Soil Nitrogen Application Regulates Photosynthetic Limitations in Plants: Insights from a Controlled Experiment

Abstract

Soil nitrogen (N) availability modulates the relative magnitudes and dominant roles of stomatal, mesophyll, and biochemical limitations to photosynthesis (denoted as l_sc, l_m, and l_b, respectively); however, the dynamic changes in these factors and their relative contributions under varying N application levels remain controversial. In this study, we quantified the responses of photosynthetic limitations in soybeans subjected to four N application levels (0, 7.5, 15.0, and 22.5 g urea m⁻²). Moderate N addition (≤15.0 g urea m⁻²) increased l_sc and l_b but decreased l_m, whereas excessive N addition (22.5 g urea m⁻²) caused a rapid increase in l_m, exceeding both l_sc and l_b. l_m was negatively correlated with mesophyll conductance (g_m) and net photosynthetic rate (A_n), whereas l_b was positively correlated with A_n. Changes in photosynthetic limitation were accompanied by increased chlorophyll and leaf N contents and increased activities of aquaporins (AQPs), carbonic anhydrase (CA), and Rubisco under moderate N addition. These results indicated that the relative contribution of l_m to the total limitation (denoted as MC_L) decreased under moderate N addition but increased sharply under high N addition, exceeding the contributions of l_sc and l_b (denoted as SC_L and B_L, respectively). Under an excessive N supply of 22.5 g urea m⁻², MC_L was the highest among the three components, although this ranking should be interpreted considering the reference-state dependence of the limitation framework. This study provides a basis for rational N fertilization in soybean production.

1. Introduction

Photosynthesis is the core physiological process of plant carbon assimilation and directly affects the growth, development, and yield of plants [1,2]. The efficiency of photosynthesis is regulated by multiple internal and external factors, among which the availability of nutrients, especially nitrogen (N), is among the most critical [3]. N is a key nutrient element for soybean growth, development, and photosynthetic processes; it directly affects Rubisco activity, stomatal development, mesophyll cell structure, and other traits, thereby regulating the partitioning of photosynthetic limitation components [4,5,6]. As a typical legume crop, soybean can establish symbioses with rhizobia to acquire atmospheric N via biological N fixation, whereas high mineral-N input typically represses nodule initiation and reduces symbiotic N-fixation capacity [7,8]. Nevertheless, the present study focuses mainly on the physiological responses of leaf photosynthetic limitations to externally supplied mineral N.

The limitations to photosynthesis in plants can be divided into three main categories: stomatal limitation (l_sc), mesophyll limitation (l_m), and biochemical limitation (l_b) [9]. l_sc results from reduced stomatal conductance (g_sc), which restricts CO₂ entry into leaves. Subsequently, l_m is governed by mesophyll conductance (g_m), which controls CO₂ diffusion from intercellular airspaces to chloroplast carboxylation sites. Ultimately, l_b results from decreased activity of photosynthesis-related enzymes (e.g., Rubisco) and impaired product synthesis, thereby constraining CO₂ fixation efficiency [10,11]. Photosynthetic limitation analysis serves as a central tool for determining the mechanism by which N regulates photosynthesis. Identifying the dominant limiting factor in the presence of different amounts of N is highly important for precise N management in soybean production. Previous studies have shown that N application can increase plant photosynthetic capacity by increasing chlorophyll content and photosynthesis-related enzyme activity [12,13,14], but the dynamic changes in the three types of photosynthetic limitations upon the application of different amounts of N and their relative contributions to photosynthesis are still controversial.

Both N deficiency and abundance can reduce the photosynthetic rate [12]. Most previous studies on N application and photosynthesis have focused on changes in the photosynthetic rate and related parameters [6,13,15], whereas quantitative analyses of photosynthetic limitations and of the mechanisms regulating these limitations remain scarce. Consequently, the reported dominant limitation under N deficiency is still debated: some studies identified l_sc as the main constraint [15], whereas others emphasized l_m or l_b [16,17]. In our latest study, Zuo et al. (2026) showed that enhanced biochemical metabolism under N addition was the dominant factor improving intraleaf CO₂ diffusion in soybean leaves [18]; however, it remains unclear whether this enhancement mainly reflects increased photosynthetic carbon assimilation after soil N addition. In addition, excessive N application can lead to nutrient imbalance [19,20], which may cause a decrease in photosynthetic efficiency, but the specific type and degree of photosynthetic limitation under excessive N still need clarification.

Accordingly, the aim of this study was to quantify the dynamic changes in the photosynthetic limitation components (l_sc, l_m, and l_b) of soybean when different amounts of N are added and identify the dominant limiting factor responsible for reduced photosynthesis under varied N conditions. We hypothesized that l_sc and l_b are the dominant constraints that lead to the increase in photosynthesis under low N conditions, that no obvious dominant limitation occurs under only moderate N conditions, and that l_m prevails only under high N conditions. The findings of this study will fill the research gap concerning the relationship between N application and photosynthetic limitations in soybean, elucidate the intrinsic mechanism through which N regulates soybean photosynthesis, and provide a theoretical basis and technical reference for the precise application of N fertilizer and for improving photosynthetic efficiency and grain yield in soybean.

2. Materials and Methods

2.1. Plant Material

Seedlings of the disease- and drought-resistant soybean (Glycine max L.) cultivar ‘Jindou 23’ (state-certified No. 2001011) were planted in individual pots at the Experimental Station of the College of Resources and Environment, Shanxi Agricultural University, Taigu, Shanxi, China (37°43′ N, 112°56′ E; 791 m a.s.l.) in late March 2021, with three replicate pots per treatment and two seedlings per pot. The pots were 12.2 L in volume, with a height of 24.0 cm and a diameter of 31.0 cm, and were filled with 7.4 L of soil. The initial soil total N content was 0.58 g kg⁻¹, and the phosphorus content was 0.88 g kg⁻¹. The baseline soil N content was determined using the Kjeldahl digestion method on air-dried soil samples passed through a 2 mm sieve prior to the addition of N and is expressed as total soil N content rather than being estimated from soil organic matter. The plant roots were blocked from the outside soil by pedestals placed under the pots. The relevant morphological parameters of the soybean seedlings when gas exchange was measured are shown in Table 1.

