Plant Growth Regulators Improve Soybean Yield in Northwest China Through Nutritional and Hormonal Regulation

Abstract

In Xinjiang, soybean yield potential is constrained by varietal limitations and abiotic stresses. Although plant growth regulators (PGRs) can enhance crop yield, their specific impacts on soybean production, endogenous hormone regulation, and nutrient dynamics in this region have not been well characterized. This study evaluated the effects of different PGR treatments on yield formation, hormone levels, and nutrient contents through a field experiment conducted in Ili, Xinjiang, from 2023 to 2025. Foliar applications of naphthaleneacetic acid (NAA, 300 g ha⁻¹), prohexadione-calcium (Pro-Ca, 450 g ha⁻¹), and iron chlorine e6 (ICE6, 45 g ha⁻¹) were applied twice—at the fourth trifoliolate and full-pod stages—with an untreated control (CK) for comparison. Compared with CK, PGR treatments increased biomass allocation to reproductive organs by 6.2%, nutrient accumulation of N, P, and K by 12.3%, 25.5%, and 6.5%, respectively, pod number by 6.92 pods per plant, seed number by 4.86 seeds per plant, and 100-seed weight by 0.47 g, resulting in 6.6–12.0% higher grain yield. Seed PGR residues were 0.009 mg kg⁻¹. PGR application enhanced reproductive organ conversion capacity, nutrient uptake efficiency, and regulated endogenous hormone levels, clarifying internal yield-formation mechanisms and offering valuable reference for soybean research, particularly in similar latitudes.

1. Introduction

Soybean (Glycine max (L.) Merr.) holds a strategic position within the Chinese agricultural landscape, functioning as a major source of grain, vegetable oil, and protein-rich feed. Owing to its remarkable nutritional attributes, it remains essential to both primary agricultural systems and the wider agro-industrial economy [1]. Xinjiang, recognized as one of the high-yield regions for soybean cultivation in China, recorded an average yield in 2022 that exceeded the national mean of 1980.08 kg ha⁻¹ by 52.6 percent [2]. Soybean cultivation is influenced by both inherent varietal characteristics and various abiotic stress factors, which can result in adverse outcomes such as pod abscission, incomplete seed filling, and inefficient nutrient translocation [3]. In Xinjiang, these constraints collectively prevent the crop from reaching its optimal yield potential, even under favorable agronomic conditions in certain areas. Conventional cultivation practices, such as regulating planting density and irrigation volume, are often ineffective in promptly mitigating sudden abiotic stresses, including short-term heatwaves and soil compaction [4,5,6]. Such limitations highlight the urgent need to identify and develop efficient and easily deployable strategies that can safeguard crop performance under rapidly changing environmental conditions.

The judicious application of plant growth regulators (PGRs) has been widely recognized as a convenient and effective agronomic approach to enhance crop yield and improve product quality by modulating physiological and biochemical processes, such as cell division, elongation, and nutrient translocation [7]. Moreover, PGRs can contribute to greater stress resilience, enabling crops to maintain productivity under suboptimal environmental conditions [8]. For example, spraying iron chlorine e6 (ICE-6) and prohexadione-calcium (Pro-Ca) has been shown to increase the effective spike number and yield of rice [9,10]. Pro-Ca increases seed weight in lentil plants [11] and promotes flower production in apples, although it may reduce fruit set [12]. The naphthaleneacetic acid (NAA) exhibits notable effects in improving fruit set, number, and quality in peppers [13], while its pre-flowering application to cucumbers can induce fruit development and enhance fruit set [14]. In soybeans, NAA, Pro-Ca, and ICE6 significantly reduce flower and pod drop [8], yet their comparative efficacy and optimal application strategies remain insufficiently studied. A previous study examined the effects of NAA, Pro-Ca, and ICE-6 on soybean; however, the investigation focused exclusively on floral and pod development [8], without extending to yield performance, biomass distribution, nutrient (N, P, K) dynamics, or seed quality assessments. The present study broadens the scope of evaluation to encompass a comprehensive set of agronomic and physiological parameters, thereby providing a more complete understanding of the effects of these PGRs. Nevertheless, although previous research has reported that the application of NAA, Pro-Ca, and ICE-6 can effectively increase crop yield, their specific influence on the physiological regulatory mechanisms underlying yield improvement in soybeans grown under the agro-climatic conditions of Xinjiang remains insufficiently understood. Furthermore, it is still unclear whether such applications lead to residual compounds in soybean seeds. Addressing these knowledge gaps is essential for developing region-specific strategies that optimize soybean productivity while ensuring seed quality and safety.

