Geographic distribution of cadmium and its interaction with the microbial community in the Longjiang River: risk evaluation after a shocking pollution accident

A shocking Longjiang River cadmium pollution accident occurred in 2012, the effects of which on microbial communities remain unclear. Alkaline precipitation technology was applied for remediation, but concerns rose about the stability of this technology. To understand the geographic distribution of cadmium and its correlation with microbes, in this study, 39 water samples and 39 sludge samples from this river and 2 soil samples from the nearby farmland were collected for chemical and microbial analyses. The Cd concentrations of all water samples were lower than 0.005 mg/L and reached the quality standards for Chinese surface water. A ranking of sludge samples based on Cd contents showed sewage outfall > dosing sites > farmland, all of which were higher than the quality standard for soil. Alkaline precipitation technology was effective for Cd precipitation. Cd was unstable; it was constantly dissolving and being released from the sludge. The Cd content of each phase was mainly influenced by the total Cd content. Over 40,000 effective sequences were detected in each sample, and a total of 59,833 OTUs and 1,273 genera were found using Illumina MiSeq sequencing. Two phyla and 39 genera were notably positively correlated with the Cd distribution, while the cases of 10 phyla and 6 genera were the opposite.


Results
The pH of river water and Cd distribution in the Longjiang River. The pH of the river water varied from 7.0 to 7.4 in different groups (Fig. 1). Sewage outfall and midstream showed a lower pH than the control. These results implied that the effects of the unregulated discharge of untreated acid sewage several years ago Scientific RepoRts | 7: 227 | DOI:10.1038/s41598-017-00280-y remain. Dosing sites showed the highest pH. This might be due to the caustic soda and lime in the sludge of dosing sites because a weak alkaline chemical precipitation technology was used to remove Cd in 2012.
The Cd content in sludge samples and the Cd concentration in river water samples are shown in Fig. 2. In the sludge samples, sewage outfall had the highest Cd content, followed by dosing sites and midstream, and farmland had a relatively lower Cd content; the differences were significant in these groups (p < 0.05). Higher Cd content in the dosing site samples indicated that some Cd was precipitated from the river water to sludge and that the weak alkaline chemical precipitation technology used to remove Cd from river water had some effects in 2012. The amount of Cd content in the farmland samples was the lowest but still beyond the environmental quality standard for soil (GB 15618-1995). In the river water samples, dosing sites had the highest Cd concentrations (Fig. 2), which implies that Cd is unstable in dosing sites and is constantly and slowly dissolving from the sludge.  The Cd concentrations of the river water in all the groups were lower than 0.005 mg/L. The quality of river water reached the second level of the national environmental quality standards of surface water (GB 3838-2002). The Cd in the sludge existed in five phases, although the total Cd content showed significant differences in different groups (p < 0.05). The composition of each phase in the different groups had no significant difference (p > 0.05). These results indicated that the Cd content of each phase was mainly influenced by the total Cd content of the sludge and had less influence from other factors.
Microbial diversity and composition of the samples. The composition and diversity of the microbial communities in 41 samples were profiled using Illumina MiSeq platform sequencing of PCR-amplified bacterial and archaeal 16S rRNA gene fragments. After trimming and quality filtering the raw reads, 4,230,394 sequences in all and more than 40,000 sequences with the shortest length of 444 bp per sample were obtained. All of the sequences were clustered into operational taxonomic units (OTUs) using a similarity threshold of 97%, yielding a total of 59,833 OTUs. The sequences were classified into the domains of Bacteria (90.65% of the sequences and 69 phyla) and Archaea (7.97% of the sequences and 4 phyla) using the Ribosomal Database Project (RDP) classifier with a confidence threshold of 80%, with 1.38% of the sequences unclassified at the domain level. The predominant microbial phyla were Proteobacteria (30.88% ± 25.27% of the reads), Bacteroidetes (11.58% ± 9.7% of the reads), Chloroflexi (10.14% ± 8.77% of the reads), Acidobacteria (8.88% ± 6.70% of the reads) and Verrucomicrobia (7.16% ± 9.13% of the reads).  Table 1) using MiSeq sequencing of the microbial 16S showed high richness and diversity, but the microbial richness and diversity in the sewage outfall and dosing sites showed a significant reduction compared to those of the control, especially in the sewage outfall samples. These results indicated that the microbial abundance was restrained and the microbial diversity reduced in the sewage outfall due to the existence of Cd, which was consistent with previous reports 35, 37 . Differences in microbial community composition in different samples. The results of the analysis of community similarity among samples (Fig. 3) showed that the sewage outfall and dosing site samples mainly clustered into two large branches of the community similarity tree. Most of the pristine and contaminated samples were separated in the tree. Microbial community composition exhibited regular changes with Cd content in different groups (Fig. 4). In all 73 phyla, the relative abundances of 29 phyla increased with the increase of Cd concentration, and the largest increase (TA06) was 5.42 times. In contrast, the abundances of 44 phyla decreased, and 3 phyla could not be even detected in the samples collected from high Cd concentration sites. Among them, 2 phyla (Gemmatimonadetes and Proteobacteria) had a significantly positive correlation and 10 phyla (GOUTA4, Fibrobacteres, Kazan-3B-28, WS2, WS4, OP8, KSB3, Planctomycetes, Chlamydiae and Caldiserica) had a significantly negative correlation with Cd distribution (Table 2). On the genus level, 1273 genera were detected from all the groups. The relative abundances of 533 genera increased with the increase in Cd concentration. The largest increase (Leucobacter) was 8 times, while the abundances of 562 genera decreased. In particular, 178 genera were not present in the sewage outfall samples. Among them, the abundances of 39 genera (such as Pseudoxanthomonas, Halomonas, and Methanomassiliicoccus) were positively correlated with the Cd distribution (p < 0.05), while those of 6 genera (such as C1_B004, Clostridium, and vadinHB04) were negatively correlated with the Cd distribution (p < 0.05) (Table S2).

