Accelerated recurrent selection

Abstract

In Accelerated Recurrent Selection (ARS) schemes, selection is based on the predicted performance of progeny families, estimated as the average of the parental families. These schemes can halve cycle time compared to simple recurrent selection methods. They also have a lower requirement for evaluation of families in yield trials, although they do require more seed production. ARS therefore provides options in cycle time, effective population size, response to selection and cost which have not been available before. Example schemes are compared by computer simulation with truncation selection and with optimal family selection, where contributions of families to the next generation are optimised to give the maximum response to selection at a specified effective population size. In many circumstances, ARS schemes compare favourably. Difficulties in combining estimates of selection intensity and of effective population size when comparing the merits of different breeding schemes are discussed. It is suggested that unless one is interested in response to selection over periods greater than 50 years, the weight given to effective population size in ranking different schemes should be small.