Use of Selection Indices Based on Multivariate Analysis for Improving Grain Yield in Rice

In order to study selection indices for improving rice grain yield, a cross was made between an Iranian traditional rice (Oryza sativa L.) variety, Tarommahalli and an improved indica rice variety, Khazar in 2006. The traits of the parents (30 plants), F1 (30 plants) and F2 generations (492 individuals) were evaluated at the Rice Research Institute of Iran (RRII) during 2007. Heritabilities of the number of panicles per plant, plant height, days to heading and panicle exsertion were greater than that of grain yield. The selection indices were developed using the results of multivariate analysis. To evaluate selection strategies to maximize grain yield, 14 selection indices were calculated based on two methods (optimum and base) and combinations of 12 traits with various economic weights. Results of selection indices showed that selection for grain weight, number of panicles per plant and panicle length by using their phenotypic and/or genotypic direct effects (path coefficient) as economic weights should serve as an effective selection criterion for using either the optimum or base index.