Use of QTL identified in breed crosses for genetic improvement in pigs

Strategies for the exploitation of multiple QTL identified in breed crosses for genetic improvement of livestock were examined through marker-assisted introgression (MAI) and marker-assisted selection (MAS). Evaluation of the effectiveness of alternate strategies for MAI and MAS were carried out using stochastic computer simulation. In general, introgression of multiple QTL and maintaining the frequency of the donor's allele at 50% requires large population sizes that are not feasible in livestock. An alternate selection approach for MAI of multiple QTL was developed for a breeding program of limited size, without requiring the selected candidates to be heterozygous at all loci during the backcrossing phase. This MAI strategy can enrich a breed with favorable alleles at multiple QTL from a donor breed at the end of intercrossing phase, even for 20 cM marker intervals around the QTL. An alternate strategy to use QTL from breed crosses is to apply MAS within the cross for development of synthetic lines. A model that is suitable for genetic evaluation was developed and evaluated based on a cross of inbred lines with previously detected QTL regions. Based on genetic gains and ease of implementation, the preferred model for MA-evaluation included fixed marker effects and polygenic effects (BM). It was shown that even marker intervals of 20 cM resulted in a superiority of MAS over BLUP selection using phenotypes only. A two-stage strategy for MA-synthetic line improvement involves QTL detection in the first and MAS on the identified QTL in the second stage. Backward elimination regression was used for QTL detection. MAS using the model BM resulted in significantly higher responses compared to selection on BLUP from phenotype, although MAS was subject to false positives and inaccurate marker estimates. Genomic selection in a cross between inbred lines on all marker intervals across the genome regardless of significance was evaluated as an alternative to 2-stage MAS on significant QTL regions only. The results showed that genomic selection is an approach of choice for genetic improvement in livestock, in particular when marker effects are treated as random (ridge regression) rather than fixed. Even though a constant prior variance was used for marker effects in random MAS, the regression of marker estimates towards zero reduced sampling errors of estimates, especially when true QTL effects were zero. The difference between genomic MAS and two-stage MAS depends on the significance threshold used in the QTL detection process. The MAS strategy should always include polygenic effects in models for genetic evaluation when using the two-stage strategy of MAS in crosses of inbred lines. With genomic MAS, polygenic effects had a limited impact on responses to selection. Benefits of MAS over BLUP were higher with low heritability. In conclusion, QTL identified QTL from sparse marker maps obtained from breed crosses of inbred lines can be utilized efficiently through MAI and MAS.