The experiment was conducted in the Experimental Base of the State Key Laboratory of Hybrid Rice in Changsha, Hunan Province, China, located at coordinates 28°11′46.97″N 113°05′4.54″E. The soil has a pH of 6.02, a concentration of fast-acting phosphorus of 23.8 mg·kg^-1, a concentration of fast-acting potassium of 90.0 mg·kg^-1, a concentration of alkaline nitrogen of 197.5 mg·kg^-1, and an organic matter content of 21.3 g·kg^-1. We used Zhongzao39 (ZZ39), an indica inbred rice cultivar, and Lingliangyou268 (LLY268), an indica hybrid rice. ZZ39, passed China’s validation in 2012, with an average of 112.2 days of full-life span, the production experiment in 2011 showed a yield of 7855 kg·hm^-2 (https://www.ricedata.cn/variety/varis/605891.htm). LLY268, the maternal cultivar is Xiangling628, the paternal cultivar is Hua268, passed China’s validation in 2008, with an average of 112.2 days of full-life span, the production experiment in 2007 showed a yield of 7720 kg·hm^-2 (https://www.ricedata.cn/variety/varis/605259.htm).

The experiment included two cultivars of ZZ39 and LLY268, together with two different nitrogen application procedures. The experimental treatments included the application of 150 kg·hm^-2 of nitrogen (N150), a control group with no nitrogen fertilizer (CK). Nitrogen was applied in three stages: basal fertilizer (two days before transplantation), tillering fertilizer (7 days after transplanting), and panicle fertilizer (during the fourth stage of panicle initiation), the nitrogen application rate followed a ratio of 6:3:1 for basal fertilizer, tillering fertilizer and panicle fertilizer. Nitrogen-based fertilizer, specifically urea, was used. All treatments were treated with 90 kg·hm^-2 of superphosphate (containing P₂O₅ 13.5%), which was applied as basal fertilizer at one time (two days before transplantation). A total of 180 kg·hm^-2 of potassium chloride, with a K₂O content of 52%, was applied, the ratio of basal fertilizer to panicle fertilizer was 1:1.

The plant was cultivated in plastic pots of 34 cm in diameter and 30 cm in height. Each pot contained 14.0 kg of paddy soil and was fertilized according to the recommended dosage for the potting region. Conversion according to Lu (Lu et al., 2016), nitrogen (N) was applied at 1.36 g/pot, superphosphate (P₂O₅) was applied at 0.83 g/pot, potassium chloride (K₂O) was applied at 1.63 g/pot. A total of 48 pots were set up for the entire experiment with 12 pots per treatment, i.e. 48 replications per treatment. The seeds were planted on March the 30^th, 2023, and moved to the pots on April the 29^th. Each pot contains 4 holes, with 3 seedlings placed in each hole. The rice plants were well-irrigated, weed controlled, pest controlled. These plants were used for the later experiments. These rice lines were headed on June the 19^th, and finally harvested on July the 17^th. The measurements were conducted on completely expanded mature leaves.

The labeling procedure follows Xiao (Xiao et al., 2019). Leaves were placed in a transparent leaf chamber measuring 50 cm×5 cm×5 cm ( Figure 1A ). The chamber was continuously supplied with ¹³CO₂ using a device specifically designed for continuous labeling ( Figures 1B, C ). The ¹³CO₂ was produced by mixing a solution of 0.2 mol·L^{-1 13}C-Na₂CO₃ (with a 99% abundance of ¹³C-Na₂CO₃) with 1 mol·L^-1 hydrochloric acid (HCl). The labeling process lasted for 3 hours, from 9:00 to 12:00 AM on a sunny day. Each treatment was labeled three times to ensure accuracy of the results. The labeling was conducted at the rice booting stage (May the12^th to May the 18^th, 2023), with each treatment being labeled in three separate repetitions. Three unlabeled rice pots were simultaneously used to evaluate the natural abundance of ¹³C in plants. The detailed information on the position of ¹³C labeled leaves and the time at which they were sampled for each treatment are shown in Table 1 .

The data were analyzed and assessed for statistical significance using Microsoft Excel 2021 and SPSS 26.0. GraphPad Prism 9 and Origin 2022 were utilized for creating plots.

