Leaf Color Chart-Based Nitrogen Management Affects Rice Enzyme Activities and Maintains Soil Nitrogen Balance

Abstract

Real-time nitrogen (N) management based on the leaf color chart (LCC) is considered a potential alternative to traditional farmer practices. However, its physiological mechanisms for enhancing rice N utilization and its effects on paddy field N balance remain unclear. We aimed to elucidate the potential enzymatic mechanisms underlying LCC’s influence on rice N use and quantify the impact of LCC on paddy field N balance. In 2022 and 2023, a single-factor randomized block design experiment was conducted during the rice planting season. Four N treatments: no N (ONF), farmers’ conventional practices + urea [FNR] as the control, LCC + urea [SSNM1], LCC + controlled-release urea [SSNM2] were administered. Rice yield and N uptake were positive correlations with nitrate reductase (NR), glutamine synthetase (GS), glutamate-pyruvate transaminase (GPT), glutamate-oxaloacetate transaminase (GOT) and glutamate dehydrogenase (GDH) activities, which were higher under SSNM1 and SSNM2 compared with FNR, but were negative correlation with proteinase activity. Moreover, SSNM1 and SSNM2 increased rice yield by 9.2% and 9.4%, N uptake by 15.4% and 15.3%, and N use efficiency by 46.9% and 65.0%, and reduced reactive N losses by 46.2% and 66.7%, respectively. The annual net soil N inputs under FNR, SSNM1, and SSNM2 were 12.6, 8.9, and 4.2 kg N ha⁻¹, respectively. LCC-based N management increased N uptake and rice yield by enhancing the activities of NR, GS, GPT, GOT, and GDH while reducing protease activity. Moreover, LCC maintained soil N supply capacity even with reduced nitrogen fertilizer application.

1. Introduction

Rice is a vital food crop and a major consumer of nitrogen (N) fertilizers [1]. According to the fertilizer use by crops report, N fertilizer application in rice fields accounted for approximately 16% of the total N input in farmland in 2018 [2]. In particular, farmers in China tend to overuse N fertilizer as an “insurance” against yield losses, primarily due to the lack of access to advanced N management information and technology [3,4,5]. This practice has resulted in substantial reactive nitrogen (RN) losses in gaseous forms (ammonia [NH₃], nitrous oxide [N₂O], nitric oxide [NO]), leaching, and runoff, thereby reducing N use efficiency [6,7,8]. Moreover, the imbalance between N inputs (fertilizer application, atmospheric deposition, irrigation) and outputs (crop uptake and RN losses) in rice fields disrupts the N cycle, further exacerbating environmental issues such as water eutrophication and soil acidification [9]. Therefore, adopting more efficient N management practices to reduce RN losses, enhance N uptake by rice, and restore the N balance in rice fields is one of the global priorities for achieving sustainable development [10].

The Leaf Color Chart (LCC), initially developed by the International Rice Research Institute (IRRI), was designed to provide farmers with a simple, cost-effective, and practical tool to dynamically monitor N nutrition in rice, optimizing both the timing and amount of N fertilizer application [11,12,13]. In recent years, the LCC has been increasingly recognized as a promising technology for N management in rice and has been implemented in various regions [14]. Using LCC-based N management in rice cultivation not only reduces N fertilizer application but also maintains or even increases rice yields. Previous reports suggest that depending on threshold settings, fertilizer type, and environmental factors, LCC can reduce N fertilizer use by 2–27%, improve N use efficiency by 11–45%, and increase grain yield by 4–28% [13,15,16]. These beneficial effects are often simplistically attributed to better synchronization of N supply with crop demand [17], with limited exploration of the underlying physiological mechanisms, such as changes in enzyme activities related to N metabolism. Furthermore, LCC-based N management has been shown to reduce RN losses in rice fields [11,18,19,20]. However, while some environmental impacts of LCC have been preliminarily assessed, these studies often present a fragmented focus on N balance in rice fields. Key components of the N cycle are inadequately considered. For example, gaseous N emissions tend to focus on N₂O while overlooking NO or NH₃, or soil N inputs and outputs, along with losses through leaching and runoff, lack detailed quantification. In summary, comprehensive assessments of the effects of LCC on N balance in rice fields, particularly studies encompassing all major N input and output pathways, remain scarce. This lack of holistic research limits our understanding of how LCC management practices influence the N cycle in rice fields and their environmental sustainability.

We hypothesized that LCC-based N management would enhance leaf N metabolic enzyme activity, reduce reactive N losses, and restore N balance in rice fields by synchronizing fertilizer inputs with crop demand. To test these hypotheses, a two-years of field experiments, used the local farmers’ conventional practices as a control to analyze the effects of LCC-based urea and slow-release urea management on rice yield, N metabolism-related enzyme activity in leave, N use efficiency, gaseous N losses (NH₃, N₂O and NO), N leaching and runoff, and soil N inputs and outputs. Additionally, N inputs from atmospheric deposition and irrigation water were measured, with a no-N treatment set as a baseline to represent background N inputs and outputs. Our study provides new insights into the mechanisms by which LCC influences N utilization in rice and further quantifies its impact on N balance in paddy fields. The findings hold significant potential for accelerating the adoption of LCC as a sustainable rice cultivation technology.

