Effects of bacillin 20 on microbial community structure and cross-domain networks in the soybean root-rhizosphere interface Authors: Bulbul Ahmed, Mahtab Nazari, Rachid Elfermi, Ahmad H Kabir, Mohamed Hijri, Donald L Smith Published: 2026 ABSTRACT AbstractBackgroundUnderstanding how biostimulants modulate plant-associated microbiomes is critical for advancing sustainable agriculture. Here, we investigated the effects of Bacillus-derived bacillin 20 on the root and rhizosphere microbiomes of soybean (Glycine max) using amplicon-based profiling, community ecology, and network analysis.ResultsMicrobial community assembly was driven primarily by plant compartmentalization, with higher bacterial richness in the rhizosphere and stronger host filtering in root-associated microbiomes. PERMANOVA analysis indicated that compartment explained most of the variation in community structure, whereas bacillin 20 treatment had no statistically significant effect. Despite no meaningful shifts in alpha diversity, bacillin 20 was associated with subtle, non-significant compositional trends in specific taxa across treatments. Indicator species and core microbiome analyses revealed compartment-specific taxa with potential roles in nutrient cycling, stress tolerance, and plant growth promotion.ConclusionBacillin 20 was associated with changes in cross-domain microbial co-occurrence patterns, including differences in network connectivity, particularly in the rhizosphere, where several fungal ASVs (e.g., ASV33, ASV5, ASV8, and ASV88) exhibited high centrality. These findings indicate that bacillin 20 is associated with changes in microbial interaction patterns while maintaining overall community diversity. Overall, treatment effects were minor relative to compartment-driven structuring of the microbiome. Together, our results suggest that microbiome-informed approaches, including the use of targeted biostimulants, may contribute to the management of plant-microbe interactions in agricultural systems. Future studies integrating multi-omics approaches will be required to elucidate the underlying mechanisms of these interactions.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12866-026-05135-z.Keywords: Biostimulant, Microbiome, Network, Bacillin 20, Soybean Introduction Plant-associated microbiomes are crucial to host health, and playing essential roles in nutrient acquisition, growth promotion, disease resistance, and tolerance to both biotic and abiotic stresses. These microbial communities are highly coordinated across the root, rhizosphere and endosphere representing distinct ecological compartments shaped by host selection and environmental filtering [1–5]. The rhizosphere typically harbors a rich and diverse community fueled by root exudates, while the root endosphere supports a more selective microbiota, often enriched in symbionts and endophytes adapted to close interaction with the host [6–9]. Increasingly, biostimulants are being explored as tools to harness and steer these microbiomes toward enhancing plant resilience and productivity in sustainable agroecosystems [10, 11]. Among emerging microbial-derived compounds, bacillin 20 has gained attention for its bioactivity and potential for microbiome modulation. bacillin 20, which was initially characterized as thuricin 17, later redefined based on revised insights into its composition and structure, is a peptide produced by Bacillus thuringiensis NEB 17 and belongs to the bacteriocin family of ribosomally synthesized peptides that inhibit related microbial strains [12–14]. Bacillin 20 is distinguished by its ability to promote plant growth without phytotoxicity or negative environmental impacts [15]. Recently, in soybean, bacillin 20 has been shown to enhance biomass production and physiological traits under both optimal and stress conditions. Notably, when co-applied with Bradyrhizobium japonicum, bacillin 20 increased root and nodule biomass by 37% and 55%, respectively, and enhanced photosynthetic rates, leaf area, and abscisic acid accumulation in roots, suggesting coordinated physiological responses related to plant water-stress responses and nutrient uptake [16, 17]. Despite its promising applications, the ecological effects of bacillin 20 on complex plant-associated microbial communities remain poorly understood, particularly regarding potential differences in its impacts across plant compartments, each of which harbors distinct microbial communities and interaction networks. Whether such treatments preserve native diversity, restructure community composition, or enhance functional interactions between cross-domain microbiome elements is a critical question for microbiome-informed biostimulant deployment. Here, we investigated how bacillin 20 modulates bacterial and fungal communities in soybean roots and rhizosphere using amplicon-based sequencing, network inference, and core microbiome profiling. Our data confirm that spatial niche is the primary determinant of microbial diversity and community composition, consistent with previous studies. Importantly, relative to untreated controls, bacillin 20 treatment was associated with detectable shifts in microbial community structure and interaction patterns, and the magnitude of these changes differed between plant compartments. These results indicate that while spatial niche defines baseline community structure, bacillin 20 can modulate microbiome composition and interactions in a compartment-dependent manner. Methods Bacillin 20 extraction and purification Bacillin 20, a bacteriocin originally isolated from Bacillus thuringiensis NEB17, was extracted and purified following established protocols with minor modifications [12, 13]. Briefly, B. thuringiensis NEB17 cultures were maintained