Correlations between sample properties and microbial community composition. The relation-
ships between environmental factors and microbial communities were evaluated by the canonical correspondence analysis (CCA) method (Fig. 5). All the environmental factors were distributed on the right side of the CCA   Table 3. Correlation between Cd content in sludge and Cd content in different phases, microbial richness (Chao1 index), and microbial diversity (Shannon index) in sludge samples. **Represents significant at the level of 0.01.
plots and were positively correlated with the first axes. The microbial communities of the sewage outfall had the highest correlation with the Cd content of the sludge. The microbial communities of other groups were subjected to relatively smaller impacts from the environmental factors. Anaeromyxobacter, Thiobacillus, Bradyrhizobium, SJA-88, SHD-14, Fusibacter, Dechloromonas, Flavisolibacter, Perlucidibaca and Methanobacterium were the top 10 most abundant genera that had significantly positive correlations with the Cd content of the sludge. Significant correlations between the Cd content in the sludge, the Cd content in different phases, microbial richness (Chao1 index) and microbial diversity (Shannon index) in the sludge samples were detected, as shown in Table 3. There were notable positive correlations (p < 0.01) between the Cd content of the five different phases in the sludge and the total Cd content in the sludge. There were no significant correlations between the Cd content of the sludge and microbial richness or microbial diversity (p > 0.05).