Table 1. Height (H) and base diameter (D) of the soybean seedlings.

Treatments	Height (H, cm)	Base Diameter (D, cm)
CK	35.3 ± 0.15 b	1.0 ± 0.15 b
LN	49.3 ± 0.12 a	1.6 ± 0.12 a
MN	48.1 ± 0.10 a	1.6 ± 0.10 a
HN	48.5 ± 0.15 a	1.8 ± 0.15 a

All the data are presented as the means ± SDs (n = 3). Different lowercase letters (a, b) in the same column indicate significant differences between N treatments at p < 0.05. CK, control; LN, low N addition; MN, medium N addition; HN, high N addition.

2.2. Experimental Design

Four N application levels, 0 (no N addition, CK), 7.5 (low N addition, LN), 15.0 (medium N addition, MN) and 22.5 g urea m⁻² (high N addition, HN), were used to simulate the ecological effects under varying intensities of atmospheric N deposition. Urea solutions with different concentrations of N were sprayed evenly onto the surface of the rhizosphere soil in the different treatment groups. The plants in the CK group were sprayed with the same volume of water. Three replicates (i.e., three potted plants) were included in each treatment group, with two seedlings retained per pot. Six fully expanded mature leaves (three leaves from each individual plant) were selected per pot for measurements of photosynthetic and physiological parameters, and the mean values were taken as the replicate values for that potted plant. All the potted slings were thoroughly watered daily to avoid water deficit.

2.3. Analysis of Photosynthesis Limitations

Gas exchange and chlorophyll fluorescence on newly and fully expanded, sun-exposed leaves were simultaneously measured using an open-flow gas exchange system (Li-6400XT; LI-COR Inc., Lincoln, NE, USA) equipped with an integrated fluorescence leaf chamber (Li-6400-40; LI-COR) from 12 to 20 July 2022. During this period, the leaves were fully light-adapted under a saturated photosynthetically active photon flux density (PPFD) of 1200 µmol m⁻² s⁻¹ provided by Li-6400 with 10:90 blue:red light for 30 min; this saturating PPFD was selected on the basis of light response curves for soybean and is consistent with that used in previous studies [5,21]. The leaf temperature, relative humidity and CO₂ concentration in the leaf chamber were maintained at 25 °C, 60 ± 5% and 400 μmol CO₂ mol⁻¹, respectively, with a CO₂ mixture. These environmental settings follow the standard configuration for photosynthetic limitation analysis [22,23]. In addition, the gas flow rate was controlled at 300 μmol s⁻¹ to ensure adequate gas exchange. After stabilization to a steady state, gas exchange parameters (i.e., net photosynthetic rate (A_n, µmol CO₂ m⁻² s⁻¹), intercellular CO₂ concentration (C_i, µmol mol⁻¹)), steady-state fluorescence (F_s) and maximum fluorescence (F_m′) were recorded.

In accordance with Grassi and Magnani (2005) and Wang et al. (2018) [22,23], the relative changes in light-saturated assimilation are sequentially expressed in terms of parallel relative changes in g_sc and g_m and in biochemical capacity (e.g., in maximum carboxylation rate, V_cmax) as follows:

$${\displaystyle \frac{\mathrm{d}{A}_{\mathrm{n}}}{A_{\mathrm{n}}}}={SC}_{\mathrm{L}}+{MC}_{\mathrm{L}}+B_{\mathrm{L}}=l_{\mathrm{s}\mathrm{c}}·{\displaystyle \frac{\mathrm{d}{g}_{\mathrm{s}\mathrm{c}}}{g_{\mathrm{s}\mathrm{c}}}}+l_{\mathrm{m}}·{\displaystyle \frac{\mathrm{d}{g}_{\mathrm{m}}}{g_{\mathrm{m}}}}+l_{\mathrm{b}}·{\displaystyle \frac{\mathrm{d}{V}_{\mathrm{c}\mathrm{m}\mathrm{a}\mathrm{x}}}{V_{\mathrm{c}\mathrm{m}\mathrm{a}\mathrm{x}}}}$$

(1)

$$l_{\mathrm{s}\mathrm{c}}={\displaystyle \frac{g_{\mathrm{t}\mathrm{o}\mathrm{t}}/{\mathrm{g}}_{\mathrm{s}\mathrm{c}}·\partial A_{\mathrm{n}}/\partial C_{\mathrm{c}}}{g_{\mathrm{t}}+\partial A_{\mathrm{n}}/\partial C_{\mathrm{c}}}}$$

$$l_{\mathrm{m}}={\displaystyle \frac{g_{tot}/g_m·\partial A_n/\partial C_c}{g_t+\partial A_n/\partial C_c}}$$

(2)

$$l_{\mathrm{b}}={\displaystyle \frac{g_{\mathrm{t}\mathrm{o}\mathrm{t}}}{g_{\mathrm{t}\mathrm{o}\mathrm{t}}+\partial A_{\mathrm{n}}/\partial C_{\mathrm{c}}}}$$

where l_sc, l_m and l_b are the limitations imposed by g_sc, g_m and biochemical capacity, respectively (0 < l_i < 1, i = sc, m, b); SC_L, MC_L and B_L are the relative contributions of g_sc and g_m and biochemical capacity to dA_n/A_n, respectively; and the sum of contributions due to g_sc and g_m is defined as the diffusional limitation (D_L = SC_L + MC_L), whereas the sum of g_m and biochemical capacity is defined as the nonstomatal limitation (NS_L = MC_L + B_L) [22]. Within the Grassi–Magnani framework, limitation coefficients (SC_L, MC_L and B_L) represent the relative contribution of each component (g_sc and g_m and V_cmax) to the relative change in A_n with respect to a defined reference state and are not absolute measures of constraint intensity. In the current study, V_cmax was replaced with J_f, which is coupled with V_cmax [23,24,25,26], and the fitted photosynthesis parameters at Ψ_pd = −0.25 MPa (CK) were used as reference values [23]. Thus,

$${\displaystyle \frac{dx}{x}}={\displaystyle \frac{x_{\mathrm{r}\mathrm{e}\mathrm{f}}-x}{x_{\mathrm{r}\mathrm{e}\mathrm{f}}}}={\displaystyle \frac{x_{0.25}-x}{x_{0.25}}}$$

(3)

where x represents the fitted g_sc, g_m or J_f and x_0.25 (e.g., x_ref) represents the x value at Ψ_pd = −0.25 MPa (CK).

