Fungal and bacterial pathogenic co-infections mainly lead to the assembly of microbial community in tobacco stems

2.1 Study site and soil sampling

The study was conducted in the tobacco field of Ji Ling village, Ma Shui Town, Leiyang city, Hunan Province, China. This region experiences a subtropical monsoon humid climate with an average annual temperature of 17.9°C, 1,640 h of sunshine, and an average annual rainfall of 1,337 mm. The soil type is predominantly loam, according to the FAO classification [37]. Fieldwork was performed in late July 2022. Since 2003, the fields in this area have followed a tobacco–rice rotation cropping system. The experimental field covered an area of over 667 m² and was managed using uniform agricultural practices. For pathogen inoculation, tobacco plants were exposed to R. solanacearum and N. falciformis. Both pathogens were inoculated via root dipping. A bacterial suspension of R. solanacearum was prepared at a concentration of 10⁸ CFU/mL and applied to the tobacco plants’ root systems. N. falciformis was inoculated by introducing a spore suspension at a concentration of 10⁶ spores/mL into the root zone. Inoculations were performed simultaneously to ensure uniform exposure to both pathogens. For endospheric sample preparation, tobacco plants’ stems exhibiting symptoms of co-infection by R. solanacearum and N. falciformis were selected based on disease (IR) signs observed in the field. An equal number of healthy tobacco plants stems (HR), showing no signs of infection, were also collected for comparison. All tobacco stems were washed sequentially with 75% ethanol, 2.5% sodium hypochlorite, and sterile water [6]. The stems were then cut into small pieces, homogenized in a mortar with PBS, and the mixture was transferred to centrifuge tubes. After standing for 30 min, the homogenate was centrifuged to remove the supernatant [7]. The resulting cell pellets (endophytic samples) were stored at −80°C until DNA extraction.

2.2 DNA extraction and amplicon sequencing

A total of 16 endospheric samples were processed consisting of both healthy and infected tobacco stems, with eight samples from each group. DNA extraction was performed using the FastDNA^TM SPIN Kit (MP Biomedicals) following the manufacturer’s protocol. The quantity and quality of the extracted DNA were evaluated using a NanoDrop Spectrophotometer (Nano-100, Aosheng Instrument Co. Ltd), ensuring that the A260/A280 ratios fell between 1.8 and 2.0, indicating good-quality DNA suitable for downstream applications. For bacterial amplicon sequencing, the V5–V6 region of the 16S ribosomal RNA (rRNA) gene was targeted to avoid amplification of chloroplast DNA. Universal primers 799F (5′-AACMGGATTAGATACCCKG-3′) and 1115R (5′-AGGGTTGCGCTCGTTG-3′) were used for amplification [6]. For fungal amplicon sequencing, the internal transcribed spacer (ITS) regions were amplified using the primers 5.8F (5′-AACTTTYRRCAAYGGATCWCT-3′) and 4R (5′-AGCCTCCGCTTATTGATATGCTTAART-3′) [38]. To facilitate high-throughput sequencing, 12 bp unique barcodes were added to the 5′-ends of both forward and reverse primers for sample identification. The PCR reactions were performed in a 50 µL reaction mixture containing 1.5 µL of dNTP mixture, 0.5 µL Taq DNA polymerase (TaKaRa, Beijing, China), 5 µL of 10× PCR buffer, 1.5 µL of each 10 µM primer, and 20–30 ng of DNA template. The thermal cycling conditions were as follows: initial denaturation at 94°C for 1 min, followed by 30 cycles of 94°C for 20 s, 57°C for 25 s, and 68°C for 45 s. A final extension at 72°C for 10 min was followed by a hold at 4°C. The amplification products were verified via gel electrophoresis to ensure successful amplification. Following amplification, the PCR products were purified using the Agencourt AMPure XP beads (Beckman Coulter, USA) to remove primer dimers and non-specific amplification products. The purified amplicons were then quantified using the Qubit 3.0 fluorometer (Thermo Fisher Scientific, USA) with the Qubit dsDNA HS Assay Kit. Equal amounts of purified amplicons from each sample were pooled, and the pool was subjected to quality control using the Agilent 2100 Bioanalyzer (Agilent Technologies, USA) to confirm the correct amplicon size distribution. Sequencing was performed using the Illumina HiSeq 2500 platform (Illumina, USA) with paired-end 250-bp reads. This high-throughput sequencing method ensured the comprehensive profiling of the microbial communities. All sequencing was conducted by Magigene Biotechnology Co., Ltd (Guangzhou, China), with the sequencing data processed and analyzed using standard bioinformatics pipelines.

