Comparative Analysis of Yield and Grain-Filling Characteristics of Conventional Rice with Different Panicle Types in Response to Nitrogen Fertilization

Abstract

This study investigated the impact of nitrogen (N) fertilization on the yield and grain filling (GF) characteristics of two conventional japonica rice varieties with distinct panicle types: Yangchan 3501 (large-panicle: spikelets per panicle $> 150$) and Nangeng 46 (medium-panicle: $100 <$ spikelets per panicle $< 150$). Field experiments were conducted over two growing seasons (2022–2023) with three N application rates (T1: 225 kg ha⁻¹, T2: 270 kg ha⁻¹, T3: 315 kg ha⁻¹). Key measurements included tiller dynamics, panicle composition, GF parameters modeled using the Richards equation, and enzyme activities related to nitrogen metabolism (Fd-GOGAT, NR) and carbohydrate transport (α-amylase, SPS). Results showed that the yield increased with higher N levels for both varieties, with Yangchan 3501 achieving higher yields primarily through increased grains per panicle (15.65% rise under T3 vs. T1), while Nangeng 46 relied on panicle number (8.83% increase under T3 vs. T1). Nitrogen application enhanced Fd-GOGAT and NR activities, prolonging photosynthesis and improving GF rates, particularly in the inferior grains of Yangchan 3501 during middle and late stages. However, a high N reduced seed-setting rates and 1000-grain weight, with larger panicle types exhibiting a greater sensitivity to N-induced changes in branch structure and assimilate allocation. This study highlights that optimizing N management can improve nitrogen-metabolism enzyme activity and GF efficiency, especially in large-panicle rice, while medium-panicle types require higher N inputs to maximize panicle number. These findings provide actionable insights for achieving high yields and efficient nutrient use in conventional rice cultivation.

1. Introduction

Rice (Oryza Sativa L.) ranks among the world’s primary food crops [1]. In the last forty years in China, enhancements in rice yield have primarily stemmed from increased sink capacity [2], with the cultivation of large panicles serving as a crucial method to augment this capacity [3]. Grain filling (GF) represents the most vital physiological process for determining grain weight in rice. Variations in GF occur within different sections of the panicle of the same variety [4]. This difference is particularly pronounced in rice varieties with an average number of grains per panicle exceeding 150 [1,5]. Prior research has identified several factors contributing to poor GF: (1) Insufficient photosynthetic capacity: The material production advantage in rice is generally more pronounced during the GF stage, whereas it is relatively lower before heading [6]. In the initial phase of grain development, the average supply of assimilates to individual spikelets is clearly inadequate, with photosynthetic assimilates primarily allocated to the middle and upper grains during filling. As a result, the filling rate of basal grains decreases, sometimes preventing any filling, which leads to lighter grain weight and lower seed-setting rates in the inferior basal grains [7]. (2) The impaired transport of assimilates: Compared to the branch’s vascular bundle of superior grains, the vascular bundle and phloem areas of the branch at the base of the inferior grains are smaller, with fewer ducts, sieve tubes, and companion cells, leading to poor transport of inorganic and organic matter [8]. (3) Insufficient starch synthesis in grains: Superior grains exhibit higher sucrose synthase activity at the early GF stage, reaching peak activity earlier [8]. Research has shown that starch synthase and ADP-glucose pyrophosphorylase activities in superior grains are markedly elevated compared to those in inferior grains during the active filling stage [9]. (4) Reduced cell division rate: Grain plumpness at maturity is closely linked to cell proliferation during the GF period [10,11]. Rice cultivars with dense panicles encompass many inadequately developed grains, exhibiting markedly diminished cell division rates and ploidy but extended division periods [12]. The suboptimal grain plumpness in large panicle type rice restricts the yield potential and adversely impacts the efficient utilization of rice nutrients.

Previous studies have shown that the external environment and cultivation conditions provide significant opportunities for regulating GF [13]. To address the issues of insufficient GF and low seed-setting rates in rice, earlier researchers identified adjusting nitrogen fertilizer (N) levels as a key cultivation strategy to enhance GF. Some studies have suggested that increasing N in the later growth stages can delay the degradation of chlorophyll and soluble proteins in leaves, thereby prolonging photosynthesis [14]. Adequate late-stage N increases the nitrogen content of leaves and enhances the activities of superoxide dismutase, peroxidase, and catalase in flag leaves during the filling stage, which aids in boosting the supply of grain polysaccharides, and ultimately increasing grain weight [15]. Additionally, low nitrogen levels combined with moderate soil drought can enhance the activities of α-amylase, β-amylase, and sucrose phosphate synthase (SPS) in rice stems and sheaths, thereby accelerating the hydrolysis of starch and its conversion to sucrose, and facilitating the transport of assimilates to grains [16,17]. While numerous studies have investigated N-mediated yield enhancement in hybrid large-panicle rice, the physiological mechanisms in conventional japonica varieties remain less understood [18]. This study employed two panicle-type conventional japonica cultivars to examine yield formation and GF characteristics under differential N regimes. Our objectives were to (1) elucidate yield component responses to N availability, and (2) reveal the physiological mechanisms underlying yield formation. These findings provide both theoretical and practical significance for achieving high-yield, high-efficiency production in conventional japonica rice systems.

