A total of 4449 rice accessions (3133 indica accessions and 1316 japonica accessions) were tested in present experiment from 2004 to 2018, and 162 test plots were distributed from 20°N to 46°N, 80°E to 130°E (Table 1 and Supplemental Table 1 and 2). As the growth period shortened with increasing latitude, the indica accessions were subsequently divided into four groups (early indica group, middle indica in the upstream region of the Yangtze River group, middle indica in the downstream region of the Yangtze River group, and the late indica group), whereas the japonica accessions were divided into three groups (northern japonica group, middle japonica group, and southern japonica group) based on the location of the test plots and the growth period of the accessions (Table. 1). The test lines were evaluated in a randomized complete block design with at least three replications, and each block was approximately 13-14 m². Seeds were sown in a seedling nursery and then transplanted into the block. The planting densities and fertilizer protocols were based on local conventional methods.

Table 1

Table 1

Table 1 Correlation between yield and other traits. Group PN GNP SSR TGW BR HR CR CD AC PH BL DTM Early indica 0.016 0.183* 0.514* 0.114* -0.058 0.105* -0.086 -0.100* 0.119* 0.006 -0.152* -0.068 Middle indica (upstream) -0.097 0.421* 0.158* 0.103* -0.209* -0.381* -0.037 -0.074 -0.098 0.217* -0.202* -0.227* Middle indica (downstream) -0.304* 0.645* 0.338* 0.103* 0.051 -0.020 -0.533* -0.500* -0.358* 0.298* -0.220* 0.539* Late indica 0.460* 0.219* 0.288* 0.038 0.044 -0.158* -0.034 -0.072 -0.127* -0.151* -0.171* 0.292* Middle japonica 0.638* 0.093 0.493* 0.300* 0.659* 0.408* 0.001 0.000 -0.399* -0.333* -0.222* 0.859* Northern japonica 0.129* 0.136* 0.046 0.053 0.060 -0.008 0.036 0.107* -0.059 0.005 -0.109* -0.014 Southern japonica -0.415* 0.622* 0.048 0.026 -0.300* -0.347* 0.029 -0.023 -0.108* 0.415* -0.075 0.489* indica -0.509* 0.724* 0.160* 0.342* -0.278* 0.110* -0.150* -0.238* -0.109* 0.552* -0.023 0.701* japonica 0.151* 0.148* 0.142* 0.001 -0.020 -0.115* -0.014 0.048 -0.035 0.027 -0.139* 0.077 Average 0.112* 0.347* 0.317* 0.096 0.193* 0.311* -0.176* -0.268* -0.251* 0.276* -0.166* 0.716* PN, Panicle number per plant; GNP, Grain number per panicle; SSR, Seed-setting rate; TGW, 1000-grain weight; BR, Brown rice rate; HR, Head rice rate; CR, Chalkiness rice rate; CD, Chalkiness degree; AC, Amylose content; PH, Plant height; BL, Blast level; DTM, Days to maturity. * , Significant at 5% level.

Table 1 Correlation between yield and other traits.

Supplemental Table 1

Supplemental Table 1

Supplemental Table 1 Specific location of the test plots in this study

Supplemental Table 1 Specific location of the test plots in this study

Supplemental Table 2

Supplemental Table 2

Supplemental Table 2 Geographic coordinate ranges and number of plots of test groups per location and rice subspecies. Group No. of plots No. of accessions Longitude range Latitude range Altitude range Early indica 37 610 108°03′ -120°40′ 20°01′ -30°57′ 4.0-600.0 Middle indica (upstream) 23 815 98°36′ -117°55′ 25°6′ -33°07′ 180.0-1284.0 Middle indica (downstream) 23 937 111°05′ -120.9° 27°03′ - 33°23′ 2.7-231.0 Late indica 30 771 109°22′ -120°40′ 24°48′ -31°01′ 6.0-252.0 Southern japonica 11 681 112°02′ -120°46′ 29°42′ -31°52′ 3.0-67.0 Middle japonica 11 520 113°41′ -119°16′ 32°07′ -35°27′ 4.5-96.4 Northern japonica 27 115 80.5°-128°42′ 37°1′ -45°51′ 2.0-1118.5

Supplemental Table 2 Geographic coordinate ranges and number of plots of test groups per location and rice subspecies.