Discussion
There was a positive correlation (Table 3) between the Cd content of the sludge and the Cd concentration in the river water in most of the groups (except sewage outfall). However, in sewage outfall, the highest Cd content was found in the sludge, and there was a lower Cd concentration in the water. This might be because the sewage outfall has been hardened with cement and there was less sludge in the sewage outfall. Although the highest Cd content occurred in the sewage outfall, the total amount of Cd in the sewage outfall was less than that found in the other groups. This explains the result of no significant correlation between the Cd content of the sludge and the Cd concentration of the river water (p > 0.05).
There were notable positive correlations (p < 0.01) between the Cd content of the five different phases in sludge and the total Cd content in the sludge. These results indicated that the Cd content of each phase in the sludge was mainly influenced by the total Cd content of the sludge rather than other factors. There were no significant correlations between the Cd content of the sludge and microbial richness or microbial diversity (p > 0.05). In particular, the farmland had the smallest correlation, which might be because there were differences in microbial taxa and microbial community structure between the river sludge samples and farmland soil samples.
There were several abundant genera in all the samples from the Longjiang River that can greatly impact the water quality of the Longjiang River and may be indicative of the water quality. For example, Geobacter is a genus of proteobacteria and the third largest genus of all the samples. Geobacter is an important metal ion-reducing bacterium that can reduce Fe(III) 38 , Pd(II) 39 and uranium (VI) 40 . Geobacter can be used for the precipitation of uranium out of groundwater 41 , as a new alternative for synthesizing Pd(0) nanocatalyst, and also has potential applications for microbial metal recovery from metal-containing waste streams 39 . Additionally, Geobacter's ability to consume oil-based pollutants with carbon dioxide as a waste byproduct has already been used in environmental clean-up of underground petroleum spills 42 . Nitrospira, from the Nitrospirae phylum, ranked the fourth in abundance in all samples. Nitrospira are nitrite-oxidizing bacteria that are important in marine and no marine habitats 43 . In water, Nitrospira takes part in the nitrification process; this is important for the biogeochemical nitrogen cycle. Desulfobulbus, with an abundance ranking thirty-sixth, is a sulfate-reducing bacterium, which is important for the precipitation of the heavy metal Cd in the Longjiang River.
There are some microbes that are sensitive to Cd, such as Labrys, Clostridium and Methanocella. Labrys is also sensitive to other metals (Fe 2+ , Cu 2+ , and Ag + ) 44 . Labrys is an important microbe for the bioremediation of contaminants in wastewater treatment, as it has the ability to biodegrade three of the most used fluoroquinolone antibiotics worldwide: ofloxacin, norfloxacin, and ciprofloxacin 45 . The decreased abundance of Labrys may cause harmful effects to the environment. In contrast, the decrease of some microbes is good for the environment. For example, some species of Clostridium are pathogenic bacteria 46 . Clostridium can produce botulinum toxin in food or wounds and cause botulism 47 and are also the causative organism of tetanus 48 . Methanocella is also a Cd-sensitive microbe that mainly exists in rice paddy soil and could produce methane 49 . The existence of Methanocella in rivers significantly contributes to methane emissions and causes global warming 50 ; the abundance in the Longjiang River could reduce the production of methane.
The results of the sample similarity tree and microbial community composition indicated that the microbial community was affected by Cd. The microbial communities of sewage outfall had the highest correlation with the Cd content of sludge. This result indicated that the microbial communities of sewage outfall are subjected to the greatest impact from the Cd of sludge. There were 39 genera showing a significantly positive correlation with the sludge Cd content, some of which were beneficial for heavy metal solidification and remediation, such as Pseudomonas, Desulfovibrio, Anaeromyxobacter, Leucobacter and Halomonas. Pseudomonas contains a Cd resistance (cadR) gene 51 . It is tolerant to other heavy metals 52 and has the ability to adsorb heavy metals 53 . In addition, Pseudomonas has the potential to promote plant growth, remove heavy metals from contaminated soil 54 and immobilize heavy metals from solution 55 . Therefore, Pseudomonas is a new potential resource for the remediation of Cd and other heavy metals. Desulfovibrio is the first reported acid-tolerant gram-negative sulfate-reducing bacteria resistant to high concentrations of metals. Some Desulfovibrio species have bioremediation potential for the treatment of metal-containing wastewater 56 , with applications such as Hg methylation 57 and U reduction 58 . There are two important protein types in Desulfovibrio. The first is orange protein (ORP, 11.8 kDa), which contains a mixed metal sulphide cluster of the type [S 2 MoS 2 CuS 2 MoS 2 ] 3− noncovalently bound to the polypeptide chain 59 . The second is cobalt-and zinc-containing adenylate kinases (AKs) 60 . A novel type of metal-binding site for three metal ions: cobalt, zinc and iron (II) is reported to be present in AKs 61 . Anaeromyxobacter is an arsenate-respiring bacterium isolated from arsenic-contaminated soil that contains three distinct arsenic resistance gene clusters (ars operons) 62 . Anaeromyxobacter could reduce not only dissolved arsenate but also arsenate adsorbed on the soil mineral phase 63 . It might play a role in arsenic detoxification from these environments. Leucobacter is a chromate-resistant strain that was reported to be able to grow in a medium containing up to 300 mM K 2 CrO 4 and showed cellular aggregation in response to chromate stress 64 . Leucobacter also showed strong Ni(II) removal efficiency by biosorption 65 . All these microbes have favourable functions for environment remediation and may therefore play important roles in the Longjiang River. In contrast, the increase in abundance of Halomonas with Cd content may be unfavourable for the environment. For example, Halomonas have been called "rust-eating bacteria"; they have been reported to be detected in the rusticles of the Titanic and could consume metal, causing rapid decay of the metal 66 . Halomonas might be unfavourable for Cd solidification because it might consume Fe of the Fe-Mn oxide bound fraction and accelerate the dissolution of Cd.
Several microbes in the samples showed significant correlations with Cd content. The relative abundances of some microbes, such as Leucobacter, in the sewage outfall samples were 8 times that of all samples. However, these populations accounted for a small proportion of the abundances in all samples. This might indicate small beneficial effects of Cd elimination on the microbial community. The adjustment and increase in beneficial microbes are the focus and direction of our future heavy metal pollution remediation work.

