Application of Broccoli Residues to Soil can Suppress Verticillium Wilt of Cotton by Regulating the Bacterial Community Structure of the Rhizosphere

[Aims] Verticillium wilt (VW) of cotton was effectively controlled by application of broccoli residues (BR) to soil. Information regarding the variation in bacterial communities in rhizosphere of cotton cultivars with different VW resistance levels under BR treatment is still lacking and nally to provide guidance for screening effective biocontrol bacteria. [Methods] Real-time uorescence quantitative PCR was used to determine the population of Verticillium dahliae, the effects of BR on the bacterial community structure in rhizosphere were determined by high-throughput sequencing technology. [Results] Results showed that control effects for susceptible cultivar (cv. EJ-1) and resistant cultivar (cv. J863) on VW after BR treatment were 51.76% and 86.15%, the population of V. dahliae decreased by 18.88% and 30.27%, respectively. High-throughput sequencing showed that ACE and Chao1 indices were increased by application of BR. Actinobacteria, Proteobacteria, Bacteroidetes, Gemmatimonadetes, Acidobacteria, and Firmicutes were the most dominant phyla, and relative abundances of these bacterial taxa signicantly differed between cultivars. Additionally, Bacillus stably increased in rhizosphere following BR treatment. Redundancy analysis (RDA) showed that relative abundances of Bacillus, Lysobacter, Streptomyces, Rubrobacter, Gemmatimonas, Bryobacter and Nocardioides were correlated with occurrence of VW. Field experiments demonstrated that dressing cotton seeds with Bacillus subtilis NCD-2 could successfully reduce occurrence of VW, and control effects for EJ-1 and J863 were 35.26% and 31.02%, respectively. [Conclusions] The application of BR changed the bacterial community structure in cotton rhizosphere, decreased the population of V. dahliae in soil, and increased the of benecial microorganisms, thus signicantly VW.


Introduction
Cotton (Gossypium hirsutum L.) is the most important source of natural textile bers worldwide and a signi cant oilseed crop (Zhang et al., 2016). Verticillium wilt (VW), caused by Verticillium dahliae, is a typical soil-borne disease and results in extensive economic losses. In China, losses of approximately 250-310 million US dollars have been reported for cotton annually due to V. dahliae (Li et al., 2015;Rehman et al., 2018). VW is particularly di cult to control due to the long-living dormant microsclerotia produced by the pathogen, which remain viable in the soil for more than two decades (Fradin and Thomma, 2006;Alstrom, 2001), as well as the inability of fungicides to contact the hyphae of V. dahlia after they spread inside the xylem (Klosterman et al. 2009). It is imperative to develop novel control strategies to control this devastating disease.
Previous studies have shown that soil-borne disease management has relied principally upon fumigation (Atallah et al. 2012; Johnson and Dung, 2010;Taylor et al., 2005). However, the application of chemical fumigants to the soil may be environmentally unfriendly (Uppal et al., 2008). Therefore, there is growing interest in the search for alternatives to fumigants for disease control. Many reports have demonstrated that the use of organic soil amendments may be a potential strategy for the control of insect pests,  Akao et al., 2017). The changes in soil microbial community structure caused by organic soil amendments provide useful information on soil health and quality (Poulsen et al., 2013). In particular, the responses of soil bacterial communities to organic soil amendments are particularly important and are believed to be one of the main drivers of disease suppression (Garbeva et al., 2004). The disease-suppressive effects of certain crop residues are well documented, such as those of broccoli, buckwheat, canola, mustard, and sweet corn (Tubeileh and  . Information regarding the variation in bacterial communities in rhizosphere soil that are affected by cotton cultivars that vary in resistance to VW following the application of broccoli residues is still lacking. It is unclear whether and, if so, how cultivar resistance against V. dahliae is related to rhizosphere bacteria. The overall objectives of this study were therefore (i) to determine the effect of BR on the incidence of VW among different cotton cultivars, (ii) to study the differences in the bacterial composition, diversity, and community structure following the application of BR, (iii) to analyze the relationship between disease incidence and the bacterial community, and (iv) to assess the effect of the addition of exogenous Bacillus subtilis on VW in the eld.

