Optimization of the number of cycles and intensity of selection, and population size in mass selection: Selection for single traits in outcrossing plant populations

Optimum number of cycles and rate of selection and optimum population size to give the highest cost efficiency in mass selection for allogamous crop p lants were investigated. The optimum values of these variables were obtaine d based on the numerical calculations of a new selection efficiency index, S/C, where S stands for the probability that the desired genetic improvemen t is achieved in the target population, and C is the cost expended for that end, Selection procedures that maximize S/C will give the greatest opportu nity of success under a certain total resource investment. The index S/C wa s calculated based on Monte Carlo simulations with a model in which the tar get population was initiated as a hybrid population of two cultivar lines t hat were genetically fixed for the trait concerned, and populations of a co nstant size were generated for subsequent selection cycles via random cross ing between the plants selected with a constant rate in each cycle, The gen es involved were assumed to be inherited independently and have no epistati c interaction. The calculations over practically possible ranges of the rel ated variables led to the following conclusions, If, as would be the case i n most selection projects for breeding a new commercial variety, the expend iture of time (years) rather than the resource expense for managing the sel ection is important, a few to several cycles of selection with a rate of ar ound 1%, testing a few thousand plants per cycle, should be nearly if not e xactly the optimum, If the desired improvement is not achieved in these cyc les, the population should be discarded and a new one should be tested. By contrast, if the time expenditure is not important, as in the case of the s election for new breeding stock lines, selection with several or more cycle s, a rate in the range of 10 to 20% and a population size smaller than 100 should be cost-efficient, Dominance is not an important modifier of the opt imum values unless the desirable genes are dominant or recessive unidirecti onally at the majority of the loci involved.