Soybean [Glycine max (L.) Merr.], with its origin traced to the Yellow River Basin of northern China, is an economically vital crop that serves as both a grain and oil source. Its seeds not only supply plant-derived proteins and oils essential to human diets, but also contain a wide array of health-promoting bioactive constituents-among them isoflavones, lecithin, polypeptides and dietary fiber (Li et al., 2008; Meng et al., 2024). Under the pressure of accelerating global climate change, shifts in regional precipitation patterns have made drought episodes increasingly common and severe. Consequently, water deficit has become one of the foremost abiotic stressors that curtail soybean yield potential and impede quality improvement  (Burner et al., 2026).
       
At the physiological level, drought stress initially disrupts the water balance in plants, leading to a notable decline in tissue water content. The ensuing water deficit drives excessive generation of reactive oxygen species (ROS), which initiate lipid peroxidation in membranes and result in a marked rise in malondialdehyde (MDA) levels, ultimately compromising cell membrane integrity (Zhao et al., 2025). In response to drought-induced oxidative stress and dehydration, plants have developed sophisticated physiological defense strategies. One branch of this defense involves the accumulation of osmolytes-including proline (Pro), soluble sugars (Ss) and soluble proteins (Sp)-that reduce cellular osmotic potential, thereby supporting water absorption and retention (Zhou et al., 2023; Sistu et al., 2023). Another branch involves the induction of antioxidant enzyme systems comprising superoxide dismutase (SOD), peroxidase (POD), catalase (CAT) and ascorbate peroxidase (APX), which act in concert to eliminate surplus ROS and limit cellular injury (Yang et al., 2021; Feng et al., 2024). Beyond these direct defense responses, endogenous phytohormones serve as indispensable signaling molecules that coordinate plant growth, differentiation, flowering and fruiting through elaborate regulatory networks (Virág et al., 2025). A substantial body of evidence indicates that drought rapidly reshapes the hormonal metabolic landscape. The drought-induced accumulation of abscisic acid (ABA) is a well-established hallmark of drought signal transduction (Ali et al., 2020; Puértolas et al., 2025). At the same time, growth-related hormones such as indole-3-acetic acid (IAA) and gibberellins (GA) undergo dynamic changes that contribute to growth suppression and the reallocation of metabolic resources under stress (Salehifar et al., 2014).
       
On this basis, the present study employed two soybean cultivars with contrasting drought tolerances as experimental materials. Dynamic changes in leaf membrane lipid peroxidation, osmolyte accumulation and antioxidant enzyme activities were continuously monitored over a 3- to 12-day period under different drought-stress intensities. Endogenous hormone levels were additionally quantitatively analyzed at key stress time points. However, previous studies of soybean drought responses have typically examined these physiological parameters in isolation and integrative data linking dynamic endogenous hormone profiles with antioxidant enzyme kinetics remain scarce for contrasting cultivars under graded osmotic stress. The study was designed to systematically characterize the physiological adaptations and dynamic response patterns of different drought-tolerant soybean varieties to varying drought intensities and durations. The results are expected to provide a theoretical foundation for elucidating the physiological regulatory mechanisms underlying drought tolerance in soybean and to support breeding efforts to develop drought-resistant cultivars.

Experimental design and plant materials
 
Two soybean cultivars with contrasting drought responses were used in this study: HN-44 (drought-tolerant) and HN-65 (drought-sensitive) (Wang et al., 2012). The experiment was conducted from May to October 2025 at the experimental station of Heilongjiang Agricultural Engineering Vocational and Technical University (Harbin, Heilongjiang, China). Seedling establishment utilized a sand-filled pot system. Cylindrical plastic vessels (both height and diameter: 30 cm) were equipped with four square drainage ports (each 1 cm × 1 cm) on the bottom surface and subsequently overlaid with fine-grade mesh to prevent substrate escape. Purified river sand was loaded into each vessel to a level 3 cm below the upper rim and the sand was thoroughly pre-saturated with water before seed introduction.
       
Six seeds, selected for uniformity in size and appearance, were sown per vessel beneath a 1-cm layer of dry sand. Daily provision of 500 mL distilled water per vessel was maintained from sowing through the cotyledon-expansion stage (VC). Upon reaching VC, thinning was performed to retain three vigorous seedlings of comparable development per vessel and the daily irrigation regime was then converted to 500 mL of complete nutrient medium per vessel. Refer to the nutrient medium formulation from Li et al., (2026).
       