Field experiment site
The experimental sites were located in Quzhou County, Hebei Province. Field trials were conducted at two sites, Field A and Field B, from 2017 to 2019. The experimental sites have a long history of cotton cultivation and occurrence of VW. The plots with at terrain, relatively uniform fertility and continuous cotton planting for more than ten years were selected as experimental elds. Soil nutrient characteristics are outlined in detail in a previous publication (Zhao et al., 2019b). Detailed information regarding the eld experiment setting is described in the following statements.

Experimental setup and design
Broccoli was planted in August 2017, and the density of the broccoli plants was approximately 41 thousand plants per hectare. After harvesting the edible part of the broccoli, the remaining parts of the plants were chopped in the eld with a grinder and mechanically incorporated into the soil with a rotovator in early November 2017 at a depth of 25 to 30 cm. The amount of broccoli residues amended into the soil was approximately 57 thousand kilograms per hectare. Parts of the eld were not amended with broccoli residues as a blank control. The susceptible cultivar Ejing 1 (EJ-1) and resistant cultivar Ji 863 (J863) were planted in late April 2018. The experimental design included four treatments: 1) susceptible cultivar EJ-1 planted without broccoli residues (EJ-1-CK); 2) susceptible cultivar EJ-1 planted with broccoli residues (EJ-1-BR); 3) resistant cultivar J863 planted with broccoli residues (J863-BR); 4) resistant cultivar J863 planted without broccoli residues (J863-CK). The experiment had a randomized complete block design with three replicates. All plots were covered with plastic lm and irrigated as necessary. For Field B, in early August 2018, broccoli planting, preparation of BR, and experimental setup and design were similar to those in Field A. After that, cotton was planted in late April 2019.

Soil sample collection and DNA extraction
Soil samples were collected at the owering and boll-forming stages in 2018 and 2019, respectively.
Within each sampling plot, three plants were randomly selected and carefully removed from the soil using a spade. The root systems of the three plants from each plot were rst vigorously shaken to remove loosely adhering soil particles, and then the remaining root systems were combined as a rhizosphere sample. Soil samples were immediately preserved at 4°C for less than 48 hr. To remove plant material, the sample was sieved through a 2.0 mm sieve and stored at -80°C for subsequent DNA extraction. DNA was

Statistical Analyses
Statistically signi cant differences (P < 0.05) in disease incidence, the disease index, DNA copies of V. dahliae, and changes in the bacterial community composition between the control and BR treatments were evaluated with Student's t-test or one-way analysis of variance (ANOVA) using SPSS. Soil bacterial diversity indices were calculated based on resampled OTU abundance matrices in MOTHUR. Principal component analysis (PCA) was performed to explore the differences in soil bacterial community composition. Redundancy analysis (RDA) was performed to examine the relationship between disease occurrence and bacterial community composition. Analysis of similarities (ANOSIM) was performed to identify the signi cant differences in bacterial community structure among treatments. Data on the differences in bacterial community composition among treatments were obtained, and the relative abundances of major taxonomic groups at the phylum and genus levels were compared. Graphs were generated with Origin 8.0 software.

