论文部分内容阅读
用23个分离群体(8个 F:群体、8个回交群体、7个三交 F_1群体)研究最小二乘法在亲本选配中的应用和主成分在最小二乘法中应用的可能性。研究结果表明;用原始性状值进行组合预测,预测的单交组合、回交组合和三交组合的位次与实际位次有较高的正相关,秩次相关系数分别为0.6905,0.7857~*和0.7143。用主成分值代替原始性状值预测,回交组合和三交组合的预测效果提高,预测位次与实际位次的秩次相关系数分别为0.8810~(**),0.8214~*。主成分预测和原始性状预测的位次间也存在很高的正相关,对于回交和三交组合分别为:0.9048~(**),0.9286~(**)。用主成分代替原始性状进行预测,作者称为主成分最小二乘法,至少对于回交组合和三交组合,主成分最小二乘法较优。
The application of least square method in parental matching and the possibility of using principal components in least square method were studied with 23 segregating populations (8 F: population, 8 backcrossed population and 7 trio F_1 population). The results showed that the combination of the original trait values predicted that there was a high positive correlation between the ranking of the single crosses, the backswap and the triple crosses, and the rank correlation coefficients were 0.6905 and 0.7857 ~ And 0.7143. Prediction results of the principal component values instead of the original trait values were improved, and the prediction results of the backcross and the triple crosses were improved. The rank correlation coefficients of the predicted positions and the actual positions were 0.8810 ~ (**) and 0.8214 ~ *, respectively. There is also a high positive correlation between the principal component prediction and the prediction of the original traits. The combinations of backcrossing and triadimetry are 0.9048 ~ (**) and 0.9286 ~ (**), respectively. The main component of the original trait instead of the prediction, the author referred to as the principal component least-squares method, at least for the backcross and three cross combinations, the principal component least squares method is better.