Controlled‐release nitrogen fertilizer management influences grain yield in winter wheat by regulating flag leaf senescence post‐anthesis and grain filling

Abstract

Controlled‐release nitrogen fertilizer (CRNF) is expected to meet the nutrient demands for crop growth with single fertilizer application, so as to achieve high‐yield, high‐efficiency, and labor‐saving cultivation. However, under current CRNF practices, a single application of CRNF in winter wheat (Triticum aestivum L.) usually fails to achieve satisfactory effects due to insufficient nitrogen (N) supply in the later growth stage, contributing to premature leaf senescence and poor grain filling. To determine a reasonable CRNF application strategy for improving the grain weight and yield of winter wheat, the responses of leaf senescence, photosynthetic capacity, grain‐filling characteristics, and grain yield to different CRNF types [polymer‐coated urea (PCU), sulfur‐coated urea (SCU), and urea‐formaldehyde (UF)] and fertilization practices [single fertilization (P1) and twice‐split fertilization (P2)] were investigated. The results showed that, compared with P1, P2 increased the activities of catalase, peroxidase, and superoxide dismutase, alleviated the accumulation of malondialdehyde, delayed flag leaf senescence, and enhanced the net photosynthetic rate (Pn) of the flag leaves, which promoted assimilate accumulation and transport to the grains. These advantages in P2 improved the grain‐filling process of wheat, with the active filling duration (T), average filling rate (G), and maximum filling rate (G_max) in P2 all being higher than those in P1, which significantly increased the 1000‐grain weight (TGW) and grain yield of wheat. Among the three CRNFs, PCU showed advantages in Pn at the milk‐ripe stage (Zadoks growth stage, GS75), dry matter accumulation post‐anthesis, T, G_max, and G, which contributed to TGW and grain yield. Overall, our study suggested that the twice‐split application of CRNF could substantially delay leaf senescence post‐anthesis and increase photosynthetic capacity, thus improving grain filling and productivity in winter wheat.

INTRODUCTION

Winter wheat (Triticum aestivum L.) accounts for more than 20% of the total calories and 20% of the protein consumed by the global population (FAO, 2016; Peña et al., 2017) and is one of the most important cereal crops in China and the world. The grain yield of wheat per unit area needs to be enhanced in order to meet the requirements of an increasing global population and decreasing arable land, through either breeding or improved agronomic management (Yan et al., 2019). Wheat production has doubled since the 1960s, and the development of genetic resources has contributed 47–60% to yield increases (Liu et al., 2021; Novoselović et al., 2000; Richards, 2000; Tian et al., 2011). However, as the wheat yield per unit area has reached very high levels in recent years, it is becoming increasingly difficult to continue improving wheat production by relying solely on the improvement of varieties (Alotaibi et al., 2020). Cultivation management also plays an important role in wheat yield increase. Wheat yield is separated into three components, and an effective approach for improving yield is to increase the contributions of yield components (spikes per unit area, kernels per spike, and grain weight) through rational management practices, such as optimizing N supply (Liang et al., 2017).

Path analysis of yield and its components by Cao et al. (2019) showed that spikes per unit area contributed the most to yield, followed by 1000-grain weight (TGW), while grain number per spike was observed to have a negative effect on yield. In contrast to grain number per spike, spikes per unit area and grain weight are more influenced by environmental conditions rather than gene regulation, indicating that they are susceptible to the regulation by cultivation measures (Du et al., 2020). Reports have shown that optimizing spike number per hectare is one of the key cultivation practices for maximizing yield in cereal crops (Cao et al., 2019; Hayashi et al., 2006). However, crop yield under increased density showed a parabolic trend (Qun et al., 2020). Under current cultivation conditions, spike number per unit area has likely reached its limits, and further increases in density may lead to severe problems, such as lodging, premature senescence, and disease aggravation (Liang et al., 2017). Therefore, an inevitable approach for increasing wheat yield is to increase the grain weight based on the appropriate spikes per area and grains per spike. Some reports have indicated that N supply is the most direct way to regulate leaf senescence post-anthesis and the accumulation of assimilates in the grains (Kitonyo et al., 2018; Tadahiko, 2004).

The application of N fertilizer, especially urea, has contributed greatly to increased crop yields in the past decades (Erisman et al., 2008). However, due to the rapid transformation to ammonia and CO₂ in moist soil (Fisher et al., 2016), nutrient losses from N fertilizers contribute significantly to low fertilizer use efficiency (Linquist et al., 2012) and environmental problems, including soil acidification, air pollution, and surface water eutrophication (Cameron et al., 2013; Dimkpa et al., 2020; Liu et al., 2016). Accordingly, scientists and the industry have been striving for the ideal fertilizer that promotes a balance among nutrient release, plant uptake, and environmental conservation (Bindraban et al., 2015; Dimkpa et al., 2020). Controlled-release nitrogen fertilizer (CRNF) offers an approach for alleviating this contradiction (Shaviv & Mikkelsen, 1993). Our previous study has confirmed that CRNF practices could affect the spike number and grain weight of wheat to a certain extent, thus affecting wheat yield, but it failed to explore the effects of CRNF on leaf senescence and the grain-filling process, as well as the relationship among leaf senescence, grain filling and yield. (Ma et al., 2021).

Grain filling is the final growth period in cereals when the fertilized ovaries develop into caryopses and determines the ultimate grain weight (He et al., 2020). It involves complex physiological and biochemical reactions, such as photosynthetic physiology, remobilization of nutrients, degradation of chlorophyll–protein complexes, and lipid peroxidation (Liang et al., 2017; Wang et al., 2014). These processes regulate plant senescence and the rate and duration of the grain-filling process (Gelang et al., 2008; Zhao et al., 2007), which largely determines the grain weight and are also susceptible to N application (Wei et al., 2019). The mechanism of grain weight regulation by the grain-filling process remains controversial. Bolaños (1995) reported that improvements in grain weight were largely supported by the prolongation of the effective filling-period duration. However, Borrás and Gambín (2010) argued that grain weight was closely determined by the rate of grain growth during the grain-filling period rather than the grain-filling duration. N rate was reported to significantly influence the rate and duration of grain filling, and grain weight was more responsive to spike N (Jiang et al., 2016; Wei et al., 2019). Excessive N affects grain filling and reduces wheat yield by extending growth and delaying maturity (Jiang et al., 2016), whereas N shortage contributes to premature senescence, which is also detrimental to nutrient transport (Roger & Kindred, 2009).

