In August 2020, a test was conducted on a Yunyan 105 tobacco field infected with tobacco brown spot caused by Alternaria alternata was used for the experiment in the area of Heishou Town, Weining County, Bijie City, Guizhou Province (26.74 N, 104.02E) of China. Before the experiment, tobacco brown spots appeared sporadously on tobacco plants (Yunyan 105) in the field, and the incidence rate was about 15–25%. The fully randomized block was designed to subject plants in an open field, when tobacco brown spot occurs sporadically in the field, treated with 57.6% Cu(OH)₂ WP (Newham, Qingdao, China), and the dosage is the recommended dosage (0.52 a.i. kg/ha). A multi-function sprayer (model: DSF01A-20-100, Guizhou Qian Fengyuan Agricultural Technology Development Co., Ltd.) was used to evenly spray the leaf surface until the droplet was lost.

At 0 days (before treatment) and then 5, 10, and 15 days, the tobacco leaf tissues of diseased and healthy plants that had been treated with copper hydroxide were cut with sterilizing scissors. Three biological replicates were conducted. Leaf samples were placed in sterile 50 mL centrifuge tubes and stored at 4°C in a cryopreservation box. One part was used for metabolic function study (5 g), and the other part was stored in the refrigerator at −80°C for application sequencing.

For designating codes to samples, the first and second letters QY mean copper hydroxide, the next letter J was used for healthy tobacco, and B for diseased tobacco leaves, followed by a number, which represents the time taken for samples to be taken, 0 for the 0 days before application, 1 for 5 days after application, 2 for 10 days after application, 3 for 15 days, and then followed by 1 for the first replicate, 2 for the second replicate, 3 for the third replicate. To be more specific, QYB33 represents the third replicate of the diseased leaves after 15 days of copper hydroxide application (Table 1). In this study, healthy and diseased tobacco leaves were selected to study the effects of copper hydroxide on phyllosphere microbial community composition and metabolism before the application of fungicide and 5, 10, 15 days after the application of fungicide.

Total DNA of tobacco samples was extracted by CTAB method (Saghai-Maroof et al., 1984). DNA concentration and purity were determined by 1% agarose gel electrophoresis. According to the concentration, DNA was diluted to 1 ng/μL using sterile water. The fungal primers ITS1-5F-F (5’-GGAAGTAAAAGTCGTAACAAGG-3′) and ITS1-1F-R (5’-GCTGCGTTCTTCATCGATGC-3′) were used as templates to amplify the fungal ITS region (Zhang et al., 2010). Bacterial primers 515F (5’-GTGYCAGCMGCCGCGGTAA-3′) and 806R (5’-GGACTACHVGGGTWTCTAAT-3′) were used to amplify the V4 region of bacteria (Thompson et al., 2017). The PCR reaction consisted of 20 μL (4 μL 5 × FastPfu Buffer, 2 μL 2.5 mM dNTPs, 0.8 μL each primer, 0.4 μL FastPfu polymerase, and 1 μL DNA and 11.8 μL water). Fungal and bacterial PCR reactions were performed on a peqSTAR thermocycler (PEQLAB Ltd., United Kingdom). Fungal PCR was set up as follows: 94°C for 5 min, followed by 35 cycles of 94°C for 1 min, 57°C for 1 min, 72°C for 1 min, and finally at 72°C for 5 min. The bacterial PCR setup consisted of the following steps: 94°C for 3 min, followed by 30 cycles of 94°C for 45 s, 55°C for 45 s, 72°C for 90 s, and finally 72°C for 7 min (Huang et al., 2021). PCR products were checked by 2% agarose gel electrophoresis and purified with Gene JET (Thermo Fisher Scientific, Waltham, MA, United States), then sent for sequencing (250 bp pairedend sequencing) according to the standard protocol. The IonS5 XL platform (Thermo Fisher Scientific, Waltham, MA, United States) was used for high-throughput sequencing at Novogene Bioinformatics Technology Co., Beijing, China.

