Abstract: The endosperm of cereal crops is a triploid tissue produced by a male gamete from the pollen grain fusing with two female nuclei. Endosperm traits refer to those characteristics that determined by the physical and chemical properties of grain endosperm. Because many quality traits are closely related to endosperms, the genetic study of endosperm traits plays a fundamental and vital role in improving grain quality. There have been several publications detailing the unique properties of endosperm traits. A series of QTL mapping models and approaches specific for triploid traits have been proposed, using single grain observation or more practical one that obtained from bulked samples. However, current methods based on bulked samples often fail to estimate two dominance effects of endosperm QTL separately, or involve a heavy work of crossing. On the basis of quantitative genetic models for triploid endosperm traits, a new mapping approach based on North Carolina Design III (NC III) and Triple Test Cross (TTC) was proposed in this study. Feasibility and efficacy of the method were investigated through simulated data. Experimental factors considered in the simulations include QTL heritability, number of plants in F₂ population, and number of endosperms collected per family. The results suggested that the two methods had satisfactorily high detection powers, which reached 100% even when the QTL heritability was 5%. But the TTC-based approach tended to have slightly higher power than the NC III-based one. Both approaches can distinguish two dominance effects successfully. But overall, the TTC-based approach can offer better estimation of either QTL position or effect, though this superiority waned as the heritability and sample size increase.

Key words: Endosperm trait, QTL mapping, North Carolina Design III, Triple Test Cross, Maximum likelihood estimation, EM algorithm