in King’s B medium and grown for 48 h at 28 ± 2 °C on an orbital shaker (150 rpm). Cultures were harvested when they reached an OD600 (OD600 of 0.8), indicative of late-log phase. Bacillin 20 was partially purified from Bacillus thuringiensis NEB17 cultures (2 L) by n-butanol extraction, followed by rotary evaporation at 50 °C under vacuum. The dried extract was resuspended in 18% acetonitrile (AcN: H₂O, v/v), centrifuged (13,000 × g, 13 min), and the supernatant subjected to reverse-phase HPLC (Waters Corporation) on a C18 column. Elution was performed using an acetonitrile–water gradient containing 0.1% trifluoroacetic acid (18% AcN for 45 min, 18–60.4% over 45–110 min, 60.7–100% over 110–115 min). Fractions were collected at 1-min intervals, and bacillin 20 was identified by a reproducible retention time of 80–82 min (≈ 60% AcN). Corresponding fractions were pooled, lyophilized, and stored at − 20 °C [14, 18, 19]. Fractions corresponding to bacillin 20 peaks were pooled, lyophilized, and stored at -20 °C until further use. Experimental design and field application Bacillin 20 was applied at concentrations of 10⁻⁹ and 10⁻¹¹ M, previously shown to enhance seed germination, growth, and stress resilience in crops including soybean, Arabidopsis, and canola [14, 19]. Field trials were conducted at McGill University’s Emile A. Lods Research Centre (45°28’ N, 73°45’ W), using soybean (Glycine max (L.) Merrill, cv. Absolute RR) from BelCan, Quebec, Canada. Five experimental treatments were used: T1-control; T2- bacillin 20 (10⁻⁹ M) foliar spray along with herbicide; T3- bacillin 20 (10⁻¹¹ M) foliar spray along with herbicide; T4 -herbicide spray at first stage + foliar spray of bacillin 20 (10⁻⁹ M) at 6 leaf stage; T5- herbicide spray at first stage + foliar spray of bacillin 20 (10⁻11 M) at 6 leaf stage. Although bacillin 20 was applied as a foliar spray, root and rhizosphere microbiomes were examined to capture plant-mediated systemic effects on belowground microbial communities, which are central to nutrient acquisition and soil–plant interactions. The experiment followed a randomized complete block design with four replicates per treatment. Plots (5 m × 14 rows) were established on clay soils, with 19 cm between rows and 80 cm between blocks. Sampling and sample processing At flowering, five healthy plants were randomly selected from each treatment. Root systems were excavated to ~ 15 cm depth, and fine roots were separated from shoots and gently shaken to retain rhizosphere soil. Roots were gently washed under sterile distilled water to remove loosely attached soil particles prior to processing. No surface sterilization was performed; therefore, the root microbiome analyzed in this study includes both endosphere and rhizoplane-associated microorganisms and should be interpreted as the root-associated microbiome rather than a strictly defined endophytic community. Accordingly, the results do not allow inference of strict endophytic selection. Following collection, roots were gently shaken to remove loosely attached soil, briefly rinsed with sterile distilled water without surface sterilization, and homogenized using sterile tools prior to DNA extraction. Rhizosphere soil was collected as soil tightly adhering to roots after gentle shaking. Approximately 0.5 g of root and 1 g of rhizosphere soil per plant were pooled into composite samples (n = 36 total). Root and soil samples were transported on ice and stored at − 20 °C prior to DNA extraction. We did not include nodulation parameters because the primary objective of this study was to characterize root and rhizosphere microbiomes rather than quantify symbiotic nitrogen fixation. Nodulation is known to vary strongly in response to environmental conditions and management practices, even within a single plant genotype, and was therefore excluded to avoid introducing additional phenotypic variability unrelated to microbiome composition. DNA extraction and amplicon sequencing Following the previously established methods [6, 20], total DNA was extracted from root and rhizosphere soil samples collected from 40 soybean plants. Total DNA was extracted from 100 mg of ground root tissue using the DNeasy Plant Mini Kit (Qiagen, Toronto, ON, Canada), and from 250 mg of rhizosphere soil using the DNeasy PowerSoil Pro Kit (Qiagen, Toronto, ON, Canada). DNA concentrations were quantified with a NanoDrop spectrophotometer, and quality was checked on 1% agarose gels. PCR amplification was conducted to target both the bacterial 16 S rRNA gene and the fungal ITS region, followed by sequencing using an Illumina MiSeq platform at the Genome Quebec Innovation Centre (Montreal, QC, Canada). The V3-V4 region of the bacterial 16 S rRNA gene was amplified using the forward primer CS1_341 (5′-ACACTGACGACATGGTTCTACACCTACGGGNGGCWGCAG-3′) and reverse primer CS2_806R (5′-TACGGTAGCAGAGACTTGGTCTGACTACHVGGGTATCTAATCC-3′) [21]. The fungal internal transcribed spacer 2 (ITS2) was amplified using the forward primer CS1_ITS3_KYO2 (5′ACACTGACGACATGGTTCTACAGATGAAGAACGYAGYRAA-3′) and reverse primer CS2_ITS4 (5′TACGGTAGCAGAGACTTGGTCTTCCTCCGCTTATTGATATGC-3′) [22]. Each PCR reaction was performed in a 25 µL volume containing 1.5× Platinum™ Direct PCR Universal Master Mix (ThermoFisher, Montreal, QC, Canada), 0.25 µM of each primer, 1.5× Platinum™ GC Enhancer, and approximately 20 ng of template DNA. Amplicons were visualized, purified, and sequenced on an Illumina MiSeq platform (San Diego, CA, United States; Montreal, QC, Canada) using 2 × 300 bp pair-end reads at the McGill Genome Québec Innovation Centre. Bioinformatics for microbiome data processing and statistical analysis Raw Illumina MiSeq sequence data were processed using the DADA2 pipeline [23] in R