To assess the recovery of Bradyrhizobium cells from soybean seeds, a controlled experiment was carried out in the Soil Biotechnology Laboratory at Embrapa Soja in Londrina, Paraná (PR), Brazil, under controlled conditions in a completely randomized design with eight replicates. The strain Bradyrhizobium japonicum SEMIA 5079 (=CPAC 15, =CNPSo 07), which is used in commercial inoculants for soybean cultivation in Brazil, was used. The strain is deposited at the “Diazotrophic and Plant Growth Promoting Bacteria Culture Collection of Embrapa Soja (WFCC Collection # 1213, WDCM Collection # 1054).

Seeds were treated with one of six doses of nickel sulfate (NiSO₄.6H₂O) (0, 60, 120, 180, 240, and 300 mg Ni kg seeds^-1) corresponding to 0, 3, 6, 9, 12, and 15 g Ni ha⁻¹. The seeds were then inoculated with a high bacterial load of 7 × 10⁹ cells mL⁻¹ to provide 1.2 × 10⁶ cells per seed. Three biological samples were transferred to sterile Erlenmeyer flasks containing 100 mL of sterile saline solution (0.85%) supplemented with Tween 80 (0.4 mL L^-1). Each sample contained 100 seeds treated with NiSO₄·6H₂O. The flasks were subjected to horizontal shaking at 150 rpm for 20 min, and the recovered saline solution was referred to as the recovery sample (Pinto et al., 2023).

The recovery sample was serially diluted (10^-1, 10^-2, 10^-3, and 10^-4), and aliquots of 100 µL of each dilution were spread onto Petri dishes containing modified-yeast mannitol agar (YMA) culture medium (Hungria et al., 2017). The medium was supplemented with 10% Congo Red solution (0.25%), actidione (667 μL L^-1 cycloheximide [84 μg mL^-1 in ethanol]), and vancomycin (333 µL L^-1 vancomycin hydrochloride [0.3 g mL^-1]) to inhibit soil or seed contaminants (Rodrigues et al., 2020). The plates were incubated for seven days in the dark at 28 ± 2°C. Colony-forming units (CFUs) were then quantified on each plate and expressed as CFUs per seed (CFU seed^-1). Based on the results, recovery was classified as low (up to 100 CFUs per 100 µL of recovery sample), medium (between 100 and 200 CFUs per 100 µL of recovery sample), or high (above 300 CFUs per 100 µL of recovery sample).

To measure BNF activity via the acetylene reduction activity (ARA) method using the in situ continuous flow protocol (Devi et al., 2010), a controlled experiment was conducted in a greenhouse at Embrapa Soja, Londrina, PR, Brazil, under greenhouse conditions in a completely randomized design with eight replicates. The soil, which was classified as a Ferralsol according to the International System (Soil Survey Staff, 2014), had the following chemical properties: pH(_CaCl2) 5.1; calcium (Ca) 26 mmol_c dm^-3; magnesium (Mg) 5 mmol_c dm^-3; aluminum (Al) 0 mmol_c dm^-3; H⁺+ Al 67 mmolc dm^-3; potassium (K) 5.8 mmol dm^-3; phosphorus (P) (resin) 16 mg dm^-3; sulfur (S) 4.8 mg dm^-3; boron (B) 0.36 mg dm^-3; copper (Cu) 0.9 mg dm^-3; iron (Fe) 49 mg dm^-3; manganese (Mn) 9.5 mg dm^-3; zinc (Zn) 2.7 mg dm^-3; organic matter 20 g dm^-3; Ni 0.2 mg dm^-3; cation exchange capacity (CEC) 99 mmol_c dm^-3. The physical properties of the soil were clay 600 g dm^-3; silt 50 g dm^-3; and sand 350 g dm^-3. The liming requirement was determined using the base saturation method to reach 70% of CEC (van Raij. et al., 1997).

