Selection response with marker-based assortative mating

Assortative mating can increase additive variance (V(A)) and, consequently, selection response. Marker-based assortative mating (MAM) - the mating of individuals that have similar marker genotypes - has not been studied. My objectives in this simulation study were to (i) compare the selection response associated with MAM, phenotypic assortative mating (PAM), and random mating (RM) of selected individuals and (ii) determine when MAM will be most useful in breeding programs. I simulated 25 generations of selection among 200 individuals, followed by MAM, PAM, or RM, in a cross between two inbreds. A total of 100 codominant marker loci and 100 quantitative trait loci (QTL) were randomly distributed across 10 chromosomes, each 100 centimorgans (cM) long. The effects of QTL were additive and followed an exponential distribution. Cumulative selection response was determined for different initial levels of heritability (h² = 0.20, 0.50, and 0.80) and different numbers of individuals selected (N = 4, 8, 16, and 32) in each generation. Compared with RM, MAM resulted in slight increases of 3 to 8% in selection response during the first several generations, particularly when h² = 0.20 and N = 32. The PAM procedure led to slight increases of 2 to 4% when h² = 0.80 and N = 32. Any advantage of assortative mating over RM dissipated by Generation 8 for MAM and Generation 15 for PAM. The loss of QTL heterozygosity was much greater with MAM than with PAM or RM. For long-term improvement, selected individuals should be intermated by RM. The MAM procedure will be most useful for short-term improvement in a biparental cross, particularly when h² and selection pressure are low.