng samples was positively related to pH and moisture content, and the abundance in Changlongjie samples was positively related to Cr, Sb, and As. Thus, ecological habitat had a significant effect on bacterial abundance. Soil pH had a positive effect on Latescibacteria and a negative effect on Acidobacteria and Firmicutes. Soil moisture had a positive effect on Proteobacteria, Bacteroidetes, and Planctomycetes. Soil Sb had a negative effect on Actinobacteria. As and Zn had negative effects on Gemmatimonadetes and Nitrospirae.

Discussion
Soil pollution with Sb from mining and manufacturing has become increasingly severe 6 . The pollution from the XKS Sb mine is due not only to Sb, but also As, Cd, Cr, Pb, and other heavy metals. Compared to traditional methods of soil remediation for heavy metals, such as physical and chemical approaches, phytoremediation is considered effective and more environmentally friendly 24 . In this study, successful plant establishment was observed in the XKS area after ecological restoration for more than 10 years. The plant species had diverse effects in altering the soil properties and heavy metal concentrations. With phytoremediation practice, the microbial diversity in the rhizosphere was remarkably changed. We investigated the composition and diversity of the bacterial communities of L. lucidum and B. papyrifera at two different sites. We observed clear differences between the two sites as well as differences between bulk soil and rhizosphere samples (Fig. 3), indicating that changes in the bacterial community structure were related to plant species and soil characteristics 25 . By comparing the bacterial communities in bulk soils or rhizospheres at the phylum level for the same tree species at two sites, we demonstrated that these communities were impacted by their ecological habitats. For example, Acidobacteria communities were more abundant in L. lucidum rhizospheres at Changlongjie than at Lianmeng, and Nitrospirae communities were more abundant in B. papyrifera bulk soils at Lianmeng than at Changlongjie. RDA revealed that bacterial community structures were influenced by pH, Cr, Sb, As, Zn, and moisture content, but pH was the dominant factor. Previous studies have shown that pH is a key factor in the distribution of bacterial populations 26,27 and bacterial community diversity is highest in soils with near-neutral pH and lowest in soils with pH < 5 21 . In this study, pH values differed between Changlongjie (4.74-5.1) and Lianmeng (6.92-7.49). It has been suggested that pH affects bacterial survival directly and/or by controlling ancillary environmental parameters that are closely related to soil pH, such as cationic metal solubility 28 . Many bacteria have near-neutral intracellular pH levels, and their growth is sustainable over a relatively narrow pH range (3-4 pH units) 29 . Indeed, pH had a significant effect on the bacterial communities in this study (Fig. 5). The effects of Cr on the bacterial communities at XKS were greater than the effects of Sb and As, which has not been reported in previous studies on the structure of XKS microbial communities. One potential explanation is that the metal speciation of Cr affected the organic acid exudation rate and the composition of root exudates, and the root exudates greatly affected the microbial activity and bacterial communities in the soil 30,31 . Plant species, as well as ecological habitats, have a substantial influence on the structure of rhizosphere-associated microbial populations. The influence of plant species has been clearly shown in several studies. Kowalchuk reported that plant species impact the rhizosphere bacterial communities of Cynoglossum officinale and Cirsium vulgare 32 . Uroz et al. 33 also reported the effects of plant species on rhizosphere bacterial communities. The rhizosphere, a unique microenvironment in terrestrial ecosystems, is described as the portion of the soil in which microorganism-mediated processes are under the influence of the root system 19 . The biomass and activity of microorganisms are higher in the rhizosphere than in bulk soils, owing to the exudation of compounds by plant roots 34 . Indeed, in our study, the diversity (Shannon index) and abundance (Chao1 index) of rhizosphere bacterial communities under each tree species were greater than in the bulk soils (Fig. 1), indicating that the trees influence the structure and diversity of the microbial community. Sun et al. 35 also reported similar results in studying changes in the rhizosphere microbial communities of Pennisetum purpureum, Typha angustifolia, and Alnus cremastogyne in copper mine tailings in China.
In parallel, Hinsinger et al. 36 reported that plants can regulate rhizosphere pH, and a major process contributing to root-induced pH changes in the rhizosphere is the release of charges carried by H + or OH − . The root-mediated pH changes in the rhizosphere depend on constraints from the enviroment 37 . In this study, we found that the soil pH in the rhizosphere of B. papyrifera was obviously elevated compared to bulk soil in Changlongjie, whereas B. papyrifera rhizosphere pH was significantly reduced in Lianmeng. These pH changes were not observed in L. lucidum. The results indicate that B. papyrifera regulates the rhizosphere soil pH better than L. lucidum. These root-mediated pH changes affect the bioavailability of many nutrients and toxic elements and improve the growth environment of rhizosphere microorganisms. This indicates that B. papyrifera can create better survival conditions for itself and its rhizospheric microbes, in addition to playing a more significant role in the restoration of heavy metal-contaminated soil.