Because A_n-C_i curves were not measured, V_cmax and J_max were derived from light-response gas-exchange-fluorescence data, and triose phosphate utilization (TPU) could not be estimated. Importantly, biochemical limitation (l_b) is also modulated by TPU in addition to V_cmax and J_max, which represents a methodological limitation of the current analysis.

g_sc was determined by the following Formula (4):

$$g_{\mathrm{s}\mathrm{c}}=g_{\mathrm{s}\mathrm{w}}/1.6$$

(4)

where g_sw is the stomatal conductance to water vapour (mol (H₂O) m⁻² s⁻¹), which was directly obtained from gas exchange measurements, and 1.6 is the ratio of the diffusivities of CO₂ and water in air.

g_m was calculated using the ‘variable J method’ described by Harley et al. (1992) [27]:

$$g_{\mathrm{m}}={\displaystyle \frac{A_{\mathrm{n}}}{C_{\mathrm{i}}-\frac{{\Gamma }^{*}\left(J_{\mathrm{f}}+8\left(A_{\mathrm{n}}+R_{\mathrm{d}}\right)\right)}{J_{\mathrm{f}}-4(A_{\mathrm{n}}+R_{\mathrm{d}})}}}$$

(5)

where J_f is the photosynthetic electron transport rate, Γ* represents the CO₂ compensation point in the absence of respiration (µmol mol⁻¹), and R_d represents the mitochondrial respiration in the light (µmol m⁻² s⁻¹).

In this study, J_f was calculated as follows:

$$J_{\mathrm{f}}={\Phi }_{\mathrm{P}\mathrm{S}\mathrm{I}\mathrm{I}}·PPFD·\alpha \beta$$

(6)

$${\Phi }_{\mathrm{P}\mathrm{S}\mathrm{I}\mathrm{I}}={\displaystyle \frac{\left(F_{\mathrm{m}}^{\prime }-F_{\mathrm{s}}\right)}{F_{\mathrm{m}}^{\prime }}}$$

(7)

where Φ_PSII is the actual photochemical efficiency of photosystem II, which is calculated according to Genty et al. (1989) [28] using Formula (7). α is the total leaf absorptance, and β is the partitioning of absorbed quantum between PS II and PS I. In this study, αβ was corrected as described by Valentini et al. (1995) [29], and its values for different N treatments are shown in Table 2.

Table 2. Values of αβ for different N treatments.

Treatments	CK	LN	MN	HN
αβ	0.36 ± 0.02 b	0.36 ± 0.01 ab	0.38 ± 0.01 a	0.37 ± 0.00 ab

All the data are presented as the means ± SDs (n = 3). Different lowercase letters (a, b) in the same column indicate significant differences at p < 0.05. CK, control; LN, low N addition; MN, medium N addition; HN, high N addition.

Γ* and R_d were measured using the Laisk method [30]. In theory, they intersect with each other at a point, and the intersection points at the x- and y-axes are considered to indicate the intercellular CO₂ partial pressure (C_i*) and R_d, respectively. In practice, three CO₂ response curves under low light and low CO₂ concentrations obtained by varying CO₂ concentrations from 150 to 40 µmol CO₂ mol⁻¹ under three PPFDs (150, 100 and 50 µmol m⁻² s⁻¹) formed a triangular range rather than a single point [31], and C_i^* and R_d were calculated as the barycentre of the triangle formed by the intersection of the three lines on the x- and y-axes, respectively, in this study. The value of Γ* was calculated using the following Formula (8) [32]:

$${\Gamma }^{*}=C_{\mathrm{i}}^{*}+{\displaystyle \frac{R_{\mathrm{d}}}{g_{\mathrm{m}}}}$$

(8)

g_tot was the total conductance to CO₂ and was calculated using Formula (9):

$$g_{\mathrm{t}\mathrm{o}\mathrm{t}}={\displaystyle \frac{1}{\frac{1}{g_{\mathrm{s}\mathrm{c}}}+\frac{1}{g_{\mathrm{m}}}}}$$

(9)

Notably, the values for biochemical and anatomical traits, including aquaporin (AQP) activity, carbonic anhydrase (CA) activity, Rubisco activity, cell wall thickness (T_cw), surface area of mesophyll cells exposed to intercellular space per unit of leaf area (S_m), chloroplast surface area facing the intercellular space per unit of leaf area (S_c), liquid-phase conductance to CO₂ (g_liq), and J_f, used in this study were obtained from our companion study in Zuo et al. (2026) [18], which was conducted using the same experimental system and plant material. The present study focused on the quantitative analysis of photosynthetic limitations, whereas the study by Zuo et al. (2026) [18] focused on internal leaf CO₂ diffusion and biochemical metabolism. Together, these two studies provide a complementary understanding of how soil N supply regulates soybean photosynthesis.

Notably, the limitation coefficients SC_L, MC_L and B_L in the Grassi–Magnani framework are defined relative to a chosen reference state. In this study, the CK treatment (0 g urea m⁻², Ψ_pd = −0.25 MPa) was used as the reference. Consequently, all the limitation values are relative to this reference and should not be interpreted as absolute constraint intensities. A reference-dependence sensitivity analysis was performed to assess the robustness of the inferred limitation rankings (see Section 3.4).

