Introduction

Plants exist in close association with vast communities of microorganisms inhabiting their internal and external environments, including the rhizosphere, phyllosphere, and internal tissues. This collective microbial community, along with its genetic and metabolic potential, is referred to as the plant microbiome (Müller et al. 2016). Bacteria, in particular, dominate in terms of both abundance and species diversity within these communities.

Plant health and the maintenance of ecosystem functions are heavily dependent on the dynamic interactions between hosts and their associated microbiomes. These microbial communities play critical roles in regulating host growth, nutrient uptake, stress responses, and ecosystem functions (Brader et al. 2017; Philippot et al. 2009; Wen et al. 2025). However, when pathogens attacked, these well-coordinated microbial assemblages can be disrupted, leading to dysbiosis and potentially impacting disease outcomes. The development of disease disturbs the physiological state of the host and triggers shifts in microbial communities.(Arnold et al. 2025; Ge et al. 2025) These alterations can spread from the infection site throughout the plant and into associated soils, disrupting the original functional networks of the plant microbiome (Huang et al. 2023; Jiao et al. 2023; Tang et al. 2025).Therefore, understanding how diseases disrupt the balance of plant–microbe interaction is crucial for elucidating disease mechanisms and improving integrated management strategies for forest diseases.

Pine wilt disease (PWD) is a destructive vascular wilt disease caused by the pine wood nematode Bursaphelenchus xylophilus (PWN). It kills pine trees rapidly and causes severe ecological damage. After colonizing the xylem, the nematode obstructs water transport and triggers systemic physiological collapse in the host (Ikeda and Kiyohara 1995; Masai et al. 1993). This vascular damage is likely to induce cascading changes in plant tissue composition and root exudates, thereby altering the dynamics of endophytic and rhizosphere microbial communities (Jiao et al. 2023). Traditional studies have predominantly focused on the pathogenic mechanisms of nematodes and their interactions with vector insects. However, growing evidence suggests that PWD also significantly reshapes the host-associated microbial community structure (Jiao et al. 2023; Kikuchi et al. 2009; Ma et al. 2020). Studies on P. thunbergii and P. massoniana have demonstrated shifts in specific bacterial and fungal taxa in stems and roots following infection (Liu et al. 2023, 2021). In particular, spatially distinct microbial compartments may exhibit divergent responses to disease progression, and clarifying such stage- and compartment-specific patterns is essential for revealing the underlying microbial dysbiosis mechanisms of PWD. Nevertheless, how the host-associated microbiome reorganizes across different infection stages and in various compartments remains poorly understood.

A thorough investigation of the plant microbiome during disease progression is essential for understanding how pathogens alter host-associated microbial communities and ecosystem processes. This understanding is key for forest nutrient cycling, as disease‑induced shifts in microbial composition can affect essential functional traits, such as nutrient cycling, pathogen antagonism, and stress tolerance. For example, diseased plants may recruit beneficial bacteria to facilitate survival, creating feedback loops that affect disease dynamics (Gao et al. 2025). Moreover, vascular diseases like PWD accelerate host senescence and demise, leading to increased litter inputs. Microbial communities colonizing decaying tissues control decomposition rates and soil carbon/nutrient dynamics (Liu et al. 2025). Indeed, community succession driven by PWD alters the structure and function of microbial communities during litter degradation (Qu et al. 2022). However, most existing studies have only compared healthy versus diseased trees, or focused solely on a single compartment (Liu et al. 2021; Ma et al. 2020). A comprehensive understanding of where and when dysbiosis arises, how it propagates within the plant, and it affects bacterial ecological functions is therefore still lacking.

To address these knowledge gaps, we conducted a comprehensive study examining how PWD affect the host microbiome in P. massoniana, a major target tree species of PWN in China. We systematically characterized the bacterial microbiome across five key compartments (needles, stems, roots, rhizosphere, and bulk soil) throughout four visual stages of disease progression (healthy, early, intermediate, and late). The objectives of this study were to: (1) quantify the stage-specific variations in bacterial diversity, community structure, and potential functional changes across different compartments; (2) identify key shifts in bacterial taxonomy and potential functions associated with disease progression. This study aims to elucidate the spatiotemporal response patterns of bacterial communities in P. massoniana to PWD, and to provide a microbiome-oriented perspective on understanding the ecological impact of this disease.

