Field experiments were carried out in the 2019 and 2020 rice seasons in Mingguang, Chaohu and Guichi counties in Anhui Province, located in the Yangtze River Delta, China, which belongs to the northern subtropical monsoon climate. The annual average temperature and precipitation are 15.2°C and 953 mm in Mingguang, 16.1°C and 1030 mm in Chaohu, and 16.1°C and 1400 mm in Guichi, respectively ( Figures 1A, B ). The physical and chemical properties of the topsoil (0-20 cm) in the experimental sites are shown in Table 1 .

The CRNs (44.5% N) were purchased from Anhui Moith Agricultural Technology Co., Ltd. in China centered on a urea pellet coated with polyurethane. In this study, two N release periods of CRNs were used. Two CRNs with the N release periods of 40 and 90 days are abbreviated as CRN1 and CRN2, respectively. The N release characteristics of CRN1 and CRN2 in 25°C water and paddy fields for 42 and 100 days were measured in the 2019 (Yuan et al., 2023).

A split-plot design was used, with the main plot being the type of N fertilizer and its application method, and the subplot being N application rate. The numbers of main plot and subplot per replicate were 2 and 5, respectively. The plot size was 5 m × 6 m. The main plot treatments were as follows: 1. blended CRN treatment: the conventional N fertilizer, CRN1 and CRN2 with a 4:3:3 ratio of the N rate applied at one-time, and 2. conventional N fertilizer treatment: the conventional N fertilizer applied at three times as basal, tillering and panicle fertilizer. The subplot treatments were N application rates at 0, 60, 120, 180, and 240 kg N ha^-1. The 0 rate was the same for both two main plot factors. Therefore, there were 9 treatments in the experiment. The amount and application method of N fertilizer in the different treatment are shown in Table 2 . Triple superphosphate and potassium chloride were applied as basal fertilizers at a rate of 39 kg P ha^-1 and 62 kg K ha^-1, respectively.

The rice variety and timing of the planting and management operations in this study are shown in Table 3 . All rice varieties used in this study were high-yield super indica rice, which are commonly grown in the region. The spacing of transplanted rice was 0.30 m (length) by 0.13 m (width).

At the anthesis and maturity, the number of tillers per unit area in each plot was measured. Based on the average number of tillers per hole, rice from three representative holes were selected for destructive sampling. The samples were divided into two parts: leaf and stem, and panicle. The panicle samples at maturity were used to determine the yield components, including the number of panicles (NP), the number of spikelets per panicle (NSP), seed-setting rate (SR), and 1000-grain weight (GW). Thereafter, the samples were dried at 105°C for 30 min and then dried continuously at 85°C for 48 h and weighed. The biomass weight was the sum of dry matter weight of shoots and panicles (Yuan et al., 2021). The dry matter samples were digested with H₂SO₄-H₂O₂. The N concentrations in the digests were determined by the Kjeldahl method (Bremner and Mulvaney, 1982).

At the final stage, the entire length of 20 rows of every plot was harvested manually to obtain the seed weight and moisture content. The rice grain yield was determined at 14% moisture content (Yang et al., 2006).

The NUE was calculated according to the following equation (Sun et al., 2012):

Based on the dry matter (DM) and N concentration, the accumulation and translocation parameters of DM and N after anthesis stage were calculated according to the following equations (Ntanos and Koutroubas, 2002; Jiang et al., 2004):

Economic benefit calculation: the output value was calculated according to the average yield and the average price of rice sold in the 2019 and 2020 (2.5 Yuan kg^-1). The labor cost of rice fertilization was 900 Yuan ha^-1 each time. The fertilizer cost was calculated according to the prices and amount of fertilizer. The average prices of urea, CRN, potassium and phosphorus fertilizer were 2.0, 2.8, 2.0 and 0.6 Yuan kg^-1 in 2019 and 2020, respectively. The other rice production costs included seed cost, irrigation cost, and disease, pest and weed management costs, etc, which were the same under all the treatments for two years. Moreover, this study mainly focused on the impact of costs associated with the type and application method of N fertilizer on the economic benefits of rice. Therefore, the net economic benefit this study was a partial net economic benefit due to the fact that the other rice production costs was not calculated.

