Cooperative Action of Fulvic Acid and Bacillus paralicheniformis Ferment in Regulating Soil Microbiota and Improving Soil Fertility and Plant Resistance to Bacterial Wilt Disease

Continuous cropping tobacco has led to soil degradation and caused soilborne bacterial wilt disease. Fulvic acid as a biostimulator was applied to restore soil and control bacterial wilt disease. ABSTRACT Excessive continuous cropping and soil degradation, such as acidification, hardening, fertility decline, and the degradation of microbial community, lead to the epidemic of soilborne diseases and cause great loss in agriculture production. Application of fulvic acid can improve the growth and yield of various crops and effectively suppress soilborne plant diseases. Bacillus paralicheniformis strain 285-3 producing poly-gamma-glutamic acid is used to remove the organic acid that can cause soil acidification and increase the fertilizer effect of fulvic acid and the effect of improving soil quality and inhibiting soilborne disease. In field experiments, the application of fulvic acid and Bacillus paralicheniformis ferment effectively reduced the incidence of bacterial wilt disease and improved soil fertility. Both fulvic acid powder and B. paralicheniformis ferment improved soil microbial diversity and increased the complexity and stability of the microbial network. For B. paralicheniformis ferment, the molecular weight of poly-gamma-glutamic acid became smaller after heating, which could better improve the soil microbial community and network structure. In fulvic acid and B. paralicheniformis ferment-treated soils, the synergistic interaction between microorganisms increased and the number of keystone microorganisms increased, which included antagonistic bacteria and plant growth-promoting bacteria. Changes in the microbial community and network structure were the main reason for the reduced incidence of bacterial wilt disease. Application of fulvic acid and Bacillus paralicheniformis ferment improved soil physicochemical properties and effectively controlled bacterial wilt disease by changing microbial community and network structure and enriching antagonistic and beneficial bacteria. IMPORTANCE Continuous cropping tobacco has led to soil degradation and caused soilborne bacterial wilt disease. Fulvic acid as a biostimulator was applied to restore soil and control bacterial wilt disease. For improving its effect, fulvic acid was fermented with Bacillus paralicheniformis strain 285-3 producing poly-gamma-glutamic acid. Fulvic acid and B. paralicheniformis ferment inhibited bacterial wilt disease, improved soil quality, enriched beneficial bacteria, and increased microbial diversity and microbial network complexity. Some keystone microorganisms in fulvic acid and B. paralicheniformis ferment-treated soils had potential antimicrobial activity and plant growth-promoting attributes. Fulvic acid and B. paralicheniformis 285-3 ferment could be used to restore soil quality and microbiota and control bacterial wilt disease. This study found new biomaterial to control soilborne bacterial disease by combining fulvic acid and poly-gamma-glutamic acid application.

T he continuous cultivation of the same crops in the same field is called continuous cropping. With the development of the scale and intensification of crops, restricted by the cultivated land area and cultivation conditions, continuous cropping has become the main planting mode of crops in most of the main crop-producing areas in China (1). Continuous cropping leads to the reduced consumption and uneven enrichment of soil nutrients, resulting in an imbalance and deterioration of soil nutrients, and then affects the absorption of soil nutrients by plants (2,3). Additionally, after serious soil degradation, it is easy to cause soilborne diseases, such as bacterial wilt disease and root rot disease, resulting in a large area of plant death and causing great economic losses (4). Bacterial wilt disease is caused by Ralstonia solanacearum, which infects more than 250 plant species (5).
Fulvic acid is a kind of yellowish-brown powder substance and is soluble in acid and alkali, containing hydroxyl group, amino group, and other groups, widely used in agriculture, medicine, and other fields (6). Fulvic acid has high quality and is the best active and resistant component of humic acid. Fulvic acid is rich in nitrogen and potassium and has the function of solubilizing phosphorus and improving fertilizer utilization rate. Fulvic acid has been widely utilized to amend soils with low fertility. Fulvic acid can also accelerate the formation of soil aggregates and improve soil physical and chemical properties, such as permeability, airing, and pH (7). In recent years, as a new biostimulator and humic acid soil modification agent, fulvic acid is widely used in facility agriculture, such as greenhouses, as well as in cash crops such as fruit trees; facilitates the assimilation of plant nutrients; and has achieved good application results (8). Fulvic acid induces disease resistance to Botrytis cinerea through the activation of the phenylpropanoid pathway (9). Although fulvic acid has been used in more and more crops, its use for biocontrol of bacterial wilt disease is still rarely reported. We had previously conducted preliminary studies on the application effectiveness evaluation of fulvic acid in tobacco cultivation and found that fulvic acid could effectively inhibit the occurrence of tobacco soilborne diseases, such as root rot disease. This study was based on previous research to further develop the application technology of fulvic acid in crop planting, establish a solid foundation for improving soil quality, and prevent the continuous cropping obstacle and soilborne plant diseases.
