Trait variations and correlations
All four measured traits showed tremendous variations in the given subpopulations across the three years (Table 1 and Additional file 1: Table S1). Among 2445 accessions, subtropical geng presented the highest TGW among all the groups primarily due to significantly longer GL and wider GW than other groups, followed by Xian-3 and intermediate geng. Although xian-1B, xian-3, subtropical geng and intermediate geng had similar GL, two xian groups (-1B and − 3) showed significantly larger RLW than two geng groups (subtropical and intermediate). Each group also showed continuous distribution in above four traits, with IRIS_313-12340 having the highest TGW (46.26 g) and IRIS_313-12336 having the widest GW (4.20 mm) in subtropical geng group; IRIS_313–9817 having the longest GL (11.73 mm) in xian-2 group; IRIS_313-15909 having the largest RLW (4.66) in basmati/sadri group,; and IRIS_313–8118 having the highest TGW (40.15 g) and longest GL (9.90 mm) in the temperate geng group.
Table 1
Performance of grain weight and grain shape in different subpopulations in 2015‒2017
Trait a | Group | Mean b | Range | CV (%) | Elite accession of each trait |
TGW (g) | Xian-1A | 24.73 de | 16.69–35.00 | 10.38 | B149 |
Xian-1B | 24.16 ef | 18.39–32.22 | 11.07 | IRIS_313-11517 |
Xian-2 | 24.55 e | 12.20–36.06 | 15.82 | IRIS_313–9817 |
Xian-3 | 26.32 bc | 13.00–38.41 | 19.64 | IRIS_313-11081 |
Intermediate xian | 24.65 e | 12.99–35.76 | 13.74 | CX182 |
Temperate geng | 25.86 cd | 17.47–40.15 | 12.39 | IRIS_313–8118 |
Subtropical geng | 31.44 a | 12.43–46.26 | 20.76 | IRIS_313-12340 |
Tropical geng | 25.66 cd | 8.20–36.31 | 16.52 | IRIS_313–7959 |
Intermediate geng | 27.86 b | 15.84–38.48 | 16.86 | IRIS_313–8121 |
Intermediate type | 25.00 cde | 15.50–35.09 | 14.78 | IRIS_313-11189 |
Aus/boro | 22.78 f | 11.73–32.48 | 16.92 | IRIS_313-11232 |
Basmati/sadri | 19.78 g | 11.54–33.75 | 22.55 | CX110 |
GL (mm) | Xian-1A | 7.75 d | 5.94–9.65 | 7.66 | B149 |
Xian-1B | 8.77 a | 6.56–10.38 | 7.56 | CX250 |
Xian-2 | 8.02 c | 5.17–11.73 | 10.77 | IRIS_313–9817 |
Xian-3 | 8.68 a | 5.70–10.72 | 10.61 | IRIS_313-11180 |
Intermediate xian | 8.33 b | 6.11–10.89 | 9.61 | IRIS_313–9841 |
Temperate geng | 7.5 de | 6.34–9.90 | 10.44 | IRIS_313–8118 |
Subtropical geng | 8.77 a | 5.90–11.45 | 13.08 | IRIS_313-11923 |
Tropical geng | 8.41 b | 5.30–11.18 | 8.98 | CX243 |
Intermediate geng | 8.52 ab | 6.54–11.67 | 13.21 | IRIS_313-10061 |
Intermediate type | 8.25 bc | 6.35–10.68 | 12.89 | IRIS_313–9818 |
Aus/boro | 7.67 de | 5.54–9.24 | 9.15 | IRIS_313-11481 |
Basmati/sadri | 7.28 e | 5.04–10.90 | 18.61 | CX110 |
GW (mm) | Xian-1A | 3.10 cd | 2.44–3.66 | 6.73 | IRIS_313-11986 |
Xian-1B | 2.65 g | 2.13–3.60 | 9.29 | B058 |
Xian-2 | 2.90 f | 1.99–3.57 | 9.89 | IRIS_313-10628 |