2.3 Sequence processing

A total of 32 samples, including 16 bacterial community samples and 16 fungal community samples, were analyzed. All raw reads from the 16S rRNA and ITS regions were uploaded to a comprehensive sequence analysis pipeline, which integrates multiple bioinformatics tools [31]. This pipeline, hosted on a private network, facilitates internal data processing and analysis. Initially, the raw reads were assigned to samples based on their barcodes using the “Detected barcodes” tool, followed by trimming of the barcode sequences [39]. Subsequently, the FLASH tool was employed to merge forward and reverse reads. The merged reads, devoid of ambiguous bases, were filtered using the Btrim tool [40]. For chimera checking, the Greengenes database was used for bacterial communities [41], and the ITS RefSeq database was used for fungal communities [42]. Singletons were retained to preserve rare species, and sequences were clustered into zero-radius operational taxonomic units (zOTUs) at a 97% similarity threshold using Unoise. For ITS gene sequences, the ITSx tool was utilized to identify and extract the ITS regions. Following the generation of OTU tables, reads were randomly resampled to 100,000 sequences per bacterial sample and 25,000 sequences per fungal sample. The Ribosomal Database Project classifier was used to assign bacterial and fungal OTUs with the Greengenes ribosomal database and UNITE database, respectively, ensuring confidence values greater than 0.8.

2.4 Network analysis using random matrix theory (RMT) and evaluation of community stability

RMT was utilized to construct bacterial community networks based on Spearman correlations, leveraging an open-access analysis pipeline available at https://inap.denglab.org.cn/ [43]. Spearman correlation coefficients were filtered using thresholds of r > 0.91 for bacterial communities and r > 0.85 for fungal communities, with a false discovery rate of <0.05. These stringent thresholds were chosen to ensure the robustness and reliability of the network connections. Applying consistent thresholds across all treatments allowed for direct comparisons of network properties, ensuring that any observed differences were attributable to treatment effects rather than variations in network construction parameters. The resulting networks were visualized using Gephi (version 0.9.2). To characterize the topological structure of the microbial ecological networks (MENs), various indices were measured, including the number of nodes and links, power-law fitting of node degrees, average degree (avgK), average clustering coefficient (avgCC), and modularity. Synergistic interactions were defined as positive correlations (r > 0.91 for bacteria and r > 0.85 for fungi), where the presence of one microorganism enhanced the growth or activity of another, suggesting a cooperative relationship. Antagonistic interactions were defined as negative correlations (r < −0.91 for bacteria and r < −0.85 for fungi), where the presence of one microorganism inhibited the growth or activity of another, indicating competition or inhibition. Statistical significance of these interactions was assessed by comparing the observed correlation coefficients to the null distribution generated by randomly rewiring the network 100 times using the Maslov–Sneppen approach [44]. A correlation coefficient was considered statistically significant if it exceeded the 95th percentile of the null distribution. For each observed MEN, the Maslov–Sneppen approach was employed to generate 100 randomly rewired networks to compare against the observed network properties.

Robustness indices were used to assess the resistance of the microbial community. The evaluation process involved two key steps: (i) calculation of abundance-weighted mean interaction strength (wMIS _i ). The wMIS _i for node i was calculated using the following equation:

where b _j is the relative abundance of species j and s _ij is the association strength between species i and j, measured by Pearson correlation coefficient. (ii) Network robustness assessment. Nodes with wMIS _i values of 0 were removed from the network. The fraction of the remaining nodes after this elimination process was reported as the network robustness. By following these steps, the robustness of the microbial network was quantitatively evaluated, providing insights into the community’s resistance to disturbances and its overall stability.