At maturity stage, following the N150 treatment, the grain yields of ZZ39 and LLY268 increased by 90.42% and 72.61% respectively compared to CK. Additionally, the N150 treatment resulted in a considerable increase in the number of panicles for both ZZ39 and LLY268 cultivars. More precisely, the number of panicles for ZZ39 showed a significant increase of 58.80%, while for LLY268, the increase was 27.78%. In addition, the N150 treatment resulted in a 48.49% increase in the number of grains per panicle for LLY268, and a 16.4% increase for ZZ39. Under N150 treatment, the seed-setting rate of ZZ39 increased significantly compared to CK by 10.0%, but there was no significant difference for LLY268. Nevertheless, there was no substantial difference between the two treatments in the thousand-grain weight of the two cultivars ( Table 2 ).

At the stage of panicle initiation, we studied the number of tillers in ZZ39 and LLY268, and measured the plant height and dry mass of both cultivars at the maturity stage. The results indicate a significant increase in the number of tillers, plant height, and dry matter weight of rice in the N150 treatment ( Figure 2 ). Specifically, the number of tillers for ZZ39 and LLY268 increased by 26.09% and 53.13% ( Figure 2A ), while plant height increased by 13.89% and 10.67% ( Figure 2B ). Additionally, the dry matter weight increased by 74.71% and 71.62% in the N150 treatment respectively, compared to CK ( Figure 2C ).

During the tillering and panicle initiation stages, we measured leaf length, width, area ( Figure 3 ), and thickness ( Table 3 ). Both cultivars showed increased leaf length and leaf widths when exposed to the N150 treatment, as compared to CK. At the tillering stage, the leaf areas of the uppermost leaves for ZZ39 and LLY268 increased by 33.93% and 33.13% respectively, compared to CK. In addition, the total leaf areas of the top three leaves of ZZ39 and LLY268 increased by 13.97% and 34.53% respectively at the panicle initiation stage, compared to CK, under N150 treatment. Nevertheless, there was no notable difference in the thickness of the leaves between the two cultivars when subjected to N150 treatment, as compared to CK, during the tillering and panicle initiation stages ( Table 3 ).

The leaf SPAD values were measured during the tillering and panicle initiation stages ( Figures 4A, B ), whilst the net saturated photosynthetic rate was measured during the panicle initiation and grain filling stage ( Figures 4C, D ). Both cultivars showed an increased SPAD values when subjected to the N150 treatment, as compared to the CK. At the tillering stage, the SPAD value of the upmost leaf of ZZ39 and LLY268 increased by 6.53% and 6.69% under the N150 treatment, respectively, compared to CK ( Figure 4A ). In comparison to the CK ( Figure 4B ), the total SPAD values of the upper three leaves in the canopy of ZZ39 and LLY268 under N150 treatment increased by 21.43% and 10.15% respectively at panicle initiation stage. Moreover, the topmost leaf of ZZ39 showed an 16.78% increase in its saturation net photosynthetic rate, while LLY268 showed a 40.73% increase, under the N150 treatment during the panicle initiation stage. Furthermore, during the grain filling stage, only the highest leaf of ZZ39 exhibited a noteworthy increase of 27.78% in its saturated net photosynthetic rate following the N150 treatment, as compared to CK ( Figure 4D ).

An overall pattern of the increase and then decrease of nitrogen (N) concentration in the uppermost leaves(flag leaves)was observed in each treatment of LLY268 and the N150 treatment of ZZ39 from the booting stage to the grain filling stage and to the maturity stage ( Figure 5A ). In contrast, the basipetal 2^nd leaves and basipetal 3^rd leaves exhibited a consistent trend of decrease or remain constant and then decrease except the N150 treatment of ZZ39, as did the N concentration in the sheaths and internodes ( Figures 5B, C ). The nitrogen concentrations in the leaves, sheaths, and internodes of N150 treatment in the two cultivars did not show a significant difference compared to that of the CK after nitrogen application. There was no noticeable disparity in the concentration of nitrogen (N) between ZZ39 and LLY268 when subjected to the same treatment. Organs located in the top layers of the canopy showed increased nitrogen (N) concentration compared to organs in the lower layers of the canopy.