One gram of diseased and healthy mixed tobacco leaves at different periods were taken and placed in a 100 mL triangular flask containing 50 mL 0.8% sterile normal saline. The flask was incubated at 180 rpm for 2 h at 28°C, and then stood for 30 min at night after shaking. 100 μL of the supernatant was added to the test well of Biolog ECO plate in turn (Grove et al., 2004). The ECO plate was placed in OmniLog incubator and incubated at 28°C for 7 days. Biolog D5E_OKA_data.exe software was used to collect the color changes in the metabolic wells of tobacco phyllosphere microorganisms during the growth process. After collection, HemI software was used to make a heat map to analyze the metabolic function of sample interfoliar microorganisms (Deng et al., 2014).

The sequencing data were spliced using FLASH and fastp software, and the chimeric sequences were filtered and removed to obtain the final Effective Tags. UPARSE software (Version 7.1) was used to cluster all Effective Tags of all samples into OTUs (operational taxa) with 97% similarity, and representative OTUs sequences were selected for species annotation (Edgar, 2013). The OTUs of fungal ITS were annotated using Qiime software (Version 1.9.1) and Unite (v8.2) database (Altschul et al., 1990; Kõljalg et al., 2013). The OTUs of Bacterial 16S rRNA were annotated using the SSUrRNA database of SILVA138.1(Wang et al., 2007; Quast et al., 2012). Fast multiple sequence alignment was performed using MUSCLE (Version 3.8.31) software to obtain phylogenetic relationships of all OTUs representative sequences (Manuel, 2013). Finally, microbial community diversity (Shannon and Simpson index), richness (Chao1 index), Goods Coverage index and beta diversity were estimated by Qiime software (Version 1.9.1) for different time series samples. Shannon and Simpson index was used to analyze the microbial community diversity of tobacco leaves. Chao 1 index and ACE index were used to analyze the community abundance in community samples (Zhang et al., 2019). Goods Coverage index was used to analyze the microbial community coverage of samples (Xie et al., 2021). PCoA and PCA were used to determine differences between community structures. Samples with higher community similarity tended to cluster together. Fungal trophic patterns were analyzed by FUNGuild database (Nguyen et al., 2016). Bacterial metabolic functions were analyzed by PICRUSt database (Langille et al., 2013). Cytoscape 3.9.1 software was used to assess the interactions between microbial communities (Williams et al., 2014; Meng et al., 2017). R software (Version 2.15.3) was used to draw the dilution curve, Rank abundance curve and species accumulation curve.

IBM SPSS Statistics 23 (IBM Corp., New York, United States) was used to compare the differences in alpha-diversity indexes between fungal and bacterial communities (Mao et al., 2020). The mean values were compared, and p ≤ 0.05 was considered statistically significant.

After quality control processing, denoising and chimeric filtering, a total of 1,557,233 fungal sequences and 1,557,862 bacterial sequences were obtained from 24 samples. At 97% similarity level, the fungal sequences were classified into 623 operational taxa (OTUs), and the bacterial sequences were classified into 773 OTUs. When the sequence number of fungi and bacteria reached 50,000 and 60,000, respectively, all sparse curves reached the plateau stage (Figures 1A,B), suggesting that sequencing depth could reflect community structure and diversity. When the Number of samples reached 24, the position of the box chart tended to be flat, indicating sufficient sampling and subsequent data analysis could be carried out (Figures 1C,D).

Diversity (Shannon and Simpson) index, richness (Chao1) index and goods coverage index were used to analyze the fungal and bacterial OTU diversity in the infected and healthy groups at different application periods. The fungal coverage index for healthy and infected tobacco samples was above 0.96 and the bacterial coverage index was above 0.85. It indicated that the sequence detection probability of the samples is high, and the sequencing results can represent the actual microbial community structure in the samples (Table 2).

Among the fungal OTUs, the diversity index and richness index of healthy tobacco leaves were higher than those of diseased leaves before and after the application of the fungicide, but the diversity indices of healthy and diseased leaves were significantly different after applying the fungicide only on day 15, at 4.33 and 1.41, respectively. After 5 days of fungicide application, the diversity and richness of healthy and diseased tobacco leaves decreased to varying degrees. After that, the diversity and richness of healthy tobacco leaves and diseased tobacco leaves continued to rise (Figures 2A–C).