Rough rice samples for quality evaluation were obtained from the center of each block at the maturation stage. After the number of panicles were counted, they were hand threshed and placed into water. The filled grains that sank were separated from the floating unfilled grains. The number of grains per panicle, 1000-grain weight, and seed setting rate were calculated. After drying, the rough rice from each block was dehulled, and the percent hulled was determined. Brown rice samples from each block were milled, and the yield of head rice was calculated. The grains that displayed chalkiness were counted, and the percentage of chalky grains was calculated as the chalkiness rate. For the chalkiness area level, 30 grains with chalkiness were randomly selected, and the ratio of the chalkiness area level to the whole kernel square was evaluated grain scanner JMWT 12 (Satake, Tokyo, Japan) following the protocol. The spikelet density was calculated as the average number of seeds per centimeter on a panicle. The amylose content was investigated using the method described in our previous study (Li et al., 2018). The blast level was evaluated based on the Chinese national rice test standard.

The data of yield potential and actual yield was obtained from Deng et al. 2019 (Deng et al., 2019). Correlation analysis and Student’ s t tests were performed using Excel 2013 (Microsoft, Redmond, WA, USA) and R studio software (Boston, MA, USA).

Rice, Oryza sativa, is a major global food staple, 90% of which is cultivated in Asian countries. Its two major subspecies are japonica and indica, differing in geographical distribution and morphological characteristics (Wang W S et al, 2018). The indica subspecies occupies 90% of the rice cultivation area, as opposed to only 10% occupied by the japonica subspecies, which is mainly distributed in northern China, Japan and Korea (Xu and Chen, 2016). China is the only country in which japonica and indica are of equal importance. Two-thirds of the rice grown in China is indica, which is cultivated mostly in low-altitude and low-latitude regions, and one-third is japonica, farmed prevalently in high-altitude and high-latitude regions (Wang Y H et al, 2018). After the introduction of semi-dwarf gene sd1, the ideal plant architecture breeding and the application of the F₁ hybrid, accompanied by the improvement of cultivation methods and field management, the rice yield of China has increased from 1.9 t/hm² in 1949 to 7.0 t/hm² in 2018. This corresponds to an increase of 268%, which is 50% higher than the average yield increase worldwide (Qian et al, 2016). These achievements have placed China in the position of a leader among the major rice-producing countries. In recent years, along with the increase in the living standard and the development of rice breeding, the objectives in rice production have changed from merely increasing the yield to improving the quality without sacrificing the yield. Rice science research and the history of rice production indicate that it is increasingly difficult to improve yield further, and even more difficult to improve quality while maintaining the yield or increase the yield while maintaining the quality. Thus, analyzing the trends and the relationship between yield and quality of indica and japonica rice in different areas can provide valuable information necessary to attain high-yielding and good-quality rice breeding in China as well as the rest of the world.

A total of 4 449 rice accessions (3 133 indica accessions and 1 316 japonica accessions) were tested from 2004 to 2018 in 162 test plots distributed from 20°N to 46°N and from 80°E to 130°E (Supplemental Table 1 and 2). Despite some fluctuations over the period of 15 years, the general trend was that japonica rice, especially northern and middle japonica, generated an 11.82% higher yield than indica rice. The middle indica, in both upstream and downstream of the Yangtze River China, had a higher yield than early or late indica. The 15-year survey documented that the yield stably increased by 14.27%, corresponding to an annual increase of 0.95%. However, a significant variation in the rate of increase was present among groups, ranking as: southern japonica> early indica> middle indica (downstream) > middle japonica> middle indica (upstream) > northern japonica > late indica. During the analyzed period, the yield of southern japonica increased by 28.20% (1.88% per year), while the yield of early indica increased only by 3.86% (0.26% per year) (Fig. 1-A).

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. Trends of yield components and quality traits of tested rice subspecies accessions in China in 2004-2018. A, Yield. B, Grain number per panicle. C, Seed-setting rate. D, Panicle number per plant. E, 1000-grain weight. F, Brown rice rate. G, Head rice rate. H, Grain length to width ratio. I, Chalkiness rice rate. J, Chalkiness degree. K, Amylose content.

