Bacillus inoculation drives fine-scale taxonomic shifts in the bacterial communities of saline–alkaline wheat rhizosphere soil Authors: Huicheng Zhao, Linqi Zhang, Mengshuai Liu, Xinhui Fu, Meiyu Liu, Peng Guo, Wenliang Yang Published: 2026 ABSTRACT AbstractBackgroundSoil salinization severely restricts wheat production in the coastal regions of northern China, and inoculation with plant growth-promoting rhizobacteria (PGPR) is a promising strategy for alleviating salt stress. However, the response mechanisms of indigenous bacterial communities to PGPR remain unclear, partly because of the limitations of short-read sequencing in terms of taxonomic resolution. This study combined pot and field experiments with PacBio third-generation sequencing (TGS) to investigate bacterial community dynamics in rhizosphere and bulk soils after Bacillus inoculation.ResultsThe dominant phyla across treatments were Proteobacteria, Acidobacteria, and Bacteroidetes, with Proteobacteria being enriched in rhizosphere soils. Bacillus inoculation altered the bacterial community structure, increasing diversity and reducing intergroup dissimilarity, particularly in the pot experiments. Key ASVs responsive to inoculation were identified by intergroup difference analysis, including the enrichment of Pseudomonas spp. (ASV4007) in the inoculated treatments, which exhibited syntrophic interactions with Bacillus. Inoculation increased the abundance of functional genes, which were annotated using PICRUSt2, involved in phosphorus and nitrogen cycling and improved bacterial co-occurrence network resilience reflected by the natural connectivity.ConclusionsThis study demonstrated that Bacillus inoculation reshaped indigenous bacterial communities to mitigate salt stress, and TGS served as a powerful tool for identifying key functional taxa and supports the development of microbe-based saline soil remediation strategies.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40793-026-00882-y. Background Soil salinization, defined as the excessive accumulation of soluble salts in the root zone, acts as a critical constraint to global agricultural sustainability and impacts approximately 1 billion hectares worldwide, with arid, semiarid, and coastal regions being particularly susceptible [1]. Wheat (Triticum aestivum L.) production in coastal saline‒alkaline regions of northern China is highly climate dependent and relies primarily on rain-fed dryland farming [2]. However, salt stress significantly decreases the wheat germination rate and yield. Soil salinity severely impairs agricultural productivity by disrupting key plant physiological processes, including photosynthesis, respiration, and enzymatic synthesis, ultimately inhibiting growth or causing plant death [3]. Salinized land accounts for approximately 7% of the global terrestrial area, with nearly one-third of irrigated croplands threatened by secondary salinization [4, 5]. Globally, approximately 800 million hectares of land experience varying degrees of salinization [6], leading to substantial reductions in crop yield. Current mitigation strategies include breeding salt-tolerant winter wheat cultivars and implementing field desalination practices, such as freezing saline water irrigation [7], to reduce topsoil salinity. Recently, plant growth-promoting rhizobacteria (PGPR)-mediated approaches for alleviating salt stress have garnered increasing attention.To address the challenges associated with salinization, ecofriendly biotechnological approaches have garnered increasing attention, among which the application of plant growth-promoting rhizobacteria (PGPR) stands out [8–11]. The rhizosphere, a thin yet extraordinarily dynamic soil layer shaped by plant roots, serves as a pivotal hub for material cycling and energy flow among plants, microbes, and their shared environment [12]. Unlike bulk soil, this microcosm harbors uniquely structured microbial communities, which are driven primarily by the continuous supply of root-derived exudates [13]. Moreover, plant roots actively absorb water and nutrients from the rhizosphere, triggering physicochemical shifts (pH gradients and oxygen depletion) [14]. These changes, combined with biochemical cues from exudates, create niche-specific microenvironments that selectively enrich beneficial microbes and suppress pathogens [15]. For example, organic acids acidify the rhizosphere to solubilize iron, whereas mucilage traps beneficial microbes near root surfaces. This dynamic interplay sustains plant growth and drives rhizosphere microbiome diversity, underscoring the role of the rhizosphere in maintaining plant and soil health [16–18]. PGPR play prominent roles in promoting crop growth and increasing yield. As indispensable components of the plant rhizosphere ecosystem, PGPR can increase plant salt tolerance through plant‒microbe interactions [19, 20]. These beneficial bacteria directly stimulate plant growth by synthesizing plant-absorbable metabolites, facilitating nutrient acquisition, or modifying the rhizosphere microenvironment via the induction of gene transcription and metabolite biosynthesis in plant cells [21]. PGPR increase plant growth and crop productivity through multiple mechanisms, including biological nitrogen fixation, phosphate/potassium solubilization, phytohormone biosynthesis, siderophore production, ACC deaminase activity, antibiotic secretion, volatile organic compound (VOC) emission, the induction of systemic resistance, and the facilitation of beneficial symbioses [22, 23]. Among PGPR, some Bacillus spp. are particularly widely and effectively applied in agriculture because of their endospore-forming capacity, which confers superior adaptability to extreme pH, salinity, and other challenging soil conditions. Bacillus spp. thrive in the rhizosphere and promote plant development through multiple mechanisms, including increased nutrient absorption, phytohormone biosynthesis, siderophore production, antioxidant activity, exopolysaccharide (EPS) secretion, osmoprotectant synthesis, enzyme (1-aminocyclopropane-1-carboxylate deaminase) activity, and the induction of systemic resistance under salt stress [24–29]. For example, the in vitro application of indole-3-acetic acid (IAA)-producing Bacillus strains to plant roots under salinity has been shown to improve root architecture parameters [30]. In salt-stressed Solanum tuberosum L., inoculation with Bacillus spp. increased the K⁺ and Ca²⁺ contents and K⁺/Na⁺ and Ca²⁺/Na⁺ ratios while reducing Na⁺ absorption, leading to increased tuber yield [31]. Specific Bacillus species (including B. subtilis, B. licheniformis, B. thuringiensis, and B. amyloliquefaciens) can increase the levels of sugars, free amino acids, crude protein, and proline in plants. For example, the salt-tolerant strain B. subtilis RH-4 promotes seed germination and plant development in salt-inhibited chickpea plants by increasing the production of proteins, carbohydrates, and osmolytes (proline, glycine betaine, and choline) [32]. Additionally, B. subtilis improves salt tolerance in wheat by reducing the malondialdehyde content and increasing the activities of antioxidant enzymes [33]. While microbial inoculation has been shown to improve salt tolerance in wheat and other crops [34], the underlying mechanisms remain unclear. Crucially, systematic research on how indigenous bacterial communities respond to PGPR inoculation in saline–alkaline soils is lacking. Microbe-based soil fertility enhancement technologies currently lag far behind conventional nutrient management approaches in farmland applications [35]. Additionally, the limitations of next-generation sequencing (NGS), such as short read lengths (typically covering V3V4 or V4 regions) and inaccurate species identification, constrain comprehensive microbial community analyses [36, 37]. Third-generation sequencing, which allows long reads and high-resolution taxonomic identification, enables fine-scale bacterial community profiling and in-depth detection of fine-scale taxonomic shifts in saline–-alkaline wheat rhizosphere bacterial communties, overcoming the short-read sequencing limitations of microbial analysis. This study addresses these gaps by performing pot and field experiments with Bacillus spp. inoculation combined with full-length 16 S rRNA gene sequencing (V1–V9 regions) via PacBio circular consensus sequencing. This study aims to compare the bacterial composition between bulk soil and the wheat rhizosphere and identify keystone microorganisms that respond to microbial inoculants. The results revealed that Bacillus inoculation enriched Pseudomonas spp, which exhibited syntrophic interactions with Bacillus. Inoculation also increased the abundance of functional genes involved in phosphorus and nitrogen cycling and improved the resilience of the bacterial co-occurrence network. These findings provide novel insights for developing biotechnological strategies to increase salt tolerance, yield, and nutrient use efficiency in wheat production systems. Methods Experimental design and sampling The field experiment was conducted in a coastal saline zone in Huanghua County, Cangzhou city, China, on the western side of Bohai Bay (117°26′30″ E, 38°28′35″ N; Supplementary Fig. 1). The annual average temperature and precipitation at this location are 12.1 °C and 642 mm, respectively. The field treatments included (1) local traditional fertilization (Field-Uninoculated) and (2) local traditional fertilization combined with spraying a Bacillus inoculum (B. subtilis and B. velezensis, 1:1, isolated from saline soil and having salt-tolerant and plant-growth-promoting characteristics). The Bacillus powder (1010 cfu/g) was applied at a rate of 30 kg ha−1, applied by spraying with water. Each experimental plot measured 36 m², and each treatment had 3 replicate plots arranged randomly. During the jointing stage of wheat (Triticum aestivum L: Jiemai 19), the inoculum was diluted with normal saline and sprayed onto the wheat leaves. Rhizosphere and bulk soil samples were collected at the wheat harvest stage in 2024. In each field, the wheat samples were carefully excavated from the soil using a shovel. Root samples with adhered soil particles were collected from each plant, placed on dry ice for transportation, and stored at -80 °C until processing. Moreover, bulk soil samples were collected from unplanted plots using a 20 cm deep soil core sampler with a diameter of 4 cm. The pot experiments were performed in a greenhouse (25–30 °C with a 16 h light/8 h dark cycle) at the Center for Agricultural Resources Research, Chinese Academy of Sciences (CAS, Shijiazhuang). The soil used for the wheat pot experiment was collected from the corresponding field. The basic soil properties were as follows: the soil pH was 8.0, the electrical conductivity (EC) was 383, the soil organic matter (SOM) content was 22.4 g/kg, the soil total nitrogen (TN) content was 1.4 g/kg, the soil total phosphorus (TP) content was 0.9 g/kg, the soil total potassium (TK) content was 16.8 g/kg, the alkali-hydrolyzable nitrogen (AN) content was 95.4 mg/kg, the soil available phosphorus (AP) content was 15.5 mg/kg, and the soil available potassium (AK) content was 322.4 mg/kg. The pot treatments included (1) no inoculation (Pot-Uninoculated) and (2) inoculation with a Bacillus inoculum (Pot-Inoculated, B. subtilis and B. velezensis, 1:1; the Bacillus powder (1010 cfu/g) was applied at a rate identical to that used in the field experiments). The same wheat cultivar was used in both the field experiments and the greenhouse experiments. Each treatment included 3 pots as replicates. The pot experiment lasted for one month, after which the rhizosphere and nonrhizosphere soils of the wheat plants were collected. The soil particles attached to the roots after gentle shaking were defined as rhizosphere soil. The attached soil particles were removed by brushing off and sieved through a 2 mm pore size sieve. A portion of each bulk soil sample was sieved through a 2 mm mesh to remove visible organic matter and stones, and the soil samples were air-dried to remove