2.4. Determination of Chlorophyll Content

The chlorophyll content was determined using the Arnon method [33]. Fresh leaf fragments (0.1 g) with the veins removed were weighed and placed in a mortar. A small amount of quartz sand and calcium carbonate, along with 2 mL of 80% acetone, was added, and the sample was ground thoroughly into a homogenate, which was then transferred to a 25 mL brown volumetric flask. The mortar, pestle, and residue were rinsed 2–3 times with a small volume of 80% acetone, the rinses were combined, and the volume was made up to 25 mL with 80% acetone and mixed well. The samples were extracted in the dark for 12 h, followed by centrifugation at 4 °C and 8000 r min⁻¹ for 10 min, after which the supernatant was collected. Using 80% acetone as the blank, the absorbance was measured at 663 nm and 645 nm using UV-Vis/Vis Spectrophotometers (Ther-moFisher Scientific, Waltham, MA, USA). The chlorophyll content was calculated using the following formulas and is expressed as mg g⁻¹ FW: Chl a = 12.7A₆₆₃ − 2.69A₆₄₅; Chl b = 22.9A₆₄₅ − 4.68A₆₆₃; and Chl total = 8.02A₆₆₃ + 20.21A₆₄₅.

2.5. Measurement of Leaf N Content

Fully expanded and healthy functional leaves were randomly selected during the gas exchange measurement period and immediately cleaned with deionized water to remove surface dust and soil residues. The samples were subsequently oven-dried at 75 °C for 24 h until a constant weight was achieved to eliminate residual moisture and then thoroughly ground into a fine and homogeneous powder using an oscillating mixer mill (MM400; Retsch, Düsseldorf, Germany). Approximately 5.0 mg of each leaf sample was accurately weighed and loaded into a C–N elemental analyser (Elementar Vario MACRO, Element, Langenselbold, Germany) to determine the leaf N content (LNC). The obtained N content was calculated and is expressed on a dry weight basis, and replicate measurements (n = 3) were performed to guarantee the accuracy and reproducibility of the experimental data.

2.6. Statistical Analysis

Normality and homogeneity of variance were confirmed prior to parametric analysis. One-way ANOVA was performed using IBM SPSS Statistics 29.0 (SPSS, Inc., Chicago, IL, USA) to determine differences in l_sc, l_m and l_b among the different N treatment groups, and means were compared using the least significant difference (LSD) multiple comparison test. In addition, regression analysis was used to describe the response of each limitation component to the gradient of added N, with the N dose treated as a continuous variable; both linear and quadratic models were fitted, and the best-fitting model was selected on the basis of R² and the significance level. Pearson correlation analysis and linear regression analysis were subsequently performed to examine the relationships between photosynthetic limitation components (l_sc, l_m and l_b) and physiological traits (g_m and A_n), as well as between A_n and the limitation components. Correlation coefficients (r), regression parameters and their significance levels were calculated. A sensitivity analysis was subsequently conducted to evaluate the effects of the limitation components on the photosynthetic carbon assimilation capacity, which was assessed as A_n and V_cmax. Differences were considered statistically significant at p < 0.05. All figures were generated using OriginPro 2021 (OriginLab Corporation, Northampton, MA, USA).

3. Results

3.1. Quantitative Changes in Photosynthetic Limitation Components

The responses of the three photosynthetic limitation indices (l_sc, l_m, and l_b) to the gradient of added N are shown in Figure 1. One-way ANOVA revealed significant differences among the N treatment groups in terms of l_sc (F = 11.261; p = 0.003), l_m (F = 3.652; p = 0.063), and l_b (F = 0.104; p = 0.955). When the N dose was treated as a continuous variable, quadratic models provided the best fit for all three components, although none reached statistical significance (l_sc: R² = 0.925, p = 0.311; l_m: R² = 0.908, p = 0.326; l_b: R² = 0.993, p = 0.108; Figure 1), likely owing to the limited number of N levels (n = 4) and the resulting low statistical power. Despite the nonsignificant regressions, l_sc increased from the CK group to the MN group and then decreased in the HN group; l_m exhibited a U-shaped trend, reaching a minimum at 10.9 g urea m⁻² and showing a sharp increase at 22.5 g urea m⁻²; and l_b gradually increased. At the highest N dose, l_m exceeded both l_sc and l_b, although this difference was not statistically significant for all pairwise comparisons, indicating a shift in the dominant limitation from mesophyll to stomatal and biochemical constraints as the N supply increased.

Figure 1. Response of photosynthetic limitation components to soil N application levels. l_sc, stomatal limitation; l_m, mesophyll limitation; l_b, biochemical limitation. The values are the means ± SDs (n = 3). Different lowercase letters (a, b, and c) denote significant differences among the N addition treatment groups at p < 0.05. Regression curves were fitted against the N application level as a continuous variable: l_sc, quadratic model, R² = 0.925, p = 0.311; l_m, quadratic model, R² = 0.908, p = 0.326; l_b, quadratic model, R² = 0.993, p = 0.108. CK, control; LN, low N addition; MN, medium N addition; HN, high N addition.

3.2. Quantitative Changes in the Relative Contribution of Photosynthetic Limitation Components to Photosynthetic Carbon Assimilation

We quantified the relative contributions of photosynthetic limitation components (SC_L, MC_L and B_L) to photosynthetic carbon assimilation with increasing N application levels (Figure 2). Overall, the three indices responded differently to increasing N supply: SC_L exhibited a biphasic response, MC_L remained the largest single component across most of the gradient, and B_L increased gradually but stayed comparatively small. Specifically, SC_L was negligible at low N levels (0%) and peaked at approximately 22% under excess N. MC_L ranged from 34% to 39% relative to the CK treatment across the N gradient, with g_m accounting for 60–70% of the total limitation. B_L increased gradually from 1% to 9% with increasing N availability. Consequently, the total photosynthetic limitation nearly doubled from 35% (control) to 70% at the highest N level. Our findings indicate that g_m was the core bottleneck of photosynthesis under N addition, whereas excess N simultaneously strengthened stomatal and biochemical constraints, providing a mechanistic basis for optimizing N management to improve plant photosynthetic efficiency.

Figure 2. Relative contribution of photosynthetic limitation components to photosynthetic carbon assimilation.

3.3. Correlations Between Limitation Components and Physiological Traits

We first examined the relationships between g_m and the three limitation components, and then assessed how each component was related to A_n (Figure 3). As shown in Figure 3I, g_m was significantly positively correlated with l_sc (r = 0.602, p = 0.039) and l_b (r = 0.584, p = 0.046) but extremely significantly negatively correlated with l_m (r = −0.880, p < 0.001). l_m acted as the dominant negative regulator that directly constrained g_m, whereas l_sc and l_b exerted positive effects on g_m, probably via a compensatory mechanism in which g_m was increased to alleviate stomatal/cuticular and biochemical constraints on photosynthesis. Thus, g_m appeared to be co-regulated by these three limitation components, with l_m as the decisive factor and l_sc and l_b as compensatory positive drivers.

