Incorporating molecular markers into applied maize breeding

Two experiments were conducted to study applying molecular marker technologies to maize breeding. Both isozymes and restriction fragment length polymorphisms (RFLPs) were used as markers, and associations were established between these markers and different quantitative traits. Comparisons were made between marker-trait associations in diallel, testcross, and line per se data sets, and also between marker assisted selection (MAS) and conventional early generation testcross selection.In the first experiment, 33 markers were used to genotype 14 inbred lines. Using genotypic data, and data from diallel, testcross, and line per se yield trials, associations between genotype and grain yield, moisture, plant and ear height, root and stalk lodging, and number of days to flower were established. Marker-trait associations were found for all traits in all three data sets.Comparisons of significant marker-trait combinations were made between the diallel and testcross data sets. No similarities were found between performance of marker-trait-allele combinations in the diallel and testcross data sets. Comparisons were also made between the testcross and line per se data sets. More locus-trait-allele matches were found than expected by random chance. In 13 of the 14 marker-trait associations that were significant in both the testcross and line per se data sets, the same allele was 'favorable' for performance.In the second experiment, selections were made between and within F$\sb2$:S$\sb4$ families. Two types of selection were used: conventional early generation selection based on F$\sb2$ testcross performance, and MAS based on a selection index incorporating marker-yield associations determined in F$\sb2$ testcrosses. From among 220 F$\sb2$:S$\sb4$ families, 20 families were chosen by conventional selection, 20 families by MAS. Within each of the MAS selected families, individual plants were selected based on favorable and unfavorable genotypes.Selection among families using either MAS or conventional selection resulted in similar F$\sb2$:S$\sb4$ testcross performance. However, neither method selected families that performed significantly better than the original hybrid to improve, nor the S$\sb0$ unselected population testcross. Within family selection, using the index, was not effective. The selection index, which was developed with F$\sb2$ testcross data, was not correlated with F$\sb2$:S$\sb4$ testcross performance.U of I OnlyRestricted to the U of I community indefinitely during batch ingest of legacy ETD