Microbial composition of diseased and healthy trees
A total of 3,601,568 (fungi) and 2,240,660 (bacteria) high-quality sequences were generated across all samples after sequence de-noising and quality filtering. The number of fungal communities was less in diseased trees than in healthy trees at all classification levels, while bacteria communities were opposite (Fig. 1a, Table S1). The analysis of the α diversity index of diseased trees and healthy trees at the OTUs level showed that there was no significant difference in microbial community richness and evenness between diseased trees and healthy trees, but the fungal diversity index was a significant difference. In addition, the community richness of healthy and diseased trees was the highest, followed by branches and trunks (Table S2).
All sequences were classified to the fungal domain and assigned to 2830 OTUs across all samples, including 14 phyla, 46 classes, 109 orders, 454 genera and 679 species. Ascomycota (59.80% of the sequence) was the most abundant phylum, followed by Basidiomycota (29.90% of the sequence), Mortierellomycota (2.63% of the sequence), Rozellomycota (1.11% of the sequence), and others (< 1.0% of the sequence) include Mucoromycota, Glomeromycota, Chytridiomycota, Entomophthoromycota, Calcarisporiellomycota, Cercozoa, Kickxellomycota, Olpidiomycota (Fig. 1b, Table S3). All sequences were classified to the bacterial domain and assigned to 6720 OTUs across all samples, including 51 phyla, 114 classes, 315 orders, 565 families, 1280 genera and 2546 species. Proteobacteria (52.87% of the sequence) was the most abundant phylum, followed by Cyanobacteria (17.71% of the sequence), Actinobacteria (10.57% of the sequence), Acidobacteria (5.31% of the sequence), Bacteroidetes (4.14% of the sequence), Firmicutes (2.86% of the sequence), Chloroflexi (2.26% of the sequence), Patescibacteria (1.48% of the sequence) and others (< 1.0% of the sequence) include WPS-2, Verrucomicrobia, Planctomycetes and so on (Fig. 1c, Table S4).
Microbial structure of diseased and healthy trees
In the analysis of the fungal community of diseased and healthy trees, it was found that the number of unique OTUs in branches, trunks and soil of healthy samples was more than that of diseased samples. The soil shared 40.30% of the OTUs (surface soil 37.34% and the deep soil 29.96%) between healthy and diseased samples, followed by trunks and branches. Only 1.13% of OTUs were shared in branches, trunks and soil, in which the soil harbored the most abundant OTUs, followed by branches and trunks (Fig. 2a). In the analysis of the bacterial community of diseased trees and healthy trees, it was found that the number of unique OTUs in branches, trunks and soil of healthy samples was more than that of diseased samples. The trunks shared 63.90% of the OTUs (upper trunk 21.64%, middle trunk 35.51%, lower trunk 41.37%) between healthy and diseased samples, followed by the soil and branches. Only 2.32% of OTUs were shared in branches, trunks and soil, of which OTUs, in which the trunks harbored the most abundant OTUs, followed by soil and branches (Fig. 2b).
PCoA analysis among fungal communities was performed based on Bray-Curtis distance with the first and second axes explaining 21.79 and 16.74% of the variance, respectively (Fig. 3a). PCoA analysis among bacterial communities was performed based on Bray-Curtis distance with the first and second axes explaining 32.79 and 21.3% of the variance, respectively (Fig. 3b). The results showed that the infection of PWN mainly affected the endophytic microbial community of branches and trunks of P. massoniana, but had little effect on the microbial community in soil.
