Impact of selenium on rhizosphere microbiome of a hyperaccumulation plant Cardamine violifolia

Cardamine violifolia is the only selenium hyperaccumulation plant found in China. It has been developed as a source of medicinal and edible products that we can consume as selenium supplements. Many planting approaches have been developed to increase the selenium content of C. violifolia for nutrient biofortification. However, the contribution of rhizosphere microbes of C. violifolia to selenium enrichment has not been investigated. In this study, four types of selenium, i.e., selenate, selenite, nanoparticles selenium from Bacillus subtilis (B. subtilis-Se), and organic selenium from yeast (yeast-Se), were added to the soil that C. violifolia was grown in, respectively. Selenate led to the greatest accumulation of selenium in C. violifolia, followed by selenite, B. subtilis-Se, and yeast-Se. Except for yeast-Se, the concentration of selenium in C. violifolia positively correlated with the amount of selenium added to the soil. Furthermore, the different types of exogenous selenium exhibited distinct effects on the rhizosphere microbiome of C. violifolia. Alpha and beta diversity analyses demonstrated that rhizosphere microbiome was more obviously affected by selenium from B. subtilis and yeast than from selenate and selenite. Different microbial species were enriched in the rhizosphere of C. violifolia under various exogenous selenium treatments. B. subtilis-Se application enhanced the abundance of Leucobacter, Sporosarcina, Patulibacter, and Denitrobacter, and yeast-Se application enriched the abundance of Singulishaera, Lactobacillus, Bdellovibrio, and Bosea. Bosea and the taxon belonging to the order Solirubrobacterales were enriched in the samples with selenate and selenite addition, respectively, and the abundances of these were linearly related to the concentrations of selenate and selenite applied in the rhizosphere of C. violifolia. In summary, this study revealed the response of the rhizosphere microbiome of C. violifolia to exogenous selenium. Our findings are useful for developing suitable selenium fertilizers to increase the selenium hyperaccumulation level of this plant.


Introduction
Selenium (Se) is an essential micronutrient with strong bioactivity for all kinds of organisms, including humans, animals, plants, and microorganisms (White and Brown 2010). However, the safety margin of Se is narrow; for adult humans, the recommended dietary reference Se intake is 55-70 µg Se/day, though the Institute of Medicine (USA) suggests a tolerable upper intake of 400 µg Se/day (EFSA Panel on Dietetic Products NaAN 2014). Excessive Se intake can cause hair and nail loss and nervous system disorders (selenosis) in humans, symptoms similar to those caused by toxic metals (Li et al. 2012;Rayman et al. 2018;White 2016). Conversely, a Se deficiency can weaken the immune system and increase the risk of health disorders including Responsible Editor: Diane Purchase Zisheng Guo, Bin Zhu and Jia Guo contributed equally to this work.
oxidative-stress-related conditions, heart disease, reduced fertility, hypothyroidism, cardiovascular disease, and various cancers (Fairweather-Tait et al. 2011;Rayman et al. 2008;Schomburg 2017;White 2016). It has been reported that Se deficiency affects hundreds of millions of people worldwide, particularly in developing countries (Joy et al. 2014;Rita and Morse 2015). For example, approximately two-thirds of people in China have a dietary Se intake of about 40 µg/ day (Wu et al. 2015), which is lower than the recommended range (WHO 2009). Thus, it is essential to develop natural, plant-based Se biofortification products for people with severe Se deficiencies worldwide (Wu et al. 2015).