excess moisture before DNA extraction. Measurement and determination of soil physicochemical properties The soil properties were measured following methods described in previous studies [38, 39], and the specific research methods are described below. The EC of the soil samples was measured in a 1:5 (wt/vol) mixture of moist soil and boiled deionized water using a DDS-307 A conductivity meter (INASA Scientific Instrument Co., Ltd., China). The soil pH was assessed with an S220 Seven Compact pH meter (Mettler Toledo, Switzerland) at a soil:water ratio of 1:2.5 (dry weight/volume). The SOM content was determined using the dichromate oxidation method, and the TN content was measured using the Kjeldahl approach. The TP content was determined by alkali fusion combined with molybdenum-antimony-anti-colorimetry. The TK content in the soil was determined by alkali fusion combined with flame photometry. The AN content in the soil was determined using the diffusion method. AP and AK were extracted from the soil after digestion with 0.5 M NaHCO3 (pH = 8.5) and 1 M NH4OAc, respectively, and their content was measured using the molybdenum blue method and flame spectrophotometry (FP640, INASA, China). Molecular and bioinformatic analyses Total genomic DNA was extracted using an OMEGA Soil DNA Kit (M5635-02; Omega Bio-Tek, Norcross, GA, USA) following the manufacturer’s protocol. Blank extraction controls were established during nucleic acid extraction by substituting the sample with sterile ultrapure water while strictly following the identical extraction protocol used for the experimental samples. The concentration and purity of the extracted DNA were determined using a NanoDrop One system (NanoDrop Technologies, Wilmington, DE, USA), and the DNA was stored at -20°C until molecular analysis. PCR amplification of the nearly full-length bacterial 16S rRNA genes was performed using the forward primer 27F (5’-AGAGTTTGATCMTGGCTCAG-3’) and reverse primer 1492R (5’-ACCTTGTTACGACTT-3’) [40]. The PCRs were conducted in a 25 µL system with a Q5 High-Fidelity DNA Polymerase Kit (NEB, #M0491L) under the following conditions: initial denaturation at 98 °C for 2 min; 25 cycles of 98 °C for 30 s, 55 °C for 30 s (annealing), and 72 °C for 90 s; and a final extension at 72 °C for 5 min. Negative controls were established by substituting the templates with sterile ultrapure water in the PCR system. Amplicons were sequenced using single-molecule real-time (SMRT) technology on the PacBio Sequel platform. Bioinformatic analysis was primarily performed using Quantitative Insights Into Microbial Ecology (QIIME 2) [41]. First, adaptors and barcodes were trimmed using cutadapt [42], and paired-end reads were denoised with the DADA2 method [43] to filter noisy sequences, remove chimeras, merge denoised reads, and dereplicate sequences. Taxonomy was assigned to amplicon sequence variants (ASVs) using the q2-feature-classifier plugin of QIIME 2 [44] against the Silva database (v138) [45] for the bacterial 16 S rRNA gene. Singletons and ASVs unassigned to the bacterial kingdom were removed. The Shannon and PD (Faith’s phylogenetic diversity for evolutionary breadth) indices were calculated to compare the alpha diversity of the bacterial communities in the different treatments, and dissimilarity index (to analyze intergroup divergence) was also calculated on the basis of Bray‒Curtis distances. Principal coordinate analysis (PCoA) based on Bray‒Curtis distances was performed to determine the differences in community composition among the treatments. Rank abundance curves depicting the distribution of the relative abundance (log2 transformed) of amplicon sequence variants (ASVs) were constructed; in the rank curve, a longer horizontal span of the curve indicates higher species richness in the sample, and a flatter curve indicates a more uniform species composition. The functional profiles of the bacterial communities were annotated using PICRUSt2 [46]. Statistical analyses Statistical analyses were performed in R (version 3.4.3) using the appropriate packages. The neutral community model (NCM) [47] was used to quantify the role of migration in shaping bacterial community structure and compare community assembly patterns using the “neutralitytestr” package [48]. The taxonomic composition of bacterial species was significantly enriched in these treatments on the basis of FDR-adjusted P values < 0.05 calculated using the “DESeq2” package [49]. The relative abundance of functional genes between the two treatments and the distributions of bacterial community diversity, including the Shannon, PD (Faith’s phylogenetic diversity for evolutionary breadth) and dissimilarity (for intergroup divergence) indices were compared with the Wilcoxon test on the basis of P < 0.05 with the R “stats” base function. Network analysis was performed using the SparCC algorithm with the “SpiecEasi” package [50]. Cooccurrence networks were constructed to reflect positive correlations (edges) between ASVs (nodes) with a SparCC rank coefficient > 0.6 and a Benjamini‒Hochberg adjusted P value < 0.05 and were visualized using the “igraph” package [51]. Permutational multivariate analysis of variance (PERMANOVA) was performed using the “vegan” package [52]. Results Bacterial communities in rhizosphere and bulk soils differed between the pot and field treatments Using PacBio third-generation sequencing (TGS), we investigated the bacterial community dynamics in rhizosphere and bulk soils after Bacillus inoculation in combined pot and field experiments, which included eight sample groups (PCKBS, PBBS, PCKRS, PBRS, FCKBS, FBBS, FCKRS, and FBRS, where P represents pot treatment, F represents field treatment, B represents Bacillus inoculation, CK represents uninoculated treatment, BS represents the bulk soil, and RS represents the rhizosphere soil). The most abundant bacterial phyla were Proteobacteria, Acidobacteria, Bacteroidetes, Gemmatimonadetes, Actinobacteria, Verrucomicrobiota, Planctomycetes, Firmicutes, Myxococcota, and Chloroflexi, which accounted for 