v4.0.2. Primer sequences were trimmed and reads were filtered based on quality (maxEE = c [2], truncQ = 2). Denoising, error correction, chimera removal, and ASV inference were performed using default settings with minor modifications. Taxonomic assignment was carried out using the SILVA 138 database [24], and ASVs identified as mitochondria or chloroplasts were excluded. The dataset was rarefied to the minimum sequencing depth using the vegan package (v2.5-6) prior to downstream analysis. Alpha diversity metrics, including Observed Richness, Chao1, Shannon, and Simpson indices were calculated using phyloseq package in R. To explicitly evaluate the effects of Bacillin 20 treatment and plant compartment, alpha diversity was analyzed using a factorial framework with Treatment (T1-T5) and Compartment (root vs. rhizosphere) as fixed factors. Two-way ANOVA was performed to test for main effects and Treatment × Compartment interactions when assumptions of normality and homoscedasticity were met. When these assumptions were violated, non-parametric alternatives including Kruskal–Wallis test followed by Dunn’s post hoc comparisons and Tukey’s HSD post hoc tests (agricolae v1.3-3) [25] were applied within compartments. Where appropriate, compartment-stratified analyses were conducted to directly assess treatment effects independently in root and rhizosphere samples. Beta diversity was assessed using Bray–Curtis dissimilarity, followed by principal coordinates analysis (PCoA) for ordination. Community-level differences were tested using PERMANOVA (adonis function, 999 permutations) [26] with a model including Treatment, Compartment, and their interaction term (Treatment × Compartment). This approach allowed us to quantify the relative contribution of spatial niche versus Bacillin 20 treatment to microbial community structure. Differential abundance analysis was conducted in a compartment-stratified manner to directly evaluate treatment effects within root and rhizosphere separately. Taxa were tested using LEfSe (Kruskal–Wallis p < 0.05; LDA score > 3.0). Additional methods using edgeR [27] and Metastat) [28] were used for cross-validation of abundance trends. Multiple testing correction was applied where appropriate using the Benjamini-Hochberg false discovery rate (FDR). Indicator species and core microbiota analyses were performed using custom R scripts using the package indicpecies [29]. Core taxa were defined as genera present at ≥ 1% relative abundance in ≥ 80% of samples within each compartment. Feature importance and classification accuracy were assessed using Random Forest models. Co-occurrence networks were constructed from filtered ASV tables (relative abundance > 0.01%, present in ≥ 20% of samples) using Spearman correlation (p > 0.6, FDR-adjusted p < 0.05) [30]. Networks were visualized using igraph [31], and Zi-Pi analysis [32] was used to classify nodes into ecological roles such as module hubs, connectors, and peripherals. Results Microbial diversity differs between compartments but remains stable across bacillin 20 treatments Bacterial alpha diversity We characterized bacterial alpha diversity in soybean roots and rhizosphere across bacillin 20 treatment using Chao1, Observed Richness, Shannon, and Simpson indices (Fig. 1a). Bacterial alpha diversity significantly varied between root and rhizosphere compartments. The rhizosphere consistently exhibited higher Chao1 (mean ± SD: 173.3 ± 17.6) and Observed Richness (170.5 ± 16.9) values compared to the root (Chao1: 135.2 ± 30.7; Richness: 132.1 ± 29.9), reflecting increased taxonomic richness in the rhizosphere (p < 0.05). Shannon diversity was also elevated in the rhizosphere (4.72 ± 0.14) relative to the root (4.38 ± 0.36), suggesting greater evenness of taxa (p < 0.05). Simpson index values did not differ significantly (both ~ 0.98–0.99), indicating similar dominance structure in both compartments. These findings suggest that spatial niche filtering exerts strong control over bacterial diversity, with rhizosphere communities being more taxonomically complex than root (Fig. 1). Bacillin 20 application did not significantly (p > 0.96, Tukey’s HSD; Table S1) alter alpha diversity of either bacterial or fungal communities in soybean root or rhizosphere compartments. Across treatments (T1-T5), Chao1, observed richness, Shannon, and Simpson indices showed no significant differences within compartments (Fig. S1). For both microbial groups, diversity patterns differed between root and rhizosphere, with consistently higher diversity observed in the rhizosphere; however, no treatment-driven shifts in richness or evenness were detected in either compartment. This indicates that bacillin 20 application did not perturb the richness or evenness of bacterial communities in either root or rhizosphere. Overall, the spatial compartment remains a dominant factor shaping bacterial alpha diversity in soybean, and bacillin 20 application preserves the underlying bacterial diversity structure across treatments. Alpha diversity of root and rhizosphere microbiome and effect of bacillin 20 treatments. Violin plots represent alpha diversity indices including Chao1 (species richness estimate), Observed Richness, Shannon (diversity), and Simpson (evenness) for bacterial and fungal communities in the root (blue) and rhizosphere (red) of soybean plants. A bacterial communities. Rhizosphere samples exhibited significantly higher Chao1, Richness, and Shannon diversity compared to roots (p < 0.05, letters indicate statistically distinct groups based on Tukey’s HSD test), while Simpson index did not differ significantly between compartments. Each point represents an individual biological replicate. Boxes denote interquartile range and median. B fungal