2.5 Statistical analysis

All statistical analyses were performed using the analysis pipeline available at https://dmap.denglab.org.cn/. Alpha diversity, representing within-sample diversity, was evaluated using indices such as Shannon, Simpson, and Chao1. Beta diversity, which measures between-sample diversity, was analyzed using Bray–Curtis dissimilarity and UniFrac distances. To assess community dissimilarity, we employed multiple methods: multi-response permutation procedure, analysis of similarities, and permutational multivariate analysis of variance. Prior to conducting these analyses, we confirmed that soil physicochemical variables and diversity indices followed a normal distribution. These analyses provide insights into the diversity and compositional differences within and between microbial communities. To pinpoint specific differences between group means, we utilized both the least significant difference test and Tukey’s post-hoc test. The structural features of the observed networks were compared to those of random networks using a one-sample Student’s t-test, ensuring a robust statistical framework for interpreting network differences.

Moreover, to quantify the niche width of microbial communities, we utilized Levins’ niche breadth (B) index, as described by Logares et al. [45] and Jiao et al. [46]. The index is calculated using the following formula:

where p _i is the proportion of individuals using resource i. This method allows for the assessment of the range of conditions and resources that microbial species can utilize, providing insights into their ecological versatility and adaptability.

3.1 Substantial shifts in pathogenic bacteria and fungi under infected conditions

In the comparison of microbial community composition between healthy (HR) and infected stems (IR), significant shifts in the relative abundance of key pathogenic bacteria were observed. R. solanacearum, a primary bacterial pathogen, exhibited a dramatic increase in abundance (P < 0.001), constituting 56.97% of the bacterial community in infected stems, compared to a minimal presence in healthy stems (1.44%). This substantial rise underscores R. solanacearum’s dominant role in the microbial community under disease conditions. Similarly, N. falciformis another fungal pathogen associated with plant diseases, showed a marked increase in infected stems (P < 0.001), rising from 1.33% in healthy samples to 13.36% in infected samples (Figure 1). This suggests that R. solanacearum and N. falciformis gains a competitive advantage or thrives under the altered environmental conditions present in infected stems.

Figure 1

Changes in pathogenic microbiome composition. The left panel illustrates the composition changes and response ratios of the fungal pathogens N. falciformis and Sarocladium kiliense. The right panel shows the composition changes and response ratios of the bacterial pathogen R. solanacearum. Statistical significance is indicated as follows: P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

3.2 Impact of infection on microbial diversity and community structure in plant stems

Using a 97% similarity threshold, our analysis identified 10,044 bacterial zOTUs and 1,115 fungal zOTUs. Notably, the Chao1 index and observed richness for fungal communities were significantly higher in healthy stems (Chao1: 262, richness: 222) compared to infected stems (Chao1: 214, richness: 167), whereas no significant differences were observed between healthy (Chao1: 1,887, richness: 1,729) and infected (Chao1: 1,766, richness: 1,643) stems in the bacterial community (Figure S2, ANOVA, P  <  0.005). The Venn diagram analysis further illustrated the unique and shared OTUs within the bacterial and fungal communities of healthy and infected stems. In the bacterial community, healthy stems contained 4,090 unique OTUs, while infected stems had 3,159 unique OTUs, with 2,795 OTUs shared between both conditions (Figure S3(a)). Similarly, in the fungal community, healthy stems harbored 489 unique OTUs, and infected stems contained 400 unique OTUs, with 226 OTUs common to both (Figure S3(b)). These findings underscore the distinct microbial compositions associated with health and disease states, as well as the presence of microbial taxa that persist regardless of infection status.