Drought treatments were imposed when seedlings reached the three-trifoliate-leaf stage (V3). Four osmotic stress levels were created by dissolving polyethylene glycol 6000 (PEG-6000) in nutrient solution: 0% (CK, well-watered control), 7.5% (T1, mild stress), 15% (T2, moderate stress) and 22.5% (T3, severe stress). Throughout the treatment period, 500 mL of the respective solution was supplied to each container daily. Physiological tissue harvesting was performed during the morning hours (08:00-09:00) on days 3, 6, 9 and 12 post-treatment onset, targeting the second-youngest fully expanded leaf from the apex. Three independent biological replicates were secured per treatment × time combination. To ensure representativeness of the sample, leaf material from all three co-cultivated seedlings was collected.
 
Determination of physiological parameters
 
Determination of MDA content and relative water content
 
Malondialdehyde (MDA) quantification was performed using the thiobarbituric acid (TBA) colorimetric assay. For determination of leaf relative water content (RWC), the second-youngest fully expanded leaves were swiftly excised at designated sampling moments and immediately transferred to an ice-chilled container. Upon arrival at the laboratory, the fresh mass (m₁) was recorded promptly. Leaf samples were subsequently submerged in distilled water within 50 mL centrifuge tubes for a 2-hour rehydration interval. Following gentle removal of surface moisture, turgid mass (m₂) was determined. Samples were then oven-dried at 65°C until mass stabilization, yielding dry mass (m₃). RWC was computed following the protocol of Li et al., (2000):.

 
Determination of antioxidant enzyme activities
 
Superoxide dismutase (SOD) activity was quantified using the nitroblue tetrazolium (NBT) photoreduction assay (Flohe and Ötting, 1984). Peroxidase (POD) activity was measured using guaiacol as the electron-donor substrate, whereas catalase (CAT) activity was assessed by monitoring the decline in ultraviolet (UV) absorbance attributable to H₂O₂ decomposition. Ascorbate peroxidase (APX) activity was determined by monitoring the oxidation-linked reduction in ascorbic acid absorbance at 290 nm in the presence of H₂O₂. Assay execution for POD, CAT and APX conformed to methodologies delineated by Wang and Huang (2015).
 
Determination of osmolytes
 
Proline (Pro) quantification applied the sulfosalicylic acid extraction coupled with the ninhydrin colorimetric procedure. Soluble sugar (Ss) measurement utilized the anthrone-sulfuric acid colorimetric technique, while soluble protein (Sp) determination employed the Coomassie brilliant blue G-250 dye-binding assay. Osmotic adjustment substance assays adhered to protocols specified by Wang and Huang (2015).
 
Quantification of endogenous hormones
 
Leaf contents of IAA-Glu, GA3, SA and ACC were quantified by high-performance liquid chromatography (HPLC) according to the procedure described by Wang and Huang (2015). Abscisic acid (ABA) was not included in this hormone panel.
 
Statistical analysis
 
Raw data were organized and tabulated in Microsoft Office Excel 2010. All statistical analyses were performed using IBM SPSS Statistics (Version 21.0, IBM Corp., Armonk, NY, USA). Mean comparisons among treatments were conducted via Duncan’s multiple range test (α = 0.05). Before ANOVA, normality was assessed using the Shapiro–Wilk test (p>0.05) and homogeneity of variance was assessed using Levene’s test (p>0.05). A multifactor ANOVA was used to evaluate the independent and interactive effects of cultivar, PEG concentration and stress duration on each measured variable. Figures were generated using Origin 9 (OriginLab Corp., Northampton, MA, USA).

Effects of drought stress on leaf RWC and MDA content in different soybean varieties
 
Leaf relative water content (RWC) and malondialdehyde (MDA) levels are critical physiological indicators that reflect plant water status and the extent of cellular membrane damage. As illustrated in Fig 1, an increase in both PEG concentration and stress duration led to a progressive decline in leaf RWC, accompanied by a continuous accumulation of MDA in both soybean cultivars. The most severe physiological disruptions were recorded under the T3 treatment on day 12.


