Growth indices of M. floridulus (Lab.)
In general, the addition of L. jeotgali MR2 and K. michiganensis TS8 significantly (p < 0.05) increased plant height, fresh weight and dry weight of M. floridulus (Lab.) after 9 weeks of planting. (Fig. S1). Compared with the CK group (33.07 cm), the plant height was increased by 41.22%, 59.81% and 20.40%, respectively, in the MR (46.70 cm), TS (52.85 cm) and MT (39.82 cm) groups (Fig. 1a). In addition, fresh plant weight (Fig. 1b) and dry weight (Fig. 1c) were also increased after inoculating this two PGPB: Compared with the CK group (3.06 g; 1.18 g), the fresh weight in the underground and overground was increased by 0.35–1.01 and 1.28–3.05 times, respectively in the other groups (4.14–6.16 g; 2.68–4.77 g); compared with the CK group (0.84 g; 0.29 g), the dry weight in the underground and overground parts was increased by about 2.5 and 1.3 times, respectively in the other groups (2.99–3.10 g; 0.67–0.71 g). Overall, K. michiganensis TS8 showed the highest growth-promoting efficiency.
Antioxidant enzyme activities and chlorophyll content in M. floridulus (Lab.)
To some extent, the activities of SOD, POD and CAT in plants were changed by the addition of MR2 and TS8 (Fig. 2). Compared with the CK group (SOD, 1164.14 U·g− 1; POD, 373.02 U·g− 1; CAT, 34.54 nmol·min− 1·ml− 1) in the underground parts, MR2 inoculants could significantly increase the activities of POD (1682.49 U·g− 1) and CAT (74.63 nmol·min− 1·ml− 1); TS8 inoculants could significantly increase the POD activity (857.87 U·g− 1); and the mixed strains promoted the activities of these three enzymes (SOD, 1939.97 U·g− 1; POD, 2715.59 U·g− 1; CAT, 83.50 nmol·min− 1·ml− 1). Compared with the CK group in the aboveground parts, only the CAT activity was significantly increased by the TS8 inoculant, while the additions of MR2 and mixed strains decreased it. In addition, in the MR group (077 mg·g− 1; 0.26 mg·g− 1), compared to the CK group (1.27 mg·g− 1; 0.42 mg·g− 1), the chlorophyll a and b content in the leaves was only significantly reduced by 37.16–39.05%.
Metal(loid)s contents in M. floridulus (Lab.)
Compared with the CK group (5.49 mg/kg), the three different treatments significantly reduced As and Cu accumulation in the underground part of M. floridulus (Lab.) by 47.91%-77.60% and 29.06%-37.31%, respectively (Table 1). However, Cd accumulation in the subterranean part of M. floridulus (Lab.) was significantly increased by 64.71% and 76.47% in MR and MT groups; Pb accumulation was enhanced considerably by 40.41% in the TS group. Compared with the CK group in the overground part, Cd accumulation was significantly increased by 4.54 and 3.24 folds in the MR and TS groups, and Pb accumulation was significantly enhanced by 61.62% and 102.22% in the TS and MT groups. Besides, the concentrations of Zn showed no significant differences among all groups.
Table 1
Different metal(loid)s concentrations in M. floridulus (L.) underground and overground parts under the different treatments. (Unit: mg·kg− 1)
| Underground | Overground |
CK | MR | TS | MT | CK | MR | TS | MT |
As | 5.49 ± 0.31a | 1.23 ± 0.04c | 1.76 ± 0.28c | 2.86 ± 0.24b | 0.94 ± 0.08ab | 1.50 ± 0.19ab | 2.50 ± 1.29a | 0.71 ± 0.03b |
Cd | 0.34 ± 0.04b | 0.53 ± 0.15a | 0.43 ± 0.06ab | 0.56 ± 0.05a | 0.33 ± 0.11b | 1.83 ± 0.44a | 1.40 ± 0.27a | 0.43 ± 0.24b |
Cu | 26.43 ± 4.43a | 18.52 ± 1.62b | 16.57 ± 1.26b | 18.75 ± 0.72b | 23.32 ± 2.92b | 17.29 ± 1.10 c | 22.53 ± 1.65b | 35.76 ± 7.91a |
Pb | 17.84 ± 0.71b | 9.04 ± 1.40c | 25.05 ± 0.67a | 13.65 ± 3.03bc | 9.90 ± 2.33b | 9.43 ± 0.21b | 16.00 ± 3.81a | 20.02 ± 0.51a |
Zn | 58.92 ± 5.57a | 46.87 ± 12.11a | 43.99 ± 0.93a | 56.26 ± 4.37a | 65.39 ± 5.74a | 55.15 ± 8.02a | 54.44 ± 6.83a | 51.83 ± 10.10a |
Soil physicochemical properties and enzymes activities
Soil enzyme activities including SOD, POD and CAT, were shown in Fig. 3a-d. Compared with the CK group, the activities of three enzymes were significantly increased in the MR and MT groups by 48.95-354.17%. TS8 showed an insignificant effect on the enzyme activities. Besides, the pH values were significantly lower in the three experimental groups (MR: 6.67; TS: 6.61; MT: 6.54) than in the CK group (6.84). MR2 and mixed strains significantly improved the contents of available phosphorus (19.07% and 23.02%) and potassium (15.34% and 17.79%) in soil compared with the CK group (Fig. 3e-h). In addition, TS8 also significantly improved the available potassium content by 10.99%. Still, it significantly reduced the content of soil organic matter (41.80 ± 6.63 g·kg− 1) (Fig. 3c). The content of TN in the MR, TS and MT groups showed no significant difference compared with the CK group (Fig. 3d).