Conclusions
In this study, the geographic distribution of cadmium contents and its correlation with the bacterial community composition in the Longjiang River were analysed.
Although the water quality of the Longjiang River met the second level national environmental quality standards of surface water, the Cd contents in the farmland samples were still beyond the environmental quality standard for soil (GB 15618-1995).
Cd was precipitated from the river water to sludge, and the weak alkaline chemical precipitation technology used to remove Cd from the river water has been effective since 2012. However, the Cd in the sludge was not stable, as it was constantly dissolving and being released from the sludge to the water. The Cd content of every phase showed significant and positive correlations with the total Cd content of the sludge. The Cd content of each phase in the sludge was mainly influenced by the total Cd content of the sludge rather than other factors.
Microbial abundance and microbial diversity were limited in the sewage with higher Cd content. The abundances of 2 phyla (Gemmatimonadetes and Proteobacteria) and 39 genera (such as Pseudoxanthomonas, Halomonas, and Methanomassiliicoccus) had significantly positive correlations with the Cd distribution (p < 0.05), while those of 10 phyla (GOUTA4, Fibrobacteres, Kazan-3B-28, WS2, WS4, OP8, KSB3, Planctomycetes, Chlamydiae and Caldiserica) and 6 genera (such as C1_B004, Clostridium, and vadinHB04) had significantly negative correlations with the Cd distribution. The taxa positively correlated with Cd content might be beneficial for Cd precipitation and remediation.

Materials and Methods
Sampling site information. Guangxi is located in the western region of southern China (east longitude 104°26′-112°04′, north latitude 20°54′-26°24′). This region has abundant rainfall and no obvious seasonal variations. Guangxi Zhuang Autonomous Region is rich in mineral resources and is called the "home of nonferrous metal". The Longjiang River is located in Guangxi Zhuang Autonomous Region. The Longjiang River basin is a non-ferrous metal producing region and a grain producing region. A serious cadmium (Cd) pollution accident occurred in the Longjiang River in 2012.

Sample collection. Sampling began at the junction of Guizhou province and Guangxi Zhuang Autonomous
Region and ended at the Hexi water plant in July of 2014; the sampling area is approximately 307 km in length with sampling intervals of approximately 8 km. 39 water samples and 39 sludge samples from this river and 2 soil samples from the nearby farmland were collected, and a detailed map is provided in Fig. S1. Two 500 mL water samples and two 500 g sludge samples from each river sampling site and two 500 g soil samples from each farmland site were collected using sterile plastic bottles. The water samples were collected directly from the river surface. The sludge and soil samples were collected with a Luoyang shovel. Only the middle part of the samples in the Luoyang shovel, which represented the part approximately 5 cm-10 cm under the surface, was collected in the sterile plastic bottles. All the samples for chemical analysis were stored at 4 °C before analysis. For the microbial community analysis, the sludge samples were centrifuged at 5,000 g for 10 min, and the pellets were stored at −80 °C before DNA extraction.

Physicochemical analysis of the samples.
To estimate the amount of Cd bound to different phases in the soils and sludge, the sequential extraction procedure, consisting of a series of chemical extractions, was carried out using the following steps 7 : Collection of Exchangeable Fraction. 1.000 g of dry sludge sample was weighed and transferred to a 50 mL capped centrifugable bottle. 15 mL of 1 M MgCl 2 (pH = 7.0) was added, and the sample was mechanically shaken for 2 h. Then, it was centrifuged for 15 min at 8000 rpm, and the exchangeable fraction was collected from the supernatant.
Collection of Carbonate and Sulphide Bound Fraction. 15 mL of 1 M NaAc (pH = 5.0) was added to the residue, and it was mechanically shaken for 2 h. Then, it was centrifuged for 15 min at 8000 rpm, and the carbonate and sulphide bound fraction was collected from the supernatant.