Bacillus paralicheniformis strain 285-3 was isolated from soil sample collected from Shanxi province, China. B. paralicheniformis strain 285-3 produced poly-gamma-glutamic acid (g -PGA) with an average molecular weight of 580 kDa. The g -PGA is an environmentally friendly biomaterial that has a broad application for food, medicine, agriculture, environmental protection, and other fields (10). In this study, fulvic acid powder (FA) extracted from yeast molasses fermentation was added in culture medium of 285-3 to remove the organic acid that could cause soil acidification and increase the fertilizer effect of fulvic acid and the effect of improving soil quality and inhibiting soilborne bacterial wilt disease.

RESULTS
Antagonism of fulvic acid and B. paralicheniformis ferment on bacterial wilt disease. Fulvic acid powder (FA) treatment significantly (P , 0.0001) reduced the disease incidence and disease severity index of tobacco bacterial wilt (Fig. 1). Compared with the control group (CK), the disease incidence and disease severity index in FA treatment was reduced by 95.63% and 99.05%, respectively. Disease severity index in the unheated B. paralicheniformis ferment (BP) and heated B. paralicheniformis ferment (IBP) treatments also significantly decreased compared with that of the control. Application of fulvic acid or heat-treated or untreated fermentation broth of B. paralicheniformis 285-3 all effectively controlled bacterial wilt disease. Fulvic acid powders exhibited the best control effect on bacterial wilt disease among all treatments. The application of fulvic acid and B. paralicheniformis ferment had a promoting effect on tobacco plants' growth, among which BP and FA treatments had an obvious promoting effect on plant height. Improvement in soil physicochemical properties and plant growth by fulvic acid and B. paralicheniformis ferment. The organic matter and alkali-hydrolyzable nitrogen contents of the soils treated with IBP, BP, and FA were significantly greater than those of the control CK at 18 days and 36 days posttransplantation (Fig. 2), indicating that the application of fulvic acid and B. paralicheniformis ferment increased the fertility and nitrogen supply levels of the soil and that the application of B. paralicheniformis ferment had better effect. The available phosphorus and potassium contents of FA treatment were significantly (P , 0.01) greater than control CK at 18 days and 36 days posttransplantation. The available potassium content of IBP and BP treatments was significantly greater than control CK at 36 days posttransplantation. Application of B. paralicheniformis 285-3 ferment and fulvic acid powder both improved soil physicochemical properties.
Microbial community changed after application of fulvic acid and B. paralicheniformis ferment. At 18 days posttransplantation, the number of bacterial operational taxonomic units (OTUs) unique to IBP, BP, FA, and CK treatments was 408, 344, 375, and 315, respectively, and the common number of OTUs was 5,096 (see Fig. S1 in the supplemental material). The number of bacterial species observed in IBP, BP, and FA treatments was greater than that in control CK. Application of fulvic acid and B. paralicheniformis ferment increased bacterial species at 18 days posttransplantation.
At 36 days posttransplantation, the number of species (i.e., OTU number) observed in the IBP, BP, FA, and CK treatments was 2,621, 2,433, 2,453, and 1,976, respectively. Compared with the 18 days posttransplantation results, the common OTU number of four treatments was significantly reduced; the unique OTU number of IBP, BP, and FA treatments all increased; and the unique OTU number of CK decreased (see Fig. S2 in the supplemental material). The number of bacterial species observed in IBP, BP, and FA treatments was higher than that in control CK. The IBP group contained the greatest number of OTUs.
The alpha diversity can reflect the microbial community's diversity and richness. There was no significant difference between the alpha diversity of IBP, BP, and FA treatments and the control CK at 18 days posttransplantation (see Table S1 in the supplemental material). The alpha diversity of IBP, BP, and FA treatments was different from the control at 36 days posttransplantation ( Table 1). The higher the community diversity and the more

Fulvic Acid Application Changes Microbiota
Microbiology Spectrum uniform the species distribution, the greater the Shannon index. The Shannon indices of IBP and FA treatments were significantly (P , 0.05) greater than that of the control CK. The overall Shannon index at 18 days posttransplantation was greater than that at 36 days posttransplantation. Simpson indices characterize the diversity and evenness of species distribution within the community, and species were evenly distributed within each treatment community. The Chao1 index was used to estimate the total number of species included in the community, and the total number of species in IBP, BP, and FA treatments were significantly (P , 0.05) greater than in the control CK. The total number of species of all treatments at 18 days posttransplantation was greater than those at 36 days posttransplantation. The ACE index was used to estimate the OTU numbers in the community. ACE indices of IBP, BP, and FA treatments were significantly (P , 0.05) greater than those of the control CK at 36 days posttransplantation. The ACE index of all treatments at 36 days decreased compared to that at 18 days posttransplantation. Application of fulvic acid and B. paralicheniformis ferment increased microbial diversity at 36 days posttransplantation, and heat-treated B. paralicheniformis ferment had the better effect.