96.3–99.9% of all bacterial taxa (Fig. 1a). Proteobacteria was enriched in the rhizosphere soil samples in both the pot and field treatments. To investigate the separation of bacterial communities among all the samples, we further visualized the Bray–Curtis distances using principal coordinate analysis (PCoA) (Fig. 1b). Bray–Curtis distance-based ordination indicated that the bacterial community assemblages were nonrandom and structured by group-specific determinants. The results clearly revealed that soil bacterial community separation was driven by niche (bulk soil and rhizosphere, R2= 0.242, P = 0.001), experiment type (pot and field, R2= 0.239, P = 0.001) and inoculation (Bacillus and noninoculated, R2= 0.054, P = 0.252) according to the PERMANOVA results. We also compared bacterial community diversity indices, including the Shannon diversity, PD (Faith’s phylogenetic diversity for evolutionary breadth), and dissimilarity (for intergroup divergence) indices (Fig. 1c). These results revealed extremely significant overall differences among the groups, confirming the strong regulatory effect of the different treatments on the bacterial communities. Bacterial communities of the rhizosphere and bulk soils in the pot and field treatments. a Taxonomic composition of the bacterial communities in the eight sample groups (PCKBS, PBBS, PCKRS, PBRS, FCKBS, FBBS, FCKRS, and FBRS, where P represents pot treatment, F represents field treatment, B represents Bacillus inoculation, CK represents uninoculated treatments, BS represents the bulk soil, and RS represents the rhizosphere soil). b Principal coordinate analysis (PCoA) plot of the bacterial community on the basis of the Bray‒Curtis distance. c Boxplots of the distributions of bacterial community diversity, including the Shannon, PD (Faith’s phylogenetic diversity for evolutionary breadth) and dissimilarity (for intergroup divergence) indices. Statistical significance is denoted by * (P < 0.05), ** (P < 0.01), and *** (P < 0.001) The Shannon index, which reflects species richness and evenness, showed marked intergroup differences. Samples such as PCKBS, PBBS, PCKRS, PBRS, and FCKBS presented relatively high Shannon diversity indices (median values > 8), whereas FBBS, FCKRS, and FBRS presented reduced diversity. FBRS had the lowest Shannon index (median value ~ 6) with a narrow interquartile range, indicating low species richness and evenness. Assessment of phylogenetic diversity (PD), a measure of evolutionary richness, exhibited a similar trend: groups PCKBS and PBBS maintained high PD values (median ~ 35–40), whereas groups FCKRS and FBRS showed substantial declines. Compared with the other groups, the FBRS group presented the lowest PD value (median ~ 15), suggesting a depauperate phylogenetic composition. The dissimilarity index (Bray–Curtis distance) revealed that the dissimilarity of the FBRS group was the greatest (broad interquartile range, median ~ 0.45), indicating greater divergence from the other groups. In contrast, PCKBS, PBBS, and PCKRS presented lower dissimilarity, indicating the existence of more conserved community structures within these groups. Our multilevel analysis revealed the variations in bacterial community composition, structural organization, and diversity across the rhizosphere and bulk soils in the pot and field treatments. The bacterial community dynamics differ between the Bacillus-inoculated and uninoculated treatments The rank–abundance curves and frequency–abundance distributions collectively revealed the richness, evenness, and community assembly patterns of bacterial communities in the Bacillus-inoculated and uninoculated treatments. The rank‒abundance curves (Fig. 2a) show ASVs ranked by decreasing relative abundance (log2-transformed, y-axis). Compared with those of the uninoculated groups (blue), the curves of the inoculated groups (red) were more extended at lower relative abundances (higher ranks). In the curves of the inoculated treatments, the smooth downward slope and the longer, flatter span on the abscissa indicated that Bacillus inoculation increased species richness and the evenness of the species composition. In contrast, the uninoculated treatments produced steep downward curves, suggesting a high proportion of dominant bacterial groups and low diversity. This pattern implied greater taxonomic richness and potentially greater evenness in the inoculated communities, as ASVs were more uniformly distributed across abundance ranks, indicating that Bacillus inoculation profoundly influenced the compositions of the bacterial communities. Bacterial communities in the rhizosphere and bulk soils subjected to Bacillus inoculation and uninoculated soils. a Rank abundance curves depicting the distribution of the relative abundance (log2 transformed) of amplicon sequence variants (ASVs) in the Bacillus-inoculated (red) and uninoculated (blue) groups. The abscissa represents the number of ASVs sorted by abundance from high to low, and the ordinate represents the abundance of ASVs. The abscissa value at the extension endpoint of the sample curve is the number of ASVs in that sample. b Abundance occurrence relationships for the Bacillus-inoculated and uninoculated treatments, with the mean relative abundance (log10-transformed) and frequency of occurrence. The blue solid line represents the fitted model, and the dashed lines denote variability We used the neutral community model (NCM) [47] to quantify the role of migration in shaping bacterial community structure and compared community assembly patterns influenced by Bacillus inoculation. Relationships between the mean relative abundance (log10, x-axis) and occurrence frequency (y-axis) are shown in Fig. 2b. The Bacillus-inoculated group exhibited a weaker model fit (R² = 0.455) than the uninoculated group did (R² = 0.515), indicating that the abundance of ASVs in the inoculated groups was less consistent across samples. This discrepancy suggested that Bacillus inoculation introduced greater stochasticity in the taxonomic distribution, in