The same strain and treatments used in Experiment I were used in Experiment II. Soybean seeds (cultivar BRS 1001 IPRO) were inoculated with a high bacterial load of 7 × 10⁹ cells mL^-1 to provide 1.2 × 10⁶ cells seed^-1. Three seeds were sown in a polyvinyl chloride (PVC) pot (15 cm in diameter and 30 cm in height) filled with the clay soil described above. The pots were housed under controlled irrigation conditions in a greenhouse with the temperature regulated to maintain 28°C both day and night. Each pot contained 2 kg of soil, and the irrigation frequency was determined based on the soil water retention curve, maintaining soil moisture at 70–90% of field capacity. In addition, sterilized N-free standard nutrient solution was applied (De Oliveira-Francesquini et al., 2017). After one week, the seedlings were thinned to one plant per pot, six replicates were performed per treatment.

ARA was measured during phenological stages V₆ – R₃ (Fehr et al., 1977) as illustrated in Figure 1. On the day before sampling, all pots were watered thoroughly and left to drain overnight. The next morning, a lid was sealed around the base of each plant, and an acetylene:air mixture (1:9) was introduced into the pots through an inlet at the bottom at a flow rate of 1 L min^-1 (Figures 1A, B). When the steady state was reached after 15 min of flow (Figures 1C, D), three gas samples (1 mL each) were collected from the outlet of each pot using syringes (Figure 1E). Once all gas samples were collected, acetylene-free air was flushed through all pots for 1 h to remove any remaining acetylene. Flushing the pots with air ensured that the plants were exposed to acetylene for only a very short period each day.

The concentration of ethylene (C₂H₄) in the gas samples was determined using a gas chromatograph equipped with a flame ionization detector (FID) (Thermo Scientific; TRACE™ 1310; Waltham, MA) (Figure 1F). The run time to measure the ethylene concentration was 1 min, and the concentrations were expressed in µL L^-1. All details, including the carrier gas, flow rate, and temperatures, as well as whether concentrations were determined with reference to a calibration curve, are described by Devi et al. (2014).

Microbial activity and root respiration were assessed by monitoring carbon dioxide (CO₂) in the soil using an Arduino prototype. The experiment was conducted at Embrapa Soja, Londrina, PR, Brazil (27° 16’ 60’’ S, 50° 35’ 7’’ W) under greenhouse conditions in a completely randomized design with eight replicates. Experiment III followed the standard protocol used in Experiment II with respect to controlled greenhouse conditions, soil attributes, cultivar, strains, and treatments.

Figure 2 presents a simplified diagram of the Arduino Uno R3 (ATmega32) prototype, which comprised an SD module, sensor, LCD display, power supply, USB cable, protoboard and its use. Code was uploaded via an Arduino IDE. Sensor data were processed, displayed, and recorded on an SD card. The prototype was placed in a sealed tube to protect against water, secured to prevent cable disconnection, and connected to a stabilized electrical network. The full methodology is detailed in Kormann et al. (2024). The Arduino prototype monitoring system was installed by inserting a 15-cm-long, 27-mm-diameter PVC tube into the top 12.5 cm of the soil, leaving 2.5 cm of the tube above the surface. The tube was positioned approximately 5 cm from the pot center, where soybean seeds were later planted. The tube’s bottom was sealed with Parafilm™ to prevent water from entering and damaging the sensor. An MQ-135 sensor was placed 12 cm below the soil surface inside the tube, which was then covered with Parafilm™ to block water entry during irrigation. The soil was prepared to prevent compaction around the tube and ensure that gas concentrations could be measured through the tube’s base.

The Arduino prototype was used to record CO₂ every hour throughout the soybean cycle. The data were then used to calculate the average daily concentration of CO₂ in the soil air. All details of validation or calibration are described by Kormann et al. (2024).

