Delaying Application and Reducing the N Rate Enhances Grain Yield and N Use Efficiency in No-Tillage, Direct-Seeded Hybrid Rice

Abstract

The no-tillage, direct-seeded (NTDS) rice cropping system has attracted considerable attention because of its social, economic, and environmental benefits. However, very limited research has been conducted on optimizing nitrogen (N) management practices to enhance grain yield and N use efficiency (NUE) of rice grown in NTDS. An N fertilization field experiment with two rice hybrids was conducted in 2020 and 2021; the experiment consisted of three N rates (180, 153, and 0 kg N ha⁻¹; N₁₈₀, N₁₅₃, and N₀, respectively) and two N-application ratios split among the basal, seedling (three-leaf stage), mid-tillering, and panicle initiation stages (50%, 0%, 30%, and 20% and 0%, 30%, 40%, and 30%; R₁ and R₂, respectively). Although the N rate was 15% lower in the N₁₅₃ treatment than in the N₁₈₀ treatment, grain yield for N₁₅₃ was equal to or slightly higher than that for N₁₈₀. N₁₅₃ had a higher agronomic efficiency of applied N (AE_N), partial factor productivity of applied N (PFP_N), and recovery efficiency of applied N (RE_N) compared to N₁₈₀ by 10.1–24.7%, 15.0–20.1%, and 1.8–12.6%, respectively. Grain yield, AE_N, PFP_N, and RE_N in R₂ were higher than those in R₁ by 0.8–4.6%, 2.4–19.7%, 0.7–4.6%, and 3.5–30.0%, respectively. The increase in grain yield was due to improvement in the sink size that resulted from larger panicles, higher biomass production at maturity, which was partially attributable to increased biomass accumulation after heading, and a higher harvest index. Our results suggest that it is feasible to simultaneously improve grain yield and NUE in hybrid rice under NTDS through delayed and reduced N application rate, and current breeding programs need to target large panicle size as a primary objective for NTDS rice.

1. Introduction

Rice is one of the most important food crops in the world, because it supports a large percentage of the global population, and rice yield must increase by 116 million tons annually by 2035 to meet the growing demand for food due to projected population growth and economic development [1]. China, the largest rice producer and consumer in the world, will need to produce about 20% more rice by 2030 to meet domestic needs [2]. However, rice grain yields have shown declining or stagnant trends in most major rice-producing provinces of China, and the annual growth rate averaged −0.3% from 1998 to 2006 [3]. In order to break the stagnation in rice yield and meet the food requirement of an increasing population, great efforts should be made to develop new rice cultivars with higher yield potential and better abiotic or biotic stress resistance to enhance the average rice yield of farmers’ fields [4]. As of 2021, 135 rice cultivars with high yield potential and good rice quality have been approved as super rice by the Ministry of Agriculture of China [5].

Rice yields depend upon not only the cultivar characteristics but also the crop management practices employed [6]. Crop breeding and crop management have played equal roles in increasing the production of major grain crops in the past in worldwide [7], but at present, the relative contributions of crop improvement and crop management to rice yield increase has shifted to 30% and 70%, respectively [8]. In China, conventional tillage (by ploughing and harrowing) and manually transplanted rice, which demands a large amount of energy and labor input, is still the dominant technology for land preparation and rice establishment, and accounts for 70% of the total rice planting area. With the rapid development of the social economy and urbanization in China, labor availability is limited because more and more young farmers are leaving the countryside for jobs in the cities, leaving the older farmers behind [9]. To sustain rice production and productivity into the foreseeable future, alternative rice production systems are urgently required. The combination of no-tillage and direct seeding (NTDS), which greatly reduces the inputs required for both land preparation and rice establishment operations, is an alternative to conventional tillage and transplanting (CTTP) because it requires less labor and improves rice yield if the crop is properly managed [10,11,12].

Each crop cultivation method requires a series of supporting agronomic practices to achieve high rice yields [13]. Since the concept of NTDS in China is new, there are relatively few insights into N dynamics and fertilizer N use. Nitrogen is the most important yield-limiting nutrient for rice, and the most difficult to manage. In addition, N has several transformations and loss mechanisms, and the efficient accumulation and utilization of N is dependent on agronomic, genetic, and environmental factors. At the same time, to maximize rice grain yield, most farmers apply more fertilizer than the minimum amount required for maximum rice growth; in particular, excessive use of chemical N fertilizer has been one of the major problems confronting rice production in China [7,14]. Moreover, most farmers usually apply N in two split dressings (as basal and topdressing) within the first two weeks of the rice-growing season. Because of the improper application of N (i.e., the higher application rate and at the wrong time) and the methods used for N application [15], only 20–30% of the N is taken up by the crop plants and a large proportion is lost to the environment [16]. This results not only in waste of the applied N and increased production costs but also a host of environmental problems, such as soil acidification, groundwater pollution, and river and lake water eutrophication [14,17]. Similar trends to those occurring in China have also been documented in other Asian countries [18].