Se is not essential for higher plants, but it is for several microalgal strains. In general, the concentrations of Se in plants grown in seleniferous soil are less than 25 mg/kg dry weight, but some Se-hyperaccumulator species can accumulate over 1000 mg/kg Se in their plant tissues (White 2016). Many Se-hyperaccumulating plants have been discovered from the order Brassicales. For example, two well-known Se hyperaccumulators are the desert prince's plume (Stanleya pinnata) and two-grooved milkvetch (Astragalus bisulcatus). These were found to contain Se of up to 0.1-1.5% and 0.6% (dry weight), respectively (Freeman et al. 2006). In China, Cardamine violifolia is an important Se-hyperaccumulating plant. It was discovered in 2010 in Enshi, where Se mine drainage areas contain high levels of Se. Some hyperaccumulator cultivars of C. violifolia have also been reported, such as C. enshiensis and C. hupingshanesis (Lei et al. 2010;Yuan et al. 2013). C. violifolia growing near Se mine tailing was found to accumulate Se of up to 20,000, 18,000, and 44,000 mg/kg dry weight in its leaves, stems, and roots, respectively, making it a Se hyperaccumulator (Yuan et al. 2013). Though the Se concentration in C. violifolia is too high for human consumption, the plant can be mixed with other foods or medicines that we can consume for Se biofortification (http:// www. enshi. gov. cn/ xw/ esxw/ 202103/ t2021 0318_ 11084 88. shtml) (Wu et al. 2020).
Though many plants with high Se-tolerance have been discovered, there have been few studies investigating the mechanisms responsible for such high Se-tolerance in plants. S. pinnata contains approximately 80% Se-methylselenocysteine (MeSeCys) and 20% selenocystathionine, and its high Se-tolerance was found to be related to its levels of ascorbic acid, glutathione, total sulfur, and nonprotein thiols, as well as having been partly attributed to increased antioxidants and upregulated sulfur assimilation (Freeman et al. 2010). Further studies indicated that when selenocysteine (SeCys) was excessive, it could be methylated to form methyl-SeCys. Methyl-SeCys does not enter proteins but supports the high Se-tolerance of hyperaccumulators. However, in C. violifolia, the total Se accumulated was mostly in the form of selenolanthionine Ouerdane et al. 2020), indicating a different Se-tolerance mechanism at play. Comparative transcriptomics analysis provided certain clues indicating that storage function, oxidation, transamination, and selenation play roles in the high Se-tolerance of C. violifolia (Zhou et al. 2018). In addition to plants' physiological metabolism, many researchers have found that plant root-associated microbes help with supporting tolerance to toxic compounds Morgan et al. 2005;Stolz et al. 2006). However, the root-associated microbes in C. violifolia have not been investigated.
Se intake is dependent on its chemical form in the root substrate, which affects its bioavailability, mobility, and nutritional value (Thiry et al. 2012). Se commonly has four valence states in the natural environment: selenide (Se 2− ), elemental Se (Se 0 ), selenite (Se 4+ ), and selenate (Se 6+ ). For human health, it is generally believed that ingesting organic Se compounds such as selenomethionine (SeMet) and SeCys (Se 2− ) are safer than with inorganic forms (Se 4+ and Se 6+ ) (Lunøe et al. 2011), and some studies have found that nanoparticles of Se (Se 0 ) are less toxic among the four types of Se (Shakibaie et al. 2013;Wadhwani et al. 2016). However, the optimum form of Se to generate its maximum uptake and transformation differs among plants. Inorganic Se compounds, such as selenate and selenite, have been commonly used to grow selenized vegetables as they are more easily available than organic Se compounds for biofortification (Michela et al. 2017;Schiavon and Pilon-Smits 2017). For example, two-grooved milkvetch preferred to uptake selenite and SeCys in nutrient-solution, but not SeMet (Williams and Mayland 1992). Indian mustard shoots accumulated the greatest amount of Se through the application of selenate, followed by those supplied with SeMet and selenite (Zayed et al. 1998). When spring canola (Brassica napus) was grown hydroponically, organic forms of Se were taken up at a greater rate than inorganic forms (Kikkert and Edward 2013). For nanoparticles of Se from microbes, no application in plants has yet been investigated.
The best Se compound for uptake and transformation by C. violifolia remains unclear. To understand the Se hyperaccumulation mechanisms in C. violifolia, it is significant to investigate the rhizosphere physical, chemical, and biological processes, which may affect the bioavailability, uptake, distribution, and transformation of Se in the plant. Thus, in this study, we examined the effects of the different types of exogenous Se (selenate, selenite, nanoparticles Se from Bacillus subtilis (B. subtilis-Se), and organic selenium from yeast (yeast-Se)) with various levels of these applied on the growth and Se accumulation of C. violifolia, and analyzed the corresponding response of root-associated bacteria. The results are useful for understanding the Se accumulation mechanisms and extending our knowledge of how we can develop Se biofortification or phytoremediation techniques in the future.