Xian-3 | 2.94 f | 2.14–3.92 | 11.82 | IRIS_313-11898 |
Intermediate xian | 2.89f | 2.02–3.89 | 9.57 | IRIS_313-10739 |
Temperate geng | 3.30 b | 2.37–3.88 | 7.09 | B111 |
Subtropical geng | 3.51 a | 2.44–4.20 | 10.77 | IRIS_313-12336 |
Tropical geng | 3.04 de | 2.23–4.15 | 12.7 | IRIS_313-10645 |
Intermediate geng | 3.19 bc | 2.35–3.82 | 11.85 | B144 |
Intermediate type | 3.00 def | 2.17–3.67 | 11.01 | IRIS_313-11189 |
Aus/boro | 2.95 ef | 2.11–3.77 | 10.2 | IRIS_313-11483 |
Basmati/sadri | 2.71 g | 2.12–3.43 | 13.84 | IRIS_313-11289 |
RLW | Xian-1A | 2.52 g | 1.77–3.66 | 12.29 | B239 |
Xian-1B | 3.36 a | 2.24–4.32 | 13.24 | IRIS_313–7808 |
Xian-2 | 2.81 de | 1.73–4.35 | 17.15 | IRIS_313-12082 |
Xian-3 | 3.00 b | 1.99–4.19 | 15.03 | IRIS_313-11683 |
Intermediate xian | 2.93 bc | 1.95–4.30 | 15.16 | CX363 |
Temperate geng | 2.31 h | 1.81–3.75 | 15.67 | IRIS_313–8160 |
Subtropical geng | 2.52 fg | 1.74–3.63 | 14.18 | IRIS_313-11923 |
Tropical geng | 2.84 cd | 1.80–4.20 | 17.85 | IRIS_313–8062 |
Intermediate geng | 2.75 def | 1.75–4.48 | 22.17 | IRIS_313-10061 |
Intermediate type | 2.83 cde | 2.00–4.56 | 21.69 | CX46 |
Aus/boro | 2.64 efg | 1.73–4.25 | 14.95 | IRIS_313–8390 |
Basmati/sadri | 2.77 def | 1.88–4.66 | 26.16 | IRIS_313-15909 |
a TGW: thousand grain weight, GL: grain length, GW: grain width, RLW: ratio of grain length to grain width |
b The letters indicate multiple comparisons result at the significant level 0.05 |
The phenotypic pairwise correlations between the measured traits were similar in three years (Additional file1: Table S2). TGW was significantly positively correlated with GL and GW, while it showed slightly negative correlations with RLW.
Basic statistics of markers
For the 4.8 M high-quality SNP data, the number of markers per chromosome ranged from 197,757 on chromosome 9 to 422,795 on chromosome 1. The length of the chromosome varied from 23.01 Mb for chromosome 9 to 43.27 Mb for chromosome 1. The whole-genome size was 373.24 Mb with an average marker density of 9085 SNPs/Mb (Additional file 2: Fig. S2; Additional file 1: Table S3).
Population structure and linkage disequilibrium (LD) decay
The 3D plot showed that five principal components (PC) can be retained as effective, exhibiting the distribution of each accession in the rice diversity panel (Additional file 2: Fig. S3). The LD decay in the whole panel was much faster than any other panels. The maximum LD was 0.47, 0.64, 0.69, 0.53, 0.68 and 0.67 in the whole panel, aus/boro, basmati/sadri, xian, temperate geng, and tropical geng, respectively. LD reached half of its initial value at around 300 kb in the basmati/sadri, temperate geng and the whole panel, 250 kb in the aus/boro panel, 130 kb in the xian panel and 260 kb in the tropical geng panel (Additional file 2: Fig. S4).