communities. Each dot represents an individual sample. The central black line indicates the median. Statistical comparisons were performed using ANOVA followed by Tukey’s HSD test; groups sharing the same letter are not significantly different at P < 0.05 Fungal alpha diversity We also examined alpha diversity indices using Chao1, Observed Richness, Shannon, and Simpson on fungal communities. Across all indices, fungal communities in the root and rhizosphere compartments did not show statistically significant differences (p > 0.05), although trends indicated slightly higher diversity within the root compartment (Fig. 1b). Specifically, Chao1 and Observed Richness indices, which reflect species richness, were generally higher in roots. In addition, Shannon and Simpson indices, which incorporate both richness and evenness, indicated a similar trend toward elevated diversity in the root microbiome. Despite these observations, no statistically significant variation was observed between root and rhizosphere for any index (Kruskal–Wallis, p > 0.05), suggesting that fungal community diversity is compartmentally stable under baseline conditions. Fungal alpha diversity under the five bacillin 20 treatment (T1-T5) showed no significant treatment effects on any alpha diversity index within either compartment (Kruskal-Wallis, all p > 0.05). The observed patterns showed considerable overlap across treatments (Fig. S1b), indicating that bacillin 20 did not markedly alter fungal richness or evenness in soybean associated communities. Together, these findings suggest that fungal alpha diversity in soybean root and rhizosphere is resilient to bacillin 20 perturbation, and that community structure is not strongly compartment specific, in contrast to patterns often observed for bacterial communities. Beta diversity patterns in bacterial and fungal communities Principal coordinate analysis (PCoA) of Bray–Curtis dissimilarity revealed a pronounced separation between root and rhizosphere bacterial communities, indicating strong compartmental effects on community composition. While overall beta diversity patterns showed limited separation among Bacillin 20 treatments, subsequent ASV-level analyses were performed to detect more subtle, taxon-specific responses that may not be captured by global distance-based metrics. For bacterial communities (Fig. 2A), samples clustered primarily according to compartment along the first principal coordinate (PCoA1, explaining 25.26% of total variation), separating root and rhizosphere assemblages irrespective of Bacillin 20 treatment. Although treatments (T1-T5) showed some overlap within each compartment, PERMANOVA indicated that compartment identity explained a larger proportion of community variation than treatment (Adonis: R = 0.093), while treatment effects were not statistically significant (p = 0.996) (Fig. 2A & Fig. S2A). Principal Coordinate Analysis (PCoA) of microbial communities in soybean root and rhizosphere. (A) bacterial and (B) fungal community composition associated with soybean roots and rhizospheres under five Bacillin 20 treatments (T1-T5). Each point represents an individual sample, colored by treatment and shaped by compartment (root or rhizosphere). Dashed ellipses indicate 95% confidence intervals for each compartment. Percentages on axes denote the proportion of total variance explained by each principal coordinate. PERMANOVA (Adonis) results are shown in each panel, indicating that compartment identity explains more variation in community structure than treatment effects A similar pattern was observed for fungal communities (Fig. 2B). Root- and rhizosphere-associated fungal assemblages formed distinct clusters along PCoA1 (27.19% of variation), with tighter clustering of rhizosphere samples compared with roots, suggesting greater compositional stability in soil-associated fungal communities. PERMANOVA analysis showed that compartment significantly explained variation in bacterial community structure (R² = 0.25, p < 0.001), whereas treatment had a minimal and non-significant effect (R² = 0.093, p > 0.05), with no significant interaction between factors (Treatment × Compartment, p = 0.996, p > 0.05). Together, these results demonstrate that spatial niche differentiation between root and rhizosphere is the dominant factor shaping both bacterial and fungal beta diversity in soybean, whereas Bacillin 20 treatments exerted comparatively minor effects on overall community composition. Compartmental shifts in taxonomic composition of bacterial communities Bacterial community composition was primarily structured by plant compartment (root vs. rhizosphere), while Bacillin 20 treatment did not induce statistically significant compositional shifts within compartments. The rhizosphere microbiome was dominated by taxa such as Planctomycetota, Verrucomicrobiota, and Chloroflexota (Fig. S3A), whereas root-associated communities showed relatively higher representation of Pseudomonadota, Bacteroidota, and Actinomycetota, reflecting niche-specific selection. At finer taxonomic resolution, distinct genera exhibited compartment-specific distributions, as illustrated in heatmap (Fig. 4A; Table S3), indicating ecological filtering along the root–rhizosphere continuum. LEfSe analysis further confirmed significant compartment-driven differences (Fig. 5A) but did not detect any statistically significant treatment-driven differential taxa within compartments. Although some variation in relative abundance across treatments was observed at genus-level (Fig. 3B-C; Fig. S6A-B), these patterns were not statistically supported. Overall, these results indicate that bacterial community composition is strongly structured by spatial compartment, whereas Bacillin 20 does not