Ordination analyses, visualized through non-metric multidimensional scaling (NMDS) based on Bray–Curtis dissimilarity and detrended correspondence analysis (DCA) plots, revealed clear differences in microbial community structures between HR and IR. In the NMDS plots (Figure S4(a) and (c)), the separation between microbial communities in HR (blue triangles) and IR (red circles) was pronounced, with distinct clustering patterns and low stress values (stress = 0.084 and 0.001), indicating a strong fit for the data. These clusters suggest that infection significantly alters microbial composition, leading to the emergence of distinct communities in infected samples compared to healthy ones. The DCA plots (Figure S4(b) and (d)) reinforced these observations, showing a marked divergence between HR and IR microbial communities along the primary ordination axes. The spread and separation of data points in these plots highlight the extent of compositional shifts driven by infection. Further analysis using dissimilarity tests revealed significant differences between healthy and infected soil (Tables S1 and S2). Overall, these results indicate that infection not only impacts microbial diversity but also induces substantial changes in community structure, resulting in distinctly different microbial ecosystems in infected stems compared to healthy ones.

3.3 Shifts in microbial community composition and structure between healthy and infected stems

The comparison of microbial community composition between HR and IR at both phylum and genus levels reveals significant shifts in microbial taxa associated with infection. At the phylum level, the bacterial community in HR is predominantly composed of Pseudomonadota, followed by Bacteroidota and Actinomycetota (Figure 2(a) and (c)). However, in IR, there is a noticeable shift, with an increase in the relative abundance of Pseudomonadota, while other phyla such as Bacteroidota and Actinomycetota see a reduction. In the fungal community, Ascomycota remains dominant in both HR and IR, but shifts in the relative abundance of other phyla, such as Mortierellomycota and Chytridiomycota, are observed. At the genus level, Ralstonia, a known pathogen, shows a significant increase in IR, reflecting its critical role in disease development (Figure 2(b) and (d)). Other genera, such as Stenotrophomonas and Pseudomonas, also see an increase in infected samples, while the abundance of beneficial or neutral genera decreases. In the fungal community, genera such as Neocosmospora and Sarocladium are more abundant in IR, highlighting their involvement in the infection process, whereas other genera like Mortierella and Fusarium decrease. These findings illustrate how infection alters the microbial community structure, promoting the proliferation of pathogenic taxa while suppressing non-pathogenic or beneficial microbes, likely contributing to disease symptoms and reduced plant health in infected stems.

Figure 2

Composition changes in pathogenic microbiome at phylum and genus levels. (a) and (c) Relative abundance of bacterial and fungal communities at the phylum level in healthy stem (HR) and infected stem (IR). (b) and (d) Composition changes at the genus level for bacterial and fungal communities in HR and IR.

Further analysis using Random Forest and linear discriminant analysis effect size revealed significant differences in the microbial communities of HR and IR. In the bacterial communities (Figure 3(a)), key genera such as Acidibacterium, Ralstonia, and Stenotrophomonas were identified as important discriminators between HR and IR. Notably, Ralstonia exhibited higher importance in IR, underscoring its prominent role in the disease state. The fungal communities also displayed distinct differences, with genera such as Cylindrocarpon, Neocosmospora, and Sarocladium being more prominent in IR, further highlighting the shifts in microbial populations due to infection (Figure 3(b)). The cladograms provide a phylogenetic perspective, illustrating the differential abundance of bacterial and fungal taxa in HR and IR (Figure 3(c) and (d)). Red and green branches represent taxa significantly more abundant in IR and HR, respectively, visually demonstrating the impact of infection on microbial community structure. The enrichment of specific pathogenic taxa in infected stems likely contributes to disease progression and severity. Overall, these analyses highlight the profound changes in microbial communities associated with stem health, where infection leads to the dominance of pathogenic bacteria and fungi.

Figure 3

Biomarker analysis and phylogenetic distribution of microbial communities. (a) and (b) Top 21 bacterial and fungal zOTUs identified as biomarkers in healthy stem (HR) and infected stem (IR) using the Random Forest model. These biomarkers highlight significant differences in microbial communities between the two conditions. (c) and (d) Cladograms depicting the phylogenetic distribution of the most differentially abundant taxa in HR and IR for bacterial and fungal communities, respectively. Each circle’s diameter corresponds to the relative abundance of the taxa, with different colors indicating the most differentially abundant taxa. The circles represent phylogenetic levels from domain to genus, illustrating the hierarchical relationships and the shifts in community composition due to infection.