To evaluate the effects of Ni application on soybean nutritional, physiological, and production components and grain yield, field trials were conducted over two growing seasons (2022/2023 and 2023/2024) at three experimental sites in distinct regions of Brazil: (I) Carolina Farm in the municipality of Sengés, Paraná State (PR), southern Brazil; (II) the Lageado Experimental Farm, College of Agricultural Sciences (UNESP/FCA), Botucatu, São Paulo State (SP), southeastern Brazil; and (III) the Experimental Farm of the School of Engineering, Campus Ilha Solteira (UNESP/FEIS), Selvíria, Mato Grosso do Sul State (MS), central-western Brazil, all of them in a completely randomized design with eight replicates. Supplementary Table 1 provides comprehensive experimental information for the three sites. Prior to sowing, the soybean seeds were treated with one of six doses of nickel sulfate (NiSO₄.6H₂O) corresponding to 0, 60, 120, 180, 240, and 300 mg Ni kg^-1 and were then inoculated with a high bacterial load of 7 × 10⁹ cells mL^-1 to provide 1.2 × 10⁶ cells per seed. Dolomitic lime (28% calcium oxide—CaO, 18% magnesium oxide—MgO, and 81% calcium carbonate equivalent—%ECaCO₃) was applied 60 days before the experiments to increase base saturation in the topsoil (0.00–0.20 m) to 70%, following the procedure outlined by Cantarella et al. (2022).

Supplementary Table 2 details the physical, chemical, and biological properties of the soil at a depth of 0.00–0.20 m, as well as the climate classification at each site according to the Köppen-Geiger system (Alvares et al., 2013). Physical attributes were determined following the methodology described by Donagema et al. (2017), and chemical properties were analyzed according to van Raij. et al. (1997). The autochthonous bacterial population capable of nodulating soybean was estimated using the most probable number (MPN) method with soybean plants, as described by O’Hara et al. (2016). Climatological data recorded during the experiments are shown in Supplementary Figure 1.

The nutritional status of the plants—based on leaf concentrations of N, P, K, Ca, Mg, S, Cu, Fe, Zn, Mn, and Ni—was assessed across the different treatments. Diagnostic leaves were collected at the R₂ phenological stage (full flowering). Specifically, the third fully expanded trifoliate leaf from the apex downward was sampled, properly labeled, and dried in a forced-air circulation oven. The dried material was ground using a Wiley-type mill and stored in labeled plastic bags for nutrient analysis (AOAC, 2019).

Photosynthetic performance was assessed through non-destructive diagnostic leaf measurements at the R₂ stage using a portable gas exchange system (CIRAS-3, PP Systems Inc., Amesbury, MA, USA). Instrument settings were standardized to 380–400 μmol mol^-1 atmospheric CO₂, photosynthetically active radiation (PAR) of 1,100 μmol photons m^-2 s^-1 (provided by LED light), chamber temperature between 25°C and 27°C, and relative humidity between 60% and 70%. A minimum equilibration period of 3 min was used for each measurement. All measurements were performed between 10:00 a.m. and 12:00 p.m. The following parameters were recorded: net photosynthetic rate (A; μmol CO₂ m^-2 s^-1); stomatal conductance (gs; mol H₂O m^-2 s^-1); intercellular CO₂ concentration (Ci; μmol mol^-1); transpiration rate (E; mmol H₂O m^-2 s^-1); water use efficiency (WUE; μmol CO₂ mmol^-1 H₂O), calculated as the A/E ratio; and carboxylation efficiency, calculated as the A/Ci ratio.

Urease activity was assessed in vivo (Hogan et al., 1983) in diagnostic leaves at the R₂ stage (Fehr et al., 1977). Discs of fresh tissue (100 mg) were incubated for 3 h in test tubes containing 8 mL of 50 mM phosphate buffer (pH 7.4), 0.2 M urea, and 0.6 M n-propanol. After incubation, 0.5 mL of the supernatant was added to 2.5 mL of Reagent I (0.1 M phenol and 170 μM sodium nitroprusside), which was followed by the addition of 2.5 mL of Reagent II (0.125 M NaOH, 0.15 M Na₂HPO₄·12H₂O, and 3% NaOCl). The reaction was conducted in sealed tubes under constant agitation in a water bath at 37°C for 35 min. The ammonium concentration was measured spectrophotometrically at 625 nm using an NH₄Cl standard curve. Urease activity was expressed as μmol N–NH₄⁺ g^-1 fresh weight h^-1.

