Application of multivariate statistical methods in detection of effective traits on bread wheat (Triticum aestivum L.) yield under moisture stress condition

To evaluate performance of multivariate statistical methods in detection of the most important effective traits on grain yield of 20 promising bread wheat lines and for determination of the role of each trait on yield changes, an experiment was conducted based on randomize complete block design with 3 replications in 2011-2012. Results of principal components analysis showed that three first principal components explained 72 % of the total variance. Accordion to the results, spike harvest index, ELWR and harvest index had highest effects on two first principal components. Cluster analysis by Ward’s method grouped the lines in 4 clusters. Discrimination function analysis confirmed the fourgroups derived from cluster analysis. Stepwise regression analysis indicated that harvest index, biomass and RWC were the most important effective traits on economic yield and explained 94% of its total variance. Results of path analysis showed that harvest index and biomass had maximum positive direct and negative indirect effect on economic yield, respectively, which can be used for selection of wheat varieties under terminal moisture stress condition.