Chemicals and reagents
The biochemical reagents used in this study were analytical grade. Selenite and selenate were bought from Sigma-Aldrich (St. Louis, MO, USA). Yeast-Se (Se content 2000 mg/kg, 98% organic Se, ≥ 40% protein, ≤ 6% water) was purchased from Angel Yeast (Yichang, China). The B. subtilis strain used in this study can tolerate up to 120 mM selenite. Nanoparticles of Se were prepared by B. subtilis fermentation in an LB medium with 10 mM selenite sodium (Se content 1500 mg/kg, 90% Se nanoparticles with 50-200 nm diameter).

C. violifolia growth and treatment
C. violifolia was grown and treated as previously reported (Wu et al. 2020). In brief, after germination and growth for one month in peat soil, the seedlings were transplanted and cultivated for three months in the trial field in a greenhouse in a the nutritional substrate mixture of peat soil and vermiculite (3:1). Then, plants with similar growth were transplanted into flowerpots with a diameter of 32 cm and height of 29 cm for one month of adaptation. Four Se solutions with various concentrations (0, 50, 100, 200, 400, and 800 mg/L) were separately added to the substrates. Once every 20 days, 750 mL Se solution was added to the flowerpots by root application for three times. The control group was treated by adding equal amount of deionized water. Each treatment mode had three replicates. After 90 days of cultivation, the plants were harvested for subsequent analysis.

Soil sampling, DNA extraction, and 16S rRNA gene sequencing
After the loosely bound soil of C. violifolia was shaken off, tightly associated soil samples were collected. The samples were divided into two: half were air-dried, ground, and sieved through a 2-mm screen for physiochemical analysis, and the other half were stored at − 20 °C for total DNA extraction.
Total genomic DNA was extracted from 0.2 g soil using a PowerSoil DNA Isolation Kit according to the manual protocol (Mobio Laboratories, USA). The extracted DNA concentration was measured using a NanoDrop 1000 spectrophotometer (Thermo Scientific, USA). After amplifying the V4-V5 region of the bacterial 16S rRNA gene using the primer sets F515 (5′-GTG CCA GCMGCC GCG G-3′) and R907 (5′-CCG TCA ATCMTTT RAG TTT-3′), the bacterial community composition was investigated by high-throughput sequencing the amplified products using an Illumina HiSeq 2500 instrument from Novogene Biotechnology (Beijing, China). The raw reads were filtered and analyzed using QIIME 1.9.1 (Caporaso et al. 2010). The sequences with a similarity of above 97% were clustered into operational taxonomic units (OTUs) using UCLUST clustering (Edgar 2010). The bacterial taxa dominance was analyzed based on taxonomy information.

Se quantification
The total Se concentration in plants and soils was measured using hydride generation atomic fluorescence spectrometry (HG-AFS) as previously reported (Wu et al. 2020). Briefly, samples of approximately 1 g were weighed, treated using 8 mL of nitric acid in digestion bottles, and heated in a microwave digestion system. The digestion process was 130 °C for 2 min, 150 °C for 2 min, and 180 °C for 15 min. The solution was evaporated to 2 mL, transferred to a 50-mL volumetric flask, and diluted to 50 mL with ultrapure water. The blank group was prepared using the same method. Then, the digestion solution was extracted, mixed with 2.0 mL of 6 M HCl, and fixed to 10 mL in a 15-mL tube for Se determination. The Se content was calculated with a standard curve.