To address the aforementioned unresolved issues, this study investigates whether foliar application of the plant growth regulators NAA, Pro-Ca, and ICE-6 can increase reproductive biomass allocation, promote nutrient accumulation, and modulate endogenous hormone profiles to improve soybean yield and seed quality in Northwest China. We hypothesize that these PGRs will achieve such improvements while keeping residue levels in seeds within safety thresholds. The findings provide novel insights into how PGR application modulates endogenous hormone contents and nutrient levels in soybean, potentially promoting soybean yield improvement.

2. Materials and Methods

2.1. Experimental Site

This field study was conducted from 2023 to 2025 in Qapqal County (43°83′ N, 81°20′ E), Ili Prefecture, Xinjiang Uygur Autonomous Region, China. The experimental soil was classified as sandy loam. Within the 0–30 cm depth, the soil exhibited the following properties across the three consecutive growing seasons: pH values of 7.9, 8.1, and 8.2; available phosphorus (as P₂O₅) contents of 17.4, 17.7, and 16.9 mg kg⁻¹; and available potassium (as K₂O) levels of 131.4, 134.1, and 137.6 mg kg⁻¹; total nitrogen contents of 0.069%, 0.073%, and 0.074% for the years 2023, 2024, and 2025, respectively. Soil agrochemical properties were determined before planting. Soil pH was determined potentiometrically in a 1:2.5 soil-to-water suspension, following the Soil pH in water method (ISO 10390:2005) [15]. Available phosphorus was measured using the Olsen P method (0.5 mol L⁻¹ sodium bicarbonate extraction, pH 8.5; Sinopharm Chemical Reagent Co., Ltd., Beijing, China) [16]. Available potassium was extracted with 1 mol L⁻¹ ammonium acetate (pH 7.0; Sigma-Aldrich, St. Louis, MO, USA) and determined by flame photometry according to the Available potassium by ammonium acetate extraction protocol [17]. Total nitrogen was determined by the Kjeldahl method [18]. All analyses were conducted in triplicate to ensure accuracy and reproducibility.

Meteorological data for this study were obtained from the county agrometeorological bureau serving the experimental site. The site layout and observation elements were designed in accordance with the standards of international organizations, such as the World Meteorological Organization. In the study area, mean air temperatures during April–September in 2015–2022 ranged from 13.2 °C to 25.3 °C, and precipitation totals from 0.38 mm to 5.53 mm. Records for 2023–2025 fell within these ranges, indicating that the experimental years were representative of the long-term climatic conditions. Meteorological conditions during the soybean growing season in the Ili region (April–September) are presented in Figure 1.

2.2. Experimental Design and Management

The study spanned three independent growing seasons (2023–2025). For statistical rigor, data from each season were first analyzed separately before being combined, when appropriate. A randomized complete block design was employed, with each treatment replicated three plots. Four foliar spray applications were administered, and treatments were randomly allocated to experimental units within each block. Foliar treatments were applied twice during the growing season: once at the fourth trifoliolate stage (31 May) and once at the full-pod stage (12 July). The soybean growth stages were classified according to the system of Fehr and Caviness (1971) [19], which distinguishes vegetative stages (V1–Vn) based on the number of fully developed trifoliate leaves and reproductive stages (R1–R8) indicate beginning bloom, full bloom, beginning pod, full pod, beginning seed, full seed, beginning maturity, and full maturity, respectively. At each application time, the treatments included naphthaleneacetic acid (NAA) at 300 g ha⁻¹, prohexadione-calcium (Pro-Ca) at 450 g ha⁻¹, iron chlorine e6 (ICE-6) at 45 g ha⁻¹, and an equivalent volume of water as the control (CK). The 5% NAA was obtained from Zhengzhou Zhengshi Chemical Co., Ltd. (Zhengzhou, China); the 8% Pro-Ca from Chengdu Guanzhi Agricultural Technology Co., Ltd. (Chengdu, China); and the 0.01% ICE-6 from Nanjing Biotek Biological Engineering Co., Ltd. (Nanjing, China).

