Bio-fertilizer amendment shaping the R. pseudostellariae-associated bacterial community structure
Replanting disease is a typically negative feedback of plant-soil, and causes a significant wreck in mediating crop performance. In the current study, we found that bio-fertilizer application significantly increased the yield and quality (i.e., contents of heterophyllin B, total polysaccharide, and saponin) of R. pseudostellariae compared to second- and third-year monocultures (Fig. 1). Then, we collected the microbiota that were tightly attached to the surface and endogeny of leaves and roots. High-throughput sequencing showed that the roots contained a significantly higher alpha diversity of bacteria than the leaves (Figure S1 and S2A). PCA revealed significant differences in community composition between leaf and root microbiota (Figure S2B). The bio-fertilizer treatment significantly decreased the bacterial alpha diversity in both leaves and roots compared to the third-year monoculture treatment (Fig. 2). The results were in line with a previous study suggesting that the application of fertilizer would lead to lower bacterial diversity in leaves . Meanwhile, the PCA analysis showed that the bacterial community composition was generally separated when comparing bio-fertilizer and different years of consecutive monoculture treatments in each of two compartments (Fig. 2C). This implied that the application of bio-fertilizer altered the bacterial community structure in leaves and roots under agricultural field conditions.
Continuous monoculture and bio-fertilizer altered the R. pseudostellariae-associated bacterial community composition
Continuous monoculture significantly increased the dominant abundance of Proteobacteria and Firmicutes and decreased the Actinobacterial phyla in leaves. Meanwhile, bio-fertilizer treatment significantly enhanced the abundance of Proteobacteria, Firmicutes, and the ratio of Firmicutes/Proteobacteria and decreased Actinobacteria in leaves compared to the consecutive monoculture treatments. In addition, the bio-fertilizer treatment significantly lowered the abundance of Proteobacteria and Bacteroidetes, and the bio-fertilizer treatment increased the abundance of Acidobacteria in roots under the third-year monoculture treatment (Fig. 3). Although the key community members of microbiotas were overlap in the leaves and roots, the overall bacterial community structure and composition were different. In line with previous observations, the abundance of Firmicutes was markedly reduced and Proteobacteria were enriched; this contributed to shaping the bacterial community in leaves and subsequently causing leaf disease . This result indicated that the bio-fertilizer might influence the leaf and root microbial community, possibly through plant microbe–microbe interactions.
Ralstonia, Pseudomonas, and Paenibacillus were significantly more abundant in leaves than in roots (Figure S3 and S4). Continuous monoculture significantly increased the abundance of pathogenic Ralstonia and Fusarium oxysporum in the two compartments. However, the bio-fertilizer treatment significantly decreased the abundance of Ralstonia and F. oxysporum and increased Bradyrhizobium in the leaves and roots under the third-year monoculture treatment. Previous studies have reported that Ralstonia  and F. oxysporum  were among the most important plant pathogens. Pseudomonas , Paenibacillus , Bradyrhizobium , and Streptomyces  have been recognized as microbial antagonists and biological control agents in agricultural production. Furthermore, bio-fertilizer has been shown to have a positive effect on the abundances of Pseudomonas and Streptomyces in roots and on Paenibacillus in leaves compared to consecutive monoculture treatments, suggesting that newly available habitat niches resulting from a relative decrease of pathogens can be filled by functionally different bacteria. The potential mechanisms and soil ecological processes that are responsible for these findings might be the bio-fertilizer improving the rhizosphere micro-environment by increasing the abundance of indigenous beneficial microbes . This could then alter the bacterial communities in roots and leaves. The soil microbiome was previously shown to be the major source of the leaf bacterial microbiota [10, 34–36]. Previous studies suggested that the microbes were able to be transported from the rhizosphere soil to the plant phyllosphere through xylem vessels and aerosols [9, 10]. Therefore, we assumed that the root microbiome might also serve as an important reservoir of beneficial microbes for leaves under bio-fertilizer treatment. Interestingly, the abundances of beneficial Pseudomonas and Paenibacillus were significantly increased in the leaves under continuous monoculture treatment. This might be due to the leaves harboring certain suppressive bacteria that can restrict pathogens and increase resistance .
Relationships between microbial communities and the yield and quality of R. pseudostellariae
Linear discriminant analysis (LDA) indicated that continuous monoculture tended to decrease the indicator taxa in both compartments (Figure S5 and S6). Random forest regression modelling showed that the genera belonging to Proteobacteria, Actinobacteria, and Firmicutes were incorporated as the main biological predictors for yield and quality in the two compartments (Fig. 1E and 4). The genera Pseudomonas, Ralstonia, Acidovorax, Conexibacter, and Streptomyces were the dominant predictors of yield and quality. These beneficial and deleterious biomarkers that have been shown to have a significant impact on microbial communities when comparing consecutive monoculture and bio-fertilizer treatments and should be taken into consideration for new strategies to improve plant health and agricultural productivity through suppressing the activity of pathogens and promoting beneficial microbes. Our results showed that the majority of the co-occurrence network connections were linked to phyla of Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes, and Acidobacteria (Fig. 5). In addition, the modularity values of networks decreased as the number of consecutive monoculture years increased (Table S2). Previous studies have suggested modules as niches [37, 38], and the lower modularity values may therefore be linked to stronger niche overlap and interspecific competition of leaf and root microbiota under consecutive monoculture regimes. Moreover, the positive correlations of microbiota under bio-fertilizer treatment were highest among all of the treatments (Table S2), suggesting that bio-fertilizer treatment enhanced the ecological commensalism or mutualism of microorganisms.
Structural equation models (SEMs) indicated that consecutive monoculture treatment had a negative and indirect effect on the yield and quality by influencing leaf and root microbiota (Fig. 6A). The bacterial diversity and richness of roots had a significant negative effect on the bacterial diversity and richness of leaves under consecutive monoculture regimes, while the opposite pattern was observed under bio-fertilizer application. This may be due to replanting disease decreasing cell density in leaves and roots and causing malformed organs of R. pseudostellariae . Furthermore, the addition of bio-fertilizer had a more direct influence on leaf microbiota than on root microbiota, indicating that leaf microbiota was more sensitive to bio-fertilizer than microbiota of the roots. This might be due to the root endophytes being more influenced by host genetic control  or phyllosphere microbes being more vulnerable to anthropogenic disturbance than soil microbes . Spearman correlation analysis indicated that yield and quality were negatively correlated with the abundance of potentially pathogenic Ralstonia in the two compartments and positively correlated with beneficial Pseudomonas, Streptomyces, and Bradyrhizobium in roots (Fig. 6B). Moreover, the abundance of Ralstonia was significantly negatively correlated to beneficial bacteria. The beneficial microbiota also negatively affected the abundance of pathogenic F. oxysporum (Fig. 7). This indicated that plant-microbe–microbe interactions in turn not only impacted microbial abundance, but it also impacted plant disease by antagonizing plant pathogens.