The difference in the microbial community composition of the four treated soils was analyzed by principal-coordinate analysis (PCoA), and the distance between the treatment samples represented the degree of difference. At 18 days posttransplantation, the samples of IBP treatment were clustered, concentrated, and separated from CK samples. At 36 days posttransplantation, IBP, BP, and FA communities were all quite different from control CK, indicating that both fulvic acid and B. paralicheniformis ferment application changed soil bacterial community structures (Fig. 3). Based on multi-response permutation procedures (MRPP) analysis, microbial communities of IBP, BP, and FA treatments were all significantly different from the control CK both at 18 days and 36 days posttransplantation ( Table 2). Analysis of similarity (ANOSIM), multivariate analysis of variance (Adonis), and analysis of molecular variance (AMOVA) all showed that BP, IBP, and FA communities were significantly (P , 0.05) different from CK at 36 days posttransplantation. Community difference between fulvic acid and B. paralicheniformis ferment-treated soils and control soil at 36 days was greater than that at 18 days posttransplantation.
Soil microbial network changed after application of fulvic acid and Bacillus paralicheniformis ferment at 36 days posttransplantation. At 36 days posttransplantation when compared to those of the control, the microbial networks of FA, BP, and IBP treatments had a greater number of nodes, closer microbial connection, and higher complexity; that is, each treatment (FA, BP, IBP) increased the size and complexity of soil microbial network (Fig. 5). The close similar threshold (S t ) value was obtained for the FA (0.94), BP (0.94), IBP (0.93), and CK (0.95) networks. Applying these thresholds, four networks were constructed for the CK-, FA-, BP-, and IBP-treated soils, respectively (Fig. 5). The CK, FA, BP, and IBP networks consisted of 339, 885, 810, and 1,265 nodes (OTUs) and 307, 913, 1,029, and 1,875 edges (connections), respectively, indicating that the FA, BP, and IBP networks had a larger size with more nodes and links than the control CK network. IBP network had the largest size among the four networks. After application of fulvic acid and B. paralicheniformis ferment, microbial network had far more complicated interactions in terms of network size, connectivity, and clustering coefficient when compared with the control CK network (Table 3). In the FA, BP, and IBP networks, 66.8% (610/912), 53.7% (553/1,028), and 57.5% (1,078/1,874) of the nodes had positive interactions with others, indicating that the microbial community under fulvic acid or B. paralicheniformis ferment-treated soils formed a close organization via synergistic  interactions among different species. In the control CK network, 70.9% (217/306) of the nodes had negative interactions, suggesting that there were more competition and antagonism among different bacterial species in the untreated soil. Connectivity represents the number of links between a node and other nodes. Higher average connectivity (avgK) means a more complex network. The avgK of FA (2.06), BP (2.54), and IBP (2.96) networks was higher than that of the control CK network (1.81), suggesting that the FA, BP, and IBP networks were more complex than the CK network. All curves of network connectivity distribution were fitted well with the power-law model (R 2 values varied from 0.89 to 0.98), indicative of scale-free networks. Geodesic distance represents the shortest path between two nodes. The average network distance (avgGD) between all pairs of nodes was 3.85, 9.59, 5.44, and 9.62 edges for the CK, FA, BP, and IBP networks, respectively, suggesting that OTUs in the FA, BP, and IBP networks were more highly connected with each other when compared with the control CK network. Clustering coefficient describes how well a node is connected with its neighbors. Average clustering coefficient (avgCC) is used to measure the extent of module structure present in a network. The avgCC of BP (0.12) and IBP (0.12) was higher than that of CK (0.08), indicating that nodes in the BP and IBP network tended to better cluster together when compared with those in the CK network. Modularity is used to demonstrate a network that can be naturally divided into communities or modules. Each module in the microbial network is considered as a functional unit that consists of several elementary bacteria and performs an identifiable task (11). Modularity (M) values of the FA, BP, IBP, and CK networks were all higher than the threshold value for modular structure (0.4), suggesting that these four networks were all modular (12). The geodesic distance, clustering coefficient, and modularity of these four phylogenetic a An A value greater than 0 indicates that differences between groups are greater than within-group differences. A significance value less than 0.05 indicates a significant difference. Fulvic Acid Application Changes Microbiota Microbiology Spectrum molecular ecological networks (pMENs) were significantly different from those of the corresponding random networks with the same network size and number of links, indicating that these microbial networks showed typical small-world characteristics. The geodesic distance, clustering coefficient, and modularity of these pMENs were significantly different between control and the three treatments (i.e., FA, BP, IBP). These results indicated that the network composition and structure of fulvic acid and B. paralicheniformis ferment-treated soils differed from the untreated soil. Application of fulvic acid and B. paralicheniformis ferment shifted microbial network structure. The IBP network had higher connectivity and geodesic distance than the FA and BP networks, indicating that application of heat-treated B. paralicheniformis ferment increased complexity of network compared to fulvic acid powder and B. paralicheniformis ferment application. The number of unique nodes in IBP, BP, and FA treatments was greater than control CK. The IBP network had the most unique nodes followed by the FA network (see Fig. S3 in the supplemental material).