The soybean cultivar XND-3 (Xinjiang Agricultural University, Urumqi, China), characterized by medium maturity and sub-limited pod formation, was used in this study. Sowing was carried out on 22 April. Each subplot measured 2.1 m in width and 10 m in length, with a plant spacing of 5 cm and a row spacing of 45 cm. Drip irrigation was employed.

Field management followed conventional local agricultural practices. Throughout the soybean growth cycle, five irrigation events were conducted, with a total water volume of 3750 m³. Prior to sowing, nitrogen fertilizer (70 kg ha⁻¹), phosphorus fertilizer (150 kg ha⁻¹), and potassium fertilizer (30 kg ha⁻¹) were incorporated into the soil via rotary tillage. Additionally, urea was applied at a rate of 150 kg ha⁻¹ through the drip irrigation system.

2.3. Data Collection

2.3.1. Yield Determination

At the full maturity stage, seed yield was determined by harvesting plants from five randomly located 1 m² quadrats within each plot. The harvested seeds from all quadrats were pooled, and the mean yield per plot was calculated to provide a representative estimate. All soybean plants within each selected area were harvested. After threshing, the number of pods, effective seeds, and seed weight were determined. Moisture content was adjusted to the standard value of 13.5% using a grain moisture meter. Yield was calculated at 13.5% moisture content, expressed as kg ha⁻¹, by converting the harvested fresh seed weight per 1 m² to a per-hectare basis. The 100-seed weight was also determined. Measurements for each 1 m² area were repeated five times, and mean values were used for subsequent analyses.

2.3.2. Plant Sampling and Nutrient Analysis

Plant sampling was carried out at four key soybean growth stages: beginning flowering (R1), beginning pod formation (R3), beginning seed fill (R5), and beginning maturity (R7). At each stage, three plants per plot were harvested and separated into vegetative organs (stems, leaves, and petioles) and reproductive organs (flowers, pods, and seeds). All samples were oven-dried initially at 120 °C for 30 min to deactivate enzymatic activity, and subsequently dried at 80 °C until a constant weight was obtained.

The accumulation of nitrogen (N), phosphorus pentoxide (P₂O₅), and potassium oxide (K₂O) was determined separately for each organ type. The dried samples were ground into a fine powder using a sample mill (Retsch ZM 200, Retsch GmbH, Haan, Germany) and digested with a concentrated H₂SO₄–H₂O₂ solution (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China). Nitrogen content was quantified using the Kjeldahl method [18], phosphorus content was determined spectrophotometrically via the vanadate–molybdate yellow colorimetric method, and potassium content was measured by atomic absorption spectrometry (AAnalyst™ 400, PerkinElmer Inc., Waltham, MA, USA) [20]. Nutrient accumulation (g m⁻²) was calculated by multiplying the nutrient concentration of each organ by its dry weight per unit ground area.

2.3.3. Phytohormone Extraction and Quantification

At 6, 12, 18, and 24 h after foliar application at the full pod stage (R4), fully expanded and healthy leaves from the middle of the inverted third node position were collected from soybean plants in each treatment plot. Samples were immediately wrapped in aluminum foil, flash-frozen in liquid nitrogen, and stored at −80 °C until analysis.

Cytokinin (CTK), indole-3-acetic acid (IAA), abscisic acid (ABA), and gibberellic acid (GA₃) contents were determined using HPLC–MS/MS (Agilent 1290–6460). Briefly, 0.5 g of leaf tissue was ground in a liquid nitrogen-pre-chilled mortar and extracted with 5 mL of n-propanol/water/HCl (2:1:0.002, v/v) at 100 rpm for 30 min at 4 °C, followed by the addition of 2 mL dichloromethane and further shaking for 30 min. After centrifugation (13,000 rpm, 5 min, 4 °C), 2 mL of the organic phase was frozen (−80 °C, 14 h), lyophilized (24 h), reconstituted in 200 μL of 50% methanol, vortex-mixed, and filtered (0.22 μm) for analysis [21].

