CHANGES IN THE GENETIC PROPERTIES OF CMS-39 MAIZE POPULATION SUBMITTED TO MASS SELECTION FOR PROLIFICACY

Recurrent selection aims to increase the frequency of favorable alleles. However,genetic variability must be maintained for sustained progress from selection and there are manyoptions of breeding methods to reach this objective. Mass selection for prolificacy can be useddue to the high positive genetic correlation between number of ears per plant (prolificacy, withhigh heritability) and grain yield. In the evaluation of six cycles for prolificacy in the CMS-39maize population, differing from some literature results that did not attain gain for grain yield, itwas obtained genetic progress. One of the possible causes for failing to increase yield mayhave been the lack of enough genetic variability in the traits submitted to selection. The presentstudy was carried out to test the formulated hypothesis and to check for changes in thecomponents of genetic variance of the original (CO) compared to the fifth (CV) selection cyclein CMS-39 population. Two hundred S1 progenies and corresponding half and full sib progeniesfrom C0 and CV were assessed at C0 and CV, totaling 400 progenies in each selection cycle.These assessments were carried out in two locations in southern Minas Gerais, Lavras andIjaci, using a triple 10 x 10 lattice design. The traits evaluated were husked ear yield andprolificacy. The estimates of the genetic variance components and heritability suggested thepresence of variability in the CMS-39 population indicating that the previous unsuccessfulselection could not be attributed to lack of genetic variability. However, the heritability estimates(h2) for the prolificacy trait were not higher than that of husked ear yield, showing that indirectselection would not be a good alternative in this case. No reduction in the estimates of geneticvariance components was shown in the CV compared with the CO population, showing thatthe selection method used did not exploit the genetic variance available. The estimates of theD1 component, which is the covariance between the mean effects of the alleles and the dominanceeffects of the homozygotes, were negative in all situations, indicating that the frequency of thefavorable alleles for the traits studied in the CMS-39 population must be low. The additivecomponent explained, on average, more than 75% of the total genetic variance indicatingthe predominance of the additive effects in the control of all the traits considered.