3.4 Infection-induced shifts in niche width

The analysis of niche width between healthy stems and infected stems reveals significant ecological shifts driven by infection. Bacterial communities in infected stems exhibit a markedly narrower niche width compared to those in healthy stems, with a statistically significant reduction (P < 0.01) (Figure 4a). This reduction suggests that bacterial communities in IR are confined to fewer ecological niches, indicating a decrease in overall diversity and potential functional redundancy. Similarly, fungal communities also show a substantial decrease in niche width in IR compared to HR (P < 0.001) (Figure 4b), highlighting the impact of infection in reducing ecological diversity within the fungal microbiome.

Figure 4

Niche breadth analysis of microbial communities. (a) Niche width of bacterial communities in healthy stem (HR) and infected stem (IR). The bar plot indicates a significant reduction in niche width for bacterial communities in IR compared to HR (P < 0.01**). (b) Niche width of fungal communities in HR and IR. The bar plot shows a significant decrease in niche width for fungal communities in IR compared to HR (P < 0.001***). (c) Mean values of relative abundance for various bacterial genera in HR and IR. Significant differences in abundance between HR and IR are indicated, highlighting the impact of infection on specific bacterial taxa (P < 0.05*, P < 0.01**, P < 0.001***). (d) Mean values of relative abundance for various fungal genera in HR and IR. The bar plot reveals significant changes in the abundance of certain fungal genera due to infection (P < 0.05*, P < 0.01**, P < 0.001***).

Further analysis of niche width at the genus level reveals additional insights into the ecological shifts caused by infection (Figure 4(c) and (d)). For example, in bacterial communities, genera such as Ralstonia show a significant increase in niche width in IR, indicating that these pathogens are adapting to and thriving within the narrower ecological niches imposed by disease conditions (Figure 4(c)). In contrast, the niche width of beneficial or neutral bacterial genera is reduced in IR, suggesting a constriction of their ecological roles within the infected environment. Similarly, in fungal communities, pathogenic genera like Neocosmospora demonstrate a wider niche width in IR, reinforcing the notion that infection fosters a more specialized and less diverse microbial community (Figure 4(d)). Collectively, these findings indicate that infection drives substantial reorganization of both bacterial and fungal communities, characterized by a loss of niche diversity and the rise of pathogenic taxa. This shift is likely a key factor contributing to the observed disease symptoms and underscores the critical role of microbial community dynamics in plant health and disease progression.

3.5 Infection-induced departure from neutrality in microbial community structure

The neutral community model (NST) was applied to assess bacterial and fungal communities across different conditions, including all samples, healthy stems, and infected stems. The NST analysis revealed distinct patterns in the adherence of bacterial and fungal communities to neutral theory (Figure 5). Bacterial communities displayed a broader distribution of points (Figure 5(a)), indicating a stronger alignment with neutral theory, where a greater proportion of species exhibit behaviors consistent with neutral processes (R ² = 0.437). In contrast, fungal communities showed a loose distribution (R ² = 0.302), suggesting that while some species follow neutral theory, others may be influenced by additional non-neutral factors (Figure 5(d)). When focusing on healthy stems, bacterial communities continued to closely follow neutral patterns (R ² = 0.395), reinforcing the notion that neutral processes predominantly shape these communities under healthy conditions (Figure 5(b)). However, fungal communities, though still largely neutral, exhibited slight deviations, implying that other ecological processes may also be influencing their composition in healthy stems (Figure 5(e)). In infected stems, a noticeable shift was observed. Both bacterial (R ² = 0.232) and fungal (R ² = 0.205) communities demonstrated significant departures from neutral theory, particularly within the fungal communities (Figure 5(c) and (f)). This deviation suggests that deterministic factors – such as the selective pressures exerted by infection – are becoming more influential, altering the microbial community structure in infected stems and reducing the role of neutral processes. Overall, these findings highlight the importance of neutral processes in structuring microbial communities in both healthy and infected stems. However, the presence of infection introduces deterministic forces that disrupt this balance, particularly in fungal communities, where infection drives a more pronounced departure from neutrality. This shift likely contributes to the observed differences in microbial community composition and function between healthy and infected stems.