contrast with the more predictable abundance‒frequency relationships in uninoculated systems. A comparative analysis of the differences in bacterial community dynamics between the Bacillus-inoculated and uninoculated treatments highlighted that Bacillus inoculation altered both the internal abundance structure (rank abundance) and cross-sample distribution patterns (frequency–abundance) of the bacterial communities. Collectively, these findings emphasize that Bacillus inoculation perturbs the intrinsic structure of the local microbiome. Key ASVs are enriched in Bacillus-inoculated vs. uninoculated treatments We identified bacterial community shifts between the Bacillus-inoculated and uninoculated treatments in the rhizosphere and bulk soils by leveraging ASV-level proportional dynamics and rigorous statistical inference. We also determined the taxonomic composition of bacterial species that were significantly enriched in these treatments (FDR-adjusted P values < 0.05). The left panel of Fig. 3, which shows the mean ASV proportions, reveals pronounced disparities in composition. The Bacillus-inoculated treatments (shown in red) demonstrated taxonomic skewing, with specific ASVs (for instance, ASV1201, ASV4007, and ASV152) exhibiting increased relative abundances, indicating that inoculation enabled these taxa to occupy distinct ecological niches. In contrast, the uninoculated treatments exhibited a distinct assemblage characterized by the prevalence of indigenous ASVs (ASV4178, ASV286, and ASV1076). This phenomenon confirms that exogenous Bacillus introduction altered the ecological niches of the native bacterial community. The right panel of Fig. 3, which integrates the mean proportional differentials with 95% confidence intervals and FDR-adjusted P values, provides robust statistical validation. For example, inoculation-associated ASVs (ASV1201, FDR = 0.0158) presented significant positive effect sizes with nonoverlapping confidence intervals, reliably confirming their enrichment. Collectively, these analyses reveal the Bacillus inoculation-driven shifts in bacterial taxon abundance, identifying significantly enriched/depleted ASVs with statistical support, which is critical for elucidating microbiome responses to exogenous Bacillus input. These findings highlight inoculation as a potent driver of bacterial community reconfiguration and targeted enrichment. Proportions of key amplicon sequence variants (ASVs) between the Bacillus-inoculated and uninoculated treatments. The left panel shows the quantified group-specific ASV enrichment, contrasting the ASV abundance profiles of the inoculated (red) and uninoculated (blue) bacterial communities. The right panel illustrates the difference in mean proportions with 95% confidence intervals and associated P values Key ASVs enriched in Bacillus-inoculated vs. uninoculated treatments differ in terms of their taxonomic characteristics To identify the key ASVs enriched in the Bacillus-inoculated vs. uninoculated treatments, we elucidated the taxonomic compositions of the bacterial species significantly enriched in the Bacillus-inoculated and uninoculated treatments in the rhizosphere and bulk soils (FDR-adjusted P values < 0.05) through alluvial-based taxonomic profiling. In the Bacillus-inoculated treatments (Fig. 4a), Proteobacteria was the dominant phylum. Within Proteobacteria, Gammaproteobacteria and Pseudomonadales emerged as pivotal taxa, cascading into specific genera (such as Pseudomonas). A previous study confirmed that plant-beneficial indigenous Pseudomonas species exhibit syntrophic cooperation with Bacillus inoculants in the plant rhizosphere [53]. In contrast, the uninoculated treatments were dominated by taxa such as Rhizobiales (Fig. 4b), a taxon with well-documented nitrogen-fixing traits. These alluvial diagrams allow visualization of the quantified taxonomic flow, revealing that the Bacillus-inoculated treatments induced significant shifts; they prioritized Pseudomonadales-affiliated ASVs for potential plant‒microbe interaction functions. For example, the taxonomic streamlining driven by Bacillus inoculation may enhance targeted biogeochemical cycling. In summary, Bacillus inoculation reconfigured bacterial taxonomic hierarchies, restructuring bacterial community assembly from the phylum level to the ASV level. The diagrams reveal distinct bacterial community structures between the two groups, highlighting the inoculation-driven shifts in taxonomic composition and the differential distribution of specific ASVs across hierarchical levels. Taxonomic composition of key amplicon sequence variants (ASVs) in the Bacillus-inoculated and uninoculated treatments. Each colored band represents a taxonomic rank (phylum, class, order, family, genus, or ASV), and the lengths of the connections reflect the relative abundance of the taxon Key functional gene prediction reveals that Bacillus spp. provide key phosphorus and nitrogen metabolic functions To compare the functional genes involved in the soil phosphorus and nitrogen cycle, we perdormed functional gene prediction analysis. The bar chart shows the relative abundance of key functional genes associated with phosphorus and nitrogen cycling processes, including inorganic P solubilization, organic P mineralization, nitrogen fixation, nitrification, denitrification, and dissimilatory nitrate reduction to ammonium (DNRA), in the Bacillus-inoculated and uninoculated treatments. Genes involved in inorganic P solubilization (pqq, ppx, and ppa) and organic P mineralization (phoN, phn, kdsC, and glpK) presented significantly different relative abundance patterns between the two treatments (P < 0.05) (Fig. 5a). Specifically, the relative abundances of certain inorganic P solubilization genes (pqqE, pqqD, pqqB, and ppx) and organic P mineralization genes (phn and glpK) were significantly greater in the Bacillus-inoculated samples than in the uninoculated samples (P < 0.05). These findings indicate that Bacillus inoculation significantly influences the expression of these genes, potentially