The LEfSe analysis showed that the abundance of some fungal taxa differed between the healthy and diseased samples in the branches (HB or DB), upper trunks (HTU or DTU), middle trunks (HTM or DTM), lower trunks (HTL or DTL), surface soil (HTS or DTS) and deep soil (HS or DS), respectively (LDA > 4.8, p < 0.05) (Fig. 4). In the branches, the class Eurotiomycetes, the orders Capnodiales and Xylariales, the family Sporocadaceae, and the genus Pestalotiopsis were more abundant in the healthy trees, whereas the orders Ophiostomatales and Botryosphaeriaceae; the families Ophiostomataceae and Botryosphaeriaceae, the genera Graphilbum and Diplodia were had a higher abundance in the diseased trees (Fig. 4a). In the upper trunks, the phylum Ascomycota, the classes Sordariomycetes and Eurotiomycetes, the orders Hypocreales and Eurotiales, the families Hypocreaceae and Nectriaceae, and the genus Trichoderma were more abundant in the healthy trees, whereas the classes Saccharomycetes and Agaricomycetes, the orders Saccharomycetales and Polyporales, the family Ganodermataceae, and the genus Candida were had a higher abundance in the diseased trees (Fig. 4b). In the middle trunks, the phylum Basidiomycota, the class Tremellomycetes, and the order Eurotiales were more abundant in the healthy trees, whereas the phylum Ascomycota, the classes Sordariomycetes and Saccharomycetes, the orders Saccharomycetales, Ophiostomatales and Xylariales; the family Ophiostomataceae, and the genus Graphilbum were had a higher abundance in the diseased trees (Fig. 4c). In the lower trunks, the order Hypocreales, the families Nectriaceae, Hypocreaceae and Aspergillaceae; and the genera Fusarium, Trichoderma and Penicillium were more abundant in the healthy trees, whereas the class Saccharomycetes, the orders Saccharomycetales and Ophiostomatales; the family Ophiostomataceae, and the genus Graphilbum were had a higher abundance in the diseased trees (Fig. 4d). In the surface soil, the order Russulales, the family Russulaceae, and the genus Russula were more abundant in the healthy trees, whereas the order Cantharellales, the family Clavicipitaceae, and the genus Membranomyces were had a higher abundance in the diseased trees (Fig. 4e). In the deep soil, the phylum Ascomycota, the order Russulales, the family Russulaceae, and the genus Russula were more abundant in the healthy trees, whereas the phylum Basidiomycota, the class Tremellomycetes, the order Tremellales, the family Trimorphomycetaceae, and the genus Saitozyma were had a higher abundance in the diseased trees (Fig. 4f).
Analysis of the abundance of fungal taxa in different parts of healthy trees are as follows. The class Dothideomycetes, the orders Xylariales, Botryosphaeriaceae and Chaetothyriales; the families Sporocadaceae, Teratosphaeriaceae, Botryosphaeriaceae, Cladosporium and Mycosphaerellaceae; the genera Pestalotiopsis, Devriesia, Diplodia and Cladosporium were more abundant in the branches. The family Hypocreaceae and the genus Trichoderma were more abundant in the upper trunks. The class Saccharomycetes, the orders Saccharomycetales, Pleosporales and Cantharellales; and the genus Candida were more abundant in the middle trunks. The phylum Ascomycota, the class Sordariomycetes, the order Hypocreales, the family Nectriaceae, and the genus Fusarium were more abundant in the lower trunks. The phyla Basidiomycota, the class Agaricomycetes, the order Atheliales, the families Clavicipitaceae, Atheliaceae and Herpotrichiellaceae; the genera Cladophialophora and Tylospora were more abundant in the surface soil. The phyla Mortierellomycota, the classes Mortierellomyceres and Leotiomycetes; the orders Tremellales, Mortierellales and Helotiales; the families Trimorphomycetaceae and Mortierellaceae; the genera Saitozyma and Mortierella were more abundant in the deep soil (Fig. S1a).
Analysis of the abundance of fungal taxa in different parts of disease trees are as follows. The order Botryosphaeriales, the families Botryosphaeriaceae, Bionecteiaceae and Chrysozymaceae; the genera Diplodia, Capronia, Ophiostoma, Hamamotoa and Clonostachys were more abundant in the branches. The classes Agaricomycetes and Saccharomycetes, the orders Saccharomycetales and Polyporales, the families Ganodermataceae and Pichiaceae, and the genus Kuraishia were more abundant in the upper trunks. The phyla Ascomycota, the class Sordariomycetes, the orders Hypocreales and Xylariales; the families Nectriaceae, Hypocreaceae and Sporocadaceae; the genera Xenoacremonium, Trichoderma, Trigonosporomyces, Fusarium and Neopestalotiopsis were more abundant in the middle trunks. The orders Ophiostomatales, Russulales and Eurotiales; the families Ophiostomataceae and Trichocomaceae, and the genus Talaromyces were more abundant in the lower trunks. The phyla Basidiomycota and Mortierellomycota; the classes Mortierellomycetes and Cystobasidiomycetes; the orders Mortierellales, Chaetothyriales and Filobasidiales; the families Mortierellaceae, Aspergillaceae, Russulaceae, Clavicipitaceae, Herpotrichiellaceae and Teratosphaeriaceae; the genera Penicillium, Mortierella, Pestalotiopsis and Devriesia were more abundant in the surface soil. The classes Tremellomycetes and Leotiomycetes; the orders Tremellales, Helotiales and Trichosporonales; the families Trimorphomycetaceae and Trichosporonaceae; the genera Saitozyma and Apiotrichum were more abundant in the deep soil (Fig. S1b).
