Grain-Filling Characteristics in Extra-Large Panicle Type of Early-Maturing japonica / indica Hybrids

: Early-maturing japonica / indica hybrids (EJIH) have recently been released, performing a yield potential of 13.5 t ha − 1 and greater yield increase over conventional japonica rice (CJ) and hybrid indica rice (HI) in production. More spikelets per panicle and improved grain-ﬁlling efﬁciency underlined the basis for the superior yield performance of EJIH. However, few studies are available on the panicle traits and grain-ﬁlling characteristics of EJIH, as well as their differences to CJ and HI. In our study, two EJIH, two CJ, and two HI cultivars with similar growth patterns were grown in the same ﬁelds. EJIH had a 12.2–18.8% increased ( p < 0.05) grain yield relative to CJ and HI, mainly attributed to their higher daily grain yield. Although it had a lower panicle per m 2 , EJIH exhibited 28.0–38.3% more ( p < 0.05) spikelets per m 2 from an increase of 58.0–87.8% ( p < 0.05) in spikelets per panicle than CJ and HI. Compared with CJ and HI, EJIH had a higher single panicle weight and more grains in the six parts of the panicle, especially in the upper secondary branches (US) and middle secondary branches (MS). EJIH exhibited a higher leaf area index (LAI), leaf area duration (LAD), leaf photosynthetic rate, and SPAD values after heading, which helped increase shoot biomass weight at heading and maturity and post-heading biomass accumulation. For CJ and HI, the grain-ﬁlling dynamics of grains in the six parts were all well simulated by the Richards equation. For EJIH, the grain-ﬁlling dynamics of grains in the lower secondary branches (LS) were well ﬁtted by the logistics equation, with the Richards equation simulating grain positioning on the other ﬁve parts. EJIH had a lower mean grain-ﬁlling rate (GR mean ) and longer days and grain ﬁlling amounts (GFA) during early, middle, and late stages than CJ and HI. Our results suggest EJIH gave a yield advantage over CJ and HI through a higher daily grain yield. The panicle traits and grain-ﬁlling characteristics differed greatly among the three cultivar types. Compared with CJ and HI, EJIH had lower GR mean and higher days and more grains in the panicle during early, middle, and late stages, which contributed to an increased GFA after heading, improved ﬁlled-grain efﬁciency, and higher grain yield.


Introduction
The dominated cropping system in the lower reaches of the Yangtze River, China, is a rotation of rice and wheat [1]. In this system, rice is sown after harvesting wheat, and wheat is sown after harvesting rice. Proper sowing date is a critical basis for achieving a high yield of rice and wheat [2][3][4]. In recent years, late-maturing rice cultivars with high yield performance have been widely planted by farmers. However, these late-maturing rice cultivars are always harvested late, which delays the sowing date of wheat and easily cause a yield penalty [5,6]. Hence, rice cultivars with an early maturity and high yield in 2018. Generally, the rice experienced similar mean temperature, sunshine hours, and rainfall during the rice-growing period from May to October at two years ( Figure 2).  Two EJIH, two CJ, and two HI cultivars were field grown in this study. The two EJIH cultivars were Yongyou 2640 and Yongyou 1640; the two CJ cultivars were Yangjing 4227 and Zhendao 14; and the two HI cultivars were Fengyou 293 and Zhongnanyou 1. The selection of these rice cultivars was based on their popularity and large planting area in production. Furthermore, these rice cultivars share a similar growth pattern, and the total growth period ranged from 149 d to 153 d in this study. Detailed information on the year of release, cross information, and growth period of rice cultivars are listed in Table 1.
The experiment design was a randomized block, and each experimental plot covered 35 m 2 (7 m × 5 m) with three replicates for two years. Rice seeds were sown in a seedling nursery on 23 May and transplanted with three seedlings per hill on 12 June. The hill spacing was 30 cm row spacing and 13 cm plant spacing. The management practices were the same in all experimental plots. N was applied at 270 kg urea-N ha −1 in total at a ratio of 3:3:2:2 1 d before transplanting, 1 week after transplanting, at panicle initiation, and during the penultimate-leaf appearance. The P and K fertilizers were both applied once as base dressing, at an application rate of 180 kg P 2 O 5 ha −1 and 150 kg K 2 O ha −1 , respectively. Irrigation regimes, pest, disease, and weed management followed local agricultural recommendations.

