Abstract

A maize population was subjected to three different selection procedures to determine the best method for its improvement in a study conducted at Ibadan, Nigeria. Considerable genetic variability existed within the population for effective selection for each of the traits considered, using any of the procedures namely full-sib, half-sib and S1 family selection. Heritability estimates were from moderate to large for most traits. Direct selection for grain yield is expected to result in favourable increases in other traits except that the selected materials will flower earlier. Since increases in plant height due to selection for grain yield is undesirable, it was concluded that the best selection method for yield improvement that will increase height, the least would be full-sib family selection.

Key words: Recurrent selection, predicted responses, maize.