The LEfSe analysis showed that the abundance of some bacterial taxa differed between the healthy and diseased samples in the branches (HB or DB), upper trunks (HTU or DTU), middle trunks (HTM or DTM), lower trunks (HTL or DTL), surface soil (HTS or DTS) and deep soil (HS or DS), respectively (LDA > 4.8, p < 0.05) (Fig. 5). In the branches, the phylum Cyanobacteria, the class Oxyphotobacteria and the order Chloroplast were more abundant in the healthy tree, whereas the phyla Proteobacteria and Actinobacteria, the classes Gammaproteobacteria, Actinobacteria and Alphaproteobacteria; the orders Xanthomonadales and Enterobacteriales; the families Xanthomonadaceae, Rhodanobacteraceae and Enterobacteriaceae; the genus Pseudoxanthomonas and Dyella were had a higher abundance in the diseased tree (Fig. 5a). In the upper trunks, the phylum Cyanobacteria, the class Oxyphotobacteria, the order Chloroplast, and the genera Serratia were more abundant in the healthy trees, whereas the phylum Proteobacteria, the class Gammaproteobacteria, the order Enterobacteriales, the family Enterobacteriaceae, and the genera Pantoea were had a higher abundance in the diseased trees (Fig. 5b). In the middle trunks, the phylum Cyanobacteria, the class Oxyphotobacteria, the orders Chloroplast and Enterobacteriales, and the family Enterobacteriaceae were more abundant in the healthy trees, whereas the order Xanthomonadales, the family Xanthomonadaceae, and the genera Pseudoxanthomonas were had a higher abundance in the diseased trees (Fig. 5c). In the lower trunks, the phylum Cyanobacteria, the class Oxyphotobacteria, and the order Chloroplastwere more abundant in the healthy trees, whereas the class Bacteroidia was had a higher abundance in the diseased trees (Fig. 5d).
Analysis of the abundance of bacterial taxa in different parts of the healthy trees are as follows. The phylum Cyanobacteria, the class Oxyphotobacteria, the order Chloroplast were more abundant in the branches. The genus Serratia was more abundant in the upper trunks. The phylum Proteobacteria, the class Gammaproteobacteria, the order Enterobacteriales and the family Enterobacteriaceae were more abundant in the middle trunks. The order Betaproteobacteriales, the family Burkholderiaceae, and the genus Burkholderia-Caballeronia-Paraburkholderia were more abundant in the lower trunks. The phyla Actinobacteria, the classes Alphaproteobacteria and Actinobacteria were more abundant in the surface soil. The phyla Acidobacteria and the class Acidobacteriia were more abundant in the deep soil (Fig. S1c). Analysis of the abundance of bacterial taxa in different parts of the disease trees. The order Xanthomonadales, the family Rhodanobacteraceae, and the genus Dyella were more abundant in the branches. The phyla Proteobacteria, the class Gammaproteobacteria, the order Enterobacteriales, the family Enterobacteriaceae, and the genus Pantoea were more abundant in the upper trunks. The family Xanthomonadaceae and the genus Pseudoxanthomonas were more abundant in the middle trunks. The orders Ophiostomatales, Russulales and Eurotiales; the families Ophiostomataceae and Trichocomaceae, and the genus Talaromyces were more abundant in the lower trunks. The class Alphaproteobacteria was more abundant in the surface soil. The phyla Actinobacteria and the class Actinobacteria were more abundant in the deep soil (Fig. S1b).
Microbial correlation of diseased and healthy trees
Fungal correlation results showed there were 334 strong taxon–taxon correlations in healthy trees, positive correlation demonstrated a double (218 vs. 116, ratio = 1.88) increase in the number of negative correlations among them. There were 707 strong taxon–taxon correlations in diseased trees, positive correlation demonstrated an equal (406 vs. 301, ratio = 1.35) number of negative correlations among them. At the phylum level, Ascomycota has the most correlations degrees (cd) (778) in both type of trees. At the class level, Microbotryomycetes had significant correlation only in healthy trees, however Geminibasidiomycetes, Umbelopsidomycetes and Wallemiomycetes were closed in diseased trees. Interestingly, although both groups contain Agaricomycetes and Sordariomycetes, the degrees in healthy trees were an obviously increase in diseased trees, they are 381:87 and 314:118 respectively (Fig. 6a). Among the diseased trees, some genera with the highest abundance were Membranomyces, Oidiodendron and Ganoderma. Also, some genera such as Membranomyces, Ganoderma, Tomentella, Menispora, Ophiostoma, Hamamotoa, Graphilbum, Xenoacremonium, Cytospora, Clonostachys and Entomocorticium were only exist in diseased trees. Among the healthy trees, some genera with the highest abundance were Candida (27), Geminibasidium (23) and Mortierella (23). Also, there were some genera such as Geminibasidium, Bifiguratus, Fusarium, Cladosporium, Lasiodiplodia, Tylospora, Phialemoniopsis, Umbelopsis, Paraconiothyrium, Capnobotryella, Neopestalotiopsis, Pestalotiopsis, Catenulostroma and Wallemia. Surprisingly, although Apiotrichum, Devriesia, Kuraishia, Penicillium and Trichoderma have existed both diseased and healthy trees, the abundance of Apiotrichum, Devriesia, Kuraishia and Penicillium in diseased trees were triple to fivefold as in healthy trees, the abundance of Trichoderma in healthy trees were triple as in diseased trees.