Metal(loid)s morphology and contents in soil
The PGPB inoculants showed some changes in the metal(loid)s content of the soil (Table 2). The concentrations of As, Pb and Zn in the MR (23.91 mg/kg; 2768.66 mg/kg; 818.61 mg/kg), TS (25.11 mg/kg; 2567.52 mg/kg; 851.56 mg/kg) and MT (25.81 mg/kg; 2750.17 mg/kg; 740.61 mg/kg) groups were significantly decreased by 15.27–21.50% and 14.59%-25.72% compared with the CK group (30.46 mg/kg; 2942.89 mg/kg; 977.04 mg/kg). The concentrations of Cd were also significantly decreased by 8.64–15.52% in the MR and TS groups. However, Cu concentrations were not significantly different between the CK group and the PGPB inoculation groups. Figure 3i showed that PGPB could significantly enhance the remediation efficiencies of metal(loid)s, including As, Cd, Pb and Zn. However, the best treatment for removing different metal(loid) was different, such that MR2 showed the best effect on remediating Cd and As, TS8 on remediating Pb, and the mixed strains on remediating Zn.
We further explored the speciation distributions of metal(loid) using the method of BCR sequential extraction (Table S1 and Figure S2). The residual fraction was the dominant fraction of soil As (86.95%-90.23%), Cu (77.67%-79.35%), and Zn (81.50%-82.77%), the reducible fraction occupied the most proportion (80.20%-80.88%) of Pb, while four fractions of Cd were relatively uniformly distributed (Figure S2). It showed that PGPB significantly (p < 0.05) decreased the fraction concentrations of metals (except Cu) to some extent, which suggested PGPB could change the transformation of speciations (Table S1). For example, compared with the CK group (0.85 mg/kg; 1.43 mg/kg), MR2 (0.61 mg/kg; 1.17 mg/kg) significantly reduced the oxidisable and residual fractions.
Table 2
Total concentration of different metal(loid)s under the different treatments. (Unit: mg·kg− 1)
| CK | MR | TS | MT |
As | 30.46 ± 0.49a | 23.91 ± 1.05b | 25.11 ± 2.01b | 25.81 ± 1.57b |
Cd | 5.09 ± 0.17a | 4.30 ± 0.24c | 4.65 ± 0.06b | 4.80 ± 0.21a |
Cu | 78.61 ± 9.6a | 79.42 ± 1.95a | 86.13 ± 3.25a | 79.95 ± 2.45a |
Pb | 2942.89 ± 30.44a | 2768.66 ± 125.12b | 2567.52 ± 37.57c | 2750.17 ± 78.4b |
Zn | 977.04 ± 49.01a | 818.61 ± 25.83bc | 851.56 ± 37.92b | 740.64 ± 15.93c |
Factor of bioconcentration and translocation of M. floridulus (Lab.) for metal(loid)s
In comparison with the control (0.07; 0.0054), MR2 (0.18) showed significantly (p < 0.05) enhanced the BCF of M. floridulus (Lab.) for Cd by 157.14% and TS8 (0.0091) significantly (p < 0.05) enhanced the BCF of M. floridulus (Lab.) for Pb by 68.52% (Table 3). For the TF of M. floridulus (Lab.) for Metal(loid)s (except Zn), significant differences were detected between the CK group and bacterial-inoculated groups (Table 3). MR2 significantly (p < 0.05) enhanced As (1.22), Cd (1.22) and Pb (1.22) translocation by 6.18, 2.56, 0.89 folds, respectively; TS8 significantly (p < 0.05) enhanced As (1.42), Cd (3.26) and Cu (1.36) translocation by 7.35, 2.36, 0.55 folds, respectively; MT mixed culture significantly (p < 0.05) enhanced As (0.25), Cu (1.91) and Pb (1.47) translocation by 0.47, 1.17, 1.67 folds, respectively.
