Resumen

MONTESINOS-LOPEZ, Osval Antonio et al. Best linear unbiased familiar predictor for partial diallel experiments with maternal effects. Agric. Téc. Méx [online]. 2009, vol.35, n.3, pp.245-256. ISSN 0568-2517.

In plant or animal breeding the experiments with diallel crosses (complete Griffing's designs) have been frequently used to estimate parameters and test hypothesis, which is really important for decision making in breeding programs. Nevertheless, the complete Griffing's designs are useful when the number of parental lines is reduced. With a high number of parental lines is difficult to prepare, stablish and conduct the trials in the field. In this situation a partial diallel experiment is recommended, which can be symetric or asymetric. An option is to use the partial diallel designs, which test a subset from the total number of crosses that is possible to develop among the parental lines. These trials can either be symetric or asymetric. In addition is frequent that researchers conform groups of families or parents that implies the consideration of estimating group or family effects. Nevertheless, so far no research has been conducted to estimate the best unbiased lineal predictor of group or family effect taking into account maternal effects in partial designs under a mixed model. For the above reason, in this research the best unbiased lineal predictor of familias for general combining ability in parcial diallel designs, including maternal effects, are derived, and a computer algorithm in SAS-IML commands that allows for the application of the proposed methodology.

Palabras llave : diallel crosses; general combining ability; mixed effects model; partial diallel experiments.