NTDS combines no-tillage with direct seeding, and both crop management practices can cause accumulation of rice roots and soil N in the surface soil layer compared with CTTP [19,20]. No-till/direct seeded rice has more roots distributed in the soil layer coincident with the higher N content, indicating that more N fertilizer may not necessarily be required to achieve high grain yields in the NTDS rice cropping system. However, previous studies have demonstrated that no-tillage rice requires extra time to take up basal N, resulting in an increase in N loss and a consequent increase in the N fertilizer requirement [21,22]. In China, two split applications of N at mid-tillering and panicle initiation are recommended for no-tillage rice, and reducing the N application rate from 150 to 120 kg ha⁻¹ could simultaneously improve grain yield and NUE in no-tillage rice production [23]. Recently, Zhang et al. [24] reported that loss of N due to runoff was higher in direct-seeded rice than in transplanted rice, and reducing the rate of N applied at sowing and also delaying the first N application significantly increased grain yield, N uptake, and NUE of dry directly seeded rice compared with applying a higher rate of N at sowing [25]. In the directly seeded rice system, there is a long period of growth from sowing to the three-leaf stage, during which the rice seedlings are very small; therefore, the N uptake during this stage is very low [15]. In fact, the N demand of directly seeded rice seedlings before the three-leaf stage can be met by the rice seeds themselves as well as the endogenous soil nutrient supply. Thus, a large proportion of the N applied basally at sowing is lost to the environment. These results indicate that less N fertilizer may be required at the early growth stage in the directly seeded rice system.

Due to phenological and physiological differences, the traditional N management practices applied to transplanted rice are not suitable for the NTDS rice cropping system [3]. Fertilizing schemes appropriate for NTDS have become the focus of new studies to improve rice grain yield and NUE and while simultaneously reducing environmental risk. The rate and timing of N fertilization should be adjusted to match the N supply to the N requirement of the rice crop throughout the growing season. At present, while many studies have been conducted to determine the response of rice to N management practices, the majority have focused on the effect of N management practices on transplanted rice [16,21]. This means that limited information is available on N management practices that enhance grain yield and N use efficiency (NUE) of hybrid rice grown under NTDS. Based on the growth performance and physiological characteristics of no-tillage rice and direct-seeded rice crops, we hypothesize that the rice grain yield and NUE under NTDS can be simultaneously improved by delaying the timing of the first N application and reducing the N application rate. To test this hypothesis, an N fertilization field experiment was conducted with the following aims: (1) to address the effects of delaying the first N application time and reducing the N application rate on grain yield and yield components of two hybrid rice cultivars; (2) to examine the effects of delaying the first N application time and reducing the N application rate on N uptake and NUE of two hybrid rice cultivars.

2. Materials and Methods

2.1. Experimental Sites

An N fertilization field experiment was performed in the 2020 and 2021 growing seasons in Luxian County (29°10′ N, 105°23′ E, 280 m asl), located in Sichuan Province, China. The soil in the experimental paddy field was dark purple paddy soil with the following properties in the upper 20 cm layer: pH of 4.5, organic matter of 28.3 g kg⁻¹, total N of 1.7 g kg⁻¹, total phosphorus of 485.5 mg kg⁻¹, total potassium of 38.9 g kg⁻¹, NaOH hydrolysable N of 149.5 mg kg⁻¹, Olsen P of 6.4 mg kg⁻¹, and NH₄OAc extractable K of 132.0 mg kg⁻¹. In the two years, the fertilization field experiment was conducted in the same paddy field.

2.2. Experimental Design and Cropping Management

Two hybrid rice varieties, ‘Nei6you107’ (long growth duration) and ‘Chuankangyou6276’ (medium growth duration) were used in this study. The two hybrid rice cultivars were developed at the Sichuan Academy of Agricultural Sciences, and the seeds were also provided by the Sichuan Academy of Agricultural Sciences. The hybrid rice cultivar ‘Nei6you107’ has been widely grown by farmers in southwest China because of its good rice quality, high grain yield, and good high-temperature tolerance.

In both years, the experiment included five N treatments; an N omission plot (N₀, CK) and a factorial combination with two N rates and two N split-application ratios (namely R₁ and R₂) among the basal (BS, 1 day before sowing), seedling (SD, three leaves), mid-tillering (MT), and panicle initiation (PI) stages. The two N split-application ratios for R₁ and R₂ were 50% (BS) + 0% (SD) + 30% (MT) + 20% (PI) and 0% (BS) + 30% (SD) + 40% (MT) +30% (PI), respectively. The two N rates were high N (180 kg N ha⁻¹, N₁₈₀) and medium N (a 15% reduction N rate, 153 kg N ha⁻¹, N₁₅₃), and the urea was used as N source. The nitrogen doses were determined according to our previous study [15]. Plots were arranged in a split-plot design with N treatment as the main plot and the hybrid rice cultivars as the sub-plots. Each treatment was replicated three times and the sub-plot size was 35 m². The single applications of superphosphate for phosphorus (P) and potassium chloride for potassium (K) were applied at doses of 67.5 kg P₂O₅ ha⁻¹ and 150 kg K₂O ha⁻¹. P was applied as basal, and K was applied in two splits: 50% as basal and 50% at the panicle initiation stage.