Chromatographic separation was performed on an Eclipse Plus C18 column (2.1 mm × 50 mm, 1.8 μm, 35 °C; Agilent Technologies, Santa Clara, CA, USA) using a mobile phase of (A) 0.1% formic acid in water and (B) 0.1% formic acid in acetonitrile at 0.3 mL min⁻¹, with a 2 μL injection volume. Compounds were identified and quantified by retention times and calibration curves from external standards [22].

2.3.4. Seed Quality Assessment and Maximum Residue Limit Determination

Protein and oil contents were determined according to the official protocols of the American Oil Chemists’ Society (AOCS, 2009) [23]. Seeds from individual plants were pooled and ground using a universal cutting mill (Tecno Dalvo, Santa Fe, Argentina). Nitrogen content was measured by the Kjeldahl method and converted to protein using a factor of 6.25. Oil was extracted with n-hexane in a Soxhlet apparatus, the solvent removed by rotary vacuum evaporation at 40 °C, and oil content determined gravimetrically.

Maximum residue limits (MRLs) of NAA, Pro-Ca, and ICE-6 in soybean seeds were determined according to the Chinese local standard DB 37/T 3760-2019 [24]. Homogenized seed samples (5.0 g) were extracted with methanol (25 mL) by shaking for 30 min, centrifuged at 4000 rpm for 10 min, and filtered through a 0.45 μm membrane. NAA was quantified via HPLC on a C18 reversed-phase column using methanol–water (70:30, v/v) at 1.0 mL min⁻¹, detection at 280 nm, and external standard calibration. Pro-Ca was analyzed by HPLC with methanol–water (60:40, v/v) at 1.0 mL min⁻¹, detection at 230 nm, and 20 μL injection volume. Fe-dihydro-porphyrin was quantified by UV–Vis spectrophotometry at 410 nm (Soret band), using external standards. Results were expressed in mg kg⁻¹ and compared with the MRLs specified in GB 2763-2021 [25].

2.4. Statistical Analysis

All data were compiled and organized using Excel 2021 (Microsoft Corp., Redmond, WA, USA). Data were subjected to analysis of variance (ANOVA) in SPSS software (version 26.0; SPSS Inc., Chicago, IL, USA), and treatment means were compared using the Least Significant Difference (LSD) test at the 0.05 probability level. Figures were generated using Origin 2025 (OriginLab Corp., Northampton, MA, USA) and PowerPoint 2021 (Microsoft Corp., Redmond, WA, USA).

3. Results

3.1. Yield Components

Across the three experimental years, plant population (10⁴ plants ha⁻¹) remained generally unchanged among treatments, with differences within years <7% and no statistical significance (

Table 1. Effects of PGR treatments on soybean yield and yield components across three growing seasons (2023–2025).

Year	Treatment	Plant Population (104 Plants ha−1)	Pod Number (per Plant)	Seed Number (per Plant)	100-Seed Weight (g)	Yield (kg ha−1)	Increase Over CK (%)
2023	CK	40.11 a	41.76 c	64.82 c	18.67 b	4849.78 c	-
	NAA	39.67 a	55.47 a	73.31 a	18.70 b	5433.80 a	12.0
	Pro-Ca	40.34 a	50.07 b	69.93 b	18.91 b	5330.72 ab	9.9
	ICE-6	40.33 a	47.49 b	65.21 c	19.87 a	5220.95 b	7.7
2024	CK	43.67 a	42.38 b	60.45 b	18.25 c	4802.14 b	-
	NAA	43.11 a	50.42 a	66.08 a	18.40 c	5230.19 a	8.9
	Pro-Ca	42.00 a	50.60 a	65.22 a	18.76 b	5123.64 a	6.7
	ICE-6	42.89 a	48.00 a	62.16 b	19.58 a	5209.61 a	8.5
2025	CK	43.00 a	43.87 c	57.03 b	18.36 b	4489.31 c	-
	NAA	42.11 a	49.18 a	64.05 a	18.41 b	4952.61 a	10.3
	Pro-Ca	43.22 a	47.51 b	60.33 ab	18.43 b	4783.48 b	6.6
	ICE-6	40.33 a	47.56 b	64.37 a	19.05 a	4935.82 a	9.9
Source of variance
Y		**	ns	**	**	**
T		ns	**	**	**	**
Y × T		ns	**	ns	ns	ns

Note: For a given trait, treatments with the same letter within a year were not significantly different, based on Duncan’s multiple range test at p < 0.05, using a general linear model. ** represent a significant difference at the 1% level; ns represents no significant difference at the 5% level.

