增效位点在稻米碾磨品质分子预测中的效应及其与遗传效应的关系

Effect of Effect-Increasing Loci in Predicting Milling Quality Traits of Indica Rice and It's Relation to the Genetic Effects

利用重庆和泸州两个环境下种植的两套杂交水稻不完全双列组合(按NCII设计),结合SSR和AFLP标记,按照单向分组的方差分析法筛选与F_2碾磨品质表现相关的阳性位点和增效位点,分别就此两类位点建立相应的预测模型,同时采用包括基因型×环境互作效应的种子遗传模型对这两套材料进行遗传效应分析,旨在分析水稻碾磨品质的分子预测效果及其与遗传效应之间的关系.结果表明:(1)增效预测模型较阳性预测模型稳定性好,精确度高.糙米率、精米率和整精米率的增效预测模型可决系数分别为0.6467、0.6516和0.7265,而阳性预测模型则为0.4053、0.4981和0.6897;增效预测模型的剩余平方和分别为0.8104、0.8011和4.4508,而阳性预测模型则为0.9826、0.9673和6.2676;增效预测模型的预测变异系数为6.79%、7.27%和5.02%,而阳性预测模型则为9.12%、8.13%和6.09%.(2)增效预测模型预测效果因材料和性状的不同存在差异,套内预测好于套间预测,固定不育系预测好于固定恢复系预测;预测效果以整精米率最好,精米率次之,糙米率稍差;(3)不同性状和材料的预测效果受环境互作的影响不同,糙米率受环境互作影响大于精米率大于整精米率.因此,可根据不育系和恢复系材料特性,在一定环境条件下建立碾磨品质性状的预测模型,或者选择遗传主效应表现良好,同时环境互作效应表现较为稳定的亲本建立预测模型,可能将会获得较为理想的预测效果.

In order to analyze the prediction effect hybrid performance of milling quality traits of indica rice by molecular markers especially effect-increasing loci （ILs）, and to analyze the relation between the prediction effect and the genetic effects of brown rice percentage （BRP）, milled rice percentage （MRP） arid head rice percentage （HRP）, polymorphism of 174 SSR and 31 AFLP markers for 32 parents were screened. By loci-screening performance of 3 F2 milling quality traits, using different parents of two half-diallel sets （NCII design） under two environments （Chongqing and Luzhou）, positive loci （PLs） and effect-increasing loci （ILs） were first obtained. Based on this, the mathematic models for prediction of hybrid performance by molecular markers were set up by the stepwise regression analysis. In the meanwhile, by using genetic model including genotype ~ environment （GE） interaction effects for quality traits of endosperms in cereal crops, the genetic effects of 3 milling quality traits in different environments were analyzed. Results indicated as follows： （1） the prediction effect of IL-prediction models （ILMs） was better than that of PLprediction models （PLMs）. ILMs are highly precise, strongly robust and have the certain practical value. R2 of ILMs of BRP, MRP and HRP were 0.6467, 0.6516 and 0.7265, respectively; That of PLMs were 0.4053, 0.4981 and 0.6897. The residual errors of ILMs of BRP, MRP and HRP, were 0.8104,0.8011 and 4.4508, respectively; That of PLMs were 0.9826,0.9673 and 6.2676.CV% ILMs of BRP, MRP and HRP, were 6.Trespectively; That of PLMs were 9.12%,8.13% and 6.09%. （2） The prediction effect of ILMs was different for different parents and traits. The prediction effect in same set of combination was better than that of between 2 sets of combination; The prediction effect of same sterile lines was better than that of same restorers lines; The prediction effect of HRP was better than that of MRP And BRP, respectively.（3） It was different that the prediction effects were influenced by GE interaction effects for different parents and milling quality traits. BRP was influenced by GE interaction effects was larger than MRP and HRP. Consequently, it may be more effective to set up prediction models with same sterile lines and same restorers lines under different environments, or to set up prediction models with the parental material less influenced by GE interaction effects.