Evaluation of Alfalfa Synthetic Varieties: Prediction of Yield in Advanced Generations and Average Clone Effects1

Abstract

First, second, and third generations of nine synthetic varieties derived from combinations of two to nine clones selected from the alfalfa (Medicago sativa L.) variety ‘Cherokee’ were harvested in field plots for three seasons to evaluate the effectiveness of a prediction formula for yields and to estimate genetic factors affecting yield. The Syn 1 yields were predicted as the average of all possible Fls. Yields of F1s were previously determined in a nine clone diallel. The Syn 2 yields were predicted as a function of the Syn 1 predictions adjusted by subtracting the average S1 yields of the clones in the synthetic variety.Observed yields were compared to predicted yields using product moment correlation analyses. The predictions of Syn 1 yields were inaccurate while those of Syn 2 were very good. For Syn 2 the overall means of the predicted and observed yields were the same with a highly significant correlation (r = 0.86).A genetic model was developed to determine average effects of the individual clones in synthetic varieties, the linear generation effect, the quadratic generation effect, and the interaction of the linear generation effect and the clone effects on the yields of the synthetics. All effects except the quadratic generation effect and the interaction term were statistically significant. The average effects of the individual clones suggested that two of the nine clones were much inferior to the others and may have been misclassified in the initial selection of clones. The advantages of this modeling approach are outlined.