GENETIC VARIANCES IN AN EIGHT PARENT HALF DIALLEL OF OATS
Twenty-eight progenies with their eight parent cultivars of Avena saliva L. (2n = 6x = 42) were grown in F1, F2 and F3 in separate years; the F1 as spaced plants, the F2 and F3 as dense seeded populations. Additive genetic variance constituted most of the phenotypic variance of eight traits (heading date, plant height, stem diameter, grain yield and four components of yield) according to a Griffing Method 4, Model II analysis. Similarly, additive × year interactions were more important than nonadditive × year interactions. A Hayman-Jinks analysis of the same material but with the parents included showed that the additive component was 2 to 16 times larger than the dominance components in the F1 However in the F2 and F3 the dominance components became larger than the additive components for most traits instead of declining in importance as expected. Further, tests of fit to the hypotheses underlying the Hayman-Jinks analysis were negative in 8 of 24 cases. It is postulated that these discrepancies result from epistatic variance which caused an upward bias in the dominance estimates. The calculation and uses of two estimates of narrow-sense heritability are discussed.