Comparing the keystone microorganisms of four networks. Topological roles of nodes were shown in Fig. 6. According to the threshold values of within-module connectivity (Zi) (2.5) and among-module connectivity (Pi) (0.62), all nodes were divided into peripherals (specialists), connectors (generalists), and module hubs (generalists), but no nodes fell into network hubs (supergeneralists) (13). Most of the nodes (99.1%, 98.2%, 98.8%, and 97.4% for the CK, FA, BP, and IBP networks, respectively) were peripherals. Totals of 93.8%,  92.5%, 91.5%, and 89.4% of peripherals in the CK, FA, BP, and IBP networks had no links to other nodes outside of their own modules (Pi = 0). Totals of 3, 16, 10, and 33 keystone microorganisms were identified in the CK, FA, BP, and IBP networks, respectively ( Table 4). Application of fulvic acid and B. paralicheniformis ferment increased the number of keystone microorganisms.
In the CK network, 3 generalists (3 module hubs) belonged to Acidobacteriota, Actinobacteriota, and Bdellovibrionota, respectively. OTU288 and OTU569 were respectively closely related to Oligoflexus and Atopobium. OTU74 belonging to Acidobacteriota was positively correlated to disease severity index (r = 0.95; P = 0.05). In the FA network, 16 generalists (16 module hubs) belonged to Acidobacteriota, Actinobacteriota, Bacteroidota, Cyanobacteria, Firmicutes, Gemmatimonadetes, and Proteobacteria, respectively. OTU446 and OTU763 were respectively closely related to Streptomyces and Bacillus. OTU763 was negatively correlated to disease severity index (r = 20.99; P = 0.02). OTU43 was closely related to rhizobium Mesorhizobium. OTU480 was closely related to Bryobacter. OTU1458 was closely related to "Candidatus Solibacter." OTU3541 was closely related to denitrifying bacterium Thermomonas. OTU7318 was closely related to Sphingomonas and negatively correlated to disease severity index (r = 20.95; P = 0.049). In the BP network, 10 generalists belonged to Acidobacteriota, Actinobacteriota, Chloroflexi, Crenarchaeota, Cyanobacteria, Firmicutes, Gemmatimonadetes, and Proteobacteria, respectively. OTU185 was closely related to phototrophic Gemmatimonas. OTU1512 was closely related to "Candidatus Koribacter." OTU446 was closely related to Streptomyces. OTU1287 was closely related to Bacillus. In the IBP network, 33 generalists (30 module hubs, 3 connectors) belonged to Acidobacteriota, Actinobacteriota, Bacteroidota, Chloroflexi, Firmicutes, Gemmatimonadetes, Kapabacteria, Methylomirabilota, Proteobacteria, and Verrucomicrobiota, respectively. OTU35 was closely related to Devosia. OTU260 was closely related to Dongia and negatively correlated to disease severity index (r = 20.98; P = 0.03). OTU352 was closely related to plant-beneficial Ramlibacter. OTU490 was closely related to beneficial bacteria Rhodanobacter. OTU494 was closely related to nitrosifying bacteria Ellin6067. OTU881 was closely related to Paenarthrobacter. OTU1428 was closely related to rhizobium Mesorhizobium. OTU3275 was closely related to beneficial bacteria Rhodanobacter. OTU4557 and OTU4754 were closely related to Sphingomonas. OTU5164 was closely related to Microbacterium. OTU5196 was closely related to plant growth promoting rhizobacteria  Bryobacter. OTU5443 and OTU7120 were closely related to Chujaibacter. OTU7262 was closely related to Catellatospora. It could be seen that keystone microorganisms were definitely different between the four groups. Application of heat-treated B. paralicheniformis ferment enhanced the microbial network complexity and increased the number of keystone microorganisms. In FA, BP, and IBP networks, keystone microorganisms included some plant-beneficial taxa like Bacillus, Ramlibacter, Sphingomonas, Catellatospora, and Mesorhizobium, which might improve soil health, inhibit bacterial wilt disease, and indirectly benefit the plant growth.