The effect of N on the grain weight of cereals is complex. Senescence and photosynthesis—especially that of the flag leaves, which are highly regulated by N supply—play a vital role in modulating the grain-filling process, nutrient remobilization, and dry matter accumulation (Liu et al., 2013; Wu et al., 2012). Photosynthesis is an important process in dry matter production, and any accumulation of dry matter comes from the current photosynthesis and re-allocation of assimilates, which has a strong interaction with leaf senescence (Parry et al., 2011). One of the main factors inducing leaf senescence is damage to the pericellular membrane by reactive oxygen species (ROS) or free radicals, thereby destroying normal cell metabolism (Kapoor et al., 2019). N has been shown to regulate the scavenging system comprising antioxidants, such as superoxide dismutase (SOD, E.C.1.15.1.1), catalase (CAT, E.C.1.11.1.6), and peroxide (POD, E.C.1.11.1.7), to alleviate the cellular damage caused by ROS (Gaju et al., 2014; Misra & Gupta, 2006), thereby relieving leaf senescence. As a vital component element for chlorophyll–protein complexes, N also affects assimilate production by directly influencing leaf photosynthetic efficiency (Gaju et al., 2014; Tang et al., 2005).

The application of CRNF has the potential to enhance the efficient use of nutrients, as N can be utilized by the plant before being volatilized or leached (Azeem et al., 2014). However, the mechanism of CRNF in increasing wheat yield remains controversial due to the differences in types and application practices (Farmaha & Sims, 2013; Zheng et al., 2016). Li et al. (2021) showed that split application of a mixture of CRNF and common urea could maintain higher net photosynthetic rate and achieve higher biomass production post-anthesis, but the effect of such fertilization practice on grain filling was not mentioned. Moreover, no study has attempted to reveal the function of CRNF on leaf senescence and grain filling in wheat. Therefore, on the basis of the previous study, this paper aims to investigate the response of different types and application practices of CRNF to leaf senescence, grain filling and yield. The objectives of this study were to (1) evaluate the differences in flag leaf senescence, photosynthetic capacity, and assimilate accumulation and distribution under different types and application practices of CRNF; (2) reveal the regulatory mechanisms of leaf senescence and photosynthetic performance on the grain-filling process of wheat; and (3) develop an effective CRNF approach for improving the grain weight and grain yield of winter wheat. This study offers new insights into the mechanisms underlying the grain-filling process of wheat under twice-split application of CRNF and provides a theoretical and practical basis for promoting high-yielding and high-efficiency wheat cultivation and ensuring world food security.

MATERIALS AND METHODS Materials and site description

A two-year field experiment (during the wheat-growing seasons of 2017–2018 and 2018–2019) was conducted at the Agricultural Experiment Station (32°39′N, 119°42′E) of Agricultural College, Yangzhou University, Yangzhou, China. The experimental site is a summer rice (Oryza sativa L.)–winter wheat cropping system area where the soil texture is silty loam. The nutritional characteristics of the topsoil (0–20 cm depth) at the experimental site before sowing were soil organic matter of 14.17 g kg⁻¹, total N of 1.03 g kg⁻¹, available N of 69.86 mg kg⁻¹, available phosphorus of 44.93 mg kg⁻¹, and available potassium of 107.38 mg kg⁻¹. The accumulated temperature, precipitation, and sunshine duration in the experiment represented high-yield potential conditions (Table 1). The reasonable accumulated temperature and sunshine duration after anthesis (Zadoks stage, GS60) were beneficial to the grain filling and dry matter accumulation of wheat.

TABLE 1 Climatic conditions during the wheat-growing season at the experimental site in 2017–2018 and 2018–2019

Indicators	SD-SE	SE-TS	TS-JS	JS-BS	BS-AS	AS-MS	WGP
2017–2018
Time (d)	8	60	57	22	15	45	207
Accumulated temperature (℃)	116	397	213	267	226	950	2169
Accumulated precipitation (mm)	0	79	73	58	19	279	509
Accumulated sunshine duration (h)	53	263	271	137	104	257	1085
2018–2019
Time (d)	8	57	59	21	20	48	213
Accumulated temperature (℃)	105	411	223	235	268	949	2191
Accumulated precipitation (mm)	49	159	148	13	5	62	437
Accumulated sunshine duration (h)	23	174	137	134	109	242	819

Abbreviations: AS, Anthesis stage; BS, Booting stage; JS, Jointing stage; MS, Maturity stage; SD, Sowing date; SE, Seedling emergence; TS, Tillering stage; WGP, Whole growth period.

The cultivar of wheat chosen in the experiment was ‘Yangmai 23’, a high-yielding winter wheat variety widely used in local production, which was bred by Lixiahe Institute of Agriculture Sciences, Jiangsu, China (approved by the National Crop Varieties Certification Committee in 2014). The three types of CRNF selected in the experiment were as follows: polymer-coated urea (PCU, 45% N, with a longevity of 90–120 days), sulfur-coated urea (SCU, 37% N, with a longevity of 90–120 days), and urea–formaldehyde (UF, 26% N, with a longevity of 90 days). All CRNFs were freely provided by Hanfeng slow-release fertilizer (Jiangsu) Co., Ltd., China. Superphosphate (12% P₂O₅) and potassium chloride (60% K₂O) were also used in the experiment.

Experimental design

The experiment was laid out in a split-plot design with the type of CRNF (PCU, SCU, and UF) as the main plot and the fertilization practices (single fertilization and twice-split fertilization) as the subplot. The single fertilization treatment (P1) included 100% CRNF applied before sowing; the twice-split fertilization treatment (P2) included 60% CRNF applied before sowing, with 40% CRNF applied at the re-greening stage (GS27). Each treatment was conducted with three replications and with an application rate of 225–112.5–112.5 (N–P₂O₅–K₂O) kg ha⁻¹. Phosphate and potassium fertilizers were both surface-applied to the soil as basal fertilizer, and CRNF was also applied to the soil surface. The size of each plot was 2.7 m wide × 6 m long with 10 rows each spaced 0.27 m apart. The wheat was sown by a plot seeder on November 3, 2017, and November 1, 2018, with the seeding quantity of 165 kg ha⁻¹. At the three-leaf stage, the seedling density was adjusted to 225 × 10⁴ plant ha⁻¹ in all treatments. The other measures of field management were performed with reference to the guidelines for high-yield practices.