enhancing the associated biological processes. With respect to nitrogen cycling, genes such as nif (nitrogen fixation), amo and hao (nitrification), nap and norC (denitrification), and nir (DNRA) also presented significantly different patterns in relative abundance between the two treatments (P < 0.05) (Fig. 5b). Notably, the abundances of nitrogen fixation genes (specifically, nif), denitrification genes (nap and norC), and DNRA genes (nirB) were significantly greater in the Bacillus-inoculated samples (P < 0.05). The effect of Bacillus inoculation on the relative abundances of these functional genes suggests that Bacillus spp. could modulate the potential of bacterial community-mediated phosphorus and nitrogen cycling in the studied system. Bar charts comparing the abundances of functional genes between the Bacillus-inoculated and uninoculated treatments. Genes involved in the phosphorus a and nitrogen b cycles in the rhizosphere and bulk soils, as predicted by PICRUSt2. Statistical significance is denoted by * (P < 0.05), ** (P < 0.01), and *** (P < 0.001) Bacillus inoculation improves bacterial co-occurrence network resilience We also conducted a co-occurrence network analysis to identify the relationships in the bacterial community in the Bacillus-inoculated and uninoculated treatments in both the rhizosphere and bulk soils. Four independent co-occurrence network diagrams were constructed: inoculated (bulk soil), uninoculated (bulk soil), inoculated (rhizosphere), and uninoculated (rhizosphere), which revealed differences in the niches between the bulk soil and rhizosphere soil (Fig. 6a). Natural connectivity analysis (Fig. 6b) was used to assess network resilience upon node removal. Compared with the uninoculated treatments, the Bacillus-inoculated treatments maintained greater connectivity across all node removal gradients, indicating that Bacillus inoculation induced structural shifts to increase bacterial community stability. Specifically, after 45 nodes were removed, the inoculated network retained a connectivity of ~ 0.1, whereas that of the uninoculated network decreased to below 0.05, suggesting that inoculation reinforced keystone species interactions or modular redundancy. These results demonstrate that Bacillus inoculation shapes bacterial community interaction patterns and strengthens network robustness against structural perturbations. Additionally, it serves as a taxonomic blueprint for refining bacterial inoculation strategies in agroecosystems, where tailored communities could increase crop productivity and soil resilience. Bacterial co-occurrence network of amplicon sequence variants (ASVs) in the Bacillus-inoculated and uninoculated treatments. a Cooccurrence network topology. Each network node (single circle) represents one ASV. Network edges are shown as straight lines connecting the nodes and indicate significant cooccurrences on the basis of SparCC correlation analysis (r > |0.06|, P < 0.05). b Natural connectivity dynamics in the rhizosphere and bulk soils of the Bacillus-inoculated and uninoculated treatments Discussion PGPR inoculation improves soil health in saline soil Salinity stress severely compromises soil biological activity, depletes fertility-promoting microorganisms, and reduces microbial diversity, thereby disrupting soil aggregate formation and biogeochemical cycling [54]. Under high-salinity conditions, halotolerant microbes outcompete other taxa for resources, leading to the loss of beneficial soil microorganisms, decreased microbial diversity, and a degraded community structure [55, 56]. Restoring microbial biodiversity and soil functionality is therefore crucial for establishing sustainable agricultural systems in saline–alkaline soils. PGPR strains are recognized for their safety, efficiency, resource conservation ability, and environmental friendliness, demonstrating broad application prospects in agriculture. These microorganisms play pivotal roles in sustainable agricultural development: they promote soil formation and development, drive biogeochemical cycling, maintain soil structure and fertility, and improve plant health and stress resistance. PGPR inoculation addresses these challenges by reintroducing key microbial functions, enhancing soil biological health, and imporving the development of fertile tillage layers in saline–alkaline soils, ultimately supporting sustainable crop production in these marginal environments. Improving soil microbial diversity and promoting the fertilization of microbial communities through the use of microbial agents has become a research hotspot in saline–alkaline land improvement [57]. However, conventional microbial agents have a limited influence on overall biological networks because of their environmental sensitivity, poor stability, and low colonization capacity. Our results revealed that the bacterial community PD index was significantly greater in the pot bulk soil than in the field treatments, and the effects of Bacillus inoculation on the pot bulk soil were more clearly distinguished from those in the field trials. The observed differences between the field and pot experiments were primarily driven by the high environmental heterogeneity of field soils, including the spatial variability of soil physicochemical properties, natural microbial colonization and field management practices, which inevitably weakened the regulatory effect of exogenous Bacillus inoculation compared with that under controlled pot conditions. Although pot experiments clarified the direct effects of Bacillus inoculation on bacterial communities, the field trials better reflected its actual application potential in saline–alkaline wheat fields, highlighting the need for field validation of the results of laboratory-based inoculation studies. Notably, the specialized metabolic functions of keystone taxa sustain soil microbiome stability, and our study indicated that tailored microbial communities can increase the resilience of native bacterial communities in agricultural ecosystems (Fig. 6b). These findings