Next, we examined the changes in the characteristics of rice determining its yield among the seven groups over the 15 years. The northern japonica group showed the largest fluctuation in panicle number and had the highest number of panicles per plant among the seven groups. The middle japonica, early indica and late indica groups had comparable panicle numbers per plant, which were slightly higher than the country-wide average, while the panicle number per plant in the middle indica (upstream) was lower. The number of panicles per plant slightly decreased during the 15-year survey (Fig. 1-D). The southern japonica, middle indica (upstream) and middle indica (downstream) groups showed a higher grain number per panicle, and large fluctuation in this parameter was identified in southern japonica. The northern japonica and early indica groups had the lowest grain number per panicle. With the exception of middle japonica, the grain number per panicle increased in all groups over the 15 years (Fig. 1-B). The seed-setting rate in the seven groups exhibited large fluctuation. The seed-setting rates were higher than the average in all the three japonica groups, and lower in the indica groups, with the only exception of early indica. In the entire country, the seed-setting rate had an increasing trend over the 15 years (Fig. 1-C). The value of the 1000-grain weight was consistently the highest in the middle indica (upstream) group, while the middle indica (downstream) group had a value higher than the average during the first seven years of the survey, but later decreased to the average level. The early indica, late indica and southern japonica groups all had similar 1000-grain weight, and northern japonica had the lowest value among the seven groups. Except for the middle indica (downstream), 1000-grain weight among the seven groups was stable or slightly declined over the 15-year period (Fig. 1-E).

Although a significant positive correlation was observed between yield and panicle number per plant in average of all groups, the panicle number per plant had a significant positive correlation with yield in the japonica group and a significant negative correlation for the indica group (Table 1). There were significant positive correlations between the yield and panicle number per plant in the northern japonica, middle japonica and late indica groups, and significant negative correlations in the southern japonica and middle indica (downstream) groups. The number of grains per panicle showed a significant positive correlation to the yield in all the seven groups, except for middle japonica. Similarly, the seed-setting rate was positively correlated with the yield in all the groups, except for the northern japonica and southern japonica groups. Finally, a positive correlation between the yield and 1000-grain weight was present in all the seven groups. Taken together, whether indica or japonica, or northern or southern region are considered, the general strategy of yield improvement is to increase the grain number per panicle and seed-setting rate.

Overall, little change over the 15-year period was observed in the brown rice rate and head rice rate. These parameters remained high in the three japonica groups and were lower in the four indica groups (Fig. 1-F and -G). Among the seven groups, the middle indica (upstream) group had the lowest brown rice rate (Fig. 1-F), and the early indica group had the lowest head rice rate (Fig. 1-G). The grain length to width ratio of japonica was stable with a value of less than 2.0, whereas the grain length to width ratio of indica had a value of more than 2.5. The value of grain length to width ratio tended to increase in the later years of the survey (Fig. 1-H). The chalkiness traits showed a complex change among the seven groups (Fig. 1-I to -K). Specifically, the values of the chalkiness rate (Fig. 1-I) and chalkiness degree (Fig. 1-J) remained low in the three japonica groups, a large fluctuation of values was noted in the middle indica (upstream) group, and a declining trend was observed in the late indica, early indica and middle indica (downstream) groups. The three japonica groups had stable amylose content, which was lower than in the indicagroups (Fig. 1-K). A remarkable increase in amylose content took place in the indica groups, particularly in the middle indica (downstream) and late indica, which had already exhibited a similar level of amylose content as the japonica groups. The gap between the amylose content in the indica and japonica groups gradually shrunk over time due to improved methods of rice breeding.

There was a country-wide significant positive correlation between the yield and the brown rice rate and head rice rate. However, a large variability was observed in each group (Table 1). The brown rice rate of middle indica (upstream), and southern japonica had a significant negative correlation to yield, whereas that of the middle japonica showed a significant positive correlation with the yield. The head rice rates of early indica and middle japonica were positively correlated with the yield, and those of the middle indica (upstream), late indica and southern japonica were negatively correlated with the yield. The comprehensive analysis demonstrated that a fine-tuned balance between yield and milling quality was observed in the middle indica (downstream) and middle japonica groups, while an inverse relationship between yield and milling quality was present in the middle indica (upstream) and southern japonica groups. The amylose content and chalkiness degree correlated negatively with the yield in most of the groups, except for the amylose content in the early indica and chalkiness degree in the northern japonica.

