| Reciprocal recurrent selection (RRS) was designed to improve the hybrids between two populations (A and B). Six cycles (C1 to C6) of full-sib RRS have been conducted using W577, a maize double-cross (DC) hybrid [(W64A x A295)(W374R x Oh43)]. The initial populations (C0) were the F2 derived from the parental single hybrids of the DC. Selection was based on a selection index calculated as grain yield (GY) divided by grain moisture at harvest (GM). The objectives of this study were to: (1) estimate response to RRS; (2) quantify genetic variance; (3) estimate inbreeding depression, and (4) quantify molecular genetic diversity in both parental populations. Objectives 1, 2, and 3 were investigated from phenotypic data, for eight agronomic traits, obtained from three experiments conducted at two Wisconsin locations. Objective 4 was investigated from SSR marker data obtained from sampled S 1 families from both parental populations at C0 and C5. The selection index increased linearly in W577 AxB (5.1% cycle-1) and in both parental populations (7.5% cycle%-1, on average). The target trait GY increased, but not linearly, in W577 AxB (2.8 g plant -1 cycle-1) and in both parental populations (2.3 g plant-1 cycle-1, on average), whereas GM decreased linearly in W577 AxB (-0.7% H2O cycle -1) and in the parental populations (-1.1% H2O cycle-1, on average). Differential response for GY and GM may have resulted from their differences in heritability. Phenotype-based estimation of genetic variance indicated no significant changes for GY from C0 to C5, and a significant decrease for GM in both parental populations. Inbreeding depression for GY was significant and similar at C0 and C5. No significant inbreeding depression or heterosis was observed for GM at either C0 or C5. Molecular analyses indicated a significant reduction in variation from C0 to C5 in both parental populations, consistent with the expected loss of variation under the genetic drift model. Functional alleles influencing phenotypic-based variation, and neutral alleles influencing molecular-based variation may have been responsible for difference between both approaches. This research provides evidences that populations derived from a double-cross hybrid may be suitable for RRS. |