Bacterial correlation results showed there were 637 strong taxon–taxon correlations in healthy trees, positive correlation demonstrated a fivefold (535 vs. 102, ratio = 5.25) increase in the number of negative correlations (absolute value of Spearman correlation > 0.5 and false discovery rate-corrected p < 0.05) among them. There were 866 strong taxon–taxon correlations in diseased trees, positive correlation demonstrated equal (465 vs. 401, ratio = 1.01) the number of negative correlation (absolute value of Spearman correlation > 0.5 and false discovery rate-corrected p < 0.05) among them. At the phylum level, Proteobacteria has the most degrees in both healthy (641) and diseased trees (805). At the class level, Ktedonobacteria exists in healthy trees, while Bacteroidia exists only in diseased trees. Interestingly, Oxyphotobacteria only owned one node in the figure, but the size in healthy trees was tenfold (30 vs. 3, ratio = 10) bigger than it in diseased trees (Fig. 6b). The bacteria group displayed a co-occurrence network with a strong positive correlation among genera. Among the diseased trees, some genera with the highest abundance were Acidothermus, Enterobacter and Pseudomonas. Also, some genera such as Erwinia, Fluviicola, Novosphingobium, Gryllotalpicola, Nocardioides, Terriglobus, Acidipila, Curtobacterium, Chitinophaga, Lactobacillus and Edaphobacter were only exist in diseased trees. Among the healthy trees, some genera with the highest abundance were Candidatus (34), Ralstonia (34) and Rhodococcus (34). Also, some genera such as Rhodococcus, Occallatibacter, Kosakonia, Brevundimonas, Massilia, Silvimonas, Serratia and Stenotrophomonas were only exist in diseased trees. Surprisingly, although Dyella and Pantoea have existed between diseased and healthy trees, the abundance in diseased trees was twice as in healthy trees.
We constructed the correlation model of bacteria and fungi in healthy trees and diseased trees by SPLS. The results showed in healthy trees, Trichoderma and Stenotrophomonas are positively correlated and have the highest correlation (the SPLS coefficient is 4.08), and Fusarium and Pantoea are negatively correlated and have the highest correlation (the SPLS coefficient is − 1.96). However, In the disease trees, Candida and Pantoea are positively correlated and have the highest correlation (the SPLS coefficient is 0.75), and Saitozyma and Pseudoxanthomonas are negatively correlated and have the highest correlation (the SPLS coefficient is − 0.78). In addition, we found that bacteria with high abundance in healthy trees as Kosakonia, Brevundimonas and Serratia were positively correlated with fungi Cutaneotrichosporon (the SPLS coefficients are 0.38, 0.33 and 0.27), while they were negatively related with fungi Russula (the SPLS coefficients are − 0.14, − 0.19 and − 0.23) (Fig. 7a). Bacteria that were more abundant in diseased trees as Erwinia, Fluviicola, Novosphingobium, Gryllotalpicola, Nocardioides, Lactobacillus and Dyella were positively correlated with fungi Graphilbum (the SPLS coefficients are 0.21, 0.04, 0.05, 0.03, 0.05, 0.14 and 0.39), while they were negatively related with fungi Saitozyma (the SPLS coefficients are − 0.47, − 0.06, − 0.10, − 0.06, − 0.08, − 0.22 and − 0.71) (Fig. 7b). They are important candidate microorganisms involved in the pathological mechanism of PWN.