Nitrate reductase (NR) activity (Hewitt and Nicholas, 1964) was measured in diagnostic leaves at the R₂ stage (Fehr et al., 1977). The leaves were ground in liquid nitrogen with a 1:2 ratio (g:mL) of extraction buffer composed of 25 mM Tris-HCl (pH 8.5), 1 M EDTA, 1 mM DTT, 1% BSA, 20 μM FAD, and 200 mM leupeptin. After centrifugation at 14,000 × g for 10 min at 4°C, 200 μL of the supernatant was added to the reaction buffer to a final volume of 0.5 mL. The reaction buffer contained 50 mM HEPES-KOH (pH 7.6), 10 mM MgCl₂ (for active NR) or 5 mM EDTA (for total NR), 10 μM FAD, 3% casein, and 1 mM DTT. For total NR activity, the extract was pre-incubated for 10 min with 11 μL of a solution containing 250 mM AMP and 500 mM EDTA. The reaction was initiated by adding 25 μL of 5 mM NADH diluted in 100 mM potassium phosphate buffer (pH 7.0) and incubated for 30 min. The reaction was stopped with 62.5 μL of 500 mM zinc acetate. The solution was centrifuged, and the supernatant was used to determine nitrite concentration by colorimetric analysis (Hageman and Reed, 1980).

Ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) activity (Reid et al., 1997) was measured in diagnostic leaves at the R₂ stage (Fehr et al., 1977). Activity was calculated based on the change in absorbance between 0 and 1 min and expressed as μmol min⁻¹ mg⁻¹ protein.

At the full flowering stage (R₂), five plants per plot were sampled for evaluation of nodulation and root and shoot dry mass. The ureide-N content in petioles plus leaves was determined (Hungria, 1994) by colorimetric quantification (Herridge et al., 2022). To measure the total ureide concentration (allantoin and allantoic acid) as a BNF indicator, a 300-μL aliquot was incubated with 500 μL of Solution 1 (50% [v/v] 0.5 N NaOH and 50% [v/v] 0.15 N HCl) at 100°C for 5 min and subsequently cooled to room temperature. The mixture was then combined with Solution 2 (11.5% [v/v] 0.4 M phosphate buffer at pH 7, 11.5% [v/v] phenylhydrazine, 70% [v/v] 0.65 N HCl at −20°C, and 7% [v/v] potassium ferricyanide). The ureide-N concentration was determined colorimetrically at 535 nm.

In both growing seasons (2021/22 and 2022/23), plants at the R₂ stage (Fehr et al., 1977), were harvested and separated into shoots and roots. The roots were washed to remove substrate particles. Nodules were removed from the roots, oven-dried at 65°C for 72 h, counted, and weighed. At harvest, the 100-grain weight and grain yield were determined. For grain yield, 2-m lengths of four rows were harvested (1.8 m²), threshed and weighed. The weight was adjusted to 13% moisture and then converted to kilograms per hectare. To determine the 100-grain weight, 10 samples of 100 grains were weighed, and the results were adjusted to 13% moisture.

The greenhouse experiments were arranged in a completely randomized design, whereas the field trials followed a randomized complete block design. Prior to analysis, data were tested for normality (Shapiro and Wilk, 1965) and homogeneity of variances (Levene, 1960). Upon confirmation of these assumptions, data were subjected to one-way analysis of variance (ANOVA) with the F-test applied at the 10% significance level. When significant differences were detected, data were compared using the least significant difference (LSD) test with a significance level of 10%. The agronomic variables were standardized through Z-score transformation in order to ensure comparability across different measurement scales. The standardized values (Z-scores) were computed by subtracting the mean and dividing by the standard deviation of each variable (only to agronomic assessments). Subsequently, a mean Z-score was calculated for each treatment group by averaging the Z-scores of all variables included in the analysis. This mean Z-score served as the response variable in a multiple polynomial regression model, with Ni doses applied through seed treatment as the explanatory variable.

To assess the effects of treating soybean seeds with Ni on the viability of Bradyrhizobium, soybean seeds were treated with one of six doses of Ni prior to inoculation with a high dose (1.2 × 10⁶ cells seed^-1) of Bradyrhizobium: 0, 60, 120, 180, 240, or 300 mg of Ni per kg of seed (mg kg⁻¹ seed) (Experiment I). After incubation for 20 min, the recovery of viable Bradyrhizobium from the seeds was determined.

