This study takes an integrated approach to evaluate different bacterial-based treatments for improving crop growth and performance in common bean as a model crop. Different formulations, including live bacterial cells, cell-free extracts, consortia, and combinations with reduced fertilizer inputs, were evaluated. Treatments were assessed across crop developmental stages and experimental conditions and at different biological response levels. Overall, treatment-dependent effects were observed in seed germination, seedling vigor, plant growth, reproductive development, and physiology, with responses varying by strain, formulation, and plant developmental stage. This work moves beyond a single application strategy or response endpoint by examining multiple complementary dimensions of bacterial treatment performance within one integrated study, providing a broader view of both their practical versatility and the factors that shape their efficacy. This evaluation provides a more comprehensive framework for understanding the potential and limitations of bacterial-based biostimulants and demonstrates that treatment performance cannot be reliably inferred from isolated early-stage responses or individual crop parameters alone. Instead, efficacy is shaped by multiple interacting factors, such as the continuity of crop responses across development and the persistence of treatment effects under agronomically relevant conditions.

Biopriming of seeds with bacterial cultures showed accelerated early germination processes (Figure 1A), although all treatments achieved similar germination percentages after additional incubation. Similar responses have been reported with PGPRs accelerating the timing of radicle emergence in seeds, without altering the final germination percentages (Rozier et al., 2019). Accelerated early germination processes may reflect enhanced mobilization of seed reserves. Consistent with this, genomic annotation of the tested bacterial strains identified hydrolytic enzyme-related genes including phytases and amylases (Kaddoura et al., 2026). These can mobilize seed reserves by hydrolyzing starch and phytate, releasing soluble sugars and inorganic phosphate required for early physiological processes during germination (Al Hijab et al., 2024). Bacterial treatments also improved seedling vigor parameters. The treatments enhanced root system development during early seedling growth through coordinated increases in root length and biomass, indicating structural development rather than simple elongation (Figures 1B–D). Notably, BS-114 displayed more pronounced SRL values compared with PPW-26. Bacterial treatments induced distinct root architectural responses through which similar improvements in seedling vigor were achieved (Figures 1B, G). BS-114 seedlings displayed a stronger elongation-oriented response, whereas PPW-26 exhibited a more balanced pattern between elongation and structural development. Distinct root architectural strategies suggest early modulation of developmental and hormonal regulation pathways. Similar observations were reported in soybean, where inoculation with B. subtilis promoted root development and altered its architecture (Moraes et al., 2025). These responses are consistent with the previously characterized ability of the tested strains to produce indole-related compounds, which may contribute to modulation of root developmental processes (Kaddoura et al., 2026). Moreover, culture-based recovery, followed by validation against the expected MLST allele sequences, provided strong evidence that BS-114 internally colonized common bean roots during early seedling establishment. However, the persistence of BS-114 beyond this early developmental stage was not evaluated, and colonization was not confirmed for BS-120 or PPW-26. Importantly, the absence of confirmed internal colonization does not rule out plant growth-promoting activity, as beneficial bacteria may still act through root-surface interactions. Endophytic colonization may provide additional routes of interaction within plant tissues; however, the present study design cannot determine whether the effects of BS-114 were dependent on internal colonization or root-surface activity. Together, these findings indicate that bacterial treatments influenced early development and root architecture during seedling establishment.

Under controlled greenhouse conditions, treatment effects were expressed differently across vegetative and reproductive stages. Plant height showed a stronger response than stem diameter, while total fresh mass did not increase consistently across bacterial treatments (Figures 2D, E). This suggests that vegetative responses are trait-specific and cannot be interpreted simply as overall growth stimulation. Additionally, the LB control provided important context for separating nutrient-driven growth from developmental performance associated with the bacterial treatments. Although LB promoted vegetative fresh mass, this response was not accompanied by proportional improvements in reproductive development. Similarly, NPK enhanced some vegetative traits but did not consistently produce the strongest reproductive outcomes.