Figure 3. Correlations between g_m (I) and A_n (II) and limitation components (l_sc, l_m and l_b). l_sc, stomatal limitation, l_m, mesophyll limitation, l_b, biochemical limitation. The fitting lines are shown at the level of p < 0.05. I-a, b, c are correlations between g_m and l_sc, l_m and l_b, respectively; II-a, b, c are correlations between A_n and l_sc, l_m and l_b, respectively.

As illustrated in Figure 3II, linear regression analyses revealed divergent associations between A_n and the three photosynthetic limitation components. No statistically significant linear correlation was observed between A_n and l_sc (p > 0.05), indicating that g_sc did not have a meaningful linear effect on A_n across the soil N supply range of 0–22.5 g urea m⁻². Conversely, a strong and statistically significant negative linear correlation was observed between A_n and l_m (r = −0.776, p = 0.008), demonstrating that elevated mesophyll limitation was associated with a pronounced reduction in photosynthetic CO₂ assimilation, with l_m explaining a large proportion of the variance in A_n. Furthermore, a moderate and statistically significant positive linear correlation was identified between A_n and l_b (r = 0.686, p = 0.020), indicating that a higher biochemical limitation was linked to increased A_n, likely reflecting compensatory physiological adjustment. Collectively, these findings indicated that l_m acted as the primary negative linear predictor of A_n, l_b functioned as a significant positive linear predictor, and l_sc had no significant linear predictive power for the net photosynthetic rate, a pattern consistent with the regulatory dynamics of g_m described above.

3.4. Reference-Dependent Sensitivity Analysis of the Relative Contribution of Photosynthetic Limitation Components to Photosynthetic Carbon Assimilation

The Grassi–Magnani limitation framework is reference-state dependent. To test the robustness of the limitation-component rankings, the LN and MN treatments were separately adopted as alternative reference states to recalculate SC_L, MC_L and B_L (Table 3).

Table 3. Relative contributions of photosynthetic limitation components to photosynthetic carbon assimilation calculated using the LN and MN treatments as alternative reference states.

Treatment	Relative Contribution (%) (LN as Alternative Reference)			Relative Contribution (%) (MN as Alternative Reference)
	SCL	MCL	BL	SCL	MCL	BL
CK	5.92 ± 1.15 c	12.01 ± 2.43 b	4.61 ± 1.30 b	3.19 ± 0.56 c	19.29 ± 2.09 a	6.29 ± 0.75 a
LN	0 d	0 c	0 c	8.81 ± 1.08 b	9.01 ± 0.12 b	5.40 ± 0.88 a
MN	10.34 ± 1.28 b	11.21 ± 1.69 b	5.94 ± 1.46 b	0 d	0 c	0 b
HN	25.58 ± 2.90 a	21.90 ± 2.95 a	11.55 ± 1.47 a	22.52 ± 2.27 a	19.18 ± 2.13 a	5.73 ± 0.87 a

All the data are presented as the means ± SDs (n = 3). Different lowercase letters (a, b, c and d) in the same column indicate significant differences at p < 0.05. SC_L, MC_L and B_L are the relative contributions of g_sc and g_m and biochemical capacity to dA_n/A_n, respectively.

When the LN treatment was used as the reference, MC_L (12.01 ± 2.43%) > SC_L (5.92 ± 1.15%) > B_L (4.61 ± 1.30%) for the CK treatment, whereas SC_L (25.58 ± 2.90%) > MC_L (21.90 ± 2.95%) > B_L (11.55 ± 1.47%) for the HN treatment. When the MN treatment was used as the reference, MC_L was still the largest component in the CK group (19.29 ± 2.09%), and SC_L (22.52 ± 2.27%) and MC_L (19.18 ± 2.13%) were comparable in the HN group, with BL remaining the minor contributor (5.73 ± 0.87%). The HN treatment consistently produced the highest values for all three contribution indices across reference states.

Comparison with the original results in which the CK treatment was used as the reference (Section 3.2, where MC_L dominated the total limitation) showed that switching reference states altered the numerical order of SC_L and MC_L, especially for the HN treatment. Nevertheless, the sum of the diffusional limitations (SC_L + MC_L) always constituted the major fraction of the total photosynthetic limitation, whereas B_L remained the smallest component regardless of reference selection. These results confirm that individual limitation rankings are reference-dependent rather than absolute; however, the qualitative conclusion that diffusional limitations dominate photosynthetic constraints under varying soil N levels remains robust.

3.5. Sensitivity Analysis of the Limiting Components of Photosynthetic Carbon Assimilation

On the basis of the linear regression analysis between the photosynthetic limitation components (l_sc, l_m and l_b) and photosynthetic carbon assimilation (A_n and V_cmax) presented in Table 4, the VIF for all the photosynthetic limitation factors was greater than 1.000, indicating the absence of multicollinearity among the independent variables. For A_n (left panel), l_b had the largest standardized coefficient (beta = 0.568), followed by l_sc, with a moderate positive effect (beta = 0.261, p = 0.357), whereas l_m exhibited a negative standardized coefficient (beta = −0.515) that was not statistically significant (p = 0.087). The overall model F-statistics for l_sc and l_b (F = 2.519, df = 11) and for l_m (F = 3.603, df = 11) further suggested limited explanatory power. For V_cmax (right panel), all three factors showed weak and nonsignificant effects, with l_b having a weak negative influence (beta = −0.279, p = 0.406), whereas l_sc (beta = 0.075, p = 0.820) and l_m (beta = 0.097, p = 0.765) had negligible positive effects. Similarly, the model F-values were also low (F = 0.435 for l_sc and l_b; F = 0.095 for l_m), confirming that the relationships were not significant. These results suggested that all three limitation components might contribute to variation in A_n, with l_b and l_m showing marginal trends (p ≈ 0.06 and 0.09, respectively) and l_m exerting the strongest negative effect, whereas the effect of l_sc was relatively weak. In other words, compared with l_sc, l_b (biochemical capacity) and l_m (g_m) had relatively stronger influences on A_n. Nevertheless, none of the coefficients reached the conventional significance threshold (p < 0.05), and the limited sample size (n = 3) meant that these results should be regarded as suggestive rather than conclusive. To further evaluate the combined effects of the three limitation components on A_n, a multiple regression analysis was performed; its results were consistent with the simple regression and sensitivity analyses, again indicating that l_m had the strongest negative association with A_n (marginally significant), whereas l_b and l_sc had weaker positive effects, although none of the coefficients reached statistical significance at p < 0.05, probably owing to the limited sample size.