Effects of broccoli residues on VW of different cotton cultivars
Broccoli residues had a signi cant impact on the disease incidence and disease index of cotton VW (P 0.05). Compared with the blank control (no broccoli residues), the disease incidence of cultivar EJ-1 decreased by 38.76% and 53.50% and the disease index decreased by 46.47% and 57.04% in Field A and Field B, respectively. The disease incidence of cultivar J863 decreased by 100% and 63.42% and the disease index decreased by 100% and 72.30% in Field A and Field B, respectively. The average control effects for EJ-1 and J863 were 51.76% and 86.15%, respectively (Fig. 1).
Effect of broccoli residues on DNA copies of V. dahliae in soil When compared with those in the blank control soils, the DNA copies of V. dahliae in the soils associated with the different cotton cultivars were signi cantly reduced by the BR treatment (Fig. 2)

Alpha Diversity Of The Bacterial Community
The alpha diversity of the bacterial community was expressed by the ACE and Chao1 indices in our study ( Fig. 3). In Field A, the ACE index for EJ-1 ranged from 2503 (CK) to 2667 (BR), and the Chao1 index ranged from 2501 (CK) to 2624 (BR), which were greater by 6.55% and 4.92%, respectively. The ACE index for J863 ranged from 2603 (CK) to 2652 (BR), and the Chao1 index ranged from 2585 (CK) to 2617 (BR), which were greater by 1.89% and 1.24%, respectively. In Field B, the ACE index for EJ-1 ranged from 3690 (CK) to 3751 (BR), and the Chao1 index ranged from 3466 (CK) to 3541 (BR), which were greater by 1.64% and 2.15%, respectively. The ACE index for J863 ranged from 3949 (CK) to 3972 (BR), and the Chao1 index ranged from 3520 (CK) to 3655 (BR), which were greater by 0.59% and 3.85%, respectively. These results indicate that the ACE and Chao1 indices were increased by the application of BR at the different eld sites.

Bacterial Community Structure Analyses
Principal component analysis based on the OTU composition was used to study the effect of broccoli residues on the soil bacterial community structure associated with the different cotton cultivars. Figure 4 shows plots of the sites in the plane of the rst two principal coordinates based on the soil bacterial communities in Field A and Field B, respectively. The results show that the bacterial community structure associated with the different cultivars was located in the same quadrant after the application of broccoli residues, while that of the blank controls of the different cultivars was located in different quadrants, which indicates that the bacterial community structure changed and tended to be the same after the application of broccoli residues. In addition, the rst principal component (PC1) and the second principal component (PC2) of the bacterial community structure at the OTU level in rhizosphere soil were found to explain 34.07% and 16.36% of all variables in Field A, and 24.83% and 21.29% of all variables in Field B, respectively. The cumulative contribution rates of variance of the two principal components reached 50.43% and 46.12%, respectively. In addition, ANOSIM indicated that the BR treatment contributed signi cantly to the separation of the CK treatment (R = 0.9815, P = 0.001, Field A) and (R = 0.6481, P = 0.002, Field B).

Comparison Of Bacterial Community Composition
Among all sequences, unknown sequences were classi ed as "other group". In Field A, the dominant bacterial phyla were Proteobacteria, Actinobacteria, Acidobacteria, Gemmatimonadetes, Chloro exi, Bacteroidetes, Planctomycetes, Rokubacteria, Nitrospirae, Verrucomicrobia, Latescibacteria, Firmicutes and Patescibacteria, and these phyla accounted for more than 95% of the total sequences in each sample (Fig. S1). The changes in the relative abundances of the dominant bacterial taxa associated with the different cotton varieties after the application of broccoli residues were compared at the phylum level (Fig. 5). Notably, all dominant bacterial phyla associated with J863 increased in abundance following the application of broccoli residues, while for EJ-1, the dominant bacterial phyla were in uenced to different degrees by the application of the broccoli residues. Among them, Gemmatimonadetes, Rokubacteria, Nitrospirae, Verrucomicrobia and Firmicutes increased. The most abundant group was Firmicutes, which increased by approximately 2.36 and 1.41-fold for EJ-1 and J863, respectively, when compared with the values for CK. However, Proteobacteria, Chloro exi, Bacteroidetes, Planctomycetes, Latescibacteria and Patescibacteria decreased following the application of broccoli residues (Fig. 5A). In Field B, the dominant bacterial phyla were Actinobacteria, Proteobacteria, Acidobacteria, Chloro exi, Gemmatimonadetes, Firmicutes, Bacteroidetes, Planctomycetes, Rokubacteria, Patescibacteria, Entotheonellaeota, Nitrospirae and Verrucomicrobia, and these phyla accounted for more than 95% of the total sequences in each sample (Fig. S2). For the cultivars EJ-1 and J863, the dominant bacterial phyla were in uenced to different degrees by treatment with broccoli residues. Actinobacteria, Gemmatimonadetes and Firmicutes were increased by the application of broccoli residues. The most abundant group was also Firmicutes in Field B, and the fold changes were 1.42 and 1.27 for EJ-1 and J863, respectively. Acidobacteria decreased as a result of the application of broccoli residues. In addition, Proteobacteria, Bacteroidetes, Patescibacteria, Entotheonellaeota, Nitrospirae and Verrucomicrobia decreased for EJ-1, while the opposite tendency was observed for J863 (Fig. 5B). Based on the results for the different cultivars and eld sites, the relative abundances of Actinobacteria, Gemmatimonadetes and Firmicutes in the soil increased after the application of broccoli residues.