The northern japonica, southern japonica, middle indica (upstream) and middle japonica groups had similar growth periods, whereas middle indica (downstream) showed a shorter growth period. The late indica and early indica groups exhibited the shortest growth period among the seven groups (Supplemental Fig. 1). No evident change in growth period was noted during the 15-year survey. The three japonica groups were more resistant to rice blast than the four indica groups, but large fluctuations occurred over time. Overall, the blast resistance improved during the duration of the survey. The plant height in the four indica groups was ranked as: middle indica (downstream) > middle indica (upstream) > late indica> early indica (Supplemental Fig. 1). The plant height in the three japonica groups was the same as the average of all the seven groups. In all groups, the height of plants remained constant during the 15-year period. The panicle length in the three indica groups was stable over time, and the panicles were the longest in the middle indica (upstream) and middle indica (downstream) groups. Among the four indica groups, the early indica had the shortest panicles. The panicle length in the three japonica groups was significantly shorter than in the four indica groups. Moreover, an increasing trend in the panicle length was noted in southern japonica and a decreasing trend in the middle japonica. Overall, no apparent change in panicle length was observed throughout the 15-year survey. The spikelet densities in the middle and southern japonica groups were significantly greater than those in the other five groups. The spikelet density in the four indica groups showed a similar trend as the northern japonica group, except for an abnormality in 2013. The spikelet density was slightly increased overall during the 15-year survey.

Supplemental Fig. 1.

	Figure Option ViewDownloadNew Window
	Supplemental Fig. 1. Other important agronomy traits for test accessions from 2004 to 2018. A, Trends of days to mature. B, Trends of blast level. C, Trends of plant height. D, Trends of panicle length. E, Trends of spikelet density for rest accessions from 2004 to 2018.

Next, the relationships among grain length to width ratio, yield and quality were analyzed (Supplemental Fig. 2). A significant positive correlation between the yield and grain length to width ratio was present in the middle indica (upstream), middle indica (downstream), late indica and all the three japonica groups. The grain length to width ratio correlated negatively with the brown rice rate in the japonica groups, and there was no significant correlation between grain length to width ratio and brown rice rate in the four indica groups. The grain length to width ratio significantly correlated with the head rice rate in all the four indica groups, and the correlation between grain length to width ratio and head rice rate was significant in the three japonica groups, except for the middle japonica. The grain length to width ratio had a significant negative correlation with the chalkiness traits in all the seven groups. However, the correlation coefficient was markedly higher in the indica than in the japonica groups. The 1000-grain weight was positively correlated with the yield and brown rice rate in all the seven groups. A significant negative correlation between 1000-grain weight and head rice rate was determined in all groups, but the correlation coefficient in indica was much higher than in japonica. A significant positive correlation between 1000-grain weight and chalkiness traits was present in both the indica and japonica groups. Again, the correlation coefficient in indica was evidently higher than in the japonica group. Interestingly, the correlation between spikelet density and quality traits showed a large difference between the indica and japonicagroups. The spikelet density was positively correlated with the brown rice rate and head rice rate in the indica groups, but negative in the japonica groups. Additionally, spikelet density showed a significant negative correlation to the chalkiness traits in the indica groups but a positive correlation in the japonica groups.

Supplemental Fig. 2.

	Figure Option ViewDownloadNew Window
	Supplemental Fig. 2. Relationship of grain shape, 1, 000, grain weight, and spikelet density with yield and quality. A, Correlation of grain shape to yield and quality traits. B, Correlation of 1000-grain weight to yield and quality traits. C, Correlation of spikelet density to yield and quality traits. * means significance at the 5% level.

