INBREEDING DEPRESSION RATES IN MAIZE POPULATIONS BEFORE AND AFTER RECURRENT SELECTION

Limited empirical data are available to determine how inbreeding depre ssion rates for quantitative traits have changed with long-term germpl asm enhancement methods. Experiments were conducted to estimate inbree ding depression rate of 16 quantitative traits in unselected maize (Ze a mays L.) populations and after recurrent selection. The populations per se (S-0) and seven bulked inbred subpopulations (S-1 to S-7), deve loped by single-seed descent in BSSSCO, BSCB1CO, BS26, BSSS(R)C9, BS13 (S)C3, and BSCB1(R)C9, were evaluated in randomized complete-block exp eriments conducted at five locations for 2 years. The objectives of th is study were to determine the inbreeding depression rates for 16 plan t and ear traits, to determine the relative merit of half-sib followed by S-2 progeny selection in BSSS and reciprocal recurrent half-sib se lection in BSSS and BSCB1 for changing inbreeding depression rates, an d to determine if the mean performance of a trait and the level of inb reeding in BSSS, BSCB1, and their advanced cycle populations was expla ined by the linear (additive) model. The rate of inbreeding depression decreased for all traits in BS13(S)C3, BSSS(R)C9, and BSCB1(R)C9 exce pt for yield (g plant(-1)), 300-kernel weight, and days to anthesis in BS13(S)C3; yield (q ha(-1)) in BSSS(R)C9; and yield (g plant(-1)), nu mber of ears plot(-1), 300-kernel weight, and ear-leaf width in BSCB1( R)C9. The rates of inbreeding depression for grain yield in BS13(S)C3 and BSSS(R)C9 were not different, but rate of inbreeding depression fo r BS13(S)C3 was significantly (P <.05) less than for BSSSCO. The rate of inbreeding depression for grain yield in BSCB1(R)C-9 was significan tly (P <.01) less than in BSCB1CO. Rate of inbreeding depression was r educed for 13 of 16 traits in selected populations of BSSS. The reduce d rates of inbreeding depression for most traits in the improved popul ations of BSSS and BSCB1 suggest either that these populations were se gregating at fewer loci or that selection has increased the allele fre quencies beyond 0.5. The linear model explained more than 90% of the v ariation among generations for most traits, suggesting that epistasis was not important.