Sampling and chemical analyses Grain yield and yield components

Grain yield was determined by 1.08 m² (four 1-m-long rows) harvest areas in each plot with three replicates and standardized to 13% moisture content. Three subsamples of 1000 kernels were randomly selected from the yield measurement sample to calculate TGW. All of the spikes from each harvest area were counted to estimate spike number per hectare. More than 50 spikes from each treatment were randomly selected to determine kernel number per spike by estimating a mean value.

Sampling and determination of grain filling

According to the method by Jiang et al. (2016), a total of 250 single stems of relatively uniform size and the same anthesis date were selected and tagged in each treatment. Twenty tagged samples were collected at 7-day intervals from anthesis (GS60) to 35 days after anthesis (DAA) and then dried in an oven at 105°C for 1 h and then at 80°C until constant weight. Richards equation (Richards, 1959) was fitted to the process of grain filling to obtain the characteristic grain-filling parameters. The number of DAA (t) was treated as the independent variable and the TGW as the dependent variable (W). The basic parameters were calculated using the following equation:[Image Omitted. See PDF]where W is the TGW (g), A is the ultimate TGW (g), and t is the DAA. B, K, and N are the parameters set by the regression equation. R² is the coefficient of the fitness of the equation. The derivation of Eq. (1) is the growth rate Eq. (2).[Image Omitted. See PDF]

In our analysis, the following secondary parameters were adopted to describe the filling characteristics: days from anthesis to reaching the maximum filling rate (T_max•G), the maximum filling rate (G_max), the average filling rate (G), and the active filling duration (T). The formulas were as follows:[Image Omitted. See PDF]

Putting T_max•G into (2), G_max could be calculated.[Image Omitted. See PDF][Image Omitted. See PDF]

In order to divide the filling process into the slow-increase period, fast-increase period, and slight-increase period, two inflection points t₁ and t₂ were calculated with the growth rate Eq. (2). When the grain weight reached 99% A, the filling was assumed to end, at which the time point was represented as t₃.[Image Omitted. See PDF][Image Omitted. See PDF][Image Omitted. See PDF]

It was proved that the slow-increase period of the filling process was T₁ (t₁−0), the fast-increase period T₂ (t₂−t₁), and the slight-increase period T₃ (t₃−t₂). The filling rates G₁, G₂, and G₃, and the filling contribution rations P₁, P₂, and P₃ of the slow-increase, fast-increase, and slight-increase periods were, respectively:[Image Omitted. See PDF][Image Omitted. See PDF]

Chlorophyll content and photosynthetic parameters

The chlorophyll relative content (SPAD value) of the flag leaves was measured using a chlorophyll meter (SPAD 502, Minolta Camera Co., Ltd., Chiyoda City, Japan) at 7-day intervals from anthesis (GS60) to 28 DAA. At the anthesis (GS60) and milk-ripe (GS75), the photosynthetic parameters of the flag leaves, including the net photosynthetic rate (P_n), stomatal conductance (G_s), transpiration rate (T_r), and intercell CO₂ concentration (C_i), were measured using a portable photosynthesis machine (LI-6400, LI-COR, Lincoln, NE, USA). The measurement was performed at 8:30–11:30 a.m. on a sunny and windless day.

Assays of antioxidant enzyme activities on the flag leaves

Ten flag leaves were randomly collected at 7-day intervals from anthesis (GS60) to 28 DAA from the tagged samples of 2.3.2. Fresh flag leaves were immediately submerged in liquid N and then stored at −80°C until biochemical assays were performed. For enzyme extraction, fresh flag leaf samples were cut into 0.5-g pieces and ground in a mortar with liquid N and extracted in 5-mL potassium phosphate buffer solution (pH 7.8) containing 0.2 mol L⁻¹ KH₂PO₄ and 0.2 mol L⁻¹ K₂ HPO₄. The homogenate was centrifuged at 10,000 × g for 20 min at 4°C, and then the supernatant was obtained for enzyme analyses.

The total SOD activity of the flag leaves was measured spectrophotometrically according to the inhibition in the photochemical reduction of nitroblue tetrazolium (NBT) (Giannopolities & Rise, 1977). The CAT and POD activities of the flag leaves were determined by following the changes in absorbance at 240 and 470 nm, respectively (Zhang & Kirkham, 1996). The lipid peroxidation of the flag leaves was measured in terms of the malondialdehyde (MDA) content according to Dhindsa et al. (1981), and the concentration of MDA was calculated using an extinction coefficient of 155 mM⁻¹ cm⁻¹.

Dry matter accumulation and distribution

At anthesis (GS60) and maturity (GS92), 20 plant samples were collected from each plot, divided into leaves, stems and sheaths, spikelet rachis and glume, and kernels, and then dried at 80°C until constant weight for dry matter (DM) determination. The parameters related to dry matter accumulation and distribution were calculated following Luo et al. (2018).[Image Omitted. See PDF][Image Omitted. See PDF][Image Omitted. See PDF][Image Omitted. See PDF]where DT is dry matter translocation post-anthesis, DA is dry matter accumulation post-anthesis, GY is grain yield at maturity (GS92), DTTG is the contribution of dry matter translocation post-anthesis to grain, and DATG is the contribution of dry matter accumulation post-anthesis to grain.

Statistical analysis of the data

SPSS 19.0 (SPSS, Inc., Chicago, IL, USA) was used to fit the grain-filling equation. Microsoft Excel 2010 was used for calculations, and Sigmaplot 10.0 (Systat Software, Inc., San Jose, CA, USA) was used to draw figures. Analysis of variance (ANOVA) procedures for a split-plot design were used to analyze the main and interactive effects of N fertilizer type (the main-plot factor) and fertilization practice (the subplot factor) using DPS 7.05 (Zhejiang University, Hangzhou, China). A cross-year ANOVA was applied to analyze N fertilizer type and fertilization practice effects cross years and the interaction with year, where N fertilizer types and fertilization practices were regarded as fixed effects and years and replicates were random effects. Data from each sampling were analyzed separately. When a significant year ×treatment interaction was not observed, the two years of data were pooled for mean comparisons. Means were tested by least significant difference tests at p < 0.05 (LSD_0.05).