Table 3
Bioconcentration factors and translocation factors of M. floridulus (L.) in the different metal(loid)s
| Bioconcentration factors | Translocation factors |
CK | MR | TS | MT | CK | MR | TS | MT |
As | 0.14 ± 0.01a | 0.05 ± 0.01b | 0.08 ± 0.03b | 0.08 ± 0.02b | 0.17 ± 0.02c | 1.22 ± 0.19a | 1.42 ± 0.69a | 0.25 ± 0.02b |
Cd | 0.07 ± 0.03b | 0.18 ± 0.04a | 0.13 ± 0.03ab | 0.11 ± 0.02b | 0.97 ± 0.23b | 3.45 ± 0.18a | 3.26 ± 0.20a | 0.77 ± 0.35b |
Cu | 0.33 ± 0.05a | 0.23 ± 0.02b | 0.21 ± 0.04b | 0.30 ± 0.03a | 0.88 ± 0.28c | 0.93 ± 0.13c | 1.36 ± 0.05b | 1.91 ± 0.41a |
Pb | 0.0054 ± 0.0011b | 0.0033 ± 0.0005c | 0.0091 ± 0.0022a | 0.0057 ± 0.0006b | 0.55 ± 0.11b | 1.04 ± 0.18a | 0.64 ± 0.14b | 1.47 ± 0.39a |
Zn | 0.062 ± 0.012a | 0.059 ± 0.009a | 0.054 ± 0.010a | 0.74 ± 0.014a | 1.11 ± 0.10a | 1.18 ± 0.15a | 1.24 ± 0.17a | 0.92 ± 0.12a |
Structure, diversity, and compositions of soil bacterial communities
We tested the bacterial community using 16S rRNA sequencing method, and the number of sequenced reads in each sample was more than 30,000. NMDS analysis showed clustering of the samples among different groups (Fig. 4a). Dissimilarities tests also showed significant differences in bacterial communities' structure among different groups (Table S2). Venn diagram (Fig. 4b) showed the numble of common OTUs was 1039, and the unique OTUs were 459, 350, 271 and 205, respectively, in CK, MR, TS and MT groups. There was a lower diversity of bacterial species in the MR, TS and MT groups than in the CK group (Fig. 4c). It suggested that inoculations of functional bacteria significantly affected the structure and diversity of the bacterial community.
There were 22 phyla, 130 families, 156 genera and 3303 OTUs. The community composition shifted with bacterial inoculants. At the phylum level (Fig. 4d), the dominant phyla (relative abundance > 1%) were Proteobacteria (37.46–52.10%), Actinobacteria (26.90-34.82%), Acidobacteria (8.91–11.10%), Gemmatimonadetes (4.87–7.54%), Chloroflexi (2.17–3.19%) and Nitrospirae (2.33–3.26%). Compared with the CK group, the groups with bacterial inoculants showed increases in the relative species abundance in the phylum Proteobacteria and showed decreases in the phylum Acidobacteria, Actinobacteria, Chloroflexi, Gemmatimonadetes and Nitrospirae (Fig. 4e).
At the family level, the dominant families (relative abundances > 1%) were Xanthomonadaceae, Nocardioidaceae, Gaiellaceae, Enterobacteriaceae, Hyphomicrobiaceae, Syntrophobacteraceae, Micromonosporaceae, Sphingomonadaceae, Micrococcaceae, Intrasporangiaceae, Oxalobacteraceae, etc.. The Enterobacteriaceae, Pseudomonadaceae, Comamonadaceae, Rhizobiaceae, Cytophagaceae, Rhodocyclaceae and Ellin517 families showed significant increases in relative abundance, and significantly decreased in the family Hyphomicrobiaceae, Syntrophobacteraceae, Micromonosporaceae, Nocardioidaceae and Gaiellaceae in the bacterial inoculant groups (Table S3). At the genus level (Table S4), the relative abundances of Serratia and Enterobacter, the same as MR2 and TS8 affiliated with the family Enterobacteriaceae, significantly increased in bacterial inoculant groups (Fig. 4f).