       
A comparison between the two genotypes revealed that HN-44 exhibited superior water retention capacity and stronger resistance to membrane lipid peroxidation. As depicted in Fig 1A and 1B, the reduction in RWC was relatively marginal for both cultivars during the initial phase of stress (day 3). However, as the drought intensified, HN-65 experienced a markedly faster rate of water loss compared to HN-44. By day 12 under the severe T3 treatment, the RWC in HN-44 and HN-65 had plummeted by 60.47% and 72.09%, respectively, relative to their corresponding controls (CK).
       
Furthermore, cellular dehydration severely exacerbated membrane lipid damage (Fig 1C and 1D). Under well-watered conditions, MDA contents remained at low baseline levels in both varieties. However, upon exposure to drought stress, MDA accumulation in HN-65 was consistently higher than in HN-44 across all time points. By the 12^th day of the T3 treatment, the MDA contents in HN-44 and HN-65 had surged to 2.97-and 3.71-fold of their respective CK values. These results indicate that under drought conditions, HN-44 is highly capable of maintaining cellular water homeostasis and effectively mitigating water-deficit-induced membrane lipid peroxidation, thereby demonstrating greater drought tolerance.
 
Impact of drought stress on SOD and POD activities in different soybean cultivars
 
Water deficit conditions generally trigger elevated antioxidant enzyme activities in plant tissues, providing a biochemical shield against ROS-induced cellular injury. Regarding superoxide dismutase (SOD) activity (Fig 2A and 2B), both genotypes exhibited a significant dose-dependent elevation in response to increasing PEG concentrations during the early stress phase (days 3-6). However, as the stress period extended to days 9 and 12, the SOD activity profile shifted to a parabolic trend, characterized by an initial rise followed by a subsequent decline. On day 9, SOD activities in both cultivars peaked under the T2 treatment, registering remarkable increases of 59.8% and 48.2% for HN-44 and HN-65, respectively, compared to their well-watered controls. By day 12, the enzymatic activities under the moderate (T2) and severe (T3) stress conditions were markedly suppressed in both varieties. Nevertheless, HN-44 maintained relatively high SOD activity under the mild-stress (T1) regime, indicating a more durable regulatory capacity.


       
In terms of peroxidase (POD) activity, both cultivars generally exhibited an upward trend in parallel with increasing stress severity and duration (Fig 2C and 2D). Overall, the POD response in HN-44 leaves was not only more prompt but also greater in magnitude. On day 9, the POD activities in HN-44 and HN-65 under the T3 treatment increased substantially by 61.5% and 49.3%, respectively, relative to the CK. By day 12, although the enzymatic activities slightly waned under high-concentration treatments, HN-44 still maintained a 39.7% increase over its control under the T3 condition, whereas the increment in HN-65 dwindled to merely 27.7%. Collectively, these observations indicate that HN-44 maintains a more resilient SOD and POD defense capacity than HN-65. Particularly under severe or prolonged water deficit, its antioxidant defense mechanism is less inhibited, enabling more efficient scavenging of intracellular ROS.
 
Effects of drought stress on CAT and APX activities in different soybean varieties
 
CAT and APX cooperate to decompose H₂O₂ and sustain intracellular redox balance under oxidative conditions. As depicted in Fig 3, the overall enzymatic dynamics of CAT and APX in both genotypes exhibited a bell-shaped response pattern as PEG concentration and stress duration escalated. Notably, the drought-tolerant cultivar HN-44 demonstrated a markedly superior capacity for enzymatic upregulation relative to HN-65.


       
During the initial 3 to 9 days of the stress period, both cultivars exhibited a continuous increase in CAT activity in parallel with intensifying drought severity (Fig 3A and 3B). By day 9, CAT activities culminated under the severe T3 treatment, with HN-44 and HN-65 exhibiting striking surges of 122.07% and 117.31%, respectively, over their unstressed controls (CK). As the stress prolonged to day 12, a fluctuation characterized by an initial upsurge followed by a drop was observed across the concentration gradient. Nonetheless, CAT levels in the T3 groups remained significantly elevated above the CK baseline, retaining increments of 45.30% for HN-44 and 32.19% for HN-65. This confirms that the enzymatic defense machinery of HN-44 is less compromised by prolonged, severe water deficit.
       