Prior to sowing, the field was treated with herbicide one week before direct seeding. Sterilized rice seeds were germinated by soaking in tap water for 72 h and incubating between thick layers of cotton cloth in an incubator at 35 °C. The pre-germinated seeds were manually broadcast onto the wet soil surface at a rate of 22.5 kg ha⁻¹ on 15 March of both years. Weeds were controlled by applying a pre-emergence herbicide at 5 days after sowing, and a post emergence herbicide at the mid-tillering stage. Water management practices were as follows: water was drained off completely before sowing and soil saturation was maintained from sowing to the three-leaf stage of the rice seedlings, after which the fields were flooded to a depth of about 3–10 cm until the plants reached physiological maturity. Insects and diseases were controlled using chemicals to avoid yield loss.

2.3. Measurement Methods

Plants were sampled from an area of 0.48 m² in each sub-plot at the heading and maturity stages. At heading, plants were separated into stem, leaf, and panicle, after which the plant samples were oven-dried at 70 °C to constant weight to determine the biomass. At maturity, panicles were counted to calculate the number of panicles per m⁻², after which plants were separated into straw and panicles. Panicles were hand-threshed and the filled spikelets were separated from the unfilled spikelets by submerging them in tap water, after which both the filled and unfilled spikelets were air dried. After oven-drying to a constant weight at 70 °C, the dry weights of the straw, filled, and unfilled spikelets were determined. Three sub-samples (30 g each) of the filled spikelets and all unfilled spikelets were counted to calculate the number of spikelets per panicle, the percentage of filling grain, and the grain weight. The total biomass was calculated as the sum of the total dry matter of straw, filled spikelets, and unfilled spikelets. The harvest index was calculated as the ratio of the total dry weight of the filled spikelets to the total biomass. Grain yield was determined from a 10 m² area in the middle of each sub-plot and adjusted to the standard moisture content of 140 mg H₂O g⁻¹.

The rice straw, filled, and unfilled spikelets of the mature samples were ground into powder using a small grinding machine, after which the N concentrations in the straw and filled and unfilled spikelets were determined using an autoanalyzer (Integral Futura, Alliance Instruments, Frépillon, France). The total N uptake and N harvest index were calculated.

The NUEs of the agronomic efficiency of applied N (AE_N), partial factor productivity of applied N (PFP_N), and recovery efficiency of applied N (RE_N) were calculated using the following formulas given by Jiang et al. [15,26]:

AE_N (kg kg⁻¹) = (GY − GY₀)/FN

PFP_N (kg kg⁻¹) = GY/FN

RE_N = (TN − TN₀)/FN × 100

where GY₀ and GY represent the grain yield in the no-N treatment and the other treatments that received N, respectively; TN₀ and TN are the total N uptake in the aboveground plant parts in the no-N treatment and the other treatments that received N, respectively; and FN is the amount of N fertilizer applied.

2.4. Statistical Analysis

Daily weather data for the experimental site (daily minimum and daily maximum temperatures) during the rice-growing season were obtained from the local meteorological bureaus. The Statistix 8 software package (Analytical Software, Tallahassee, Florida, USA) was used to perform an analysis of variance (ANOVA) on the data. The ANOVA statistical model included replication, year (Y), treatments (including N₀ and four combinations of two N rate and two N split-application ratios, T), cultivar (C), the two-factor interactions of Y × T, Y × C, and T × C, and the three-factor interaction of Y × T × C. The significance level was set at the 0.05 probability level. Means of five treatments were compared based on the least significant difference test (LSD) at the 0.05 probability level for each cultivar in both years.

3. Results

3.1. Weather Conditions and Growth Duration

The daily maximum and minimum temperatures from sowing (SW) to heading (HD) were 0.6–1.1 °C and 0.2–0.3 °C higher in 2020 than in 2021, respectively (

Table 1. Daily maximum and minimum temperatures and sunshine hours during the rice-growing seasons in 2020 and 2021.

Cultivar	Year	Maximum Temperature (°C)		Minimum Temperature (°C)		Sunshine Hours (h)
		SW-HD a	HD-MA	SW-HD	HD-MA	SW-HD	HD-MA
Nei6you107	2020	27.0	33.0	19.3	25.2	529.3	175.9
	2021	26.4	34.9	19.0	26.1	330.1	214.7
Chuankangyou6276	2020	26.8	30.6	18.4	24.1	500.3	126.4
	2021	25.7	33.5	18.2	25.5	279.2	194.7

^a SW-HD, the period from sowing to heading; HD-MA, the period from heading to maturity.

), whereas from HD to maturity (MA), the daily maximum and minimum temperatures were lower in 2020 than in 2021 by 1.9–2.9 °C and 0.9–1.4 °C, respectively. The duration of sunshine from SW to HD was 199.2–221.1 h longer in 2020 than in 2021, whereas from HD to MA, the duration of sunshine was 38.8–68.3 h shorter in 2020 than in 2021.

The growth duration from SW to HD was 3–4 d shorter in 2020 than 2021 (Figure 1), whereas from HD to MA, the growth duration was 2–3 d longer in 2020 than 2021. Rice hybrid ‘Nei6you107’ had a longer growth duration from SW to HD than did ‘Chuankangyou6276’ by 13–14 d, and a similar but smaller difference between them was observed for the growth duration from HD to MA.