Network interactions of bacterial communities shifted under the bacterial inoculant
To discern possible microbial interactions in response to bacterial inoculants, RMT-based network approaches were used to construct molecular ecological networks (MENs) using 16S rRNA sequencing data. Table S5 shows the main topological properties of the empirical MENs of the microbial communities in the four groups. Using the same threshold (0.940), their correlations were greater than 0.722. This indicates that the degree distributions in the two constructed molecular ecological networks fit the power-law model well. There were more nodes and links in the CK group (475 nodes and 846 links) than those in the MR (392 and 765), TS (365 and 556) and MT (340 and 653) groups (Fig. 5a-d). It showed interactions of bacterial communities could be disrupted by bacterial inoculants, especially strain TS8. However, the proportion of positive links was increased in the bacterial inoculant groups (Table S5).
There were 3 module hubs and 3 connectors in the CK group, whereas only 1 connector in the TS group, and it was interesting that 8 (2 module hubs and 6 connectors) and 7 (1 module hub and 5 connectors) key nodes in the MR and MT groups, respectively (Fig. 5e). The key nodes were mainly affiliated to the phylum Proteobacteria, Actinobacteria and Chloroflexi (Fig. 5f).
Linkage of parameters in soil/plant and bacterial communities
To explore the interactions between soil/plant parameters and bacterial communities, all soil/plant parameters, e.g. antioxidant enzymes, nutrients, pH, chlorophyll, were considered and the Mantel test was performed (Table 4). The results showed that the parameters in soil (r = 0.363, p = 0.001) and plant (r = 0.397, p = 0.001) had a significant correlation with bacterial communities. Further analysis found that SOD/pH/available phosphorus in the soil, SOD/POD in underground tissues, and SOD/CAT/chlorophyll in overground tissues were significantly (p < 0.05) correlated with the structure of bacterial communities. In addition, we found soil bacterial communities showed more close relationships with the antioxidant enzymes in soil than in plant tissues, especially in the overground tissues.
To explore the relationships between different phyla/genera/typical OTUs and soil and plant parameters, Mantel tests were also performed. Similar results were found that fewer bacterial groups showed significant relationships with the parameters in the overground tissues. Phylum Proteobacteria and Bacteroidetes were significantly (p < 0.05) and positively correlated with AK, AP, CAT, and SOD in soil and SOD and POD in the underground tissues (Figure S3). Some genera, e.g., Enterobacter, Serratia, Sphingomonas, Pseudomonas, Ramlibacter, and Janthinobacterium, had significant positive relationships with the parameters in soil and underground tissues while showing a significantly negative relationship with soil pH (Figure S4). The typical OTUs, including 21 key nodes in the co-occurrence network (hubs and connectors) and 11 OTUs affiliated to Enterobacteriaceae same as inoculated strains MR2 and TS8 (Figure S5). It was interesting that the OTUs (OTU_4, OTU_3357, OTU_282, OTU_2645 and OTU_17) affiliated to Enterobacteriaceae were significantly and positively correlated to the soil nutrient and antioxidant enzymes in soil and root of Miscanthus, while the key nodes (OTU_6, OTU_264 and OTU_223) in the ecological network was on the contrary.