2. Materials and Methods

2.1. Experimental Materials and Field Site

The experimental site is located at the farm in Shanhe Town of Wuchang City, Heilongjiang Province, China (44.73° N, 127.23° E). This region has a temperate monsoon climate, with characteristics of a cold temperate humid continental climate. The average temperatures during the rice growth period in 2022 and 2023 were 17.6 and 17.8 °C, respectively, with total rainfall amounts of 669.0 and 783.7 mm, respectively (Figure 1). The main soil (silty clay loam) properties (0–20 cm depth) were as follows: bulk density (BD) of 1.08 g cm⁻³, soil pH of 6.3, soil organic carbon (SOC) and total N (TN) contents of 28.63 and 1.52 g kg⁻¹, nitrate (NO₃⁻) of 14.36 mg kg⁻¹, ammonium (NH₄⁺) of 17.11 mg kg⁻¹, available phosphorus (Olsen P) of 16.32 mg kg⁻¹, and available potassium (AK) of 95.63 mg kg⁻¹.

Only one season of rice (Oryza sativa L.) is planted each year, and the cultivated variety is Daohuaxiang No. 2. Fertilizers applied during the rice growth period included urea (46% N), Controlled-release urea (CRU) (44% N), potassium chloride (60% K₂O), and superphosphate (12% P₂O₅). The CRU (Controlled-Release Urea) was provided by Dongkai Green Treasure Pearl (Shandong) Ecological Technology Co., Ltd. (Heze, China) with polyurethane as the coating material. Its release characteristics are detailed in Supplementary Table S1. The Leaf Color Chart (LCC) from the International Rice Research Institute (IRRI) was used to assess the color of rice leaves and guide fertilization management. LCC has four green strips, with color ranging from yellow green to dark green (2 to 5).

2.2. Experimental Design and Agronomic Management

In 2022 and 2023, four N fertilizer management methods (FNR, SSNM1, SSNM2, ONF) were designed using random blocks with three replicates. FNR, the local farmers’ urea application practice as the control, involved applying 80 kg N ha⁻¹ one day before transplanting, 60 kg N ha⁻¹ at 7 days after transplanting (DAT), and 60 kg N ha⁻¹ at 42 DAT (the panicle initiation stage). SSNM1 involved applying 80 kg N ha⁻¹ of urea one day before transplanting, followed by measuring the LCC value every 7 days until 42 DAT. Additional 20 kg N ha⁻¹ of urea was applied only when the LCC value < 3. SSNM2 was based on SSNM1, with urea replaced by CRU. ONF denoted the absence of N fertilizer application. All treatments were applied with 60 kg P₂O₅ ha⁻¹ and 120 kg K₂O ha⁻¹ one day before rice transplanting. Rice was sown on 9 April 2022, and 6 April 2023; transplanted on 11 May 2022, and 7 May 2023; and harvested on 22 September 2022, and 19 September 2023. The manual transplanting density was 30 cm × 16 cm, with two seedlings per hill. Each plot was separated by ridges (35 cm wide and 35 cm high) that were covered with plastic film. The area of each plot was 30 m² (5 m × 6 m). For the four experimental treatments, irrigation and application of pesticide were the same. According to the local conventional irrigation-drainage practices, rice fields were maintained under deep water irrigation (5–10 cm depth) after transplanting until the end of the tillering stage, when fields were drained. From the booting stage to one week before harvest, alternate wetting and drying irrigation management was implemented. In accordance with local practices, rice straw is typically harvested, baled, and reserved for subsequent use; consequently, straw was removed from the experimental plots in this study. Under similar management practices as described above, pre-trials conducted in 2021 determined that the optimal LCC threshold for reducing N fertilizer input while increasing yield was 3 (Supplementary Figure S1).

2.3. Leaf Color Chart Value

At least 10 disease-free rice plants were randomly selected per plot, ensuring a uniform plant population. The topmost, youngest, fully expanded leaf from each hill or plant was selected for measurement. The middle part of each leaf was placed against the LCC, and its color was compared with the chart’s color panels. Measurements were taken under shade to avoid direct sunlight, which could affect readings. When the leaf color fell between two shades, the average of the two values was recorded (e.g., a color between 3 and 4 was recorded as 3.5). LCC measurements were conducted every 7 days starting from the beginning of tillering (7 DAT) and continued until after panicle initiation (42 DAT).

2.4. Enzyme Activities

At the initiation tillering stage (ITS), peak tillering stage (PTS), booting stage (BS), grain filling stage (GFS), and maturity stage (MS), the fully expanded top three leaves of the main stem of rice were collected, with 10 g of samples stored at –80 °C. Each experimental treatment was performed in triplicate. Before enzyme activity determination, leaf samples were homogenized following the method of [21]. Nitrate reductase (NR) activity was measured according to the method of [22] and expressed as the amount of nitrite ions (nmol) produced per gram of fresh weight (FW) per minute through the reduction in nitrate ions. Glutamine synthetase (GS) activity was measured following the method of [21] and expressed as the amount of L-glutamate γ-monohydroxamate (µmol) produced per gram of FW per minute. The activities of Glutamate-oxaloacetate transaminase (GOT) and glutamate-pyruvate transaminase (GPT) were determined using a colorimetric activity assay following the method of Ning et al. (2012) [23]. The activities were expressed as the amount of pyruvate (µmol) produced per gram of FW within 30 min. In addition, the activity of glutamate dehydrogenase (GDH) was determined using a quartz cuvette-based microplate assay provided by the kit (BIOSS, Beijing, China) (http://www.bioss.com.cn/upload/datasheet/AK079_201.pdf) (accessed on 5 February 2021). The GDH activity was expressed as the amount of nicotinamide adenine dinucleotide (NADH) (µmol) consumed per gram of FW per hour. One unit (1 U) of protease activity is defined as the activity of enzyme that increases the absorbance at 280 nm (A280) by 0.01 per hour. Protease activity is expressed per milligram of protein, and the protein content in enzyme extracts was determined using the Bradford method (1976) [24].