In recent years, indications have appeared for the stagnation of yield in major rice-producing areas of Asia, suggesting a lack of genetic gain in the yield potential in the rice improvement program. The stagnant yield of semi-dwarf indica inbred varieties has been observed in the tropics since the introduction of IR8 in 1966 (Peng et al, 1999). An experiment using California rice also revealed that the improvements in rice yield achieved by breeding are offset by inherent yield declines over time (Espe et al, 2018). Peng et al (2010) found that the grain yield of IR8 (the first modern rice variety of the green revolution), from 1996 to 2003 has decreased by 15% compared to that in 1960s, and this reduction resulted from the lack of adaptation to environmental change. In China, based on the reports of the National Bureau of Statistics, the relationship between the average rice production (y) and time (x) is described by the equation: y = 52.87x - 99679 (R² = 0.9396). It can be calculated that the yields in 2004 and 2018 were 6.27 and 7.01 t/hm², respectively. The present analysis showed that for the new varieties, the relationship is described by y = 93.748x + 179926 (R² = 0.8289), indicating the production of 7.94 t/hm² in 2004 and 9.05 t/hm² in 2018. Thus, the introduction of new varieties gave 26.32% and 29.10% higher yields than the national average in 2004 and 2018, respectively. Given the worldwide stagnation of the yield, China has achieved an outstanding breeding success. Deng et al 2019 conducted a detailed spatial analysis of rice production potential in China. Utilizing the data from that study, we have compared rice production potential, actual yield and the yield of new varieties (Supplemental Fig. 3). The results documented that the yield of new varieties was close to the yield potential in low and middle latitude areas. However, the actual yield in these regions remained low.

Supplemental Fig. 3.

	Figure Option ViewDownloadNew Window
	Supplemental Fig. 3. AAAYield potential, actual yield and yield of new varieties in China. A, Yield potential. B, Yield of new varieties. C, Actual yield in China. D, Ratio of yield of new varieties to yield potential. E, Ratio of actual yield to yield of new varieties. F, Ratio of actual yield to yield potential.

Given the extremely large range of rice-farming areas in China, differences exist in ecological conditions, rice variety characteristics, cultivation methods, field management practices and growth periods. The present study showed that the yields of indica were 8.27 and 8.37 t/hm² in 2004 and 2018, respectively. The yields of japonica were 8.46 and 9.88 t/hm² in 2004 and 2018, respectively. Thus, the increases in the yield of indica and japonica were 13.27% and 17.22%, respectively. These results demonstrate that the breeding of japonica rice resulted in better yield improvement than the breeding of indica rice. The grain length to width ratio, 1000-grain weight and spikelet density were the major agronomical differences between these subspecies. Moreover, these traits were closely related to the yield and quality of rice and became the main target of breeding improvement for both breeders and scientists. The present study shows that the grain length to width ratio had little effect on the brown rice rate, but was significantly negatively correlated to the head rice rate in indica, while it affected primarily the brown rice rate in japonica. The head rice rate and chalkiness traits in japonica were superior to those in indica. However, the improvement of appearance quality in indica was significantly greater. The higher value of grain shape benefited the chalkiness traits but negatively affected the head rice rate. Although the improved grain length to width ratio in indica can result in a better appearance quality, it may also lead to the expense of the head rice rate. Unlike the middle and northern japonica groups, the grain length to width ratio of southern japonica correlated negatively with the head rice rate. However, the correlation coefficient between grain length to width ratio and chalkiness traits did not reach statistical significance. The present study showed that 1000-grain weight had a greater negative effect on head rice rate and chalkiness traits, and the effect was more pronounced in indica than in japonica. Thus, the increase in 1000-grain weight might be the best approach to increase the yield of japonica rice varieties. Increasing the spikelet density not only increased the yield but also improved the appearance quality and milling quality in indica. However, the appearance and milling qualities in japonica decreased, which might be caused by the higher spikelet density in this subspecies.

Since the beginning of the 21st century, significant progress in the improvement of the milling quality of indica has been achieved, resulting in narrowing the gap between indica and japonica subspecies. However, despite the improvement in the appearance quality of indica, the gap between indica and japonica continues to be present (Xu et al, 2016). The present analysis also demonstrated that the brown rice rate and head rice rate in the indica and japonica groups were stable during the 15-year survey, and the japonica displayed better milling quality. The chalkiness trait of indica, particularly in the late, middle (downstream) and early indica groups, was significantly decreased over time. No clear change was observed in the amylose content of japonica, while the amylose content of middle (downstream) and late indica has decreased markedly to reach a level similar to japonica.