Effect of disease severity on the structure and diversity of the phyllosphere microbial community in tobacco

Tobacco target spot is a serious fungal disease and it is important to study the similarities and differences between fungal and bacterial community under different disease severities to provide guidance for the biological control of tobacco target spot. In this study, tobacco leaves at disease severity level of 1, 5, 7 and 9 (S1, S5, S7, and S9) were collected, both healthy and diseased leaf tissues for each level were sampled. The community structure and diversity of fungi and bacteria in tobacco leaves with different disease severities were compared using high-throughput sequencing technology. The results indicated that there was a significant differences in the community structure of fungi and bacteria for both healthy and diseased samples depending on the disease severity. In both healthy and diseased tobacco leaves for all four different disease severities, the most dominant fungal phylum was Basidiomycota with a high prevalence of genus Thanatephorus. The relative abundance of Thanatephorus was most found at S9 diseased samples. Proteobacteria represent the most prominent bacterial phylum, with Pseudomonas as predominant genus, followed by Pantoea. The relative abundance of Pseudomonas was most found at S7 healthy samples. In fungal community, the Alpha-diversity of healthy samples was higher than that of diseased samples. In contrast, in bacterial community, the Alpha-diversity of healthy samples was lower than that of diseased samples. LEfSe analysis showed that the most enrich fungal biomarker was Thanatephorus cucumeris in diseased samples. Clostridium disporicum and Ralstonia pickettii were the most enrich bacterial biomarker in healthy samples. FUNGuild analysis showed that the pathotroph mode was the most abundant trophic modes. The relative abundance of pathotroph mode in diseased samples changes insignificantly, but a peak at S5 was observed for healthy samples. PICRUSt analysis showed that most bacterial gene sequences seem to be independent of the disease severity. The results of this study provide scientific references for future studies on tobacco phyllosphere microecology aiming at prevention and control of tobacco target spot.

Tobacco target spot is a serious fungal disease and it is important to study the similarities and differences between fungal and bacterial community under different disease severities to provide guidance for the biological control of tobacco target spot. In this study, tobacco leaves at disease severity level of 1, 5, 7 and 9 (S1, S5, S7, and S9) were collected, both healthy and diseased leaf tissues for each level were sampled. The community structure and diversity of fungi and bacteria in tobacco leaves with different disease severities were compared using high-throughput sequencing technology. The results indicated that there was a significant differences in the community structure of fungi and bacteria for both healthy and diseased samples depending on the disease severity. In both healthy and diseased tobacco leaves for all four different disease severities, the most dominant fungal phylum was Basidiomycota with a high prevalence of genus Thanatephorus. The relative abundance of Thanatephorus was most found at S9 diseased samples. Proteobacteria represent the most prominent bacterial phylum, with Pseudomonas as predominant genus, followed by Pantoea. The relative abundance of Pseudomonas was most found at S7 healthy samples. In fungal community, the Alpha-diversity of healthy samples was higher than that of diseased samples. In contrast, in bacterial community, the Alpha-diversity of healthy samples was lower than that of diseased samples. LEfSe analysis showed that the most enrich fungal biomarker was Thanatephorus cucumeris in diseased samples. Clostridium disporicum and Ralstonia pickettii were the most enrich bacterial biomarker in healthy samples. FUNGuild analysis showed that the pathotroph mode was the most abundant trophic modes. The relative abundance of pathotroph mode in diseased samples changes insignificantly, but a peak at S5 was observed for healthy samples. PICRUSt analysis showed that most bacterial gene sequences seem to be independent of the disease severity. The results of this study provide scientific references for future studies on tobacco phyllosphere microecology aiming at prevention and control of tobacco target spot.