), indicating that PGR application did not affect final stand density. In contrast, reproductive traits exhibited marked and quantifiable improvements. Pod number per plant was consistently higher under NAA than CK, with increases ranging from 12.1% to 32.8%, while Pro-Ca and ICE-6 showed moderate gains of 8.3–19.9% depending on year. Seed number per plant followed a similar trend, with NAA elevating values by 9.3–13.1% over CK; ICE-6 increased seed numbers by up to 12.9%, and Pro-Ca by 5.8–7.9%.

For seed size, expressed as 100-seed weight, ICE-6 frequently produced the largest seeds, exceeding CK by 3.8–7.3%. NAA and Pro-Ca attained smaller but still positive changes (0.2–2.8%), whereas CK consistently recorded the smallest values. Grain yield (kg ha⁻¹) was markedly increased by PGR application, with NAA providing the largest improvement (8.9–12.0% over CK), Pro-Ca yielding an average gain of 7.7%, and ICE-6 achieving an average increase of 8.7%.

Statistical analysis showed that the year (Y) effect was highly significant (p < 0.01) for most traits, with the exception of pod number. Treatment (T) effects were significant (p < 0.01) for pod number, seed number, 100-seed weight, and grain yield. The year × treatment (Y × T) interaction was significant only for pod number (p = 0.006), indicating that the impact of PGRs on pod setting varied slightly among years. Overall, PGRs application enhanced key yield components—particularly pod number, seed number, and seed weight—by measurable margins, translating to consistent whole-plant yield gains, with NAA generally favoring reproductive capacity, ICE-6 improving seed size, and Pro-Ca delivering balanced improvements.

3.2. Biomass Allocation and Accumulation

Across three consecutive years, PGR application consistently shifted biomass allocation toward reproductive organs during late reproductive stages, with clear differences among treatments. Reproductive allocation remained below 5.0% at R1–R3 for all treatments, then increased sharply to 15.0–26.2% at R5 and 42.3–54.0% at R7 (Figure 2A). NAA reached the upper range (up to 54.0%) in 2023 and 2025, while Pro-Ca achieved similar maxima in 2024, significantly exceeding CK (which plateaued at 50.0–52.2%). ICE-6 maintained intermediate proportions (48.6%), yet still surpassed CK in most years (p < 0.05). Correspondingly, vegetative allocation decreased from 96.1% at R1–R3 to 46.0–57.7% at R7.

Biomass accumulation progressed steadily from early to late stages (Figure 2B). At R5, biomass reached 377.5–515.4 g m⁻², with NAA and ICE-6 exceeding CK by 15.6–33.0% (driven mainly by reproductive biomass gains of 31.3–60.6 g m⁻²). At R7, totals peaked at 822.4–1090.1 g m⁻²; NAA recorded the highest in 2023 (1049.8 g m⁻², 54.0% reproductive) and 2024 (1090.1 g m⁻², 48.3% reproductive), while Pro-Ca topped 2025 results (1019.8 g m⁻², 50.3% reproductive). ICE-6 achieved 995.5–1017.5 g m⁻² with reproductive proportions of 46.7–52.8%, consistently outperforming CK (822.4–931.0 g m⁻², 42.4–47.1% reproductive) in total and reproductive biomass.

Statistical analysis showed that the Y × T interaction was not significant (p = 0.133) for biomass allocation ratios of both vegetative and reproductive organs. This suggests that the relative partitioning of biomass between organs in response to treatments was consistent across years. The biomass accumulation in both vegetative organs (p = 0.04) and reproductive organs (p = 0.07) was highly significantly affected by the Y × T interaction (p < 0.01). This indicates that treatment effects on biomass allocation differed substantially between years.

Overall, NAA preferentially enhances reproductive sink capacity and late-stage accumulation, Pro-Ca delivers balanced yet sometimes superior reproductive biomass, while ICE-6 yields moderate but steady gains in both allocation and total biomass. CK consistently exhibited the lowest reproductive proportions and accumulation, underscoring the positive impact of targeted PGR application on yield-related biomass traits across multiple seasons (Figure 2).