2.5. Grain Yield and Nitrogen Uptake

After rice matured, an area of about 3 m² was selected from the central part of each plot for grain yield measurement. and then the moisture content of rice grains was measured with a PM-8188-A portable grain moisture (Kett Electric Laboratory, Tokyo, Japan). The rice grains were weighed and converted into the yield at 14% water content. Moreover, the fifth hill of representative plants was selected, killed at 105 °C, and dried to constant weight at 80 °C. The samples dried were crushed and the N contents of rice leaves, stems and roots (0–20 cm depth) were measured by the ECS 4024 CHNSO Classic Analyzer (Costech Analytical Technologies, Milan, Italy). The rice N uptake (NP) was calculated as the sum of the N content in each organ multiplied by its corresponding dry biomass. Each treatment had three biological repeats. N use efficiency (NUE_diff) using the N difference approach were calculated using the following equations [7]:

$${\mathrm{NUE}}_{\mathrm{diff}} (\%) = {\displaystyle \frac{{\mathrm{NP}}_{\mathrm{t}}-{\mathrm{NP}}_0}{{\mathrm{N}}_{\mathrm{rate}}}}$$

(1)

where NP_t and NP₀ represent the rice N uptake (kg N ha⁻¹) in plots with N fertilization and without N fertilization, respectively. Y_t and Y₀ are the rice yield (kg ha⁻¹) in plots with N fertilization and without N fertilization, respectively. Moreover, N_rate is the N fertilizer application rate under the fertilized plots.

2.6. Gaseous Nitrogen Emissions

The samples of N₂O and NO were collected with a static closed chamber at intervals of 3–10 days (8:00–10:00 a.m.) after rice transplantation. After the seedlings had been transplanted, the gases were continuously sampled until the rice had been harvested. During the sampling period, two rings were inserted for each plot into the soil at a depth of ~5 cm surrounding two rice plants, and the chambers were temporarily placed in the groove of the rings. The chamber was sealed by manually adding water to the base when it was unsubmerged. After the chamber had been closed for 0, 8, 16 and 24 min, the gas samples were collected successively with a 20 mL syringe and then transferred to 20 mL vacuum glass bottles. Moreover, after the chamber had been closed for 0, 24 and 48 min, 2 L of air were collected into gas sampling bags using a pump. The gas samples were stored for no longer than 2–3 days before they were measured. The concentrations of N₂O were analyzed using a gas chromatograph (Shimadzu GC-14 B, Kyoto, Japan). The measurement parameters of gas chromatograph and calculation methods of N₂O emission flux and cumulative emissions in this study were consistent with those reported by Tang et al. (2025) [25]. Using dark chamber sampling and aluminized film gas bags for sample storage can prevent the photolysis of nitrogen dioxide (NO₂) to NO. However, NO can be oxidized to NO₂ by the initial ozone (O₃), hydroperoxyl (H₂O₂), and organic peroxy radicals (RO₂) present in the chamber. Therefore, the sum of NO and NO₂ fluxes was used to represent the final NO flux [26]. The concentrations of NO and NO₂ were measured using a chemiluminescence nitrogen oxide analyzer (42C, Thermo Electron Corporation, Waltham, MA, USA). The sample flow rate was set to 800 mL min⁻¹ to ensure sufficient sample volume and maintain optimal reaction efficiency. Ozone, generated by the internal ozone generator, was supplied at a concentration of 250 ppm to react with NO. The reaction chamber temperature was maintained at 45 °C to optimize the chemiluminescence reaction. NO₂ was converted to NO in a molybdenum converter heated to 325 °C, with a conversion efficiency above 95%. Data acquisition was performed at 1 Hz, with an optional 10 s average to reduce short-term fluctuations. The NO emission flux and cumulative emissions were also calculated using the methodology described in Tang et al. (2025) [25]. Similarly, the sampling frequency for NH₃ was consistent with that for N₂O and NO, and the batch-type airflow enclosure method [25] was used to measure NH₃ volatilization. The calculation of NH₃ volatilization flux and cumulative emissions followed the methodology described by Tang et al. (2025) [25].

