Changes in Gene Frequency During Mass, Modified Ear‐To‐Row, and S 1 Selection: A Simulation Study 1

Computer simulation studies have indicated that S 1 recurrent selection is more efficient than mass or modified ear‐to‐row selection for improving corn ( Zea mays L.) populations (Choo and Kannenberg, 1979). The present study extends the comparison of mass, modified ear‐torow, and S 1 selection in terms of genetic drift and changes of gene frequency. Severe genetic drift occurred in S 1 selection under high (5%) selection intensity. The desirable genes lost during S 1 selection included both major (A = 2) and minor (A = 1) genes, with some losses occurring during the very early cycles of selection. Fewer and only minor genes were lost during mass selection. Gene losses under modified ear‐to‐row selection were intermediate. In all cases, genetic drift resulted is losses of alleles of 0.1 initial frequency, with losses being less severe under complete domtnance than under additive gene action. The loss of desirable genes depended almost entirely on their initial frequency and not on the magnitude of their individual effects. The results indicate that if 20 or fewer S 1 lines are recombined in each cycle, then another germplasm pool should be maintained as a source of extra genetic variability. S 1 selection brought the fastest change of gene frequency regardless of the gene effect and the initial frequency. When gene frequency exceeded 0.5, the rate of increase in gene frequency during S 1 selection was similar under the additive and the complete dominance models, but in mass or modified ear‐to‐row selection the rate of increase was much slower under the complete dominance model.