3.3. Nutrient Accumulation in Soybean

Across all three growing seasons, the whole-plant accumulation of N, P₂O₅, and K₂O in soybean exhibited consistent responses to PGR treatments (NAA, Pro-Ca, and ICE-6) compared with the control (CK) at the R1, R3, R5, and R7 growth stages (Figure 3). For N, all PGRs increased accumulation by 6.3–23.7% over CK across growth stages, with NAA and ICE-6 producing greater gains at R7 (Figure 3A). P₂O₅ accumulation increased by 3.6–56.4%, with the largest rises at R7 under ICE-6. (Figure 3B). Whole-plant K₂O accumulation was also consistently higher under PGR treatments than CK, with the gap widening after R3; by R7, NAA, Pro-Ca, and ICE-6 maintained the highest levels (Figure 3C).

Considering statistical significance, whole-plant N accumulation differed mainly at R5 and R7 (Figure 3A), with NAA consistently higher than CK, and ICE-6 and Pro-Ca also showing increases at several stages. For P₂O₅, differences were most frequent at R7 (Figure 3B), when all PGR treatments exceeded CK, especially ICE-6. K₂O differences occurred mainly from R3 to R7, with PGR treatments generally higher than others, though the magnitude was limited.

Statistical analysis showed that the Y × T interaction was highly significant (p < 0.001) for N and P contents, but not significant (p = 0.508) for K content. This suggests that the effects of treatments on N and P varied considerably between years, whereas K content responses were relatively stable across years.

3.4. Phytohormone Content

Across all three years, the overall trends were consistent. In response to PGR treatments, CTK, IAA, and GA₃ concentrations in soybean leaves generally increased compared with CK, whereas ABA concentrations tended to decrease (Figure 4). The most pronounced increases occurred in CTK under NAA and ICE-6, particularly at 24 h (Figure 4a–c), with Pro-Ca also promoting modest increases (by approximately 27.0–83.2%). IAA showed a slight upward shift (3.5–31.0%) across treatments (Figure 4d–f), and GA₃ rose notably under NAA and ICE-6 at 6 h (10.0–31.2% increases relative to CK), with smaller increases under Pro-Ca (Figure 4g–i). ABA exhibited a downward trend relative to CK in all PGR treatments (Figure 4j–l), with reductions reaching 6.8–20.5% at later time points (18–24 h).

Between 6 and 12 h after spraying (HAS), CTK levels in the PGR groups were significantly higher than those in the CK group (p < 0.05). In all years examined, GA₃ in the PGR groups increased significantly at 6 HAS, although differences among treatments gradually diminished thereafter. ABA reductions in the PGR groups were mainly observed at 18–24 HAS, whereas IAA showed an upward trend but seldom reached statistical significance.

Statistical analysis showed that the Y × T interaction was highly significant (p < 0.001) for CTK content, while the interactions for IAA (p = 0.202), GA₃ (p = 0.222), and ABA (p = 0.235) were not significant. This indicates that treatment effects on CTK varied substantially between years, whereas the responses of IAA, GA₃, and ABA to treatments were relatively consistent across years.

3.5. Seed MRL and Protein, Oil Contents

As shown in

Table 2. Residual concentrations of plant growth regulators in soybean seeds (2023–2025).

Year	Treatment	Residual Concentrations (mg kg−1)
2023	CK	0.000 c
	NAA	0.009 ab
	Pro-Ca	0.012 a
	ICE-6	0.006 b
2024	CK	0.000 c
	NAA	0.004 b
	Pro-Ca	0.008 a
	ICE-6	0.002 bc
2025	CK	0.000 c
	NAA	0.013 b
	Pro-Ca	0.020 a
	ICE-6	0.009 b
Source of variance
Y		**
T		**
Y × T		*

Note: According to the Chinese national food safety standard GB 2763-2021, the maximum residue limit (MRL) for soybean seeds is 0.05 mg kg⁻¹. For a given trait, treatments with the same letter within a year were not significantly different, based on Duncan’s multiple range test at p < 0.05, using a general linear model. * and ** represent a significant difference at the 5% and 1% levels, respectively; ns represents no significant difference at the 5% level.

, residue levels of NAA, Pro-Ca, and ICE-6 in soybean seeds remained well below the maximum residue limit (0.05 mg kg⁻¹), with mean values of 0.009, 0.013, and 0.006 mg kg⁻¹, respectively.