2.7. Nitrogen Content in Paddy Soil, Nitrogen Leaching and Runoff

The lysimeter was made of PVC with a pipe diameter of 10 cm and a length of 80 cm. The bottom of the pipe was sealed, and the top was fitted with a removable cap. Three rows of small holes (6 mm in diameter) were drilled evenly within a 15 cm section, starting 15 cm from the bottom of the pipe. Each row contained five holes, and the pipe was wrapped with two layers of nylon mesh to prevent sediment intrusion. The lysimeter was buried 60 cm into the soil, with the perforated section positioned below 30 cm underground, ensuring that all holes were located beneath the plow pan of the paddy field. This setup allowed for the collection of water samples passing through the plow pan. Leachate samples from the paddy field were collected at regular intervals of 7–14 days during the rice growing season. A PVC container was used to collect runoff generated by rainfall and irrigation. The collector had an area of 2 m × 2 m, with 30 cm buried underground and 50 cm above ground. The collector was connected to the field ridge through a PVC pipe, ensuring that when the water depth in the plot exceeded 30 cm, the excess water would flow into the collector. During the rice-growing period, a total of 16 runoff events occurred. After each event, the runoff volume was recorded, and water samples were collected. Leachate and runoff water samples were stored at 4 °C. The concentrations of TN, NH₄⁺, and NO₃⁻ were determined using the alkaline potassium persulfate digestion ultraviolet visible (UV) spectrophotometric method (GB 1184-89) [27], the indophenol blue colorimetric method (GB/T 8538-1995) [28], and the UV spectrophotometry method (HJ/T 346-2007) [29], respectively. The cumulative N leaching (NL_total) and runoff (NR_total) volume were calculated using the following equations:

$${\mathrm{NL}}_{\mathrm{total}} (\mathrm{kg} \mathrm{N} {\mathrm{ha}}^{-1}) = {\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{C}}_{\mathrm{i}}\times {\mathrm{V}}_{\mathrm{i}}\times \mathrm{B}}{{10}^6}}$$

(2)

$${\mathrm{NR}}_{\mathrm{total}} (\mathrm{kg} \mathrm{N} {\mathrm{ha}}^{-1})={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{CS}}_{\mathrm{i}}\times {\mathrm{VS}}_{\mathrm{i}}\times \mathrm{BS}}{{10}^6}}$$

(3)

where C_i and CS_i represent the N concentrations (mg N L⁻¹) in the leachate and runoff samples collected during each event, respectively. V_i and VS_i denote the recorded volumes (L) of leachate and runoff for each event, respectively. B is a constant (159,155) used to convert the leachate volume collected by the lysimeter into leachate volume per hectare [30]. Similarly, BS is a constant (333) used to convert the runoff volume collected by the runoff collector into runoff volume per hectare.

Paddy soil from a depth of 0–20 cm was collected from each plot before rice transplanting and after harvest using the five-point sampling method. The soil samples in each plot were mixed with three replications, and put into centrifuge tubes. The suspensions in the tubes, considered as the soil solution samples, were collected after centrifuging at 4500× g at 24 °C for 30 min. The remaining solid soil without suspensions were extracted with 2 mol L⁻¹ potassium chloride (KCl) for 1 h at 24 °C. The soil solution and extract were refrigerated at 4 °C after filtering through a microporous membrane (0.45 μm). Moreover, the soil TN was extracted by digesting the soil with alkaline potassium persulfate solution (5 g L⁻¹ potassium persulfate [K₂S₂O₈] containing 10 g L⁻¹ sodium hydroxide [NaOH]) at high temperature (121 °C, 30 min). After cooling, the extract was centrifuged and filtered following the same procedure described above, and the supernatant was stored for subsequent analysis. The methods for determining TN, NH₄⁺, and NO₃⁻ in the soil extract were the same as those used for water samples.

2.8. Atmospheric Nitrogen Deposition and Nitrogen Content in Irrigation Water

The atmospheric N deposition was collected using an automatic wet and dry deposition sampler (APS-3A, Changsha Xianglan Scientific Instrument, Changsha, China). The precipitation volume in the dustfall collector was recorded and collected monthly. During sampling, rainwater was used to rinse the inner walls of the polyurethane foam to simultaneously collect both wet and dry deposition nitrogen. Moreover, ZJPVC-BT01-DN80 irrigation flow meter (Zhongjiang Energy Saving Electronics, Shunde, China) installed in the irrigation pipeline was used for irrigation monitoring and control. The total seasonal water input was calculated as the sum of irrigation water from transplanting to harvest. The TN content in rainwater (TN₁) and irrigation water (TN₂) was determined using the methods described above (Section 2.7), and the calculation equations are as follows:

$${\mathrm{TN}}_1 (\mathrm{kg} \mathrm{N} {\mathrm{ha}}^{-1}) = {\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{Z}}_{\mathrm{i}}\times {\mathrm{P}}_{\mathrm{i}}\times \mathrm{K}}{{10}^6}}$$

(4)

$${\mathrm{TN}}_2 (\mathrm{kg} \mathrm{N} {\mathrm{ha}}^{-1})={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{ZS}}_{\mathrm{i}}\times {\mathrm{PS}}_{\mathrm{i}}\times \mathrm{KS}}{{10}^6}}$$

(5)

where Z_i and ZS_i represent the N concentrations (mg N L⁻¹) in the rainwater and irrigation water collected during each event, respectively. P_i and PS_i denote the recorded volumes of rainwater (mm) and irrigation water (m³) for each event, respectively. K is a constant (10,000) used to convert rainfall (mm) into rainfall volume (L) per hectare. KS is a constant (3.3 × 10⁵) used to convert irrigation water (m³) per plot into irrigation water (L) per hectare.