Introduction
Tobacco (Nicotiana tabacum L.) is one of the most important economic crops in China (Wang et al., 2016). However, the tobacco production is constrained by many tobacco leaf spot diseases (Zhu, 2002). In between, tobacco target spot is one of the most destructive types (Shew and Main, 1990). The pathogen of tobacco target spot is Rhizoctonia solani, and its teleomorph is Thanatephorus cucumeris (Elliott et al., 2008). Typical symptoms of tobacco target spots firstly appear on the old leaves as round watery spots, the tobacco leaves are chlorotic, and with yellow halo. Then the spots become a diammer of 2 to 20 cm, with concentric ring lines and dead spots (Sun et al., 2022). In last 5 years, tobacco target spot was the most severe and most common disease happened in tobacco in China with losses reaching up to 100% (Yang et al., 2014). It believes that a better understanding of pathogens from a plant microbial community perspective may lead to other ways of controlling disease in a more sustainable manner.
Changes of phyllospheric fungi and bacteria can potentially be used to evaluate the effects of efficient disease management (Huang et al., 2021a). With the development of sequencing technology, the number of studies on phyllosphere microorganisms has gradually increased in recent years, creating a new era of phyllosphere microbial community research (Hassani et al., 2018;Fitzpatrick et al., 2020;Gong and Xin, 2020). Plant phyllospheric fungi and bacteria exist in certain or all life stages of the host plant. They live in the leaves, stems, flowers, and fruits of the plant, and establish a coevolutionary relationship with the plant (Xie et al., 2021). Phylloshperic fungi and bacteria can colonize the surface (i.e., epiphytic microbiota) or interior of leaf tissue (i.e., endophytic microbiota; Agler et al., 2016). Phylloplane microorganisms include many potentially beneficial, pathogenic, and antagonistic microbes affecting plant health and productivity (Lindow and Brandl, 2003). Carlstrom et al. reported that substantial interactions exist among strains, and that a preestablished community is largely resistant to subsequent perturbations (Carlstrom et al., 2019). In tobacco, these sequencing technology have been used widely to study the structure and diversity of phyllosphere microbial community for many diseases, including tobacco brown spot caused by Alternara altanata (Dai et al., 2022), leaf spot caused bay Didymella segeticola (Huang et al., 2021a), tobacco pole rot caused by Rhizopus oryzae , and tobacco powder mildew caused by Golovinomyces cichoracearum DC (Huang et al., 2021b).
An increasing body of evidence suggests that the microbial communities can positively or negatively affect the fitness and health of plants by promoting disease (Bertelsen et al., 2001). The disease severity is an important factor affecting the quality of roasted tobacco. The more severe the disease is, the lower is the quality of roasted tobacco . Disease severity largely affects the structure and diversity of phyllosphere microbial community. The results of a previous study on the influence of the disease severity on the community structure and diversity of the tobacco leaf spot disease caused by Didymella segeticola showed that the community diversity of phyllosphere fungi decreased with the increase of the disease severity (Huang et al., 2021a). Similarly, Luo et al. (2017) studied the structure and diversity of phyllosphere bacterial community of pumpkin powdery mildew under different disease severities and identified that there was a significant difference in the bacterial community structure under different disease severities. Zhang et al. studied the structure and diversity of phyllosphere fungal community of pumpkin powdery mildew at different disease severities and concluded that the abundance of dominant genera in the community increased with the increase of the disease severity (Zhang et al., 2018). Based on the results reported in Zhang et al. (2019), most genera abundance decreased with the powdery mildew pathogen Erysiphe increasing. Although many studies on the effect of disease severity on the phyllosphere microbial community have been carried out, the effect of disease severity on the structure and diversity of tobacco target spot phyllosphere microbial community is still poor understood.
Due to serious damage of tobacco target spot, it is important to understand how the microbiota changes on tobacco leaves at different tobacco target spot disease severities. Therefore, the goals of this study were to identify the diversity and community structure of phyllosphere fungi and bacteria in tobacco target spot under different disease severities and analyzed differences depending on the disease severity. The results of this study will provide some potential guidance for the efficient prevention and control of tobacco target spot.

Sampling sites and sampling strategy
In August 2021, one tobacco field with serious tobacco target spot in Zheng'an County (28°55′N, 107°44′E), Zunyi City, Guizhou Province of China, was selected as the sampling site. The 1, 5, 7 and 9 disease severities leaves were randomly selected using criteria based on the Chinese National Standard (GB/T 23222-2008). For 1, 5, 7 and 9 disease severities leaves (hereinafter referred to as S1, S5, S7, S9), the proportion of the lesion area of Frontiers in Microbiology 03 frontiersin.org every leaf ranges from 0 to 1%,11 to 20%,21 to 40%,and 41 to 100%,. Randomly selected leaves were divided into two parts with and without obvious leaf spots and two parts (as shown in Figure 1F) were cut with sterilized scissors and put in to 50-mL sterile centrifuge tubes, which were labelled diseased samples and healthy samples. For coding samples, the letter D was used for diseased samples and H for healthy samples, which was followed by the number 1 for leaves at 1 disease severity level, 5 for leaves at 5 disease severity, 7 for leaves at 7 disease severity, 9 for leaves at 9 disease severity. For example, D1 was diseased samples from leaves at 1 disease severity. Three biological repeats were conducted. Diseased and healthy samples of the same disease severity were taken from the same leaves. Sample information was provided in Table 1. After sampling, the samples were placed into a low-temperature storage box and immediately transported to the laboratory and stored at −80°C until further use.