2. Materials and Methods

2.1. Experimental Design

The experiments took place at the Yangzhou University farm (Yangzhou, Jiangsu, China, 32°39′ N, 119°42′ E) during the rice growing seasons of 2022–2023. Soil basal fertility was as follows: organic matter 31.43 g kg⁻¹, total N 1.6 g kg⁻¹, available N 120.3 mg kg⁻¹, available P 37.2 mg kg⁻¹, available K 91.5 mg kg⁻¹, pH 6.5. The rice (Oryza sativa) varieties Yangchan 3501 (large-panicle: spikelets per panicle $> 150$) and Nangeng 46 (medium-panicle: $100<$ spikelets per panicle $< 150$) were utilized (early-maturing late japonica inbred rice with a growth period of 150–155 days). Seeds were sown on May 25 using 25-day-old blanket seedlings. Seedlings were transplanted at the single-heart and triple-leaf phases, with four per insertion point. Hill spacing was 30 cm × 11.7 cm, containing four seedlings each. The experiment employed a split-plot design with nitrogen fertilizer as the main plot factor and cultivar as the subplot factor. The experiment was performed in triplicate for every intervention, and the plot size for every intervention was 15 m² (5 m × 3 m). Three nitrogen treatments were applied: low nitrogen (T1:225 kg ha ⁻¹), medium nitrogen (T2: 270 kg ha ⁻¹), and high nitrogen (T3: 315 kg ha ⁻¹). Ridges covered with plastic film were used to separate sub-districts, ensuring isolated irrigation and drainage. Disease, insect, and weed control followed conventional high-yield rice cultivation protocols. N was introduced in three stages: 35% as basal fertilizer, 35% at tillering initiation, and 30% at panicle initiation utilizing urea (46.4% N) as the nitrogen source. Panicle fertilizer was applied twice to promote flower development and protection. Each experimental area was given a foundational treatment of calcium superphosphate (P₂O₅ content: 12%) at 135 kg P₂O₅ ha ⁻¹ and potassium chloride (K₂O content: 60%) at 135 kg K₂O ha ⁻¹ during the panicle formation phase. The study site was flooded post-transplantation until seven days before maturity. Using chemical methods, intensive control measures were implemented to address insects, diseases, and weeds.

2.2. Observations

2.2.1. Tiller Dynamics

For each plot, a random observation point was selected, and at each point, 10 hills were examined. The number of tillers was recorded at various stages: transplanting, middle tillering (20 days after transplanting), joining, booting, heading, milk ripening (20 days after heading), wax ripening (35 days after heading), and maturity.

2.2.2. Grain-Filling Determination

Panicles with relatively uniform blossoming time and size were selected at the heading (with the blade sheath approximately 5 cm drawn from the top of the panicles) and marked with tags. In each plot, 200 panicles were tagged. From flowering to maturity, 10 tagged panicles were selected every 5 days. Superior and inferior grains were picked, and unfertilized vacant grains were eliminated. Then, the grains were dried, threshed, and weighed. Superior and inferior grains were categorized as follows: The whole spike was divided into upper, middle, and lower portions. If the spike could not be divided equally, the integer part of the average was taken from the upper and lower portions, and the excess was classified as the middle part. Grains directly on the main branch at the top of the panicle (excluding the second grain at the top) were deemed superior grains, while grains on the subordinate branch stalk at the lower panicle (excluding the first grain from the top) were categorized as inferior grains.

2.2.3. Determination of Enzymes

At 15 and 30 days post-flowering, the flag leaves of the primary stem were selected, quickly wrapped in tin foil, and flash-frozen in liquid nitrogen before being transported to the laboratory for the determination of nitrate reductase (NR) and ferredoxin-dependent glutamate synthase (Fd-GOGAT). Simultaneously, the main stem was also wrapped in tin foil, frozen in liquid nitrogen, and brought to the lab to assess the activities of α-amylase and SPS.

Nitrate reductase (NR) activity assay: Leaf tissue (0.1 g) was homogenized under liquid nitrogen using a ball mill (MM400, Retsch, Haan, Germany). The powdered tissue was suspended in 1 mL of extraction buffer (100 mM potassium phosphate buffer, pH 7.5, containing 12% (v/v) 1-propanol and 100 mM KNO₃). The homogenate was centrifuged at 8000× g for 10 min at 4 °C. The reaction mixture, consisting of 10 µL supernatant, 75 µL 100 mM KNO₃, 90 µL distilled water, and 25 µL NADH solution (prepared in 100 mM PBS, pH 7.5), was incubated in a microplate. Absorbance at 340 nm was recorded at t = 1 min (A₁) and t = 6 min (A₂), with enzyme activity calculated as ΔA = A₁ − A₂ using a microplate reader (Infinite M200 Pro, Tecan, Männedorf, Switzerland).

Ferredoxin-dependent glutamate synthase (Fd-GOGAT) activity assay: Tissue homogenates were prepared as described above using extraction buffer (100 mM PBS, pH 7.4, containing 1 mM EDTA and 5 mM β-mercaptoethanol). After centrifugation (10,000× g, 10 min, 4 °C), the assay was initiated by adding 100 µL of reaction mixture (10 mM glutamine and 2 mM α-ketoglutarate in 100 mM Tris-HCl, pH 7.5) and 50 µL of methyl viologen (1 mM) to the 50 µL supernatant. Following 5 min incubation at 30 °C, 50 µL sodium bicarbonate (100 mM) and sodium dithionite (50 mM) were added. Controls received distilled water instead of supernatant. After 30 min at 30 °C, reactions were terminated at 95 °C (5 min). Absorbance was measured at 450 nm after centrifugation (5 min, 4 °C), with activity expressed as ΔA = A_meas − A_kontrol [19].

