Genotypic Variances and Covariances of Six Quantitative Characters in Oats 1

VARIOUS authors—for example, Lush (10), Frankel (5), Burton (2), Panse and Bokil (12)—have emphasized the utility of estimates of genetic variance components as a basis for prediction of the response of quantitative characters to selection in breeding programs. Smith (16), Robinson et al. (14), and Comstock and Robinson (3) pointed out that when the criterion of selection gives weight to two or more characters, the magnitudes of genetic and environmental covariances also have a bearing on responses to selection that are to be expected. Smith (op. cit.), Hazel (8), and Lerner et al. (9) have considered the principles involved in construction of a selection criterion that gives optimum weight to each of several characters. They pointed out that estimates of genetic and environmental covariances among the characters as well as of the genetic and non-genetic variances of the separate characters are required. Comstock and Robinson (3) make a point of the fact that variance resulting from interaction of genotype and environment is frequently a source of upward bias in estimates of genetic variance. For example, estimates of genetic