significantly alter taxonomic profiles. Taxonomic profiles of fungal communities in root and rhizosphere of soybean and under Bacillin20 treatments. Taxonomic structure differs substantially between root and rhizosphere compartments but remains consistent across bacillin 20 treatments. "Others" include low-abundance taxa grouped for clarity. A Genus-level composition of fungal communities in root and rhizosphere. B Genus-level fungal profiles across Bacillin20 treatments (T1-T5) in root. C Genus-level fungal profiles across bacillin 20 treatments (T1-T5) in rhizosphere Cladograms of differentially enriched bacterial and fungal taxa in soybean root and rhizosphere compartments. A LEfSe analysis based on relative abundance profiles identified bacterial clades significantly enriched in root (green) and rhizosphere (orange) samples. Taxa are shown from phylum to genus level; only those with LDA scores >2.0 and P < 0.05 are displayed. Highlighted rings indicate major clades driving compartment-specific differences. B Cladogram of fungal taxa differentially abundant between root (green) and rhizosphere (orange) compartments. Circle size reflects effect size; color intensity denotes statistical significance. Taxa are annotated by order (o_), family (f_), or genus (g_). Notable rhizosphere-enriched genera include Fusarium, Microdochium, and Gongronella, while Paraphoma, Talaromyces, and Alfimbriata were enriched in root tissues Taxonomic profiles of bacterial communities in root and rhizosphere of soybean and under Bacillin20 treatments. Taxonomic structure differs substantially between root and rhizosphere compartments but remains consistent across Bacillin 20 treatments. "Others" include low-abundance taxa grouped for clarity. A Genus-level composition of bacterial communities in root and rhizosphere. B Genus-level bacterial profiles across bacillin 20 treatments (T1-T5) in root. (C) Genus-level bacterial profiles across bacillin 20 treatments (T1-T5) in rhizosphere To further characterize conserved members of the soybean-associated microbiota, we defined the core microbiome as genera present at ≥ 1% relative abundance in ≥ 80% of samples across compartments. Under this criterion, the rhizosphere bacterial core community was dominated by Terrimicrobium, Sphingomonas, Zavarzinella, and unclassified but consistently detected lineages (e.g., UBA2421 and UBA10511), collectively comprising over 60% of the normalized relative abundance (Fig. 6A). The consistent presence and high relative abundance of UBA2421 across both compartments suggest that this lineage represents a dominant and stable component of the soybean-associated microbiome. Core microbiome composition of the root and rhizosphere compartments.Bar plots show the strict core microbial genera (≥1% relative abundance in ≥80% of samples) in the rhizosphere (left) and root (right) compartments. Each bar represents an individual sample, with colors indicating genera based on the Viridis palette. Because only taxa meeting these strict core criteria are displayed, total relative abundance may not reach 100% in some samples. A Bacterial core communities shared several dominant genera, including UBA2421, Terrimicrobium, and Chthoniobacter, though their relative abundances differed between compartments, indicating compartment-specific structuring. B Fungal core communities also differed between compartments, with Fusarium and Cladosporium dominating the rhizosphere, while Talaromyces, Tumularia, and Solicoccozyma were more abundant in the root endosphere Core microbiome assembly across soybean root and rhizosphere compartments Fungal communities displayed clear compartmentalization between roots and rhizosphere. At the family level, Pleosporaceae, Didymellaceae, and Aspergillaceae were dominant in the rhizosphere, while Corynesporascaceae and Phaeosphaeriaceae were enriched in roots (Fig. S4A). Members of the family Mortierellaceae were predominantly detected in the rhizosphere, indicating a preference for soil- and rhizosphere-associated niches rather than close association with root tissues. Genus level profiles revealed further compartment specific patterns, such as roots harboring high relative abundances of Corynespora, Phaeosphaeria, and Neocamarosporium, whereas the rhizosphere favored Mortierella, Aspergillus, and Cladosporium (Fig. S4B). Heatmap and LEfSe analyses confirmed strong compartment-driven structuring of fungal communities, with no statistically significant treatment effects detected within compartments (Figs. 4A–B and 5B). Although descriptive shifts in relative abundance were observed across treatments, these patterns were not statistically supported. Overall, these findings indicate that fungal community composition, like that of bacteria, is primarily shaped by spatial compartment rather than Bacillin 20 application. Applying the same core microbiome criterion (≥ 1% relative abundance in ≥ 80% of samples), fungal core communities exhibited clear compartment-specific patterns (Fig. 6B). The rhizosphere core was dominated by Fusarium, Didymella, Clonostachys, and Cladosporium, representing a relatively consistent assemblage across samples. In contrast, the root-associated core comprised fewer taxa, including Talaromyces, Tumularia, and Paraphoma, with Talaromyces frequently dominating. Although some genera (e.g., Fusarium and Didymella) were detected in both compartments, their relative abundances differed markedly, indicating compartment-specific enrichment. Overall, the root compartment exhibited a smaller and less evenly distributed core community, consistent with stronger host filtering and niche specialization along the root–rhizosphere continuum. Indicator species analysis revealed compartment specific