provide insights for bacterial community management to optimize ecosystem functions, offering a framework for designing bacterial inoculants to increase nutrient cycling or stress resistance in agricultural systems [58, 59]. Therefore, there is an urgent need to exogenously introduce PGPR that can activate indigenous microorganisms to exert their functions, thereby restoring biodiversity in saline–alkaline lands, reconstructing multitrophic biological networks, achieving biological soil reformation, and promoting the long-term maintenance of soil biological health and the formation of suitable tillage layers in saline-alkaline lands [60]. Therefore, improving saline–alkaline soils is a gradual process that requires long-term, direct, and continuous application of PGPR to improve soil quality. Systematic identification of PGPR-inducible indigenous microbes acts as an essential bridge The identification of PGPR-responsive indigenous microorganisms serves as a key link between basic research and practical applications, with its core significance lying in the efficient utilization of indigenous microbial resources to achieve the goals of increased agricultural production, ecological protection, and economic sustainability. Specifically, indigenous microorganisms are microbial communities that have coevolved with local plants and soils over time that exhibit greater adaptability to and colonization of local environments (soil types, climatic conditions, and crop varieties). Identifying indigenous microorganisms responsive to PGPR enables the screening of strains with strong matches to local crops; through inoculation or in situ activation, this can significantly improve the colonization efficiency of PGPR, thereby more effectively promoting crop growth (including root development and nutrient uptake) and increasing yields [61–62]. Moreover, PGPR can replace some fertilizer functions by fixing nitrogen, solubilizing phosphorus and potassium, and secreting phytohormones, which tend to be more stable in indigenous strains. Additionally, some indigenous microorganisms can reduce disease occurrence by antagonizing pathogens by secreting antibiotics or inducing systemic resistance [63], thereby reducing pesticide use and increasing the safety of agricultural products by lowering pesticide residue levels. We used the NCM to quantify the roles of migration and chance in shaping the bacterial community structure [64] and used rank‒abundance curves to depict the distribution of the relative abundance (log2-transformed) of ASVs in the Bacillus-inoculated group (red) and uninoculated group (blue). Our results revealed that the inoculated group presented altered dominance dynamics, evidenced by the rank‒abundance curves, and attenuated abundance‒occurrence coupling, likely due to the introduction of exogenous microbial taxa. In contrast, the uninoculated group retained a more “pristine” state, where abundance and occurrence showed a stronger inherent correlation. These insights improve our understanding of how exogenous microbial inputs reconfigure community assembly rules, a critical consideration for microbiome manipulation in agroecosystems (for instance, via biofertilizer application) [65]. These findings provide more targeted solutions to address challenges such as global food security and climate change. Moving forward, studies should endeavor to disentangle the functional ramifications of these structural perturbations and establish links between microbial community architecture and ecosystem-level processes. For example, coupling these taxonomic insights with metagenomic functional analyses can elucidate trait-driven community responses [66]. Similarly, screening for salt-tolerant indigenous PGPR in saline areas can increase crop growth under stress, facilitating the translation of laboratory discoveries into effective agricultural practices for saline soil remediation. Future investigations should also integrate metatranscriptomic and metabolomic datasets to resolve the functional implications of these compositional shifts, thereby bridging taxonomic reconfiguration with ecosystem-scale processes. Third-generation sequencing enables the precise identification of indigenous microorganisms that are responsive to PGPR This study focused on deciphering bacterial community dynamics in rhizosphere/bulk soils under pot and field conditions, particularly by exploring shifts induced by inoculation with Bacillus spp. (typical PGPR), including changes in taxonomic composition, ASV-level dynamics, functional gene profiles, and co-occurrence network resilience. Third-generation sequencing (TGS) technologies (PacBio and Nanopore) have unique advantages in addressing these research objectives, as supported by the findings of previous studies [67, 68]. TGS generates reads spanning the entire 16 S rRNA gene (10–18 kb), avoiding the assembly errors of hypervariable regions inherent to short-read sequencing. In fact, long read lengths increase the taxonomic resolution for closely related species. This capability is critical for distinguishing phylogenetically similar but functionally distinct indigenous species, a key challenge in this study. For example, the present study revealed Pseudomonas as a pivotal genus enriched in the inoculated treatments (Fig. 4a), which exhibited syntrophic cooperation with the inoculated Bacillus spp. These Pseudomonas strains often share high sequence similarity in partial 16 S regions, but TGS resolves their fine-scale genetic differences, enabling accurate classification at the ASV level (ASV1201 and ASV4007 in Fig. 3). In this study, the inoculated treatments exhibited enrichment in specific taxa such as Pseudomonadales, whereas the uninoculated treatments retained unique indigenous ASVs (such as ASV4178). By avoiding the overestimation of dominant groups or underestimation of rare taxa, TGS ensures reliable quantification of fine-scale shifts in abundance. This accuracy underpins the study’s analysis of diversity indices (Shannon, PD; Fig. 1c) and dissimilarity