Materials and methods

Study sites and sample collection

The study was conducted in a P. massoniana Lamb. forest within the Sun Yat-sen Mausoleum Park, located on the southern slope of Purple Mountain, Nanjing, Jiangsu Province, China (118° 48′24″–118° 53′ 04″ E, 32° 01′57″–32° 06′ 15″ N). The region experiences a subtropical monsoon climate zone, with a mean annual precipitation of 1000–1050 mm and approximately 2213 h of sunshine (Deng et al. 2014b). The soils are classified as acidic to slightly acidic yellow–brown and gray-brown soils (Deng et al. 2014a). Pine wilt disease was first reported in this area in 1982 (Zhao 2008), and the forest has since suffered from continuous PWD outbreaks, making it a representative site for studying disease progression under natural conditions.

Twenty trees were selected to represent four distinct stages of PWD severity, including (i) healthy (H): no visible foliar symptoms and no detection of B. xylophilus; (ii) early stage (E): < 10% of needles showing chlorosis; (iii) intermediate stage (I): 40–60% of needles chlorotic or wilted; and (iv) late stage (L): > 80% of needles browned and wilted (Li et al. 2022). The presence of B. xylophilus in symptomatic trees was confirmed in the laboratory using nematode isolation and species-specific PCR (Kikuchi et al. 2009). Five biological replicate trees were selected per disease stage.

Within each disease stage, replicate trees were randomly dispersed throughout the sampling site, covering an area of ~ 4.2 ha, and the minimum distance between adjacent sampled trees was ≥ 30 m. Trees with highly divergent height and diameter at breast height (DBH) were excluded during selection to randomize tree‑size effects across disease stages. From each tree, samples were systematically collected from five compartments. For needles, approximately 45 needles were collected from the mid-canopy from three directions (120° apart) and pooled as one sample. For stems, samples were collected using a sterilized 10 mm increment borer at 1.5 m height along three directions spaced 120° apart and pooled. For roots and rhizosphere soil, fine roots (< 2 mm diameter) with adhering soil were excavated from three directions around the tree base, placed in sterile bags. In the lab, the rhizosphere soil was collected by shaking method (Liu et al. 2021). The roots were then vortex-washed sequentially in sterile water to remove residual soil. For bulk soil, soil cores (20 cm depth) were taken approximately 1.5 m from the tree trunk from three direction after removing surface litter, and pooled as one soil sample. All samples were immediately placed on ice, transported to the laboratory, and stored at −20°C for further process. This design yielded a total of 100 samples (4 disease stages × 5 compartments × 5 biological replicates).

DNA extraction, PCR amplification, and illumina MiSeq sequencing

Total genomic DNA was extracted from soil samples using the Soil DNA Kit (Omega Bio-tek, Norcross, GA, USA) and from plant tissues (needles, stems, roots) using the Plant Genomic DNA Kit (Tiangen, Beijing, China), following the manufacturers’ protocols. The concentration and purity of DNA were checked with a NanoDrop ND-1000 spectrophotometer.

The V3-V4 hypervariable region of the bacterial 16S rRNA gene was amplified using the universal primers 338 F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) (Qu et al. 2022). PCR reactions were performed in triplicate for each sample in a 20 µL mixture containing 4 µL of 5 × FastPfu Buffer, 2 µL of 2.5 mM dNTPs, 0.8 µL of each primer (5 µM), 0.4 µL of FastPfu Polymerase (TransGen, Beijing), 0.2 µL of BSA, and 10 ng of template DNA. The thermocycling conditions were: 95 °C for 3 min; 27 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 45 s; followed by a final extension at 72 °C for 10 min (Ma et al. 2020). The pooled PCR products were purified using the Agencourt AMPure XP kit (Beckman Coulter, USA). Sequencing libraries were constructed with the NEBNext Ultra II DNA Library Prep Kit (NEB, USA) and quantified. Paired-end sequencing (2 × 300 bp) was performed on an Illumina MiSeq 6000 platform (Magigene, Guangzhou, China) following standard protocols. The raw data generated were deposited in the National Center for Biotechnology Information database under the accession number PRJNA1474273.