α-Amylase activity assay: Homogenates were prepared in distilled water and centrifuged (5000 rpm, 25 min, 25 °C). Diluted supernatant (75 µL, 1:10) was incubated with 75 µL substrate solution (1% soluble starch in 100 mM citrate buffer, pH 4.8) at 40 °C for 5 min. The reaction was terminated by adding 150 µL DNS reagent (3,5-dinitrosalicylic acid in 1 M NaOH with potassium sodium tartrate) and heating at 95 °C for 5 min. Absorbance at 450 nm was measured against appropriate controls.

Sucrose-phosphate synthase (SPS) Activity Assay: The supernatant obtained after centrifugation (8000× g, 10 min, 4 °C) was incubated with 45 µL assay buffer (50 mM Tris-HCl, pH 7.5, containing 5 mM MgCl₂, 5 mM fructose-6-phosphate, and 5 mM UDP-glucose) at 25 °C for 10 min. Reactions were terminated with 15 µL of 2 M NaOH and heated at 95 °C (10 min). After cooling, 210 µL of 30% HCl and 60 µL of 0.1% resorcinol were added, followed by incubation at 95 °C (30 min). The absorbance at 480 nm was measured against blank and standard controls.

All enzymatic activities were calculated as the difference in absorbance between experimental and control measurements (ΔA = A_meas − A_kontrol). All assay reagents were sourced from Suzhou Michy Biomedical Technology Co., Ltd. (Suzhou, China).

2.2.4. Grain Yield

At the maturity stage, 50 points were sampled in each plot to calculate the effective panicle number. Based on the average panicle number, 10 points were selected for seed examination to investigate the quantity of spikelets per panicle, the proportion of filled grains, and the mass of 1000 grains. One hundred points were manually collected, and the grain yield was weighed and modified to a moisture content of 135 kg ⁻¹.

2.3. Data Analysis

Data analysis adhered to the approach outlined by Gong Jinlong et al. [20], employing the Richards equation to model the GF process and calculate the corresponding filling characteristic parameters. The Richards equation used for GF growth analysis is presented below:

W = A/(1 + Be^−Kt)^1/N

(1)

where W is the weight of a kernel (mg); A is the final weight of a kernel; t is the time after flowering (d); and B is the initial parameter, K is the growth rate parameter, and N are equation parameters. R² represents the coefficient of determination. The GF rate G (mg kernel⁻¹ d⁻¹) was obtained by taking the derivative of Equation (1):

G = AKBe^−Kt/N(1 + Be^-Kt)^{(N + 1)/N}

(2)

Using these parameters, the following calculations were made:

R₀ (initial grain-filling rate) = K/N

(3)

W_max (weight of a kernel at the time of maximum GF rate) = A(1 + N) ^−1/N

(4)

GR_mean (mean GF rate within the whole filling stages) = AK/2(N + 2)

(5)

D (the whole active grain-filling stage) = 2(N + 2)/N

(6)

$$\begin{gathered} \mathrm{The} \mathrm{GF}’\mathrm{s} \mathrm{early}, \mathrm{middle}, \mathrm{and} \mathrm{late} \mathrm{stages} \mathrm{were} \mathrm{categorized} \mathrm{as} 0–{\mathrm{T}}_1, {\mathrm{T}}_1–{\mathrm{T}}_2, \mathrm{and} {\mathrm{T}}_2–{\mathrm{T}}_3 \\ {\mathrm{T}}_1 = -\ln [({\mathrm{N}}^2 + 3 \mathrm{N} + \mathrm{N}{({\mathrm{N}}^2 + 6 \mathrm{N} + 5)}^{1/2})/2 \mathrm{B}]/\mathrm{K} \end{gathered}$$

(7)

T₂ = −ln[(N² + 3 N − N(N² + 6 N + 5)^1/2)/2 B]/K

(8)

T₃ = T₉₉ (effective GF duration) = −ln{[(100/99 A)^N − 1]/B}/K

(9)

Based on these stages and the accumulation of GF material, the mean grain-filling rate (MGR) for the early, middle, and late stages was calculated, along with the ratio of GF material contributing to the A value (RGC) at each stage.

Data entry, processing, and mapping were conducted using Microsoft Excel 2003. The means were compared based on the least significant difference (LSD) test at the 0.05 probability level by using DPS 7.05. The grouting process equations was performed using DPS software. Due to consistent trends observed over two years, GF dynamics were primarily described using the data from 2023.

3. Results

3.1. Yield

As the dosage of N increased, the yield of both rice panicle types exhibited an upward trend, with marked differences observed between T1 and T3 treatments (

Table 1. Impacts of N on rice yield of different panicle types.

Year	Variety	Treatment	No. of Panicles (104 ha−1)	Spikelets Per Panicle	Total Spikelets (104 ha−1)	Seed-Setting Rate (%)	1000-Grain Weight (g)	Harvested Yield (t ha−1)
2023	Nangeng 46	T1	322.05 b	111.68 b	35,967.19 c	93.09 a	27.92 a	9.20 b
		T2	326.10 b	119.77 a	39,056.26 b	93.06 a	27.82 a	9.66 a
		T3	344.78 a	120.00 a	41,373.95 a	92.17 a	27.75 a	9.91 a
	CV (%)		3.66	4.04	6.99	0.57	0.31	3.74
	Yangchan 3501	T1	237.17 b	171.05 c	40,567.07 c	88.46 a	27.81 a	9.67 c
		T2	248.06 a	177.14 b	43,939.22 b	87.79 ab	27.33 b	10.22 b
		T3	253.50 a	187.68 a	47,577.45 a	85.87 b	27.48 ab	10.91 a
	CV (%)		3.38	4.71	7.97	1.54	0.87	6.02
2022	Nangeng 46	T1	320.25 b	115.14 b	36,873.59 b	88.59 a	27.81 a	8.70 b
		T2	329.03 b	123.71 a	40,702.86 a	86.56 a	27.34 b	9.29 a
		T3	351.45 a	120.08 ab	42,202.12 a	85.67 a	27.33 b	9.57 a
	CV (%)		4.82	3.59	6.87	1.73	1.00	4.86
	Yangchan 3501	T1	241.85 a	166.71 b	40,319.23 b	86.46 a	27.58 a	9.25 b
		T2	253.43 a	170.87 b	43,301.46 ab	83.78 ab	27.55 a	9.61 ab
		T3	256.58 a	179.13 a	45,959.00 a	82.37 b	27.48 a	10.16 a
	CV (%)		3.09	3.67	6.54	2.47	0.18	4.70