As shown in Figure 3A, Ni doses above 120 mg kg⁻¹ seed had a detrimental effect on the recovery of Bradyrhizobium, reducing the CFUs per seed to levels significantly below the optimal threshold required for effective nodulation and plant growth. At the lowest Ni dose (60 mg kg^-1), CFU counts remained high, approximately 8.7 × 10⁴ CFU seed^-1, compared with 9.0 × 10⁴ CFU seed^-1 in the control (0 mg kg^-1), indicating minimal adverse effects on Bradyrhizobium viability. However, a progressive decline in CFU counts occurred as the Ni dose increased from 120 mg kg^-1 onwards.

The dilution series in Figure 3B (A: 10^-2, B: 10^-3, and C: 10^-4) provides a clear visual representation of how increasing Ni doses negatively influenced the microbial population on the soybean seeds. Across all dilutions (10^-2, 10^-3, and 10^-4), the number of colonies was visibly higher on the plates from the control and the 60 mg kg^-1 treatment, indicating optimal conditions for the survival and recovery of Bradyrhizobium cells. At the highest Ni doses (240 and 300 mg Ni kg^-1), colonies were sparse or nearly absent, particularly in the 10^-3 and 10^-4 dilutions, suggesting that Ni doses above 240 mg kg⁻¹ drastically reduced the viability of Bradyrhizobium cells. These results emphasize the importance of optimizing Ni doses in agricultural applications to maintain inoculant efficacy.

Since one of the goals of treating soybean seeds with Ni is to enhance BNF, we examined the effects of Ni treatment on the production of ethylene from acetylene, an indicator of nitrogenase activity, using the ARA method in pot experiments (Experiment II). As shown in Figure 1, the ethylene concentration (µL L^-1) varied significantly as a function of both Ni dose (p ≤ 0.001) and sampling time in days after emergence (DAE) (p ≤ 0.001). However, there was no significant effect of the interaction between Ni dose and sampling time (p ≤ 0.797), suggesting that the response to Ni fertilization was consistent across phenological stages.

At all sampling times, ethylene levels were highest in the control and the 60 mg Ni kg⁻¹ treatment, indicating potentially greater nitrogenase activity at these lower Ni doses. Ethylene concentrations dropped significantly at a dose of 120 mg Ni kg^-1 and were lowest in the 180, 240, and 300 mg Ni kg^-1 treatments, suggesting a dose-dependent reduction in ethylene concentrations.

Regardless of Ni dose, ethylene concentrations were lowest at 36 DAE and increased gradually until peaking around 54 DAE. Ethylene levels at 60 DAE were comparable to those at 54 DAE, indicating a stabilization or plateau in ARA activity in the later stages of development. These results indicate that treating soybean seeds with lower doses of Ni may promote nitrogenase activity during the early flowering stage, likely enhancing BNF and increasing ARA. Conversely, higher Ni doses appear to inhibit ethylene production, suggesting a negative impact on nitrogenase activity.

To evaluate the impact of seed treatment with Ni on microbial activity and root respiration, an Arduino prototype (Figure 2) was used to assess soil concentrations of CO₂ in pot experiments (Experiment III). The results, which are presented in Figure 2, demonstrated that the CO₂ concentration (µL L^-1) in soil cultivated with soybean was significantly affected by both Ni dose (p ≤ 0.001) and sampling time (DAE) (p ≤ 0.001). However, the effect of the interaction between Ni dose and sampling time was not significant (p = 0.823), suggesting that the response to Ni dose was consistent across phenological stages.

CO₂ levels were consistently higher in the control treatment and the 0 and 60 mg Ni kg⁻¹ seed treatments across most of the soybean growth cycle and peaked around the flowering stage (R₁ to R₄ stages). This trend suggests a direct correlation of these specific Ni doses with enhanced microbial and root respiration activity, consistent with increased CO₂ levels in the soil environment. The CO₂ concentrations gradually decreased after the flowering stage across all treatments, indicating reductions in microbial activity and root respiration as the plants progressed to later growth stages.

Treatments with higher doses of Ni (120, 180, 240, and 300 mg Ni kg⁻¹) consistently exhibited lower CO₂ levels than the control and lower Ni dose treatments, suggesting that treating seeds with elevated Ni doses inhibited microbial processes and reduced root activity. These effects are also consistent with the decreases in Bradyrhizobium recovery (Figure 3).