Bioinformatic processing of sequence data

The raw sequence data were processed using the Mothur software (v.1.48.0) according to the Standard Operating Procedure (SOP). Briefly, reads were merged, and adapters or primers were trimmed using Cutadapt (v.1.15). Sequences with ambiguous bases, lengths outside the expected range, or homopolymers > 8 bp were removed. High-quality sequences were aligned against the SILVA reference database (v.138.1) (Schloss Patrick et al. 2009). Chimeric sequences were identified and removed using UCHIME. The remaining sequences were clustered into operational taxonomic units (OTUs) at a 97% similarity threshold. OTUs represented by fewer than 10 reads across all samples were discarded to reduce spurious noise. Taxonomy was assigned using the SILVA database with an 80% confidence threshold.

To standardize sequencing effort for downstream comparable analyses, all samples were rarefied to the lowest number of sequences per sample (subsampling without replacement) prior to calculating diversity metrics.

Prediction of bacterial functional potential

The potential metabolic functions of the bacterial communities were predicted using Functional Annotation of Prokaryotic Taxa (FAPROTAX, version 1.1). This database collates functional traits (e.g., nitrification, methylotrophy) based on the literature for cultured prokaryotes. The OTU table was mapped against the FAPROTAX database to generate a community functional profile.

Statistical analyses

All statistical analyses and visualizations were performed using R software (version 4.5.2). Community α-diversity was assessed using the Chao1 index (richness), Shannon index (diversity), and Shannon’s evenness. α-diversity indices among different disease stages in each compartment were analyzed by one-way ANOVA with Duncan’s multiple range test using the agricolae package with significance level of P < 0.05. Community β-diversity was analyzed using Bray–Curtis dissimilarities based on rarefied OTU abundances, calculated with the vegan package. Permutational multivariate analysis of variance (PERMANOVA; 9999 permutations) was used to test for significant differences in community structure among disease stages and among compartments. Principal coordinates analysis (PCoA) was conducted based on Bray–Curtis distances and visualized using the ggplot2 package.

Venn diagrams were constructed using VennDiagram package to visualize the unique and shared OTUs among different disease stages. An OTU was considered present if its relative abundance exceeded 0.01% in at least one sample. Differential abundance analysis of bacterial phyla and genera, as well as specific functional groups across disease stages were performed using one-way ANOVA on arcsine square-root transformed relative abundance data for each compartment separately. P-values were adjusted for multiple testing using the false discovery rate (FDR) correction where appropriate.

Results

Bacterial community diversity during PWD progression

The impact of disease progression on bacterial α-diversity was most evident in the stem and bulk soil (Fig. 1, Table S1). In stems, the Chao1 index revealed a significant increase at the intermediate stage compared to the early stage (Fig. 1a, P < 0.05), while both Shannon index and Shannoneven index were significantly lower at the late stage compared to all other stages. A similar trend was observed in bulk soil, where all three α-diversity indices (Chao1, Shannon, Shannoneven) decreased significantly reduced at the late stage (P < 0.05). In contrast, no significant changes in α-diversity were detected in the needles, roots, or rhizosphere soil across the disease stages.

Fig. 1
Fig. 1
Full size image

Bacterial community α-diversity in different tissues of Pinus massoniana across four disease stages (H, healthy; E, early; I, intermediate; L, late). (a) Chao1 index; (b) Shannon index; (c) Shannoneven index. Different letters indicate significant differences among stages within the same tissue (one‑way ANOVA followed by Duncan’s test, P < 0.05). Error bars represent standard deviation

Bacterial community structure across different disease stages

Bacterial community structure exhibited significant differences on both the disease stage and the compartment. Principal coordinate analysis (PCoA) explained 37.2% of the total variation in bacterial community structure across different compartments and disease stages (Fig. 2a, Table S2). In the stem, bacterial community at the intermediate stage and late stage were significantly distinct from those at the healthy and early stages (Fig. 2b). In the rhizosphere soil, bacterial communities differed significantly between healthy and early disease stage, and diverged markedly at the late stage compared to both early and intermediate stages. (Fig. 2c). Similarly, bacterial community in the bulk soil at the late stage was significantly different from those at the healthy and intermediate stages (Fig. 2d). However, no significant differences in bacterial community structure were observed in needle and root at any disease stage (Fig. S1, Table S2).

