Studies On Molecular Genetic Basis Of Yield-Related Traits And Its Heterosis In Japonica Rice (Oryza Sativa L.)

Rice (Oryza sativa L.) growing area in China is 30,670,000 ha each year. Among them, indica hybrid rice planting area is 17,330,000 ha and accounts for more than 80 percent of China indica rice planting area and 50 percent of China’s rice growing area. Japonica rice growing area in China is 8,280,000 ha annually. The area planted with japonica hybrid rice only occupied 3 percent of the total area of japonica rice in China. Therefore, great space exists for developing japonica hybrid rice, compared with indica hybrid rice, in which great achievement had been made. The major reason for low speed of japonica hybrid rice development is that competitive heterosis of hybrid cultivar is not conspicuous in yield, compared with conventional cultivar in japonica rice. Dissecting molecular genetic basis of yield-related traits and its heterosis is helpful to improve competitive heterosis of hybrid cultivar in yield by molecular marker-assisted selection (MAS). Four studies were carried out by using the recombinant inbred line population ("Xiubao RIL population" for short hereinbelow) contained 254 lines derived from a cross between Xiushui 79 (japonica cultivar variety) and C Bao (japonica restorer line) and their parents in this study. Firstly, genetic segregation analysis of the five panicle traits were conducted by using the mixed major gene plus polygene inheritance model for P1, P2, F1, B1, B2 and F2 generations of two crosses, which were made by using two lines having panicles with the most spikelet number and two lines having panicles with the least spikelet number selected from Xiubao RIL population. Secondly, dynamic QTL analysis of tiller number (TN) and seedling height (SH) in different investigated stages were performed by using Xiubao RIL population across environments. Thirdly, unconditional QTL mapping and conditional QTL mapping of growing duration (GD), plant height (PH) and panicle number per plant (PN) were performed by using Xiubao RIL population in two environments. Finally, QTLs of ten yield-related traits and their mid-parental heterosis were detected by using of the Xiubao RIL population and the two backcross populations. The main results are as follows:1. The transgressive segregation of the five traits except spikelet number per panicle (SNP) was observed in three segregation generations (B1、B2 and F2) in both of the two crosses. The result indicated the lines having panicles with the most spikelet number polymerized all exhibited positive alleles from two parents, whereas the lines having panicles with the least spikelet number polymerized all exhibited negative alleles from two parents. By using major gene-polygene mixed inheritance models, genetic analyses showed that SNP, filled grain number per panicle (FGP), panicle length (PL) and secondary branch number per panicle (SBN) were controlled by two major genes plus polygenes. The four traits were mainly governed by major genes. Primary branch number per panicle (PBN) was controlled by one major genes plus polygenes. The trait was mainly governed by polygenes.2. Increase and decrease of TN were controlled by different loci in rice of all development stages. Tiller numbers of 254 recombinant inbred lines and two parents, Xiushui 79 and C Bao, were recorded every 14 days until maturity across two environments, Nanjing and Sihong. Genetic effects for TN at different measuring stages were estimated by the mixed line model and the best linear unbiased prediction method. Static loci and dynamic loci affecting tiller numbers were detected by using unconditional and conditional QTL mapping methods. Thirteen unconditional additive QTLs were identified for TN at nine stages. For the identical locus detected at various stages, positive alleles came from the identical parent. Seven of the 13 conditional additive QTLs were detected from stage 1 to stage 4 when TN increased. Xiushui 79 carried positive alleles for qTN4 and qTN7.1, and C Bao carried positive alleles for qTN2.1, qTN5.1, qTN5.2, qTN9.1 and qTN10. The remaining 6 loci (qTN2.2, qTN3, qTN8.1, qTN8.2, qTN11.1 and qTN11.2) were detected between stage 5 and stage 9 when TN decreased. Alleles which decreased tiller mortality were except for qTN8.2, from Xiushui 79. Within the 13 conditional QTLs detected, number of elite alleles contained by the RILs extremely significantly positive correlated with productive number per plant (r=0.347**) of the lines. These results indicate that tiller morphogenesis and mortality are controlled by different loci, and it is possible to enhance productive panicles per plant by pyramiding the elite alleles at different