The net economic benefit was calculated according to the following equation:

Rice yield, NUE, and the accumulation and translocation parameters of dry matter and N after the anthesis stage were analyzed as split-plot analyses of variance (Equation 1), where the type and application method of N fertilizer were assigned as the main plot factor, and the N application rate as the sub-plot factor. The analyses were performed separately for each site and location, and differences in treatment means were assessed using the least significant difference test. Also, the relationships between parameters were analyzed by using the Pearson correlation method. The analyses were conducted with the SPSS 20.0 software (SPSS 22.0, IBM Corp., Armonk, NY, USA), and all analyses were reported as significant at P ≤ 0.05. The figures were created using the Origin 2022 software (OriginLab Corp., Northampton, MA, USA).

Compared to the conventional N fertilizer at the equal N application rate, the blended CRN treatments increased the grain yield by an average of 0.9 - 6.7% at the three experiment sites across the two years ( Figures 2A-F , P< 0.05). The rice yield increased significantly with the increase of N application rate ( Figures 2A-F , P< 0.01) with no significant interaction of N type × rate ( Figures 2A-F ). The rice yield in 2020 was, on average, 18.3% lower than that in 2019. The N type had consistent effects on rice yields at the three experiment sites across the two years ( Figures 2A-F ).

The relationship between rice yield and N fertilizer rate can be simulated by a quadratic equation ( Figures 2A-F ). Under the blended CRN treatments, the highest average yield was 9604 kg ha^-1 at 200 kg N ha^-1 ( Figures 2A-F ). Compared with the conventional N fertilizer treatments, the maximum yield of the blended CRN treatments increased by 109-309 kg ha^-1, with an average increase of 1.6%, and the corresponding N application rate decreasing by 10-32 kg ha^-1, with an average decrease of 14.7% ( Figures 2A-F ).

Compared with the conventional N fertilizer, the blended CRN application significantly increased the number of spikelets per panicle and seed-setting rate by averages of 1.6% and 2.0%, respectively. In contrast, the blended CRN application decreased the number of panicles by 2.0% ( Table 4 , P< 0.05). Increasing N application rate significantly increased these two parameters ( Table 4 , P< 0.01). The N type had consistent effects on rice components across the three experiment sites and two years ( Table 4 ).

With the equal N application rate, NUE was significantly higher under the blended CNR than the conventional N fertilizer treatments ( Figures 3A-F , P< 0.05). The increases were 5.0-16.2%, 3.2-28.4%, 10.4-21.8%, and 4.7-18.7% at 60, 120, 180 and 240 kg N ha^-1, respectively. Increasing N application rate decreased NUE ( Figures 3A-F , P< 0.01). There was no interaction between N type and rate on NUE ( Figures 3A-F ). The treatment effects on NUE followed similar trend at the three experiment sites over the two years ( Figures 3A-F ). According to the quadratic equation, the blended CNR treatments had 31.9% higher than the conventional N fertilizer treatments at the highest average grain yield.

Compared with the conventional N fertilizer, the use of blended CRN (Equation 2) significantly increased pre-anthesis DM translocation (Equation 3) (Pre-DMT), pre-anthesis N translocation efficiency (Equation 4) (Pre-DMTE), and the contribution of Pre-DMT to grain yield (Pre-DMC) by 0.8-36.7%, 0.4-25.3%, and 0.7-37.0%, respectively, but decreased the contribution (Equation 5) of pre-anthesis DM accumulation to grain yield (Post-DMC) by 0.3-25.9% ( Supplementary Table S1 ; Table 5 , P<0.05). Increasing N dose significantly increased Pre-DMT, Post-DMA, Pre-DMTE and Pre-DMC ( Supplementary Table S1 , P<0.05), but decreased Post-DMC at Chaohu and Guichi experiment sites ( Supplementary Table S1 ; Table 5 , P< 0.05).

