Theoretical Comparison of Recurrent Selection Methods for the Improvement of Self‐Pollinated Crops

Abstract

Choosing a recurrent selection strategy depends on the resources of the breeder and on the efficiency of the possible selection schemes. An experimental comparison is difficult and time consuming, but a theoretical comparison may provide some guidelines. This study was conducted to compare theoretically the efficiency of different recurrent selection methods of improving the line value of a population, in this case, of self‐pollinated crops. We considered five methods that are well adapted to breeding autogamous species: single‐doubled‐haploid descent selection with a 4‐yr (SDH1) or 5‐yr cycle (SDH2); selection on S1 family (S1; selection on S2 bulk (S2B) and on S2 family (S2F); and two other methods‐selection on half‐sib progeny (HSP) and full‐sib family (FSF)‐ used for the improvement of per se or combining ability values. Analytical expressions for the corresponding expected genetic advances (per growing season) were derived and compared according to the values of the narrow sense heritability and the variance components in an inbred system. SDH1 always appeared as the most efficient method. SDH2 and S2F were the next best methods for realistic values of parameters. S1 efficiency was restricted to very high heritabilities and additive variances. Gain due to higher gene fixation in the S2 families balanced the short cycle of S1. As the proportion of non‐additive and environmental effects increased, SDH2 became more efficient than S2F, and S2F more efficient than S1. S2B was always the least efficient except when the possibility to reduce the between‐plot environmental variance through a larger plot size was taken into account. In that case, S2B could be more efficient than S2F for medium to low heritabilities. HSP and FSF appeared only marginally as the second or third most efficient methods. SDH is the best method to improve autogamous crops even though cycle length is 5 yr; however, if haplo‐diploidization technologies are not available or are too expensive, a selection scheme based on S2 family may be the most economical method.