Statistical analysis
The data obtained by sequencing were analyzed with QIIME 2 (Hall and Beiko 2018). Initially, all reads were trimmed to a minimum length of 250 bp and at least a Phred score of 20 using DADA2. The remaining sequences were clustered into OTUs. The alpha diversity was characterized by the Shannon index, observed OTUs, and evenness. A Kruskal-Wallis test was used for differential analysis of the alpha diversity. The beta diversity was quantified by the Bray-Curtis distance, followed by differential analysis using a Mann-Whitney U test. Additionally, the dissimilarity of the microbial profiles was visualized by non-metric multidimensional scaling (NMDS) using the Bray-Curtis distance. Taxonomic information for each OTU was assigned using the SILVA v.132 taxonomy classifier (Parks et al. 2018). Quality OTUs were clustered at the 97% level of identity and a taxa table affiliated at the species level was used for analysis in the next steps. All the taxa with abundances of more than or equal to 0.01% (or 0.001%)s in at least 5% (or 15%) of the samples were kept for the following analysis. Linear regression was performed to test the correlation between Se treatment and the abundance change of each taxon using the "lm" function in R. The p and R values of the results of the linear regression were adjusted by an FDR of 5% using the Benjamini-Hochberg method for multiple comparisons. Additionally, a taxa table affiliated at the phylum level was prepared and the rhizosphere microbiome profiles of four Se sources were compared to a plant control at the phylum level using a Mann-Whitely U test, followed by adjustment of p-values using the same FDR method.

Different levels of exogenous Se available to C. violifolia
The experiments were designed to examine the effects of exogenous Se on its enrichment and the rhizosphere microbiome of C. violifolia (Table 1). Four Se sources with different concentrations (0, 50, 100, 200, 400, and 800 mg/L) were added to soils, and the Se concentrations in the soils and C. violifolia were compared in different treatments. The total Se content of C. violifolia rose with an increase of the Se concentration applied to the soil, and it was highest under the selenate treatment (6808 mg/kg), followed by selenite (1520 mg/kg), B. subtilis-Se (446 mg/kg), and yeast-Se (128 mg/kg) (Fig. S1). Consistent with our previous work (Wu et al. 2020), among these different Se sources, selenate was the best for increasing the Se concentration in the leaves of C. violifolia, followed by selenite, B. subtilis-Se, and yeast-Se in the concentrations tested. The concentration of Se in the roots of C. violifolia was positively related to supplementation of B. subtilis-Se, selenite, and selenate in the soils. However, the addition of yeast-Se in the soil was not linearly related to the Se concentration in the roots of C. violifolia.
To analyze the exogenous Se availability to C. violifolia, the ratios of Se concentrations in this plant to those in the soils were investigated in different treatments. When there was no supplementation of exogenous Se, the concentration of Se in the soil was 1.54 ± 0.4 mg/kg. The addition of these four types of Se increased the Se concentration in the soils. However, only the selenate group had a higher availability than the plant control group (C. violifolia without Se addition) (Fig. 1). In addition, the ratio was not decreased when the concentration of exogenous selenate was increased. These data suggested an unknown mechanism in C. violifolia for actively assimilating selenate. The ratios of the other three Se sources were lower than that of the plant control. The ratio reduced in higher concentrations of yeast-Se and selenite and fluctuated for B. subtilis-Se, indicating that C. violifolia could not utilize high concentrations of yeast-Se, selenite, or B. subtilis-Se. A lot of the extra yeast-Se, selenite, and B. subtilis-Se could not be absorbed by C. violifolia; instead, it might have remained in the soil or volatilized, which might have benefited C. violifolia growth.