Figure 5

NST analysis for bacterial and fungal communities. (a) Overall fit of total bacterial communities to the neutral model. (b) and (c) Distribution of bacterial taxa in healthy stems (HR) and infected stems (IR), respectively. (d) Total fungal communities’ fit to the neutral model. (e) and (f) Alignment of fungal communities in HR and IR to the neutral model. R-squared represents the goodness-of-fit of the neutral model to the observed data, with higher values indicating a stronger influence of neutral processes. Nm denotes the immigration rate, or the number of individuals migrating into the community per generation, with higher values suggesting greater connectivity and dispersal within the community.

3.6 Network analysis of microbial interactions reveals synergistic and antagonistic dynamics during pathogen co-infection in soil ecosystems

We utilized molecular ecological network (MEN) analysis to investigate the changes in microbial interactions following pathogen co-infection. To ensure comparability between different networks, we applied the same threshold values (0.91 for bacterial communities and 0.85 for fungal communities) when constructing MENs for both healthy and infected soil samples. The overall topological indices revealed that the average path lengths (GD) of all networks ranged from 3.898 to 11.219. These values are close to the logarithm of the total number of network nodes and exceed those of the corresponding random networks (Tables S3 and S4), indicating that the MENs exhibit typical small-world network characteristics. The modularity of the networks, ranging from 0.588 to 0.924 for both bacterial and fungal communities, was also significantly higher than that of their respective random networks, suggesting that all constructed networks possess a modular topology. These critical topological features enabled us to proceed with further analysis of the networks.

Further analysis revealed significant differences in the complexity and connectivity of microbial interactions under healthy versus diseased conditions. The ecological networks of HR and IR showed that healthy stems had more interconnected and complex network structures, with numerous nodes and links representing interactions between bacterial and fungal groups (Tables S5 and S6). In contrast, networks in infected stems were notably sparser, with fewer connections and more isolated nodes, indicating a disruption in microbial interactions due to the infection. Z _i –P _i analysis results indicated that the number of keystone species in the bacterial community network was lower in healthy soil (2 nodes) compared to infected soil (3 nodes). However, for the fungal community network, the number of keystone species was higher in healthy soil (20 nodes) than in diseased soil (1 node) (Figure S5). Network stability analysis showed a significant decrease in the robustness of endophytic bacterial and fungal communities after infection (P < 0.001), with bacterial community robustness decreasing from 0.373 to 0.353 and fungal community robustness decreasing from 0.433 to 0.341 (Figure S6(a)). Additionally, infection markedly increased network vulnerability, rising from 0.121 and 0.027 to 0.249 and 0.287, respectively (Figure S6(b)). This suggests that co-infection disrupted the original defense systems of the network, reducing the stability of the endophytic communities.

To gain deeper insights into the dynamics of co-infection, we focused on the pathogen-associated subnetworks – comprising nodes directly linked to pathogens within the broader network. This approach allowed us to explore the interactions between pathogens and antagonistic microbiomes, which may play critical roles in modulating the microbial community’s response to infection. Our analysis revealed that R. solanacearum, a known pathogen, exhibited positive correlations with a diverse set of bacteria, including Rhizobium, Dyadobacter, Bdellovibrio, Chryseobacterium, Devosia, Haloterrigena, Oxalicibacterium, and Sphingobium (Figure 6(c)). These positive associations suggest potential cooperative or synergistic interactions that could influence the persistence and virulence of Ralstonia in the soil environment. Conversely, Ralstonia showed negative correlations with Bdellovibrio bacteriovorus, Chitinophaga pinensis, Brevundimonas, Massilia, and Nitrospira, indicating potential antagonistic interactions that might inhibit its proliferation or pathogenicity. Similarly, within the fungal community, N. falciformis, a pathogenic fungus, was positively correlated with Preussia terricola and Endogonomycetes, suggesting possible cooperative relationships that could enhance its survival or pathogenicity (Figure 6(f)). On the other hand, Neocosmospora exhibited a negative correlation with Chytridiomycetes, which may represent an antagonistic interaction, potentially limiting its impact on the host. These findings highlight the complex interplay between pathogens and other microbial species during co-infection, revealing both synergistic and antagonistic relationships that could significantly influence disease outcomes.