QTL detection by GWAS
A total of 22 important QTL were consistently identified for the four traits in at least two years (Table 2; Additional file 2: Fig. S5A–S5F). For TGW, six QTL were detected on chromosomes 3, 5, 7, 9 and 11. Among them, qTGW3.2, qTGW5, and qTGW7 were stably detected in more than one panel in the three years. Two QTL, qTGW3.1 and qTGW9, were detected only in the whole panel while qTGW11 were just detected in the xian panel.
Table 2
Important QTL consistently identified for grain weight and grain shape at least two years by GWAS
Trait a | QTL | Chr. | Peak SNP a | QTL region (Mb) | Allele b | P | Panel | Year of QTL detected | Previously cloned gene/QTL |
TGW (g) | qTGW3.1 | 3 | rs3_4504988 | 4.30–4.70 | C/T | 2.59E-08 | Whole | 2015, 2016 | |
qTGW3.2 | 3 | rs3_16733441 | 16.53–16.93 | T/G | 2.25E-08 | Whole | 2015–2017 | GS3 (Fan et al., 2006) |
| 3 | rs3_16733441 | 16.60–16.86 | T/G | 1.78E-11 | Xian | 2015–2017 | |
qTGW5 | 5 | rs5_5371529 | 5.17–5.57 | C/A | 7.22E-10 | Whole | 2015–2017 | qSW5/GW5 (Liu et al., 2017) |
| 5 | rs5_5361877 | 5.23–5.49 | T/A | 1.38E-10 | Xian | 2015–2017 | |
qTGW7 | 7 | rs7_28305040 | 28.10–28.50 | A/G | 8.12E-18 | Whole | 2015–2017 | FZP/BFL1 (Bai et al., 2017) |
| 7 | rs7_28762557 | 28.63–28.89 | G/A | 7.37E-08 | Xian | 2015–2017 | |
| 7 | rs7_29143304 | 28.94–29.34 | T/G | 6.23E-06 | Tropical geng | 2016, 2017 | |
| 7 | rs7_28322588 | 28.12–28.52 | T/C | 1.04E-09 | Basmati/sadri | 2015–2017 | |
| 7 | rs7_28287280 | 28.08–28.48 | A/G | 9.14E-12 | Aus/boro | 2015–2017 | |
qTGW9 | 9 | rs9_21404841 | 21.20–21.60 | C/T | 2.47E-08 | Whole | 2015, 2017 | |
qTGW11 | 11 | rs11_3019935 | 2.88–3.14 | T/C | 5.17E-07 | Xian | 2016, 2017 | |
GL (mm) | qGL3 | 3 | rs3_16733441 | 16.53–16.93 | T/G | 1.97E-73 | Whole | 2015–2017 | GS3 (Fan et al., 2006) |
| 3 | rs3_16733441 | 16.60–16.86 | T/G | 8.65E-59 | Xian | 2015–2017 | |
| 3 | rs3_16733441 | 16.53–16.93 | T/G | 2.54E-15 | Temperate geng | 2016, 2017 | |
| 3 | rs3_16733441 | 16.53–16.93 | T/G | 5.16E-09 | Tropical geng | 2015–2017 | |
| 3 | rs3_16728308 | 16.52–16.92 | C/T | 4.75E-09 | Aus/boro | 2016, 2017 | |
qGL4 | 4 | rs4_29308534 | 29.15–29.42 | C/T | 1.79E-06 | Whole | 2015, 2017 | |
qGL5 | 5 | rs5_5373491 | 5.17–5.57 | A/G | 2.49E-21 | Whole | 2015–2017 | qSW5/GW5 (Liu et al., 2017) |
| 5 | rs5_5361894 | 5.23–5.49 | A/G | 2.09E-18 | Xian | 2015–2017 | |
| 5 | rs5_5351264 | 5.15–5.55 | T/A | 3.44E-08 | Tropical geng | 2016, 2017 | |
| 5 | rs5_5288638 | 5.08–5.48 | G/A | 2.58E-07 | Aus/boro | 2015–2017 | |