This study was conducted across three field sites characterized by a natural gradient of soil-available Ni (0.4–0.6 mg dm^-3) and pronounced textural variability, ranging from sandy to clayey soils. Importantly, despite this edaphic variability, the response pattern to seed-applied Ni was consistent across environments, with similar dose–response behavior observed at all sites.

To determine whether the effects of Ni seed treatment on microbial activity in pot experiments translated into changes in soybean productivity, field experiments were performed over two growing seasons (2022/23 and 2023/24) at three sites in Brazil: Selvíria, Botucatu, and Sengés (Experiment IV). The results are presented in Table 1.

Treating soybean seeds with Ni significantly affected both the number of nodules (NN) and the nodule dry weight (NDW) at all sites in both growing seasons (Table 1). Among the individual treatments, both the 60 and 120 mg Ni kg^-1 treatments significantly increased NN and NDW compared with the control (p < 0.01). In the 120 mg Ni kg^-1 treatment, NN and NDW both increased by 17.8% (2022/23) and 8.6% (2023/24) in Selvíria, 13.9% in Botucatu (2022/23), and 13.3% in Sengés (2022/23). At doses ≥ 180 mg Ni kg^-1, NN and NDW dropped sharply by 20%–30%, indicating detrimental effects on Bradyrhizobium spp. survival or root colonization.

Although shoot dry weight (SDW) and root dry weight (RDW) did not differ significantly among the treatments, treating soybean seeds with Ni did significantly affect soybean grain yield and associated yield components (Table 1). Grain yields were consistently highest in the 120 mg Ni kg^-1 treatment at all locations, with statistically significant increases over the control. In Selvíria, the grain yield in the 120 mg Ni kg⁻¹ treatment reached 5756 kg ha^-1 in 2022/23 and 4410 kg ha^-1 in 2023/24, representing increases of 3.4% and 4.0%, respectively, compared to the control. Similarly, in Botucatu, yields peaked at 5100 kg ha^-1 and 4253 kg ha^-1 in the 120 mg Ni kg^-1 treatment in the two seasons, corresponding to improvements of 3.4% and 1.9% over the control. In Sengés, the same dose resulted in yields of 4324 kg ha^-1 and 4566 kg ha^-1 in 2022/23 and 2023/24, corresponding to increases of 1.7% and 6.0% compared to the control. The 100-grain weight followed a similar trend. The 120 mg Ni kg^-1 treatment increased the 100-grain weight by up to 3.5%, while higher doses resulted in reductions of 2.5%–5.9%. Shoot and root dry weights were also maximized at 120 mg Ni kg^-1, with average increases of 5%–7% over the control, although the differences were not always statistically significant.

At higher Ni rates (240 and 300 mg kg^-1), a clear yield penalty was observed. In Selvíria, yields dropped by 1.0%–4.4% in the 240 and 300 mg Ni kg^-1 treatments compared to the control, while in Botucatu, the 300 mg Ni kg^-1 treatment reduced grain yield by 2.1%–6.2%. In Sengés, the 300 mg Ni kg^-1 treatment reduced grain yield by 1.7% in 2022/23 and by 5.3% in 2023/24 compared to the control, confirming that excessive Ni levels.

To explore the mechanisms underlying the effects of Ni seed treatment on soybean productivity, the levels of macro- and micronutrients in diagnostic leaves of soybean in the field experiments were determined at the R₂ phenological stage (full flowering) (Experiment IV). As shown in Supplementary Table 3, leaf macronutrient concentrations did not differ significantly among the Ni treatments in all site-years. The levels of N, P, K, Ca, Mg, and S were stable, indicating that Ni application did not impair nutrient uptake or translocation. Similarly, micronutrient concentrations (Supplementary Table 4) showed no significant changes, although a slight numerical increase in leaf Ni content was observed, particularly at higher Ni doses. Across all sites, leaf Ni content increased by 10%–20% in the 300 mg Ni kg^-1 treatment compared to the control (which had values ranging from 0.42 to 0.60 mg kg^-1), although this increase was not statistically significant.