Figure 6

Intra-domain network analysis of bacterial and fungal communities. (a) Bacterial network of healthy samples. (b) Bacterial network of infected samples. (c) Bacterial sub-network of infected samples. (d) Fungal network of healthy samples. (e) Fungal network of infected samples. (f) Fungal sub-network of infected samples. Different colors represent distinct phyla within the bacterial and fungal communities. The size of each node reflects its node degree, indicating the number of direct connections it has within the network. In all networks, red links denote negative correlations between nodes, which may indicate competitive or antagonistic interactions.

3.7 Synergistic and antagonistic relationships in fungal–bacterial co-infections

The bipartite network analysis of fungal and bacterial interactions reveals the intricate and complex relationships between pathogens and their associated microbial communities under co-infection conditions (Figure 7(a) and (b)). R. solanacearum demonstrates both direct and indirect associations with various fungal species, suggesting that this pathogen may manipulate the surrounding microbial community to create an environment more conducive to its proliferation, thereby promoting disease progression (Figure 7(c) and (d)). Specifically, Chrysosporium lobatum, Psathyrella spp., Orbilia spp., Fusicolla acetilerea, N. falciformis and Cladophialophora spp. are directly associated with R. solanacearum, indicating their potential roles as facilitators in the bacterial invasion process (Figure 7(e) and (f)). These fungi may either weaken the host plant’s defenses or engage in synergistic interactions that enhance the bacterium’s virulence. Conversely, other fungi such as Alternaria angustivoidea, Schizothecium spp., Coprinellus flocculosus, and Arthrobotrys arthrobotryoides have been identified as possible antagonists. These species may inhibit R. solanacearum by competing for resources, producing antimicrobial compounds, or altering microbial community dynamics in a way that reduces the bacterium’s ability to establish infection. The network analysis also reveals N. falciformis with a diverse array of bacterial species (Figure 7(e) and (f)). The network analysis centered on N. falciformis highlights its extensive interactions with a wide range of bacterial species, suggesting its significant role in shaping the microbial community associated with tobacco plants. N. falciformis is closely linked to several bacterial taxa, including Pseudomonas chlororaphis, Burkholderia sp., Stenotrophomonas maltophilia, and Rhizobium jaguaris, which may serve as facilitators in its invasion and pathogenicity. Conversely, the network also identifies bacteria such as Pseudorhodofrax spp. and Sphingobium spp. Additionally, we identified a direct interaction between R. solanacearum and N. falciformis (Figure 7(f)), providing compelling evidence of synergistic pathogen cooperation under co-infection conditions. This direct relationship suggests that these two pathogens may collaborate to enhance their invasion efficiency, potentially by jointly manipulating the host’s immune responses or by creating a more favorable environment for mutual survival and proliferation.

Figure 7

Interdomain ecological networks of bacterial–fungal associations in healthy and infected samples. (a) Interdomain ecological network illustrating bacterial–fungal associations in healthy stem samples (HR). This network represents the overall interactions between bacterial and fungal communities in a healthy state. (b) Interdomain ecological network of bacterial–fungal associations in infected stem samples (IR), showing how these interactions are altered in the presence of infection. (c) Sub-network focusing on pathogenic microorganisms within the healthy stem samples, highlighting the interactions specific to pathogenic taxa like R. solanacearum and N. falciformi. (d) Sub-network of pathogenic microorganisms in the infected stem samples, detailing how the interactions between these pathogens and other microbial taxa change under infection conditions. (e) Targeted sub-network in healthy stem samples, displaying only the interactions between each pathogenic microorganism (R. solanacearum and N. falciformis) and their associated nodes. (f) Targeted sub-network in infected stem samples, showing the interactions of each pathogenic microorganism (R. solanacearum and N. falciformis) and their associated nodes under infection.