qGL7 | 7 | rs7_28289869 | 28.08–28.48 | T/C | 4.62E-19 | Whole | 2015–2017 | FZP/BFL1 (Bai et al., 2017) |
| 7 | rs7_28787400 | 28.65–28.91 | C/G | 7.12E-07 | Xian | 2016, 2017 | |
| 7 | rs7_28290297 | 28.09–28.49 | G/A | 1.40E-10 | Tropical geng | 2015–2017 | |
| 7 | rs7_28287280 | 28.08–28.48 | A/G | 7.82E-08 | Basmati/sadri | 2016, 2017 | |
qGL10 | 10 | rs10_13616240 | 13.45–13.65 | A/C | 1.19E-07 | Whole | 2015–2017 | |
qGL11 | 11 | rs11_2044476 | 19.00–21.50 | G/A | 1.22E-06 | Aus/boro | 2016, 2017 | |
GW (mm) | qGW3 | 3 | rs3_16733441 | 16.53–16.93 | T/G | 6.82E-09 | Whole | 2015–2017 | GS3 (Fan et al., 2006) |
| 3 | rs3_16717406 | 16.58–16.84 | G/A | 2.21E-08 | Xian | 2015–2017 | |
qGW5 | 5 | rs5_5371772 | 5.17–5.57 | G/A | 1.84E-77 | Whole | 2015–2017 | qSW5/GW5 (Liu et al., 2017) |
| 5 | rs5_5361894 | 5.23–5.49 | A/G | 1.96E-71 | Xian | 2015–2017 | |
| 5 | rs5_5364791 | 5.16–5.56 | T/C | 3.73E-10 | Temperate geng | 2015–2017 | |
| 5 | rs5_5375786 | 5.17–5.57 | G/A | 2.07E-25 | Tropical geng | 2015–2017 | |
| 5 | rs5_5346606 | 5.14–5.54 | T/A | 2.16E-19 | Aus/boro | 2015–2017 | |
qGW7 | 7 | rs7_24898274 | 24.69–25.09 | A/T | 2.59E-16 | Whole | 2015–2017 | GL7/GW7 (Wang et al. 2015a; Wang et al. 2015b) |
| 7 | rs7_24902815 | 24.70–25.10 | A/G | 3.17E-07 | Tropical geng | 2015–2017 | |
qGW8 | 8 | rs8_26504638 | 26.30–26.70 | A/G | 3.58E-10 | Aus/boro | 2016, 2017 | GW8 (Wang et al., 2012) |
RLW | qRLW1 | 1 | rs1_3657795 | 3.45–3.85 | A/G | 1.42E-07 | Aus/boro | 2015, 2017 | |
qRLW3 | 3 | rs3_16733441 | 16.53–16.93 | T/G | 2.50E-55 | Whole | 2015–2017 | GS3 (Fan et al., 2006) |
| 3 | rs3_16733441 | 16.60–16.86 | T/G | 4.00E-43 | Xian | 2015–2017 | |
| 3 | rs3_16733441 | 16.53–16.93 | T/G | 4.11E-07 | Temperate geng | 2015–2017 | |
| 3 | rs3_16728485 | 16.52–16.92 | G/C | 6.65E-11 | Tropical geng | 2016, 2017 | |
| 3 | rs3_16686074 | 16.48–16.88 | G/A | 7.33E-08 | Basmati/sadri | 2015–2017 | |
qRLW4 | 4 | rs4_29317460 | 29.15–29.42 | C/A | 3.68E-09 | Whole | 2015–2017 | |
| 4 | rs4_29309086 | 29.10–29.50 | C/G | 1.14E-08 | Aus/boro | 2015–2017 | |
qRLW5 | 5 | rs5_5371609 | 5.17–5.57 | A/G | 3.59E-69 | Whole | 2015–2017 | qSW5/GW5 (Liu et al., 2017) |
| 5 | rs5_5361894 | 5.23–5.49 | A/G | 1.25E-67 | Xian | 2015–2017 | |
| 5 | rs5_5364791 | 5.16–5.56 | T/C | 9.49E-12 | Temperate geng | 2015–2017 | |
| 5 | rs5_5375793 | 5.17–5.57 | G/A | 5.48E-25 | Tropical geng | 2016, 2017 | |
| 5 | rs5_5375764 | 5.17–5.57 | A/G | 5.56E-15 | Aus/boro | 2015–2017 | |
qRLW7 | 7 | rs7_24533303 | 24.33–24.73 | T/C | 4.89E-13 | Whole | 2015–2017 | GL7/GW7 (Wang et al. 2015a; Wang et al. 2015b) |