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹; CV: Coefficient of variation. Values followed by diverse lowercase letters are markedly different within the column at the p = 0.05 level.

). Based on the average data from the two-year experiment, Nangeng 46 showed an increase of 0.26 t ha⁻¹ and 0.79 t ha⁻¹ under T3 treatment compared to T2 and T1, respectively, while Yangchan 3501 increased by 1.07 t ha⁻¹ and 0.61 t ha⁻¹, respectively. The variation coefficients for panicle number and spikelets per panicle in Nangeng 46 were 3.66–4.82% and 3.59–4.04%, respectively, whereas for Yangchan 3501, they were 0.09–3.38% and 3.67–4.71%, respectively. Compared to T2 and T1, the total spikelets in the Nangeng 46 population under T3 treatment increased by 14.72% and 4.78%, respectively, while Yangchan 3501 showed increases of 15.65% and 7.22%, respectively. The seed-setting rate and 1000-grain weight for both rice types decreased with higher N application, with minimal variation between different N treatments. In summary, the increase in N markedly enhanced panicle quantity and spikelets per panicle, which were the primary contributors for the higher rice yield. Under identical N conditions, Yangchan 3501 yielded higher amounts than Nangeng 46, with the most significant yield difference between the two varieties observed under T3. The primary reason for Yangchan 3501’s higher yield was its markedly greater number of spikelets per panicle compared to Nangeng 46.

3.2. Differences in Yield Formation

3.2.1. Tillering Dynamics of Rice Populations

N markedly impacted tiller numbers at both the middle tillering and jointing stages for the two rice types (

Table 2. Impacts of N application on tillering dynamics of rice populations with different panicle types.

Year	Variety	Treatment	Transplanting (104 ha−1)	Middle Tillering Stage (104 ha−1)	Joining (104 ha−1)	Booting Stage (104 ha−1)	Heading Stage (104 ha−1)	Milky Stage (104 ha−1)	Waxy Stage (104 ha−1)	Ratio of Productive Tillers to Total Tillers (%)
2023	Nangeng 46	T1	112.73 a	338.93 c	415.80 c	385.28 c	364.20 c	352.35 b	331.95 b	77.45 a
		T2	112.95 a	378.23 b	451.50 b	397.19 b	376.35 b	362.55 ab	344.93 ab	72.24 ab
		T3	112.65 a	424.88 a	513.30 a	416.05 a	398.70 a	384.83 a	363.13 a	67.18 b
	CV (%)		0.14	11.30	10.72	3.88	4.61	4.53	4.52	7.11
	Yangchan 3501	T1	112.05 a	248.85 c	330.00 c	293.03 b	275.10 c	258.68 c	245.18 b	71.87 a
		T2	112.13 a	265.58 b	353.33 b	306.30 ab	288.90 b	275.63 b	266.25 a	70.21 a
		T3	112.43 a	276.60 a	373.13 a	319.80 a	303.08 a	285.30 a	263.25 a	67.94 b
	CV (%)		0.18	5.30	6.13	4.37	4.84	4.93	4.42	2.82
2022	Nangeng 46	T1	112.80 a	354.53 c	425.40 c	391.43 b	373.43 b	348.45 c	334.28 b	75.29 a
		T2	112.88 a	401.10 b	461.18 b	409.50 ab	384.45 b	363.75 b	347.70 ab	71.36 ab
		T3	110.33 a	438.38 a	528.76 a	436.05 a	402.68 a	380.93 a	365.70 a	66.47 b
	CV (%)		1.30	10.56	11.13	5.44	3.82	4.46	4.52	6.22
	Yangchan 3501	T1	111.83 a	254.63 b	331.43 b	297.90 c	282.68 c	266.09 c	248.93 b	72.97 a
		T2	111.68 a	272.85 a	353.33 a	312.68 b	296.25 b	280.43 b	266.40 a	71.74 a
		T3	112.13 a	274.58 a	368.40 a	325.88 a	309.38 a	293.10 a	274.28 a	69.65 a
	CV (%)		0.20	4.13	5.30	4.48	4.51	4.83	4.93	2.35

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹; CV: Coefficient of variation. Values followed by diverse lowercase letters are markedly different within the column at the p = 0.05 level.

