The Genetic Variance-Covariance Structure of the Louisiana Sugarcane Breeding Population.

Abstract

Genetic variances estimates were obtained for ten traits of the Louisiana sugarcane breeding population. The variances and resulting statistics were derived from a population of three offspring each from 40 biparental crosses of 28 elite parents. The offspring and their parents were grown in replicated tests in five environments (year-location combinations). Narrow and broad-sense heritabilities, genetic correlations, genetic coefficients of variation and genetic gain estimates were estimated. To represent clonal populations in different stages of selection, the population was divided into three subpopulations: offspring, noncommercial parents and commercial parents. For each subpopulation, the genotype x environment (GE) variation was partitioned into genotype x year, genotype x location and genotype x year x location components. Genotypic and phenotypic path analyses were also performed for each population for sucrose yield, cane yield, sucrose concentration and stalk weight. Nonadditive genetic variance predominately determined genotype. The results suggested the most effective cross performance prediction would be derived from the progeny performance of specific biparental crosses. Genetic variances diminished with selection. GE variance and its relative importance to genetic variance estimates fluctuated with the trait and population but at times was substantial. It was suggested selection for increased genotypic stability, in the sense of reduced GE variances, would be not effective without better delineation of the specific environments interacting with specific genotypes. Broad-sense and phenotypic correlations were generally concordant and positive. In several circumstances additive genetic correlations were stronger and occasionally of different sign than the broad-sense genetic and phenotypic correlations. It was hypothesized that the disparity between correlation estimates likely resulted from use of nonrandomly mated elite parents. Although estimation assumptions were not classically upheld, the relationships were between relatives of interest and of populations specific to the sugarcane breeding program. Cane yield was the major determining influence on sucrose yield and was not affected by population. The relationship among stalk weight, stalk diameter, stalk number and stalk density strengthened with selection and in one case turned negative in sign.