RESULTS Grain yield

Year, fertilizer type, and fertilization practice significantly affected the grain yield of wheat, but no significant interaction between them was observed (Figure 1). PCU P2 achieved the highest grain yield of 8321.60 kg ha⁻¹ in 2017–2018 and 9456.51 kg ha⁻¹ in 2018–2019, and significant differences between PCU P2 and UF P2 were observed. The effect of fertilization practice on grain yield was much higher than that of CRNF type. Regardless of CRNF type, P2 significantly increased grain yield compared with P1. It was observed that both spikes and TGW had significant positive correlations with grain yield in the two years (Table 2). The path coefficient analysis showed that the direct path coefficients for yield components with respect to grain yield were all positive and were in the order of TGW > spikes > kernels per spike, which indicated that the contributions of yield components to grain yield were TGW > spikes > kernels per spike (Table 2).

TABLE 2 Path coefficient analysis showing direct and indirect effects of yield components on grain yield of wheat

Yield component	Correlation coefficient with yield	Direct path coefficient	Indirect path coefficient
			Spikes	Kernels per spike	TGW	Total
2017–2018
Spikes	0.914*	0.446	–	−0.272	0.741	0.469
Kernels per spike	−0.635ns	0.319	−0.380	–	−0.573	−0.954
TGW	0.979**	0.793	0.416	−0.231	–	0.186
2018–2019
Spikes	0.982**	0.555	-	−0.230	0.657	0.427
Kernels per spike	−0.752ns	0.297	−0.430	-	−0.619	−1.050
TGW	0.954**	0.694	0.525	−0.265	-	0.260

Note:

TGW, 1000-grain weight. ^ns, no significant; ^*, significant at p < 0.05; ^**, significant at p < 0.01.

SPAD value of the flag leaves

Figure 2 shows the influence of different types and application practices of CRNF on the SPAD values obtained for the flag leaves after anthesis (GS60). At the anthesis stage (GS60), the type and application practice of CRNF showed no significant difference in the SPAD value of the flag leaves (Table S1). However, over time, the decline rate of the SPAD value in P1 gradually increased compared with P2. Compared with those of PCU P1, SCU P1, and UF P1, the SPAD values at 35 DAA of PCU P2, SCU P2, and UF P2 were 33.65%, 35.62%, and 38.07% higher on average in the two years, respectively. The types of CRNF significantly affected the SPAD value at the later stage of grain filling, with PCU maintaining obviously higher SPAD values at 28 DAA and 35 DAA compared with SCU and UF.

Antioxidant enzyme activities of the flag leaves

Figure 3 shows the influence of different types and application practices of CRNF on the antioxidant enzyme activities of the flag leaves post-anthesis. The MDA content of the flag leaves post-anthesis showed a gradual increase over time, increasing slowly from 0 DAA to 14 DAA and then increasing rapidly thereafter. The types of CRNF showed no significant difference in MDA content from 0 DAA to 14 DAA, but a significant difference was observed at 21 DAA and 28 DAA (Table S2). Compared with P1, P2 significantly delayed the increase in MDA content from 14 DAA to 28 DAA in three types of CRNF.

The CAT activity of the flag leaves post-anthesis showed a trend of a unimodal curve with the increase in days after anthesis (GS60), which increased from 0 DAA to 7 DAA and then gradually decreased (Figure 3). The types and application practices of CRNF showed no significant effect on the CAT activity of the flag leaves at the anthesis stage (GS60), but a significant effect was detected from 14 DAA to 28 DAA (Table S2). At the milk-ripe stage (GS75), The CAT activity in P2 was significantly higher than that in P1 in PCU, SCU, and UF, which played an important role in scavenging hydrogen peroxide in cells, thus reducing damage to flag leaves caused by hydrogen peroxide. In all treatments, PCU P2 had the highest CAT activity at 28 DAA, which was 52.32% higher than PCU P1, on average, in the 2 years.

The POD activity of the flag leaves post-anthesis showed a rapidly increasing trend from 0 DAA to 14 DAA, followed by a rapid decrease (Figure 3). Among the three types of CRNF, the POD activity of the flag leaves in P2 was significantly higher than that in P1 from 7 DAA to 28 DAA, which was beneficial to eliminate the toxicity caused by hydrogen peroxide, phenols, and amines. The effect of CRNF type on POD activity was mainly reflected in the middle and late stage of grain filling, with a significant difference from14 DAA to 28 DAA (Table S2).

The SOD activity of the flag leaves post-anthesis exhibited a similar trend to CAT activity, which increased from 0 DAA to 7 DAA and then gradually decreased (Figure 3). The types of CRNF had significant effects on SOD activity from 14 DAA to 28 DAA, while fertilization practices showed a significant difference from 0 DAA to 28 DAA (Table S2).

Photosynthetic characteristics of the flag leaves

The photosynthetic parameters of the flag leaves are shown in Figure 4. The net photosynthetic rate, stomatal conductance, and transpiration rate showed similar trends, in that they decreased significantly in the milk-ripe stage (GS75) compared with at the anthesis stage (GS60), and the decrease of P1 was significantly higher than that of P2. The intercell CO₂ concentration showed the opposite trend to the net photosynthetic rate, increasing in the milk-ripe stage (GS75), and PCU P2, SCU P2, and UF P2 were significantly higher than PCU P1, SUC P1, and UF P1 in the two years. At the anthesis stage (GS60), PCU P2, SCU P2, and UF P2 showed no significant difference in the net photosynthetic rate, stomatal conductance, transpiration rate, and intercell CO₂ concentration, while at the milk-ripe stage (GS75), PCU P2 showed obvious advantages in photosynthetic parameters, with significantly higher net photosynthetic rate and stomatal conductance values compared with UF P2. As indicated in Table S3, the types of CRNF had no significant effect on the photosynthetic parameters at the anthesis stage (GS60), but the effect was significant at the milk-ripe stage (GS75). Fertilization practices showed significant effects on photosynthetic parameters at both anthesis (GS60) and the milk-ripe stage (GS75).