The dynamic fluctuations of APX closely mirrored those of CAT, yet the physiological superiority of HN-44 was even more pronounced (Fig 3C and 3D). Under the T3 regime on day 9, APX activities peaked in both genotypes. Specifically, HN-44 registered a staggering 165.0% enhancement relative to the CK, distinctly outpacing the 131.4% increase observed in HN-65. By the 12^th day, the APX response curve across treatments shifted to an inverted U-shape, cresting at the moderate T2 level. At this juncture, HN-44 maintained a robust 119.07% elevation over the control, whereas the increment in HN-65 had contracted to 74.27%.
 
Effects of drought stress on osmolyte accumulation in different soybean varieties
 
To sustain cellular water influx during episodes of water deficit, plants proactively synthesize and sequester osmo-protectants-namely proline (Pro), soluble sugars (Ss) and soluble proteins (Sp)-which serve to depress the leaf osmotic potential. According to Fig 4, the progressive intensification and prolongation of the drought treatments led to a substantial upward trajectory in all three osmolyte pools in both genotypes. Notably, the magnitude of this accumulation was consistently more pronounced in HN-44 than in its susceptible counterpart, HN-65.


       
A distinct dose-dependent enrichment of Pro was observed as the stress severity escalated (Fig 4A and 4B). By the end of the experimental period (day 12) under the extreme T3 regimen, Pro concentrations in both cultivars soared to their respective maxima, yielding dramatic expansions of 22.78-fold for HN-44 and 16.37-fold for HN-65 relative to the well-watered baselines. Throughout the entire stress timeline, the absolute Pro inventory in HN-44 consistently overshadowed that of HN-65, underscoring a superior capacity for osmotic adjustment in the tolerant genotype.
       
The temporal dynamics of Ss closely mirrored those of Pro (Fig 4C and 4D). Even a brief (3-day) exposure to mild stress (T1) was sufficient to elicit an accumulation of Ss, with the margin of increase widening progressively as the stress duration increased. On day 12 under the T3 condition, the S-pool peaked, registering impressive increases of 154.74% and 123.36% above the CK values for HN-44 and HN-65, respectively.
       
In contrast, while the relative amplification of the Sp pool was less drastic than those observed for Pro and Ss, distinct genotypic discrepancies were nonetheless evident. On day 12 of the T3 treatment, the Sp contents in HN-44 and HN-65 were elevated by 51.39% and 35.44%, respectively, compared to their controls. Interestingly, while the early-stage (day 3) mild and moderate stresses (T1, T2) failed to induce significant Sp fluctuations in HN-65 relative to the CK, HN-44 exhibited a statistically significant Sp enrichment across all stress intensities and sampling time points.
 
Analysis of variance for physiological parameters
 
A three-way ANOVA was conducted to evaluate the main and interactive effects of genotype, PEG concentration and stress duration on all measured physiological parameters (Table 1). All three main factors-genotype, PEG concentration and sampling day-significantly affected leaf water status, membrane lipid peroxidation, antioxidant enzyme activities and osmolyte accumulation (p<0.001).


       
Regarding interactive dynamics, the genotype × time interaction was highly significant (p<0.001) across all evaluated traits. This statistically confirms that as the water deficit prolonged, the physiological trajectories of HN-44 and HN-65 diverged considerably rather than following a uniform pattern. Furthermore, a significant three-way interaction (Genotype × concentration  × time, p<0.05) was detected for the variables MDA, SOD, CAT, APX, Pro and Ss. This finding underscores that the magnitude of physiological divergence between the two soybean cultivars is intricately modulated by both the severity of osmotic stress and the duration of treatment.
 
Impact of drought stress on endogenous hormone profiles in different soybean varieties
 
The dynamic equilibrium of endogenous phytohormones serves as the core signaling hub orchestrating plant physiological responses to environmental adversity. Building upon our preceding evaluations of water status, membrane lipid peroxidation and antioxidant enzyme activities, day 9 of the stress period emerged as a critical juncture for physiological adaptation. At this stage, the cellular antioxidant defense machinery was maximally mobilized, resulting in the most pronounced biochemical discrepancies between HN-44 and HN-65. Consequently, foliar tissues harvested on day 9 were selected for the targeted quantitative profiling of endogenous hormones. The analytical outcomes revealed that escalating PEG concentrations induced dramatic shifts in hormonal pools, characterized by distinct genotypic response patterns.
       