Across the three experimental years, treatment performance remained relatively consistent, with PGR application significantly increasing soybean seed protein content while slightly reducing oil content (Figure 5). All PGR treatments produced significantly higher protein levels than CK (p < 0.05), with the largest enhancement observed under NAA (p < 0.01, in 2024 and 2025), which exceeded CK by 6.4–9.0%. Pro-Ca and ICE-6 also showed positive effects, increasing protein content by 2.7–8.3% compared to CK. Conversely, PGR application reduced seed oil content by 2.2–6.1% relative to the control, indicating a consistent trade-off between protein gain and oil yield across seasons.

Statistical analysis revealed that residual concentrations were significantly affected by the Y × T interaction (p = 0.015), while the main effects of Y and T were both highly significant (p < 0.001) (Table 2). The interaction indicated that treatment effects varied between years, with some treatments showing greater reductions in residual concentrations in 2024 than in 2023 and 2025. The Y × T interaction was not significant for both protein (p = 0.544) and oil (p = 0.350) contents in soybean seeds. This indicates that treatment effects on seed protein and oil concentrations remained relatively consistent across different years.

3.6. Correlation Analysis

Correlation analysis revealed that soybean yield was most strongly and positively associated with cytokinin content (CTK; r = 0.89), indole-3-acetic acid (IAA; r = 0.89), phosphorus content (P₂O₅; r = 0.85), seed number (SN; r = 0.86), and hundred-seed weight (100SW; r = 0.72). Moderate positive correlations were also observed with potassium content (K₂O; r = 0.68), nitrogen content (N; r = 0.60), pod number (PN; r = 0.59), gibberellic acid (GA₃; r = 0.55), and reproductive biomass allocation ratio (RBAR; r = 0.48). In contrast, abscisic acid (ABA) showed a negative correlation with yield (r = −0.29), and oil content was negatively related to yield (r = −0.53). Total biomass showed no significant correlation with yield (r = 0.056, p > 0.05). Protein content displayed a low positive correlation (r = 0.40). These results suggest that yield improvement is closely linked to higher reproductive allocation, key phytohormones (CTK and IAA), seed traits (SN, 100SW), and phosphorus and potassium nutrition (Figure 6).

4. Discussion

The findings of this study demonstrate that different plant growth regulators (PGRs) enhanced soybean yield through distinct physiological pathways, with overall trends remaining consistent across years despite certain inconsistencies among treatments.

Overall yield improvement from PGR application was substantial (Table 1). This interpretation is corroborated by previous research demonstrating similar yield reductions under episodic heat stress [26,27,28,29]. Nevertheless, yields in 2025 were lower than those in 2023 and 2024, likely due to a greater proportion of unfilled seeds resulting from short-term high-temperature episodes, with the average maximum temperature in July reaching 34.6 °C compared with 31.9 °C in the other two years (Figure 1). In contrast, the yield increase ratio under PGR treatment in 2025 exceeded that in 2024, possibly due to the higher nitrogen content during the R5 stage [30] (Figure 3A) and elevated ABA levels during the R4 stage (Figure 4j–l), which together may have enhanced the plants’ tolerance to abiotic stress [31] and promoted seed filling.

As shown in Figure 2, NAA induced the most pronounced increase in the proportion of biomass allocated to reproductive organs in 2023; however, this effect was less stable across years compared with that of Pro-Ca and ICE-6. The observed phenomenon may result from reduced ABA levels (Figure 4j–l) in the PGR treatments, potentially diminishing ABA-mediated heat tolerance [32]. In 2025, a reduction in vegetative biomass accumulation following NAA application was observed, possibly resulting from heat-stress-induced leaf abscission [33]. Another plausible explanation is that the capacity of NAA to mitigate abiotic stress is inherently less consistent, a hypothesis supported by previous reports [34,35,36].

Interannual patterns of dynamic changes in whole-plant nitrogen (N), phosphorus (P), and potassium (K) contents were generally consistent; however, the overall increase in N content in 2023 was greater than that observed in 2024 and 2025 (Figure 3A). This difference may reflect up-regulation of genes involved in nitrogen fixation and amino acid biosynthesis [37], or, under the regulation of endogenous hormone signaling [38], temperatures during the R5–R7 growth stages became more favorable for plant development, thereby enhancing rhizobial proliferation and activity, which in turn triggered a short-term surge in nitrogen accumulation [39]. While the precise mechanisms remain unclear, future investigations will aim to elucidate these processes.

