Many economically important quantita- tive traits in animals and plants are measured re- peatedly over time. These traits are called dynamic traits. Mapping QTL controlling the phenotypic pro- files of dynamic traits has become an interesting topic for animal and plant breeders. However, statistical methods of QTL mapping for dynamic traits have not been well developed. We develop a composite in- terval mapping approach to detecting QTL for dy- namic traits. We fit the profile of each QTL effect with Legendre polynomials. Parameter estimation and statistical test are performed on the regression coefficients of the polynomials under the maximum likelihood framework. Maximum likelihood estimates of QTL parameters are obtained via the EM algorithm. Results of simulation study showed that composite interval mapping can improve both the statistcial power of QTL detecting and the accuracy of parameter estimation relative to the simply interval mapping procedure where only one QTL is fit to each model. The method is developed in the context of an F2 mapping population, but extension to other types of mapping populations is straightforward.