stages.3. Genetic effects of loci affecting SH were different at different growing stages. SH of 254 recombinant inbred lines and two parents, Xiushui 79 and C Bao were measured at nine investigated stages by subjected to three different environments. Genetic analysis was conducted by using the same method as mentioned above "2". The result showes that fifteen unconditional additive QTLs were identified at nine different developmental stages. For the identical unconditional additive locus detected at various stages, alleles with positive effect came from the identical parent. And the additive effect increased with the plant growth. Sixteen conditional additive QTLs and sixteen epistatic QTL pairs involved in SH were identified at nine measurement stages. It shows that these loci of SH exhibited the temporal expression pattern. Total additive genetic effect and total expained phenotypic variability of conditional QTL shows multimodal distribution in whole development stages. The result indicated genetic effects of loci affecting SH were different at different growing stages. Total expained phenotypic variability of epistatic QTL significant less than that of additive QTL from t1|t0 to t8|t7, whereas both of them was consistent in t9|t8. It reflected that the additive effect was the major genetic effect at the period from sowing to 98d after transplanting, whereas SH was controlled by both additive effect and epistatic effect during 98d and 112d. Effect of GxE interaction was small during all developmental stages.4. Applicable elite allele of target trait can be mined by the combination of unconditional with conditional mapping. Unconditional QTL mapping and conditional QTL mapping were conducted for GD, PH and PN using Xiubao RIL population. The RIL population consisted of 254 lines and two parents were planted in two environments, Nanjing and Sihong. Result showed that additive effects were major in all of QTLs for GD, PH and PN detected by the two methods. After GD was adjusted to an identical level, RM80-160bp was detected as an applicable elite allele for PN, with additive effect 0.71. After PN was adjusted to an identical level, RM448-240bp was detected as an applicable elite allele for GD, with additive effect 4.64. After PH was adjusted to an identical level, RM80-160bp was detected as an applicable elite allele for PN, with additive effect 0.62, and RM448-240bp was detected as an applicable elite allele for GD, with additive effect 3.89. These applicable elite alleles could be used to improve target traits without influencing the adjusted trait.5. The heterosis in japonica rice is attributable to the orchestrated outcome of additive by non-additive and dominant by dominant interactions. QTLs of GD, PH, PN, PL, SNP, spikelet ferlitity (SF), spikelet density (SD), PBN, SBN and secondary branch distribution density (SPD) were detected by using phenotypic value in Xiubao RIL population, and BCF1 phenotypic value and mid-parental heterosis value in the two backcross populations, XSBCF1 and CBBCF1.78 M-QTLs (Main-effect QTLs) were identified in the 3 population. The percentage of phenotypic variance explained by each QTL ranged from 2.4%to 41.9%. 79.5%(62) of the QTLs detected showed an additive effect,11.5%(9) a partial-to-complete dominant effect, and 9.0%(7) an overdominant effect.114 pairs of QTL were detected in the 3 populations showing digenic interactions. Among them,58 pairs of E-QTL were detected in RIL population, and the percentage of phenotypic variance explained by each pair of QTL ranged from 1.7% to 8.0%, with an average 3.7%. In XSBCF1 population,29 pairs of E-QTL were detected.17 pairs of E-QTL were detected by using XSBCF1 phenotypic value, and the percentage of phenotypic variance explained by each E-QTL ranged from 10.9% to 78.5%, with an average 29.8%.12 pairs of E-QTL were detected by using mid-parental heterosis value (HMP), and the percentage of phenotypic variance explained by each E-QTL ranged from 15.0% to 71.8%, with an average 46.5%. In CBBCF1 population,27 pairs of E-QTL were detected.15 pairs of E-QTL were detected by using BCF1 phenotypic value, and the percentage of phenotypic variance explained by each pair of E-QTL ranged from 2.7% to 64.4%, with an average 29.7%.14 pairs of E-QTL were detected by using the mid-parental heterosis value (HMP), and the percentage of phenotypic variance explained by each pair of E-QTL ranged from 21.2% to 64.1%, with an average 36.2%.2 pairs of E-QTL were detected by using both BCF1 phenotypic value and HMP value in CBBCF1 population. These results showed that additive×non-additive and dominant×dominant interactions effect were the primary genetic basis of heterosis in Xiubao crosses in japonica rice.