2.9. Statistical Analysis of Data

All statistical analyses were conducted using the SPSS 17.0 analytical software package. The effects of fertilization methods on rice yield, enzyme activities, rice N utilization, gaseous N emissions, N leaching and runoff were evaluated through one-way analysis of variance (ANOVA). Mean differences among multiple groups were compared using the least significant difference (LSD) test, with a significance level set at 0.05. A one-sample t-test was used to assess changes in soil nitrogen content before and after rice cultivation under different fertilization treatments, with significance levels set at 0.05, 0.01, and 0.001. Additionally, the contribution of N fertilizer to rice yield, rice N utilization, gaseous N emissions, N leaching and runoff was calculated as the difference between treatments with and without N fertilization. The contribution of fertilizer to soil N input was defined as the difference between N fertilizer application rate and other N outputs (including rice N uptake, gaseous N losses, N leaching and runoff). All data met ANOVA assumptions (normality by Shapiro–Wilk test, p > 0.05; homogeneity of variance by Levene’s test, p > 0.05) without requiring transformations. Net ecosystem economic benefit (NEEB) and N loss per unit yield are shown in Supplementary Text S1, Table S2 and Figure S2.

3. Results

3.1. Leaf Color Chart (LCC) Value and Nitrogen Reduction Amount

In 2022 and 2023, the LCC values (3.3–3.4) of SSNM1 were above 3.0 only at 14 DAT, while during other periods they ranged from 2.2 to 2.7 (Figure 2a,b). In contrast, the LCC values (3.1–3.6) of SSNM2 exceeded 3.0 at 7 and 21 DAT, while remaining between 2.5 and 2.9 at other sampling times (Figure 2a,b). Compared with FNR, the N application rates in SSNM1 and SSNM2 were reduced by 10% and 20%, respectively, over the two years (Figure 2c,d).

3.2. Nitrogen Metabolism-Related Enzyme Activities in Rice Leaves

In 2022 and 2023, compared to ONF, N application treatments (FNR, SSNM1, and SSNM2) increased the activities of NR, GS, GPT, GOT, and GDH while reducing proteinase activity throughout the growth stages (p < 0.05) (Figure 3a–l). There were no significant differences in the activities of NR, GS, GPT, GOT, GDH, and proteinase between SSNM1 and SSNM2 (p > 0.05) (Figure 3a–l). At ITS, the activities of NR, GS, GPT, GOT, and GDH under SSNM1 and SSNM2 were lower than those under FNR, while proteinase activity was higher. However, this trend was reversed during other growth stages (Figure 3a–l). Specifically, compared with FNR, the activities of NR, GS, GPT, GOT, and GDH under SSNM1 were 35.7%, 34.5%, 34.5%, 35.2%, and 30.0% lower at ITS, respectively, while proteinase activity was 53.9% higher (mean of two years) (p < 0.05). Compared with FNR, the activities of NR, GS, GPT, GOT, and GDH under SSNM1 were 16.3–26.6%, 25.2–52.3%, 19.4–54.0%, 17.1–28.7%, and 20.3–25.6% higher in the subsequent growth stages, respectively, while proteinase activity was 19.6–31.8% lower (averaged across two years) (p < 0.05). Similarly, compared with FNR, the activities of NR, GS, GPT, GOT, and GDH under SSNM2 were 36.2%, 32.2%, 33.6%, 34.2%, and 29.4% lower at ITS, respectively, while proteinase activity was 52.9% higher (mean of two years) (p < 0.05). Compared with FNR, the activities of NR, GS, GPT, GOT, and GDH under SSNM2 were 16.7–31.4%, 24.9–53.3%, 19.7–51.1%, 15.2–27.4%, and 21.0–33.9% higher during the remaining growth stages, respectively, while proteinase activity was 18.1–29.8% lower (averaged across two years) (p < 0.05).

3.3. Rice Yield and Nitrogen Use Efficiency

Over the two years, soil and fertilizer contributed 49.3–61.5% and 38.5–50.7% of rice yield, respectively, and 42.5–54.6% and 45.4–57.5% of plant N uptake (Figure 4a,b). There were no significant differences in rice yield and plant N uptake between SSNM1 and SSNM2 (p > 0.05) (Figure 4a–c). However, both treatments increased rice yield, plant N uptake, and ANE compared with FNR (Figure 4a–c). Specifically, the two-year average rice yield, plant N uptake, and NUE_diff under SSNM1 were 9.2%, 15.4%, and 46.9% higher than those under FNR, respectively (p < 0.05). Similarly, the two-year average rice yield, plant N uptake, and NUE_diff under SSNM2 were 9.4%, 15.3%, and 65.0% higher than those under FNR, respectively (p < 0.05).

Pearson correlation analysis showed that rice yield and N uptake were positive correlations with NR, GS, GPT, GOT and GDH activities, but were negative correlation with proteinase activity (

Table 1. The Pearson correlation coefficient between enzyme activities and rice yield and N uptake (n = 8).

