A total of 2,618 experimental materials were sourced from the National Rice Regional Trial Standard Sample Library in Fuyang, Hangzhou, encompassing the main representative hybrid rice varieties approved by national and provincial governments over the past 40 years. The breeding age and ecological classification data for all experimental materials were obtained from the National Crop Variety Testing and Operation Management Platform (http://202.127.45.151/NTP/login.jsp) and the National Rice Data Center (https://www.ricedata.cn/variety/). The phenotype data for rice appearance quality were derived from the phenotypic experiments of hybrid rice (Gu et al., 2023), while the genotype data were obtained from the sequencing results of previous studies and the “RiceNavi” system (Wei et al., 2021).

Based on the ecological characteristics of rice cultivation in China and the latest method for dividing the same adaptive ecological regions, the experimental materials were classified into six ecotypes: early-season cultivation of indica in South China (E1), late-season cultivation of indica in South China (E2), mid-season cultivation of indica in the upper reaches of the Yangtze River (E3), early-season cultivation of indica in the middle and lower reaches of the Yangtze River (E4), mid-season cultivation of indica in the middle and lower reaches of the Yangtze River (E5), and late-season cultivation of indica in the middle and lower reaches of the Yangtze River (E6). These ecotypes cover the most important ecological types in the southern rice region of China ( Supplementary Table S1 ). The chronological classification of materials was divided into three stages—before 2000 (Y1), 2001–2010 (Y2), and 2011–2020 (Y3)—with varieties accounting for 13.77%, 48.37%, and 34.16%, respectively ( Supplementary Table S2 ).

The details can be found in Supplementary Table S4 .

An investigation was conducted on the quality traits of 2,618 hybrid rice varieties, analyzing eight phenotype data ( Table 1 ). The results revealed substantial variation among the quality traits, with coefficients of variation ranging from 6.54% to 87.56%. The coefficients of variation for chalky grain ratio (CR) and chalkiness degree (CS) were the highest, at 64.36% and 87.56%, respectively. The coefficient of variation for grain shape ranged from 6% to 8%. The significant variation observed among various quality traits provides favorable conditions for screening high-quality germplasm resources.

We conducted a correlation analysis on the phenotype data of related traits, and the results indicated that there was a certain correlation among various traits, with most traits showing extremely significant correlations ( Table 2 ). There was a very significant negative correlation between grain length (GL) and grain width (GW), a very significant positive correlation between GL and grain length–width ratio (GLWR), and a very significant negative correlation between GW and GLWR. Additionally, there was a highly significant positive correlation between CR and CS. There was a highly significant negative correlation between chalkiness traits (CR and CS), alkali spreading value (ASV), and gel consistency (GC) and a highly significant positive correlation with amylose content (AC). These findings suggest that there is a significant negative or positive correlation among rice quality traits. Therefore, in the breeding process aimed at improving the quality traits of hybrid rice, it is essential to comprehensively consider various traits to achieve a perfect balance.

The quality traits exhibited different trends over time, specifically manifesting as increased grain length, decreased grain width, a significant reduction in chalkiness, and a gradual decrease in amylose content. The trend of alkali spreading value changed with the evolution of time, with relatively minor fluctuations ( Figure 1 ). This result indicates that the high-quality rate of hybrid rice has been gradually increasing in China.

This study classified 2,618 materials into six ecotypes (E1, E2, E3, E4, E5, and E6) based on different planting areas and maturity periods ( Supplementary Table S1 ). In different ecotypes, quality traits exhibited varying degrees of variation. The results indicated that the variation amplitude of appearance quality (AQ) and cooking and taste quality (CTQ) traits differed across ecotypes ( Figure 2 ). The quality of hybrid rice varieties in ecotypes E2 and E6 was relatively high. In contrast, the quality of hybrid rice varieties in ecotype E4 was relatively poor, particularly in terms of chalkiness and gel consistency.

Phenotypic analysis was conducted on the six different ecotypes according to their chronological changes. The results indicate that with the change of age, there is also a certain range of changes in AQ and CTQ traits ( Figure 3 ). Grain shape exhibited a small range of changes, with variation differences of 1–2 mm. There were significant differences in the variation of chalkiness traits, with CR ranging from 15% to 55% and CS ranging from 2.5% to 22.5%. The range of changes in other traits varied, with the smallest being the range of changes in ASV, with a difference of less than 1%. The largest is the change in AC, with a range of changes of approximately 10%. Additionally, with the passage of time, the AQ and CTQ traits showed a favorable trend of change. Except for E4, in the other five ecotypes, the same quality trait exhibited similar changes, with GL becoming longer and GW becoming shorter. The CR and CS gradually decreased, and the ASV became higher. However, GC and AC showed consistent changes in each ecological type, with higher GC and lower AC.