Se is indispensable for humans, but it is not clear if the trace element Se is indispensable for plants. Many studies found that Se showed beneficial effects for various plants. It promoted growth, increased the tolerance to toxic metals, conferred resistance to pathogens, and protected plants from oxidative damage (Djanaguiraman et al. 2010;Pandey and Gupta 2015;Schiavon and Pilon-Smits 2016). However, the beneficial effects of Se were usually identified at a low level, and a high concentration  . 1 Ratio of selenium concentration in C. violifolia to that in the soil after supplementation with one of four selenium sources, including nanoparticles of selenium from Bacillus subtilis, organic selenium from yeast, selenate, or selenite. There is no selenium addition for the plant control of exogenous Se application inhibited plant growth or led to Se toxicity (Hui et al. 2011). Our results showed that high concentrations of yeast-Se (> 200 mg/L) had a significant inhibitory effect on C. violifolia, indicating that organic Se or other components of yeast-Se with high concentrations might be toxic to this plant. Thus, the concentration of Se in the roots of C. violifolia did not increase with the amount of Se in the soil. Furthermore, our previous study indicated that applying a high concentration of exogenous Se affected the plant nutrition quality indexes, elements' uptake, and antioxidant responses (Wu et al. 2020). Selenate (200 mg/L) was the best option for C. violifolia to accumulate Se of the different forms of Se tested in this study, which might be because selenate was more readily assimilated through the transporter in the root cell membrane (Wu et al. 2020). Therefore, the source and concentration of Se should be taken into account when exogenous Se fertilizers are applied. In addition, other soil physiochemical parameters, especially the total and bioavailable (mostly sulfate) sulfur contents, also need to be considered, as sulfur has the similar uptake pathways to those of plants with Se (Sors et al. 2005) and it may affect the effectiveness of Se fertilizers.

Influence of Se sources on the diversity of C. violifolia rhizosphere microbiome
The rhizosphere microbiota has been reported to help plant growth by acquiring nutrients or protecting against harmful substance in the soil (Mendes et al. 2013). To investigate which rhizosphere microbes may promote Se enrichment in C. violifolia, the first thing is to characterize the impact of Se on the rhizosphere microbiome of C. violifolia. In this study, microbial profiles were compared under different exogenous Se treatments (Table 1) to identify the bacterial species with increased abundances in higher concentrations of exogenous Se. Alpha rarefaction plotting of Shannon/OTU suggested that the sequencing depth was enough to cover most of the species in the rhizosphere microbiome. The alpha diversity was assessed using three indices, the Shannon index, observed OTU, and evenness. The related differences were analyzed using a Kruskal-Wallis test. Most of the Se treatments did not affect the Shannon diversity in the rhizosphere microbiome of C. violifolia except for 800 mg/L B. subtilis-Se and 200 mg/L yeast-Se (Fig. 2b). The different alpha diversities in these two sets of microbiomes were caused by both changes of OTU numbers and microbial evenness ( Fig. 2a and c Supplementation with B. subtilis-Se reduced Fig. 2 Observed OTUs (a), Shannon index (b), and evenness index (c) of the rhizosphere microbiome affected by four types of selenium addition, including nanoparticles of selenium from Bacillus subtilis, organic selenium from yeast, selenate, and selenite. There is no selenium addition in the plant control group. "No plant" indicates samples without either plant or selenium addition. *p-value ≤ 0.05 in the Kruskal-Wallis test the observed OTUs at 100, 400, and 800 mg/L and decreased the evenness and Shannon indices at the 800 mg/L level (Fig. 2), indicating that some rhizosphere microbes could be inhibited or killed by B. subtilis-Se. In 200 mg/L yeast-Se treatment, more species were observed and the microbial evenness was greater (Fig. 2c). However, higher concentrations of yeast-Se reduced the alpha diversity to the level of the control group. One potential is that 200 mg/L yeast-Se might contain nutrients that help the growth of some rhizosphere microbiota, but a low concentration (< 100 mg/L) does not influence soil microbes, and a high concentration (≥ 400 mg/L) impairs the growth of rhizosphere microbes. As B. subtilis-and yeast-Se were not purified Se compounds, the effects of these on the rhizosphere microbiome might have been induced by other components of these Se sources. Furthermore, soil microbes with high proteolytic activity were favored and selectively supported when yeast-Se with a high protein content was added instead of inorganic Se.
Nevertheless, the influence of B. subtilis-and yeast-Se on the root microbiome is thought to be involved in the observed changes.