Table 4
The interactions between parameters in soil, plant and bacterial communities
Parameters | r | p |
Soil parameters | whole | 0.363 | 0.001 |
Antioxidant enzyme in soil | SOD in soil | 0.644 | 0.001 |
POD in soil | 0.191 | 0.103 |
CAT in soil | 0.026 | 0.345 |
Nutrient | AP | 0.379 | 0.001 |
AK | -0.028 | 0.500 |
OM | -0.080 | 0.618 |
TN | 0.160 | 0.056 |
- | pH | 0.293 | 0.008 |
Plant parameters | whole | 0.397 | 0.001 |
Antioxidant enzyme in underground tissues | SOD in underground tissues | 0.347 | 0.001 |
POD in underground tissues | 0.421 | 0.001 |
CAT in underground tissues | -0.066 | 0.714 |
Antioxidant enzyme in overground tissues | SOD in overground tissues | 0.142 | 0.026 |
POD in overground tissues | -0.092 | 0.721 |
CAT in overground tissues | 0.145 | 0.040 |
Chlorophyll | Chlorophyll a | 0.345 | 0.003 |
Chlorophyll b | 0.498 | 0.001 |
Antioxidant enzyme | Antioxidant enzyme in soil | 0.492 | 0.001 |
Antioxidant enzyme in underground tissues | 0.305 | 0.001 |
Antioxidant enzyme in overground tissues | 0.055 | 0.234 |
Key factors affected the growth and metal uptake of Miscanthus
Spearman’s tests were conducted to explore the key factors that affected the growth and metal uptake of Miscanthus (Fig. 6). We found the height, fresh and dry weight had significantly (p < 0.05) positive relationships with the antioxidant enzymes in soil/underground tissues and soil nutrients, except for total nitrogen. Different metals showed different responses to environmental parameters. The higher contents of AK, AP, SOD, POD and CAT in soil and underground tissues might enhance the translocation factor of Pb/Cu and the remediation efficiency of Zn/Pb/As. AK and organic matter could affect the BCF of Cd, AP and SOD in underground tissues could affect the BCF of Zn, total nitrogen could affect the BCF of Zn and Cu, and chlorophyll could affect the BCF of metals except for Cu. In addition, low pH benefited Miscanthus for metals uptake.
Then, we explored the relationships between the key microbial population (genera/OTUs) and the growth/metal uptake of M. floridulus (Lab.). We found that some genera played their roles in affecting both growth and metal uptake (Figure S6), including Bradyrhizobium, Catellatospora, Candidatus_Solibacter, DA101, Massilia, Mycobacterium and Pilimelia, and some genera affected only growth or metal uptake, e.g., Enterobacter, Bacillus, Pseudomonas and Sphingomonas. The results indicated that the growth of Miscanthus showed significantly positive relationships with the OTUs (OTU_4, OTU_3357, OTU_282, OTU_2645 and OTU_17) affiliated with Enterobacteriaceae while showing negative relationships with the key nodes (OTU_6, OTU_489 and OTU_223) in the ecological network (Figure S7). The correlation analysis showed the remediation efficiency of all metals showed few correlations with the selected OTUs. In contrast, the BCF and TF of metals showed different responses to the variations of the bacterial community. Among these, the BCF of Cu and Pb and the TF of Cu, Pb and As were closely correlated to some OTUs. The correlation coefficient was the opposition between the BCF of Cu and Pb. For example, OTUs affiliated with Enterobacteriaceae showed positive relationships with the BCF of Pb while negatively correlated with the BCF of Cu.
The above 47 parameters were classified into plant chlorophyll, antioxidant enzymes in overground tissues, antioxidant enzymes in underground tissues, antioxidant enzymes in soil, soil nutrition, soil pH, bacterial OTUs affiliated to Enterobacteriaceae and bacterial OTUs in the ecological network, and further correlation analysis was conducted (Fig. 6c). It was found that the roles of antioxidant enzymes in overground tissues and bacterial OTUs in the ecological network were relatively weaker than other parameters in the metal uptake, and antioxidant enzymes in overground tissues and chlorophyll play weaker roles in miscanthus growth. Chlorophyll might be the most critical factor in Cd uptake; soil pH might be in As uptake; OTUs affiliated to Enterobacteriaceae and antioxidant enzymes in underground tissues might be in Cu uptake; Chlorophyll and antioxidant enzymes in underground tissues might be in Pb uptake; and antioxidant enzymes in underground tissues might be in Zn uptake. Besides, Miscanthus trait indices were significantly correlated with As, Pb and Zn remediation efficiencies. There was a significant positive correlation between the BCF or TF of cadmium and its removal efficiency, TF of As and Pb showed a significant and positive correlation with removal efficiencies. However, it was not guaranteed that the higher BCF or TF enhanced the remediation efficiency for metals, e.g., Zn.
Analysis of predictive functional enzyme and genes
The relative abundances of the enzyme about ROS, such as glutathione peroxidase, superoxide dismutase, and glutathione synthase, were the most abundant in TS (Table S6). Genes function prediction results showed that the genes related to metal transport about Mn, Zn, Cd, Cu et al. (Table S7) were ABC.ZM.A, corA, corC, zntB, zipB, ycnJ, mtsB. Meanwhile, the relative abundances of the genes ABC.ZM.A, corC, and zntB were the most in MT.