Potato dextrose agar (PDA) medium (potato 200 g, glucose 15 g, agar 15 g for 1 L) was used for fungal culture and activation. Nutrient agar (NA) medium (yeast extract 5 g, peptone 10 g, sodium chloride 10 g, agar 15 g for 1 L) was used for bacteria culture and activation. MS medium (Coolaber Science & Technology Co., Ltd., Beijing, China) was used for banana tissue culture seedlings propagation. Banana seedling substrate and coconut bran were purchased from Yuxi Letu Technology Co., Yuxi, China.

Resistant cultivar Yunjiao No. 1 (AAA) and susceptible cultivar Brazilian (AAA) propagated by the Agricultural Environmental and Resources Institute, Yunnan Academy of Agricultural Sciences, were used in this study to determine the effects of QST713 on bananas (Zhou et al., 2023). The tissue culture plantlets cultured at 25°C with 16 h/8 h (light/dark) for 20 days were transplanted to the seedling substrate for cultivation. When the banana plantlets had grown with 2–3 more new leaves, they were transplanted into plastic pots (25 cm × 18.3 cm) containing transplanting soil in which the seedling substrate (Banana Planting Substrate, Yunnan Yuxi Leshi Technology Co. Ltd., Yunnan, China) was mixed with fresh soil (native horticultural red soil Ferralosols, which bananas have never been planted in previously) at a ratio of 1:1. One banana plant was planted in each pot. There were 50 banana plantlets in each treatment, and each treatment was replicated three times (Table 1). Routine fertilizer and watering were applied in greenhouse management.

Bacillus amyloliquefaciens strain QST713 was provided by Bayer’s Crop Science Department (Leverkusen, Germany). When banana plantlets were transplanted and produced 6 to 7 leaves, a 100 mL QST713 solution (5 mL QST713 original agent labeled as EZU 1631901*1.0 + 95 mL water) was provided to each plant, the roots around the banana plant being drenched. QST713 was applied every 4 weeks, three times in total.

Foc TR4 previously isolated by our laboratory in Xishuangbanna, Yunnan Province was used in this experiment (Zhang et al., 2018, 2020). TR4 spore suspension was prepared with 1 × 10⁷ CFU/mL after culturing in PDB medium at 28°C for 48 h. Seven days after the second application of QST713, the soil around each banana plant was drenched with 100 mL TR4 spore suspension (BTR4 + QST713, YTR4 + QST713). The TR4 treatments of Brazilian and Yunjiao No. 1 cultivar was inoculated with TR4 at the same time (BTR4, YTR4). The control treatments of Brazilian and Yunjiao No. 1 cultivar were inoculated with normal tap water (BCK, YCK). There were six different treatments in the experiment, each replicated three times. Each treatment was a total of 150 banana plants, 50 banana plants/treatment/replication (Table 1).

The banana pot experiment was carried out in the greenhouse of the Agricultural Environment and Resources Institute, Yunnan Academy of Agricultural Sciences. The indoor average illumination time was about 12 h, the average indoor temperature was 32°C (day)/22°C (night), and the average humidity was about 50%–60%. The banana growth traits (plant height, pseudostem girth, leaf number, leaf length, and leaf width) were investigated at 0 day, 28 days, 56 days and 97 days after the first application of QST713. The measurement of plant height was the range from the base of the pseudostem to the intersection of the newest two leaves. Pseudostem girth was measured as the pseudostem diameter 1 cm from the base using vernier calipers. The leaf number was counted as all functional leaves, and the length and width of the first expanded leaf were measured (Fan et al., 2021). After 62 days of TR4 inoculation, the fresh weights of the above and below ground parts of the banana plants were weighed.

In order to evaluate the efficacy of the QST713 agent for the control of banana wilt in pot experiments, we investigated the incidence on each plant 62 days post-inoculation with TR4 according to the lesion characteristics of bananas. The disease severity index (DSI) standards from Zhang’s et al. (2018) protocol were referred to. DSI was graded from 0 to 4 according to the necrosis degree of the banana corm (“0” represented healthy plants with no internal symptoms and “4” represented severe wilting or that the plant had died).

At 9 weeks (62 days) after inoculation with TR4, the rhizosphere soils of 15 plants of each treatment (CK, TR4, TR4 + QST713) were collected with a sterilized soil shovel. After pulling out the banana plant from the pot and shaking the roots vigorously, soil attached to the surface of the root was regarded as rhizosphere soil and was collected. Before collecting soil samples, tweezers were used to remove residual root system and other sundries from the collected mixed rhizosphere soil. Soil samples were stored at −20°C before DNA extraction. In total, 15-pot-plant soils were sampled, and three pots of soil were mixed into one sample (thus five soil sample replicates) for sequencing. Approximately 20 g rhizosphere soil was collected per plant.

