Bacterial inoculates were prepared in nutrient broth for 24 h at 30°C under constant agitation at 90 rpm. The cell concentration in each culture was estimated by CFU counting in nutrient agar and was standardized to 10⁴ ml⁻¹ using 0.85% saline solution. The fungal mycelia were cultivated in a PDA medium for 14 days at 30°C. Subsequently, the plates were superficially washed using 10 ml of 0.85% saline solution per plate, and the spore concentration was determined by counting them in a Neubauer chamber (hemocytometer) using an optical microscope (BA210, MOTIC, Canada) (40–100 × magnification). The spore concentration of the cultures was adjusted to 10⁵ ml ⁻¹.

This test was conducted to verify the colonizing potential of the strains to be tested in a hydroponic system. The test was conducted using G. max seeds of the Bonus 8579 RSF IPRO cultivar (BRASMAX, Brazil). The seeds were disinfected to remove epiphytes, successively washed in running water, and agitated in water and neutral detergent (Tween) for 5 min. Then, the seeds were immersed in 70% ethanol for 1 min, followed by immersion in sodium hypochlorite (2.5% active chlorine) for 1 min and 30 s, and then again in 70% ethanol for 30 s. Finally, the seeds were washed three times in autoclaved distilled water and planted in plastic trays containing autoclaved sand as substrate, where the seedlings remained for 7 days in a BOD Camera (SSBODU320, PROLAB, Brazil) at 25°C (77F) and 12/12 h photoperiod. During this period, the seedlings were aseptically watered once a day with sterilized water. The plants were then carefully removed from the sand, and the roots were detached and washed four times with sterilized distilled water under vigorous agitation. The roots were then submerged in the previously prepared inoculum solutions and deposited in Petri dishes containing GELP medium. Plates containing roots immersed in autoclaved distilled water were used as a negative control and incubated at 28°C for 48 h. Subsequently, the roots were prefixed with glutaraldehyde, post-fixed with osmium tetraoxide, dehydrated in an ethanol series, transferred to amyl acetate, and critically dried in a dryer with carbon dioxide (CO₂), following the methodology described by Ghosh et al. (2016). Then, they were coated with gold using an ion jet to evaluate microbial colonization under a scanning electron microscope (Jeol JSM-IT300LV, JEOL USA, Inc., Peabody, MA, US).

The experiment was conducted in a greenhouse at the Plant Tissue Culture Laboratory of the IFGoiano Rio Verde campus (17° 48' 15.9' ‘S – 50° 54' 19,5” W), from July to August 2020, under an average temperature of 26.85°C and relative humidity of 23.7%. Soybean seeds of the same cultivar were subjected to the same disinfestation and germination treatment in autoclaved sand, where the seedlings remained until they reached a mean size of 15 cm, and then transferred to 4 L pots containing the hydroponic nutrient solution proposed by Hoagland and Arnon (1950) with half ionic strength for 15 days of adaptation. After this period, the plants were subjected to a nutrient solution with full ionic strength.

The inoculation of microorganisms occurred during phase R5 of soybean development. For this, 10 ml of the previously prepared inoculate solutions were added to the nutrient solution along with 10 g L⁻¹ of aseptic glucose to stimulate microbial growth in the nutrient solution. Biometric, physiological, and anatomical evaluations were conducted after 10 days of exposure to the tested strains. Plants grown in a nutrient solution without adding microorganisms and grown in a nutrient P-free solution (without the addition of CaHPO₄) were used as control treatments.

The in vivo experiment was conducted in randomized blocks, with nine treatments with microorganisms (eight strains + Biomaphos®) and two control treatments (plants grown in a nutrient solution without microorganisms and plants grown in P-free nutrient solution). All treatments were evaluated in five repetitions, with each repetition consisting of two plants per pot. The results obtained for biometric analyses, tissue P content, and physiological and anatomical analyses were subjected to ANOVA to evaluate the treatment effect. When significant, the effects were evaluated using the Scott-Knott test at 5% probability.