Collection of Fe and Mn Oxide Bound
Fraction. 20 mL of 0.04 M NH 2 OH · HCl was added to the residue, leached for 5 h at 96 °C and shaken occasionally. Then, another 10 mL of 0.04 M NH 2 OH · HCl was added, the solution was centrifuged for 15 min at 8000 rpm after cooling, and then the Fe and Mn oxide bound fraction was collected from the supernatant.
Scientific RepoRts | 7: 227 | DOI:10.1038/s41598-017-00280-y Collection of Organic Matter Bound Fraction. 3 mL (0.02 M) of HNO 3 and 10 mL of H 2 O 2 (30%) was added to the residue, the pH was adjusted to 2.0 with HNO 3 , and the solution was heated for 2 h at 85 °C. Then, another 3 mL of H 2 O 2 was added (adjusted to pH = 5.0 with HNO 3 ), and the residue was maintained at 85 °C for 3 h. Five millilitres of NH 4 Ac (3.2 M) was added after cooling, and the residue was diluted to 20 mL with deionized water and shaken for 1 h. After that, it was centrifuged for 15 min at 8000 rpm, and the organic matter bound fraction was collected from the supernatant.
Collection of Residual Fraction. 5 mL of HF and 10 mL of HClO 4 were added to the residue, which was heated until the liquid was completely evaporated, and the above steps were repeated three times. Then, 1 mL of HClO 4 was added, and the solution was heated until the liquid was completely evaporated and white smoke fumed, following which 0.5 mL of concentrated HCl was added, and deionized water was used to adjust the solution to 25 mL.
After the sequential extraction, all the samples were filtered with super membrane filters (0.2 mm pore size, Sigma-Aldrich, MO, USA) and analysed using ICP-OES 67,68 . The river water samples were analysed using ICP-OES directly after filtration with super membrane filters (0.2 μm pore size, Sigma-Aldrich).
DNA extraction and sequencing. The total DNA of sludge samples was extracted using an UltraClean Soil DNA Isolation Kit (MO BIO) according to the manufacturer's instructions, and the quality and concentration of the extracted DNA was measured using NanoVue plus. 16S rRNA gene amplifications of microbes were conducted with the 340F/805 R primer set 340F: CCTACGGGNGGCWGCAG and 805R: GACTACHVGGGTATCTAATCC, which amplifies the V4 region of the 16S rDNA gene 69 . The amplification products were confirmed by electrophoresis. High throughput sequencing targeting 16S rRNA genes was conducted on an Illumina MiSeq platform 70 Data processing. Paired-end reads of the original DNA fragments from high throughput sequencing were merged using FLASH 72 and assigned to each sample according to their unique barcodes. The 16S rRNA genes were processed and analysed using the open-source software QIIME 73,74 , and in-house Perl scripts were used to analyse alpha (within samples) and beta (among samples) diversity. First, sequence reads were filtered (fastq maxee = 0.5 and fastq trunclen = 289), replication was removed, and singletons were discarded 75 . The Chimera Slayer (CS) tool was used for chimaera detection 18 . Then, the CD-HIT package 76 and the QIIME script "pick_de_ novo_otus.py" 73 were used to identify OTUs by making a OTU table, and sequences with ≥97% similarity were assigned to the same OTUs 77 . Representative sequences for each OTU were selected, and the RDP classifier was used to annotate taxonomic information for each representative sequence 78 . To compute the alpha diversity, the OTU table was rarefied, and two metrics were calculated: Chao1 estimates microbe abundance, and the Shannon index is used to estimate the number of unique OTUs found in each sample. Rarefaction analysis was used to quantify the representativeness of the sequencing dataset 79 . Hierarchical cluster analysis was carried out using Bray-Curtis similarity based on the abundance of all OTUs in the stats package of R 80 .

Statistical analysis.
When performing the statistical analysis, 41 samples were divided into 6 groups, which were branch river control (n = 3), upstream control (n = 6), sewage outfall (n = 6), dosing sites (n = 7), midstream (n = 16), and farmland (n = 2) ( Table S1). The physicochemical indices were statistically analysed with separate one-way analyses of variance (ANOVA). The correlations (correlation between Cd content of sludge and Cd concentration of river water; the correlation between Cd content of five different phases in sludge and the total Cd content in sludge; the correlation between Cd content in sludge and microbial richness (Chao1 index); the correlation between Cd content in sludge and the alpha diversity (Shannon index) of the communities in the sludge samples; the correlation between Cd content in sludge and microbial structure) were analysed with bivariate correlation, and these statistical analysis were performed using SPSS 19.0 for Windows 81 . To evaluate the effects of environmental factors on overall functional community structures, CCA was implemented with the CANOCO 4.5 software package 82 .