enrichment of microbes Indicator species analysis demonstrated distinct bacterial enrichment patterns between the root and rhizosphere microbiomes of soybean (Fig. 7A, Table S1). Indicator species analysis was used to identify compartment-associated taxa and was not designed to test treatment effects. Using Indicator Value (IndVal) scores, which integrate specificity and fidelity, several genera showed strong compartmental associations. Bradyrhizobium, Terrimicrobium and UBA2421 were significantly enriched in the root compartment, exhibiting strong indicator values (> 0.8; -log₁₀ p ≥ 2), consistent with their preferential association with root-associated niches. Conversely, taxa including Chthoniobacter, AV80, and FW60-2_bin2 were predominantly associated with the rhizosphere, highlighting their adaptation to the root-soil interface. The strength of association varied among taxa with multiple genera showing statistically significant shifts in abundance between compartments (P < 0.05). These findings underscore the phylogenetic diversity of key bacterial taxa driving microhabitat partitioning within the plant microbiome and establish a framework for targeted functional validation. Quantitative enrichment of indicator taxa in root and rhizosphere compartments. Bubble plots showing significantly enriched indicator genera in the root and rhizosphere compartments based on Indicator Value (IndVal) analysis, calculated as the product of specificity and fidelity. The x-axis represents the IndVal score, with higher values indicating stronger compartment-specific association. Bubble size corresponds to the –log₁₀(p) value, providing a quantitative measure of statistical significance, while colors denote the enriched compartment (blue: root; red: rhizosphere). (A) bacterial indicator genera, and (B) fungal indicator genera. This representation allows direct comparison of enrichment strength and significance across taxa and compartments Similarly, fungal communities displayed clear compartment-specific signatures (Fig. 7B, Table S2). Indicator species analysis was used to identify compartment-associated taxa and was not designed to test treatment effects. Fungal communities in the root were characterized by Tumularia, Corynespora, and Paraphoma, primarily members of the order Pleosporales. These taxa showed strong indicator values and significant abundance shifts (P < 0.05), consistent with preferential association with root-associated niches. In contrast, the rhizosphere hosted distinct indicator fungi, including Fusarium, Didymella, Talaromyces, and Solicoccozyma, reflecting the broader taxonomic diversity typical of the root–soil interface. These patterns indicate strong ecological filtering between compartments and support compartment-specific fungal community assembly in soybean. Bacillin 20 is associated with changes in microbial co-occurrence networks To assess whether bacillin 20 influenced microbial association patterns, cross-domain co-occurrence networks were constructed using bacterial and fungal ASVs from soybean root and rhizosphere compartments. Significant associations were defined based on correlations (|r| > 0.3, P < 0.05). In the root compartment (Fig. S7), the network comprised 210 ASVs connected by 398 significant edges. Bacteria–bacteria interactions accounted for 60.8% of edges, followed by bacteria–fungi (23.4%) and fungi–fungi (15.8%) interactions. Several fungal ASVs (e.g., ASV33, ASV5, ASV8, ASV88) showed high connectivity and were linked to multiple bacterial ASVs, including ASV2, ASV23, and ASV70. The rhizosphere network (Fig. S8) included 302 ASVs and 652 significant associations, indicating higher network complexity compared to the root compartment. Bacteria–bacteria interactions represented 60.6% of edges, followed by bacteria–fungi (24.7%) and fungi–fungi (14.7%) interactions. Fungal ASVs (e.g., ASV13, ASV88, ASV35) exhibited relatively high centrality and formed multiple associations with bacterial taxa. When root and rhizosphere datasets were combined (Fig. 8), the resulting network showed increased node centrality and edge density compared to the root-only network. Across compartments, several fungal ASVs (including ASV33, ASV5, ASV8, and ASV88) consistently displayed high connectivity and were associated with multiple bacterial partners (e.g., ASV2, ASV23, ASV70, ASV100). Cross-domain co-occurrence network of bacterial and fungal ASVs in soybean roots under bacillin 20 treatment. The network is based on significant Spearman correlations (|r| > 0.3, P < 0.05) among 210 ASVs. Edges represent positive associations, colored by interaction type: bacteria–bacteria (blue), fungi–fungi (orange), and cross-domain (purple). Node color indicates domain affiliation (blue: bacteria, orange: fungi), and edge thickness reflects correlation strength (weight). Key fungal ASVs (e.g., ASV5, ASV8, ASV33, ASV88) formed hub-like structures through extensive cross-domain and intra-kingdom linkages, highlighting enhanced microbial connectivity under bacillin 20 influence Importantly, no significant treatment-associated differences were detected in taxonomic abundance (based on differential abundance analysis; data not shown). However, differences in network properties, including edge density, node centrality, and the proportion of cross-domain associations, were observed between conditions. These results indicate that, although bacillin 20 did not significantly alter the relative abundance of individual taxa, it was associated with changes in microbial co-occurrence patterns, including network connectivity, node centrality, and cross-domain interactions. This suggests that bacillin 20 influences the structure of microbial associations rather than taxonomic composition. Because