The expected efficiencies of half-sib, testcross and S1 progeny testing methods in single population improvement

Summary

Expressions are given for the expected rate of response to selection using half-sib, S1 or testcross progeny tests for biallelic loci with arbitrary dominance and epistatic properties. It is shown that the value of a tester is not necessarily expected to be a simple function of either its phenotype or of the number of recessive alleles it carries, and that neither downward selection nor inbreeding are expected to be consistently successful in isolating superior testers. Calculation of expected annual responses to selection under a variety of genetic models shows that, while mass selection is the most efficient method at high heritabilities, S1 testing is expected to be the best method for low heritabilities when replicated block trials are used. Conventional half-sib testing may have an advantage with complete randomisation of large families, especially for genetic models involving overdominance, deleterious recessives or epistasis in biennial or perennial crops. The use of the lowest homozygote tester was inferior to S1 testing for all models including simple directional dominance. Assumptions of low environmental homoeostasis of S1 material did not alter the general conclusions.

Monte Carlo simulations of breeding procedures confirmed the overall superiority of S1 testing in terms of response per cycle, and a system of mass selection followed by S1 testing gave a higher average response per year than did either mass selection or S1 testing alone.