| 7 | rs7_24537476 | 24.40–24.66 | C/T | 3.67E-07 | Xian | 2015–2017 | |
| 7 | rs7_24902815 | 24.70–25.10 | A/G | 8.02E-07 | Tropical geng | 2016, 2017 | |
qRLW8 | 8 | rs8_26504638 | 26.30–26.70 | A/G | 1.17E-14 | Aus/boro | 2015–2017 | GW8 (Wang et al., 2012) |
Note: For trait abbreviations, see Table 1. |
a The peak region in the chromosome. |
b Allele effect with respect to the minor allele. |
Six QTL for GL were mapped on chromosomes 3–5, 7, 10 and 11. The qGL3, qGL5 and qGL7 were consistently detected in more than one panel in the three years. The qGL4 and qGL10 were just detected in the whole panel whereas qGL11 was only detected in the aus/boro panel.
Four QTL were detected for GW on chromosomes 3, 5, 7 and 8. Three QTL, qGW3, qGW5, and qGW7 were stably detected in more than one panel in the three years. One QTL, qGW8 was detected only in aus/boro panel over the three years.
Six QTL affecting RLW were identified on chromosomes 1, 3–5. 7 and 8. Four QTL, qRLW3, qRLW4, qRLW5, and qRLW7 were consistently detected in more than one panel in the three years. Two QTL, qRLW1 and qRLW8 were detected only in the aus/boro panel.
We specified QTL with overlapping QTL regions identified for different traits or in different years as the same QTL, which lead to the identification of 12 different QTL (qTGW3.1, qTGW3.2/qGL3/qGW3/qRLW3, qTGW5/qGL5/qGW5/qRLW5, qTGW7/qGL7, qTGW9, qTGW11, qGL4/qRLW4, qGL10, qGL11, qGW7/qRLW, qGW8/qRLW8 and qRLW1).
Candidate genes for important QTL
The seven newly identified QTL regions were used for gene-based association analysis and haplotype analysis to identify candidate genes. In the region of 4.30–4.70 Mb on chromosome 3 harboring qTGW3.1, 9,177 non-synonymous SNPs of 38 genes were used for gene-based association analysis in the whole panel. SNPs with -log10(p) above threshold 6.0 centered around four candidate genes, Os03g0184400, Os03g0185200, Os03g0186600 and Os03g0189400 (Fig. 1a). Haplotype analyses indicated that Os03g0189400 showed no significant differences between various haplotypes (Additional file2: Fig. S8). Significant differences for TGW were observed among different haplotypes at Os03g0186600, Os03g0184400 and Os03g0185200 in the whole population. Of them, Hap3 with average TGW of 26.81 g,26.73 g and 26.81 g for Os03g0184400, Hap5 with average TGW of 28.19 g, 28.86 g and 28.90 g for Os03g0185200, and Hap3 with average TGW of 26.03 g, 26.13 g and 26.51 g for Os03g0186600 were favorable alleles across the three years (Fig. 1b–d).