From the booting stage to grain filling stage, the nitrogen content in the upper three leaves of canopy in CK decreased faster than that of N150 treatment in both cultivars, while the N content in the leaves of N150 treatment drop faster than CK from grain filling stage to maturity stage in two cultivars ( Figure 6A ). The N content in upper three sheaths of CK in ZZ39 dropped faster that of N150 treatment in ZZ39 from booting stage to the grain filling stage and maturity stage. The N content of flag sheaths had similar decreasing trend for CK and N150 treatment in LLY268 ( Figure 6B ), and the upper three internode have higher N content in grain filling stage ( Figure 6C ).

In both treatments of two cultivars, we observed a consistent trend of the carbon concentration changes in leaves, i.e. an initially increase followed by a gradual decrease from the booting to the grain filling and maturity stages ( Figure 7A ). In contrast, the carbon concentration in sheaths exhibited little changes during growth ( Figure 7B ). Following the application of nitrogen, the levels of carbon in the leaves, sheaths, and internodes of the N150 treatment in ZZ39 were increased compared to CK ( Figures 7A–C ). The concentration of carbon (C) in the sheaths and internodes were marginally elevated in LLY268 compared to ZZ39( Figure 7A, C ), nevertheless, there was no notable difference in carbon (C) concentration between the two cultivars under N150 treatment. The carbon concentration in different tissues in the upper layers of a canopy continuously exceeded that in tissues at the lower layers of a canopy, respectively of their growth locations.

The carbon (C) content of the leaf and sheaths of both cultivars in each treatment consistently decreased from the booting stage, except to the flag leaves and the basipetal 2^nd leaves from the grain filling stage to maturity stage ( Figure 8A, B ). Nevertheless, the carbon content of the internodes had an initial increase from booting stage to grain filling stage, and followed by a subsequent decrease from grain filling stage to maturity stage ( Figure 8C ). As for the carbon content under two treatments, the carbon content in the leaves of LLY268 increased by 17.32% compared to CK at the booting stage. The carbon concentration in sheaths and internodes showed little changes between different treatments at the booting, grain filing and maturity stages. Under the CK treatment, the total carbon content in leaves and sheaths of ZZ39 and LLY268 were reduced by 19.98% and 12.93% respectively. By comparison, the N150 treatment resulted in a decrease of just 12.89% for ZZ39 and 5.96% for LLY268 ( Figure 8A, B ). As for the two rice cultivars, the total carbon contents of leaves, sheaths, and internodes in ZZ39 were higher than that of LLY268 in both experimental circumstances at the booting stage, showing an average increase of 34.35%.

During the booting stage, after the assimilation of ¹³CO₂ and its conversion into photosynthate by leaf photosynthesis, the assimilated carbon were mainly allocated to leaves for leaf growth and maintenance of leaf function ( Figures 9A, D ). Afterwards, photosynthate was distributed from leaves to sheaths and internodes, and finally to panicle and other organs ( Figures 9B, C, E, F ).

During the booting stage, almost 61.1% of the photosynthate in the flag leaves were used directly within the flag leaves under both treatments. The remaining photosynthate was allocated to the panicle through the basipetal first sheath and the basipetal 2^nd internode. The rate of photosynthate export from the flag leaves was higher under the CK treatment in comparison to the N150 treatment ( Figure 9A ). On average, 48.2% of the photosynthates from the two cultivars were transported out of the flag leaves under CK treatment, compared to just 29.6% after the N150 treatment at the booting stage ( Figure 9A ). Meanwhile, the photosynthate of basipetal 2^nd leaves of both cultivars were distributed in the basipetal 2^nd leaves by an average of 38.05% under both treatments ( Figure 9D ). The remaining photosynthate was allocated to the panicle through the basipetal 2^nd sheaths and basipetal 3^rd internodes. The rate of transfer of photosynthetic products out of the basipetal 2^nd leaves was higher under the CK treatment compared to the N150 treatment. In the CK treatment, approximately 48.2% of the photosynthetic products from both cultivars were transported out from the basipetal 2nd leaves. However, under the N150 treatment, only 29.6% of the photosynthetic products were transferred out ( Figure 9D ). During the grain filling stage, around 94.4% of the photosynthetic products from the flag leaves of both cultivars under both treatments were exported from the flag leaves. The majority of these photosynthate were allocated into the basipetal 2nd internode (24.56%), followed by the flag internode (10.2%), and the panicle (48.94%) ( Figure 9B ). In addition, approximately 93.18% of the photosynthate originating from the basipetal 2nd leaves were exported out. The distribution of these products was mostly concentrated in the basipetal 3^rd internode (18.6%), basipetal 2nd internode (6.3%), flag internode (20.7%), and panicle (44.2%) ( Figure 9E ). At the maturity stage, almost 81.71% of photosynthate in the flag leaves of both cultivars was transported to the panicle under both conditions. In the case of ZZ39, the average of CK and N150 treatments showed that 78.07% of the photosynthate was transported to the panicle. In contrast, the LLY268 treatments had a higher transfer rate of 85.35% ( Figure 9C ). In addition, about 83.37% of the photosynthate from the basipetal 2nd leaves in two variations were transported to the panicle. In the case of the ZZ39, the transfer rate was 90.63% on average for both CK and N150 treatments. For the LLY268, the transfer rate was 76.11% ( Figure 9F ).