DISCUSSION
Application of fulvic acid and B. paralicheniformis ferment improved soil fertility and inhibited bacterial wilt disease. Fulvic acid is an organic aromatic substance with high activity and is a biostimulator. In the present study, it was first found that fulvic acid powder and the fulvic acid fermented by B. paralicheniformis could inhibit bacterial wilt disease (Fig. 1). Biocontrol efficacy of fulvic acid (99.1%) was better than the chitosan and chitosan-derived nanoparticles reported previously (62%) (14) and biological control agents (49.3%) (15). In our previous study, fulvic acid did not exhibit antibacterial activity in vitro and could not directly inhibit the growth of pathogen R. solanacearum causing the widespread disease known as bacterial wilt. Fulvic acid and g -PGA produced by B. paralicheniformis might induce tobacco plant defense responses that indirectly inhibited bacterial wilt disease. It has been reported previously that fulvic acid induces resistance to Botrytis cinerea through the activation of the phenylpropanoid pathway (9). Xu et al. (16) found that application of g -PGA increased the fresh weight, chlorophyll content, proline content, and antioxidant enzyme activity of rape seedlings.
Application of fulvic acid and B. paralicheniformis ferment increased soil organic matter (Fig. 2). Zhang et al. (17) found that exogenous fulvic acid (900 kg ha 21 ) applications, which derived from the biological fermentation of maize straw, increased soil organic carbon by 26.93%. FA (fulvic acid, 150 kg ha 21 ), which was derived from the yeast molasses fermentation, increased soil organic carbon by 62.05%. Different sources of fulvic acid have different effects on improving soil organic matter. Fulvic acid extracted from the molasses fermented by yeast had a better effect on improving soil carbon content than that extracted from maize straw ferment. The BP and IBP treatments increased soil organic carbon by 64.96% and 73.38%, respectively. Fulvic acid and g -PGA produced by B. paralicheniformis showed additive effect on improving soil carbon and nitrogen nutrients. Fulvic acid application also increased soil N, P, and K contents. At 36 days posttransplantation, application of fulvic acid powder increased available phosphorus in soil by 185.69% compared with that of soil untreated with fulvic acid, which had better effect than the fulvic acid extracted from paper mill effluent used in a previous study (47.5 to 57.5% increase of phosphorus) (18). At 36 days posttransplantation, IBP, BP, and FA treatments increased alkali-hydrolyzable nitrogen in soil by 101.25%, 127.07%, and 90.43%, respectively. Few studies reported the effect of fulvic acid on soil alkali-hydrolyzable nitrogen content. At 36 days posttransplantation, IBP, BP, and FA treatments increased available potassium in soil by 53.19%, 43.27%, and 51.49% compared with that of the untreated soil, which showed better effect than potassium fulvic acid used in previous study (4.23% increase in soil available potassium) (19). Fulvic acid could provide potassium necessary for the growth of crops. However, Wang et al. (18) found that application of fulvic acid extracted from paper mill effluent had no remarkable effects on soil pH, inorganic nitrogen, available potassium, and organic matter. The effects of fulvic acid from different sources on soil nutrition varied, and the fulvic acid extracted from yeast fermentation broth had a better effect of improving soil fertility. Effect of fulvic acid and B. paralicheniformis ferment on soil properties at 18 days differed from that at 36 days posttransplantation, and it was supposed that fulvic acid and g -PGA in soil was absorbed by plants as the plant grew.