Table 4. Linear regression analysis of the effects of photosynthetic limiting factors on photosynthetic carbon assimilation capacity.

Parameter	An (μmol CO2 m−2 s−1)					Vcmax (μmol CO2 m−2 s−1)
	Beta	VIF	p	F-Value	df	Beta	VIF	p	F-Value	df
lsc (%)	0.261	1.012	0.357	2.519	11	0.075	1.012	0.820	0.435	11
lm (%)	−0.515	1.000	0.087	3.603	11	0.097	1.000	0.765	0.095	11
lb (%)	0.568	1.012	0.063	2.519	11	−0.279	1.012	0.406	0.435	11

Beta is the standardized coefficient, VIF is the variance inflation factor, p is the significance coefficient, p < 0.05 indicates statistical significance, and df is the degree of freedom.

3.6. Effects of Soil N Application on Chlorophyll Content

N application significantly altered the leaf chlorophyll content, which followed a parabolic trend—initially increasing but then decreasing—with increasing N levels (Table 5). Compared with the CK treatment, all the N treatments significantly increased the chlorophyll a (Chl a), chlorophyll b (Chl b), and total chlorophyll (total Chl) contents (p < 0.05). Compared with the CK treatment, the MN treatment resulted in the highest chlorophyll contents, with the contents of Chl a, Chl b, and total chlorophyll increasing by 59.4%, 90.0%, and 66.7%, respectively. The chlorophyll levels in the LN group were intermediate, whereas those in the HN group slightly decreased, indicating that an excessive N supply does not further promote chlorophyll accumulation.

Table 5. Effects of different N treatments on the chlorophyll and leaf N contents of soybean plants.

Treatments	Chl a (mg g−1 FW)	Chl b (mg g−1 FW)	Total Chl (mg g−1 FW)	LNC (g·kg−1)
CK	3.2 ± 0.2 b	1.0 ± 0.1 c	4.2 ± 0.3 c	26.2 ± 2.1 c
LN	4.5 ± 0.2 a	1.6 ± 0.2 b	6.1 ± 0.4 b	35.7 ± 1.8 b
MN	5.1 ± 0.5 a	1.9 ± 0.1 a	7.0 ± 0.3 a	44.9 ± 3.2 a
HN	4.6 ± 0.4 a	1.7 ± 0.1 ab	6.3 ± 0.5 ab	37.1 ± 2.3 b

CK, control; LN, low N addition; MN, medium N addition; HN, high N addition; Chl a, chlorophyll a content; Chl b, chlorophyll b content; total Chl, total chlorophyll content; LNC, leaf N content. Different lowercase letters (a, b and c) in the same column indicate significant differences at p < 0.05.

3.7. Effects of Soil N Application on Leaf N Content

The LNC was significantly affected by soil N application (Table 5). Compared with the control, the LNC increased by 36.3% in the low N application group (7.5 g urea m⁻²) and by 71.4% in the medium N application group (15.0 g urea m⁻²), reaching a maximum value of 44.9 ± 3.2 g kg⁻¹ in the MN group. However, in the high N application group (22.5 g urea m⁻²), the LNC decreased to 37.1 ± 2.3 g kg⁻¹, representing a 17.4% reduction relative to that in the MN group, but remained 41.6% higher than that in the CK group. Multiple comparisons revealed that the LNC in all N-treated groups was significantly greater than that in the CK group (p < 0.05). Notably, no significant difference was detected between the LN and HN groups (p > 0.05), whereas the LNC was greatest in the MN group, significantly differing from that in all the other treatment groups (p < 0.05).

3.8. Biochemical and Anatomical Traits Underlying Photosynthetic Limitations

To provide mechanistic insight into the shifts in photosynthetic limitation components, key biochemical and anatomical traits, including AQP activity; CA activity; Rubisco activity; and T_cw, S_m, S_c, g_liq, and J_f, obtained from our companion study (Zuo et al., 2026) [18] were summarized in Table 6. Compared with the CK treatment, moderate N addition (≤15.0 g urea m⁻²) increased CA activity and Rubisco activity, with the highest values observed in the MN group (CA: 223.8 U L⁻¹; Rubisco: 836.8 U L⁻¹). However, in the HN group, Rubisco activity decreased to 807.4 U L⁻¹, which was significantly lower than that in the MN group ($p<0.05$), whereas CA activity continued to increase to 238.3 U L⁻¹. AQP activity differed among groups, being lowest in the LN group (604.0 U L⁻¹) and highest in the HN group (678.3 U L⁻¹). With respect to leaf anatomical traits, T_cw decreased from the CK group (0.36 μm) to the LN group (0.25 μm) but then increased in the HN group (0.35 μm), whereas S_m increased from the CK group (6.52 μm²) to the MN group (16.87 μm²) but decreased in the HN group (13.66 μm²). S_c did not significantly differ among the treatment groups. g_liq was highest in the MN group (4.27 × 10⁻³ mol CO₂ m⁻² s⁻¹), which is consistent with the lowest l_m observed in the MN group.

Table 6. Values of key biochemical and anatomical traits in the different soil N addition treatment groups. All the data are presented as the mean ± SDs (n = 3). Different lowercase letters (a, b, c and d) indicate significant differences at p < 0.05. AQPs, aquaporins; CA, carbonic anhydrase; Rubisco, ribulose-1,5-bishosphate carboxylase; T_cw, cell wall thickness; S_m, surface area of mesophyll cells exposed to intercellular space per unit leaf area; S_c, chloroplast surface area facing the intercellular space per unit leaf area; J_f, photosynthetic electron transport rate; g_liq, liquid-phase conductance to CO₂. All the data presented in the table were obtained from our previous study (Zuo et al. (2026)) [18].

