Genetic diversity for restriction fragment length polymorphisms and heterosis for two diallel sets of maize inbreds.

Changes that may have occurred over the past 50 years of hybrid breeding in maize (Zea maize L.) with respect to heterosis for yield and heterozygosity at the molecular level are of interest to both maize breeders and quantitative geneticists. The objectives of this study were twofold: The first, to compare two diallels produced from six older maize inbreds released in the 1950's and earlier and six newer inbreds released during the 1970's with respect to (a) genetic variation for restriction fragment length polymorphisms (RFLPs) and (b) the size of heterosis and epistatic effects, and the second, to evaluate the usefulness of RFLP-based genetic distance measures in predicting heterosis and performance of single-cross hybrids. Five generations (parents, F1; F2, and backcrosses) from the 15 crosses in each diallel were evaluated for grain yield and yield components in four Iowa environments. Genetic effects were estimated from generation means by ordinary diallel analyses and by the Eberhart-Gardner model. Newer lines showed significantly greater yield for inbred generations than did older lines but smaller heterosis estimates. In most cases, estimates of additive x additive epistatic effects for yield and yield components were significantly positive for both groups of lines. RFLP analyses of inbred lines included two restriction enzymes and 82 genomic DNA clones distributed over the maize genome. Eighty-one clones revealed polymorphisms with at least one enzyme. In each set, about three different RFLP variants were typically found per RFLP locus. Genetic distances between inbred lines were estimated from RFLP data as Rogers' distance (RD), which was subdivided into general (GRD) and specific (SRD) Rogers' distances within each diallel. The mean and range of RDs were similar for the older and newer lines, suggesting that the level of heterozygosity at the molecular level had not changed. GRD explained about 50% of the variation among RD values in both sets. Cluster analyses, based on modified Rogers' distances, revealed associations among lines that were generally consistent with expectations based on known pedigree and on previous research. Correlations of RD and SRD with f1 performance, specific combining ability, and heterosis for yield and yield components, were generally positive, but too small to be of predictive value. In agreement with previous studies, our results suggest that RFLPs can be used to investigate relationships among maize inbreds, but that they are of limited usefulness for predicting the heterotic performance of single crosses between unrelated lines.