Fig. 2
Fig. 2
Full size image

Principal coordinates analysis (PCoA) of bacterial community structure based on Bray–Curtis dissimilarity with permutational analysis of variance (PERMANOVA) in all compartments (a), stem (b), rhizosphere soil (c) and bulk soil (d) of Pinus massoniana across disease stages (H, healthy; E, early; I, intermediate; L, late)

The composition of both shared (core) and stage-unique OTUs varied significantly among compartments, reflecting the spatial gradient of disease impact (Fig. 3). In needle, the core community constituted only 8.8% of OTUs shared across different disease stages, and 10.3% in stems, while, stem showed a higher proportion of unique OTUs during the late stages (46.2%, Fig. 3b).

Fig. 3
Fig. 3
Full size image

Venn diagrams showing the number of shared and unique bacterial operational taxonomic units (OTUs) to each disease stage (H, healthy; E, early; I, intermediate; L, late) in needle (a), stem (b), root (c), rhizosphere soil (d) and bulk soil (e) of Pinus massoniana

In roots, the core OTU proportion was 18.5%, with diseased stages (early, intermediate, and late) exhibiting higher percentages of stage-specific OTUs (8.06%–15.3%) compared to the healthy stage (3.73%, Fig. 3c). Bulk and rhizosphere soil retained a high proportion of shared OTUs across all stages (35.4% and 31.9%, respectively), however, the unique stage-specific OTUs remained low (≤ 7.62%; Fig. 3d and e).

Bacterial community composition at phylum level

After quality filtering, a total of 99,797,701 high-quality bacterial sequences were obtained and were clustered into 13, 533 OTUs, assigned to 48 phyla. The overall bacterial community was dominated by Proteobacteria (49.2%), Actinobacteriota (25.6%) and Acidobacteriota (12.1%) (Fig. 4a). In needle and stem tissues, Proteobacteria was the dominant phylum across all disease stages (70%–85% and 50%–90%, respectively). Communities in needles showed no significant changes in the top 10 phyla among disease progression (Figs. 4b, S2a). In contrast, communities in stems exhibited significant stage-dependent shifts (Fig. 4b), with Proteobacteria abundance increasing significantly at the early and late stages (P < 0.05). The abundance of Actinobacteriota was the highest in healthy stems, which decreased significantly at early stage, increased at the intermediate stage, and dropped to its lowest level at the late stage (P < 0.05). Acidobacteriota exhibited higher relative abundances at the healthy and intermediate stages, but was significantly reduced in the early and late stages (P < 0.05). Similarly, Bacteroidota was more abundant in the healthy, early, and intermediate stages, and declined significantly in the late stage (P < 0.05) (Fig. 4c-f).

Fig. 4
Fig. 4
Full size image

Bacterial community composition at phylum level. (a) Relative abundance of the top 10 bacterial phyla across different compartments and disease stages (H, healthy; E, early; I, intermediate; L, late) in Pinus massoniana; (b) Significant shifts in the relative abundance of key phyla across disease stages within stem; (c-f) Phyla exhibiting significant differences in stem bacterial communities across disease stages. Different letters indicate significant differences among stages. Error bars represent standard deviation

In belowground compartments, roots were dominated by Abditibacteriota and Acidobacteriota. Abditibacteriota reached its highest abundance during the healthy phase and gradually declined with disease progression, while Acidobacteriota was significantly enriched in the early and late stages (Fig. S2b). The core phyla in the soil (Proteobacteria, Actinobacteriota, Acidobacteriota) were consistent with those in aboveground compartments, showing no significant phylum-level changes in either rhizosphere or bulk soil across disease stages (Figs. S2c, 2 d).

Genus level composition

At the genus level, 48.1% of the sequences were assigned to 1 539 known genera. In the aboveground compartments, needle tissues maintained a relatively stable community structure. The healthy needle was dominated by core genera including Marivita, Endobacter, and Sphingomonas, while Massilia showed enrichment at the early disease stage but decreased in abundance at both the intermediate and late stages (Fig. S3a). In the stems, significant shifts in genera were observed throughout disease progression. The healthy stems had core genera Methylobacterium and Acidothermus. In the early stage of disease, the relative abundance of Methylobacterium increased significantly, and Acidothermus also showed an increase (P < 0.05). In the intermediate and late stages, the overall relative abundance declined significantly, with Pseudomonas emerging as the sole dominant genus in the late stage (Fig. 5a-c).

