Response to Four Cycles of Combined Half‐Sib and S1Family Selection in Maize

Effective recurrent selection for intrapopulation improvement in maize (Zea maysL.) should be most effective when both inbred and noninbred progenies are evaluated. The objective of this study was to examine the effectiveness of evaluating both half‐sib (HS) and selfed (S1) families during selection to improve population per performance and inbred productivity. Four cycles of combined HS and S1selection for increased grain yield and reduced grain moisture were completed in a narrow‐based maize population derived from inbred lines; A635, W182E, and W64A. For each cycle, both HS and S1progenies were produced on prolific plants by selfing the bottom ear and exposing the upper ear to random pollen from the population. Half‐sib and S1progenies were then evaluated in separate trials. Plants with superior HS and S1progeny performance, i.e., above average yield and below average grain moisture in both trials, were selected. Half‐sib progeny were used for recombination through the first two cycles. In the third cycle, S1progeny were recombined. Both progeny types were recombined to form two subpopulations in the fourth cycle. Over four selection cycles, yield of the population increased by 3.5% cycle−1, and grain moisture decreased by 1.5% cycle−1. Corresponding increases in yield were not detected in testcrosses with inbred lines A554 and Wl17, but cycle four testcrosses had significantly lower grain moisture than corresponding testcrosses with the initial population. The yield of bulked composites from random, selfpollinated plants increased by 20 g plant‐1 over four cycles. Broad‐sense heritability estimates were 0.34 and 0.76 for HS and S1families, respectively, for yield, and 0.68 and 0.88 for moisture. The genetic covariance of HS and Sa progenies was compared to the variance among HS progenies for grain yield and moisture in order to estimate D1, the covariance of additive and homozygous dominance effects. A large negative value for D̂1for yield, −309±126 g2plant−2, indicated that simultaneous improvement of both inbred and testcross performance may be difficult. However, the magnitude of D̂1may have been affected by linkage disequilibrium in the initial population.