Based on the genome resequencing of 2,618 hybrid rice varieties from previous studies (Gu et al., 2023), we utilized the “RiceNavi” system (Wei et al., 2021) to perform genotype analysis on genes related to quality traits ( Supplementary Table S3 ). We organized the genotype and phenotype data and conducted a correlation analysis of R ² ( Table 3 ). The results showed that GS3 has the highest correlation with GL, GW, and GLWR. The Waxy gene has the highest correlation with CR, CS, GC, and AC. ALK has the highest correlation with ASV.

We further analyzed the allele forms of 16 genes in hybrid rice and found that most of the genes affecting AQ and CTQ in breeding were applied in homozygous genotypes, while a small number were applied in heterozygous genotypes, such as GS5, GS6, Chalk5, Waxy (Wx^lv/Wx^a/Wxⁱⁿ/Wx^op/Wx^hp ), Waxy (Wx^a ), and ALK(ALK-4328C>T) ( Figure 4 ). The results have a certain guiding significance for the genetic improvement of maintainer and restorer lines in future hybrid rice breeding.

We have integrated representative approved varieties from the past 40 years, sequenced each sample to obtain genotype data, and screened out excellent genes that significantly impact AQ and CTQ. We analyzed the allele frequency of these excellent genes. The results showed that with the change of year interval, the allele frequency of excellent genes exhibited a hierarchical change ( Figure 5 ). Among these, the frequency of the altered genotype in GS3, GS5 (GS5-1), GW8 (GW8-5174C>A), Chalk5, and ALK (ALK-4328C>T) has been increasing year by year. The frequency of the reference genotype in GW5 has been decreasing year by year. The frequency of the alterant genotype in Waxy (Wx^lv/Wx^a/Wxⁱⁿ/Wx^op/Wx^hp ) and Waxy (Wx ^a) has been decreasing year by year.

In hybrid rice breeding, combinations of reference alleles of GL3.3, GW3p6, GW5, GW7, BG2(2), OsSNB, and OsARE2; alterant alleles of GS3, GL3.2, OsLG3, BG(1), and GW8; and heterozygous alleles of GS5, GS6, and Chalk5 may result in better AQ. The combination of the reference alleles of Waxy (Wx^mp ), Waxy (Wx^op/hp ), and Waxy (Wxⁱⁿ /Wx^mw ); the alterant allele of ALK(AIK-3797A>G); and the heterozygous alleles of Waxy (Wx^lv/Wx^a/Wxⁱⁿ/Wx^op/Wx^hp ), Waxy (Wx^a ), and ALK(AIK-4328C>T) may result in better CTQ.

We further investigated the allele frequency of variations in 16 genes affecting AQ and CTQ across different ecotypes ( Figure 6 ). We found that grain length showed a slight increase from E1 to E6, with the largest change observed in E3 ( Figure 3 ). According to the variation of allele frequencies of excellent genes, the allele frequency of the GL3.2 heterozygous genotype was the lowest in E3. Grain width reached its highest in E3 and its lowest in E1 ( Figure 2 ). From the variation of allele frequencies of excellent genes, the GS5 heterozygous genotype had the lowest variation allele frequency in E3, while the GW5 reference genotype and GW7 reference genotype had relatively low allele frequencies in E1.

Chalkiness exhibited different changes across the six different ecotypes, with CR and CS being the highest in E4 and lower in E1 and E2 ( Figure 2 ). The allele frequencies of Chalk5, a major gene affecting chalkiness, varied among the six ecotypes. The Chalk5 heterozygous genotype had higher allele frequencies in E1, E2, and E3. The allele frequencies of the Chalk5 reference genotype were higher in E4. AC differed across different ecotypes. In E1, E2, E3, E5, and E6, AC showed a decreasing trend. E4 consistently exhibited high AC ( Figure 2 ). By analyzing the variation allele frequency of Waxy, the allele frequencies of Waxy (Wx^lv/Wx^a/Wxⁱⁿ/Wx^op/Wx^hp ) and Waxy (Wx ^a) heterozygous genotypes were the highest in E4.

Among the six different ecological types, ASV was the highest in E2, but the ALK reference genotype had the highest allele frequency in E4. The combined alleles of ecotypes with better AQ tended to exhibit GL3.3, GW3p6, GW5, GW7, and OsSNB as reference alleles and GS3, GL3.2, OsLG3, BG2(1), OsARE2, and GW8 (gw8 ^Basmati) as alterant alleles, with GS5, GS6, GW8(GW8-5174C>A), and Chalk5 as heterozygous alleles. The combined alleles of ecotypes with better CTQ tended to exhibit Waxy (Wx^mp ), Waxy (Wx^op/hp ), and Waxy (Wxⁱⁿ /Wx^mw ) as reference alleles; ALK(ALK-3797A>G) as an alterant allele; and Waxy (Wx^lv/Wx^a/Wxⁱⁿ/Wx^op/Wx^hp ), Waxy (Wx ^a), and ALK (ALK-4328C>T) as heterozygous alleles. These results indicate a certain correlation between genes and quality, but quality is determined by comprehensive factors, including environmental factors.