Measurements
At heading and maturity, five hills of plants were sampled to determine leaf area index (LAI) and shoot biomass. LAI was determined through a leaf area meter (LI-3100C, Lincoln, NE, USA). Sampled plants were subdivided into panicles, leaves, and stems, which were then placed well in Kraft paper bags, and shoot biomass was recorded after 80 h of oven-drying at 75 • C.
Plants heading and flowering at the same date were labeled as single stems with similar growing conditions, and 500 single stems were selected in each plot. The single plants were collected from heading to maturity, and 15 samples were collected at 4-d intervals to determine the grain-filling dynamics of rice cultivars. Total sampling times of EJIH, CJ, and HI were 17, 14, and 13, respectively, based on their grain-filling period (Table 1). At each sampling, rice panicles were divided into six parts, grains in the UP, US, MP, MS, LP, and LS. The specific classification was as follows, first, rice panicles were separated equally into upper, middle, and lower parts, based on the number of the primary branches; then the three parts were divided into primary and secondary branches according to the grain position.
For each rice cultivar, SPAD value was determined at 12, 24, and 36 days after heading (DAH), leaf photosynthetic rate was determined at 15, 30, and 45 DAH. SPAD values of flag leaf were determined by soil-plant analysis development meter (SPAD-502 plus). The flag leaf photosynthetic rate was performed through two photosynthetic instruments (LICOR-6400, Lincoln, NE, USA), which was conducted around 9:30 h to 11:00 h under sunny conditions. At maturity, two hundred representative hills (7.8 m 2 ) of plants in each plot excluding border plants were harvested for determining grain yield at 14% moisture. In addition, one hundred representative hills of plants in each plot were collected for measuring panicle traits and grain yield components.

Model Analysis of Grain Filling
In this study, the Richards equation and logistic equation were adopted to simulate rice grain-filling dynamics. The equation (Richards or logistic) with a better fit coefficient was selected to simulate the rice grain-filling process.
The Richards equation was expressed as W = A 1 + Be −kt − 1 N , where W represents the grain weight, A represents the final grain weight, and t represents the days after heading. The parameters B, K, and N were computed by the regression equation.
After parameters of the Richards equation were estimated, the maximum grain-filling rate (GR max ), mean grain-filling rate (GR mean ), days achieving the maximum grain-filling rate (D max ), and effective grain-filling period (EP) were calculated by follows [24]: The rice grain-filling process could be divided into early (0-t 1 ), middle (t 1 -t 2 ), and late stages (t 2 -t 3 ).
The logistic equation was expressed as W = A 1 + Be −kt −1 , where W represents the grain weight, A represents the final grain weight, and t represents the days after heading. B and K are parameters computed by the regression equation.
After parameters of the logistic equation were estimated, the GR max , GR mean , D max , and EP were calculated by follows [24]: The rice grain-filling process could be divided into early (0-t 1 ), middle (t 1 -t 2 ), and late stages (t 2 -t 3 ).

Data Analysis
The data was processed using analysis of variance (ANOVA) with the least significant difference through SPSS 17.0 software (SPSS Inc., Chicago, IL, USA). ANOVA showed that there were no significant (p ≥ 0.05) differences in grain yield and the related agronomic and physiological traits between the two study years (Table 2), and data averaged across two years were used for the following analysis. ns, non-significance, *, and **, significant at 0.05 and 0.01 probability levels, respectively.

Grain Yield and Panicle Traits
EJIH showed a 12.2% increase (p < 0.05) in grain yield over CJ, and an 18.8 increase (p < 0.05) over HI. Similarly, EJIH had a higher daily grain yield than CJ and HI (Table 3). EJIH showed 58.0-87.8% more (p < 0.05) spikelets per panicle and 28.0-38.3% more (p < 0.05) Agriculture 2021, 11, 1165 7 of 17 spikelets per m 2 , although with less panicles per m 2 than CJ and HI. The filled-grain percentage of EJIH (nearly 88%) was similar to that of CJ and HI. The grain weight of CJ was close to HI, and higher (p < 0.05) than that of EJIH (Table 4).
HI showed the highest panicle length, followed by EJIH and CJ (p < 0.05), respectively. EJIH had 69.4% higher (p < 0.05) single panicle weight than CJ, and 36.9% higher (p < 0.05) than HI. Compared with CJ and HI, EJIH showed an increased number of grains on six parts of the panicle, especially in the US and MS position (Table 5).