Among the bacterial OTUs, the diversity index of healthy tobacco leaves before application of fungicide was lower than that of diseased leaves, while the richness index was higher than that of diseased leaves, but there was no significant difference. After 5 days of fungicide application, the diversity and richness of healthy and diseased tobacco leaves increased to varying degrees. After 15 days of fungicide application, the diversity index of healthy tobacco leaves was higher than that of diseased leaves, which might be due to the greater effect of copper hydroxide on healthy leaves (Figures 2D–F).

The ITS dataset showed that 48.93% of OTUs were mainly divided into Ascomycota and Basidiomycota. The “other” group in Figure 3 includes all organisms that are not recognized or have a relative abundance of <0.1%. Before copper hydroxide treatment, the phyllosphere microorganisms of healthy and disease tobacco leaves were mainly divided into Ascomycota (66.52 and 92.85%) and Basidiomycota (7.30 and 2.47%). With the application of copper hydroxide, the abundance of Ascomycota in diseased leaves changed little over time, but the abundance of Ascomycota in healthy leaves was reduced by 19.59% compared to before application (Figure 3A). The relative abundance of Basidiomycota in healthy and diseased leaves decreased by 5.88 and 2.28% at 5 days after application, and thereafter increased by 10.4 and 6.4% at 15 days after application (Figure 3B). The “other” group showed a similar pattern in healthy and diseased leaves, with abundance declining at 5 days and peaking at 15 days., when the relative abundance of healthy leaves was 40.04% and diseased leaves only 4.42% (Figure 3C).

At the genus level, the 10 most common fungal genera are shown in Figure 4. The main genera of healthy tobacco leaves and diseased tobacco leaves before fungicide application were Alternaria (38.91 and 84.79%), Boeremia (5.47 and 0.41%), Cladosporium (15.69 and 6.67%), Symmetrospora (5.96 and 2.25%), Epicoccum (1.09 and 0.19%), and Phoma (2.13 and 0.26%). After the application of copper hydroxide, Alternaria in healthy and diseased leaves showed a general downward trend, decreasing by 21.16 and 40.82%, respectively, at 15 days (Figure 4A). The relative abundance of Boeremia did not change much with time in healthy leaves, but increased significantly in diseased leaves and reached a peak of 33.11% at 5 days (Figure 4B). Issatchenkia was not detected in diseased leaves and reached a peak of 9.21% in healthy leaves at 5 days (Figure 4C). The relative abundance of Stagonosporopsis in healthy leaves changed little over time and reached a peak of 9.03% at 15 days in diseased leaves (Figure 4D). The relative abundance of Cladosporium showed a similar pattern in healthy and diseased leaves, decreasing by 14.77 and 5.81% at 5 days, respectively (Figure 4E). The relative abundance of Symmetrospora showed similar trends in healthy and diseased leaves, decreasing by 4.87 and 2.1% at 5 days, respectively (Figure 4F). Pseudeurotium was not detected in diseased leaves and reached a peak of 3.82% at 15 days in healthy leaves (Figure 4G). Thanatephorus was also not detected in diseased leaves, but was detected in healthy leaves at 0 days and 15 days, peaking at 2.43% on 15 days (Figure 4H). Epicoccum reached a peak of 4.12% at 10 days in healthy leaves, but changed little over time in diseased leaves (Figure 4I). Phoma also showed little change over time in diseased leaves, while peaking at 2.51% at 15 days in healthy leaves (Figure 4J). The “other” group showed little change over time in diseased leaves but reached a peak of 47.64% at 15 days in healthy leaves (Figure 4K).