The qTGW9 was identified for TGW in the region of 21.20–21.60 Mb on chromosome 9, 9,593 non-synonymous SNPs of 35 genes were used for gene-based association analysis in the whole panel. SNPs with -log10(p) above threshold 6.0 centered around three candidate genes, Os09g0544300, Os09g0544400, and Os09g0544800 (Fig. 2a). No haplotype differences were found in CDS regions of Os09g0544300, and Os09g0544800 (Additional file2: Fig. S9). Three major haplotypes within Os09g0544400 in the CDS region were detected in the whole panel with Hap1 associated with the highest TGW of 24.89 g, 25.13 g, and 25.47 g across the three years (Fig. 2b).
The qTGW11 was detected for TGW in the region from 2.88 Mb to 3.14 Mb on chromosome 11 in the xian panel, harboring 9,948 non-synonymous SNPs of 26 genes. SNPs with -log10(p) above the threshold 5.0 hits only one candidate genes, Os11g0163600 (Fig. 3a). However, significant differences in Os11g0163600 in the CDS region were detected only in 2016 and 2017 (Additional file2: Fig. S10). Hap2 of the three major haplotypes was the favorable allele with the TGW of 24.78 g, 25.29 g, and 25.40 g over the three years (Fig. 3b).
For qGL4/qRLW4 in the region of 29.15–29.42 Mb on chromosome 4, 9,638 non-synonymous SNPs of 23 genes were used for gene-based association analysis in the whole panel. Os04g0580500, Os04g0580700, and Os04g0580800 were presumed to be candidate genes (Fig. 4a–b). No haplotype differences for GL or RLW were found in the CDS region of Os04g0580500 and Os04g0580800 (Additional file2: Fig. S11). However, significant differences in GL and RLW were found in the CDS regions of Os04g0580700. Six haplotypes at Os04g0580700 were identified with Hap5 having the highest GL and Hap1 having the highest RLW (Fig. 4c).
qGL10 was predicted in the region of 13.45–13.65 Mb on chromosome 10 covering 15,535 non-synonymous SNPs of 13 genes. These SNPs were used for gene-based association analysis in the whole panel. Two genes (Os10g0399700 and Os10g0400100) harboring SNPs with -log10 (P) larger than the threshold 6.0 were identified (Fig. 5a). Highly significant differences for qGL10 among different haplotypes in the CDS regions were observed at the two genes in the whole population (Additional file2: Fig. S12). Three major haplotypes for Os10g0399700 and four major haplotypes for Os10g0400100 were detected, and Hap3 of Os10g0399700 (8.39 mm, 8.42 mm and 8.70 mm) and Os10g0400100 (8.39 mm, 8.43 mm and 8.71 mm) were associated with significantly longest GL in the three years (Fig. 5b‒c).
In the region from 19.00 to 21.50 Mb on chromosome 11 harboring qGL11, 7,866 non-synonymous SNPs of 32 genes were used for gene-based association analyses in the aus/boro panel. SNPs with -log10(p) above the threshold 5.0 were found to locate in two candidate genes, Os11g0143700 and Os11g0144500 (Fig. 6a). Haplotype differences were found in the CDS region only for Os11g0143700 (Additional file2: Fig. S13). Two major haplotypes of Os11g0143700 were found. Of them, Hap2 for Os11g0143700 (7.78 mm, 7.80 mm and 8.21 mm) was associated with significantly longer GL in the three years (Fig. 6b).
On chromosome 1, a high peak of qRLW1 was found, 8,116 non-synonymous SNPs of 23 genes were used for gene-based association analysis in the predicted candidate region from 3.45–3.85 Mb in the aus/boro panel. SNPs with -log10(p) above threshold 5.0 centered around two candidate genes, Os01g0171000 and Os01g0171100 (Fig. 7a). Haplotype differences were found in the CDS regions of both Os01g0171000 and Os01g0171100 (Additional file2: Fig. S14). Hap2 for Os01g0171000 and Hap2 for Os01g0171100 were found to harbor the highest RLW (Fig. 7b–c).