). The coefficients of variation at the middle tillering stage were 10.56–11.30% for Nangeng 46 and 4.13–5.30% for Yangchan 3501, and at the jointing stage, they were 10.72–11.13% for Nangeng 46 and 5.30–6.13% for Yangchan 3501. Tiller numbers increased with higher N application during the middle tillering and jointing stages, with marked differences observed. Nangeng 46 exhibited higher tiller numbers compared to Yangchan 3501 across all three N treatments, and this difference persisted from the middle tillering stage to maturity. The panicle rate of Nangeng 46 under T1 and T2 treatments was markedly higher than that of Yangchan 3501. Thus, increasing N application rates enhanced tiller formation and the number of panicles at maturity, although it also reduced the panicle rate. The influence of N on the tiller number and panicle rate was more pronounced in medium-panicle-type rice compared to large-panicle-type rice.

3.2.2. Characteristics of Rice Panicles

Analyzing the characteristics of rice panicles at various N levels revealed that the number of primary branch stems increased with higher nitrogen application (

Table 3. Impacts of N on panicle composition of rice with different panicle types.

Year	Variety	Treatment	Primary Branches				Secondary Branches
			NB	GPB	TG	SSR (%)	NB	GPB	TG	SSR (%)
2023	Nangeng 46	T1	12.24 a	5.27 b	64.51 b	96.30 a	18.64 a	2.53 a	47.18 a	88.71 a
		T2	12.62 a	5.56 a	70.17 a	96.38 a	19.26 a	2.58 a	49.60 a	88.35 a
		T3	12.36 a	5.58 a	68.91 a	96.60 a	20.11 a	2.54 a	51.10 a	86.18 b
	CV (%)		1.57	3.14	4.38	0.16	3.81	0.93	4.01	1.56
	Yangchan 3501	T1	15.66 c	5.23 a	81.87 c	90.88 a	29.38 b	3.04 a	89.18 b	86.23 a
		T2	16.33 b	5.25 a	85.65 b	91.08 a	29.94 b	3.06 a	91.48 b	84.72 a
		T3	16.79 a	5.33 a	89.49 a	91.45 a	32.30 a	3.04 a	98.19 a	80.77 b
	CV (%)		3.51	1.02	4.45	0.32	5.06	0.34	5.04	3.36
2022	Nangeng 46	T1	12.29 a	5.22 b	64.15 b	92.37 a	19.35 a	2.64 a	50.99 a	83.82 a
		T2	12.70 a	5.60 a	71.09 a	92.46 a	19.70 a	2.67 a	52.61 a	78.58 b
		T3	12.41 a	5.56 a	68.97 a	92.61 a	19.33 a	2.65 a	51.11 a	76.28 b
	CV (%)		1.67	3.82	5.22	0.13	1.09	0.68	1.76	4.86
	Yangchan 3501	T1	15.37 c	5.25 a	80.62 c	87.42 a	29.33 a	2.94 a	86.09 b	85.56 a
		T2	16.04 b	5.20 a	83.30 b	87.68 a	29.48 a	2.97 a	87.56 ab	80.07 ab
		T3	16.76 a	5.28 a	88.38 a	87.65 a	29.95 a	3.03 a	90.74 a	77.21 b
	CV (%)		4.31	0.77	4.69	0.16	1.08	1.61	2.70	5.24

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹; NB: No. of branches; GPB: grains per branch; TG: total grain; SSR: seed-setting rate; CV: Coefficient of variation. Values followed by diverse lowercase letters are markedly different within the column at the p = 0.05 level.

). Specifically, under T3 treatment, Nangeng 46 and Yangchan 3501 exhibited a 4.40–74.82 and 3.84–5.08 increase in the number of primary branch stems compared to T1, respectively. In contrast, the number of secondary branch stems in Nangeng 46 did not show marked variation across different nitrogen treatments. For Yangchan 3501, however, the number of secondary branch stems was markedly higher under T3 treatment compared to T1. The number of main branches, seed-setting rate of the main branches, grains per primary branch, grains per secondary branch, and secondary branches of Nangeng 46 did not exhibit marked differences under varying nitrogen levels. Conversely, in Yangchan 3501, the number of primary branches increased markedly with higher N application, while the seed-setting rate of secondary branches decreased. These results indicate that N fertilizer levels have differential effects on panicle characteristics in different rice types, with large-panicle rice being more influenced by N levels. Yangchan 3501 demonstrated clear advantages over Nangeng 46 regarding the quantity of primary branches, secondary branches, and grains per secondary branch, despite having a lower seed-setting rate across different branches.

3.2.3. Dynamic Characteristics of GF

The amount of N had a marked impact on the dynamic increase in grain weight (

Table 4. Characteristics of rice GF under different N dosages.

Items	Parameters	Nangeng 46						Yangchan 3501
		Superior Grain			Inferior Grain			Superior Grain			Inferior Grain
		T1	T2	T3	T1	T2	T3	T1	T2	T3	T1	T2	T3
Parameters of the Richards equation	A	21.3199	21.2374	21.0886	18.9672	20.0804	19.6029	21.0141	20.9260	21.1011	17.3411	18.4149	18.0013
	B	0.0469	1.3494	3.2427	1477.5000	654.9941	1722.5000	0.8057	1.7759	1.8721	482.3312	159.2927	399.8321
	K	0.1188	0.1434	0.1643	0.2033	0.1921	0.2065	0.1305	0.1487	0.1536	0.1796	0.1614	0.1760
	N1	0.0100	0.2004	0.3560	1.7188	1.6703	1.7890	0.1385	0.2413	0.2436	1.3456	1.1831	1.3322
	R2	0.9975	0.9978	0.9970	0.9993	0.9988	0.9992	0.9972	0.9976	0.9965	0.9986	0.9978	0.9989
Grain-filling parameters	R0	11.88	0.72	0.46	0.12	0.12	0.12	0.94	0.62	0.63	0.13	0.14	0.13
	Tmax	13.01	13.30	13.45	33.23	31.09	33.27	13.49	13.42	13.28	32.75	30.38	32.41
	Wmax	7.88	8.54	8.96	10.60	11.15	11.05	8.24	8.54	8.62	9.20	9.52	9.53
	GRmax	0.93	1.02	1.09	0.79	0.80	0.82	0.94	1.02	1.06	0.70	0.70	0.72
	I	36.97	40.19	42.51	55.88	55.54	56.36	39.20	40.83	40.86	53.07	51.69	52.96
	GRmean	0.63	0.69	0.74	0.52	0.53	0.53	0.64	0.69	0.72	0.47	0.47	0.48
	D	33.84	30.69	28.68	36.58	38.21	36.70	32.77	30.15	29.21	37.26	39.44	37.87
	T99	51.73	45.37	41.43	55.82	54.99	55.50	48.74	44.35	43.22	58.33	58.84	58.51