Subsequently, all variables were evaluated using a correlation matrix and combined in a principal component analysis (PCA). Since these variables had different measurement units, a correlation PCA was performed which was constructed using standardized data with a mean of 0 and a standard deviation of 1. The number of components was chosen as a function of the eigenvalues (>1) and explained variance (above 80%). The variables were also evaluated using Pearson's correlation coefficient, and the strength of the relationship was analyzed through R-values and the significance of the interaction at 5% probability. All statistical tests were performed in the R software version 4.0.4 (R Core Team, 2021), using the “ExpDes.pt” (Ferreira et al., 2014) and “FactoMineR” (Husson et al., 2010) packages.

The biometric variables plant height, stem diameter (SD), root length (RL), number of leaves (LN), and biomass were evaluated. For biomass evaluation, the plants were segmented into leaves, stems, and roots, and the material was dried in a forced-air oven at 65°C to a constant mass. Subsequently, the dry mass (DM) of each plant part was determined. The sum of the values corresponding to the biomass of each structural component of the plants (LDM + SDM + RDM) enabled the determination of the total DM (TDM).

Leaves, stems, and roots were oven-dried at 65°C until they reached a constant weight (Oven; SL-102/1152, SOLAB, Brazil) and crushed in a Willey mill with a 20 mm-mesh sieve (R-TE-680, TECNAL, Brazil). Finally, the P content was estimated according to the procedure described by Malavolta et al. (1997).

The diameter of the root pot elements was evaluated as an indication of root development because it is commonly associated with P availability to plants (Rosolem and Marcello, 1998). Thus, root samples were fixed in glutaraldehyde solution (2.5%) and paraformaldehyde (4%) in sodium cacodylate buffer (pH 7.2) and added to 5 mM calcium chloride (Karnovsky, 1965). The roots were cross-sectioned into 5-μm-thick slices and stained with toluidine blue at pH 6.8 (O'Brien et al., 1964). In these sections, the diameter of the vessel elements was measured using the Anati Quanti 2 software (Aguiar et al., 2007). Each repetition was measured 15 times, and the treatments were evaluated in quintuplicate.

Gas exchange was evaluated from 7:00 AM to 10:00 AM on the third leaf counted from the apex of the plant using an infrared gas analyzer with a fluorometer (LI-6800xt, LI-COR Inc., Lincoln, USA) and photosynthetically active radiation (1,000 μmol photons m⁻² s⁻¹) at a block temperature of 27°C and relative humidity of ~70%. The net photosynthesis rate (A) (μmol CO₂ m⁻² s⁻¹), stomatal conductance (g_sw) (mol of H₂O m⁻² s⁻¹), transpiration (E) (H₂O m⁻² s⁻¹), and internal carbon concentration (Ci) (mmol m⁻² s⁻¹) were measured (see Hunt, 2003).

The concentration of photosynthetic pigments was evaluated using leaf disks (three fresh matter mass disks of 5 mm diameter each). These disks were covered with a DMSO solution and saturated with calcium carbonate (CaCO₃) (Santos et al., 2007). Subsequently, they were stored in tubes wrapped in Al foil for 24 h at 65°C. The absorbance of the obtained extract was determined by spectrophotometry at 664, 649, and 480 nm. Chlorophyll a, b, and total and carotenoid concentrations were determined according to the methods described by Wellburn (1994).

Chlorophyll a fluorescence OJIP transient was determined using a portable FluorPen FP 100 fluorometer (Photon Systems Instruments, Drasov, Czech Republic). The third leaf of all sample units was previously dark-adapted for 30 min for complete oxidation of the photosynthetic electron transport system. They were later subjected to a 3,000 μmol m⁻² s⁻¹ blue light pulse, with minimum fluorescence (F₀) and maximum fluorescence (F_M). These values were used to estimate several PSII bioenergetic indices, according to the methods described by Strasser et al. (2000). The values of the specific light absorption flow per reaction center (ABS/RC), energy flow captured per reaction center at t = 0 (TRo/RC), electron transport flow per reaction center (ETo/RC), specific energy dissipation flow at the level of the antenna complex chlorophylls (DIo/RC), photosynthetic performance index (Pi_Abs) that incorporates energy cascade processes from the first absorption events to PQ reduction, the maximum quantum yield of primary photochemistry (PHI_Po), the probability of an exciton moving an electron through the electron transport chain after quinone (PSI_O), and the quantum yield of electron transport (PHI_Eo) after dark adaptation of the leaves (30 min) were determined in this study.