Compared with the conventional N fertilizer (Equation 6), the blended CRN increased the pre-anthesis N translocation (Pre-NT) and the contribution of pre-NT to grain N (Equation 9) (Pre-NC) by 1.0-19.8% and 0.2-8.7%, respectively, while decreased the contribution of post-anthesis N translocation to grain N (Equation 7) (Post-NC) by 1.0-24.8% ( Supplementary Tables S2 ; Table 6 , P< 0.05). The fertilizer (Equation 8) type did not affect the pre-anthesis N translocation efficiency (Pre-NTE) ( Table 6 , P > 0.05). Increasing N rate significantly increased the Pre-NT, post-anthesis N uptake (Post-NU), and Post-NC, but decreased Pre-NTE and the contribution of pre-NT to grain N (Pre-NC) by 0.1-23.6% and 0.1-20.0%, respectively ( Supplementary Table S2 ; Table 6 , P< 0.05). The effects of N type and dose on N accumulation and translocation were consistent across the three experiment sites over two years ( Table 6 ).

Dry matter and N accumulation, translocation, and utilization are closely related to yield and NUE. Rice yield was correlated positively with NP, NSP, Pre-DMT, Pre-DMC, Pre-NT, Post-NU, Post-NC but negatively with Pre-DMTE, Post-DMC, Pre-NTE, Pre-NC and NUE ( Table 7 , P<0.05). The NUE was correlated positively with Post-DMC and Pre-NC, but negatively with yield, NP, Pre-DMT, Pre-DMC, Post-NU and Post-NC ( Table 7 , P<0.05).

The effects of N type and application rate on economic benefits were assessed by average yields across three sites over two years ( Table 8 ). The N fertilization increased fertilizer cost, economic output and net economic benefit increased by 38.5-179.0%, 33.5-67.5% and 21.5-68.8%, respectively (Equation 10). Compared with the conventional N fertilization, the blended CRN treatments reduced the labor cost by 1800 Yuan ha^-1, but increased the fertilizer cost, economic output and net economic benefit by 4.5-9.8%, 0.2-5.2% and 8.8-15.3%, respectively. Using quadratic equation simulation, the highest net benefit was 21450 Yuan ha^-1 under the blended CNR at an application rate of 189 kg ha^-1 and NUE of 47.3%. In comparison, the highest net economic benefit was 19161 Yuan ha^-1 under the conventional N fertilization at an application rate of 247 kg ha^-1 and NUE of 31.9%. The blended CNR treatment had 11.9% higher net profit, but the applied N rate was 23.6% lower N dose and NUE was 50.4% higher NUE than the conventional N fertilization.

The final yield of rice results from the translocation of pre-anthesis stored DM and post-anthesis accumulation of photosynthetic products (Kumar et al., 2006). Appropriate N fertilizer type, rate and management can improve the photosynthetic rate and increase biomass and N accumulation and translocation, and hence yield (Artacho et al., 2009; Wang et al., 2018; Ju et al., 2021). The blended CRN treatment increased pre-anthesis DM translocation (Pre-DMT) and the contribution of Pre-DMT to grain yield (Pre-DMC) compared with the conventional N fertilizer treatment ( Supplementary Tables S1 , Table 5 , P<0.05), which was consistent with the results of Wang et al. (2018). In addition, rice yield was positively correlated with Pre-DMT and Pre-DMC ( Table 7 , P<0.01), in agreement with the findings by Pal et al. (2017). This may be related to the fact that the blended CRN treatment decreased ineffective tillering to increase panicle number ( Table 4 ), improved post-anthesis root function, especially cytokinin transport to leaves, and increased the accumulation of photosynthetic assimilates (Gu et al., 2017). The blended CRN might play a role in enhancing the sink and source sizes.

Rice yield also depends on yield components, including the number of panicles (NP), the number of spikelets per panicle (NSP), seed-setting rate (SR), and 1000-grain weight (GW). Previous studies have shown that one-time application of CRN increased NP, NSP, and SR of rice, thereby increasing yield (Guang et al., 2018; Wang et al., 2018). In this study, compared with the conventional urea use, one-time application of CRN significantly improved rice yield, mainly due to the increased NSP ( Table 4 ). The positive correlation between NSP and yield in this study ( Table 5 ) was consistent with previous studies (Artacho et al., 2009; Lan et al., 2021). Although the blended CRN treatment of significantly reduced the NP ( Table 4 ), the contribution of the NP and NSP to rice yield is a trade-off process (Sui et al., 2013). Due to its slow N release at the tillering stage (Yuan et al., 2023), CRN decreased the NP of rice by 3.8-11.7% and increased the NSP by 11.8-21.8% during the heading stage compared to the common urea (Dun et al., 2023). Moreover, the lack of a significant N type effect on GW in this study ( Table 4 ; P>0.05) was consistent with previous reports (Sui et al., 2013; Wang et al., 2018), suggesting that the GW is a stable trait determined by rice genotype.