The 16S dataset showed that 37.28% of OTUs were mainly divided into Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria and Fusobacteria. Before copper hydroxide treatment, the phyllosphere microorganisms of healthy and disease tobacco leaves were mainly divided into Proteobacteria (32.9 and 70.56%) and Firmicutes (55.84 and 4.33%) (Figure 5). With the application of copper hydroxide, the relative abundance of Proteobacteria and Firmicutes showed similar patterns in healthy and diseased leaves. On the fifth day, Proteobacteria increased by 38.1 and 28.14%, while Firmicutes decreased by 54.54 and 4.33%, respectively (Figures 5A,B). The relative abundance of Bacteroidetes had no obvious pattern, but was not detected at 15 days (Figure 5C). The relative abundance of Actinobacteria showed a similar pattern in healthy leaves and diseased leaves, decreasing by 0.87 and 2.16%, respectively, at 15 days, when Actinobacteria could not be detected (Figure 5D). Fusobacteria was only detected in healthy leaves at 5 days, while unidentified Bacteria was only detected in diseased leaves at 10 days (Figures 5E,F). The “other” group showed the opposite pattern in healthy and diseased leaves, with a 17.32% increase in abundance in healthy leaves and a 19.43% decrease in abundance in diseased leaves at 15 days (Figure 5G).

At the genus level, the 10 most common bacterial genera are shown in Figure 6. Before copper hydroxide treatment, the phyllosphere microorganisms of healthy and disease tobacco leaves were mainly divided into Pantoea (30.74 and 29.87%), Pseudomonas (0.43 and 13.85%) and Sphingomonas (0 and 16.88%). After the application of copper hydroxide, the relative abundance of Pantoea showed a similar pattern in healthy and diseased leaves, decreasing by 29.87 and 27.27% at 5 days, respectively (Figure 6A). The relative abundance of Ralstonia peaked at 61.9 and 30.30% in both healthy and diseased leaves at 10 days (Figure 6B). The relative abundance of Pseudomonas did not change with time in healthy leaves but reached a peak of 30.30% at 10 days in diseased leaves (Figure 6C). The relative abundance of Sphingomonas changed little over time in healthy leaves and peaked at 26.41% at 5 days in diseased leaves (Figure 6D). The relative abundance of Massilia did not change with time in healthy leaves but reached a peak of 19.05% at 5 days in diseased leaves (Figure 6E). The relative abundance of Erwinia peaked at 3.03 and 9.96% in healthy and diseased leaves at 15 days (Figure 6F). The relative abundance of Methylobacterium was not detected in both healthy and diseased leaves at 0 days, and reached a peak of 4.33% at 15 days in healthy leaves and 6.93% at 5 days in diseased leaves (Figure 6G). unidentified Enterobacteriaceae was not detected in healthy leaves but reached a peak of 2.16% in diseased leaves at 15 days (Figure 6H). Bacteroides was detectable in diseased tobacco leaves only at 0 days, but was detectable in healthy tobacco leaves at 5 and 10 days (Figure 6I). Serratia was detected only in healthy leaves and diseased leaves at 15 days, when the peak values were 0.87 and 1.30% (Figure 6J). The relative abundance of The “other” group did not change much over time in diseased leaves, but reached a peak of 40.69% at 15 days in healthy leaves (Figure 4K).

Principal Co-ordinates Analysis (PCoA) plots were used to determine the spatial distribution of fungal and bacterial communities (Figures 7A,C). In the fungal community, all diseased groups tended to cluster together, suggesting that copper hydroxide has less effect on the phyllosphere microorganism of diseased samples. However, the dispersion degree among healthy samples is large, indicating that copper hydroxide has a great influence on healthy groups (Figure 7B). In the bacterial community, the samples of the diseased group and the healthy group were more discrete (Figure 7D).

The FUNGuild database was used to analyze the ecological function of the fungi in the samples in this study (Figure 8A). Fungal OTUs can be subdivided into nine groups according to nutrient patterns. Before the application of copper hydroxide, pathogen-saprophytic-symbiotic (84.94%) and pathogen-symbiotic (6.67%) were the dominant ecological functions of fungi in the disease group. The fungal dominant ecological functions of the healthy group were Pathotroph-Saprotroph-Symbiotroph (39.14%), Pathotroph-Symbiotroph (15.69%), Pathotroph (8.39%), Pathotroph-Saprotroph (3.30%) and Unassigned (32.66%). After treatment with copper hydroxide, the prediction of ecological function of fungi was changed. With the increase of time, the relative abundance of pathotroph-Saprotroph-Symbiotroph and Pathotroph-Symbiotroph decreased by 40.71 and 4.83%, respectively, in the diseased group, while the relative abundance of Pathotroph increased by 41.31%. The relative abundance of Pathotroph, Saprotroph and Unassigned increased by 6.25, 5.25 and 19.18%, respectively, in the healthy group. Meanwhile, the relative abundance of Pathotroph-Saprotroph-Symbiotroph and Pathotroph-Symbiotroph decreased by 21.09 and 9.77%, respectively.