Chlorophyll a fluorescence in G. max leaves was also evaluated using the modulated Imaging-PAM fluorometer to obtain fluorescence images. Initially, the initial fluorescence (F0) and maximum fluorescence (Fm) were determined in leaves pre-adapted to the dark for 30 min. Then, it was possible to calculate the potential quantum yield of photosystem II (PSII) (Genty et al., 1989). Sequentially, the fluorescence in the light-adapted sample before the saturation pulse (F) and Fm in a light-adapted sample (Fm') were used to obtain the effective quantum yield of photochemical energy conversion in PSII, ΦII = (Fm′-F)/Fm′.

The exposition of G. max seedlings to the microorganisms tested did not affect the height of the plants; the means ranged between a minimum of 84.8 cm in plants inoculated with Epicoccum keratinophilum (SC4) and a maximum of 93.8 cm in plants treated with P. alvei (PA12). However, the development of plants in a control solution without P was significantly affected by the unavailability of this element, with a mean height for these plants of only 64.25 cm (Supplementary Figure 3A). Similar results were obtained for RL, which remained statistically equal in plants inoculated with different microorganisms and in a solution without microorganisms, with the means ranging between a minimum of 46.5 cm in plants inoculated with E. keratinophilum (SC4) and without microorganisms and a maximum of 93.8 cm in plants treated with P. alvei (PA12). Plants kept in a nutrient solution without P also had their RL affected, with a mean of 34 cm (Supplementary Figure 3B).

Standard deviation means also followed the same pattern: not differing between plants grown in a solution without microorganisms and inoculated plants, ranging from a minimum of 0.38 mm in non-inoculated plants to a maximum of 0.46 mm in plants treated with P. alvei (PA12). As described for the previous variables, plants subjected to a nutrient solution without P also presented low SD development (0.16 mm) (Supplementary Figure 3C). However, the LN was significantly affected by the inoculation treatments, with the highest values for plants inoculated with P. alvei (PA12), L. fusiformis (PA26), and Penicillium sheari (SC15) (21.2, 20.2, and 19.6 leaves, respectively). The absence of P in the solution also affected the LN in the plants, with a mean of only 3.25 leaves in plants developed with this treatment (Figure 2A).

The dry matter mass of leaves, stems, and roots did not differ between plants inoculated with microorganisms and plants grown in a solution without microorganisms. Leaf dry matter means varied between a minimum of 4.39 g in non-inoculated plants and a maximum of 5.2 g in plants treated with L. fusiformis (PA26) (Supplementary Figure 4A). Shoot dry matter mass followed a similar trend, ranging from a minimum of 4.05 g in non-inoculated plants and a maximum of 5.31 g in plants treated with L. fusiformis (PA26) (Supplementary Figure 4B). Root dry matter also followed the pattern described above, ranging from a minimum of 2.91 g in non-inoculated plants to a maximum of 4.12 g in plants treated with L. fusiformis (PA26) (Supplementary Figure 4C). However, plants grown in a solution without P presented the lowest values for LDM, SDM, and RDM (0.196, 1.036, and 0.847 g, respectively).

For TDM, the highest values were observed for plants inoculated with P. alvei (PA12) and L. fusiformis (PA26) (14.63 and 14.43 g, respectively). Plants grown in solution without microorganisms had a mean of 11.35 g TDM, but plants kept in solution without P had the lowest values (2.08 g) (Figure 2B).