The beta diversity was quantified using the Bray-Curtis distance and the difference between each sample and the plant control was calculated using a Mann-Whitney U test (Fig. 3a). Furthermore, ordination analysis was performed via NMDS (Fig. 3b). The root microbiome was less similar to the plant control with B. subtilis-Se supplementation, suggesting an impact of B. subtilis-Se on the rhizosphere microbiome of C. violifolia in a concentration-dependent manner (Fig. 3). This phenomenon was consistent with the reduced alpha diversity found by B. subtilis-Se treatment, which may be because some microorganisms were inhibited or killed by exogenous B. subtilis-Se (Fig. 2). Although selenate was the best Se source for C. violifolia (Fig. 1), the alpha diversity of the rhizosphere microbiome was not affected by selenate (Fig. 2), and the beta diversity was only affected by 50 mg/ Fig. 3 a Beta diversity of the rhizosphere microbiomes as measured using Bray-Curtis distances and as visualized by NMDS. The rhizosphere microbiomes are color-coded by the supplemented selenium source and the concentration of the selenium sources is marked using different shapes. b Bray-Curtis distance from each rhizosphere microbiome to the rhizosphere microbiome of the plant control. "No plant" indicates samples without either plant or selenium addition. *p-value ≤ 0.05, **p-value ≤ 0.01 in the Mann-Whitney U test mL of selenite (Fig. 3). A reasonable explanation is that most of selenate was assimilated by C. violifolia, and as a result, selenate had a low concentration in the soil (Fig. 1b), which might have limited its influence on the rhizosphere microbiome. As such, selenate was the closest group to the plant control in NMDS figure (Fig. 3b). The beta diversity of the "no plant" group was significantly different from that of the plant control (Fig. 3), suggesting that C. violifolia affected the soil microbiome. Significantly less evenness of the microbiome in plant control samples than in the "no plant" samples illustrated how the existence of C. violifolia might enrich some microorganisms in the soil (Fig. 2).

Species affected by Se sources in rhizosphere microbiome of C. violifolia
We trained a QIIME 2 feature classifier using the SILVA database to identify species in the rhizosphere microbiome of C. violifolia. The top five phyla in the rhizosphere microbiome of C. violifolia were Actinobacteria (42.3%), Proteobacteria (32.4%), Bacteroidetes (7.1%), Firmicutes (3.8%), and Patescibacteria (3.7%) (Fig. 4). The rhizosphere microbiome profiles of four Se sources were compared to the plant control at the phylum level using a Mann-Whitely U test. The application of B. subtilis-and yeast-Se slightly changed the bacterial abundance at the phylum level compared to selenate and selenite (Fig. 4). B. subtilis-Se application enriched the abundance of Deinococcus-Thermus, Dependentiae, and Patescibacteria, and reduced that of Epsilonbacteraeota, Elusimicrobia, Chlamydiae, and Acidobacteria. Yeast-Se application enriched the abundance of Patescibacteria, Acinobacteria, and Firmicutes, and reduced that of Epsilonbacteraeota, Spirochaetes, Elusimicrobia, Chlamydiae, and Acidobacteria. The change profiles were similar for selenate and selenite application, and they both increased Deinococcus-Thermus, Halanaerobiaeota, Acitnobacteria, Patescibacteria, and Dependentiae, and reduced Epsilonbacteriaeota, Spirochaetes, Verrucomicrobia, Chlamydiae, and Acidobacteria. The only differences were that selenite also increased Firmicutes and reduced Elusimicrobia (Fig. 5).