Both salicylic acid (SA) and indole-3-acetic acid-glutamate (IAA-Glu) exhibited a robust positive correlation with stress severity. Notably, HN-44 outpaced HN-65 in both absolute accumulation and the relative amplitude of increase. The fluctuation in SA was particularly striking; under the extreme T3 regime, the SA reservoir in HN-44 expanded by an impressive 165.90% relative to the control, whereas the equivalent treatment in HN-65 elicited a much more modest 64.42% increment. A parallel trend was observed for IAA-Glu dynamics. The T3 condition triggered enhancements of 88.35% and 44.93% over the baseline in HN-44 and HN-65, respectively.
       
Conversely, the accumulation of 1-aminocyclopropane-1-carboxylic acid (ACC)-a direct precursor to ethylene biosynthesis-was markedly suppressed. Compared to their respective well-watered counterparts, the T3 treatment reduced ACC content by 38.67% in the susceptible HN-65, while the reduction in the tolerant HN-44 was slightly milder at 32.66% (Fig 5).


 
Two-way ANOVA of endogenous hormone profiles
 
A two-way ANOVA was applied to partition the individual and combined effects of genotype and drought treatment on endogenous hormone contents (Table 2). Both genotype and drought regime independently exerted highly significant effects on all four assayed phytohormones (p<0.001).


       
Regarding the interaction terms, a robust genotype × treatment interactive effect was detected across all four evaluated hormones (p<0.01 or p<0.001). Notably, salicylic acid (SA) accounted for the largest proportion of variance, with a magnitude that overwhelmingly surpassed those of the other hormonal variables. This compelling statistical evidence suggests that SA serves as a pivotal mediator driving the divergence in drought tolerance between the two soybean cultivars.
       
Water scarcity constrains plant development by limiting cellular hydration and disturbing normal metabolic pathways (Pamungkas and Farid, 2022). Leaf RWC responds rapidly to changes in water availability, providing a direct measure of tissue water status and the capacity to retain moisture under deficit conditions (Hu et al., 2026). Although stomatal closure reduces water loss through transpiration, this protective response simultaneously restricts CO₂ entry into leaves, thereby diminishing photosynthetic carbon fixation. The resulting energetic imbalance within chloroplast photosystems promotes ROS overproduction. These highly reactive molecules subsequently attack membrane lipids, producing MDA and other peroxidation products that compromise membrane function (Rao et al., 2025). In the present experiment, increasing PEG concentration and longer stress exposure led to progressive declines in RWC accompanied by rising MDA levels in both cultivars, confirming earlier reports on soybean responses to water deficit (Hossain et al., 2024). Across all sampling times, however, HN-44 maintained higher RWC and lower MDA than HN-65, indicating that the tolerant cultivar was more effective at preserving cellular hydration and limiting membrane damage.
       
Antioxidant enzyme systems represent a critical line of defense against oxidative injury. When drought-induced ROS generation surpasses basal scavenging capacity, plants typically enhance enzyme activities to restore redox homeostasis (Wang et al., 2025). SOD catalyzes the dismutation of superoxide into H₂O₂, which is subsequently detoxified by POD, CAT and APX (Mishra et al., 2023; Guo et al., 2024; Fathi et al., 2025). In this study, enzyme activities generally rose from day 3 to day 9, reflecting active upregulation of the antioxidant machinery during the early and middle phases of stress. The subsequent decline under prolonged or severe treatments suggests that excessively high stress can impair enzyme synthesis, stability, or overall redox buffering. Compared with HN-65, HN-44 maintained higher and more stable enzyme activities throughout the stress period, supporting the interpretation that a robust and durable enzymatic defense system contributes to its drought-tolerant phenotype.
       
Osmotic adjustment through compatible solute accumulation also plays a central role in drought adaptation. Under reduced water potential, solutes such as Pro, Ss and Sp help maintain turgor and protect cellular structures (Dong et al., 2020; Song et al., 2022). The present results showed that all three osmolytes increased with stress intensity in both cultivars, but accumulation was consistently greater in HN-44. The particularly pronounced increases in Pro and Ss relative to Sp suggest that these smaller molecules are more efficient for rapid osmotic regulation under severe stress. The comparatively modest and delayed accumulation of Sp observed here is consistent with its secondary role in osmotic adjustment: soluble proteins are multifunctional macromolecules whose de novo synthesis is energetically costly, so under intense stress plants may preferentially allocate resources to smaller, more rapidly accumulated osmolytes. Beyond lowering osmotic potential, Pro can also scavenge ROS and stabilize proteins and membranes (Haghpanah et al., 2024). The stronger accumulation of Pro and Ss in HN-44 thus likely contributed to both improved water retention and enhanced cellular protection.
       
