News

Genomic selection: More allele frequency changes and fixation of favorable and unfavorable loci

article_published_on_label
September 14, 2023

In order to understand the long-term consequences of selection on complex traits, it is important to investigate the impact of selection on the changes in allele frequencies of causal loci and new mutations.

Scientists from Wageningen University & Research (WUR) and the French National Research Institute for Agriculture, Food, and Environment (INRAE) investigated those changes in allele frequencies for different selection methods using a simulation study. Moreover, the impact of non-additive effects (dominance and epistasis) on those changes was investigated.

Genetic selection has been applied for many generations in animal and plant populations as well as in experimental species (e.g. mice, fruit flies, and butterflies). This has resulted in a considerable improvement in desirable traits of those populations without depleting variation. In the past decade, pedigree selection – where animals or plants are selected on the basis of their ancestry – has been replaced more and more by genomic selection. Given the worldwide implementation of genomic selection, it is important to get a better understanding of the long-term consequences.

Results

“All in all, our results show that genomic selection results in slightly larger and faster changes in allele frequencies of causal loci than pedigree selection and much larger and faster changes than phenotypic selection,” says Yvonne Wientjes, first author of the paper. “The presence of non-additive effects limits the change in allele frequency, because non-additive effects can change the selection pressure and direction on a locus over generations. Loci with a larger statistical additive effect change on average more in allele frequency over 50 generations. However, the selection pressure on a locus is also largely affected by the linkage phase with other loci and by the accuracy by which the effect can be estimated.”

The scientists also found that genomic and pedigree selection fixated in general more loci after 50 generations, particularly more unfavorable alleles, than phenotypic selection. This increase was more related to hitchhiking with genomic selection compared to pedigree selection, showing the importance to minimize hitchhiking in genomic breeding programs to limit the loss of favorable alleles that are important for long-term genetic improvement.

Phenotypic selection was best in creating new genetic variance by maintaining mutations, followed by genomic and pedigree selection with OP (own performance) records, and finally genomic selection without OP records. “This shows that OP records are important to use in selection to optimally exploit new mutations that are essential for long-term genetic gain,” says Wientjes.

Relevance for practical breeding programs

Although the aim of the simulations was to compare the long-term impact of different selection methods relevant to livestock and plant breeding on allele frequency changes, it was necessary, when simulating the breeding program, to make several simplifying assumptions. “One assumption is that we focus only on a single trait, while in practical breeding programs, selection takes place on an index combining multiple traits,” says Wientjes. However, this does not mean that the research is not valuable for practical breeding programs. “The number of causal loci underlying the index is generally larger than in our simulations, and loci can have antagonistic pleiotropic effects on the different traits in the selection index. This can reduce the selection pressure on each locus, but not the ranking of the selection methods in terms of their impact on allele frequency changes. Moreover, the weights of the different traits in the index can change over time or new traits can be added. This can also change the selection pressure and direction on a locus over time, which is in a way comparable to our scenarios with epistasis. Given that the presence of epistasis only reduced the absolute differences between the selection methods, but not the ranking, we expect that a changing selection index will not result in reranking of the method.”

The publication ‘The long-term effects of genomic selection: 2. Changes in allele frequencies of causal loci and new mutations’ is part of the project ‘(R)evolution of traits? Quantifying the genetic change in traits over generations as a result of Genomic Selection’ (with project number 16774) of the research programme Veni which is (partly) financed by the Dutch Research Council (NWO).