Phosphorus content did not differ in leaves of plants subjected to the different treatments evaluated, and the means ranged between a minimum of 0.67 g in plants grown without P and a maximum of 1.30 g kg⁻¹ in plants treated with P. alvei (PA12) (Supplementary Figure 5A). Stem P content was affected by the inoculation treatments, being higher in plants inoculated with P. alvei (PA12) and L. fusiformis (PA26) (1.25 and 1.25 g kg⁻¹, respectively) (Figure 2C). The P content in roots was higher for plants inoculated with Codinaeopsis sp. (328EF), H. insecticola (33EF), P. alvei (PA12), L. fusiformis (PA26), B. thuringiensis (SC10), and P. sheari (SC15) (1.50, 1.57, 1.40, 1.45, 1.35, and 1.35 g kg⁻¹, respectively) (Supplementary Figure 5B). The evaluation of total P content showed the highest means in plants inoculated with the bacteria P. alvei (PA12) and L. fusiformis (PA26) (3.95 and 3.77 g kg⁻¹), while the lowest means were verified in plants growing in a solution without microorganisms (2.58 g kg⁻¹) and P-free (2.13 g kg⁻¹) (Figure 2D).

Microbial inoculation significantly affected gas exchange rates in G. max. The net CO₂ assimilation rate (A) was higher in plants inoculated with the strains P. alvei (PA12), L. fusiformis (PA26), B. cereus (SC5), and P. sheari (SC15) at 8.41, 7.43, 9.65, and 8.78 μmol CO₂ m⁻² s⁻¹, respectively (Figure 3A). However, these plants presented low rates of internal CO₂ (Ci) concentration, respectively 235.56, 219.9, 230.85, and 230.24 mmol m⁻² s⁻¹, while the highest means were verified in plants grown in a solution without P (280.32 mmol m⁻² s⁻¹) and inoculated with Necropsied sp. (328EF) and Biomaphos® (256.96 and 251.16 mmol m⁻² s⁻¹) (Figure 3B). A behavior similar to that observed for A was observed for stomatal conductance (Gsw), with the highest means obtained for inoculation treatments with Necropsied sp. (328EF), P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), and B. cereus (SC5) strains (0.072, 0.099, 0.077, 0.098, and 0.109 mol m⁻² s⁻¹, respectively), and for plants grown in solution without microorganisms (0.07 mol m⁻² s⁻¹) (Figure 3C). The transpiration rate (E) was higher in plants grown in a solution without microorganisms (0.0016) and inoculated with Necropsied sp. (328EF), Biomaphos®, P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), and B. cereus (SC5) at 0.0021, 0.0018, 0.0024, 0.0018, 0.0023, and 0.0026 E, respectively (Figure 3D).

Chlorophyll a content increased in plants inoculated with H. insecticola (33EF), P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), E. keratinophilum (SC4), and B. cereus (SC5) at 42.437, 49.240, 47.717, 47.571, 41.469, and 46.052 μmol m⁻², respectively (Figure 4A). Chlorophyll b content also increased following a pattern similar to that observed for chlorophyll a, but it was higher in plants subjected to inoculation with Necropsied sp. (328EF), H. insecticola (33EF), P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), E. keratinophilum (SC4), and B. cereus (SC5) strains (11.644, 13.723, 13.069, 13.726, 12.231, 12.269, and 12.704 μmol m⁻²) (Figure 4B).

Microbial inoculation also affected the carotenoid content, which was higher in plants inoculated with H. insecticola (33EF), P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), and B. cereus (SC5), with mean values of 10.458, 12.275, 12.084, 11.034, and 11.382 μmol m⁻² (Figure 4C). Similar to other chlorophylls, the total chlorophyll content varied, with the highest means obtained in plants treated with H. insecticola (33EF), P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), E. keratinophilum (SC4), and B. cereus (SC5) (56.161, 62.309, 61.444, 59.802, 53.738, and 58.756 μmol m⁻²). However, much lower mean values were observed in plants grown in a solution without P (33.832 μmol m⁻²) (Figure 4D).