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹; A: The final weight of a kernel; B: Initial parameter; K: Growth rate parameter; N₁: Shape parameter; R²: The determination coefficient of the equation; R₀: Initial grain-filling; GR_max: Maximum grain-filling rate; T_max: The time to reach the maximum grain-filling rate; W_max: Weight of a kernel at the time of maximum grain-filling rate; I: The percentage of W_max to A; GR_mean: Mean grain-filling rate; D: Active grain-filling period; T₉₉: Effective grain-filling time.

). This experiment employed the Richards equation to model the GF process for superior and inferior rice grains under varying N levels. Using days after flowering (t) as the independent variable and grain weight (W) as the dependent variable, the parameters estimated included the final weight of a kernel (A), initial parameter (B), growth rate parameter (K), shape parameter (N), and the determination coefficient (R²). As illustrated in Table 4, the fitting coefficients for the GF process of superior and inferior grains in both rice varieties under different nitrogen conditions exceeded 0.99, suggesting that the Richards equation effectively described the GF process for both rice varieties. Figure 1 displays the Richards simulation curve for the increase in single-grain mass. The final weight of a kernel (A) for Nangeng 46’s superior grain exhibited minimal change with increased N application, while the inferior grain under T2 treatment was 0.4775 mg and 1.1131 mg heavier than under T3 and T1, respectively. Similarly, Yangchan 3501’s superior and inferior grains demonstrated that inferior grains under T2 treatment were 0.4136 mg and 1.0738 mg heavier than those under T3 and T1, respectively. The final kernel weight of Nangeng 46 was higher than that of Yangchan 3501, regardless of being superior or inferior grain, likely due to the higher grain number per panicle in Yangchan 3501. For both rice types, the shape parameter (N₁) for superior grains was less than 1, with values T3 > T2 > T1, indicating a left-skewed single-grain quality growth curve. For inferior grains, the N₁ was greater than 1, with the smallest value for T2, indicating a right-skewed growth curve. These findings suggest that the GF of superior and inferior grains in both rice types is asynchronous: superior grains rapidly increased in weight after flowering, whereas inferior grains experienced a slower initial weight increase before accelerating after significant superior grain weight gain. Increasing N altered the dynamics of GF but did not markedly change the filling pattern between superior and inferior grains. Additionally, it was observed that the mismatch in GF among superior and inferior grains was more pronounced in large-panicle rice.

Figure 2 illustrates the Richards curve simulation of the GF rate for superior and inferior grains under varying N levels. The impact of N on the initial grain-filling rate (R₀) was comparable across the different panicle types of rice. However, the R₀ of inferior grains was less affected by N compared to superior grains. The maximum grain-filling rate (GR_max), mean grain-filling rate (GR_mean), and kernel weight at the time of maximum grain-filling rate (W_max) for both rice types increased with higher N application. Additionally, the time to reach the maximum grain-filling rate (T_max) and the effective grain-filling duration (T₉₉) were prolonged (Table 4). Both the GR_max and GR_mean of inferior grains showed an increase. Among the varieties, Nangeng 46 exhibited a higher GR_max and GR_mean than Yangchan 3501, regardless of whether the grain was superior or inferior. The time to reach T_max, W_max, and the percentage of W_max relative to the final kernel weight (I) in Nangeng 46 were all greater than those in Yangchan 3501. In contrast, the active grain-filling period (D) and T₉₉ were shorter for Nangeng 46 compared to Yangchan 3501.

According to the two inflection points of the filling rate curve, the GF process was segmented into three stages: early, middle, and late (

Table 5. Characteristics of rice GF stages under different N dosage.

Items	Parameters	Nangeng 46						Yangchan 3501
		Superior Grain			Inferior Grain			Superior Grain			Inferior Grain
		T1	T2	T3	T1	T2	T3	T1	T2	T3	T1	T2	T3
Early stage of grain filling	D	4.87	6.00	6.71	25.84	23.32	25.91	5.66	6.27	6.35	24.89	21.90	24.41
	GF	1.59	2.23	2.71	5.38	5.61	5.68	2.00	2.33	2.35	4.31	4.26	4.44
	MGR	0.33	0.37	0.40	0.21	0.24	0.22	0.35	0.37	0.37	0.17	0.19	0.18
	RGC	7.46	10.50	12.83	28.35	27.92	28.97	9.53	11.12	11.16	24.83	23.15	24.69
Middle stage of grain filling	D	16.28	14.60	13.48	14.79	15.53	14.72	15.66	14.30	13.86	15.72	16.96	16.00
	GF	13.00	12.90	12.72	10.33	10.98	10.62	12.79	12.69	12.80	9.74	10.48	10.12
	MGR	0.80	0.88	0.94	0.70	0.71	0.72	0.82	0.89	0.92	0.62	0.62	0.63
	RGC	60.96	60.74	60.33	54.48	54.70	54.17	60.85	60.65	60.64	56.16	56.90	56.22
Late stage of grain filling	D	30.58	24.77	21.25	15.19	16.14	14.87	27.41	23.78	23.01	17.72	19.99	18.10
	GF	6.52	5.90	5.45	3.07	3.29	3.11	6.01	5.70	5.74	3.12	3.49	3.26
	MGR	0.21	0.24	0.26	0.20	0.20	0.21	0.22	0.24	0.25	0.18	0.17	0.18
	RGC	30.58	27.76	25.84	16.17	16.39	15.86	28.62	27.23	27.20	18.01	18.95	18.08

