Prediction Of Yield Component Performances In Hybrid Rice (Oryza Sativa Indica L.) Using Molecular Marker Technology

With the development of social economy, greater demand is put on world food production due to the decreasing arable lands and increasing global population. Heterosis utilization has become an important approach to increase crop yield since its success in maize in 1930s. Favorable economic and social benefits have been obtained due to its extensive utilization in both crops (maize, rice, rape, sorghum, wheat, etc.) and vegetables. Heterosis is a very complex phenomenon and is generally evaluated by combining ability. These demand large numbers of test-crosses and field evaluations, and which are labor-consuming and time-consuming, while the results are prone to be erratic due to different environments. Methods of morphological indices and biochemical markers, etc. were adopted to predict heterosis in earlier studies, and these results might be referred but could not be directly utilized due to lower accuracy. With the appearance of molecular markers in 1980s, study on prediction of heterosis has stepped into a new stage. In recent years, RFLP, RAPD, AFLP and SSR etc. have been reported to predict heterosis by means of NCII design,etc.Limitations of the former studies on heterosis prediction with molecular marker were mainly manifested in two aspects. Firstly, total loci were utilized to predict heterosis in former research. In fact, polygenes function with different dosages and directions (positive or negative) of genetic effects on heterosis. Consequently, effects of different loci related would be counteracted if they were not distinguished, and negative correlations might also appear in different environments. This might be responsible for the contrary conclusions on heterosis prediction by means of genetic distance (GD). Positive markers were once screened for heterosis prediction and preferable results were obtained, however descrepancy existed due to different materials and environments. Secondly, correlaion between heterosis and GD was analysed, but prediction for performance of crosses with different parents was not conducted yet. Because the prediction results varied with both materials and traits, even though there was high correlations between heterosis and GD for given materials, the conclusion could not be concluded to other combinations with different parents. To resolve the above problems, this research studied the feasibility of prediction for F₁ yield components performances using various materials under two environments with effect-increasing loci (IL) screened by F₁ performances, which will provide a new strategy for heteosis prediction.1. Genetic difference changes of main restorer and sterile lines in different stages of hybrid riceMany good restorers and sterile lines of WA have been breeded for 30 years since the success of three-line hybrid rice. Study on genetic diversity dynamics of widely used sterile and restorer lines of hybrid rice at different stages will help to choose parents efficiently and to improve hybrid rice yield. In present research, 31 pairs of AFLP primers and 174 pairs of SSR primers were used to detect GDs of 19 restorers and 13 sterile lines of WA-type hybrid rice that were widely used at different stages. 32 rice materials were the parents of two half-dellel sets in present prediction study. All polymorphic bands were recorded by 0/1 method. GD of Nei & Li between all possible pairs of parents were calculated using AFLP and SSR data. Chi-square distances between the parents were calculated based on the standardized values of 15 agronomic traits of 32 parents. 32 rice parents were clustered based on GD of Nei and Li and on Chi-square distances by UPGMA respectively.(1) 535 (AFLP) and 550 (SSR) clear polymorphic bands were detected among 32 parents, and on average 17.3 and 3.2 distinguishable bands were observed after amplification with each primer pair, respectively.(2) The result of dendrogram based on AFLP or SSR markers was similar to that on agronomic traits analyses in grouping of 32 parents of restorers and sterile lines, but a few differences between marker and trait dendrograms were detected in diversity among sterile or restorer lines.(3) There were relatively larger genetic differences among 32 parents, and the GDs between sterile and restorer lines were larger than those among restorers as well as among sterile lines. Larger genetic difference between restorers and sterile lines was the basis of high yield of hybrid rice.(4) The GDs among early-stage restorers were lower than those among middle-stage and among late-stage restorers. Similar trend was observed on the GDs between sterile lines and restorers in 3 stages. These trends were positively correlated with the yield trend of hybrid rice, i.e. the yield of hybrid rice was largely improved in the middle-to-late 1980s and was stunted after 1990s. This indicated that GDs based on molecular makers can be used as one reference in choosing restorers in breeding.(5) The GDs among late-stage sterile lines were larger than those among early-stage sterile lines, but the GDs between early-stage sterile lines and restorers showed no significant difference from those between the late-stage sterile lines and restorers. It was necessary to introduce or improve sterile lines and restorers for raising yield of hybrid rice.2. Prediction of F₁ yield components performance with GD on molecular markers151 F₁ produced in two half-diallels sets (7×10 and 9×9) were planted in two locations (Chongqing and Luzhou, sichuan). By loci-screening from SSR and AFLP polymorphic bands data using F₁ performances of seven yield components by two-group method. PL and IL were obtained and utilized to predict the hybrid yield components performances of crosses with different parents, changed environment, and fixed parents, respectively.(6) Correlations between parental GDs based on IL with F₁ performances were larger than those between GD based on PL and the traits as well as between GD based on total loci and the traits. Mean of the former were above 0.7.(7) Predictions between two environments conducted by IL were better for yield component traits (except grain weight per plant), but those by PL and IL were unfaithful between two sets.(8) IL was more veracious than PL in predicting performances of combinations with the fixed parents in Set 2, and predictions of crosses with fixed restorers were better than those of fixed CMS lines. For example, predictions for panicle numbers of crosses with the fixed restorers by total IL of two type of markers were above 0.7; Those for grain setting percentage and grain weight per panicle of crosses with the fixed restorers and of grain setting percentage in crosses with the fixed sterile lines were above 0.6. These conclusions are helpful to facilitating molecular prediction of heterosis by mean of elite sterile lines and restorers.3. Prediction of hybrid yield components performance with hybrid marker valuePL and IL related to corresponding yield component trait were screened by three-group method from AFLP and SSR loci data. Some PL had only additive or dominant effect, while the other had both of them. Marker types of hybrid (1/1 or 1/0 or 0/0) could be conducted by parents' mark types (1 or 0) of each cross. Marker value of each cross for each PL were obtained by marker types of hybrid and marker effect. Prediction model were founded by stepwise regression analysis for marker values using yield component performances.(9) In prediction model, marker value of F₁ had highly positive relationship to corresponding hybrid performances for seven yield component traits, being about 0.9. The least was above 0.7 for grain weigh per plant.(10) Prediction by model founded by marker value for yield component traits performances of crosses with different parents were low. Only 2/5 of the all predictions with the total of two sorts of markers were significantly correlated, and among those, 1/3 being negative.(11) In all, prediction with marker value for crosses with fixed restorers were better than those for crosses with fixed steriles, with mean of 0.50 and 0.29 respectively.(12) It was interesting that in the prediction with marker value of crosses with fixed parents, prediction effects with PL were similar to those with IL. But the latters were better than the formers when predicted with GD of molecular marker. This indicated that part interferential loci in PL were eliminated and the prediction precision of PL was improved.