Treatments	AQPs Activity (U L−1)	CA Activity (U L−1)	Rubisco Activity (U L−1)	Tcw (μm)	Sm (μm2)	Sc (μm2)	Jf (μmol e− m−2 s−1)	gliq (10−3 mol CO2 m−2 s−1)
CK	643.7 ± 27.8 a	198.8 ± 3.4 d	690.5 ± 4.5 c	0.36 ± 0.04 a	6.52 ± 0.24 c	3.69 ± 0.58 a	131.5 ± 13.3 a	3.10 ± 0.50 a
LN	604.0 ± 13.0 b	214.7 ± 3.3 c	807.1 ± 3.4 b	0.25 ± 0.03 b	13.48 ± 2.47 b	3.74 ± 0.91 a	132.0 ± 15.9 a	3.13 ± 0.80 a
MN	644.7 ± 44.5 a	223.8 ± 4.5 b	836.8 ± 15.5 a	0.28 ± 0.04 ab	16.87 ± 0.08 a	2.71 ± 0.50 a	153.2 ± 12.2 a	4.27 ± 0.84 a
HN	678.3 ± 8.0 a	238.3 ± 2.6 a	807.4 ± 1.2 b	0.35 ± 0.09 a	13.66 ± 0.55 b	3.28 ± 0.61 a	157.0 ± 37.0 a	3.54 ± 0.71 a

These patterns were consistent with the changes in l_m: the lowest l_m in the MN group coincided with the highest CA activity, Rubisco activity, S_m, and g_liq, whereas the sharp increase in l_m in the HN group coincided with decreased Rubisco activity and S_m and increased T_cw. These results suggest that N-induced changes in biochemical capacity and leaf anatomy underlie the observed shifts in mesophyll limitation.

4. Discussion

Photosynthetic limitations are key determinants of plant photosynthetic efficiency, and their nature and magnitude are modulated by N supply levels [21,34]. Upon entering a plant, N is preferentially allocated to photosynthesis-related enzymes and electron transport components, thereby increasing the capacity for photosynthetic carbon assimilation [35,36]. The results of the present study revealed a nonlinear response of photosynthesis to N addition. Quantitative limitation analysis revealed that moderate N application (≤15.0 g urea m⁻²) progressively increased l_sc and l_b but significantly decreased l_m (see Figure 1). Consequently, g_sc and biochemical metabolic capacity (V_cmax, which increased with increasing N supply, as shown in our previous study by Zuo et al. (2026) [18]), emerged as the dominant drivers of photosynthetic changes. These improvements were primarily attributed to increased activities of CA and Rubisco, as well as to favourable modifications in leaf anatomical traits—including reduced T_cw, increased surface areas of mesophyll cells and chloroplasts exposed to intercellular airspaces per unit of leaf area (S_m and S_c), and elevated liquid-phase diffusional conductance (g_liq)—as detailed in our earlier study by Zuo et al. (2026) [18] (see Table 6). In contrast, a high N supply redirected leaf N partitioning from the photosynthetic apparatus towards storage and structural pools, thereby increasing l_m and suppressing photosynthesis [37]. Additionally, high N application reduced photosynthetic capacity by decreasing Rubisco activity and diverting leaf N into nonphotosynthetic storage pools [38]. Consistent with these reports, our study revealed that at 22.5 g urea m⁻² (high N, HN), l_m increased rapidly and significantly surpassed both l_sc and l_b (see Figure 1), a pattern differing from those reported by some previous studies suggesting that excessive N primarily increases l_sc [15,39]. Instead, our results indicated that excessive N increased both l_m and l_b. The concurrent decline in the LNC in the HN group (see Table 5) likely reduced substrate availability for CA and Rubisco synthesis, directly reducing g_m and carboxylation efficiency. Several factors likely contributed to this outcome: (i) nitrate accumulation in leaves damaged chloroplast structure, reducing g_m; (ii) the synthesis and activity of photosynthesis-related enzymes (CA and Rubisco) were inhibited (see Table 6), lowering CO₂ fixation efficiency; and (iii) an imbalanced leaf C/N ratio might have disrupted the transport and accumulation of photosynthetic products, further intensifying l_b. In the HN group in the present study, g_m gradually surpassed g_sc and biochemical metabolic capacity as the dominant factors limiting photosynthesis (see Figure 2). This shift was attributed mainly to high-N-induced alterations in mesophyll cell structure and Rubisco activity. The increased T_cw and decreased S_m under high N application (see Table 6) not only lengthened the CO₂ diffusion pathway [40] but also reduced the contact area between CO₂ and chloroplastic carboxylation sites [23], directly decreasing the amount of CO₂ available for carbon assimilation. Concurrently, the marked decline in Rubisco activity under high N application (see Table 6) substantially decreased carboxylation efficiency.

Appropriate N application (≤15.0 g urea m⁻²) increased the chlorophyll and leaf N contents (see Table 4 and Table 5) and subsequently increased Rubisco activity, decreased l_m, and consequently improved photosynthetic efficiency. A clear parabolic response of the LNC to increasing N supply was observed; the maximum LNC was observed in the MN group (44.9 g kg⁻¹), followed by a subsequent decline in the HN group (37.1 g kg⁻¹) (see Table 5). This LNC trajectory closely mirrored the opposite trend in l_m, which decreased progressively from the CK group to the MN group but then increased sharply in the HN group (see Figure 1), suggesting a mechanistic linkage between leaf N status and the efficiency of CO₂ diffusion from the substomatal cavity to chloroplast carboxylation sites.