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹; D: Active grain-filling period; GF: Grain filling; MGR: Mean grain-filling rate; RGC: The ratio of the grain-filling that contributed to the final grain weight.

). The findings indicated that N increased the early filling days of superior grains, enhancing the GF, MGR, and the ratio of grain-filling contribution to the final grain weight (RGC), while having minimal impact on the middle and late stages of superior grains. Additionally, N extended the filling duration in the middle and late stages, increased GF, and improved RGC in the middle and late stages of inferior grains in Yangchan 3501, but had a negligible effect on the inferior grains of Nangeng 46.

3.3. Related Enzyme Activity

3.3.1. Nitrogen Metabolizing Enzyme Activity in Leaves

The Fd-GOGAT and NR in the leaves of the two panicle types of rice at 15 days after flowering followed the order T1 < T2 < T3 under different N treatments, with marked differences among treatments (

Table 6. The activity of nitrogen metabolism enzymes in rice leaves at the GF stage under different N treatments.

Variety	Treatment	15 Days After Flowering		30 Days After Flowering
		Fd-GOGAT	NR	Fd-GOGAT	NR
Nangeng 46	T1	46.64 b	425.98 c	37.12 b	128.92 b
	T2	47.52 b	478.41 b	42.58 ab	279.54 a
	T3	54.14 a	518.36 a	48.23 a	323.10 a
CV (%)		8.29	9.77	13.03	41.78
Yangchan 3501	T1	46.98 c	467.26 c	40.75 c	156.71 c
	T2	52.73 b	502.42 b	45.87 b	314.46 b
	T3	62.12 a	546.14 a	54.85 a	364.44 a
CV (%)		14.17	7.82	15.14	38.93

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹. Fd-GOGAT: ferredoxin-dependent glutamate synthase; NR: Nitrate reductase; CV: Coefficient of variation. Values followed by diverse lowercase letters are markedly different within the column at the p = 0.05 level.

). As growth progressed, the activity of both enzymes decreased to varying degrees at 30 days after anthesis, with Fd-GOGAT showing a smaller decrease. The enzyme activities also maintained the order T1 < T2 < T3 across different nitrogen treatments. At 15 days post-flowering, the activities of Fd-GOGAT and NR in Yangchan 3501 were 0.34–7.98 and 24.01–41.28 higher, respectively, than those in Nangeng 46 under various nitrogen treatments. At 30 days post-flowering, the activities of Fd-GOGAT and NR in Yangchan 3501 were 3.29–6.63 and 27.8–41.35 higher, respectively, than those in Nangeng 46.

3.3.2. α-Amylase and SPS Activity in the Stem

The α-amylase and SPS activity in the stalks of the two panicle types at 15 days after flowering followed the order T1 > T2 > T3, with marked differences among treatments (

Table 7. Activity of α-amylase and SPS enzymes in stems at different N dosages at filling stage.

Variety	Treatment	15 Days After Flowering		30 Days After Flowering
		α-Amylase	SPS	α-Amylase	SPS
Nangeng 46	T1	16.22 a	621.08 a	17.21 a	406.31 a
	T2	15.30 b	544.77 b	16.60 a	361.87 b
	T3	14.61 c	469.96 c	15.34 b	310.13 c
CV (%)		5.25	13.86	5.84	13.39
Yangchan 3501	T1	16.74 a	542.03 a	17.25 a	354.48 a
	T2	16.30 b	525.96 b	16.68 b	328.25 a
	T3	15.56 c	487.68 c	15.89 c	279.19 b
CV (%)		3.68	5.38	4.1	11.92

Note: T1: 225 kg ha⁻¹; T2: 270 kg ha⁻¹; T3: 315 kg ha⁻¹. SPS: Sucrose phosphate synthase; CV: Coefficient of variation. Values followed by diverse lowercase letters are markedly different within the column at the p = 0.05 level.

). As growth progressed, α-amylase activity increased at 30 days after anthesis, while SPS activity exhibited a decreasing trend. The enzyme activity consistently followed the order T1 > T2 > T3 across different nitrogen treatments. The α-amylase activity of Yangchan 3501 was comparable to that of Nangeng 46 under various N treatments. At 15 days post-flowering, the SPS activity of Yangchan 3501 was 79.05 and 18.81 lower than that of Nangeng 46 under T1 and T2 treatments, respectively. At 30 days post-flowering, the SPS activity of Yangchan 3501 was 51.83, 33.62, and 30.94 lower than that of Nangeng 46 under T1, T2, and T3 treatments, respectively.

