Further Evidence on Testcross Versus Line Performance in Maize1

Abstract

A total of 169 S1 lines out of KrugIII synthetic were evaluated as (1) lines per se, (2) testcrosses to an unrelated tester, and (3) testcrosses to the parental population. The three highest‐ and three lowest‐yielding from each of the three evaluation series were used to study their intercross behaviour. The 18 lines were intercrossed. Composites of HH, HL, and LL crosses within each set of 6 lines and crosses between sets (H1H1, H1L1, L1H1 and L1L1) provided the material for test evaluation in 1965‐66. In each evluation series the superior lines [>(p̄ + σp)] had been selected to form new cycle populations. These populations (Cycle IV) together with the parental, KIII synthetic, were included in the performance trials.The LL, HL, and HH groups from line per se selections exhibited a linear trend for yield. Selection on the basis of an unrelated tester resulted in the HL group ≥ the HH group. Intercrosses of lines selected on the basis of the parental population as the tester exhibited a linear yield trend but with evidence of partial dominance for high yield. Highest yields resulted from intercrosses of lines selected on the basis of performance in testcrosses with the parental population.With respect to the newly derived populations, that based upon selection from testcrosses with the parental population (KIV‐3) exhibited a 15% increase in yield relative to KIII synthetic. Line per se selection resulted in a 4% gain in yield, whereas no gain resulted from selection based upon evaluation in testcrosses with the unrelated tester parent.