3.2. Grain Yield and Yield Components

The grain yield was significantly affected by year (Y) and treatment (T), but not by the cultivar (C) (

Table 2. Effects of different N management practices on grain yield (t ha⁻¹) in hybrid rice grown in the no-tillage, direct-seeded (NTDS) cropping system.

Cultivar	Treatment		2020	2021
	N Rate (N) a	N Split-Application Ratio (R) b
Nei6you107	N0		5.70 c	6.87 c
	N180	R1	8.43 b	8.92 b
	N180	R2	8.58 ab	9.21 ab
	N153	R1	8.57 ab	8.97 ab
	N153	R2	8.78 a	9.38 a
	Mean		8.01	8.67
Chuankangyou6276	N0		5.74 b	6.40 b
	N180	R1	8.56 a	9.01 a
	N180	R2	8.81 a	9.18 a
	N153	R1	8.41 a	8.92 a
	N153	R2	8.59 a	8.99 a
	Mean		8.02	8.50
Analysis of variance
Year (Y)			**
Treatment (T)			**
Cultivar (C)			ns
Y × T			ns
Y × C			ns
T × C			ns
Y × T × C			ns

Note: Within the column for each cultivar, means followed by the same letters are not significantly different according to LSD at p = 0.05. ** mean significance at p < 0.05 and p < 0.01 levels, respectively. ns denotes non-significance. ^a N₁₈₀, N₁₅₃, and N₀ are the 180, 153, and 0 kg N ha⁻¹ treatments, respectively. ^b R₁ and R₂ are N split-applications among the basal (one day before direct-seeding), seedling (three leaves), mid-tillering, and panicle initiation stages at ratios of 5:0:3:2 and 0:3:4:3, respectively.

). The interactive effects of Y × T, Y × C, T × C, and Y × T × C were not significant for grain yield. For the hybrid rice Nei6you107, the grain yield under combination of N₁₅₃ and R₂ split application was significantly higher than that under combination of N₁₈₀ and R₁ split application by 4.2% in 2020 and by 5.2% in 2021. For the hybrid rice Chuankangyou6276, there was no significantly difference in grain yield among the four combinations of two N rates and two N split-application ratios in both years. Under the same N split-application ratio, although the N rate was 15% lower for N₁₅₃ compared to N₁₈₀, grain yield under N₁₅₃ was equal to or higher than that under N₁₈₀. The grain yield for the R₂ split application was higher than that for R₁ by 1.8–3.3% for N₁₈₀ and 0.8–4.6% for N₁₅₃. The difference in grain yield between two hybrid rice cultivars was small.

The panicles per m², sipikelets per panicle and spikelets per m² were significantly affected by year, treatment, and cultivar (

Table 3. Effects of different N management practices on yield components of hybrid rice grown in the no-tillage, direct-seeded (NTDS) system.

Year	Cultivar	Treatment		Panicles m−2	Spikelets Panilce−1	Spikelets m−2 (×103)	Grain Filling (%)	Grain Weight (mg)
		N Rate (N) a	N Split-Application Ratio (R) b
2020	Nei6you107	N0		205.5 d	121.0 bc	24.9 c	85.8 a	29.1 a
		N180	R1	328.9 a	122.0 bc	40.1 a	80.7 b	29.5 a
		N180	R2	325.6 ab	127.0 b	41.3 a	81.8 ab	29.2 a
		N153	R1	311.1 b	116.8 c	36.3 b	83.9 ab	29.6 a
		N153	R2	291.1 c	140.2 a	40.7 a	83.0 ab	28.3 a
		Mean		292.4	125.4	36.7	83.0	29.1
	Chuankangyou6276	N0		198.9 c	138.6 b	27.5 c	83.4 a	27.0 a
		N180	R1	315.6 a	140.2 b	44.3 ab	80.8 ab	27.5 a
		N180	R2	302.2 a	155.6 a	47.1 a	81.9 a	27.4 a
		N153	R1	280.0 b	158.7 a	44.5 ab	78.1 b	27.2 a
		N153	R2	275.0 b	164.0 a	45.1 b	82.5 a	27.6 a
		Mean		274.3	151.4	41.7	81.3	27.3
2021	Nei6you107	N0		188.9 c	136.2 c	25.7 c	84.9 a	29.7 a
		N180	R1	247.2 b	145.5 bc	35.9 b	84.3 a	29.7 a
		N180	R2	250.0 ab	156.0 ab	38.9 b	84.1 a	28.8 b
		N153	R1	273.6 a	136.7 c	37.3 b	84.8 a	29.2 ab
		N153	R2	256.9 ab	164.8 a	42.3 a	84.3 a	29.0 ab
		Mean		243.3	147.8	36.5	84.5	29.3
	Chuankangyou6276	N0		190.3 c	152.6 c	28.8 c	81.5 a	28.3 a
		N180	R1	251.4 a	166.8 abc	41.6 a	78.3 a	29.0 a
		N180	R2	234.7 ab	187.4 a	44.0 a	81.0 a	28.6 a
		N153	R1	241.7 ab	159.1 bc	38.4 b	78.5 a	29.4 a
		N153	R2	213.9 bc	182.1 ab	38.7 b	79.9 a	29.0 a
		Mean		226.4	169.6	38.3	79.8	28.9
Analysis of variance
Year (Y)				**	**	**	ns	**
Treatment (T)				**	**	**	ns	ns
Cultivar (C)				**	**	**	**	**
Y × T				**	ns	**	ns	ns
Y × C				ns	ns	**	*	**
T × C				*	ns	**	ns	ns
Y × T × C				ns	ns	**	ns	ns

