Abstract
A simple genetic algorithm (GA) was used to find a mating design which would produce a higher genetic gain with a lower inbreeding level. Stochastic simulation was conducted to compare the genetic and inbreeding levels between GA mating and random mating. Selection for both mating schemes was based on the predicted breeding values of best linear unbiased prediction (BLUP). When sires had similar relationships with dams, GA failed to find mating design that satisfied the constraints because GA searched for the mating design with a lower inbreeding level than random mating among selected animals. The percentage of successful replicates ranged from 70.5% to 100.0% for a combination of 3 heritabilities and 4 penalty coefficients. The GA reduced 7.8% to 11.4% of inbreeding level (P<0.001) compared to random mating after 9 generations of selection and mating when both males and females had phenotypes. When the trait was measured only on females, GA reduced 8.1% to 14.8% of inbreeding level (P<0.001). Although the increase in genetic level was small, the ability of GA to reduce inbreeding and maintain genetic level was demonstrated.
