engleski

Long Term Selection Effects on Inbreeding under Finite Locus Models

The true coefficients of inbreeding in biological populations are unattainable, hence estimation of inbreeding from pedigree data is a frequently used method. Inbreeding can also be measured from genotype data when pedigrees are not available. Both methods are appropriate for traits that are not under selection. A simulation with a two-locus bi-allelic model was performed to study how well inbreeding estimates from pedigree and genotypic frequencies correspond to true values when loci controlling the trait are under selection. Different models were used, depending on the number of loci controlling the trait (two loci and 50 pairs of loci), the mating strategy (random mating or truncated phenotypic selection), the initial gene frequency of the favorable allele (0.2, 0.5 and 0.7), and linkage within a pair of loci (recombination rate 0.5 or 0.02). Selection was applied under 11 genetic models that differed in the presence and intensity of additive, dominance, and epistatic (additive by additive, additive by dominance, and dominance by dominance) effects. Phenotypic values were generated with broad sense heritability equal to 0.30 in the first generation. Populations of 40 male and 80 female animals were simulated for up to 300 cycles of selection. Generations were overlapping and the average generation interval was around three cycles. When random mating was practiced as a selection strategy, both inbreeding measures corresponded closely to the true values. When selection was applied, neither estimates of the inbreeding coefficient from pedigree nor from genotypic frequencies corresponded to true values. Bias was present in both directions, thus leading to overestimation or underestimation depending on the genetic model and the initial gene frequency. Although linkage did not influence the sign of the observed bias, it enforced the effects of selection, increasing the magnitude of the bias. The same patterns of bias were observed in models with two loci and with 50 pairs of loci, except that they were stronger and sharper for models with two loci. Results presented show that caution is necessary whenever conclusions (e.g. on inbreeding depression) are based on inbreeding coefficients estimated from pedigree or genotypic frequencies. General guidelines on when and what kind of caution has to be taken when using such estimates are also provided.

Inbreeding; Two-locus model; Epistasis; Selection; Linkage

nije evidentirano