Dry matter accumulation and distribution

Under the different types of CRNF, dry matter translocation post-anthesis in P2 showed a lower trend than that in P1, but significant difference was only observed in PCU of 2017–2018 (Table 3). In the three types of CRNF, dry matter accumulation post-anthesis in P2 was significantly higher than that in P1. Both CRNF type and fertilization practice showed significant effects on translocation and accumulation of dry matter post-anthesis. In 2017–2018, the contributions of dry matter accumulation post-anthesis to grain in P2 were significantly higher than that in P1 for the three types of CRNF, and similar trends were observed in 2018–2019. Compared with dry matter accumulation post-anthesis, the contribution of dry matter translocation post-anthesis to grain showed an opposite trend in P1 and P2. Under twice-split fertilization, the contributions of dry matter accumulation post-anthesis to grain in PCU were higher than that in SCU and UF, but the difference was not significant.

TABLE 3 Dry matter translocation from vegetative organ to kernels post-anthesis under different types and application practices of controlled-release nitrogen fertilizer (CRNF)

Fertilizer	Fertilization practice	DT (kg ha−1)	DA (kg ha−1)	DTTG (%)	DATG (%)
2017–2018
PCU	P1	2448.36a	5467.17b	30.95ab	69.05bc
	P2	1956.92b	6620.34a	22.82c	77.18a
SCU	P1	2214.90ab	4731.67c	31.89a	68.11c
	P2	2108.98b	5964.67ab	26.13bc	73.87ab
UF	P1	2221.32ab	4789.62c	31.68a	68.32c
	P2	2188.29ab	6271.02a	25.88c	74.12a
2018–2019
PCU	P1	2289.96ab	5793.39c	28.34ab	71.66bc
	P2	1805.85b	8206.55a	18.03c	81.97a
SCU	P1	2436.19a	5102.68c	32.31a	67.69c
	P2	2200.66ab	6981.86b	23.96bc	76.04ab
UF	P1	2631.77a	5211.44c	33.57a	66.43c
	P2	2376.01a	7211.90ab	24.80bc	75.20ab
F-value
Year (Y)		10.67ns	161.64*	7.68ns	7.68ns
Fertilizer type (F)		9.56*	20.38**	12.43*	12.43*
Fertilization Practice (P)		9.81*	795.59**	162.09**	162.09**
Y×F		31.27**	0.65ns	2.90ns	2.90ns
Y×P		1.44ns	45.33**	4.33ns	4.33ns
F×P		5.32*	1.35ns	1.24ns	1.24ns
Y×F×P		0.49ns	3.63ns	0.03ns	0.03ns

Abbreviations: DA, dry matter accumulation post-anthesis; DATG, contribution of dry matter accumulation post-anthesis to grain; DT, dry matter translocation post-anthesis; DTTG, contribution of dry matter translocation post-anthesis to grain; P1, single application of CRNF; P2, twice-split application of CRNF; PCU, polymer-coated urea; SCU, sulfur-coated urea; UF, urea-formaldehyde.

Values within a column and for the same year followed by different letters are significantly different at p < 0.05. ^ns, no significant; ^*, significant at p < 0.05; ^**, significant at p < 0.01.

Grain-filling characteristics

The decision coefficients of the grain-filling process equation for different types and application practices of CRNF were above 0.99, indicating that the Richards equation suitably fitted the wheat grain-filling process (Table S4). A represented the ultimate 1000-grain weight fitted by the equation, which showed that P2 had obvious advantages in grain weight compared with P1. The grain-filling characteristic parameters of wheat differed obviously under the different types and fertilization practices of CRNF (Table 4). The effects of fertilization practices for the three types of CRNF on the active filling duration, maximum filling rate, and average filling rate were all in the order of P2 > P1. Days reaching the maximum filling rate of P1 appeared earlier than that of P2, and the maximum filling rate of P2 was significantly higher than that of P1. When applied twice, PCU had a longer active filling duration and higher average filling rate compared with SCU and UF in both years. As shown in Table 4, the grain-filling process was divided into the slow-increase, fast-increase, and slight-increase periods according to the Richards equation. During the three grain-filling periods, the fertilization practices had no significant effect on the duration of slow-increase period, but significantly affected the duration of fast-increase and slight-increase periods, and the average filling rate of P2 was all significantly higher than that of P1 in the fast-increase and slight-increase periods. The grain-filling contribution to the ultimate grain weight was in the order of fast-increase >slow-increase >slight-increase period, being higher in the slow-increase period for P1 but lower in the fast-increase and slight-increase periods of P1 compared with P2, but no significant difference was observed.

TABLE 4 Grain-filling characteristics on per unit area basis under different types and application practices of controlled-release nitrogen fertilizer (CRNF) in 2017–2018 and 2018–2019

