分子标记遗传效应预测杂交水稻产量性状

Prediction of F_1 Yield Using Genetic Effects of Molecular Marker in Indica Rice(Oryza sativa L.)

选用13个水稻不育系和19个恢复系按NCII设计配制两套不完全双列杂交组合151个，2005年分别在重庆和泸州种植，结合AFLP和SSR标记位点的遗传效应，预测7个产量性状值。用两套组合F1，产量性状对标记多态性位点进行筛选，获得阳性位点和增效位点及其加性和（或）显性效应值，通过逐步回归构建分子标记遗传效应预测产量性状模型，并对不同亲本组合（套间预测）及固定亲本组合产量性状进行了预测。结果表明：（1）阳性和增效位点进行套间预测的效果，绝大多数不理想（预测值与实际值的相关系数为一0．55～0．45），且预测效果不稳定；（2）阳性和增效位点对第二套固定亲本组合的预测效果相近，均优于套间预测，其中对固定恢复系组合的预测好于对固定不育系；（3）对固定亲本组合大部分产量性状的预测达到较理想水平，其中增效位点对固定不育系组合结实率、固定恢复系组合有效穗数、阳性位点对固定恢复系组合穗着粒数、单穗重的预测在0．6以上。用分子标记遗传效应预测产量性状，只需少数标记，应用方便，特别是对水稻育种亲本选配具有重要指导意义。

One hundred and fifty-one rice hybrids produced in two sets of half-dialell mating crosses designed according to NCⅡ , planting respectively in Chongqing and Luzhou, Sichuan Province of China in 2005 were used to predict the F1 performances of seven yield traits through genetic effects of marker loci of AFLP and SSR. Positive loci （PL）and effect-increasing loci （IL）were screened respectively by yield traits of two sets of crosses and their ad- ditive or/and dominant effects were used to founded the prediction model for these yield traits. Predictions of yield traits of crosses derived from different parents （ prediction between sets） and fixed parents were conducted by the model founded. The results were as follows： （ 1 ） Most predictions between sets were not ideal and changeable. The correlations between predictive values and actual ones ranged from - 0.55 to 0.45. （2） The prediction effects of crosses with by PL were similar to those by IL in Set 2, and all better than that of between sets. Predictions of cros- ses with fixed restorers were more valuable than those of crosses with fixed sterile lines. （3） Most predictions of crosses with fixed parents were relatively better. The predictions by IL of seed setting rate of crosses with fixed sterile lines,panicle number of those of fixed restorers,and those by PL of spikelets per panicle and grain weight per panicle of crosses with fixed restorers were above 0. 6. Because the number of PL and IL in prediction model were much less, the prediction models was conveniently applied and particularly, and it was important for the institute by which the models were founded in parent-selection in breeding.