	Rice Yield	N Uptake	NR	GS	GPT	GOT	GDH	Proteinase
rice yield		0.997 **	0.996 **	0.971 **	0.985 **	0.992 **	0.956 **	−0.995 **
N uptake	0.997 **		0.996 **	0.982 **	0.992 **	0.996 **	0.970 **	−0.991 **
NR	0.996 **	0.996 **		0.980 **	0.988 **	0.998 **	0.963 **	−0.997 **
GS	0.971 **	0.982 **	0.980 **		0.992 **	0.989 **	0.985 **	−0.964 **
GPT	0.985 **	0.992 **	0.988 **	0.992 **		0.994 **	0.990 **	−0.977 **
GOT	0.992 **	0.996 **	0.998 **	0.989 **	0.994 **		0.978 **	−0.991 **
GDH	0.756 *	0.970 **	0.963 **	0.985 **	0.990 **	0.978 **		−0.946 **
Proteinase	−0.995 **	−0.991 **	−0.997 **	−0.964 **	−0.977 **	−0.991 **	−0.946 **

NR, Nitrate reductase; GS, Glutamine synthetase; GPT, Glutamate-pyruvate transaminase; GOT, Glutamate-oxaloacetate transaminase; GDH, Glutamate dehydrogenase. * and ** represent significant at 5% and 1% levels, respectively.

).

3.4. Gaseous Nitrogen Emissions from Rice Fields

The peak fluxes of NH₃, N₂O, and NO emissions occurred after each topdressing application and subsequently declined (Figure 5a–f). Compared to FNR, SSNM1 and SSNM2 exhibited higher frequencies of NH₃, N₂O, and NO emissions peaks but with lower peak values (Figure 5a–f). Over the two years, the peak emission fluxes of NH₃, N₂O, and NO under FNR were 1.8–2.8 kg N ha⁻¹ d⁻¹, 121.4–203.0 µg N m⁻² h⁻¹, and 66.7–125.6 µg N m⁻² h⁻¹, respectively. In comparison, the peak emission fluxes under SSNM1 were 0.3–0.9 kg N ha⁻¹ d⁻¹, 64.9–164.6 µg N m⁻² h⁻¹, and 30.1–42.6 µg N m⁻² h⁻¹, respectively. Lastly, and with the lowest values, the peak emission fluxes under SSNM2 were 0.3–0.4 kg N ha⁻¹ d⁻¹, 58.3–95.9 µg N m⁻² h⁻¹, and 25.0–40.8 µg N m⁻² h⁻¹, respectively.

Over the two years, soil and fertilizer contributed 7.6–28.6% and 71.4–92.4% of the cumulative NH₃ emissions, 12.4–19.9% and 80.1–87.6% of the cumulative N₂O emissions, and 8.2–16.5% and 83.5–91.8% of the cumulative NO emissions, respectively (Figure 5g–i). The cumulative emissions of NH₃, N₂O, and NO in 2022 and 2023 followed a consistent pattern: SSNM2 < SSNM1 < FNR (p < 0.05) (Figure 5g–i). Compared with FNR, the two-year average NH₃, N₂O, and NO emissions under SSNM1 were reduced by 45.2%, 19.2%, and 22.0%, respectively. Furthermore, over two years, the average NH₃, N₂O, and NO emissions under SSNM2 were 70.4%, 34.4%, and 36.7% lower, respectively, than those under FNR.

3.5. Nitrogen Leaching and Runoff in Rice Fields

Across the two years, soil and fertilizer contributed 4.5–20.5% and 79.5–95.5% of NH₄⁺ leaching, 4.6–16.8% and 83.2–95.4% of NO₃⁻ leaching, and 4.8–17.3% and 82.7–95.2% of TN leaching, respectively (Figure 6a,b). The leaching amounts of NH₄⁺, NO₃⁻, and TN in 2022 and 2023 followed the consistent trend: SSNM2 < SSNM1 < FNR (p < 0.05) (Figure 6a,b). Compared to FNR, the two-year average leaching amounts of NH₄⁺, NO₃⁻, and TN under SSNM1 were reduced by 55.6%, 54.0%, and 54.9%, respectively. Under SSNM2, the NH₄⁺, NO₃⁻, and TN leaching amounts (mean of two years) were 71.3%, 69.2%, and 69.0% lower, respectively, than those under FNR.

The N runoff in paddy fields was lower than the RN loss through leaching (Figure 6a–d). Over the two years, soil and fertilizer contributed 5.9–15.8% and 84.2–94.1% of NH₄⁺ runoff, 5.6–16.4% and 83.6–94.4% of NO₃⁻ runoff, and 6.2–16.5% and 83.5–93.8% of TN runoff, respectively (Figure 6c,d). The runoff amounts of NH₄⁺, NO₃⁻, and TN in both years followed the trend SSNM2 < SSNM1 < FNR (p < 0.05) (Figure 6c,d). Compared with FNR, the NH₄⁺, NO₃⁻, and TN runoff under SSNM1 were reduced by 31.6%, 31.2%, and 30.9%, respectively, on average over the two years. Moreover, NH₄⁺, NO₃⁻, and TN runoff (mean of two years) under SSNM2 were reduced by 58.4%, 58.0%, and 57.5%, respectively, compared with FNR.