patterns, revealing that the inoculated treatments better maintained conserved community structures than the uninoculated treatments did [69]. TGS achieves > 99% accuracy via circular consensus sequencing, enabling precise ASV clustering and supporting the robust ASV level analysis in this study. ASVs represent the finest-resolution taxonomic units, and the reliable identification of ASVs is critical for dissecting inoculation-induced community reconfiguration. The study leverages ASV-level data to identify significantly enriched taxa (FDR-adjusted P < 0.05; Fig. 3) and track shifts in the rank‒abundance curves (Fig. 2a). TGS-derived high-quality sequences were seamlessly integrated with the multiomic and ecological analysis results in this study. For example, TGS-resolved ASVs served as nodes in the co-occurrence network analysis (Fig. 6), revealing that inoculation resulted in the formation of more resilient networks with stronger interactions among keystone species [70, 71]. Additionally, the ability of TGS to capture full-length 16 S sequences supports phylogenetic diversity (PD) analysis (Fig. 1c), clarifying that inoculated treatments retain higher evolutionary richness—a pattern masked by short reads owing to incomplete phylogenetic information. In summary, TGS overcomes the limitations of short-read sequencing in terms of taxonomic resolution, quantification accuracy, and ASV-level reliability, providing a technical foundation for this study for deciphering PGPR-induced bacterial community shifts, identifying key indigenous species, and linking community structure to ecosystem functions, ultimately offering insights for advanced mechanistic models of microbiome modulation and informing precision microbiome engineering in agrobiotechnology. In terms of technical application, although TGS has shown significant advantages in this study, this study still has certain constraints. For example, the cost of TGS is relatively high compared with that of short-read sequencing, which may limit its large-scale application in extensive soil sample studies. In addition, processing TGS data requires more computational resources and more complex bioinformatics analysis pipelines, which may pose challenges for some research teams with limited technical resources. In terms of optimizing technical applications, efforts can be made to reduce the cost of TGS and simplify the data processing process. For instance, developing more efficient sequencing protocols and bioinformatics analysis tools can promote the widespread use of TGS in soil microbiome research. Conclusions This study confirms that Bacillus inoculation can regulate soil microbial communities to ameliorate saline–alkaline soils and highlights the irreplaceable role of third-generation sequencing (TGS) in deciphering soil microbiome dynamics. In terms of practical applications, the identification of PGPR-responsive indigenous microbes (Pseudomonas spp.) not only provides a theoretical basis for targeted microbial inoculant development but also bridges the gap between laboratory research and field practices. This means that in saline–alkaline regions where agricultural production is constrained by soil conditions, the adoption of Bacillus-based microbial management strategies to activate indigenous microbial functions may reduce the reliance on chemical fertilizers and align with the global trend of eco-friendly agriculture. It would also be valuable to explore the molecular mechanisms through which key enriched taxa interact with Bacillus and host plants, as this could deepen our understanding of the synergistic effects of microbes in rhizospheres. Overall, the results of this study confirm the potential of PGPR inoculation to ameliorate saline–alkaline soils through microbial community regulation and underscore the value of TGS in deciphering complex soil microbiome dynamics, offering practical insights for ecofriendly agricultural practices in saline regions. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements This study was supported by the Chinese Academy of Sciences. We acknowledge Xinzhen Wang and Baozhi Yang for their assistance with field sampling. Abbreviations Alkali-hydrolyzable nitrogen Available phosphorus Available potassium Amplicon sequencing variant Deoxyribonucleic acid Exopolysaccharide Electrical conductivity False discovery rate Indole-3-acetic acid Next Generation Sequencing neutral community model Plant growth-promoting rhizobacteria Principal coordinates analysis Phylogenetic diversity Soil organic matter Total nitrogen Total phosphorus Total potassium Third-generation sequencing Volatile organic compound Author contributions H. Z, P. G, and W. Y conceived and designed the project. H. Z, M. Y. L, and X. F performed the experiments, conducted the microbiological analysis, and created the graphs. H. Z wrote the paper. L. Z, M. S. L, P. G, and W. Y reviewed the paper. Funding This study was supported by the Chinese Academy of Sciences. Data availability The sequencing data of this study have been deposited in the National Center for Biotechnology Information (NCBI) under BioProject number PRJNA [1293161](https:/www.ncbi.nlm.nih.gov/bioproject/PRJNA1293161) , and the code is available at Github: https://github.com/huichengzhao/Taxonomic-changes-in-saline-alkaline-wheat-bacterial-communities-driven-by-Bacillus-inoculation. Declarations Ethics approval and consent to participate Not applicable. 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. Contributor Information Huicheng Zhao, Email: hczhao@sjziam.ac.cn. Peng Guo, Email: guopeng@hebust.edu.cn. Wenliang Yang, Email: wlyang@issas.ac.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The sequencing data of this study have been deposited in the National Center for Biotechnology Information (NCBI) under BioProject number PRJNA [1293161](https:/www.ncbi.nlm.nih.gov/bioproject/PRJNA1293161) , and the code is available at Github: https://github.com/huichengzhao/Taxonomic-changes-in-saline-alkaline-wheat-bacterial-communities-driven-by-Bacillus-inoculation.