Application of fulvic acid and B. paralicheniformis ferment changed soil bacterial community diversity, structure, and composition. The effect of fulvic acid and B. paralicheniformis ferment on microbiota became more obvious at 36 days than at 18 days posttransplantation. So, the following analysis was focused on microbiota of 36 days. Application of fulvic acid and B. paralicheniformis ferment increased bacterial community Fulvic Acid Application Changes Microbiota Microbiology Spectrum diversity and numbers in soil (Table 1). Fulvic acid extracted from yeast molasses fermentation was rich in various metabolites, nitrogen, and carbon sources (i.e., amino acids, proteins, organic acids) that might provide nutrients for bacterial growth in rhizosphere soil, thus increasing bacterial diversity and quantity. For the BP and IBP treatments, fulvic acid was fermented with B. paralicheniformis 285-3. The g -PGA produced by B. paralicheniformis 285-3 provided a nitrogen source for soil bacteria growth and possibly enhanced bacterial stress resistance. Microbiota in IBP treatment had greater Shannon and ACE indices than in BP and FA treatments. Heat-treated g -PGA had a smaller molecular weight than unheated g -PGA, and thus was more easily absorbed and utilized by microorganisms, and was easier to improve plant immune resistance (Fig. 1) based on a previous study (20). Accordingly microbial community structures of IBP, BP, and FA treatments differed from the control CK (Fig. 3). Application of fulvic acid and B. paralicheniformis ferment increased the abundances of plant beneficial bacteria in soil, which had the potential to promote plant growth. Relative abundance of Dyella in BP and FA treatments was greater than in control CK (Fig. 4). Dyella has shown various plant growth-promoting activities and the potential as a biological control agent (21). Relative abundance of Luteibacter in BP and FA treatments were greater than in control CK. Guglielmetti et al. (22) found that Luteibacter had the ability to produce molecules able to chelate ferric ions and solubilize monocalcium phosphate, produce indole acetic acid, and promote root development. Relative abundance of Arthrobacter in IBP was greater than in control CK. Krishnan et al. (23) found that Arthrobacter possessed plant growth beneficial traits such as positive growth on 1-aminocyclopropane-1-carboxylic acid and production of indole acetic acid and siderophore. The relative abundance of Caulobacter in BP treatment was greater than in control CK. Berrios (24) found that several Caulobacter strains harbored the potential to enhance plant biomass. The relative abundances of Bacillus aryabhattai, Bacillus sporothermodurans, and Gemmatimonas in the FA treatment were greater than in control CK. Deng et al. (25) reported that Bacillus aryabhattai had phosphate-solubilizing and nitrogen-fixing functions that promote plant growth. Osaki et al. (26) found that B. sporothermodurans produced volatile compounds that showed inhibitory activity against pathogenic fungi. Mujaki c et al. (27) reported that Gemmatimonas had the capacity for anoxygenic photosynthesis. Plant beneficial bacteria enriched in fulvic acid and B. paralicheniformis ferment-treated soil might improve soil fertility and inhibit bacterial wilt disease.
Application of fulvic acid and B. paralicheniformis ferment increased soil bacterial network complexity and stability. The microbial networks of fulvic acid and B. paralicheniformis ferment-treated soils had more complicated interactions than the control CK network at 36 days posttransplantation (Table 3). Qi et al. (28) reported that microbial network of healthy soil was more complex than susceptible soil network, which was conducive to enhancing soil health and disease suppression. The increase in bacterial network complexity after application of fulvic acid and B. paralicheniformis ferment possibly enhanced soil health and inhibited pathogen R. solanacearum infection. The IBP network was the most complex and had the largest size among all treatments. It was supposed that fulvic acid and g -PGA in fermentation broth had a synergistic action to enhance the stability of the microbial network.
Compared with untreated soil, more synergistic interactions among bacterial species were found in fulvic acid and B. paralicheniformis ferment-treated soils (Fig. 5). The synergy between microorganisms can promote microbial growth, complete more complex metabolic processes, produce new metabolites, and occupy more ecological niches. For example, Corrêa et al. (29) observed the synergy between the enzymes produced by Trichoderma reesei RUT-C30 and Penicillium oxalicum to break down the cellulose fraction of sugarcane straw. It was speculated that the microbiota of soil treated with fulvic acid and B. paralicheniformis ferment was more stable and could resist the infection of foreign phytopathogens compared to untreated soil. Bacterial species in the untreated soil had more competition and antagonism. The antagonism and competition among microorganisms mean that different microorganisms compete for space and nutrients, which inhibit the growth of some microorganisms, decrease the microbial species, and reduce microbial diversity. The microbial community of untreated soil was loose and unstable and susceptible to phytopathogen. Based on the theory of Cray et al. (30), pathogen R. solanacearum might act as the microbial weed species and struggle to dominate in the control soil. Why could the alien species R. solanacearum invade native communities and become widespread in untreated soil? It was speculated that low microbial community diversity and unstable network structure of untreated soil favored the pathogen infection; while an increase of microbial diversity and a more stable network of fulvic acid and B. paralicheniformis ferment treated soil was conducive to resist pathogen infection.