Relationships between the occurrence of VW and bacterial community composition
The relationships between the occurrence of VW and bacterial community composition in Field A and Field B were studied with RDA (Fig. 7). For Field A, the RDA that was performed with the genera and disease incidence data showed that the rst two RDA components could explain 52.3% of the total variation (Fig. 7A). As shown by their close grouping and by the vectors, the disease incidence of cultivar J863 was positively related to the abundant genera Gemmatimonas, Pontibacter, RB41, Blastococcus and Massilia after the application of broccoli residues, and it was negatively related to Bacillus, Lysobacter, and Nitrospira. However, the disease incidence for EJ-1 treated with BR was positively related to the abundant genera Streptomyces, Rubrobacter, Bryobacter and Nocardioides, and it was negatively related to Gemmatimonas, Pontibacter, RB41, Blastococcus and Massilia.
For Field B, the RDA that was performed with the genera and disease incidence data showed that the rst two RDA components could explain 47% of the total variation (Fig. 7B). The disease incidence in the BR treatment for the cultivars (including J863 and EJ-1) was positively related to the abundant genera Bacillus, Nocardioides, RB41, Rubrobacter, and Arthrobacter and negatively related to Streptomyces, Nitrospira, Sphingomonas, and Lysobacter.

Control effect of the exogenous application of Bacillus subtilis NCD-2 on VW
The control effect of the application of B. subtilis on cotton VW was investigated in our study. As indicated in Fig. 8, the control effect of BS on the disease at the boll-forming stage for EJ-1 was 38.55%, while that for J863 was 26.73%. Additionally, further study at boll opening showed that the control effect for EJ-1 was 31.96%, while that for J863 was 35.31%. The average control effects for EJ-1 and J863 were 35.26% and 31.02%, respectively.