3.6. Nitrogen Content in Paddy Soil

Compared with the pre-cultivation levels in 2022, the soil NH₄⁺ content under the ONF treatment decreased by 36.5% and 62.3% after harvest in 2022 and 2023, respectively, while the NO₃⁻ content decreased by 17.0% and 41.0%, respectively (p < 0.05) (Figure 7). However, the TN content only showed a reduction of 3.9% after harvest in the second year (p < 0.05). Additionally, the three fertilization treatments (FNR, SSNM1, and SSNM2) did not significantly affect soil NH₄⁺, NO₃⁻, and TN content (p > 0.05).

3.7. Nitrogen Balance in Rice Fields

In 2022 and 2023, RN loss from paddy fields (gaseous N emissions, N leaching, and runoff) primarily derived from fertilizers (77.3–93.5%) rather than from soil (6.5–22.7%) (Figure 8a–f). Compared with FNR, both SSNM1 and SSNM2 reduced fertilizer RN losses (Figure 8a–f). Specifically, fertilizer RN losses averaged 34.7% (69.4 kg N ha⁻¹), 19.3% (34.8 kg N ha⁻¹), and 12.3% (19.6 kg N ha⁻¹) under FNR, SSNM1, and SSNM2, respectively, over the two years. Gaseous N emissions were the primary pathway of fertilizer RN loss, accounting for 48.2%, 26.1%, and 14.3% under FNR, SSNM1, and SSNM2, respectively. Moreover, under the FNR treatment, soil contributed more to rice N uptake than fertilizers, whereas the opposite was observed under SSNM1 and SSNM2 (Figure 8a–f). Furthermore, the soil N input and output under FNR, SSNM1, and SSNM2 treatments were nearly balanced, with a net input of only 4.2–13.4 kg N ha⁻¹ over the two years. The contributions of atmospheric N deposition and irrigation water to soil N inputs were only 6.2–9.2% of the contributions from fertilizers (Figure 8a–f).

4. Discussion

4.1. Effects of Leaf Color Chart-Based N Management on Enzyme Activities, N Utilization and Rice Yield

Plants primarily acquire inorganic N from the soil in the forms of NO₃⁻ and NH₄⁺ [31]. NR and GS are key enzymes in N assimilation. NR catalyzes the reduction of NO₃⁻, marking the first step of N assimilation in plants [32], while GS facilitates the combination of NH₄⁺ with glutamate to form glutamine, which subsequently generates more glutamate through the glutamine synthetase–glutamate synthase (GS-GOGAT) cycle [33]. Additionally, NH₄⁺ can be directly converted to glutamate via GDH [34]. GOT and GPT are the most active transaminases, catalyzing the formation of aspartate and alanine from glutamate, respectively [35]. These enzymes work synergistically through shared substrates and products to complete N assimilation and transformation. Adequate supplies of NO₃⁻ and NH₄⁺ may stimulate the activity of these enzymes to meet the increased N demands of plants. Therefore, treatments (SSNM1 and SSNM2) with higher N uptake rates maintained consistently high activities of NR, GS, GOT, GPT and GDH throughout most growth stages (Tabel 1 and Figure 3a–j). In contrast, proteinase activity, which decomposes proteins to supplement amino acids and other metabolites under N deficiency or stress conditions [21,36], showed an opposite trend compared to the activities of NR, GS, GOT, GPT, and GDH (Table 1). Notably, at 7 DAT, the FNR treatment received a substantial N application of 60 kg N ha⁻¹, whereas SSNM1 received only 20 kg N ha⁻¹, and SSNM2 received none (Figure 2a,b). As a result, FNR exhibited higher activities of NR, GS, GOT, GPT and GDH than SSNM1 and SSNM2 at the ITS, whereas proteinase activity showed the opposite trend (Figure 3). In this study, N management based on the LCC aimed to maintain appropriate N levels in rice. Since SSNM1 and SSNM2 used the same leaf color thresholds, there were no significant differences in the six enzyme activities or N uptake between these treatments, despite differences in N fertilizer types (Figure 3 and Figure 4b). However, compared to urea, CRU has a longer fertilizer release cycle at the same N application rate [37]. This characteristic enabled SSNM2 to require the least amount of N fertilizer while achieving the highest NUE_diff (Figure 2b,c and Figure 4c). Furthermore, N is a key component of nucleic acids, proteins, and other nitrogenous compounds and serves as the foundation of carbon metabolism in crops [31]. Consequently, similar to the differences in N uptake between fertilization treatments, there was no significant difference in grain yield between SSNM1 and SSNM2, both of which outperformed FNR (Figure 4a,b). These results align with the positive correlation between rice yield and nitrogen uptake [38,39]. In summary, LCC-based N management enhances rice N uptake and grain yield by sustaining high levels of NR, GS, GOT, GPT and GDH enzyme activities over the long term while maintaining low protease activity. Moreover, combining LCC with controlled-release N fertilizer was more advantageous in improving N use efficiency.