Total genomic DNA was extracted from soil samples using the OMEGA Soil DNA Kit (D5625-01) (Omega Bio-Tek, Norcross, GA, United States) following the manufacturer’s instructions, and the isolated DNA was stored at −20°C. The concentration of extracted DNA was determined using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States), and the quality was checked by agarose gel electrophoresis.

PCR amplification of the bacterial 16S rRNA genes V3–V4 region was performed using the forward primer 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and the reverse primer 806R (5′-GGACTACHVGGGTWTCTAAT-3′). PCR amplification of the fungal ITS1 region was performed using the forward primer ITS5F (5′-GGAAGTAAAAGTCGTAACAAGG-3′) and the reverse primer ITS1R (5′-GCTGCGTTCTTCATCGATGC-3′) (Caporaso et al., 2011; Kozich et al., 2013). The PCR contained 5 μL of buffer (5×), 0.25 μL of Fast pfu DNA Polymerase (5 U/μL), 2 μL (2.5 mM) of dNTPs, 1 μL (10 μM) of each forward and reverse primer, 1 μL of DNA template, and 15.75 μL of ddH₂O. Thermal cycling consisted of initial denaturation at 98°C for 5 min, followed by 25 cycles consisting of denaturation at 98°C for 30 s, annealing at 53°C for 30 s, and extension at 72°C for 45 s, with a final extension of 5 min at 72°C. The amplification products were sent to Personalbio (Personal Biotechnology Co., Shanghai, China) for high-throughput sequencing on the Illumina Miseq platform. Average length of the sequences was 337.77 for 16S rRNA genes, and 269.12 for ITS1 region. QIIME2 2019.4 (Bolyen et al., 2019) was used to performed microbiome bioinformatics. We finally obtained 249,376 ASVs for bacterial communities, and 20,109 ASVs for fungal communities. The sequential data of all samples were loaded into NCBI with accession numbers PRJNA 949124 (16SrRNA) and PRJNA949129 (ITS).

Kruskal Wallis rank sum test was used to test the difference between Chao1 and Shannon indices. Principal coordinates analysis (PCoA) was used to analyze the visual difference of different community compositions. Permutational MANOVA (PERMANOVA) was used to analyze the differences between community compositions. Co-occurrence patterns were reconstructed by Hmisc package in R and Gephi 0.9.2. Gephi 0.9.2 was used to calculate the modular index. If the Spearman’s correlation coefficient was >0.70 and p < 0.05, then co-occurrence was considered robust (Gao et al., 2021). Network stability was measured by the proportion and modularity of negative or positive correlations. The correlation between growth and relative abundance of bacterial genera was calculated by R and visualized with the ggpubr package. Wilcoxon rank sum test was used to analyze the relative abundance of dominant bacteria in the bacterial community between two different treatments. Excel 2019 and IBM SPSS Statistic 25.0 software was used for data processing and statistical analysis, and univariate analysis (LSD and Duncan) in SPSS Statistic 25.0 software was used for variance analysis.

The results of our study showed that the microbial products have a significant effect on promoting plant growth. At 28 days post application (dpa) of QST713 on cultivar Brazilian, there were no significant difference among the CK, TR4, and TR4 + QST713 treatments in plant growth. At 56 dpa, the plant height (36.57 ± 0.79 cm), pseudostem girth (26.48 ± 0.51 mm), leaf length (36.07 ± 0.87 cm), and leaf width (18.00 ± 0.46 cm) of TR4 + QST713 treatments were significantly higher than CK (32.26 ± 1.03 cm, 24.36 ± 0.31 mm, 32.3 ± 1.24 cm, 16.16 ± 0.64 cm) and TR4 (31.36 ± 0.17 cm, 24.07 ± 0.19 mm, 32.69 ± 0.39 cm, 15.67 ± 0.23 cm) treatments. There was no significant difference in leaf number. At 97 dpa, the plant height (42.37 ± 1.36 cm) and pseudostem girth (32.15 ± 0.57 mm) of TR4 + QST713 treatment were significantly higher than the corresponding traits in the TR4 treatment (36.40 ± 0.93 cm, 29.37 ± 0.22 mm) and CK (37.94 ± 1.25 cm, 28.72 ± 1.32 mm), and the leaf length (37.03 ± 1.52 cm) and leaf width (16.82 ± 0.93 cm) were higher than TR4 treatment (32.38 ± 0.98 cm, 13.77 ± 0.06 cm) but not significantly different from the control (36.33 ± 1.65 cm 16.36 ± 0.83 cm) (Figures 1A, 2; Supplementary Figures S1A–S4A).