Fertilizer	Fertilization practice	T (d)	G (kg ha−1 d−1)	Tmax•G (d)	Gmax (kg ha−1 d−1)	Slow-increase period			Fast-increase period			Slight-increase period
						T1 (d)	G1 (kg ha−1 d−1)	P1 (%)	T2 (d)	G2 (kg ha−1 d−1)	P2 (%)	T3 (d)	G3 (kg ha−1 d−1)	P3 (%)
2017–2018
PCU	P1	37.79ab	299.31d	19.18bc	454.58d	13.34a	164.02c	26.28a	11.69ab	400.11 cd	56.18a	12.77ab	114.40bc	17.54a
	P2	38.92a	326.75a	19.91a	495.96a	13.96a	172.22ab	26.07a	11.89ab	436.46a	56.27a	13.06a	124.69a	17.66a
SCU	P1	37.22b	291.40e	19.09c	444.25e	13.25a	171.77ab	27.66a	11.67ab	391.40d	55.52a	12.30b	112.58c	16.83a
	P2	38.48ab	318.01b	19.77a	483.62b	13.84a	174.57a	26.77a	11.86ab	425.82b	55.94a	12.78ab	122.01a	17.29a
UF	P1	37.09b	291.87e	19.04c	444.54de	13.27a	167.11bc	27.31a	11.55b	391.56d	55.68a	12.27b	112.46c	17.00a
	P2	38.68a	305.94c	19.62ab	464.94c	13.61a	171.13ab	26.52a	12.03a	409.30c	56.06a	13.05a	117.15b	17.42a
2018–2019
PCU	P1	41.87ab	291.50c	21.10abc	441.57c	14.69a	151.54c	25.32ab	12.82bc	388.39c	56.62ab	14.36ab	110.60c	18.06a
	P2	43.82a	320.09a	21.67a	483.60a	14.96a	162.22ab	24.30b	13.42ab	425.03a	57.08a	15.44a	120.52a	18.62a
SCU	P1	42.08ab	273.34d	20.80c	414.40d	14.19a	153.96bc	25.63ab	13.21abc	364.57d	56.48ab	14.67ab	103.96d	17.89a
	P2	43.65a	315.59a	21.50ab	477.35a	14.73a	167.74a	24.75ab	13.53ab	419.68a	56.88ab	15.38a	119.23ab	18.37a
UF	P1	40.51b	269.53d	20.96bc	410.85d	14.67a	154.29bc	27.62a	12.57b	361.96d	55.53b	13.26b	104.10d	16.85a
	P2	43.59a	307.79b	21.45ab	465.82b	14.66a	166.77a	24.97ab	13.58a	409.61b	56.78ab	15.35ab	116.47b	18.25a
F-value
Year (Y)		916.07**	195.89*	291.59*	175.64**	7.45ns	9.51ns	7.34ns	108.36**	753.63**	7.45ns	7.71ns	557.99**	7.13ns
Fertilizer type(F)		2.41ns	11.67**	2.70ns	90.84**	1.49ns	2.81ns	1.83ns	0.74ns	80.55**	1.94ns	0.88ns	31.26**	1.76ns
Fertilization Practice (P)		37.75**	847.16**	42.77**	544.81**	6.80ns	65.53**	5.42ns	14.77**	77.54**	5.51ns	8.98*	230.02**	5.45ns
Y × F		1.01ns	1.46ns	0.36ns	1.43ns	0.65ns	0.46ns	0.65ns	1.17ns	1.36ns	0.68ns	0.79ns	0.95ns	0.63ns
Y × P		2.17ns	46.62**	0.17ns	28.56*	0.68ns	12.18*	0.87ns	1.92ns	24.61**	0.81ns	1.52ns	10.85*	0.93ns
F × P		0.95ns	5.89*	0.21ns	4.47ns	0.58ns	0.12ns	0.47ns	1.36ns	4.10ns	0.50ns	0.68ns	2.59ns	0.42ns
Y × F × P		0.34ns	10.44*	0.11ns	6.73*	0.14ns	1.20ns	0.39ns	0.22ns	5.91*	0.45ns	0.26ns	2.92ns	0.35ns

Note:

G, average filling rate; G₁-G₃, filling rate of slow-increase, fast-increase, and slight-increase period, respectively; G_max, maximum filling rate; P1, single application of CRNF; P₁-P₃, filling contribution ration of slow-increase, fast-increase, and slight-increase period, respectively; P2, twice-split application of CRNF; T, active filling duration; T₁-T₃, duration of slow-increase, fast-increase, and slight-increase period, respectively; T_max•G, days reaching the maximum filling rate; PCU, polymer-coated urea; SCU, sulfur-coated urea; UF, urea–formaldehyde.

Values within a column and for the same year followed by different letters are significantly different at p < 0.05. ^ns, no significant; ^*, significant at p < 0.05; ^**, significant at p < 0.01.

Table 5 shows the correlations between the grain-filling parameters and the yield. The active filling duration, average filling rate, maximum filling rate, and days reaching the maximum filling rate were significantly positively correlated with the yield. In the slow-increase, fast-increase, and slight-increase periods, the duration of slow-increase and slight-increase period and the filling rate of fast-increase and slight-increase period showed significant positive correlations with the yield in 2017–2018; the filling rate of slow- and fast-increase period, the duration of fast- and slight-increase period and the filling contribution ration of fast-increase period showed significant positive correlations with TGW in 2018–2019. A negative correlation was observed between the filling contribution ratio of slow-increase period and TGW in the two years. The results of the two years suggested that the filling rate of fast-increase and slight-increase period and the duration of slight-increase period could play an important role in the formation of grain weight.

TABLE 5 Correlation coefficients between grain-filling parameters and yield

Parameter	Yield
	2017–2018	2018–2019
Active filling duration (T)	0.925**	0.945**
Average filling rate (G)	0.985**	0.956**
Days reaching the maximum filling rate (Tmax•G)	0.983**	0.947**
Maximum filling rate (Gmax)	0.985**	0.957**
Duration of slow-increase period (T1)	0.993**	0.584ns
Filling rate of slow-increase period (G1)	0.613ns	0.929**
Filling contribution ration of slow-increase period (P1)	−0.693ns	−0.785
Duration of fast-increase period (T2)	0.748ns	0.874*
Filling rate of fast-increase period (G2)	0.985**	0.957**
Filling contribution ration of fast-increase period (P2)	0.687ns	0.778ns
Duration of slight-increase period (T3)	0.804*	0.877*
Filling rate of slight-increase period (G3)	0.985**	0.958**
Filling contribution ration of slight-increase period (P3)	0.699ns	0.791ns

Note:

^ns, no significant; ^*, significant at p < 0.05; ^**, significant at p < 0.01.

DISCUSSION Contributions of grain weight to wheat grain yield under different types and application practices of CRNF

In the present study, we compared the effects of three types of CRNF under two fertilization practices and found that the effect of fertilization times on grain weight and yield was greater than that of CRNF type, and the twice-split application of CRNF significantly increased grain weight and wheat yield compared with the single application of CRNF (Figure 1 and Table S4). This result may be due to the difference in N supply in the early and later growth stage of wheat under different fertilization practices. Our previous study showed that different fertilization times of CRNF resulted in differences in nutrient release from fertilizer and N uptake in winter wheat, significantly affecting the spikes per unit area and grain weight (Ma et al., 2021). This study further found that the type of CRNF also significantly influenced grain weight (F_2017-2018 = 165.88, p < 0.01) and wheat yield (F = 15.10, p < 0.01), indicating that the differences in N release mechanism of CRNF also affected its N supply to wheat, thus affecting grain weight and wheat yield (Trenkel, 2010). PCU or SCU is prepared by wrapping a protective coating around the urea nucleus, and their N release performance is mainly controlled by the properties and thickness of the coating material but is also affected by environmental factors, such as temperature, moisture, and soil pH (Trenkel, 2010). According to Azeem et al. (2014), the nutrient release of polymers and sulfur coatings follows a “diffusion mechanism” and “failure mechanism”, respectively, which could explain why the patterns of N release from PCU and SCU exhibited sigmoidal release and parabolic release patterns, respectively (Trenkel, 1997). UF, based on urea and formaldehyde condensation compounds, releases N by the enzymatic hydrolysis of urease or other biological catalysts (Chalk et al., 2015). The degradation and subsequent N release from UF are driven by the size and activity of the soil microflora and factors that influence microbial activity, such as soil moisture and temperature (Alexander & Helm, 1990; Trenkel, 1997). N release from UF is unstable due to the large fluctuation and randomness of these factors in the soil (Trenkel, 1997). The different nutrient release characteristics of the three CRNFs resulted in distinctions in N loss in the soil and N uptake by crops. In contrast to SCU and UF, the sigmoidal release of PCU avoided the waste of N in the early stage and ensured sufficient N supply in the middle and later stage, and contributed to the improvement of N use efficiency, thus promoting grain filling and crop yield (Trenkel, 2010).

