Fields of efficiency of breeding methods for per se value or combining ability in plant breeding

Abstract

Starting from classical expressions of genetic advance for per se value or general combining ability, the fields of efficiency of 6 breeding methods (mass selection, half-sib family selection, full-sib family selection, half-sib progeny test, full-sib progeny test and S1 selection) were determined with the following assumptions: no epistasis, absence of genotype × environment interaction, same number of plants at the level of the test whatever the selection methods and same effective size in one cycle. Genetic advance was considered at 3 levels: per cycle, per generation and per unit of time with one off-season generation for intercrossing. To simplify the comparisons among methods involving families, a sufficiently large number of plants per family was considered, which permitted an examination of relationships between the heritabilities of the various testing systems. For the 5 breeding methods involving family or progeny testing, 2 types of 'fields' were determined, those that were a function of heritability and the correlation ρ{variant}S1, between additive effect for S1 value and additive effect for per se value (for a given part of the additive variance (a2) and a given ratio rS12 of additive variance in S1 to classical additive variance), and those that were a function of heritability and of the part of additive variance (for given values of ρ{variant}S12 and rS12). On a per cycle basis, among methods with progeny and family testing, S1 selection is the best in a large range of situations from medium to low heritabilities. Per unit of time, without and more clearly with one off-season generation for intercrossing, the field of efficiency of S1 selection restricted and full-sib family recurrent selection is the most efficient in a large range of realistic situations. From comparisons involving mass selection and family selection without an off-season generation, it appears that mass selection is the best method in several situations, not necessarily at high heritability. However, the advantage of mass selection at medium to low heritability disappear with the use of one off-season generation for intercrossing in family selection. © 1993.