Shoot Biomass, LAI, LAD, Leaf Photosynthetic Rate, and SPAD Values
Compared with CJ and HI, EJIH exhibited higher (p < 0.05) shoot biomass weight at heading, as well as maturity. For example, shoot biomass at heading of EJIH was 13.8% and 29.1% higher than that of CJ and HI, respectively. EJIH also showed 15.3% and 25.8% more (p < 0.05) shoot biomass accumulation from heading to maturity than CJ and HI, respectively. The harvest index of CJ and HI was around 0.49, and both were higher (p < 0.05) than that of EJIH (Table 6). Generally, EJIH showed 9.5% and 12.5% higher (p < 0.05) LAI at heading than CJ and HI, respectively, with similar results observed at maturity. EJIH had 35.4% and 55.6% higher LAD from heading to maturity, compared with CJ and HI, respectively (Table 7). . Means followed by different letters within a column are significantly different at the 5% probability level.
EJIH exhibited a higher (p < 0.05) flag leaf photosynthetic rate at both 30 and 45 DAH, relative to CJ and HI. Similarly, EJIH had higher SPAD values of flag leaf at 12 DAH than CJ and HI (p < 0.05). SPAD values for flag leaf at 24 and 36 DAH of EJIH were higher (p < 0.05) than that of CJ and HI (Figure 3).
EJIH exhibited a higher (p < 0.05) flag leaf photosynthetic rate at both 30 and 45 DAH, relative to CJ and HI. Similarly, EJIH had higher SPAD values of flag leaf at 12 DAH than CJ and HI (p < 0.05). SPAD values for flag leaf at 24 and 36 DAH of EJIH were higher (p < 0.05) than that of CJ and HI (Figure 3).

Grain-Filling Dynamics and Simulated Equations
For EJIH, grains in the UP and US positions exhibited similar dynamics in the grainfilling process, which were different from grains in the other position. For CJ, grains in the UP, US, MP, and MS showed similar grain-filling dynamics and were different from grains in the LP and LS. For HI, grains in the six parts all showed similar grain-filling dynamics ( Figure 4).

Grain-Filling Dynamics and Simulated Equations
For EJIH, grains in the UP and US positions exhibited similar dynamics in the grainfilling process, which were different from grains in the other position. For CJ, grains in the UP, US, MP, and MS showed similar grain-filling dynamics and were different from grains in the LP and LS. For HI, grains in the six parts all showed similar grain-filling dynamics (Figure 4).
For EJIH, the grain-filling dynamics of grains in the LS position were well fitted by the logistic equation, while the Richards equation was used for grains in the other positions. For CJ and HI, the grain-filling processes of six parts in the panicle were all wel simulated by the Richards equation (Table 8).   For EJIH, the grain-filling dynamics of grains in the LS position were well fitted by the logistic equation, while the Richards equation was used for grains in the other positions. For CJ and HI, the grain-filling processes of six parts in the panicle were all well simulated by the Richards equation (Table 8).

Grain-Filling Characteristics
For EJIH and CJ, GR max and GR mean of grains in the UP were the highest, followed by MP, MS, US, LP, and LS, respectively. For HI, GR max and GR mean of grains in the UP were the highest, followed by US, MP, MS, LP, and finally LS. For the three cultivar types, GR max and GR mean of grains varied in the six parts, with a smaller difference in CJ and greater difference in HI. D max and EP also varied across grains in the six parts of each cultivar type, and the difference was smaller in CJ and greater in EJIH. Generally, EJIH exhibited lower (p < 0.05) GR max and GR mean across grains in the six parts, while higher (p < 0.05) D max and EP, compared with CJ and HI (Table 9). For the three cultivar types, early stage was the highest for the grain-filling characteristic 'days', followed by middle and then late stages; while GR mean and GFA of middle stage were the highest, followed by early and then late stages. Compared with CJ and HI, EJIH showed consistently higher (p < 0.05) days and GFA, but a lower (p < 0.05) GR mean during early, middle, and late stages. For example, GFA during early, middle, and late stages of EJIH was 21.4%, 39.6%, and 72.5% higher than that of HI, respectively (Table 10).