Microbial function of healthy and diseased trees
There were 649 OTUs (22.9%) matched in the FUNGuild analysis for the predicted resource utilization function of fungi. These OTUs were assigned to 20 functional guilds. Overall, the guild animal pathogen had the highest abundance (24.8%), followed by endophyte (16.5%), plant-pathogen (14.9%), ectomycorrhizal (13.7%), fungal parasite (9.2%), wood saprotroph (6.5%), ericoid mycorrhizal (3.1%), soil saprotroph (2.2%), dung saprotroph (1.7%), plant saprotroph (1.5%), epiphyte (1.5%), animal endosymbiont (1.5%), arbuscular mycorrhizal (0.9%), orchid mycorrhizal (0.5%), leaf saprotroph (0.5%), bryophyte parasite (0.3%), litter saprotroph (0.2%), lichenized (0.2%), animal parasite (0.2%) and algal parasite (0.2%) (Fig. 8a). In healthy trees, animal pathogen, plant pathogen and saprotroph are the top three functional guilds. Among them, animal pathogen (46.39%) is mainly concentrated in the lower trunk, plant pathogen (20.39%) is mainly concentrated in the branches, and saprotroph (69.82%) is mainly concentrated in the upper trunk. However, plant pathogen, parasite and saprotroph are the top three functional guilds of diseased trees. Among them, plant pathogen (29.12%) is mainly concentrated in the lower trunk, parasite (36.52%) is mainly concentrated in the branches, and saprotroph (67.88%) is mainly concentrated in the upper trunk. Interestingly, parasitic fungus in soil of diseased trees (16.43%) is significantly higher than soil of healthy trees (6.49%), and the endophyte in soil of diseased trees (5.08%) is significantly less than soil of healthy trees (0.92%).
The COG analysis was utilized for the predication of resource utilization function of bacteria. The functions of COG were assigned to 23 functional guilds. Among them, the guild amino acid transport and metabolism had the highest abundance (10.1%), followed by energy production and conversion (7.3%), cell wall/membrane/envelope biogenesis (6.6%), translation, ribosomal structure and biogenesis (6.6%), inorganic ion transport and metabolism (6.4%), carbohydrate transport and metabolism (6.0%), transcription (5.7%), replication, recombination and repair (4.5%), posttranslational modification, protein turnover, chaperones (4.5%), coenzyme transport and metabolism (4.4%), signal transduction mechanisms (4.1%), lipid transport and metabolism (3.7%), nucleotide transport and metabolism (2.9%), intracellular trafficking, secretion, and vesicular transport (2.1%), secondary metabolites biosynthesis, transport and catabolism (1.9%), defense mechanisms (1.7%), cell motility (1.3%), cell cycle control, cell division, chromosome partitioning (1.0%), and others (< 1.0%) (Fig. 8b).
The results of PICRUSt2 analysis for the predicted function showed the OTUs were assigned to six functional groups and 52 sub-groups. The top functional annotations among them were the group Hydrolases (34.72%), followed by transferases (23.35%), oxidoreductases (23.34%), isomerases (6.94%), ligases (6.03%) and lyases (5.61%). The LEfSe analysis showed that the abundance of some functional groups differed in abundance between the diseased and healthy samples in the branches, trunks, and soil, respectively (LDA > 3.0, p < 0.05). In the branches, the abundance of Acting on carbon-nitrogen bonds other than peptide bonds and the acyltransferases were higher in healthy branches, whereas the glycosyltransferases had a higher abundance in diseased branches (Fig. 9a). In the upper trunks, the abundance of acting on the CH-OH group of donors, intramolecular oxidoreductases, acting on NADH or NADPH, acting on carbon-nitrogen bonds other than peptide bonds, carbon-oxygen lyases, forming carbon-sulfur bonds and acting on the aldehyde or oxo group of donors was higher in healthy upper trunks, whereas the intramolecular transferases, acting on ester bonds, acting on acid anhydrides and transferring phosphorus-containing groups had a higher abundance in diseased upper trunks (Fig. 9b). In the middle trunks, the abundance of acting on the CH-OH group of donors and acting on paired donors with incorporation or reduction of molecular oxygen was higher in healthy middle trunks, whereas the acting on acid anhydrides and transferring phosphorus-containing groups had a higher abundance in diseased middle trunks (Fig. 9c). In the lower trunks, acting on the CH-OH group of donors, acting on paired donors with incorporation or reduction of molecular oxygen, acting on NADH or NADPH, carbon-oxygen lyases, acting on the aldehyde or oxo group of donors and acting on carbon nitrogen bonds other than peptide bonds were higher in healthy trunks, whereas the forming carbon-oxygen bonds, acting on acid anhydrides and transferring phosphorus-containing groups had a higher abundance in diseased trunks (Fig. 9d).