PICRUSt for functional prediction of KEGG database based on 16S sequencing data, The results showed that 24 samples from 4 different periods contained 6 known biological metabolic pathways at the primary functional level (Figure 8B): “environmental information processing,” “cellular processes,” “genetic information processing,” “human diseases,” “metabolism” and “organic systems.” For the prediction of bacterial function, the gene sequences of the diseased and healthy groups were mainly related to “metabolic” function, followed by “genetic information processing” and “environmental information processing.” Copper hydroxide treatment had little effect on the prediction of bacterial function.

Biolog ECO plates contain carbon sources used by most microorganisms in nature. There are 31 carbon sources, including carbohydrates, carboxylic acids, amino acids, phenolic acids, polymers and amines. The effects of copper hydroxide application on phyllosphere microbial metabolic function were significantly different between the diseased and healthy groups (Figure 9). Before the application of copper hydroxide, the diseased group microorganisms could efficiently metabolize 24 carbon sources, while the metabolic capacity of D-xylose, glucose-1-phosphate, D, L-a-glycerol, L-threonine, α-ketobutyric Acid, α-cyclodextrin and 4-hydroxybenzoic Acid was weak. Healthy group microorganisms can efficiently metabolize 28 carbon sources in addition to D-Xylose, 2-Hydroxy Benzoic Acid and α-Ketobutyric Acid.

The metabolic function of the diseased group was greatly affected after the application of fungicide. After 5 days of copper hydroxide application, the phyllosphere microorganisms in the diseased group increased the metabolism of 11 carbon sources, including D-galactolactone, I-erythritol, 4-hydroxybenzoic Acid, α-cyclodextrin, L-threonine, glycogen, D-glucosamine, Glycine-L-glutamic Acid, Glucose-1-phosphate, α-ketobutyric Acid, D, L-a-glycerol. The microorganisms in the diseased group can efficiently metabolize 29 carbon sources in addition to D-Xylose, 2-Hydroxy Benzoic Acid and α-Ketobutyric Acid. After 10 days of copper hydroxide application, the metabolism capacity of the phyllosphere microorganisms in the diseased group was further improved except for the three carbon sources of 2-Hydroxy Benzoic Acid, L-Phenylalanine and Phenylethyl-amine. After 15 days of copper hydroxide application, the metabolic capacity of the diseased group was further improved, except D-Xylose and 4-Hydroxy Benzoic Acid.

After the application of copper hydroxide, the metabolic function of phyllosphere microorganisms in healthy group was opposite to that in diseased group. After 5 days of copper hydroxide application, the metabolism of 31 carbon sources among the microorganisms in the healthy group was inhibited, and there was no effective carbon source. After 10 days of copper hydroxide application, the metabolic function of phyllosphere microorganisms in healthy group gradually recovered, but there was still no efficient carbon source. After 15 days of copper hydroxide application, the metabolic function of phyllosphere microorganisms in healthy group was further restored. The phyllosphere microorganisms in healthy group can efficient metabolism of β-Methyl-D-Glucoside, D-Mannitol, D-Glucosaminic Acid, L-Asparagine, Glycyl-L-Glutamic Acid, the rest of the carbon metabolism is still suppressed.