DNA extraction, PCR amplification, and high-throughput sequencing
The total DNA of each sample (n = 24) was extracted from the leaf tissues using the MP FastDNA SPIN Kit for Soil (MP Biomedicals, Santa Ana, United States). The purity and concentration of sample DNA were detected with gel electrophoresis. The extracted DNA was placed in a centrifuge tube and diluted to a concentration of 1 ng/μL with sterile water, which was used as DNA template. The fungal ITS1-5F region was amplified using the primers ITS1F (5'-CTTGGTCAT TTAGAGGAAGTAA-3′) and ITS2R (5'-GCTGCGTT CTTCATCGATGC-3′) (Xiang et al., 2020a). Furthermore, 338F (5'-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5'-GGACTACHVGGGTWTCTAAT-3′) (Xiang et al., 2020b) were used to amplify the bacterial 16S V3-V4 region. The PCR amplification was carried out based on methods described in the references . The PCR products were recovered by using the AxiPrepDNA gel recovery kit (AXYGEN Company, Silicon Valley, United States), eluted with Tris HCl, and detected by using 2% agarose gel electrophoresis. The amplification products were amplified using an Illumina Miseq platform and the Miseq library was constructed and sequenced on the Miseq computer.

Data processing
After the Illumina Miseq platform sequencing, 2.3 GB raw reads were generated. All sequences were clustered according to 97% similarity with UPARSE (Version 7.1) and chimeras were filtered during operational taxonomic unit (OTU) clustering using the cluster otus command. Subsequently, unidentified bases and sequences annotated as plastids and mitochondria were removed. The OTUs were annotated with SILVA 132 and UNITE version 7.2 for the bacterial 16S rRNA and fungal ITS, respectively, set the confidence threshold to 70% (Edgar and Robert, 2013;Urmas et al., 2013). The Alpha-diversity and Betadiversity were calculated with Qiime software (version 1.9.1) and the R statistics package (R Core Team, 2017) was utilized to draw the Principal Co-ordinate Analysis (PCoA) diagram and create the rarefaction curve (Version 2.15.3) and Venn diagrams (Version 3.0.3). The linear discriminant analysis (LDA) effect Whole plant (A) and leaves displaying disease severity at (B) S1, (C) S5, (D) S7, and (E) S9, respectively. (F), sampling parts display of diseased and healthy samples on tobacco leaves.
Frontiers in Microbiology 04 frontiersin.org size (LEfSe) method 1 was used to detect the potential biomarkers. The LDA = 4 discriminant score was used as the threshold value and a significant α of 0.05 was used to divide the specific biomarkers of each disease severity sample (Segata et al., 2011). Fungal nutritional modes were analyzed with the FUNGuild database (Nguyen et al., 2016). Bacterial metabolic functions were analyzed with the bioinformatics software package PICRUSt (Langille et al., 2013). Software of Cytoscape 3.9.1 2 was used to evaluate the interactions among microbial community rate. The correlation was highly significant (p < 0.05) and with Spearman's rho >0.6 and <−0.6 were considered to be correlated and selected for network analysis. (Williams et al., 2014;Meng et al., 2017).

Statistical analysis
IBM SPSS Statistics 23 (IBM Corp., New York, United States) software was used to perform one-way ANOVA on measured data. The mean value and standard error were used to express the measured results. The Duncan's new multiple range test was employed to test the significance of differences between different treatments. A p ≤ 0.05 was considered to be statistically significant.

Quality of total fungal and bacterial sequence data
With the increase in number of sequences, the rarefaction curve gradually approached the plateau phase. When the number of fungal and bacterial sequences reaches 600 and 1,000, respectively, the rarefaction curve was saturated (Figure 2), 1 http://huttenhower.sph.harvard.edu/lefse/ 2 https://cytoscape.org/ indicating that the sequencing results reflect the structure of the phyllosphere community in tobacco leaves. A total of 2,139,800 fungal sequences were obtained from 24 samples. These sequences were classified into 484 OTUs at a 97% similarity level. For bacteria, a total of 2,184,600 sequences were classified into 271 OTUs across the 24 samples. The fungal and bacterial sequences of each sample were deposited in the SRA database under accession numbers PRJNA4081 and PRJNA4091, respectively.

