Premium
Estimating Genetic Variance in Maize by Use of Single and Three‐way Crosses among Unselected Inbred Lines 1
Author(s) -
Wright J. A.,
Hallauer Arnel R.,
Penny L. H.,
Eberhart S. A.
Publication year - 1971
Publication title -
crop science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.76
H-Index - 147
eISSN - 1435-0653
pISSN - 0011-183X
DOI - 10.2135/cropsci1971.0011183x001100050026x
Subject(s) - diallel cross , biology , statistics , inbred strain , epistasis , restricted maximum likelihood , variance (accounting) , standard error , additive genetic effects , mathematics , genetic model , genetic variation , maximum likelihood , heritability , genetics , agronomy , hybrid , gene , accounting , business
Unweighted least squares and maximum likelihood procedures were used and compared for the estimation of genetic variance for eight quantitative traits in maize ( Zea mays L.). The genetic material was developed from unselected inbred lines isolated from a strain of Krug Yellow Dent maize. All possible single and 3‐way crosses were produced from 60 inbred lines, which traced back to 51 S 0 plants. The mean squares from the diallel and triallel analyses were used in estimating the genetic components of variance. Fitting the error and a six‐parameter genetic model showed that: 1) it was not possible to obtain realistic estimates of the epistatic components, although significant effects were detected in the analyses of variance; 2) the estimates of additive geuetic variance were significant for all traits for both estimation procedures; 3) the nonadditive components accounted for only a small proportion of the total genetic variance; 4) three iterations of the maximum likelihood procedure were sufficient to stabilize the estimates; and 5) the maximum likelihood procedure generally reduced the errors of the estimates. For the two‐parameter genetic model the largest proportion of the total genetic variance was additive for all traits. The estimates of deviations due to dominance were larger than twice their standard errors for all traits in the two‐parameter model for the combined single and three‐way cross data. The frequency of significant interactions with environments was higher for the additive than for the dominance effects.