DEVELOPMENT OF GRAIN SORGHUM HYBRIDS. IV. ESTIMATION OF GENETIC PARAMETERS THROUGH INCOMPLETE MATING DESIGNS

Abstract

Breeders frequently face the lack of enough seed of sorne crosses to estímate genetic parameters from complete factorial mating systems. It this study the changes of estimates of genetic variation were evaluated after eliminating sorne crosses from a set of a diallel crosses between 15 lines A and 14 lines R of sorghum (Sorghum bicolor L. Moench) simulating incomplete diallel crosses in the following four ways: 1) completely random, 2) sistematically, keeping the same number of crosses per parental line, 3) similar to 2, but before the elimination, parental lines were ordered according to their general combining ability, 4) similar to 2, but parental lines were ordered by earliness. In each elimination system, information was reduced in 10, 20, 30, 40 and 50%, thus, twenty incomplete diallels were analyzed. In each of these, eleven parameters of genetic variation were estimated (i.e., general and specific combining ability, broad and narrow sense heterosis, and additive and non additive genetic variances), which were compared with those obtained from the complete diallel system. It was concluded that: 1) Estimates of incomplete diallel systems did not change with respect to the complete one when 10% of crosses were eliminated, and 2) 30% of crosses could be lost and still reliable
estimates could be obtained if the objective is the evaluation of the GCA of the parental lines.