Our study showed that Ascomycota and Basidiomycota were the dominant phylum level in both diseased and healthy tobacco leaves in the untreated group. The relative abundance of Ascomycetes was higher in the diseased group than in the healthy group, whereas the relative abundance of Basidiomycetes was lower than that of the healthy group. A study on phyllosphere microbial diversity of boreal forest tree (Populus balsamifera L.) also found a similar phenomenon (Bálint et al., 2015). Another study conducted high-throughput sequencing of the fungal communities of 51 tree species in the tropical rain forest showed that the fungal communities on leaves were also dominated by Ascomycota and Basidiomycetes (Kembel and Mueller, 2014). Before the application of fungicide, Proteobacteria and Firmicutes were the dominant levels in both diseased and healthy tobacco leaves, while there were Bacteroidetes and Actinobacteria with low abundance. Many studies have shown that Proteobacteria is the dominant leaf group of various crops, such as lettuce (Rastogi et al., 2012), grape (Singh et al., 2019), Soybean, Shamrock and Arabidopsis (Delmotte et al., 2009), rice (Knief et al., 2012). The Gram-negative bacteria of Proteobacteria can rely on the QS mechanism to ensure that they compete with other bacteria in the Phyllosphere environment of nutrient deficiency (Müller and Ruppel, 2014). This may be the reason why Proteobacteria occupy a favorable niche and are the only dominant bacteria under the pressure of copper hydroxide.

Our data indicated that the most dominant gensera of fungi with healthy leaves prior to treatment with copper hydroxide were Alternaria and Cladosporium, whereas in diseased leaves it was Alternaria. As expected, Alternaria is the dominant genus among diseased leaves, as Alternaria alternata is the causative agent of tobacco brown spot. However, we also found high relative abundances of Alternaria, Cladosporium, and Boeremia in healthy leaves. Alternaria in healthy tobacco leaves may come from diseased tobacco leaves and cause disease once the environmental conditions are suitable. Similar results were reported by Sun et al. (2023) in tobacco target spot. Cladosporium contains a number of potential plant pathogens that can cause diseases in a variety of crops such as Lettuce (Hunter et al., 2015), tomato (Rivas and Thomas, 2005; Toju et al., 2019), cucumber (Batta, 2004), canola (Idnurme et al., 2021), tobacco (Wang et al., 2014). Boeremia also contains a number of potential plant pathogens that can cause diseases in a variety of crops such as sweet potato (Colmán et al., 2020), Panicle hydrangea (Garibaldi et al., 2018), Walnuts (Wang et al., 2022). However, no pathogenicity has been reported in tobacco. In addition, The study of Chen Y. et al. (2020) showed the potato endophytic fungus Boeremia exigua could significantly inhibit the growth of pathogenic fungi of potato late blight.

Our data indicated that the most dominant genera of bacteria with healthy leaves prior to copper hydroxide treatment were Pantoea and Weissella, whereas in diseased leaves it was Pantoea, Pseudomonas and Sphingomonas. The results showed that all the dominant genera were Proteobacteria. The study of Lv et al. (2012) showed Proteobacteria gram-negative bacteria can regulate the density of interleaf bacterial community through AHLs to occupy a favorable ecological niche. Pantoea was the dominant bacterium in both healthy and infected tobacco leaves before using fungicide, which may be due to its strong adaptability in different environments. Canamas et al. (2007) found Pantoea sp. could accumulate glycine-betaine, ectoine and amino acids in the bacterial cells, which enable the Pantoea sp. to resist environmental stress.

Analysis of alpha diversity in the healthy and diseased groups in this study showed that copper hydroxide treatment increased fungal diversity in both the diseased and healthy groups and there was a significant difference between healthy and diseased leaves at 15 days. Similar results were reported by Peixoto et al. (2021) copper hydroxide can affect soil microbial community and lead to changes in soil microbial composition. The study of Liu et al. (2022) showed that the application of Bordeaux mixture on healthy tobacco increased the diversity of fungal community. This may be due to the selectivity of fungi to copper hydroxide, so that fungicides only inhibit the metabolism of some fungi. The study of Liu et al. (2022) showed that the dominant fungal genera were Fusarium and Cercospora after spraying Bordeaux mixture on healthy tobacco leaves. After copper hydroxide exposure, the relative abundance of Alternaria and Cladosporium decreased in both the diseased and healthy groups, while six fungi genera (Boeremia, Issatchenkia, Stagonosporopsis, Symmetrospora, Epicoccum and Phoma) of the top 10 genera were increased. This may be attributed to a decrease in the abundance of pathogenic fungi, leading to the growth of less competitive microorganisms. Copper hydroxide can reduce the relative abundance of Alternaria and Cladosporium in both healthy and diseased tobacco leaves, which may indicate that copper hydroxide can be used to control tobacco brown spot. Copper hydroxide fungicides had less effect on the microflora structure of diseased leaves than healthy leaves at different times after application. Healthy tobacco leaves have long-lasting protection after the application of fungicides. Therefore, in terms of disease prevention and control, it is recommended that copper preparations be used at the early stage of the disease and when the disease index is low. Epicoccum and Symmetrospora were unique genera found in tobacco leaf phyllosphere originally infected by A. alternata. After the application of copper hydroxide, the relative abundance of Thanatephorus, Epicoccum and Phoma in diseased tobacco leaves did not change much with time, but increased significantly at 10 days in healthy tobacco leaves. This indicates that the prevention of targeted pathogens in healthy tobacco leaves should also pay attention to the prevention of diseases caused by some potential pathogens.