Discussion
It is acknowledged that japonica/indica hybrids display strong heterosis in terms of yield production [25]. However, the application of japonica/indica hybrids was extensively restricted by some constraints, such as long growth periods and grain-filling obstacles [26,27]. Recently, considerable strides have been achieved in solving such constraints and japonica/indica hybrids with an early maturity and high yield potential have become available in production, such as Yongyou 2640 in our study. Herein, EJIH exhibited a 12.2-18.8% higher (p < 0.05) grain yield than CJ and HI (Table 3), similar to prior results [12,13,15]. Considering the similar growth period among the three cultivar types, the yield advantage of EJIH over CJ and HI was mainly driven by their higher daily grain yield (Table 3). A similar growth pattern among the three cultivar types indicated a comparable amount of intercepted solar radiation during the growth period; nevertheless, EJIH had a better solar radiation use efficiency benefitting from the improved plant posture and capability of capturing light resources [28]. For rice, better solar radiation use efficiency is typically associated with greater biomass production and daily grain yield [29,30].
Analysis of yield components suggests that more spikelets per panicle contributed to the superior yield performance of EJIH over CJ and HI (Table 4). Although there is no specific definition of "panicle type", rice cultivars with more than 200 spikelets per panicle are always considered as "large panicle type" [31,32]. In the present study, EJIH exhibited spikelets per panicle of more than 250, 58.0-87.8% more than CJ and HI; hence, EJIH could be regarded as an "extra-large panicle type". Furthermore, compared with CJ and HI, the advantage that EJIH has in spikelets per panicle was detected in all six positions of the panicle, especially for US and MS (Table 5), implying that it was an effective way to promote panicle size from increasing grains positioned in the US and MS. In this study, EJIH had a similar filled-grain percentage relative to CJ and HI (Table 4) and did not exhibit the grain-filling barrier that has previously been found in japonica/indica hybrids [33,34]. From this, it can be deduced that progress has been made in developing japonica/indica hybrids with better grain-filling efficiency in China [8,25].
Compared with CJ and HI, EJIH had a higher shoot biomass weight at heading and maturity, and increased biomass accumulation after heading (Table 6). For EJIH, such an increased pre-heading biomass production benefitted larger sink size, while postheading biomass production contributed to improved grain-filling efficiency [15,27]. This result was consistent with previous studies that report how high-yielding rice typically exhibits superior capacity of biomass production not only before heading, but also after heading [35,36]. In addition, LAI at heading and maturity, and leaf photosynthetic rate, SPAD values, and LAD during the ripening period were higher (p < 0.05) in EJIH relative to CJ and HI (Table 1, Figure 3). Our results suggest that the improved leaf photosynthetic capacity and canopy structure of EJIH promoted assimilates accumulation after heading, increased efficiency of grain filling, and, finally, a superior yield performance.
Generally, the Richards equation has been adopted to simulate grain-filling of rice, and the logistic equation for maize and wheat [18][19][20]37,38]. However, few studies are available on the grain-filling dynamics of cultivars with more than 250 spikelets per panicle. For CJ and HI, the Richards equation has been considered the best-fit for modeling the grain-filling dynamics of superior and inferior spikelets [18,19]; however, superior and inferior spikelets only occupy a small proportion of the panicle and do not comprehensively reflect the grain-filling characteristics of the whole panicle, particularly for large-panicle rice cultivars. In this study, the grains in the panicle were divided into six parts, based on their positions. For CJ and HI, the grain-filling dynamics of grains in the six parts were all well simulated by the Richards equation. For EJIH, the grain-filling dynamics of grains in the LS were well simulated by the logistics equation, and also the Richards equation for grains positioning on the other five parts (Table 8). Noteworthy, grains positioning on the LS were always selected as inferior spikelets in previous studies [18][19][20]39]. These results suggest that the grain-filling dynamics of grains located in the LS (e.g., inferior spikelets) of rice cultivars with the extra-large panicle type might be better fit by the logistic equation, rather than by the Richards equation.
Prior studies have reported that the grain-filling pattern of rice could be determined by the time difference of D max between superior and inferior spikelets [20,40]. Following such a classical classification method [40], CJ was sorted into synchronous grain-filling type, HI into asynchronous type, and EJIH into a more asynchronous type in our study (Table 9). Compared with CJ and HI, this more asynchronous grain-filling process was associated more with grains located in the six parts of EJIH with extra-large panicle type. Such a grain-filling pattern of EJIH could make for a more efficient and full utilization of the assimilates produced for filling grains, contributing to the improved filled-grain percentage of EJIH. The filling efficiency of grains positioned in the lower parts often resulted in a poor filled-grain percentage of large-panicle rice cultivars [17,41]. For EJIH, grains in the lower parts (LP and LS) had longer D max and EP, compared with grains in the upper and middle parts (Table 9). This result indicates that proper management practices should be adopted especially during the later grain-filling stage, a critical period for filling grains in the lower parts.
The differences in the grain-filling characteristics between CJ and HI have been studied and reported [21][22][23]. For example, Gong et al. [22] concluded that CJ had longer days but lower GR mean during the early, middle, and late stages relative to HI. To date, grain-filling characteristics of EJIH have received less attention, as have their differences to CJ and HI. In this study, EJIH exhibited consistently lower (p < 0.05) GR max and GR mean , but higher (p < 0.05) D max and EP, compared with CJ and HI (Table 9). Furthermore, EJIH had more grains in the six parts and longer (p < 0.05) days, which helped increase GFA during early, middle, and late stages, though it lowered GR mean during these three stages (Table 10).

Conclusions
EJIH gained yield superiority over CJ and HI, mainly driven by higher daily grain yield. Great differences exist in panicle traits and grain-filling characteristics among EJIH, CJ, and HI, the three main cultivar types in rice production. EJIH exhibited more spikelets per panicle, increased single panicle weight, and increased number of grains in each of the six parts, with pronounced increases in the US and MS, compared with CJ and HI. CJ was classified as synchronous grain-filling type, HI as asynchronous type, and EJIH as more asynchronous type. Although lower GR mean during early, middle, and late stages, the days and number of grains positioned in the panicle of EJIH were higher, which together facilitated more GFA after heading, better grain-filling efficiency, and higher grain yield.