4. Discussion

Achieving high rice yield hinges on increasing the seed-setting rate and 1000-grain weight, alongside boosting the number of spikelets per unit area [21]. N is crucial in the high-yield cultivation of rice. While N markedly enhances the yield of small and medium-panicle-type rice, large-panicle-type rice can still achieve high yields with a reduced amount of N [22]. In this study, tiller quantity for both panicle types increased markedly with higher N application from the middle tillering stage. High N levels (T3: 315 kg ha⁻¹) increased tiller mortality post-jointing, likely due to competition for assimilates rather than urea toxicity. This reduction in panicle rate underscores the need for balanced N management to optimize tiller productivity. Yangchan 3501 showed fewer tillers and a lower panicle percentage across different N treatments compared to Nangeng 46. This suggests that increasing N input is advantageous for obtaining a higher panicle number and boosting the yield of medium panicle type rice [23]. No significant difference in the panicle number at maturity was observed between T2 and T3 treatments for Yangchan 3501, exhibiting coefficients of variation of 3.09–3.38% across nitrogen treatments. For large-panicle cultivars (>150 grains/panicle), reduced nitrogen application during early and mid-growth phases coupled with optimized transplanting density and seedling management can effectively stabilize panicle number and enhance nitrogen use efficiency [24]. An analysis of panicle characteristics revealed that increased nitrogen application rates promoted the quantity of primary branches and grains per primary branch. Yangchan 3501 achieved higher total grains per panicle with markedly more primary and secondary branches and grains per branch compared to Nangeng 46, thus achieving a higher yield, despite having a lower seed-setting rate under all three N conditions. Previous studies have indicated that sufficient panicle fertilizer is beneficial for secondary spikelet differentiation but may decrease seed-setting rate and 1000-grain weight, necessitating attention to flower protection and grain fertilizers for large-panicle-type varieties [25]. In this experiment, the same amount of flower-promoting and flower-protecting fertilizers were applied to both panicle types, highlighting the need for further research on the effect of the ratio of these fertilizers on yield and the underlying mechanisms.

The grain plumpness of rice is influenced by its position on the panicle, with GF occurring more rapidly in the upper part compared to the lower part [26]. It was observed that, with increased N application, the GR_max and GR_mean of Nangeng 46 surpassed those of Yangchan 3501 for both superior and inferior grains. This suggests that enhancing the GF rate is crucial for increasing grain weight [27,28]. N application improved GR_max, GR_mean, and the kernel weight at the time of the maximum grain-filling rate (W_max) for superior grains, as well as extended the time to reach the maximum grain-filling rate (T_max) and effective grain-filling duration (T₉₉). Additionally, N promoted increases in GR_max and GR_mean for inferior grains. Therefore, ensuring adequate nitrogen supply during the later growth stages is advantageous for increasing grain weight. This study suggested that both Nangeng 46 and Yangchan 3501 exhibited asynchronous filling patterns, as evidenced by the shape parameter (N₁) in the Richards model and the time interval for reaching the maximum filling rate among superior and inferior grains [29]. Thus, increasing N application altered the GF dynamics without markedly changing the filling pattern among superior and inferior grains.

The essence of GF and grain weight increase in rice is the process of starch synthesis and accumulation [10]. Enhancing the efficient transport of carbohydrates from the leaves and stem sheaths to grains is essential for achieving high rice yields [30,31]. Studies have indicated a strong positive correlation between leaf NR activity and yield during the middle and late growth stages of rice [32]. The Fd-GOGAT and NR activities in Yangchan 3501 were higher than those in Nangeng 46 at both 15 and 30 days after flowering, following the order T1 < T2 < T3 under different N treatments. It is hypothesized that a higher level of nitrogen metabolism maintains the greenness of leaves in the late growth stage, thereby extending the photosynthesis period [33], which benefits the filling time of inferior grains in large panicle type rice during the middle and later stages. Consequently, GF and RGC were enhanced. α-amylase and SPS are critical for distributing assimilates to grains [34]. The experiment revealed that the activity of α-amylase and SPS in the stems of both panicle types were ordered T1 > T2 > T3, indicating that high nitrogen conditions were not conducive to starch hydrolysis and sucrose transformation in stems, thereby hindering the transport of assimilates from the stem and sheath to grains [17,35]. However, some research suggests that carbohydrate supply should not be the primary issue, as sufficient sucrose exists at the initial grain-filling stage. The low activity of key carbon metabolism enzymes may contribute to poor GF [5]. Therefore, further studies on N’s effects on rice carbon metabolism are necessary.

5. Conclusions

Nitrogen application significantly increased grain yields in both Yangchan 3501 and Nanjing 46, though their yield formation mechanisms differed: Yangchan 3501 achieved higher yields primarily through significantly increased spikelets per panicle, while Nanjing 46 relied mainly on substantially enhanced panicle numbers. Under high nitrogen conditions (315 kg ha⁻¹), the activity of key nitrogen metabolic enzymes (NADH-GOGAT, Fd-GOGAT, and NR) in leaves were elevated, accompanied by increased maximum grain-filling rate (GR_max) and mean grain-filling rate (GR_mean). Notably, nitrogen supplementation prolonged the mid-to-late grain-filling period of inferior spikelets in Yangchan 3501, improving both the grain-filling capacity (GF) and relative contribution rate (RGC) during these stages. This physiological adaptation explains why Yangchan 3501 maintained a normal seed-setting rate and 1000-grain weight despite possessing significantly greater sink capacity than Nanjing 46. However, further research is needed to determine whether these panicle-type-specific high-yield fertilization strategies are applicable to direct-seeded rice systems, and whether fertilization timing requires optimization for different panicle-type cultivars.