co-occurrence networks are based on correlation patterns rather than absolute abundance, shifts in network topology can occur even when overall taxonomic composition remains stable. Discussion Microbial inoculants are increasingly explored as tools to enhance plant resilience and productivity, yet their influence on microbiome structure and function remains largely unexplored. This study examined how Bacillin 20 influences the structure and dynamics of soybean-associated microbiomes across root and rhizosphere compartments. Our analyses confirmed that spatial niche was the primary and statistically supported driver of microbial community composition, whereas Bacillin 20 treatment did not induce significant taxonomic shifts within compartments. Compartmental niche strongly shaped microbial diversity and composition Compartmentalization emerged as the dominant driver of microbial community structure, consistent with previous studies showing that roots and rhizosphere represent distinct ecological niches [2, 6, 20, 33, 34]. Bacterial alpha diversity was higher in the rhizosphere, reflecting its role as a dynamic interface enriched by root exudates and soil-derived taxa, whereas root-associated communities were more selective and variable, likely due to host filtering [4, 35]. Fungal communities showed similar, though less pronounced, compartmental differentiation, supporting the role of spatial filtering in plant–microbe interactions [36]. The distinct microbial assemblages between root and rhizosphere were further evidenced by differential taxonomic signatures and core microbiome profiles. The rhizosphere core bacterial community was dominated by Zavarzinella, UBA2421 and Terrimicrobium associated with nutrient cycling and soil resilience [37]. The root core was more taxonomically diverse but included known endophytes such as Sphingomonas and Paraburkholderia, reflecting functional specialization [7, 38]. Similarly, fungal cores diverged between compartments, with root tissues enriched in Talaromyces and Paraphoma, whereas rhizospheres were dominated by Cladosporium and Fusarium. These patterns suggested a finely tuned microbiome shaped by both microbial adaptation and host filtering [39]. Bacillin 20 treatment enhanced this compartmental structure by enriching niche-adapted taxa. Notably, Trichoderma and Mortierella fungi with known plant beneficial traits were elevated in the rhizosphere under certain treatments, indicating that bacillin 20 may prime the system for beneficial associations [10]. Bacillin 20 treatment did not significantly alter taxonomic composition but was associated with subtle, non-significant trends in the relative abundance of some taxa across compartments. These findings suggest that spatial niche is the primary determinant of microbial community structure, whereas Bacillin 20 effects, if present, are minor relative to compartment-driven variation. Bacillin 20 maintained alpha diversity but slightly modulated community composition Despite repeated applications of bacillin 20, we observed no significant alterations in bacterial or fungal alpha diversity across treatments. This suggested that bacillin 20 did not broadly disrupted the structure of the soybean microbiome, unlike some synthetic agrochemicals known to diminish microbial diversity [40, 41]. However, community composition analyses revealed subtle but consistent treatment-specific shifts, particularly at the genus level. Although no statistically significant treatment-driven differences in taxonomic composition were detected, some taxa exhibited non-significant trends in relative abundance across treatments [42–44]. In fungi, we observed a marked decline in potentially pathogenic taxa (e.g., Neocamarosporium) and an increase in beneficial or opportunistic genera like Fusarium and Penicillium, depending on treatment and compartment. These findings indicated that while bacillin 20 preserved community stability at the diversity level, it fine-tuned microbial composition in a compartment-specific manner. Bacillin 20 promoted cross-domain microbial interactions and network complexity Network analyses revealed that microbial co-occurrence patterns were strongly influenced by cross-domain (bacteria–fungi) interactions, particularly in the rhizosphere. Several fungal ASVs exhibited high connectivity and centrality, highlighting their key role in structuring microbial networks [45]. Bacillin 20 treatment was associated with increased network complexity and the emergence of additional cross-kingdom associations, suggesting a more integrated microbiome. Such changes in network topology may influence ecosystem stability and resilience by altering interaction patterns among taxa [46–48]. Notably, these shifts occurred without statistically significant differences in taxonomic composition, indicating that Bacillin 20 primarily reorganizes microbial interactions rather than community membership. Consistent with this, treatment explained only a small proportion of the variance in community structure (PERMANOVA: R² = 0.093 for bacteria; 0.084 for fungi), whereas compartment remained the dominant driver. Overall, these findings suggest that Bacillin 20 reshapes microbial connectivity without disrupting community composition. Implications for microbiome engineering and sustainable agriculture Our results indicate that Bacillin 20 can modulate the soybean microbiome in a targeted and non-disruptive manner by enhancing microbial interaction networks rather than inducing strong taxonomic turnover. This capacity to strengthen cross-domain connectivity while maintaining community stability is particularly relevant for microbiome-informed agricultural strategies aimed at improving plant productivity and disease