Nitrogen fertilization is an important measure to promote sustainable crop production (Li et al., 2017). There is not a simple linear relationship between N application rate and rice yield but the one-variable quadratic can describe this relationship well (Ju et al., 2021). According to the quadratic equation, the optimal N application rate differed slightly for different rice varieties under the conventional N fertilization. The optimal N application rates for the highest yields of rice varieties HD-5 and YJ-8 under the conventional N fertilizer treatment were 255 and 261 kg ha^-1, respectively (Ju et al., 2021). Under the blended CRN treatment, this study obtained the average optimal N application rate of 200 kg ha^-1, which was lower than previous results. Wang et al. (2018) recommended the N application rate of 150 kg ha^-1 for rice crop in the Yangtze River Basin of China to achieve a rice yield of approximately 7700 kg ha^-1. This yield was comparable to the rice yield of 7820 kg ha^-1 at the N application rate of 60 kg ha^-1 in our study, with the yield further increasing with increasing N dose to 120 and 180 kg ha^-1 (8540 and 9420 kg ha^-1, respectively). The reason for such a discrepancy in yield response to N fertilization was related to the different rice varieties. Wang et al. (2018) tested indica varieties of rice, while this study used super hybrid rice varieties. At the equivalent N application rate, the yield of super hybrid rice was higher than that of indica rice (Hu et al., 2020; Zhu et al., 2020). Therefore, rice varieties and their yield potential should be considered when the N fertilizer program is formulated.

Intensive agricultural production in developing countries faces the dual challenge of ensuring sustained increases in crop yields and reducing the environmental risks of excessive fertilizer inputs (Chen et al., 2011). Changing N application methods, such as deep application and split application, or fertilizer type, has achieved simultaneous improvement in rice yield and NUE, mainly through reducing N losses such as NH₃ volatilization, N runoff and NO₃ leaching (Huang et al., 2019; Liu et al., 2016; Wang et al., 2018; Yuan et al., 2023). This and previous studies demonstrated that the single application of CRN improved NUE compared with the urea split application (Wang et al., 2018; Huang et al., 2019). This improved NUE was mainly due to the slow N release and longevity (40 and 90 d) of CRNs in the paddy fields.

The NUE is closely related to grain N accumulation, which could be derived via translocation of the pre-anthesis stored N and the post-anthesis uptake of soil N (Jiang et al., 2005; Wei et al., 2017). In this study, Pre-NC under the blended CRN treatment was 74.0-88.7% which was significantly higher than that under the conventional N fertilizer treatment ( Table 6 ). The results implied that the blended CRN treatment was more conducive than the conventional N fertilizer treatment to enhancing the re-translocation of N from stems and leaves to grains. The increased yield and NUE under blended CRN fertilization had resulted from the increased Pre-NT ( Table 6 ), improved photosynthetic rate, delayed leaf senescence, and increased post-anthesis carbon assimilation (Wu et al., 2018).

Economic benefit is one of the important factors influencing farmers to adopt fertilizer management practices (Zhang et al., 2021). The relatively high price of CRNs limits their use in farming practices such as rice production (Naz and Sulaiman, 2016). However, this study suggested that the blended CRN treatments could improve the economic benefits for rice cultivation ( Table 8 ). With accelerated urbanization, labor for agricultural cultivation has become expensive (Chen et al., 2014). Compared with the conventional N fertilization, the use of blended CRN could save labor costs. Under the blended CRN treatment, labor cost saving offset the increased fertilizer cost, which ultimately increased the profitability for the rice growers ( Table 8 ). It appears to be an effective way to achieve high grain yield and NUE by optimizing the N management mode of CRN and urea blended application (Guo et al., 2019; Huang et al., 2019; Zhang et al., 2021).