Abstract: Based on the quantitative genetic model for endosperm traits, a new segregation analysis method for detecting major gene controlling endosperm traits was proposed, which includes major gene detection and its effect and variation estimation. The segregation ratios and genetic effect of major gene as well as model residual error caused by polygenes and environments were estimated by the maximum likelihood method implemented via EM algorithm. Major gene effects were tested with the likelihood ratio test (LRT) statistic. An example of the amylose content (AC) for P1, P2 and F2 endosperm generations in rice cross IR50×CP231 was used for the illustration. The results indicate that the genetic difference of AC in this cross refers to only one major gene. The major gene shows no first dominance effect. The additive effect and the second dominance effect of this major gene were estimated as 5.37 and 6.85, respectively. The genetic standard deviation caused by this major gene is about two times of the genetic standard deviation of the polygenes.

Key words: Endosperm trait, Major gene, Segregation analysis, Maximum likelihood estimation, EM algorithm