Note: Within the column for each cultivar, means followed by the same letters are not significantly different according to LSD at p = 0.05. * and ** mean significance at p < 0.05 and p < 0.01 levels, respectively. ns denotes non-significance. ^a N₁₈₀, N₁₅₃, and N₀ are 180, 153, and 0 kg N ha⁻¹, respectively. ^b R₁ and R₂ are N split-applications among the basal (one day before direct-seeding), seedling (three leaves), mid-tillering, and panicle initiation stages at ratios of 5:0:3:2 and 0:3:4:3, respectively.

). The average number of panicles per m⁻² for the N₁₈₀ rate was higher than the N₁₅₃ rate by 2.7–3.8% for ‘Nei6you107’ and 3.2–9.9% for ‘Chuankangyou6276’, while there were 10.6–12.0% and 6.5%-11.8% fewer average spikelets per panicle under N₁₈₀ than those under N₁₅₃ for ‘Nei6you107’ and ‘Chuankangyou6276’, respectively. As a result, the average number of spikelets per m⁻² for N₁₈₀ was lower than for N₁₅₃ by 6.8–9.9% for ‘Nei6you107’ and 3.3–3.7% for ‘Chuankangyou6276’. The grain filling was significantly affected by cultivar, but not by year and treatment. The average percentage of grain filling in N₁₅₃ was slightly (0.1–3.5%) higher than that in N₁₈₀, except for ‘Nei6you107’ in 2021.

The number of panicles per m² for R₁ was higher than it was for R₂ by 1.0–7.1% (except for ‘Nei6you107’ in 2021) at N₁₈₀ and 1.8–13.0% at N₁₅₃, while there were 3.3–11.0% and 3.2–17.1% fewer spikelets per panicle under R₁ than under R₂ for N₁₈₀ and N₁₅₃, respectively. As a consequence, the R₁ split ratio had fewer spikelets per m⁻² than R₂ by 2.9–7.7% for N₁₈₀ and 0.8–11.8% for N₁₅₃. The differences in the percentages of filled grains and grain weight were relatively small between the two N split-application ratios. The number of panicles per m⁻² and spikelets per m⁻² was higher in 2020 than in 2021 by 20.7% and 4.8%, respectively, while there were 12.8% fewer spikelets per m⁻², 0.6% fewer filled grains, and 3.1% lower grain weight in 2020 compared to 2021.

3.3. Biomass Production and Harvest Index

The biomass before and after heading and at maturity, BT_pre and harvest index were significantly affected by year (Y), treatment (T), and cultivar (T), except for the effect of Y for total biomass at maturity (

Table 4. Effect of different N management practices on biomass production, and harvest index (HI) of hybrid rice in the no-tillage, direct-seeded (NTDS) cropping system.

Year	Cultivar	Treatment		Biomass Production (g m−2)				Harvest index (%)
		N Rate (N) a	N Split-Application Ratio (R) b	Before Heading	After Heading	Total	BTpre c
2020	Nei6you107	N0		821.4 d	312.4 b	1133.8 d	305.7 b	54.5 a
		N180	R1	1542.9 a	355.8 b	1898.7 bc	595.8 a	50.1 b
		N180	R2	1465.6 ab	543.3 a	2009.0 a	444.5 b	49.2 b
		N153	R1	1313.2 c	507.6 a	1820.8 c	393.9 b	49.5 b
		N153	R2	1329.6 bc	572.5 a	1902.1 b	383.9 b	50.3 b
		mean		1294.5	458.3	1752.9	424.8	50.7
	Chuankangyou6276	N0		728.8 c	325.6 c	1054.4 c	294.8 c	58.8 a
		N180	R1	1238.2 a	526.2 b	1764.4 b	453.3 a	55.5 b
		N180	R2	1226.1 a	645.2 a	1871.3 a	409.8 ab	56.4 b
		N153	R1	1220.2 a	486.7 b	1706.8 b	456.4 a	55.3 b
		N153	R2	1098.7 b	679.2 a	1778.0 b	328.4 bc	56.7 b
		mean		1102.4	532.6	1635.0	388.5	56.5
2021	Nei6you107	N0		981.1 c	277.2 c	1258.4 c	368.8 ab	51.4 a
		N180	R1	1395.9 a	485.2 b	1881.1 b	412.5 a	47.8 b
		N180	R2	1253.3 b	633.8 a	1887.0 ab	307.1 b	49.9 ab
		N153	R1	1353.3 a	524.4 b	1877.6 b	400.4 a	49.2 ab
		N153	R2	1342.7 a	670.4 a	2013.0 a	364.9 ab	51.4 a
		mean		1265.3	518.2	1783.4	370.7	49.9
	Chuankangyou6276	N0		665.8 b	614.7 b	1280.6 c	49.0 c	51.8 b
		N180	R1	1023.4 a	849.0 a	1872.5 a	93.9 bc	50.4 b
		N180	R2	969.0 a	868.8 a	1837.8 a	150.3 b	55.4 a
		N153	R1	980.8 a	606.7 b	1587.5 b	279.0 a	55.8 a
		N153	R2	969.6 a	653.0 b	1622.6 b	243.7 a	55.3 a
		mean		921.7	718.4	1640.2	163.2	53.7
Analysis of variance
Year (Y)				**	**	ns	**	**
Treatment (T)				**	**	**	**	**
Cultivar (C)				**	**	**	**	**
Y × T				**	**	**	**	**
Y × C				**	**	ns	**	**
T × C				ns	**	**	**	**
Y × T × C				ns	**	**	ns	ns