4.2. Effects of Leaf Color Chart-Based N Management on Reactive N Losses in Rice Fields

Reactive nitrogen (RN), defined as all forms of N compounds excluding N₂, primarily escapes from farmlands into the atmosphere and freshwater systems through pathways such as NH₃ volatilization, N₂O and NO emissions, runoff, and leaching [9,40]. In this study, RN loss to the environment under the FNR treatment accounted for 34.7% of N fertilizer (Figure 8a,d), which approaches the upper limit of RN loss from Chinese rice fields (15.3–34.6%) [41]. However, under the SSNM1 treatment, the two-year average RN loss from fertilizers was reduced to only 19.3%, with gaseous N emissions decreasing from 17.6% to 9.9% (Figure 8b,e). During the rice growth period, most gaseous N emissions typically occurred after fertilizer application [42,43]. Excessive N fertilization resulted in an exponential rather than linear increase in gaseous N emissions, due to competition for N between crops and soil microbes [26,44]. Therefore, reducing the amount of N applied per application is critical for mitigating gaseous N emissions. The LCC-based management approach not only reduced the total amount of urea applied but also optimizes its application timing (Figure 2), thereby lowering the peak emissions of NH₃, N₂O and NO. This strategy promoted greater N uptake by the crop rather than leading to its loss to the atmosphere. Regarding N pollution in water bodies, NO₃⁻, being highly soluble and weakly adsorbed by soil, is the primary form of N lost through leaching and runoff [45]. As a result, NO₃⁻ losses were most significantly reduced under the SSNM1 treatment, with the largest decreases observed in leaching and runoff. Based on the estimated leaching and runoff coefficients for N fertilizer in rice fields, reducing N application by 20 kg was expected to decrease leaching by 0.6 kg N and runoff by 0.5 kg N [41]. However, reductions of 13.8 kg N in leaching and 3.4 kg N in runoff were observed under the SSNM1, exceeding expectations. This may result from a synergistic effect between increased rice N uptake and reduced urea application. Compared to SSNM1, SSNM2 further reduced RN losses due to the use of urease inhibitors, which lowered the amount of N fertilizer required, slowed the urease decomposition rate, and inhibited both nitrification and denitrification rates [38,46]. Overall, N management based on the LCC effectively reduced RN losses from rice fields, and its benefits were further enhanced when combined with controlled-release N fertilizers.

4.3. Effects of Leaf Color Chart-Based N Management on Soil N Balance

Soil N is an essential source of N required for crop growth. Even with N fertilizer application, more than half of the N accumulated by crops still obtained from the soil [47]. Therefore, maintaining soil N balance or even enhancing soil N supply capacity is critical for ensuring sustainable crop production. In this study, changes in soil N reserves were determined by the relative quantities of N inputs (atmospheric deposition, irrigation water and fertilizer) and outputs (plant N uptake and RN losses) (Figure 8). Without N fertilizer application, soil NH₄⁺ and NO₃⁻ content decreased year by year, and soil TN significantly declined in the second year after planting (Figure 7). soil NH₄⁺ and NO₃⁻, being readily available nutrients, could undergo significant short-term changes due to anthropogenic or environmental influences [48,49]. In contrast, TN was relatively stable because it contained organic N encapsulated within soil aggregates [33,50]. Fertilizer served as the primary pathway for soil N inputs in 2022 and 2023, average accounting for 93.4% under FNR, 93.0% under SSNM1, and 92.6% under SSNM2 (Figure 8). These values are higher than those reported by [51], likely because crop residues were removed in this study. Our findings also demonstrated that rice N uptake was the dominant pathway for soil N outputs, average accounting for 92.2% of the total (Figure 8). Furthermore, although the N fertilizer application rates decreased in the order of FNR, SSNM1, and SSNM2, the differences between soil N inputs and outputs among the three treatments were minimal (Figure 7 and Figure 8), achieving a dynamic balance [10]. This may be attributed to the soil N retention capacity reaching its maximum under all three fertilization scenarios or the benefits of split fertilizer applications, which promote soil N fixation and compensate for the reduction in fertilizer N [52]. Our findings highlight the value of LCC-based N fertilizer management in sustainable agriculture, as it reduces N fertilizer use while maintaining soil N balance.

4.4. Uncertainty and Limitations

Our study focuses on a mechanistic interpretation of how LCC influences N balance in paddy fields. However, we acknowledge that the findings may be region-specific due to the lack of multi-site validation, which limits our ability to assess uncertainties arising from environmental heterogeneity (e.g., climatic and soil property variations). Nevertheless, this research serves as a critical foundation for understanding LCC’s role in rice field nitrogen management. Furthermore, we compared the NEEB of two LCC-based approaches with FNR, revealing similar gains. While this information can support localized decision-making, fluctuations in rice grain market prices may introduce significant variability in NEEB outcomes.

5. Conclusions

Rice leaf enzyme activity was closely related to N uptake. LCC-based N management increased rice N uptake by 15.3–15.4% and grain yield by 9.2–9.4% through sustaining high levels of NR, GS, GOT, GPT and GDH enzyme activities over the long term while maintaining low protease activity. Moreover, this method effectively reduced N fertilizer application by 10–20% and RN losses by 46.2–66.7% in rice fields. Combining LCC with controlled-release N fertilizer proved to be more effective in improving N use efficiency and reducing RN losses compared with conventional urea. While LCC-based N management did not significantly alter soil N content in the short term, this outcome can be seen as a positive indication, given that soil N inputs are steadily accumulating at a low annual rate (8.9 or 4.2 kg N ha⁻¹). Based on this, we predict that LCC will further reduce N fertilizer application rates in the future as soil N content may gradually increase. Therefore, a long-term field observation is essential to validate our hypothesis.