Grain shape and chalkiness are the main traits that determine AQ. Grain shape-related traits include grain length, grain width, grain thickness, and length–width ratio, while chalkiness can be categorized into white core, white belly, and white back based on location. CTQ refers to the comprehensive evaluation of sensory indicators such as smell, color, shape, palatability, taste, and other sensory indicators after cooking under specific conditions. Current evidence suggests that the main indicators affecting CTQ are amylose content, protein content, fat content, gel consistency, alkali spreading value (gelatinization temperature), and cooking method (Zhou et al., 2019).

We analyzed the correlation coefficients of eight traits, including GL, GW, GLWR, and CR, among 2,837 hybrid rice varieties ( Table 2 ). This study demonstrated that the appearance quality traits of rice not only acted independently but also influenced each other. It was found that AC was significantly correlated with CR, CS, and GC. CR was significantly correlated with CS (0.879**). GLWR was significantly correlated with GC, CS, and CR. These results are consistent with those of previous studies (Li et al., 1995; Xu et al., 2004; Lei et al., 2010; Yang et al., 2023). Therefore, elongated rice grains generally exhibit low chalkiness. Reducing grain width, increasing grain length, and grain length–width ratio can reduce chalkiness and improve the appearance quality of rice. High chalkiness is the most prominent problem in the quality of hybrid rice and is also the main target of hybrid rice breeding for high quality. Thus, it is particularly important to select parents with low chalkiness. In terms of quality traits of hybrid rice, the selection of hybrid parents whose main goal is to improve the AQ and CTQ of hybrid rice should be based on the quality level of parents and the genetic diversity of germplasm.

In general, for high-quality rice, it is required to have longer GL, smaller GW, low CR, low CS, higher ASV, higher GC, and lower AC (Li et al., 1987; Tang, 1987; Fan et al., 2005; Wang et al., 2007). In this study, we found that the GL of hybrid rice gradually increases, GW gradually decreases, AC gradually decreases, and chalkiness gradually improves with the passage of time in China. Fang found that the number of high-quality hybrid rice varieties has been increasing year by year in China (Fang et al., 2020), and the result of this study also reflects that the quality rate of hybrid rice varieties in our country has been increasing year by year. This may be due to the improved quality of some hybrid rice parents (Feng et al., 2017).

In terms of ecotypes, the quality of rice in South China was better than that in the middle and lower reaches of the Yangtze River. In particular, the quality of E4 has not been significantly improved in recent decades, but chalkiness and ASV were even worse. There may be two main reasons for the poor quality of E4 rice. For one thing, early rice has a short growth period, and in order to increase yield, the selection process often tends to favor wide grains. The synergistic effect of chalkiness in wide-grain genes leads to an increase in chalkiness and a decrease in quality. In contrast, around mid-June, the E4 grain filling coincides with the continuous rise in temperature in the middle and lower reaches of the Yangtze River. High temperatures can lead to poor early rice enrichment, reduced polished rice, and increased chalkiness. In the future, breeders may need to develop rice varieties that can produce high-quality rice at high temperatures.

In this study, we analyzed the genotypes of 14 genes (GS3, GL3.2/CYP78A5, GL3.3/qTGW3, OsLG3, qLGY3/OsMADS1/GW3p6/OsLG3b, GS5, GW5/GSE5/qSW5, GS6/DLT/OsGRAS-32/SMOS2/D62, GW7, GE/CYP78A13/BG2, OsSNB, OsRAE2, GW8/OsSPL16, and Chalk5) regulating AQ and two genes (Waxy and ALK) regulating CTQ in 2,618 hybrid rice varieties. By studying the allele frequency of these genes at different ages, we found that breeders were intentionally or unintentionally utilizing these alleles that are beneficial for improving AQ and CTQ in hybrid rice, leading to the breeding of high-quality hybrid rice varieties. The more notable ones include GS3, which regulates GL and GW; Chalk5, which regulates chalkiness; and Waxy, which regulates AC. This directly indicates from a molecular perspective that the process of high-quality hybrid rice in China was directly related to these alleles. However, in terms of different ecotypes, some favorable allele utilization still needs to be improved, particularly in E4. In the future, using molecular breeding to pyramid excellent quality genes to improve early rice varieties may be a fast and effective method.