Simulation of cyclic single cross selection.

Computer simulation was used to compare Hallauer's cyclic single cross selection (CSCS) with reciprocal recurrent selection (RRS). Three epistatic and three non-epistatic models with 60 loci determining a single character provided the genetic base. The rate of advance over seven cycles was always greater for CSCS than for RRS on either a cycle or generation basis. No genetic conditions were found where CSCS failed to respond. The advantages of CSCS increases as the proportion of non-additive genetic variance increases. Genetic advance of the hybrid population was shown to result from the joint effects of an average change in gene frequency and complementary effects (nicking) of selection. Nicking effects accounted for most of the advance for some starting conditions. RRS generally had higher selection limits with no epistasis or low gene frequency of the dominant allele. CSCS generally had higher limits with epistasis or high frequency of the dominant allele. We suggested that CSCS begin with divergent genetic populations and strong selection intensity for three cycles. Final selection of superior single crosses was indicated when the lines were completely inbred.