Fig. 5
Fig. 5
Full size image

Bacterial community composition at the genus level. (a) Significant shifts in the relative abundance of key genus within stem across disease stages (H, healthy; E, early; I, intermediate; L, late); (b-c) Genera exhibiting significant differences in stem bacterial communities across disease stages; (d) Significant shifts in the relative abundance of key genus within root across disease stages; (e) Genus exhibiting significant differences in root bacterial communities across disease stages. Different letters indicate significant differences among disease stages. Error bars represent standard deviation

In roots, Acidothermus was the dominant genus. Mycobacterium was significantly enriched in the early stage (Fig. 5d-e, P < 0.05). In both rhizosphere and bulk soil, Acidothermus was the dominant genus. Its relative abundance decreased at early and intermediate stages, but was significantly enriched at the late stage. In rhizosphere soil, Massilia exhibited an enrichment in the early stage, but subsequently decreased in abundance thereafter (Fig. S3b-c).

Stage specific alteration in predicted functional potential

PCoA of predicted functional profiles explained 65.3% of the total community variation in P. massoniana across compartments and disease stages, with PCoA1 and PCoA2 accounting for 50.3% and 15% of the variance, respectively (Fig. 6a). Functional structure variations across disease stages were only observed in stems, where the late stage formed a distinct functional structure from all other stages (Fig. 6b; Table S3). Similar as community structure, each compartment formed significant bacterial community functional structure.

Fig. 6
Fig. 6
Full size image

Principal coordinates analysis (PCoA) of bacterial functional structure based on Bray–Curtis dissimilarity with permutational analysis of variance (PERMANOVA) in all compartments (a) and stem (b) of Pinus massoniana across disease stages (H, healthy; E, early; I, intermediate; L, late)

A total of 1, 977 OTUs (14.6% of total OTUs) were successfully matched to the FAPROTAX database, yielding 43 potential functional groups with 13 clusters. The community was predominantly composed of chemoheterophy (86.22%), followed by ureolysis (6.70%), nitrate reductions (1.79%), and predatory or ectoparasitic (1.43%) (Fig. 7a). In stems, functional changes were pronounced at the late stage (Fig. 7b), with significant enrichment in Chemoheterotrophy, ureolysis, predatory or exoparasitic, phototrophy, aromatic compound degradation, and intracellular parasites, all of which peaked significantly higher than those in the healthy or early stages (all P < 0.05). No functional shifts were detected in needles, roots, or bulk soil across any disease stage (Fig. 8).

Fig. 7
Fig. 7
Full size image

Relative abundance of major predicted bacterial functional groups. (a) Overall abundance across all samples; (b) Stage-specific changes in stems. Different letters indicate significant differences among stages (H, healthy; E, early; I, intermediate; L, late) (P < 0.05) Different letters indicate significant differences among disease stages in stem. Error bars represent standard deviation

Discussion

This study investigated the spatiotemporal dynamics of bacterial communities across five compartments of Pinus massoniana during pine wilt disease (PWD) progression. Our results reveal compartment-dependent and stage-specific responses, with the stems showing the most pronounced changes, while needles remained stable. Belowground compartments (roots, rhizosphere, and bulk soil) exhibited moderate restructuring, evident only at specific stages. These findings provide a microbiome-based perspective on the consequence of PWD and highlight the stem as the core compartment of affected by microbial dysbiosis.

Changes in bacterial community diversity during PWD progression

Bacterial community α-diversity in stems demonstrated a pronounced response to disease progression. Richness decreased at the early stage, increased at the intermediate stage, and collapsed at the late stage. Shannon diversity and evenness dropped sharply only at the late stage. This pattern reflects the physiological progression associated with PWD. At the early stage, B. xylophilus invasion triggers host defence responses, resulting in the release of antimicrobial compounds like terpene and phenolics (Sun et al. 2024; Yu et al. 2025; Zhu et al. 2024). These compounds may suppress bacterial colonization, leading to the early decline in community richness. As nematodes proliferate, they secrete effector proteins that suppress host immunity (Hu et al. 2022), while progressively tissue damage releases cellular contents (e.g., carbohydrates, amino acids) that promote fast-growing bacterial populations (Hu et al. 2020). Consequently, this contributes to the peak in bacterial richness observed during the intermediate stage. However, extensive xylem dysfunction and hydraulic failure at the late stage lead to tissue necrosis, reducing niche diversity and favouring a few copiotrophic taxa (Ikeda and Kiyohara 1995).

