AbstractFive breedings methods have been studied for their efficiency in improving the line value of a population, i.e. the value of all lines that can be derived from the population. These methods are: half‐sib family recurrent selection (HSF‐RS), full‐sib family recurrent selection (FSF‐RS), half‐sib progeny recurrent selection (HSP‐RS), recurrent selection (S1‐RS) and the single‐doubled‐haploid descent recurrent selection (SDH‐RS). All breeding methods are compared on the basis of the genetic advance per unit of time, with the constraint of the same effective size after the first cycle of selection. Two modalities of SDH‐RS are considered for an annual plant without an off‐season generation: one with a cycle in four years (SDH1‐RS) and the other with a cycle in five years (SDH2‐RS). For all methods the possibility of one off‐season generation for intercrossing is considered. Fields of efficiency of the best methods are determined considering five main parameters, the heritability, the importance of additive variance, the correlation QL between the additive effect for line value and the additive effect for theper sevalue, the length of the cycle for SDH‐RS and the selection intensities. SDH‐RS is the best method when the heritability is low, with a low proportion of additive variance and with a relatively low correlation QL between additive effects forper seand line values. Without off‐season generation, S1‐RS is the best method among the remaining realistic situations; with one off‐season generation for intercrossing it is FSF‐RS. The use of an off‐season generation reduces the domain of efficiency of SDH‐RS while an increase in the selection intensity increases this domain. With a selection intensity for SDH‐RS equal to that for S1‐RS, SDH‐RS is the best in a large range of realistic situations except for SDH2with one off
展开▼