As expected, the specific energy flows of the active reaction centers ABS/RC and DIo/RC were similarly affected by microbial inoculation; therefore, the lowest mean ABS/RC values were observed in plants grown with the microorganisms P. alvei (PA12), L. fusiformis (PA26), B. thuringiensis (SC10), P. sheari (SC15), and B. cereus (SC5) (2.986, 2.962, 3.127, 2.989, and 2.859, respectively), as well as in plants grown in a solution without microorganisms (2.935) (Figure 5A). The lowest mean DIo/RC values were obtained in plants subjected to the same treatment sequence as above (0.714, 0.686, 0.842, 0.796, 0.660, and 0.667, respectively) (Figure 5B). The means observed for TRo/RC and ETo/RC were not affected by any of the evaluated treatments. TRo/RC values ranged from a minimum of 2.193 in plants inoculated with P. sheari (SC15) and a maximum of 2.546 in plants treated with H. insecticola (33EF), whereas ETo/RC means varied between a minimum of 0.985 in plants grown in nutrient P-free solution and a maximum of 1.384 in plants treated with the fungus H. insecticola (33EF) (Supplementary Figures 6A,B).

The kinetic parameters of Chl a fluorescence were also significantly affected by inoculation with microorganisms. The maximum quantum yield of primary photochemistry (Phi_Po) was higher for plants inoculated with P. alvei (PA12), L. fusiformis (PA26), and B. cereus (SC5) (0.763, 0.769, and 0.769, respectively), as well as in plants grown in a solution without microorganisms (0.752) (Figure 5C). The probability of an exciton moving an electron through the electron transport chain after quinone (Qa) (Psi_o) was reduced in plants inoculated with Codinaeopsis sp. (328EF) (0.443) and under P-free solution (0.406) (Supplementary Figure 6C). The means observed for the quantum yield of electron transport (Phi_Eo) followed the same pattern reported previously for Psi_o, with reduced values under Codinaeopsis sp. (328EF) (0.323) and P-free solution (0.277) (Supplementary Figure 6D). The photosynthetic performance index (Pi_ABS) was higher in plants inoculated with P. alvei (PA12), L. fusiformis (PA26), and B. cereus (SC5) (1.244, 1.289, and 1.255), suggesting that these plants have better performance in converting light energy into chemical energy (Figure 5D).

Fluorescence imaging analyses confirmed the superiority of the behavioral pattern of chlorophyll a fluorescence in plants inoculated with P. alvei (PA12), L. fusiformis (PA26), and B. cereus (SC5) and the inferiority of the pattern observed in plants grown in a P-free solution, as well as in plants inoculated with E. keratinophilum (SC4), Codinaeopsis sp. (328EF), and H. insecticola (33EF) (Figure 6).

The inoculation treatments affected the diameter of the root vessel elements. Therefore, the highest means were observed in plants inoculated with P. alvei (PA12), L. fusiformis (PA26), P. sheari (SC15), E. keratinophilum (SC4), and B. cereus (SC5) (55.383, 52.223, 49.943, 49.867, and 49.32 μm, respectively). In contrast, the lowest means were observed in plants grown in a P-free solution (28.787 μm), followed by a solution without microorganisms (39.91 μm) (Figure 7).

Anatomical analyses showed superior development of vessel elements and root caliber of plants treated with P. alvei (PA12) and L. fusiformis (PA26) strains. Root development was compromised in plants grown in a P-free solution, as well as in plants not subjected to microbial action (Figure 8).

The two-dimensional graph of principal components showed a negative correlation between the parameters DIo/RC, ABS/RC, TRo/RC, and Ci and the other biometric parameters: fluorescence, photosynthetic pigment content, and gas exchange, signaling the means for plants subjected to E. keratinophilum (SC4), Codinaeopsis sp. (328EF), Biomaphos®, and H. insecticola (33EF); plants grown in P-free solution accounted for most of the variation found in these variables. Thus, these plants were mainly associated with negative photosynthetic performance indices. On the other hand, the means verified in plants inoculated with P. alvei (PA12), L. fusiformis (PA26), and B. cereus (SC5) defined most of the variation in growth parameters, P content, photosynthetic performance, and photosynthetic pigments. These symbiotic microorganisms seem to improve the overall performance of the inoculated G. max plants (Figure 9A). The negative correlation between DIo/RC, ABS/RC, TRo/RC, and Ci means and the other biometric parameters (fluorescence, photosynthetic pigment content, and gas exchange) was more evident when the graphic pattern of the correlations between all variables was analyzed (Figure 9B).