Note: Within the column for each cultivar, means followed by the same letters are not significantly different according to LSD at p = 0.05. ** mean significance at p < 0.05 and p < 0.01 levels, respectively. ns denotes non-significance. ^a N₁₈₀, N₁₅₃, and N₀ are 180, 153 and 0 kg N ha⁻¹, respectively. ^b R₁ and R₂ are N split-applications among the basal (one day before direct-seeding), seedling (three leaves), mid-tillering, and panicle initiation stages at ratios of 5:0:3:2 and 0:3:4:3, respectively. ^c BT_pre represents the product of translocation of before-heading biomass to the grains.

). The biomass before and after heading, biomass at maturity, and BT_pre for the N₁₈₀ applications were higher than those for N₁₅₃ by 5.3%, 4.4%, 5.0%, and 0.6%, respectively, while there was a 2.1% lower harvest index for N₁₈₀ treatment than N₁₅₃ treatment. Except for ‘Nei6you107’ in the N₁₅₃ treatment in 2020, the R₁ split application scheme gave 0.8–11.4% higher biomass before heading than R₂. However, there was 2.3–34.5% lower biomass after heading with the R₁ split ratio than with R₂. In general, R₁ produced lower biomass at maturity than did R₂, but higher BT_pre than R₂. The harvest index for R₂ was higher than that for R₁. The biomass before heading, BT_pre, and harvest index were 9.6%, 52.3%, and 3.5% higher, respectively, in 2020 than in 2021, while the biomass after heading and at maturity in 2020 was lower than in 2021 by 19.9% and 1.0%, respectively.

3.4. Nitrogen Uptake, N Harvest Index, and NUE

The N uptake and NHI were significantly affected by year (Y), treatment (T), and cultivar (C) except the effect of C for N uptake (

Table 5. Effect of different N management practices on nitrogen uptake (NU), nitrogen harvest index (NHI), agronomic efficiency of applied N (AE_N), partial factor productivity of applied N (PFP_N), and recovery efficiency of applied N (RE_N) of hybrid rice in the no-tillage, direct-seeded (NTDS) cropping system.

Year	Cultivar	Treatment		N Uptake (g m−2)	NHI (%)	AEN (kg kg−1)	PFPN (kg kg−1)	REN (%)
		N Rate (N) a	N Split-Application Ratio (R) b
2020	Nei6you107	N0		10.2 d	72.3 a
		N180	R1	18.7 b	63.4 c	15.2 b	46.8 b	47.1 b
		N180	R2	19.9 a	66.0 b	16.0 b	47.6 b	53.9 a
		N153	R1	17.4 c	65.7 b	18.8 a	56.0 a	46.9 b
		N153	R2	18.8 b	66.6 b	20.1 a	57.4 a	55.9 a
		Mean		17.0	66.8	17.5	52.0	50.9
	Chuankangyou6276	N0		10.4 c	75.6 a
		N180	R1	17.9 b	71.9 b	15.7 a	47.6 b	41.8 c
		N180	R2	20.1 a	72.4 b	17.1 a	49.0 b	54.2 ab
		N153	R1	17.6 b	72.1 b	17.4 a	55.0 a	47.0 bc
		N153	R2	19.7 a	73.2 b	18.7 a	56.1 a	61.1 a
		Mean		17.1	73.0	17.2	51.9	51.0
2021	Nei6you107	N0		11.3 b	75.4 a
		N180	R1	18.0 a	65.9 c	11.3 c	49.5 c	37.1 a
		N180	R2	18.2 a	70.8 b	13.0 bc	51.2 c	38.4 a
		N153	R1	17.5 a	69.9 b	13.7 b	58.6 b	40.4 a
		N153	R2	18.1 a	73.0 ab	16.4 a	61.3 a	43.8 a
		Mean		16.6	71.0	13.6	55.1	39.9
	Chuankangyou6276	N0		11.2 c	74.9 a
		N180	R1	17.8 b	71.7 a	14.5 a	50.1 b	36.1 b
		N180	R2	18.7 a	74.4 a	15.5 a	51.0 b	41.3 a
		N153	R1	17.0 b	73.2 a	16.5 a	58.3 a	37.7 ab
		N153	R2	17.6 b	74.8 a	16.9 a	58.7 a	41.5 a
		Mean		16.5	73.8	15.8	54.5	39.1
Analysis of variance
Year (Y)				**	**	**	**	**
Treatment (T)				**	**	**	**	**
Cultivar (C)				ns	**	ns	ns	ns
Y × T				**	*	*	ns	*
Y × C				ns	**	ns	ns	ns
T × C				ns	**	ns	ns	ns
Y × T × C				ns	ns	ns	ns	ns