resistance [49]. The observed compartment-specific responses further highlight the importance of spatial context in microbiome interventions. Future approaches may benefit from compartment-tailored applications that align with naturally structured microbial communities. In addition, key taxa identified in this study, including UBA2421, Bradyrhizobium, and Talaromyces, represent promising candidates for functional validation and synthetic community design [50]. Together, these findings support the potential of Bacillin 20 as a tool for developing resilient, microbiome-based strategies for sustainable crop production systems. Limitations and future directions This study provides insights into the effects of Bacillin 20 on soybean-associated microbiomes; however, several limitations should be considered when interpreting the results. First, Bacillin 20 was applied as a foliar treatment, and thus the observed responses in root and rhizosphere microbiomes likely reflect indirect, plant-mediated effects rather than direct microbial exposure. Consequently, causal inference regarding belowground microbiome modulation remains limited in the absence of root-targeted controls. Second, the experimental design included co-application of herbicide in treatments T2-T5, whereas the control (T1) did not receive herbicide. This represents a potential confounding factor, as the independent effects of Bacillin 20 cannot be fully disentangled from herbicide application. Third, this study did not include plant phenotypic measurements, such as biomass, nutrient status, or nodulation, and therefore potential links between microbiome variation and plant functional outcomes could not be assessed. Fourth, field-based conditions and the relatively small sample size (n = 4 replicates per treatment) may limit statistical power to detect subtle treatment effects in complex microbial communities. In addition, while co-occurrence network analysis identified highly connected taxa, these findings remain inferential and were not supported by functional validation (e.g., targeted qPCR or culture-based assays). Finally, the observed fungal diversity patterns, with comparable or slightly higher diversity in roots relative to the rhizosphere, differ from commonly reported trends. This may reflect ecological context, sampling strategy, or sequencing depth, and warrants further investigation [51, 52]. Future studies should integrate microbiome profiling with plant phenotypic measurements and targeted functional assays, as well as multi-omics approaches, to better resolve the mechanisms by which Bacillin 20 influences plant–microbe interactions. Conclusion This study reveals that Bacillin 20 modulated the soybean root and rhizosphere microbiomes through niche-specific, compositionally targeted, and interaction-driven mechanisms. While spatial compartmentalization remained the primary determinant of community structure, bacillin 20 selectively enriched functionally relevant microbial taxa, including nitrogen-fixing bacteria and stress-adapted fungi and enhanced cross-domain connectivity, particularly via fungal network hubs. These effects suggested that bacillin 20 acted not merely as a microbial additive, but as a systems-level modulator capable of reshaping microbiome architecture toward enhanced plant-microbe cooperation. Our findings highlight the potential of Bacillin 20 as a biostimulant capable of influencing microbial community structure and interaction patterns while maintaining overall diversity. Although the present study does not provide direct mechanistic or plant performance data, the observed compartment-specific structuring and network-level modulation identify microbial interaction dynamics as a target for future investigation. Future work should focus on functional validation using targeted assays, metabolomic approaches, and multi-omics strategies to elucidate underlying mechanisms and determine their potential agronomic relevance. Such efforts will advance microbiome engineering toward precision agriculture frameworks that consider microbial network organization. Overall, provides a foundation for the rational use of biostimulants in microbiome-based strategies, while highlighting the need for further validation linking microbiome dynamics to plant outcomes. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Supplementary Material 5. Supplementary Material 6. Acknowledgements Authors are grateful to the molecular biology laboratory facilities at the Institut de Recherche en biologie vegetale (IRBV), Montreal. Abbreviations Linear discriminant analysis High-performance liquid chromatography revolutions per minute Authors’ contributions BA: conceptualization, experimentation, processing sequencing data, statistical analysis and draft the manuscript. MN: experimentation and sampling. RE: analysis of sequencing data. AK: draft and revision of the manuscript. MH: conceptualization, and manuscript revision. DS: conceptualization, fund acquisition, supervision, and manuscript drafting. All authors contributed to the article and approved the submitted version. Funding This research project was funded by CXC Montreal in Quebec, Canada. Data availability The datasets used in this study are available online with the accession number PRJNA1304989in the NCBI SRA database. Declarations Ethics approval and consent to participate This research does not include or report the results of a clinical trial; therefore, trial registration and ethical approval was not required. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Supplementary Material 5. Supplementary Material 6. Data Availability Statement The datasets used in this study are available online with the accession number PRJNA1304989in the NCBI SRA database.