Keystone microorganisms changed after application of fulvic acid and B. paralicheniformis ferment. The keystone microorganisms of FA, BP, and IBP networks differed from those of the control network (Table 4). Some keystone microorganisms in fulvic acid and B. paralicheniformis ferment-treated soils were potential plant beneficial bacteria, which might promote the growth of plants, produced antimicrobial substances, and participated in the soil nutrient cycle. It could be speculated from previous reports that these beneficial keystone microorganisms had important roles in improving soil quality and disease suppression. For instance, Streptomyces and Bacillus were the keystone microorganisms of FA and BP networks. Bacillus was the keystone microorganism of IBP network. Bacillus and Streptomyces have been popularly used to control plant diseases, such as bacterial wilt disease (31,32). Keystone microorganism Bacillus (OTU763) was negatively correlated to disease severity index, indicating Bacillus played a key role in inhibiting R. solanacearum. Ward et al. (33) found that "Candidatus Koribacter" (related to OTU1512) could use complex substrates (e.g., chitin, hemicelluloses, cellulose, pectin, starch, and xylan) and produced antimicrobial compounds. Huo et al. (34) reported that Rhodanobacter (related to OTU490) was plant growth-promoting rhizobacteria and showed antagonistic activity against the root rot pathogen. Some of the species from the Sphingomonas (related to OTU4557 and OTU4754) have been found to improve plant growth during stress conditions and possess antimicrobial activity (35,36). Microbacterium (related to OTU5164) is characterized by high antimicrobial activity and produces the potent antibacterial lanthipeptide microvionin (37,38). It was speculated that antibacterial substances produced by keystone microorganisms could inhibit the growth of pathogens in the soil and occupy more ecological niches. Rawat et al. (39) reported that "Candidatus Solibacter" (related to OTU1458) decomposed organic carbon and carbohydrate in the soil, thus improving soil quality. The Mesorhizobium (related to OTU43), Bryobacter (related to OTU480), Devosia (related to OTU35), Paenarthrobacter (related to OTU881), and Catellatospora (related to OTU7262) have shown plant growth-promoting activities (i.e., phosphorous, potassium solubilization, nitrogen-fixing activity, and production of siderophore, indole acetic acid, and ansamacrolactams) (40)(41)(42)(43)(44)(45). Beneficial keystone microorganisms possibly promoted plant growth, inhibited bacterial wilt disease involved in the biogeochemical C and N cycles of soil, and had a beneficial ecological function of enhancing soil fertility. The application of fulvic acid and B. paralicheniformis ferment shifted the bacterial community and network; keystone microorganisms included biocontrol bacteria (i.e., Bacillus, Rhodanobacter, Streptomyces), plant growth-promoting rhizobacteria (i.e., Bryobacter, Catellatospora, Devosia, Paenarthrobacter), and soil nutrient cycling-affecting bacteria (i.e., Bryobacter, Candidatus Solibacter, Mesorhizobium, Thermomonas), which were valuable for the promotion of plant growth and health. The number of keystone microorganisms in the untreated soil was small, and no beneficial bacteria were found. The keystone microorganism OTU74 belonging to Acidobacteriota was positively correlated to disease severity index, and acidophilic Acidobacteria might easily live and grow in the acidic soil. Zhang et al. (46) reported that bacterial wilt was accompanied by soil acidification, and the abundance of R. solanacearum increased when the pH of soil reduced. It was speculated that Acidobacteria might act as a helper of pathogen R. solanacearum. This speculation will be studied in the future.

MATERIALS AND METHODS
Plot experiment. The plot experiment was conducted in tobacco fields located in Enshi State, Xuanen County (29°959N, 109°389E), with a subtropical humid climate, a mean annual temperature of 16°C, and mean annual precipitation of 1,400 mm. Soil type was yellow-brown soil (classified as Alfisols). Tobacco (Nicotiana tabacum L.) variety Yunyan 87 was planted in this area for more than 10 years. Each plant was separated by 0.55 m. Four leaf-stage seedlings were transplanted to the field in May 2021.