In contrast, needle communities showed no significant diversity changes across any disease stage, likely because they are not directly invaded by the nematode. Their microbial communities are primarily shaped by atmospheric conditions and leaf surface chemistry, which are relatively insulated from systemic physiological changes until the tree is nearly dead. This observation supports the notion that phyllosphere bacteria demonstrate a higher response threshold to PWD stress (Sun et al. 2021b).

Within belowground compartments, bacterial diversity in bulk soil decreased only at the late stage, while root and rhizosphere diversity remained unchanged. Although PWD can reduce root carbohydrate supply from roots due to stem vascular damage and alter root exudate composition (Feng-mao et al. 2012; You et al. 2025), selective cellular barriers in root tissues limit microbial colonization and migration, lowering intraradical community sensitivity to disturbances (Tian et al. 2025), thereby explaining the minimal PWD impact on root microorganisms. As environment-driven soil bacterial communities, rhizosphere α-diversity may remain unaltered under PWD (Hou et al. 2025; Jiao et al. 2023), potentially due to high microbial dispersibility (Goss-Souza et al. 2020), whereby functionally equivalent bacteria can migrate from non-rhizosphere soil into the rhizosphere. Bulk soil, with more complex co-occurrence networks and greater disturbance resistance (Yu et al. 2025), exhibited late-stage α-diversity decline likely driven by tree-death-induced root decay and soil nutrient shifts (e.g., reduced pH, P, and K) (Chu et al. 2016).

Changes in bacterial community structure during PWD progression

Key taxonomic shifts associated with disease progress was observed. β‑diversity analysis revealed that stem bacterial communities began restructuring at the intermediate stage, with the most drastic shift at the late stage. This is consistent with the proposed sequence of physiological events: early immune suppression, intermediate resource release, and final necrotic collapse. The stem bacterial community therefore serves as a sensitive indicator of disease progression.

The unique OTU number was highest in stems, with only 8.8% of OUTs shared across all disease stages; late‑stage stems harboured a remarkable 46.2% unique OTUs. This indicates that the stressful, resource‑fluctuating environment of diseased stems favours opportunistic, transient taxa (Cao et al. 2024; Chen et al. 2025; Nascimento et al. 2016). In contrast, belowground compartments retained large core microbiomes (31.9–35.4% shared OTUs), reflecting the stabilizing influence of soil heterogeneity, functional redundancy, and high dispersal rates (Goss-Souza et al. 2020; Qin et al. 2024; Yue et al. 2023).

At the phylum level, notable changes were observed in stems, with Proteobacteria increasing from 45.2% to 68.7% as the disease progressed and becoming overwhelmingly dominant at the late stage. Conversely, Actinobacteriota, typically associated with host defense (Kaur et al. 2023; Xiong et al. 2024), were most abundant in healthy stems but declined along the disease progress. Acidobacteriota, an environmental stress indicator (Tang et al. 2026; Yang et al. 2024) were reduced at early and late stages but transiently recovered at the intermediate stage. At the genus level, the community shifted from dominance by Methylobacterium and Acidothermus in healthy stems to Pseudomona dominance at the late stage. Certain Pseudomonas species are known for their ability to degrade aromatic compounds (Masai et al. 1993; Rodrigues et al. 2017; Tian et al. 2025), which become more prevalent as phenols and terpenes are released from necrotic xylem. This taxonomic shift reflects the changing chemical environment of the dying stem, consistent with findings by Hao et al. (2022), who reported that acidophilic microorganisms in pine stems become increasingly enriched with PWD progression (Hao et al. 2022).

In contrast to the dynamic shifts in stems, needle bacterial communities showed no significant changes at phylum or genus level across disease stages. This stability may reflect their primary drivers of needle microbial communities—atmospheric conditions and leaf surface chemistry, and their high intrinsic tolerance to environmental fluctuations (Sun et al. 2021a). Belowground, roots were dominated by Acidothermus and Abditibacteriota, with the former being enriched at the late stage due to increased availability of dead root residues (Barabote et al. 2009; Jiao et al. 2023). In rhizosphere and bulk soil, no significant changes occurred at the phylum level, with only minor genus‑level fluctuations (e.g., transient Massilia enrichment at the early stage). However, the rhizosphere showed significant community restructuring at both the early and late stages. Notably, at the late stage, although bulk soil exhibited a pronounced decline in bacterial α-diversity, its overall community composition remained only moderately altered. The high proportion of shared OTUs and absence of phylum-level shifts suggest that α-diversity decline mainly reflects the loss of rare taxa rather than turnover of dominant bacterial groups, indicating strong soil buffering capacity and functional resilience in these compartments (Goss-Souza et al. 2020; Hou et al. 2025; Yue et al. 2023).