Distribution and diversity of fungal and bacterial operational taxonomic units
At the OTU level, the distribution of OTUs among different disease severities was depicted with Venn diagrams (Figure 3). For fungal community, a total of 484 OTUs were obtained among which 133 and 351 belong to diseased and healthy samples, respectively ( Figures 3A,B). The number of shared OTUs among the four diseased samples was 15. Only 24, 21, 9, and 25 OTUs were identified in D1, D5, D7 and D9 of the diseased sample, respectively. The number of shared OTUs among the four healthy samples was 49. Healthy samples H1, H5, H7, and H9 only included 75, 60, 42, and 23 unique OTUs, respectively. Notably, the number of fungal OTUs in diseased samples was lower than that in the healthy samples. The diversity index (shannon index) of diseased and healthy samples ranged from 0.46 ± 0.22 (mean ± SD) to 1.17 ± 0.88 and 2.86 ± 2.12 to 5.19 ± 0.43, respectively (Table 2). For diseased samples, the shannon index was the lowest and highest at a disease severity of 9 and 5, respectively. For healthy samples, the lowest and highest value was observed at S5 and S1, respectively. Noted that the shannon index of the four diseased samples and four healthy samples significantly differed. The richness index (chao 1 index) of diseased and healthy samples ranged from 23.19 ± 7.16 to 70.50 ± 10.50 and 101.46 ± 16.53 to 132.69 ± 45.76, respectively. For diseased samples, the chao 1 index was the lowest and highest at S9 and S5, respectively. For healthy samples, the chao 1 index was the lowest and highest at S5 and S1, respectively. The richness index (ACE index) of diseased and healthy samples ranged from 34.30 ± 7.56 to 74.42 ± 6.66 and 100.30 ± 16.38 to 144.71 ± 35.14, respectively. For diseased samples, the ACE index was the lowest and highest at S9 and S5, respectively. For healthy samples, the lowest and highest values were observed at S5 and S7, respectively. The shannon, chao 1 and ACE indices showed that the diversity and richness of the fungal community in healthy samples were higher than those in diseased samples (Table 2).
For bacterial community, 271 OTUs were observed among which 134 and 137 belong to diseased and healthy samples, respectively ( Figure 3). The four diseased and four healthy samples contained 21 and 18 shared OTUs, respectively. Diseased samples D1, D5, D7, and D9 contained 26, 12, 5, and 45 OTUs, respectively. Healthy samples H1, H5, H7, and H9 only include 10, 36, 17, and 34 unique OTUs, respectively. Notably, the number of bacterial OTUs in the diseased samples was lower than that in the healthy Accordingly, for diseased samples, the lowest and highest values of the shannon, chao 1, ACE indices were observed at S1 and S5, S7 and S5, S9 and S1, respectively. For healthy samples, the lowest and highest values of the shannon, chao 1, ACE indices were observed at S1 and S9, S1 and S5, S1 and S5, respectively. In general, the Alpha-diversity index of the bacterial community in diseased samples was higher than that in healthy samples, except for groups D7 and D9.

Fungal and bacterial community composition Fungal community composition
Based on the comparison of the sequencing and ITS data, all samples contain six phyla of fungi ( Figure 4A) .54%, respectively. The relative abundance of Ascomycota in diseased groups (D1, D5, D7, D9) of four different disease severities were 3.50, 4.04, 1.63, and 2.37%, respectively. For healthy groups (H1, H5, H7, H9), the relative abundances of Ascomycota were 32.84, 15.14, 24.16, and 18.39%, respectively. In summary, the results showed that the relative abundance of Basidiomycota in diseased samples was higher than that in healthy samples. In contrast, the relative abundance of Ascomycota in diseased samples was lower than that in healthy samples (Table 3).
The top 30 fungal genera at the genus level were displayed in Figure 4B. The top 10 genera were Thanatephorus, Alternaria, Trichocladium, Fusarium, Cladosporium, Candida, Sonoraphlyctis, Zygoascus, Agrocybe, and Allomyces (Table 3). The relative abundances of various genera in different leaf samples differ. The relative abundance of these genera in diseased samples were lower than those in healthy samples, except for Thanatephorus (the genera to which the pathogen of tobacco target spot belongs). Thanatephorus is the dominant genus in both diseased and healthy samples and the relative abundance of Thanatephorus in diseased samples and healthy samples significantly differ at different disease severities. The relative abundance of Thanatephorus showed a downward at S1 ~ S5 and then an upward trend at S5 ~ S9 in diseased samples and showed a peak at S9. However, in healthy samples, the relative abundance of Thanatephorus showed an upward at S1 ~ S5 and then a downward at S5 ~ S9 and showed a peak at S5. The relative abundance of Thanatephorus in diseased and healthy samples significantly differ at S5 compared to the other three disease severities samples. The top 100 genera were shown in the maximum-likelihood tree. The results showed that the dominant fungi belong to Basidiomycota, followed by Ascomycota. Thanatephorus was the dominant fungal genus of A B