We found that the N150 treatment resulted in an increase in the dry matter weight of panicles compared to the control (CK). Specifically, the biomass of single panicles under the N150 treatment increased by 17.88% and 35.37% in the maturity stage of ZZ39 and LLY268, respectively, compared to the control ( Figure 10A ). However, the rate of growth of the panicles of the two cultivars under CK treatment was higher compared to that of N150 from the booting stage to the grain filling stage ( Figure 10B ).

Optimizing N fertilizer application can increase nitrogen use efficiency through enhancing pre-anthesis non-structural carbohydrates accumulation and post-anthesis non-structural carbohydrates translocation (Pan et al., 2016). This non-structural carbohydrates reserved in the stems and sheaths can contribute up to 1/6–1/3 of the grain weight (Okamura et al., 2018) This pattern of increased allocation to the sheath and internode before booting and export of carbon and nitrogen from these tissues after booting were clearly shown in this study ( Figures 6 , 8 ). Remarkably, before the booting stage, rice accumulates nitrogen compounds absorbed from the soil in their vegetative organs. However, after the booting stage, nitrogen is mostly allocated to the grain (Ishikawa et al., 2023) with around 80% of nitrogen in the grains is from aging tissues (Ji et al., 2023). Nitrogen application, interestingly, had little effect on the carbon or nitrogen content of different tissues, but it did enhance the nitrogen content of leaves.

In this study, we found that the basipetal 3^rd internode had a greater ability for carbon and nitrogen accumulation and transport compared to the upper internodes, consistent with previous findings (Wakabayashi et al., 2022). Nonetheless, the accumulation of carbon and nitrogen in leaves and sheaths differed from that in the internode. The flag leaves and sheath exhibited greater nitrogen buildup than the lower organs ( Figure 6 ). However, the carbon accumulation demonstrated that the basipetal 2^nd leaves and sheaths displayed enhanced performance ( Figure 8 ). These results correspond with the findings of Liang’s study (Liang et al., 2014).

The efficiency of photosynthesis and rice yield are determined by functional leaves, in particular the flag leaves and basipetal 2^nd leaves. Previous studies show that significant differences in leaf morphology and photosynthetic properties between the flag leaf and the basipetal 2^nd leaf; there is also difference in these properties caused by different leaf positions (Gao et al., 2021; Ding et al., 2022). At the stage of panicle initiation, the leaf area and SPAD of leaves at different leaf positions varied between two cultivars, specifically, the basipetal 2^nd leaves had larger values compared to the flag leaves ( Figures 3 , 4 ).

There are also significant difference in the ability of leaves at different stem positions to gain formation. At the booting stage, after labeling the leaves at different stem positions with ¹³CO₂, we found a significant proportion (ranging from 73.92% to 91.43%) of the photosynthate from flag leaves and basipetal 2^nd leaves of different cultivars were allocated to the panicles. Leaves differentially contribute to grain formation in different cultivars. For example, a greater proportion of photosynthate was allocated to grain at maturity from the basipetal 2^nd leaf compared to the flag leaf in ZZ39. In contrast, the difference in this proportion between the flag leaf and basipetal 2^nd leaf was relatively minor in LLY268 ( Figures 9 , 11 ). During the process of photosynthate transferring, photosynthate from the basipetal 2^nd leaf was transferred out of the leaf earlier compare to that of the flag leaf ( Figures 9A, D ), as a result of the difference in the maturity of the flag leaf and the basipetal 2^nd leaf.