Under the conditions of this study, TGW was the main contributing factor to wheat yield, followed by spikes per hectare and kernels per spike (Table 2). The results indicated that the increase of grain weight caused by the twice-split application of CRNF was of great significance for the improvement of wheat yield (Figure 1). Yan et al. (2019) also emphasized the importance of increasing grain weight through regulating N supply in the later growth stage for grain yield improvement. In this study, the twice-split application of CRNF coordinated N supply in the middle and later stage of wheat growth, which improved the contribution of grain weight to wheat yield to some extent. Significant positive correlations were observed between TGW and grain yield in both years (Table 2), and grain weight in P2 was significantly higher than that in P1 among the three types of CRNF, which confirmed the importance of grain weight improvement for maximizing yield. However, the effects of strong interactions between yield components cannot be ignored on grain yield (Cao et al., 2019). Spikes per hectare and TGW were observed to have a positive indirect effect on grain yield by influencing each other, while grains per spike showed a negative indirect effect on grain yield through spikes per hectare and TGW, respectively (Table 2). Hence, the interaction between spikes per hectare and TGW is another possible reason for yield increase under the twice-split application of CRNF.

Senescence of the flag leaves and contributions of photosynthetic products to grains post-anthesis under different types and application practices of CRNF

The original intention of CRNF development was to meet N demands for the entire growing season with a single application prior to sowing, so as to save labor input and improve N utilization efficiency and yield (Trenkel, 1997). However, due to the limitation of growth period, the actual application effect of CRNF applied once in wheat is controversial (Ma et al., 2021). By adjusting the fertilization practices, the present study investigated whether the twice-split application of CRNF could coordinate the N supply in the later growth stage of wheat in order to postpone flag leaf senescence, prolong the photosynthetic validity period, and increase the photosynthetic rate, so as to meet the photosynthate accumulation required for grain filling. Studies have revealed the important effects of leaf senescence on grain weight and crop yield formation (Yan et al., 2019). A study of tropical maize under limiting N conditions indicated that leaf senescence accounted for 42% of the variation in grain yield (Bänziger & Lafitte, 1997). The results of this study showed that MDA content was contrary to the trend of the SPAD value in the flag leaves and was basically consistent with the process of flag leaf senescence (Figures 2 and 3). Compared with P1, P2 significantly decreased the accumulation of MDA from 21 DAA to 28 DAA, by increasing antioxidant enzyme (e.g., SOD, CAT, and POD) activities. Crop senescence is sink-driven, and the patterns of leaf senescence modulate N fluxes from the senescing leaves to the grains, which is accelerated by N deficiency (Kitonyo et al., 2018). Sinclair et al. (1990) also pointed out that during the source–sink transition, N demand by the grains could accelerate leaf senescence. Nutrient stress due to a lack of N was found to result in leaf premature senescence, and on the contrary, high levels of N delayed the initiation of flag leaf senescence and prolonged the leaf functional period (Luo et al., 2018). Under the twice-split application of CRNF, PCU showed better advantages than SCU and UF in delaying the decrease in SPAD value, reducing the accumulation of MDA, and improving the activities of antioxidant enzymes in the flag leaves (Figures 2 and 3). It could be concluded that the effects of different types and application practices of CRNF on flag leaf senescence post-anthesis were essentially related to the regulation of N release, which ensured sufficient N supply in the later growth stage of wheat and prolonged the photosynthetic duration of the wheat flag leaves.

Photosynthesis is the key driving force affecting the accumulation and distribution of dry matter, and the effects of N on matter accumulation and grain filling may be primarily related to the duration of photosynthesis and the photosynthetic rate (Makino, 2011). In the present study, a higher net photosynthetic rate in the flag leaves of P2 was observed at the anthesis stage (GS60) compared with that in P1, and the gap between them was further widened at the milk-ripe stage (GS75) (Figure 4). There are three explanations for this situation. First, the twice-split application of CRNF with sufficient N supply made the physiological activities in the flag leaves more active, which increased the stomatal conductance, reduced cell resistance, and improved the transpiration rate. Second, it increased the absorption and utilization of carbon dioxide (CO₂) by photosynthetic cells. In addition, the accelerated leaf senescence in P1 limited the maintenance of photosynthetic capacity and reduced the duration of photosynthesis. Previous studies demonstrated that 90% of dry matter emanates from photosynthesis, and the accumulation of photosynthates post-anthesis is key to wheat yield formation (Gaju et al., 2016). In the present study, fertilization practices significantly influenced translocation and accumulation of dry matter post-anthesis. Compared with P1, P2 decreased translocation of dry matter post-anthesis, but obviously improved accumulation of dry matter post-anthesis, and the contributions of dry matter translocation and accumulation post-anthesis to grains in P1 and P2 showed the same trend (Table 3). Boosting grain dry matter accumulation required an improvement in the allocation of assimilates to the sink organs to enhance spike growth (Sierra-Gonzalez et al., 2021). Dry matter partitioning depends on the number and activity of assimilating sinks, which is closely related to the extended duration of photosynthesis and the improvement of photosynthetic rate (Kim et al., 2011; Liu et al., 2020). It is noteworthy that the contribution of pre-anthesis assimilates to the grains was significantly increased under the single application of CRNF, which probably supported a low rate of photosynthesis that resulted in the insufficient supply of post-anthesis assimilates for grain filling, thus increasing the need for the translocation of dry matter accumulation pre-anthesis (Ercoli et al., 2008). Liang et al. (2017) also pointed out that leaf senescence, which greatly affects the photosynthetic process, shows strong interactions with carbohydrate assimilation and transport post-anthesis. Overall, the twice-split application of CRNF showed advantages in preventing the premature senescence of the leaves and enhancing photosynthetic capacity, which significantly promoted dry matter accumulation and transport post-anthesis to the grains.