The trial used a randomized block design with a total of four treatments including control (CK), fulvic acid  (20). The fermentation broth was diluted five times with water, and a 200-mL dilution was applied evenly to the soil around the tobacco roots. For IBP, after 80°C and 30 min of heat-treatment, the fermentation broth was applied in the same manner as above. Two instances of fertilization were applied at 7 days and 25 days posttransplantation for tobacco. The occurrence of tobacco bacterial wilt disease in field was recorded at 60 days posttransplantation, and 30 tobacco plants were selected for each replicate of each treatment (47). Soil sample collection. Rhizosphere soil samples were collected respectively at 18 days and 36 days posttransplantation. Eight soil samples were taken separately for each treatment. After removing plant roots and other debris, the soil samples were passed through a 2-mm sieve, then placed in sterile tubes, and quickly put in dry ice and brought back to the laboratory for high-throughput sequencing of soil microorganisms. The soil alkali-hydrolyzable nitrogen (AN), available potassium (AK), available phosphorus (AP), and organic matter (SOM) contents were tested according to previous study (48).
Rhizosphere microbial community analysis. DNA was extracted from soil using a FastDNA spin kit (MP Biomedicals, USA) according to the manufacturer's protocol. DNA concentration was determined by a NanoDrop spectrophotometer (Thermo Fisher Scientific, USA) and diluted to 1 ng mL 21 . Using diluted genomic DNA as the template, specific primers with barcode, Phusion high-fidelity PCR master mix buffer (New England BioLabs), and high-efficiency high-fidelity enzyme were selected for PCR to ensure amplification efficiency and accuracy. The V4 variable region of bacterial 16S rRNA was PCR amplified with primers of 515F (59-GTGCCAGCMGCCGCGGTAA-39) and 806R (59-GGACTACHVGGGTWTCTAAT-39). The PCR products were purified using magnetic bead. Library construction was performed using the TruSeq DNA PCR-free sample preparation kit. Library was sequenced on the NovaSeq6000 platform.
After the barcode and primer sequences were removed, the reads of each sample were assembled using FLASH. Effective tags were obtained by filtering raw tags. The effective tags were clustered using the UPARSE algorithm and clustering the sequences into OTUs (operational taxonomic units) with 97% identity. Species annotation was performed by Mothur method and SILVA138 database to obtain taxonomic information. Fast multiple sequence alignment was performed using MUSCLE software to obtain phylogenetic relationships of all OTU representative sequences. The data of each sample were normalized, and the sample with the least amount of data was used as the standard. The observed OTUs, Chao1, Shannon, Simpson, and Ace indexes were calculated using QIIME software. The PCoA was performed using R software. The Metastats analysis was performed using the R software. The ANOSIM, MRPP, and Adonis analyses were performed using the R vegan package. AMOVA was performed using the Mothur software.
Microbial network construction. The microbial network was constructed using Molecular Ecological Network Analysis Pipeline (MENAP) (28). Using the tobacco disease severity index as the environmental factor, the microbial network was constructed using MENAP and Cytoscape software. First, the OTU abundance was standardized to obtain the standard relative abundance (SRA). The SRA matrix, disease severity index, and OTU annotation information were submitted to MENAP to construct the microbial network. The modules were determined using greedy modularity optimization. The network diagram was formed using Cytoscape 3.9.0 (49). The microbial network was composed of different OTUs (nodes), and the positive or negative interactions between the OTUs were represented by connections (edges) (48). The topology properties and structure of networks were described by modularity, average connectivity (avgK), geodesic distance (avgGD), and clustering coefficient (avgCC). Connectivity represents the number of links between a node with other nodes. Geodesic distance represents the shortest path of two nodes. Clustering coefficient describes how well a node is connected with the neighbor nodes (11).
Topological roles of nodes were defined by within-module connectivity (Zi) and connectivity among modules (Pi) (13). Zi represented how well a node connected to other nodes within its own module. Pi represented how well a node connected to other modules. The nodes were divided into four categories according to Zi and Pi values. Peripheral nodes (specialists) had low Zi and Pi values that had few links and almost always connected to the nodes within their own modules. Connectors (generalists) had low Zi but high Pi values that had many links to the nodes within other modules. Module hubs (generalists) had high Zi but low Pi values that had great links to many nodes in their own modules. Network hubs (supergeneralists) had high Zi and Pi values that had great links to the nodes both in their own modules , and network hubs (supergeneralists) were the keystone microorganisms. Statistical analysis. One-way analysis of variance was performed using SPSS software to test all parameters at the 0.05, 0.01, 0.001, and 0.0001 significance levels. The differences in alpha diversity and beta diversity index between groups were analyzed by Tukey and Wilcox tests using R software. Species with significant differences between groups were analyzed by t test using R software.
Data availability. The data sets generated and analyzed during the current study are available in the NCBI Sequence Read Archive (SRA) under the BioProject number PRJNA881714.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.7 MB.