Note: Within the column for each cultivar, means followed by the same letters are not significantly different according to LSD at p = 0.05. * and ** mean significance at p < 0.05 and p < 0.01 levels, respectively. ns denotes non-significance. ^a N₁₈₀, N₁₅₃, and N₀ are 180, 153, and 0 kg N ha⁻¹, respectively. ^b R₁ and R₂ are N split-applications among the basal (one day before direct-seeding), seedling (three leaves), mid-tillering, and panicle initiation stages at ratios of 5:0:3:2 and 0:3:4:3, respectively.

). The interactive effect of Y × T were significant for N uptake and NHI. The N uptake at maturity for the N₁₈₀ application rate was 1.7–6.6% higher than for N₁₅₃, whereas there was 0.7–4.3% lower N harvest indexes in N₁₈₀ than in N₁₅₃. The N uptake at maturity in R₂ was higher than that in R₁ by 1.1–12.3%; moreover, the R₂ split produced 0.7–7.4% higher N harvest indexes than did R₁. The N uptake at maturity was 3.0% higher in 2020 than in 2021, while there was 3.5% lower N harvest index in 2020 than in 2021.

The AE_N, PFP_N, and RE_N were significantly affected by year (Y) and treatment (T) (Table 5). The interactive effects of Y × T, Y × C, T × C, and Y × T × C were not significant for AE_N, PFP_N, and RE_N, except for the interactive effect of Y × T for AE_N and RE_N. The N₁₅₃ had higher AE_N, PFP_N, and RE_N than N₁₈₀ by 10.1–24.7%, 15.0–20.1%, and 1.8–12.6%, respectively. Delaying and reducing the first N application (R₂) increased AE_N by 2.4–19.7%, PFP_N by 0.7–4.6%, and RE_N by 3.5–30.0% over the R₁ treatment. The AE_N and RE_N were 15.3% and 22.5% lower in 2021 than in 2020, while the PFP_N was 5.6% higher in 2021 than in 2020.

4. Discussion

In the present study, we determined the responses to N management practices in the NTDS rice cropping system. The N rate in the N₁₅₃ treatments was 15% lower than in the N₁₈₀ treatments, though grain yield was equal to or higher in the N₁₅₃ treatments than that in the N₁₈₀ treatments in the NTDS rice cropping system (Table 2). This aligns with the results of previous studies that were performed under conventional tillage and transplanting conditions [14,21,26]. Our results indicate that high N application rates are not necessarily required to achieve high grain yields in the NTDS rice cropping system.

Matching the N supply to the requirements of the rice crop is an essential component of optimizing N management, and one way to achieve this match is by optimizing the N rate, the split application of N, and the timing of N application [15]. In general, application of a basal dose of N fertilizer in conventional transplanted rice cultivation has always been a blanket recommendation worldwide, because it is beneficial to rice reviving and tillering. In our study, delaying and reducing the first N application (R₂) produced higher grain yield than a high basal N fertilizer application (R₁) by 1.8–3.3% for N₁₈₀ and 0.8–4.6% for N₁₅₃ (Table 2). As a consequence, the AE_N, PFP_N, and RE_N in R₂ were higher than those in R₁ by 2.4–19.7%, 0.7–4.6%, and 3.5–30.0%, respectively (Table 5). The rice roots and soil nutrient characteristics in NTDS were partially responsible for the higher grain yield and NUEs in R₂ than in R₁. First, NTDS causes the distribution of rice roots and soil nutrients to the surface soil layer; as a result, less N fertilizer is required in the early growth period of rice. Second, the N uptake of rice before the three-leaf stage is very low because the rice seedlings are very small during this growth period; therefore, the N demand of rice seedlings prior to this stage can be met by the rice seeds themselves as well as the endogenous soil nutrient supply [15]. Third, no-tillage rice requires a long time to take up basal N, which means that a large proportion of the N applied basally at sowing is prone to being lost to the environment [15,21,27]. Thus, our results suggest that applying high levels of basal N fertilizer can be replaced with a delayed and reduced first N application to improve grain yield and NUEs in the NTDS rice cropping system.