Copper hydroxide treatment increases the relative abundance of Proteobacteria in healthy and diseased tobacco leaves, while decreasing the relative abundance of Firmicutes and Actinobacteria, possibly due to the fact that Firmicutes and Actinobacteria are more sensitive to copper than Proteobacteria. A study on the change of microbial diversity of soil microorganisms under copper stress (Zhang et al., 2020) also found that copper fungicides could change the results of Phyllosphere microbial community, among which Proteobacteria significantly increased and Actinobacteria significantly decreased. After the application of copper hydroxide, the relative abundance of Pantoea decreased significantly at 5 days in both diseased and healthy leaves, indicating that Pantoea is more sensitive to copper ions. Pantoea sp. has a broad range of biological activities, including control of fire blight on fruit trees (Wright et al., 2001), and suppression of several diseases caused by bacteria or fungi on various plants (Braun-Kiewnick et al., 2000; Han et al., 2000; Ortmann and Moerschbacher, 2006). The relative abundance of Pseudomonas and Sphingomonas in the diseased leaves increased on day 5, which may be due to the plant’s ability to use copper to release growth factors to promote the growth of beneficial bacteria. Pseudomonas are primarily considered plant pathogens (Buell et al., 2003; Yu et al., 2013), although some Pseudomonas species are known to inhibit leaf fungal pathogens by producing antibiotics (De Meyer and Hofte, 1997). Most research on Sphingomonas has focused on the production of growth stimulating factors in plants (Adhikari et al., 2001; Enya et al., 2007; Tsavkelova et al., 2007; Khan et al., 2014). The differences within groups were greater than those between groups, which might be due to the variable leaf surface features, such as the junctions of epidermal cell walls, the grooves of leaf veins, stomata, trichosomes and drainers, which were the priority attachment points for bacteria, leading to the uneven distribution of bacteria on the leaf surface (Müller and Ruppel, 2014). After the application of copper hydroxide, the relative abundance of Ralstonia decreased significantly on day 15 in both diseased and healthy leaves, indicating that Ralstonia is more sensitive to copper ions. The study of Chen et al. (2019) showed CuONPs significantly reduced bacterial activity and killed Ralstonia solanacearum at high concentrations.

A variety of compounds such as carbohydrates, amino acids, methanol and other organic compounds are the main sources of carbon and nitrogen for phyllosphere microorganisms. These compounds are often passively leached from the plant interior to 7% of the leaf surface. Although these sources are not abundant, they are sufficient to explain the presence and community structure of microorganisms in the phyllosphere environment (Mercier and Lindow, 2000). The metabolic level of ECO healthy group was significantly reduced after copper hydroxide treatment, which may be due to the characteristics of most phyllosphere microbiome [Vi-able But Nonculturable (VBNC)]. Compared with diseased leaves, healthy leaves have more microbial species, which may result in higher copper stress on metabolic level. Under the influence of copper fungicides, microorganisms will maintain low metabolic activity and last for a long time after entering the VBNC state (Oliver, 2010; Dinu and Bach, 2011). A study by Zhao et al. (2018) showed that copper fungicide could significantly downregulate some intermediates of the tricarboxylic acid (TCA) cycle by 1.4 ~ 2.4 times, indicating that copper could cause carbohydrate metabolism disorder.