Discussion
The use of crop residues is an important method associated with the suppression of VW, such as those of broccoli, buckwheat, canola, mustard, and sweet corn ( , 1999). Therefore, the application of broccoli residues provides a new method and ideas for the sustainable ecological control of cotton VW. In our previous study, the main potential mechanism by which broccoli residues incorporation into the culture substrate reduced the DNA copies of V. dahliae and inhibited the spread of V. dahliae was revealed by real-time PCR and confocal microscopy methods (Wang et al., 2020; Zhao et al., 2019a). However, more in-depth research should be performed to further explore this potential mechanism, especially from the perspective of rhizosphere microbiomics. Microbiome-based research has opened a new frontier that will greatly expand our knowledge of the relationship between plant disease incidence and microbiota and offer new opportunities for developing novel approaches for biocontrol. To our knowledge, although the Illumina  ) found that there was no signi cant correlation between bacterial community indices and banana Fusarium wilt following the application of bioorganic fertilizer. In the present study, although there were no signi cant differences in the alpha diversity indices following the application of BR, the values of those indices increased. Moreover, the treatment with broccoli residues had a signi cant impact on the soil bacterial community structure, which was consistent with the results of previous studies on the changes in bacterial community structure in the rhizospheric soil of eggplant (Inderbitzin et al., 2018). In addition, the bacterial community structure associated with the different resistant cotton cultivars changed in comparison to that in CK and was located in the same quadrant after the application of broccoli residues, indicating that the bacterial community structure tended to be the same after the application of broccoli residues (Fig. 4). In terms of the bacterial community composition, the analysis at the phylum level revealed that Actinobacteria, Proteobacteria, Bacteroidetes, Gemmatimonadetes, Acidobacteria, and Firmicutes were the most common phyla, but with some changes in relative abundance. This nding roughly corresponded with those of previous articles (Inderbitzin et al., 2018;Shen et al., 2014). Inderbitzin et al. (2018) found that the ve dominant phyla of soil bacteria were Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes, and Acidobacteria. Among them, Actinobacteria and Proteobacteria were more abundant after treatment with broccoli residues than in the control, while the opposite tendency of Bacteroidetes, Firmicutes, and Acidobacteria was observed. Our study found that the relative abundances of Actinobacteria and Proteobacteria also increased for the resistant cultivar J863, which was consistent with previous studies (Inderbitzin et al., 2018). However, there was no consistent conclusion for these two phyla for the susceptible cultivar EJ-1. The reason for this nding is still unclear and may be due to differences in the types or contents of root exudates among cotton cultivars, which can cause different microbial communities to be recruited. Moreover, the relative abundance of Firmicutes in the soil associated with different resistant cotton cultivars increased after the application of broccoli residues, which was not consistent with the ndings of Inderbitzin et al. (2018).
Analysis of the dominant genera also revealed signi cant differences in the bacterial communities among the different treatments. Among all the cotton cultivars, the abundance of Bacillus was increased by treatment with the broccoli residues. RDA showed that the incidence of VW might be positively related . Bacillus subtilis NCD-2 was rst isolated from cotton rhizosphere soil in Hebei Province and showed excellent biological control of soil-borne diseases (Li et al., 2005). In the present study, Bacillus signi cantly increased in abundance after the application of broccoli residues. Therefore, to verify the control effect of B. subtilis NCD-2 against VW, eld experiments were executed by seed dressing with B. subtilis. However, the control effect of strain NCD-2 on disease was approximately 35% for different cotton cultivars. Some researchers have reported that direct applications of potentially bene cial species often result in poor disease suppression due to their low survival and colonization in soil (Saravanan et al., 2003;Lugtenberg and Kamilova, 2009). Therefore, the survival or abundance of the biocontrol inoculant B. subtilis in rhizosphere soil will be studied in future research.
In addition, many studies have shown that soil physical and chemical properties such as soil nutrients, pH and organic matter are important factors affecting the structure of the soil microbial community

Conclusions
In this study, the incidence of cotton VW and the population of V. dahliae in the rhizosphere of cotton cultivars with different verticillium wilt resistance levels were decreased by treatment with BR. Highthroughput sequencing showed that bacterial diversity was increased by the application of BR. The relative abundances of Bacillus, Lysobacter, Streptomyces, Rubrobacter, Gemmatimonas, Bryobacter and Nocardioides were correlated with the occurrence of verticillium wilt. These results provide important information necessary for a better understanding of bacterial community structure in rhizosphere soil after treatment with BR.

Declarations
Data availability statement The raw sequence data reported in this paper have been deposited at the National Center for Biotechnology Information (NCBI) under accession number PRJNA734729 and PRJNA734770.
Author contributions WZ, QG, SL, and PM planned, designed the research, and experiments. WZ, PW, LD, XZ, ZS, and XL performed the experiments. WZ and PM analyzed the data and wrote the manuscript. All authors read and approved the nal manuscript.