Phytohormones coordinate environmental perception with growth regulation and stress adaptation. SA has emerged as a key regulator of plant responses to abiotic stress, including drought. In this study, SA content increased markedly with stress severity and the increase was substantially greater in HN-44 than in HN-65. The ANOVA results further showed that SA accounted for the largest share of cultivar-related hormonal variation, suggesting that strong SA accumulation is closely associated with the drought-tolerant phenotype of HN-44. IAA-Glu and GA3 also increased under drought, particularly in HN-44. IAA-Glu, an inactive conjugated form, may reflect adjustments in auxin homeostasis during stress (Isobe and Miyagawa, 2022). Although GA3 is typically associated with growth promotion, its increase under drought may reflect a complex feedback mechanism that balances growth regulation and stress tolerance (Bai et al., 2011; Man et al., 2017). ACC, the immediate precursor of ethylene, showed an opposite trend from SA, IAA-Glu and GA3, declining as stress severity increased. This reduction may result from suppressed ACC biosynthesis, accelerated conversion to ethylene, substrate limitation, or feedback regulation within the ethylene pathway (Van de Poel and Van Der Straeten, 2014). The stronger decline in HN-65 suggests that ethylene-related metabolism may be more strongly perturbed in the drought-sensitive cultivar. It should also be acknowledged that abscisic acid (ABA), a canonical drought-signaling hormone, was not included in the hormone panel of this study; the interplay between ABA and SA in shaping cultivar-specific drought responses merits investigation in future studies. Taken together, the physiological and hormonal evidence indicates that HN-44 achieves superior drought adaptation through coordinated mechanisms encompassing water conservation, reduced membrane damage, stronger antioxidant defense, greater osmotic adjustment and pronounced SA-associated hormonal regulation.
       
It should be noted that the PEG-6000 osmotic stress model used here simulates water deficit through a reduction in the water potential of the rooting medium but does not fully replicate the process of soil drying in the field, which additionally involves soil-root hydraulic feedback, altered ion availability and progressive spatial heterogeneity of soil moisture. Accordingly, the physiological responses characterized in this study may not fully reflect those elicited under natural field drought and extrapolation of these findings to field conditions should be made with caution.

This study employed a PEG-6000-mediated osmotic stress system to compare the physiological and hormonal responses of drought-tolerant HN-44 and drought-sensitive HN-65 soybean seedlings over a 12-day stress period. Water deficit reduced RWC and increased MDA in both cultivars, but HN-44 retained higher leaf water content and exhibited less membrane lipid peroxidation. Antioxidant enzymes and osmotic adjustment substances were activated in response to stress, with HN-44 maintaining stronger and more sustained enzyme activities and greater accumulation of Pro and Ss. Hormone profiling revealed that SA, GA3 and IAA-Glu increased under drought, whereas ACC declined. Among these hormonal changes, SA showed the strongest association with cultivar differences, indicating that SA accumulation may be an important contributor to the drought tolerance of HN-44. Overall, HN-44 displayed a more integrated and resilient drought-response system than HN-65, coordinating water conservation, antioxidant protection, osmotic adjustment and hormonal regulation. Future research should therefore investigate the molecular mechanisms underlying SA-mediated drought tolerance in HN-44, for example through transcriptomic and metabolomic analyses of SA signaling and validate these physiological findings under field conditions.

This work was supported by the Scientific Research Startup Project for Doctoral Talent Introduction at Heilongjiang Agricultural Engineering Vocational and Technical University (HAU-19).
 
Disclaimers
 
The views and conclusions expressed in this article are solely those of the authors and do not necessarily represent the views of their affiliated institutions. The authors are responsible for the accuracy and completeness of the information provided, but do not accept any liability for any direct or indirect losses resulting from the use of this content.

The authors declare no conflicts of interest regarding the publication of this article.