Each plant species is thought to select specific microbial populations 38 . The compartment effect (i.e., rhizosphere vs. bulk soil) was used to determine the selective effects of the rhizosphere habitat on bacteria (Fig. 4). The sequences generated from the Lianmeng and Changlongjie soil samples were analyzed independently. The rhizospheres of both tree species at Lianmeng were enriched in Alphaproteobacteria, Sphingobacteria, and OPB35_ soil_group, whereas the other bacteria enriched in the rhizosphere differed between the two species (Fig. 4). At Changlongjie, the rhizospheres of both tree species were enriched in Betaproteobacteria only. This indicates that the selection of rhizosphere microflora by these two tree species is different. Uroz et al. 33 reported the same results when comparing the rhizosphere selectivity of beech and Norway spruce trees.
Notably, most bacteria enriched in the rhizosphere belonged to the Proteobacteria, Acidobacteria, Actinobacteria, and Bacteroidetes. Researchers have repeatedly demonstrated that Proteobacteria may be the most metal-tolerant organisms found at metal-contaminated sites 39,40 . Their ecological and metabolic abilities have been reported to be adaptable to the extreme environment of mine tailings and to reduce the toxicity of heavy metals 41,42 . Acidobacteria are disintegrators of organic matter and are involved resilience in nutrient-deficient environments 43 . At the same time, Actinobacteria are related to defense against soil diseases and improvement of root nodulation efficiency 44 . Bacteroidetes are involved in the degradation of high-molecular-weight organic matter in the rhizosphere 45 . These rhizospheric microorganisms are involved in the toxicity reduction of heavy metals in soil, the prevention and control of soil diseases, the improvement of root nodulation efficiency, and adaptation to the extreme environment of mine tailings. The resistance of plants can be promoted in extreme environments, which will be helpful for phytoremediation. Therefore, we can artificially add microbes that are enriched in the rhizosphere to the rhizospheres of other plants, which will promote plant repair. This has been done previously in heavy metal-contaminated rice fields 46 .
The effects of heavy metals on bacterial communities cannot be ignored. Cr, Sb, and As affected the distribution of bacterial communities, and the levels of each differed between the plant rhizospheres and bulk soils. At both sites, Sb and Cr levels were lower in the rhizospheres of both tree species than in bulk soils, whereas As content was greater in the rhizosphere of B. papyrifera than in bulk soil, but lower in the rhizosphere of L. lucidum than in bulk soil. The diversity of the bacterial community was somewhat greater in the B. papyrifera rhizosphere than in that of L. lucidum (Fig. 1). The bacterial diversity index is negatively correlated with heavy metals in forest soils 47 . Due to the decreased concentrations of heavy metals in the rhizosphere, the bacterial diversity may showed an obvious increasing trend 48 .
Chen 12 demonstrated that the two tree species can be used as heavy metal-accumulating plants at the two sites; thus, the reduction in heavy metals in the rhizosphere may be due to absorption by plants.
This study considered the effects of tree species and XKS sites on bacterial communities. We concluded that the bacterial community structure was influenced by both ecological habitat and tree species. However, the effects of each on bacterial communities differed between acidic and near-neutral soils. In addition, B. papyrifera may serve as a better accumulator plant than L. lucidum at Changlongjie. This speculation is based on the evidence that the bacterial diversity of the B. papyrifera rhizosphere was greater than that of the L. lucidum rhizosphere, and levels of Sb, As, and Cr in B. papyrifera rhizospheres were lower than those in bulk soil, although there was a similar trend in Sb and Cr, but not As, in the L. lucidum rhizosphere. However, this is only a preliminary evaluation based on bacterial community structure and heavy metal content. Further research elucidating the mitigation effects of the tree species and the functions of the bacteria that were significantly enriched in the rhizosphere is needed.

Conclusions
We observed the rhizosphere bacterial communities of heavy metal-accumulating plants at two sites near XKS in Lengshuijiang City, Hunan Province. By analyzing the changes of plant rhizosphere microflora in two regions, we found that plant species and ecological habitats cooperatively shaped the structure of microbial communities in the rhizosphere. In particular, B. papyrifera modified the soil pH and several heavy metal levels in the rhizosphere. Also, the diversity and abundance of the microbial community in the B. papyrifera rhizosphere was higher than in bulk soil and other plants. Therefore, we hypothesize that B. papyrifera may be more suitable for phytoremediation than L. lucidum at these two sites. We also made a preliminary assessment of the significant rhizobacterial communities of each tree species at the two sites to provide a theoretical basis for joint restoration using plants and microbiota. Most bacteria enriched in the rhizosphere belonged to Proteobacteria, Acidobacteria, Actinobacteria, and Bacteroidetes. These bacterial groups have been reported to be associated with heavy metal resistance, decomposition of organic matter, enhancement of root nodulation efficiency, and adaptation to extreme environments. In future studies, we should attempt to isolate these bacteria and explore their functions in phytoremediation. This will provide a theoretical basis for combined remediation using microbes and plants. Finally, we found that the effects of Cr on the bacterial communities at XKS were greater than the effects of Sb and As, which has not been reported in previous studies.

