Theory and application of half-sib matings in forage grass breeding

Abstract

Half-sib (HS) matings, including polycross, topcross, and open-pollination, are useful in the breeding of cross-pollinated sexual perennial forage grasses to evaluate general combining ability of parental clones for synthetic cultivar development, recombine selected entries in recurrent selection programs, and obtain quantitative genetic information. The objective of this paper is to review uses of HS matings in breeding of these forage grasses with emphasis on theoretical aspects related to quantitative genetic analysis.Polycross mating with adequate replications and sufficient isolation is recommended over topcross and open-pollinated mating schemes in generating HS families for quantitative genetic studies. For the estimates of many genetic parameters to be valid, the parents must be a random sample from a random mating population in linkage equilibrium. Precision of the estimates depends on adequante sampling of the population of genotypes and environments used for evaluation.Analyses of variance on HS families and parental clones, and analysis of covariance between parent and offspring provide useful information on additivity of genetic effects and on genotype × environment interactions. Classical, narrow-sense heritability on an individual plant basis can be estimated and used to predict genetic gain from individual (mass) selection, providing that within family variance is estimable. If the forage breeder uses family selection, heritability should be estimated according to the proposed unit of selection. The selection unit must be specified in terms of numbers of replications, years, and locations. Polycross HS family selection can be readily adapted to a population improvement program in forage grass breeding.Narrow-sense heritability can also be estimated by doubling the linear regression coefficient of HS prog eny means on parental means. When HS families and parents are evaluated together in replicated experiments under similar environments, covariance analysis is recommended to remove the genotype × environment interaction covariance and environmental error covariance between parent and offspring, since these nongenetic covariances may result in inflated heritability estimates and misleading expected genetic gains from selection.