Together these observations indicate that nutrient-driven vegetative responses were not sufficient to sustain reproductive performance and that treatment effects extended beyond nutrient availability alone. Bacterial treatment effects were more strongly evident at reproductive development stage. BS-114 and to a lesser extent PPW-26, promoted earlier reproductive progression, as reflected by the earlier pod formation observed at week 6 and the higher reproductive structure formation during the initial reproductive stages (Supplementary Figure 1). This suggests that bacterial treatments advanced the onset of reproductive development, with early differences maintained through later stages. These patterns of increased reproductive structures and reduced time to onset are consistent with other PGPR reports (Kumari et al., 2015; Lyu et al., 2022). However, early increases in reproductive structures were not fully retained, as reflected by the differences between immature and mature pod counts (Figures 3C, D). This likely reflects reproductive attrition during development, yet these early advantages continued to shape final yield outcomes (Figures 3E, F). Yield responses were reflected in multiple reproductive components rather than a single trait. Increases in cumulative pod fresh mass under bacterial treatments were accompanied by increases in both pod number and average pod fresh mass, indicating that reproductive performance depended on continuity of improvements from pod initiation to filling. Despite losses during later developmental stages, bacterial treatments maintained greater reproductive output through to maturity. Overall, this suggests that bacterial treatments influenced reproductive timing, developmental continuity, and final yield formation rather than simply enhancing vegetative growth.

Greenhouse evaluation of the different bacterial formulations revealed strain- and formulation-dependent vegetative responses (Figures 4A–C). In particular, the stronger increase in stem thickness under BS-114E relative to its live-cell counterpart suggests that the effect was not solely dependent on viable cells and may have been associated with extracellular components in the extract. These observations indicated that treatment responses were influenced by bacterial formulation. Additionally, combination treatments showed responses comparable to the NPK control, without consistently exceeding its effects, suggesting that the observed plant responses were not solely attributable to nutrient input. Reproductive responses were more pronounced than vegetative responses and varied across developmental stages. However, early responses at reproductive initiation were not retained uniformly across treatments. Differences in buds and flower counts suggested variation in the progression of reproductive development (Figures 5A, B). BS-114E showed the strongest continuity from bud formation to pod number and cumulative pod mass while NPK and CS.NPK50 also performed strongly at final yield, as reflected in total pod weight and average pod mass (Figures 5C–E). Previous characterization of BS-114 showed indole-related compound production and biosurfactant activity (Kaddoura et al., 2026). Together with the current observations, this suggests that soluble bacterial metabolites may have contributed to the overall positive effects observed across both vegetative and reproductive traits. In contrast, PPW-26E, despite its strong effect at the flowering stage, showed comparatively moderate performance at the pod and yield levels. Together, these observations indicate that early reproductive responses were not sufficient predictors of final yield outcomes, as reproductive performance depended on the progression and retention of reproductive structures through later developmental stages. These patterns were consistent with those observed in GH1.

Physiological responses were primarily expressed through increases in photosynthetic rate and stomatal conductance, whereas intercellular CO₂ concentration and transpiration remained comparatively stable across treatments. Although PPW-26L, BS-114L, and CS.NPK50 showed the strongest physiological responses, these improvements did not consistently correspond to reproductive outcomes. For example, PPW-26L exhibited the highest photosynthetic rate, whereas BS-114E achieved greater reproductive performance despite more moderate physiological responses (Figure 4D, 5C–E). This contrast suggests that enhanced carbon assimilation alone was insufficient to explain differences in productivity. Within this context, source–sink relationships, particularly the balance between photosynthetic carbon supply and reproductive demand, may provide a useful framework for interpreting differences in the retention and development of reproductive structures (Soltani et al., 2019). Similarly, the stronger physiological responses under PPW-26L and BS-114L, compared with the reproductive performance of their corresponding extracts, suggest that live cells may be more closely associated with leaf-level physiological activity, whereas cell-free extracts may influence sink development or reproductive allocation. Although not all differences were significant, this pattern was observed across both strains. Notably, the bacterial consortium combined with half the fertilizer dose (CS.NPK50) sustained vegetative growth and final yield comparable to the full-dose consortium treatment (CS.NPK100) and the NPK control, while showing stronger physiological responses than either treatment, particularly in photosynthetic rate and stomatal conductance. This suggests that consortium application may help maintain productivity under reduced fertilizer input, supporting its potential role in fertilizer-reduction strategies. This is consistent with previous findings in corn, where co-inoculation with B. subtilis and Azospirillum brasilense under reduced nitrogen fertilization enhanced leaf gas exchange parameters, improved plant nitrogen- and water-use efficiency, and supported greater biomass and grain yield (Galindo et al., 2024).