The high grain yields realized for R₂ are attributed to large sink size (spikelets m⁻²), which result from large panicle size because the differences in grain filling and grain weight between them were relatively small (Table 3). The importance of sink size in enhancing grain yield has been reported in many studies [10,20,27]. Applying basal N has always been a blanket recommendation in worldwide rice cultivation because it is beneficial to promoting rice growth and tillering. The N application rate from sowing to the three-leaf stage in R₁ was 30.6–36.0 kg N ha⁻¹ higher than in R₂. As a result, the number of panicles per m² in R₁ was higher than that in R₂, while plants in R₁ had fewer spikelets per panicle than did plants in R₂. These results indicate that there is a strong compensation between the two yield components, and an increase in one component will not necessarily result in an overall increase in sink size. Overuse of basal N fertilizer in the early growth stage can induce rice plants to produce a large number of tillers that aggravate the contradiction between rice crop population size and individual plants in the NTDS rice cropping system; therefore, the number of spikelets per panicle decreased significantly. R₂ had a lower N application rate in the basal fertilizer, but a higher N application rate in the panicle initiation fertilizer than in R₁. It is generally accepted that N should be applied at panicle initiation stage, as this promotes spikelet differentiation and increases spikelets per panicle. R₂ produced plants with more spikelets per panicle than did R₁, and as a consequence, the sink size was higher in R₂ than in R₁ (Table 3). These results show that shifting part of the N from the basal fertilizer to the panicle initiation stage could improve the number of spikelets per panicle, leading to a higher sink size in the NTDS rice cropping system.

More interesting, R₂ produced a larger sink size (spikelets per m⁻²) than R₁, though R₂ consistently had a higher percentage of grain filling than R₁. Low biomass accumulation is probable after heading with R₁ compared with R₂, reducing grain filling in the two hybrid rice cultivars (Table 4). This result suggests that developing new hybrid rice cultivars with high spikelet numbers per panicle is a feasible approach to improve grain yield in the NTDS rice cropping system, and a high biomass production capacity after heading is critical to improving the percentage of grains.

Rice grain yield is determined by the aboveground biomass and harvest index. A number of previous studies have reported that further improvement of rice yield depends on increasing the biomass production, because there is little scope to further increase harvest index [27,28,29]. However, in this study, the biomass and harvest index in R₂ were equal to or higher than those in R₁ (Table 4), which indicated that both aboveground biomass and harvest index are important factors to explain the yield gap between them. In another approach, rice grain yield is the function of biomass accumulation after heading and the translocation of biomass accumulated before heading into the grains. Increasing biomass production after heading (or improving after-heading growth) can decrease BT_pre, while improving before-heading growth can increase BT_pre. In the present study, the biomass production before heading and the translocation of biomass accumulated before heading (BT_pre) to the rice grains was lower for R₂ than for R₁. Biomass accumulation after heading, which represents the current flow of photoassimilates to the rice grains, was much greater for R₂ than for R₁ in both study years (Table 4). This result suggests that high biomass production after heading is an important factor to explain the yield gap between R₁ and R₂. Huang et al. [30] reported that a high BT_pre results in a low level of storage of before-heading biomass in the straw and consequently decreases straw stiffness and plant lodging resistance. No lodging occurred in R₂ during either year of the experiment.

Reducing the N application rate and applying less N in the early growing season are the two key N management strategies for improving NUEs (AE_N, PFP_N, and RE_N) in rice production. The two N management strategies might be more efficient in NTDS rice, because the uptake of basal N takes a long time in no-tillage rice, and the N uptake of directly seeded rice is very low before the three-leaf stage. Consistently, in the present study, reducing the total N fertilizer rate from 180 kg ha⁻¹ to 153 kg ha⁻¹ resulted in higher AE_N, PFP_N, and RE_N for 153 kg N ha⁻¹ than for 180 kg N ha⁻¹ (Table 5). This result agrees with those of a previous field study, and a reduction in total N application rate is believed to improve NUEs [31]. A decrease in the N rate (30.6–36.0 kg N ha⁻¹) at the early vegetative stage resulted in higher total N uptake, N harvest index, AE_N, PFP_N, and RE_N for R₂ compared to R₁. These results suggest that reduced and delayed N application rate at the early vegetative stage can improve N uptake, N harvest index, and NUEs in the NTDS rice cropping system.

Grain yield was higher in 2021 than in 2020, though the AE_N was lower in 2021 than in 2020. Because AE_N was calculated by the ‘‘difference method’’ using the yield of the no-N plot, the low values of AE_N indicate that the grain yield in the no-N control was higher in 2021 than in 2020. The larger sink size that resulted from more spikelets per panicle and high biomass production at maturity are two important factors to explain the yield difference in the no-N control between the two years. The differences in grain yield, total N uptake AE_N, PFP_N, and RE_N between the two hybrid rice cultivars were relatively small.

5. Conclusions

Our study demonstrated that the total amount of N fertilizer applied could be reduced from 180 kg ha⁻¹ to 153 kg ha⁻¹ without sacrificing grain yield, while increasing the AE_N, PFP_N, and RE_N by 10.1–24.7%, 15.0–20.1%, and 1.8–12.6%, respectively, in the NTDS rice cropping system. Delayed and reduced N application rate at the early vegetative stage improved grain yield and NUE in the NTDS rice cropping system. The increase in sink size that resulted from more spikelets per panicle and higher biomass production after heading are critical to achieve a high grain yield. The results of our study suggest that it is feasible to simultaneously improve grain yield and NUE of hybrid rice, and reducing production inputs and environmental risk in the NTDS in cropping system through delayed and reduced N application rate, and current breeding programs need to target large panicle size as a primary objective for NTDS rice.