The effects of Ni application on photosynthetic parameters were examined by analyzing pigment accumulation and RuBisCO activity in leaves and gas exchange variables at the R₂ stage (Tables 2–4). As shown in Table 2, the 120 mg Ni kg⁻¹ treatment increased total chlorophyll (Chl t) by up to 25.2% in Botucatu (2022/23) and 8.4% in Selvíria (2023/24) compared with the control treatment. At the highest Ni dose, chlorophyll levels decreased by 7.5%–10% compared to the 120 mg Ni kg^-1 treatment in Botucatu. Consistent with the increase in chlorophyll content, the 120 mg Ni kg^-1 treatment significantly enhanced the net photosynthetic rate (A) across all locations (Table 3). In Sengés, A increased from 15.0 µmol CO₂ m^-2 s^-1 in the control to 17.7 µmol CO₂ m^-2 s^-1 in the 120 mg Ni kg^-1 treatment (+18%) in 2022/23 and from 12.3 to 13.4 µmol CO₂ m^-2 s^-1 (+8.9%) in the corresponding treatments in 2023/24. Carboxylation efficiency (CE) also improved, peaking at 0.065–0.071, representing an average increase of 23% over the control. RuBisCO activity also peaked at intermediate doses (60–120 mg Ni kg^-1) (Table 4). These results indicate that the increases in productivity at lower Ni doses (≤ 120 mg Ni kg^-1) reflect increases in photosynthetic parameters.

To evaluate the effects of Ni seed treatment on N metabolism, the leaf concentration of N-ureides, a proxy for BNF, and leaf activities of urease and NR were determined in soybean plants in the field experiment at the R₂ stage (Table 4, Experiment IV). Compared with the control, the 120 mg Ni kg^-1 treatment increased the concentration of N-ureides by up to 18%. Conversely, N-ureide concentrations declined by 8%–13% in the 240 and 300 mg Ni kg^-1 treatments. Similarly, NR activity peaked at intermediate doses. As expected, urease activity increased linearly with Ni dose. In Sengés, urease activity was 32% higher in the 300 mg Ni kg^-1 treatment than in the control. However, the increase in urease activity was not accompanied by improved BNF or yield.

Z-score transformation enabled the aggregation of heterogeneous variables on a common scale, enhancing the multivariate interpretability of treatment effects (Figure 4). Applying a second-order polynomial regression model to the mean standardized Z-scores revealed a statistically significant quadratic response (p < 0.05) the integrated agronomic traits. The model identified an optimal dose of 66.2 mg Ni kg^-1 seeds at the apex of the response curve (Figure 4B), corresponding to the maximum overall standardized performance across all evaluated variables. Beyond this critical threshold, the mean Z-scores declined consistently.

Ni is required for the activity of microbial hydrogenases, important enzymes that reduce the energy needed for N₂ reduction by nitrogenase (Nunes et al., 2003; Bagyinka, 2014). Consequently, providing Ni via seed treatment could enable early biosynthesis and activation of hydrogenase, increasing BNF efficiency. Hydrogenase activity can be assessed by ARA measurement, which is an important tool for evaluating BNF throughout the growth cycle due to its simplicity, low cost and high sensitivity (Devi and Sinclair, 2013).

Consistent with the results of previous studies showing that Ni levels influence ARA activity (Ureta et al., 2005; Yusuf et al., 2011; Lavres et al., 2016; Freitas et al., 2019; Rengel et al., 2022), we found that regardless of phenological stage, Ni doses of 0 and 60 mg kg^-1 increased ethylene production (Experiment II). Supporting these results, Lavres et al. (2016) reported that treating soybean seeds with different doses of Ni increased BNF by an average of 10%, significantly contributing to N supply, especially in grains. In another assessment of the effects of different Ni doses on soybean, Freitas et al. (2019) observed that doses of up to 60 mg Ni kg seeds^-1 increased nitrogenase activity, enhancing NN, NDW, photosynthesis, and grain yield. However, the same study showed that Ni doses above 125 mg Ni kg seeds^-1 reduced these parameters and were associated with symptoms of Ni toxicity. Similar effects were observed in our study at Ni doses of 120, 180, 240, and 300 mg kg^-1.