The carbon fixation-related gene rbcL, which encodes the large catalytic subunit of RuBisCO, showed only minor changes across treatments despite the increases observed in photosynthetic performance. Notably, BS-114L exhibited higher rbcL expression, whereas PPW-26L showed higher photosynthetic rates, indicating that transcriptional changes in carbon fixation did not correspond directly with gas exchange measurements or plant performance. The limited alignment between rbcL expression and photosynthetic performance in the live-bacteria treatments suggests that increased CO₂ assimilation was not driven by rbcL expression alone. Bacillus spp. have been previously reported to enhance photosynthetic performance and modulate photosynthesis-related processes, including carbon metabolism and RuBisCO-associated activity, although direct evidence linking beneficial bacterial inoculation to rbcL transcription remains limited (Wang et al., 2022). Similarly, inoculation with an endophytic Pseudomonas oryzihabitans CB24 was described to enhance carbon assimilation and RuBisCO activity in Pelargonium graveolens while upregulating several photosynthesis-related genes involved in light harvesting and electron transport (Deepa et al., 2025). In common bean, proteomic analysis further showed differential regulation of RuBisCO and proteins involved in its activation and assembly (Zadražnik et al., 2017). Together, these findings suggest that improved photosynthetic performance may involve broader physiological mechanisms and post-transcriptional regulation of RuBisCO rather than changes in rbcL expression alone. The nitrogen fixation-related gene nifH, which encodes the dinitrogenase reductase subunit of nitrogenase, was strongly upregulated across treatments, with the highest expression observed under BS-114L, followed by PPW-26L. Consistent with this, increased nodulation was observed (Figure 2A), indicating enhanced associations with nitrogen-fixing microorganisms. However, nifH expression and nodule formation reflect the presence of nitrogen fixation potential rather than actual nitrogen fixation rates, which are influenced by multiple environmental and physiological factors (Singleton and Stockinger, 1983; McCormick, 2018). A similar distinction was observed in soybean, where co-inoculation with Bradyrhizobium japonicum and B. aryabhattai enhanced nodule nitrogenase activity despite relatively low rhizosphere nif gene abundance, indicating that nif-related molecular responses did not directly reflect functional nitrogen fixation (Xing et al., 2022). In this study, despite these increases, nifH expression alone was insufficient to explain differences in physiological performance and yield outcomes, suggesting that broader plant–microbe and physiological processes contributed to the final response. Bacterial treatments also modulated auxin- and cytokinin-related gene expression across tissues. Auxin-related genes (AUX and AUX/IAA) were consistently upregulated, suggesting enhanced auxin transport and signaling activity. In contrast, cytokinin-related genes displayed clear tissue-specific regulation, with increased IPT3 expression in roots and reduced expression of IPT3 and LOG in leaves, indicating redistribution of hormonal regulation across tissues. Reduced PVCK expression is also consistent with reduced cytokinin degradation, further supporting modulation of cytokinin-related pathways. This pattern reflects a coordinated rebalancing of auxin–cytokinin signaling, indicating that hormonal activity was redistributed rather than uniformly elevated across the plant. Such coordinated regulation is consistent with established auxin–cytokinin crosstalk, in which cytokinin can modify auxin transport and signaling to regulate plant development (Street et al., 2016). Importantly, hormone-related regulation has also been observed following microbial inoculation, with bacterial consortia in Arabidopsis differentially modulating auxin-, cytokinin-, and jasmonate-responsive signaling, while a Bacillus-containing consortium promoted flowering and reproductive output (Guerra et al., 2026). Taken together, these responses indicate that both hormonal pathways were actively regulated in parallel, with their relative balance likely contributing to the developmental outcomes observed throughout the study rather than being driven by either pathway independently. Overall, hormone-related gene expression aligned more closely with the observed vegetative and reproductive responses than carbon- or nitrogen-related gene expression. These findings suggest that treatment effects were associated with regulation of developmental processes rather than uniform enhancement of growth.