The LNC-dependent regulation of l_m can be attributed to at least three non-mutually exclusive mechanisms. First, increased LNC under moderate N supply promoted the synthesis of CA and AQPs (see Table 6), both of which are N-rich proteins that facilitate CO₂ hydration and membrane transport, thereby increasing g_m (data available in our previous study by Zuo et al. (2026) [18]). Second, adequate leaf N availability supported the development of mesophyll anatomical traits—including by increasing S_m and reducing T_cw—which collectively lowered the physical resistance to CO₂ diffusion (see Table 6). Third, the decline in the LNC under excessive N application (HN) likely reflected a shift in N partitioning from the photosynthetic apparatus to nonphotosynthetic storage pools, leading to reduced g_m and consequently elevated l_m (see Figure 1). This interpretation is consistent with the strong negative correlation observed between g_m and l_m (r = −0.880, p < 0.001; Figure 3I) and the significant negative linear relationship between A_n and l_m (r = −0.776, p = 0.008; Figure 3II), reinforcing the view that l_m serves as the primary negative predictor of photosynthetic carbon assimilation under varying N supply. In contrast to its dominant effect on l_m, the LNC showed a weaker direct relationship with l_sc. Although moderate N addition increased both the LNC and l_sc (Figure 1), the contribution of g_sc to overall photosynthetic restriction remained relatively modest (SC_L ≤ 22%; Figure 2). These observations align with the finding that l_sc did not significantly linearly correlate with A_n (p > 0.05; Figure 3II), implying that g_sc was not a major limiting factor within the N gradient examined. Rather, the N-induced increase in l_sc might represent an incidental consequence of improved leaf structure and increased guard cell turgor [41,42] rather than a direct cause of photosynthetic enhancement. This interpretation is supported by the relatively low standardized coefficient of l_sc on A_n (beta = 0.261, p = 0.357; Table 4), indicating that stomatal factors played a subordinate role relative to mesophyll and biochemical constraints. Unlike l_m and l_sc, l_b increased with N addition across the entire gradient (see Figure 1), although the magnitude of this increase was relatively small (B_L from 1% to 9%; Figure 2), and l_b exhibited a significant positive correlation with A_n (r = 0.686, p = 0.020; Figure 3II), suggesting that greater biochemical limitation was paradoxically associated with increased net photosynthesis—a counterintuitive finding that likely reflects a compensatory mechanism. The positive correlation between A_n and l_b should be interpreted with caution because l_b is a relative limitation index that is mathematically coupled with l_sc and l_m. Under moderate N supply, the absolute biochemical capacity (Rubisco activity and electron transport rate) and g_m both increased, whereas the relative contribution of l_b to total limitation increased. This pattern reflects a coordinated adjustment among limitation components rather than evidence that A_n is unlimited by biochemical capacity. Under sufficient N conditions (≤15.0 g urea m⁻²), the increase in Rubisco activity and J_f (see Table 6) outweighs the concurrent increase in l_b, resulting in a net increase in photosynthetic carbon assimilation [43,44]. This interpretation is consistent with the sensitivity analysis showing that l_b had the highest positive standardized coefficient for A_n (beta = 0.568, p = 0.063; Table 4), although this effect was only marginally significant. However, the marginal significance (p ≈ 0.06) suggested that the positive effect of l_b might be context dependent, varying with N availability and the prevailing physiological status of the leaf. Importantly, the decrease in the LNC in the HN group did not translate to a reduction in l_b; instead, l_b continued to increase (see Figure 1). These results suggest that in the presence of excessive N, the biochemical machinery might be subjected to feedback inhibition or downregulation at the posttranslational level, even when the leaf N content remains relatively high (37.1 g kg⁻¹). Such a decoupling between the LNC and l_b in the presence of high amounts of N highlights the limitation of using total LNC as a sole proxy for photosynthetic capacity and the importance of considering N allocation among photosynthetic components.

Specifically, correlation analyses between the limitation components and physiological traits (see Figure 3), together with sensitivity analysis of photosynthetic carbon assimilation (Table 1), revealed that l_sc and l_b exerted positive regulatory effects on net photosynthetic capacity, with l_b having the strongest positive effect and l_m having the most pronounced negative effect. Hence, the influence of photosynthetic limitation components on carbon assimilation capacity is environment dependent [45,46], with l_b (i.e., biochemical metabolic capacity) being the most sensitive variable, whose effect direction varies under different model conditions.

Nevertheless, despite the clear mechanistic insights obtained in this study, several caveats should also be noted. This study was conducted under greenhouse conditions with relatively stable temperature, light, and humidity, which may differ from field environments. In addition, we did not quantify nodule-related traits in the current experiment. Since mineral nitrogen fertilization may alter nodule formation and symbiotic nitrogen-fixation activity, future investigations should include quantitative measurements of nodule development to disentangle the interactive effects between exogenous mineral N supply and biological N-fixation in soybean. Moreover, the experiments were performed solely on soybean plants, and the regulatory effects of N application on photosynthetic limitations may vary across plant species. Third, the limitation analysis was inherently reference-dependent. The choice of CK as the reference state influenced the absolute values and potentially the ranking of limitation components. Although we conducted a reference-dependence sensitivity analysis (Section 3.4), the limited number of N levels (n = 4) and replicates (n = 3) constrained the statistical power of our conclusions. Future studies should include more N levels and larger sample sizes to validate the robustness of the inferred limitation rankings. Therefore, future studies should be carried out in the field using a broader range of species to validate the present findings and to provide a more comprehensive theoretical basis for optimizing soil N management strategies.

5. Conclusions

In this study, moderate soil N addition (≤15.0 g urea m⁻²) increased l_sc and l_b but markedly decreased l_m, whereas excessive N addition (22.5 g urea m⁻²) led to a rapid increase in l_m, which exceeded both l_sc and l_b. The LNC showed a parabolic response to N supply, with a maximum at 15.0 g urea m⁻² and a decline at 22.5 g urea m⁻², and this pattern was associated with the observed shifts in the photosynthetic limitation components. The changes in net photosynthetic capacity were accompanied by altered activities of CA and Rubisco and by changes in leaf structure. These findings indicate that optimizing soil N addition could reduce l_m and that excessive N supply is associated with the greatest relative contribution of l_m, but this conclusion requires validation by studies with larger sample sizes and alternative reference states. Further field studies involving multiple species are needed to validate these conclusions and to support rational N fertilization for sustainable agriculture.