摘要
本文提出了标准差距离分组法并以中国和联邦德国4个不同生态环境,5个不同年份的5套试验资料研究用标准差距离分组法进行小麦杂种优势预测。结果表明:G:组(D^2±1/2s)出现强优势组合最多,为43个,平均杂种优势最高,为60.4%;强优势组合出现的频率也最高,为57.3%。因而可以选择遗传距离(D^2)大小为(m±1/28)的亲本进行杂交,这样有较大的把握和机会获得强优势组合。为应用遗传距离预测小麦杂种优势提供了一个量化指标。
The possibility of application of a least-square method proposed by Pederson (1981) and a least-square method based on principal component analysis proposed by Zhang Aimin on wheat breeding was studied by using the data from 23 populations,including 8 F_2 populations,8 backcross populations and 7 three-way cross F_1 populations.The main results are as follows: The least-square method proposed by Pederson (1981) was applied to wheat breeding practice to predict better crosses,whose mean values in progeny popu- lations were as close as possible to those specified for the corresponding traits of an ideal genotype or the breeding goal.Based on the sum of square deviations, the predicted crosses was ranked.The rank correlation coefficient between pre- dicted rank of crosses and actual rank of crosses were R =0.6905 for single cros- ses,R =0.7850* for backcross and R =0.7143 for three-way crosses.When prin- cipal component values instead of origional character were used in predication, the rank correla tion coefficient was R=0.6190 for single crosses,R=0.8810** for back-crosses and R=0.8214* for three-way crosses.Using the principal component values in predication,it was named 'A least-square method based on principal component analysis-PLS' by the author.Both procedures can be used in plant breeding,butPLS is better than that proposed by Pederson,at least for backcrosses and three-way crosses.
基金
国家教委博士点基金资助课题
"七五"攻关课题的一部分
关键词
小麦
杂种优势
遗传距离
parents selection
least-square method
principal component analysis
wheat breeding