Materials and Methods
site location and sample collection. The study site is located at the XKS Sb mine, Lengshuijiang City (27°30′49″-27°50′38″N, 111°18′57″-111°36′40″E), Hunan Province, Southwest China (Fig. 6, generated by Photoshop CC 2015). The mine is one of the world's major producers of Sb. It is renowned as the "world's antimony capital". In this region, mining and smelting activities have resulted in the pollution of soil, water, and sediments with Sb and other metals over the past approximately 100 years 2 . In November 2015, rhizosphere and bulk soil samples from L. lucidum and B. papyrifera were collected from two sites near the XKS mine, Lianmeng and Changlongjie. The rhizosphere soil samples were collected from soils adhering to plant roots (ca. 1-2 mm to roots) from different locations within each plant stand and systematically pooled together. Bulk soil ~50 cm from the trees was collected using a soil corer at soil depths of 0-20 cm. Each sample was randomly collected using the five-point sampling method, then pooled together into one composite sample. We set up three plots (20 × 20 m) for each tree species at each site. We took one rhizosphere and one bulk soil sample from each plot; the three soil plots were three repetitions of one sample. A total of 24 composite samples were collected (two tree species × three replicates × two sites × [rhizosphere + bulk soil]). Samples were encoded with letters indicating their collection location (L, Lianmeng; C, Changlongjie), tree species (L, L. lucidum; B, B. papyrifera), and substrate (R, rhizosphere; B, bulk soil). Each soil sample was collected and passed through a 2-mm sieve, then stored in a sterile polyethylene bag at 4 °C for measuring physicochemical parameters, and at −20 °C for DNA extraction. physicochemical analyses of soil and determination of target metals. Soil moisture content, pH, organic matter (OM), total nitrogen (TN), total phosphorus (TP), and total potassium (TK) were assessed. Moisture was estimated using the oven dry-weight method. Soil pH was measured using a glass electrode pH meter (Sartorius PB-10; Sartorius Scientific Instruments Co., Ltd., Beijing, China) in a suspension of 1 g soil in 5 mL distilled water. OM content was determined by the Walkey and Black method 49 . The determination of TN, TP, and TK were carried out in accordance with the national standards of the People's Republic of China. Briefly, 0.2 g of soil was added to 5 ml of perchloric acid and 5 ml of hydrofluoric acid for low-temperature digestion, followed by determination of TP and TK by the Mo-Sb colorimetric method and the flame photometer method, respectively. TN was extracted from 0.2 g of soil by digestion with 5 ml of concentrated sulfuric acid, and then determined by the Kjeldahl method. According to Wilson 50 , heavy metal contaminated soil was analyzed by digestion of 0.5 g of soil with 10 mL of concentrated HNO 3 , following the microwave-nitric acid method. The concentrations of Sb, As, Pb, Zn, Cr, and Cd in the soil samples were measured using inductively coupled plasma optical emission spectrometry.
High-throughput sequencing of the V3-V4 regions of 16S rRNA genes. Total genomic DNA was extracted from each soil sample using a FastDNA spin kit (MP Biomedicals, Santa Ana, CA, USA) following the manufacturer's protocol. The concentration and purity of the extracted DNA were confirmed using 1% agarose gel electrophoresis. The hypervariable V3-V4 region of the bacteria 16S rRNA gene was amplified using the 338 f/806r primer set (338 f: 5′-ACTCCTACGGGAGGCAGCA-3′, 806r: 5′-GGACTACHVGGGTWTCTAAT-3′) 51 . These primers contain a set of 8-nucleotide sequences unique to each sample. The PCR program was as follows: a 5-min initial denaturation at 95 °C, 25 cycles at 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, with a final extension at 72 °C for 10 min. PCR reactions were performed in triplicate. The high-throughput sequencing of the 16S rRNA amplicons was carried out on an Illumina Miseq PE300 sequencing platform (Illumina, Inc., San Diego, CA, USA) at Allwegene Tech, Ltd. (Beijing, China). All datasets were analyzed using QIIME, based on sequence length, quality, primers, and tags; the raw sequences were selected and the low-quality sequences were removed 52 . High-quality sequences were trimmed using the Illumina Analysis Pipeline version 2.6. The SILVA Classifier tool was used to classify the unique sequence set into operational taxonomic units (OTUs) 53 , with a threshold of 97% identity 54 .
Data analyses. All sequences were analyzed using the bioinformatics platform Mothur (v.1.33.0) 55 . The similarity of bacterial communities among different soil samples was determined using both weighted and unweighted UniFrac. PCA based on weighted UniFrac distances was performed to quantify differences in community composition at the OTU level. Rarefaction analyses were used to measure whether sequences sampled were sufficient to capture the total richness at a genetic distance of 0.03. Shannon, Chao1, observed_species, and the PD_whole-tree indices were determined for the diversity and richness of the bacterial community. RDA was used to identify the effects of environmental factors on bacterial communities, based on a relative abundance greater than 1% of bacterial communities at the phylum level, with Canoco 5.0 software. Before we did the RDA analysis, we unified the units of various environmental factors. The effects of tree species on the abundance of the different taxa at the class level were tested using analysis of variance (ANOVA) at a threshold level of P = 0.05 using IBM SPSS Statistics v. 19.0 (IBM Corp., Armonk, NY, USA).