Changes in predicted functional profiles during PWD progression

Predicted functional profiles exhibited significant changes primarily in stems at the late stage. Functional groups such as Chemoheterotrophy, ureolysis, aromatic compound degradation and intracellular parasitism were significantly enriched. Such shifts have direct implications for forest nutrient dynamics, as increased cheomheterotrophy indicates elevated microbial decomposition of dying xylem tissue. Ureolysis suggests enhanced mineralization of organic nitrogen, which is supported by field observations of decreased microbial biomass nitrogen and increased N mineralization after PWD infection (Song et al. 2025). This enrichment of aromatic compound degradation is directly linked to the breakdown of lignin-derived phenols and terpenes from necrotic wood, which may accelerate the decomposition of standing dead trees and influence carbon turnover (Masai et al. 1993; Rodrigues et al. 2017). However, no significant functional shifts were detected in needles, roots, rhizosphere, or bulk soil across disease stages. This functional redundancy (Sauma-Sánchez et al. 2024) and soil buffering (Iqbal et al. 2025) likely prevent major disruptions to biogeochemical processes until the very late stage of tree death.

Notably, the compartment-specific patterns of microbial dysbiosis identified in this study have practical implications for PWD control and forest stewardship. Because the stem showed the strongest microbial community disruption during disease progression, it may serve as a priority target for early monitoring in Pinus massoniana forests. Foresters may track shifts in stem-associated Proteobacteria and Actinobacteria to support early warning, which would help overcome the delayed diagnosis that comes from relying only on visible symptoms. These compartment-based microbial findings provide a practical microecological perspective for precise PWD management and sustainable forest stewardship (Yin et al. 2021). Our study reveals the spatiotemporal response of P. massoniana bacterial communities to PWD, identifying the stems as the core compartment experiencing microbial dysbiosis. However, several limitations should be acknowledged. First, functional predictions from 16S rRNA data reflect potential rather than actual metabolic activity; methods such as metatranscriptomics or metabolomics could confirm the inferred functional shifts. Second, our study was conducted at a single site with one pine species; generalization these findings to other hosts or environments requires further testing. Additionally, the lack of direct measurements of root exudate composition and soil nutrient fluxes limits the ability to link plant physiology to microbial changes mechanistically. Future studies should integrate plant metabolomics, soil biogeochemistry and microbiome time series to better understand the plant–microbe-soil feedbacks during disease progression.

Conclusion

This study demonstrates that bacterial community responses to PWD in Pinus massoniana are highly dependent on both compartment and disease stage (Fig. 8). The stems, serving as the initial infection site, represent the core compartment experiencing microbial dysbiosis. Significant changes in diversity, compositional turnover (with only 8.8% shared OTUs across disease stages), and a shift from dominance of Methylobacterium and Acidothermus to Pseudomonas in this compartment. Additionally, late-stage functional profile indicate enrichment of chemoheterotrophy and aromatic compound degradation. In contrast, needle communities remain stable throughout disease progression, while belowground compartments exhibit substantial resilience, maintaining large core microbiomes (with 31.9–35.4% shared OTUs) and with no phylum-level shifts. These findings enhance our understanding of PWD pathogenesis and establish a microbiome-based foundation for disease monitoring and management in P. massoniana forests.

Fig. 8
Fig. 8
Full size image

Schematic diagram of summarizing bacterial community dynamics across five compartments (needle, stem, root, rhizosphere soil, and bulk soil) of Pinus massoniana during pine wilt disease progression (healthy, early, intermediate, and late stages). Red upward and downward arrows indicate significant increase or decrease in α diversity, respectively. Double headed arrows (⇄) represent significant shifts in community composition. Circular arrows (⟲) denote significant changes in predicted functional profiles