Field trials conducted under agronomic conditions showed consistent increases in yield across all treatments relative to the NC control. Among treatments, BS-114+BS-120 and NPK exhibited the highest yields, followed by CB and PPW-26, whereas BS-114 and BS-120 showed comparatively smaller gains. The best-performing bacterial treatments showed yield responses comparable to both NPK and the commercial biostimulant (CB). These results indicate that positive yield responses to selected bacterial treatments persisted under field conditions. Although differences were not statistically significant, yield responses consistently favored bacterial treatments over the NC control.

Consortia-based treatments generally outperformed single isolates; specifically, the Bacillus consortium (BS-114+BS-120) closely matched NPK performance while the Bacillus–Pseudomonas consortium showed smaller improvements, suggesting that compatibility among strains may influence field performance. This aligns with meta-analytic evidence showing that microbial consortia generally outperform single-strain inoculants, with functional complementarity and compatibility among constituent strains contributing to their efficacy (Liu et al., 2023).

These shifts in microbial community composition were accompanied by subtle but structured changes within taxa associated with nutrient cycling and soil functionality. At the phylum level, communities across all treatments were dominated by Verrucomicrobia, Actinobacteria, Proteobacteria, and Firmicutes, consistent with previously reported core microbial taxa associated with Phaseolus vulgaris rhizosphere soils (Stopnisek and Shade, 2021). Relative to NC, Verrucomicrobia decreased most strongly under BS-114, followed by PPW-26 and CS. Verrucomicrobia, commonly associated with oligotrophic conditions and carbon turnover, are often reported to decrease with increasing nutrient availability (Navarrete et al., 2015), suggesting that this reduction may reflect treatment-associated improvements in soil nutrient availability. In contrast, Actinobacteria increased across all bacterial treatments, with the most notable increases observed under BS-114+BS-120, PPW-26, and CS, whereas Firmicutes increased mainly under BS-120 and CS. These increases are relevant as both Actinobacteria and Firmicutes are widely recognized for their roles in organic matter decomposition, nutrient mineralization, and production of antimicrobial compounds (Bhatti et al., 2017; Anderson et al., 2018), contributing to soil fertility and potential disease suppression (Cha et al., 2015). Moreover, the relative abundance of Bradyrhizobium slightly increased under BS-114+BS-120, PPW-26, and CS. As a well-established nitrogen-fixing genus in legume systems, this increase is consistent with enhanced nitrogen-cycling potential and plant-microbe interactions. Previous studies have shown that co-inoculation of Bradyrhizobium with PGPRs, including Bacillus species, can enhance nodulation, nitrogen accumulation, and N-fixation-related traits in legumes (da Costa Neto et al., 2024). The presence of Bacillus species across treatment groups is also relevant because members of this genus are associated with plant growth promotion, nutrient solubilization, and biocontrol activity (Miljaković et al., 2020). Slight increases in Candidatus Nitrocosmicus, particularly under NPK and CS, suggest increased nitrification potential (Lee et al., 2024), consistent with observed increases in nitrate levels. Similarly, the enrichment of Pseudarthrobacter in BS-114+BS-120, PPW-26, and CS suggests enhanced organic matter turnover and nutrient mobilization, as members of this genus have been associated with phosphorus solubilization and stress tolerance in soil systems (Ham et al., 2022).

Consistent with these microbial shifts, soil nutrient analysis revealed significant differences in several nutrient and soil fertility parameters, whereas C;N ratio, nitrite, and pH remained comparatively stable across treatments. Treatment-level alignments with soil nutrient changes were most evident in treatments that also showed increases in taxa associated with nutrient cycling and organic matter turnover. Differences were particularly evident for nitrogen forms, bicarbonate-extractable phosphorus, and sulfur, although no consistent directional pattern was observed across bacterial treatments. These patterns suggest. That the observed microbial shifts were associated with changes in specific nutrient pools and functional processes, including organic matter turnover, phosphorus mobilization, and nitrogen transformations. Higher phosphorus and sulfur levels under select treatments indicate improved availability of key macronutrients involved in plant development. Together, these shifts in nutrient availability indicate that treatments influenced soil nutrient dynamics rather than overall soil fertility. Despite these microbial and nutrient shifts, microbial community structure remained largely conserved across treatments, with no clear separation observed in beta diversity and only minor changes in alpha diversity. This suggests that bacterial treatment application primarily influenced existing microbial processes, particularly those related to nutrient cycling, rather than substantially restructuring the soil microbiome, a pattern consistent with short-term field experiments (Reese et al., 2018). .