Although most of the exact species were not identified, some genera differences were found between the rhizosphere microbiome profiles of four types of Se application and the plant control (Fig. 6). B. subtilis-Se application changed the most types of microbes, and it enriched the abundance of Leucobacter, Sporosarcina, Patulibacter, and Denitrobacter, while it reduced that of Mycobacterium. Yeast-Se application enriched the abundance of Singulishaera, Lactobacillus, Bdellovibrio, and Bosea. The selenite sample enriched only Bosea, and the selenate sample was found to enrich only Solirubrobacterales at the order level. Furthermore, the abundance increases of Solirubrobacterales and Bosea were, respectively, positively related to the concentrations of selenate and selenite application as quantified by linear regression (Fig. 6b and c). Fig. 4 Relative abundance of the rhizosphere microbiomes at the phylum level in the four sample groups with one of four types of selenium addition, including nanoparticles of selenium from Bacillus subtilis, organic selenium from yeast, selenate, and selenite. There is no selenium addition in the plant control. There is no selenium addition and no plant in the "no plant" group Taxa enrichment generated by selenium sources at the phylum level. The taxa abundance in the rhizosphere microbiomes supplemented with 800 mg/L of one of four selenium sources was compared to that of the plant control in each case. The difference in abundance was calculated using the ADLEx2 package in R and quantified using the per-feature median difference between two conditions. Adjusted p-values were generated with a Wilcoxon test and corrected using Benjamini-Hochberg procedure Leucobacter was reported to tolerate high concentration of toxic metals, such as chromate, and therefore, it was used for toxic metal reduction (Zhu et al. 2008). Another previous study found that B. subtilis application increased the Leucobacter abundance (Ding et al. 2021), indicating that some components or metabolites of B. subtilis benefit Leucobacter growth. In our result, B. subtilis-Se application generated the biggest increase of abundance for Leucobacter, indicating that Leucobacter might tolerate for high concentration of Se. Certain species belonging to Singulishaera were often identified in acidic environments (Kulichevskaya et al. 2008(Kulichevskaya et al. , 2012, with similar condition to yeast's growth environment. Although yeast-Se is inactive, it is speculated that some organic compounds from yeast might benefit Singulishaera enrichment. Solirubrobacterales was ubiquitous in natural environments, including rocky areas, croplands, and lakes (Khilyas et al. 2019;Liu et al. 2020), and was also found to be enriched in high-selenium-contaminated reclaimed mine soils (Rosenfeld et al. 2018), which is consistent with our results. Some reports showed that Bosea was present in the rhizosphere or root nodules of plants (Qaisrani et al. 2019;Safronova et al. 2020;Sazanova et al. 2019), and it was also frequently found to perform a denitrification function or play antioxidant protective roles (Hassan et al. 2019;Tian and Wang 2021;Wang et al. 2020;Zhang et al. 2021). These results indicated that Bosea might use selenite as an electron receptor and play key roles during the selenite reduction in the rhizosphere of C. violifolia. Taken together, these results showed that the various Se source application shaped different rhizosphere microbiome profiles. Enrichment microbes might affect the Se uptake and transform of C. violifolia by regulating the expression of Se metabolism-related genes or releasing secretory products to alter the soil properties and Se bioavailability (Hatfield et al. 2014;Yang et al. 2021). In this study, the genome sequences for most of the identified bacteria were not available because it was difficult to isolate and culture them. In future studies, it is essential to isolate these microorganisms and study their genes related to Se metabolism after genomic sequencing.

Conclusions
C. violifolia exhibited distinct trends of absorption efficiency depending on the form of exogenous Se applied. Selenate caused the greatest level of accumulation in this Fig. 6 a Enrichment of taxa by selenium sources at the species level. The abundance of taxa in the rhizosphere microbiomes supplemented with 800 mg/L selenium from one of four selenium sources was compared to that of the plant control in each case. Abundance difference was calculated using the ADLEx2 package in R and quantified using the per-feature median difference between two conditions. Adjusted p-values were generated with a Wilcoxon test and corrected using the Benjamini-Hochberg procedure (b and c). We carried out linear regression of the taxa abundance in the rhizosphere microbiome when supplemented with different concentrations of selenium sources. The p-values, R-values, and coefficients were measured using the stats package in R, and the p-values were corrected using the Benjamini-Hochberg procedure plant and thus was found to be the most suitable option for Se supplementation. High concentrations of exogenous Se had significant effects on the rhizosphere microbial profiles especially for B. subtilis-and yeast-Se application. Various microorganisms were recruited into the rhizosphere of C. violifolia under four different exogenous Se treatments. The abundance of enriched microbes such as Bosea and the taxon belonging to the order of Solirubrobacterales, respectively, correlated positively with the concentrations of selenate and selenite. This study extends our knowledge of the rootassociated microbes of C. violifolia with various Se sources applied. It will be worthwhile investigating the role of bacteria enriched by exogenous Se in the accumulation of Se in C. violifolia in the future.