As for the resistant cultivar Yunjiao No. 1, there was the same trend as Brazilian. Twenty-eight days after the first application of QST713, there were no significant differences in plant growth of QST713 + TR4, TR4, and CK. At 56 dpa, the plant height (37.27 ± 1.22 cm), pseudostem thickness (26.66 ± 0.75 mm), leaf length (39.79 ± 0.87 cm), and leaf width (18.58 ± 0.28 cm) of QST713 + TR4 treatment were significantly higher than those of TR4 treatments (31.97 ± 0.16 cm, 24.79 ± 0.27 mm, 35.43 ± 0.89 cm, 15.86 ± 0.36 cm). At 97 dpa, the plant height (41.24 ± 2.19 cm), pseudostem thickness (33.37 ± 0.52 mm), leaf length (41.5 ± 1.76 cm), and leaf width (17.91 ± 0.67 cm) of the QST713 + TR4 treatment were significantly higher than those of TR4 (34.97 ± 0.46 cm, 29.95 ± 0.41 mm, 34.40 ± 2.47 cm, 15.25 ± 0.39 cm) and the control (35.90 ± 0.53 cm, 29.66 ± 0.51 mm, 38.4 ± 0.34 cm, 15.76 ± 0.25 cm), respectively (Figures 1B, 2; Supplementary Figures S1B–S4B).

The fresh weight of aerial biomass in QST713 + TR4 (Brazilian: 0.292 ± 0.017 kg, Yunjiao No. 1: 0.313 ± 0.026 kg) treatment was significantly higher than that of TR4 (Brazilian: 0.237 ± 0.013 kg, Yunjiao No. 1: 0.234 ± 0.006 kg) and the control treatments both in two cultivars. In terms of fresh weight of underground biomass (roots), QST713 + TR4 treatment of Yunjiao No. 1 (0.079 ± 0.001 kg) was significantly higher than that of TR4 (0.067 ± 0.005 kg) but showed no statistically significant difference from CK (0.069 ± 0.001 kg). In this respect, there were also no statistically significant differences between any treatments on cultivar Brazilian (Figure 3).

Disease investigation indicated that QST713 did not decrease the incidence rate of disease of Fusarium wilt of banana. The disease index of BTR4 + QST713 (55.20 ± 3.79) was not significant reduced compared to that of BTR4 (54.17 ± 0.52). The disease index of TTR4 + QST713 (30.73 ± 5.29) was not significant reduced compared to that of TTR4 (26.64 ± 2.61). It was noticed that the disease incidence on Yunjiao No. 1 was prominently lower than Brazilian, which reflects the differences in disease-resistant characteristics of these two cultivars (Supplementary Figures S5, S6).

In order to determine the effect of QST713 on the soil microbial diversity (Chao1 and Shannon index) of two different banana cultivars (Brazilian and Yunjiao No. 1), we compared the alpha diversity of bacterial and fungal communities in rhizosphere soils. The results showed that there were no statistically significant differences in bacterial community richness (Chao1 index) between BCK, BTR4, and BTR4 + QST713 treatments in the Brazilian cultivar (Figure 4A and Supplementary Table S1). However, there was a statistical significance in bacterial community richness from YTR4 to YTR4 + QST713 treatments in Yunjiao No. 1 (Figure 4C and Supplementary Table S1). The Chao1 index of YTR4 + QST713 was significantly lower than YTR4. In addition, the Chao1 index of YTR4 + QST713 was significantly lower than that of YCK treatment (Figure 4C and Supplementary Table S1). However, there was no statistical significance in the diversity of the bacterial communities in either cultivar.

From the perspective of alpha diversity of fungal community, there was no significant difference in fungal community richness between BCK, BTR4, and BTR4 + QST713 in Brazilian cultivar (Figure 4B and Supplementary Table S1). However, in Yunjiao No. 1, compared with YCK treatment, YTR4 + QST713 and YTR4 treatments showed significantly higher fungal community richness (Chao1 index) (Figure 4D and Supplementary Table S1).