FIGURE 2
Rarefaction curves of fungal (A) and bacterial (B) OTUs across different tobacco leaf samples.
The top 30 bacterial genera at the genus level were displayed in Figure 4D and the top 10 genera at the genus level were   Figure 5B).

Spatial distribution of fungal and bacterial communities
PCoA plots were used to reveal the spatial distribution of fungal and bacterial community ( Figure 6). In fungial community, diseased groups (D1, D5, D7, and D9) with four disease severities were clustered together. In contrast, healthy groups (H1, H5, H7, and H9) with four disease severities were irregularly separated from each other ( Figure 6A). For bacterial community, all diseased and healthy groups (D1, D5, D7, and D9; and H1, H5, H7, and H9) were irregularly separated from each other ( Figure 6B). This indicated that there was a significant differences in the spatial distributions of fungal and bacterial communities in diseased and healthy samples.

Fungal and bacterial co-occurrence
Network analysis of the co-occurrence of the most abundant 50 fungial species revealed that there were 49 fungal species showing positive co-occurrence with each other, except for Thanatephorous showed a highly significant negative co-occurrence with 16 genus including Fusarium, Pantospora, Alternaria, Symmetrospora etc. (p < 0.05, r < −0.5) ( Figure 13A). Network analysis of the co-occurrence of the most abundant 50 bacterial species revealed that all 50 bacterial species showed positive co-occurrence with at least one other bacterial species (p < 0.05, r > 0.5) ( Figure 13B). There was no bacterial species to show negative co-occurrence with at least one other bacterial species.

Discussion
Microbes play an important role in plant growth and development and resistance to biotic and abiotic stresses (Luo et al., 2017). Simultaneously, phyllosphere microbes play important roles in maintaining leaf health and ecosystem balance (Baker et al., 2010). In this study, the V3-V4 hypervariable regions of the 16S rRNA and ITS region of rDNA were amplified to determine the effect of the disease severity of tobacco target spot on microbial diversity. The results of a previous study showed that disease severity reduced the diversity of microbial communities (Manching et al., 2014). The results of this study also showed that the shannon, chao 1, ACE indices decreased with the increase of disease severity. Similarly, based on a study on bacterial diversity in needles with shoot blight with different disease severities, the diversity and abundance of endophytic bacteria gradually decreased with the aggravation of the disease (Xie et al., 2021). Shen et al. reported that different tissues of eucalyptus were differently affected by diseases, which was consistent with the conclusion drawn in this study about the Alpha-diversity of Frontiers in Microbiology 12 frontiersin.org healthy tissues was higher than that of diseased tissues (Shen et al., 2016). Pathogen stimulation has a significant effect on the composition of the microbial community (Erlacher et al., 2014). The ITS dataset showed that Basidiomycota was the dominant phylum in both diseased and healthy samples at all four disease severities. The relative abundance of Basidiomycota gradually increased with the increase of disease severity, whereas the relative abundance of Ascomycota gradually decreased with the increase of disease severity, similar to the trend reported by Zhang et al. (2018) when studying pumpkin leaves affected by powdery mildew. The dominant fungal genus of both diseased and healthy leaves was Thanatephorous, which was consistent with the research results obtained for culturable fungi isolation by tissue isolation (Hou et al., 2018). The relative abundance of Thanatephorous positively correlated with the disease severity in both diseased and healthy samples, which is consistent with the results reported by Zhang et al. (2018) who pointed out that A B