High CO₂ concentrations in the environment enhance plant growth, increase symbiotic activity, and promote nodulation in soybean inoculated with Bradyrhizobium japonicum (Akley et al., 2022). According to Dacal et al. (2019), gas emission processes and soil quality are influenced by microbial activity, root respiration, decomposition of plant residues, and oxidation of organic matter. The soil bacterial community engages in intense decomposition activity, leading to the oxidation of organic compounds and the emission of CO₂ into the gas phase (Atere et al., 2020). Thus, higher levels of emission of CO₂ are indicative of soils with high respiration rates, microbial activity, and microbial diversity (Sun et al., 2024). The use of an Arduino prototype to evaluate CO₂ release (Kormann et al., 2024) in Experiment III allowed a comprehensive and continuous understanding of gas dynamics throughout the soybean crop cycle.

Throughout most of the experimental period, CO₂ concentrations were higher in the control treatment and in the treatments with Ni doses of 60 and 120 mg kg^-1; these treatments also led to the highest recovery of Bradyrhizobium from treated seeds. In all treatments, the CO₂ concentration increased over time until the flowering stage of soybean and declined thereafter. Increases in CO₂ concentrations may indicate greater microbial activity in the soil (Zapata et al., 2021). Larger microbial populations suggest an increase in microbial biomass and, consequently, greater C immobilization, which is important from the perspective of C sequestration (Rao et al., 2021).

Ni doses above a threshold value adversely affect nodulation and overall plant health (Lavres et al., 2016). The architecture of root development—including root hair formation, primary root elongation, and lateral root initiation—represents a complex interaction between root growth dynamics and nutrient availability (Moretti et al., 2020b). Higher Ni concentrations impair nodule development and function, which are essential for BNF and plant growth (Freitas et al., 2018). Ni application rates up to 60 mg kg⁻¹ consistently enhanced nodulation, and physiological vigor without adversely affecting 100-grain weight, supporting the hypothesis that moderate doses of seed-applied Ni are agronomically safe and effective. By contrast, higher doses of 180–300 mg kg⁻¹ led to sharp declines in NN and NDW, as well as N metabolism associated with BNF. These findings align with the results of previous studies indicating that Ni toxicity negatively impacts N fixation, root morphology, and phenylpropanoid pathway activity, ultimately compromising plant establishment and yield. The influence of Ni on the regulation of nodulation-related gene expression appears to be mainly mediated by its effects on phenylalanine ammonia-lyase (PAL) activity and nod gene expression. The expression of PAL, a key enzyme in the phenylpropanoid pathway, is upregulated in response to metal stress, leading to increased synthesis of phenols, flavanones, and isoflavanones (Shen et al., 2022).

The decrease in the photosynthetic rate (A) at higher Ni doses, combined with elevated intercellular CO₂ concentrations (Ci)—partly due to reduced stomatal conductance (gs), i.e., stomatal closure—suggests that decreased nodulation not only has direct detrimental effects on photosynthesis but also compromises the carbon fixation capacity of soybean plants (Pietrini et al., 2015).

Importantly, our findings indicate that an adequate supply of Ni positively regulates stomatal opening and closure, directly enhancing gas exchange processes and contributing to increased biomass accumulation. The observed improvement in photosynthetic activity may have resulted from enhanced nutrient uptake and/or reduced Ni absorption. Specifically, lower Ni doses increased RDW compared to the control. A more developed root system improves soil exploration, thereby promoting greater root distribution and activity in soybean plants. This, in turn, can indirectly support key metabolic functions and facilitate successful plant establishment.

Plants that displayed more pronounced symptoms of Ni toxicity exhibited elevated urease activity. Under abiotic stress conditions such as Ni toxicity, the activities of arginine decarboxylase and ornithine decarboxylase are typically upregulated. These enzymes catalyze arginine degradation, releasing spermidine and spermine in mitochondria and urea in the cytosol (Reis et al., 2017). Thus, the observed increase in urease activity in soybean plants exposed to Ni doses exceeding 180 mg kg^-1 likely represents a physiological response to enhanced arginine catabolism and the consequent rise in cytosolic urea levels (Macedo et al., 2016). Furthermore, urea and ammonia can be direct products of ureide degradation through the urease pathway (Todd et al., 2006).