The results showed that there were no statistically significant differences in bacterial and fungal community diversity and richness (Chao1 and Shannon index) between BCK, BTR4, and BTR4 + QST713 treatments, and there were also no statistically significant differences in bacterial and fungal community diversity (Shannon index) between YCK, YTR4, and YTR4 + QST713 treatments. However, there were significantly differences in the richness of bacterial and fungal communities (Chao1 index) between YTR4 and YTR4 + QST713 treatments. The Chao1 index of YTR4 + QST713 was significantly lower than YTR4 and YCK in the bacterial community, and the Chao1 index of YCK was dramatically lower than YTR4 + QST713 and YTR4 in the fungal community.

PCoA and PERMANOVA were used to analyze the effects of QST713 application and pathogen TR4 on the bacterial and fungal communities of banana rhizosphere microbial. From the perspective of bacterial community composition in Brazilian, compared with BCK treatment, the composition of bacterial communities treated with BTR4 was statistically significant (R² = 0.04, p < 0.05) (Figure 5A). In addition, there were significant differences between BTR4 + QST713 and BCK treatments in bacterial communities (R² = 0.06, p < 0.001) (Figure 5C). BTR4 + QST713 treatment was also significantly different from BTR4 treatment (R² = 0.06, p < 0.001) (Figure 5B). Therefore, the application of QST713 changed the composition of the bacterial community additionally to the changes brought about by TR4 alone. These changes had the same tendency in resistant cultivar Yunjiao No. 1 (Figures 5D–F). The results showed that the application of QST713 and inoculation of TR4 could significantly change the rhizosphere soil bacterial community of banana plants.

From the point of view of fungal community composition, there were no statistically significant differences between the compositions of fungal communities for the Brazilian BTR4 vs. BCK treatments (Figure 6A, p > 0.05) or BTR4 vs. BTR4 + QST713 treatments (Figure 6B, p > 0.05). Compared with BCK treatment, there was a significant difference in BTR4 + QST713 treatment in the composition of fungal communities (Figure 6C, R² = 0.08, p < 0.05). As for Yunjiao No. 1, there were significant differences between YTR4 vs. YCK (Figure 6D, R² = 0.08, p < 0.01), YTR4 vs. YTR4 + QST713 (Figure 6E, R² = 0.08, p < 0.001), and YTR4 + QST713 vs. YCK (Figure 6F, R² = 0.17, p < 0.001). It can be seen that TR4 infection and QST713 application could both change the bacterial community composition of two banana cultivars, and the fungal community composition of Yunjiao No. 1 also changed significantly.

We further used the negative correlation edges and modularity index to determine the stability of community network relationships (Figure 7; Supplementary Figure S7). The higher the ratio of negative correlation edges and modularity index, the more stable the community structure. The results showed that the ratio of modularity index and negative correlation edge of TR4 inoculation (Figures 7B,E) were significantly higher than that of CK treatment (Figures 7A,D), while the influence of TR4 + QST713 treatment lay between them (Figures 7C,F). This shows that ST713 has a significant effect on maintaining the stability of rhizosphere soil microorganism community composition in the presence of TR4.

The above diversity analysis showed that the bacterial community composition was significantly different in different treatments, but the fungal communities were significantly different between the two cultivars. The change of bacterial community could be the potential reason for the growth of the two cultivars. Therefore, Wilcoxon rank sum test of variance was used to analyze the relative abundance of dominant bacteria in the bacterial community between different treatments of the two cultivars, so as to identify the dominant bacteria with specific changes. The outcome of this study indicated that compared to the TR4 treatments, the bacteria of Pseudomonas and Bacillus were significantly up-regulated in TR4 + QST713 treatment (Figure 8), which may provide a key insight into the observed growth promotion effect on banana plants.

In order to clarify the correlation between dominant bacteria (Pseudomonas and Bacillus) and plant growth effect, the correlation analysis of the relative abundance of dominant bacteria and banana growth parameters were conducted. The outcome of this study indicated that that the relative abundance of the two genera had a significant positive correlation with plant height, pseudostem girth, and above-ground fresh weight (Figure 9). It was the same for leaf length and leaf width but had no correlation with leaf number (Supplementary Figure S8). There was a low positive correlation between Bacillus and underground fresh weight (Supplementary Figure S8). Finally, we conclude that QST713 can significantly promote the growth of banana plants by altering the microbial community of bananas, especially by increasing the relative abundance of Pseudomonas and Bacillus.