Plant breeders performed multi-trait selection using a suitable method, one of which is selection index. A multivariate analysis may be used for developing a selection index model. The objective of this experiment was to select doubled haploid rice lines with high yield potential and good agronomic characters based on the selection index. The experiment was carried out using a single factor completely randomized design with three replications, from August 2018 to December 2019. The study included 89 doubled haploid (DH) lines derived from anther culture as well as four check varieties (Ciherang, Inpari 30 Ciherang Sub1, Inpari 41, and Inpari 35 Agritan). Development of selection index model considered the results of analysis of variance, correlation analysis, path analysis, principal component analysis, and aimed plant ideotype. The findings revealed that there was variation in agronomic traits and yield among the DH lines tested. The selection index model can be formulated as I (index value) = 3*0.75 yield + 0.85 total grain per panicle – 0.85 plant height. Based on this model, 85 lines had index value and yield higher than Inpari 41. The selected DH lines were included in the subsequent field evaluation.
Keywords: correlation, heritability, path analysis, variability