Regulation of types and application practices of CRNF on the grain-filling process

Grain filling is a critical physiological process that determines the grain weight and yield of wheat, and the grain-filling rate and effective filling duration affect the filling degree of grain storage capacity (He et al., 2020). In the present study, both the filling duration and average filling rate were observed to be higher in P2 than those in P1, indicating that the twice-split application of CRNF was conductive to grain filling and contributed greatly to the formation of grain weight (Table 4). The difference in TGW between P2 and P1 observed in this study confirmed the above conclusion. PCU P2 achieved the highest TGW, which could be attributed to the highest filling duration and average filling rate. The filling rate is considered to be mainly controlled by genetics and is positively correlated with grain weight, while grain filling duration is strongly affected by environmental factors (Li et al., 2020). However, some researchers have shown that N fertilizer has significant effects on the maximum grain-filling rate and the average grain-filling rate. With the increase in N level, the maximum grain-filling rate and the average grain-filling rate increased and then exhibited a slight decrease. Instead, N fertilizer also had significant effects on the time of appearance of the maximum filling rate and ultimate TGW (Yan et al., 2019). Due to the difference in nutrient supply in the later stage, different types of CRNF significantly affected the average filling rate and maximum filling rate, and PCU showed obvious advantages compared with SCU and UF (Table 4). In the three filling periods, CRNF type had no significant effect on the filling duration and filling contribution ratio, but it significantly influenced filling rate in fast- and slight-increase periods. Plant senescence is closely related to the duration of grain filling, and postponing the premature senescence of the leaves is beneficial to prolonging the effective filling duration (Luo et al., 2018). Dry matter accumulation post-anthesis has been proved to be the main matter source of grain filling and significantly affects the maximum and average grain-filling rate (Luo et al., 2018), which is consistent with our observation. The increase in grain weight in the twice-split application of CRNF could be attributed to the extension of filling duration by delaying leaf senescence and the increase in grain-filling rate by improving the assimilate accumulation post-anthesis. Correlation coefficient analysis showed that grain yield was significantly correlated with the filling rate in the fast-increase and slight-increase periods, and with the filling duration in the slight-increase period, indicating that improving the grain-filling capacity in the two periods had positive effects on grain weight and yield formation (Table 5).

Horie et al. (2005) proposed that the reserve of NSC pre-anthesis was significantly correlated with crop growth at the early stage of grain filling. The current assimilation post-anthesis may not supply sufficient NSC for grain filling at the early grain-filling stage, because the initiation of grain filling requires a large supply of NSC (Zheng et al., 2010). In the case of insufficient supply, the NSC reserves stored in the stem pre-anthesis may be transported from the stem to the spikelet to supply grain filling. This can explain the tendency of the contribution ratio of grain filling in P1 to be higher than that in P2 in the early stage. P1 made up for the deficiency of current assimilation at the early grain-filling stage through increasing the transport of dry matter pre-anthesis to the grains (Tables 3 and 4). Some studies revealed that the poor grain filling of the inferior grains was the primary factor inhibiting the promotion of the grain weight of cereals, and thus promoting the grain filling of the inferior grain might be key for promoting the grain weight of wheat (Liang et al., 2017). The inferior grain is more sensitive to environmental factors than the superior grain, and the processes of cell division, differentiation, enlargement, and enrichment of inclusions during the development of inferior and superior grains differ significantly and are under the regulation of a range of hormones (zeatin riboside, indole-3-acetic acid, ethylene, etc.), starch synthesis-related enzymes (sucrose synthase, ADP-glucose pyrophosphorylase, etc.), and their interactions (Ahmad et al., 2018; Wei et al., 2019). Therefore, our further studies may focus on investigating the regulation of hormones and starch synthesis-related enzymes on the grain filling of superior and inferior grains under different CRNF application practices.

CONCLUSIONS

The present study demonstrated that grain weight is an important contributing factor to wheat yield, and improving grain filling to increase TGW is a crucial means of enhancing grain yield in wheat. The effects of different types and application practices of CRNF on the grain-filling process and grain weight formation of wheat were significantly different. Compared with the single application of CRNF, the twice-split application of CRNF increased the antioxidant enzyme activities to postpone flag leaf senescence and improve the flag leaf photosynthetic capacity post-anthesis, which was conducive to the accumulation and transport of assimilates to the grains. Due to rational N supply, the twice-split application of CRNF enhanced the grain-filling rate and filling duration, especially in the middle and late stage of grain filling, resulting in a significant increase in grain weight compared with a single application of CRNF. Compared with SCU and UF, PCU showed better performance in the net photosynthetic rate at the milk-ripe stage (GS75), dry matter accumulation post-anthesis, active filling duration, maximum filling rate, and average filling rate, which contributed to the formation of grain weight, and also obtained a higher TGW and grain yield. Therefore, promoting the twice-split application of CRNF, especially PCU, may be a critical strategy for the high-yield cultivation of winter wheat.

ACKNOWLEDGMENTS

This work was jointly supported by the National Key Research and Development Program of China (2018YFD0200500); the Pilot Projects of the Central Cooperative Extension Program for Major Agricultural Technologies; the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD); the Science and Technology Innovation Team of Yangzhou University; the Earmarked Fund for Jiangsu Agricultural Industry Technology System (JATS[2021]503); the Postgraduate Research & Practice Innovation Program of Jiangsu Province (XKYCX20_020). We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

CONFLICT OF INTEREST

The authors declare that they have no competing interests.

AUTHOR CONTRIBUTIONS

Quan Ma: Conceptualization, Investigation, Formal analysis, and Writing-original draft. Quan Sun: Conceptualization, Investigation, and Formal analysis. Xinbo Zhang: Investigation. Fujian Li: Investigation. Yonggang Ding: Investigation. Rongrong Tao: Investigation. Min Zhu: Conceptualization. Jinfeng Ding: Conceptualization. Chunyan Li: Conceptualization and Resources. Wenshan Guo: Conceptualization and Supervision. Xinkai Zhu: Conceptualization, Writing-review & editing, Supervision, and Project administration.