The observed allocation pattern of photosynthate from this study on early-season rice differs from those on single-season rice (Mohapatra et al., 2004), where the majority of the photosynthate from the upper three leaves during the booting stage were stored in the stems until maturity. This maybe is due to the fact that in single-season rice, which usually has a longer period of growth and a higher biomass, the photosynthate continues to be supplied to the vegetative tissue during the booting period. On the other hand, in early-season rice, which has a shorter growth period and a smaller biomass, the vegetative tissues are mostly developed before the booting stage due to the relatively short growth cycle (Li et al., 2023). As a result, the formed photosynthate can be allocated directly to the panicle, instead of first storing them in the sheath and stem. Interestingly, in low-yielding traditional rice, the flag leave is also primarily responsible for providing carbon assimilates to the grain while in the super high-yielding rice cultivar, e.g. Takanari, the contribution of the basipetal 2^nd leaf to grain formation is comparable to that of flag leaves (Mohapatra et al., 2004). These increased number of leaves contributing to grain formation might reflect the enhanced source capacities of these high-yielding cultivars. In the present experiment, the contributions of the flag leaf and the basipetal 2^nd leaf to the grain of both cultivars were similar, which is similar to the pattern of photosynthate allocation in high-yielding rice. There were deficiencies in this experiment in that the ¹³CO₂ labeling experiment was conducted for only one year, and because of the difficulty of labeling in the field, a field trial was not conducted. However, field experiments were carried out previously (Wang et al., 2021), and the data on yield composition and other data from this pot experiment are reliable.

After nitrogen application, the yield of ZZ39 increased by 90.42% and the yield of LLY268 increased by 72.61% ( Table 2 ). Our finding is consistent with the earlier finding that the response to nitrogen application is more pronounced in cultivars with larger panicles (Liao et al., 2023). Remarkably, though nitrogen application increased rice yield dramatically in both cultivars; however, the increases in yields were achieved through complexly different mechanisms. Specifically, following the application of nitrogen fertilizer, ZZ39 showed increased number of effective panicles, whereas LLY268 showed increased number of grains per panicle ( Table 2 ). These responses reflect, in some sense, an optimal response of rice to nitrogen application. Prior field experiments have revealed variations in the yield components’ response to nitrogen between the two cultivars. Notably, at a nitrogen application of 150 kg·hm^-2, the yield increases for both cultivars were similar, averaging 27.51%. However, the increase in the number of effective panicles was greater for ZZ39 compared to LLY268 regarding yield components (Wang et al., 2021). These differential responses in yield components in these two rice cultivars are underlined by their different patterns of carbon and nitrogen under nitrogen treatment ( Figures 6 , 8 , 9 ). Such near-optimal responses in these allocation patterns might reflect the artificial selection by breeders for cultivars which can adjust their growth and development patterns to gain superior yield in a relatively short growth cycle. For instance, LLY268 exhibits a greater number of effective panicles, necessitating increased transport of photosynthates from individual stems to other tillers. This is evidenced by the observation that photosynthates from the basipetal 2^nd leaf were more frequently transported to the lower internode compared to ZZ39 during the filling stage ( Figure 9E ). Concurrently, the C content indicated that the lower internode of LLY268 had a more pronounced rise during the filling stage ( Figure 8C ).Therefore, early-season rice might be used as a model to study how plants gain flexible and tailored responses to nitrogen treatment to gain higher yields.

Besides these tailored and near-optimal responses of photosynthate allocation under nitrogen treatment, we also found that the speed of photosynthate translocation was slower ( Figures 8 , 9 ), and the panicle growth rates were slower at the beginning of grain filling as well ( Figure 10 ) under nitrogen treatment. This finding is consistent with earlier findings that α-amylase and β-amylase were more efficient in the rice stalk under low nitrogen conditions (Li et al., 2018a, 2022), the higher rate of NSC translocation in sheaths under low nitrogen (Zakari et al., 2020) and constant ¹³C distribution ratio in rice under nitrogen application (Xiao et al., 2019).