Recurrent selection (RS) schemes were introduced to increase the frequency of favorable alleles for quantitatively inherited traits. The main goal of RS was to genetically improve germplasm resources for breeding programs. Data were summarized for 14 intra-population and eight inter-population maize (Zea mays L) RS programs conducted in 17 genetically broad-based populations. The intra-population programs included evaluation of half-sib and full-sib families, either S-1 or S-2 inbred progenies, and a combination of S-1 and S-2 inbred progenies. The inter-population reciprocal RS programs were restricted to either half-sib or full-sib family selection. Grain yield was the primary trait considered in selection, but selection indices that include grain moisture at harvest and resistance to root and stalk lodging also were considered in making the selections that were intermated to form the next cycle population. Approximately, 10 to 20 selections were intermated for each cycle. Estimates of the genetic variation among progenies tested (sigma(2)(g)), interactions of progenies with environments (sigma(2)(ge)), and experimental error (sigma(2)) were obtained from the combined analyses of variance for each cycle of selection and then averaged across cycles for each selection program. From the estimates of the components of variance, estimates of heritability (h(2)) on a progeny mean basis, the genetic coefficient of variation among progenies tested relative to their mean, (GCV), selection differentials (D), predicted genetic gains (Delta(G)). and least significant differences (LSD) were calculated and averaged across cycles of selection. The average estimates of sigma(2)(g) were largest for inbred progeny selection and smallest for half-sib family selection as expected. Averaged expected genetic gain across all intra-population selection programs was 3.17 q ha(-1) yr(-1) vs 2.32 q ha(-1) yr(-1) for inter-population reciprocal recurrent selection, or 2.64 q ha(-1) yr(-1) across all methods. On a per cycle and per year basis the differences among types of progeny were relatively small, ranging from (3.57 q ha(-1) yr(-1),1.78 q ha(-1) cycle(-1)) for half-sib family selection to (9.62 q ha(-1) yr(-1), 3.21 q ha(-1) yr(-1)) for S-2 inbred progeny selection. Regression analyses of the square roots of with cycles of selection suggested that genetic variation was not reduced significantly with selection. Even though AS was used to determine the primary types of genetic effects that respond to selection and contribute the expression of heterosis and could ideally support basic association and genome selection studies, the principle goal of AS is to adapt and improve genetically broad-based germplasm sources for potential use in breeding programs. Few programs have integrated RS programs with development programs to isolate unique inbred lines that have potential either as parents of hybrids or use in elite line crosses to develop recycled lines. In the past 10 years North Dakota has released 18 (out of 28) derived from RS programs, six from the NDSU EarlyGEM program (also diverse), and four from elite x elite combinations. Based on the number of progenies evaluated (25,692) in the RS programs presented and the number of inbred lines (31) that met standards for a 'B' designation and release to other maize breeders, the frequency of released lines was 0.12% or 1.2 lines per 1000 tested. However, the value of each line is different. B73 is a successful example of integrating recurrent and pedigree selection programs in order to develop outstanding cultivars. These significant gains can be realized with long-term RS selection programs. National support for them is encouraged in order to develop the next generation of maize products.