FIGURE 6
Principal Co-ordinate Analysis (PCoA) of the fungal (A) and bacterial (B) communities in the different group sample.
LEfSe analysis of fungi abundance between diseased and healthy samples of different disease severities. (A), (C), (E), (G) are LDA score identified the size of differentiation between diseased and healthy groups of 1, 5, 7, 9 disease severity, respectively, with a threshold value of 4. (B), (D), (F), (H) are the cladogram of microbial communities of 1, 5, 7, 9 disease severity, respectively. Microbiology  13 frontiersin.org the abundance of Podosphaere (pathogen of pumpkin powdery mildew) increased with the increase of disease severity. The results of this study showed that the relative abundance of eight genera in diseased samples decreased from S1 to S9, except for Thanatephorous and Allomyces. For healthy samples, the relative abundance of six genera increased from S1 to S9, except for Relative abundance of fungal functional groups (modes) based on OTU annotation table with disturbance frequency level. Thanatephorous, Trichocladium, Agrocybe, and Zygoascus. The above-mentioned two phenomena indicated the existence of synergistic or antagonistic relationships between Thanatephorous and the other nine genera. This has also been pointed out in previous studies. For example, Arnold et al. reported that foliar fungi can alter the plant disease severity (Arnold et al., 2003). Simultaneously, Busby et al. pointed out that Cladosporium, Trichoderma, and Epicoccum can alter the disease severity caused by the Melampsora rust (Busby et al., 2016).

Frontiers in
The 16S dataset showed that Proteobacteria represent the dominant phylum in both diseased and healthy samples at all Frontiers in Microbiology 15 frontiersin.org disease severities. The relative abundance of Proteobacteria gradually increased with the increase of disease severity in diseased samples, whereas it decreased with the increase of disease severity in healthy samples, which was similar to the trend reported by Xie et al. (2021) when studying the bacterial diversity in needles with different disease severities of shoot blight. The dominant bacterial genus of both diseased and healthy leaves was Pseudomonas, followed by Pantoea, which was in agreement with the results of previous studies of phyllospheric microorganisms of tobacco (Liu et al., 2020;Chen et al., 2021). It is of significance that the relative abundance of Pseudomonas and Pantoea correlate with the disease severity. Their relative abundance increased with the disease severity increasing. One can speculate that they may have a coevolutionary effect with pathogenic Thanatephorous. Genus Pseudomonas can cause diseases in tobacco; for example, Pseudomonas syringae pv. tabaci has been reported to cause tobacco wildfire (Wolf and Foster, 1918) and Pseudomonas syringae pv. angula causes tobacco angular leaf spot (Cole, 1960). Simultaneously, Pantoea agglomerans has also been reported to cause bacterial soft rot in Chinese cabbage (Guo et al., 2019). In this study, several potential bacterial pathogens, such as biocontrol bacterium Paenibacillus, were identified by high-sequence analysis (Tsai et al., 2022). Paenibacillus polymyxa reportedly protects Arabidopsis thaliana against pathogens and abiotic stress (Salme et al., 2005). Based on the prediction of the bacterial function, the cell motility, amino acid metabolism, lipid metabolism, cellular processes, and signaling in diseased samples increased with the disease severity increasing, which may be caused by the invasion of pathogenic T. cucumeris (Luo et al., 2017). It is worth noting that the turning point of the relative abundance change of Thanatephorous at S5 and the LEfSe analysis showed that T. cucumeris was the most abundant biomarker in diseased samples at S1, S7 and S9. It is speculated that the pathogen propagates in large numbers after colonization, causing leaf spots on the tobacco leaves. Other phyllosphere microorganisms compete with the pathogen for ecological niches at S5. There are two situations. One is that the pathogen occupies more ecological niches (the relative abundance increases), which increases the disease severity of tobacco leaves. The second is that pathogen occupies less ecological niches (the relative abundance decreases), and the disease severity of tobacco leaves remains unchanged (Wintermute and Silver, 2010;Perazzolli et al., 2014). In this study, Thanatephorus showed no positive co-occurrence with any fungal genera but did have a negative co-occurrence with Fusarium, Pantospora, Alternaria, etc., which may because Thanatephorus greatly perturbed the resident fungal microbiota (Hassani et al., 2018). In bacterial community, there was no bacterial species shown any negative co-occurrence with each other, which meant that the most bacterial did not exert major effects on each other.

Conclusion
The diversity and abundance of microorganisms in tobacco leaves were strongly affected by the tobacco target spot disease severity. The structure of fungal and bacterial community were consistent with the changes caused by tobacco target spot pathogen. A genus with high relative abundance in the fungal community was Thanatephorus and genera with high relative abundance in the bacterial community were Pseudomonas and Pantoea. The infestation with the tobacco target spot fungus affected the community structure and diversity of phyllosphere bacteria and fungi in tobacco. The fungal diversity and relative abundance increased with the tobacco target spot disease severity increasing. The bacterial diversity index decreased with the tobacco target spot disease severity increasing.

Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih.gov/, PRJNA887197 https://www.ncbi.nlm.nih.gov/, PRJNA887218.