Endogenous hormone levels in 2023 were also higher than those in the other two years (Figure 4), with PGR-treated plants showing more pronounced differences relative to the control (CK). This pattern may be linked to the relatively lower mean temperature in that year, whereby PGR application likely enhanced the capacity of leaves to synthesize hormones—a phenomenon consistent with findings from other studies [40]. On the other hand, this phenomenon may be attributable to the more pronounced crosstalk among CTK, IAA, GA₃, and ABA following PGR treatment, which in turn enhanced the soybean plants’ tolerance to abiotic stress [41].

Overall, the interaction between Year and Treatment (Y × T) exhibited trait-specific patterns. Residual concentrations, CTK content, N and P contents, biomass accumulation in vegetative and reproductive organs, and pod number showed significant or highly significant Y × T effects, indicating that the responses of these parameters to treatments varied substantially across years. This variability is likely associated with differences in the degree of coordination between hormones and nutrients under varying environmental conditions, as supported by previous studies [42,43].

In contrast, seed protein and oil contents, K content, biomass allocation ratios, and most yield-related traits (except pod number) exhibited no significant Y × T interaction, suggesting that treatment effects on these traits were relatively stable across years. This stability may be attributed to the strong genetic control of seed composition [44], the relatively abundant and consistent K availability in the experimental soils [45], and the inherent physiological homeostasis of biomass partitioning and yield component formation under the given management practices [46].

Conversely, traits that exhibited significant Y × T interactions tended to be more susceptible to inter-annual variation, driven primarily by short-term heat stress events (Figure 1, with the average maximum temperature in July reaching 34.6 °C). Such transient heat stress can suppress photosynthesis, accelerate leaf senescence, and alter nutrient allocation. Adequate and balanced N, P, and K fertilization plays a critical role in alleviating these adverse impacts (Figure 3, where PGR treatments markedly increased nutrient accumulation in plants): nitrogen supports photosynthetic capacity and antioxidant defense [47]; phosphorus enhances energy metabolism, root development, and membrane stability [48]; and potassium regulates osmotic balance, stomatal conductance, and enzymatic activity [49]. The synergistic effects of N, P, and K help maintain physiological homeostasis and reduce yield losses under transient heat stress [47,48,49], underscoring the importance of integrating year-specific climatic assessments—particularly the occurrence and intensity of short-term heat stress—into the evaluation of treatment efficacy.

Correlation analysis (Figure 6) indicated that the observed effects of protein and oil content on yield may not be representative, as no inherent causal relationship exists between these quality traits and yield [50]. The correlation between NPK levels and hormone contents was strong, consistent with previous findings [51,52]. Among these, the correlation between phosphorus (P) and cytokinin (CTK) was the most pronounced. Consistent with this observation, previous studies have indicated that P may regulate key genes associated with CTK signal transduction [53]. However, the nature of their interaction following PGR application in soybean remains poorly understood, and the large regional span of cultivation areas makes it difficult to determine whether such patterns are consistent across different growing regions. Addressing this knowledge gap will be a focus of future research, with the aim of elucidating the underlying physiological and molecular mechanisms.

5. Conclusions

The findings of this study demonstrate that the application of plant growth regulators (NAA, Pro-Ca, and ICE-6) modulated key intrinsic factors in soybean, including endogenous hormone levels and nitrogen, phosphorus, and potassium concentrations. This regulation enhanced biomass allocation to reproductive organs, increased yield potential, and improved seed protein content, while residue levels remained far below the safety threshold (0.05 mg kg⁻¹), indicating no human health risk.

NAA had the most pronounced effect, markedly increasing reproductive biomass allocation and nitrogen accumulation, resulting in the highest yield. Under the specific agro-climatic conditions of Xinjiang, applying NAA at 300 g ha⁻¹ significantly improved yield; however, caution is warranted when generalizing these results from this single small-scale trial.

These findings clarify the physiological mechanisms by which PGRs enhance soybean productivity and provide a basis for optimizing yield through targeted regulation of hormones and nutrients. Multi-location, multi-year trials are needed to confirm the robustness and wider applicability of these results.