Genomics of heterosis and egg production in White Leghorns

Crossbreeding is practiced extensively in commercial breeding programs of many plant and animal species, in order to exploit heterosis, breed complementarity, and to protect pure line genetic material. The success of commercial crossbreeding schemes depends on identifying and using the right combination of breeds, lines or varieties that produce the desired crossbred offspring. Currently, the selection of pure lines is based on the results of “field tests”, during which the performance of their crossbreds is assessed under typical commercial settings. Field tests are time-consuming, and also constitute a large percent of the costs of commercial crossbreeding programs. The research in this thesis therefore set out mainly to develop models for the accurate prediction of heterosis in White Leghorn crossbreds, using genomic information from their parental pure lines. Predicted heterosis could be used as pre-selection criteria, thus substantially reducing the number of crosses that need to be field-tested. In Chapter 1, I give an overview of the history of selective breeding in laying hens, and introduce heterosis and its genetic basis. In Chapter 2, based on a dominance model, we showed that a genome-wide squared difference in allele frequency between parental pure lines (SDAF) predicts heterosis in egg number (EN) and egg weight (EW) at the line level with an accuracy of ~0.5. With this accuracy, one can reduce the number of field tests by 50%, with only ~4 loss in realised heterosis. In laying hens, selection pressure is highest on the sires. We therefore went further to develop a model to predict heterosis at the individual sire level, in order to exploit the variation between sires from the same line. We found that the within-line variation between sires in our data was very small (0.7% of the variation in predicted heterosis), and most of the variation was explained by across-line differences (90%) (Chapter 3). Quantitative genetic theory shows that heterosis is proportional to SDAF and the dominance effect at a locus. In Chapter 4, we estimated variance components and dominance effects of single nucleotide polymorphisms (SNPs) on EN and EW in White Leghorn pure lines. We found that dominance variance accounted for up to 37% of the genetic variance in EN, and up to 4% of that in EW. We then used the estimated dominance effects to calculate dominance-weighted SDAFs for EN and EW between parental pure lines, and showed that prediction of heterosis based on a weighted SDAF would yield considerably different ranking of crosses for each trait, compared with a prediction based on the raw SDAF. This implies that different crosses would be selected depending on the criterion used to predict heterosis. To gain an insight into the genetic architecture of EN and EW, in Chapter 5 we performed genome-wide association studies using data on 16 commercial crossbred populations. We did not identify any significant SNPs for EN, indicating that EN is a highly polygenic trait with no large quantitative trait loci segregating in the populations studied. For EW, however, we identified several significant SNPs. One explanation for these results is that EN has been under intense directional selection for several decades, whereas EW has been under less-intense, stabilising selection. Finally, in the general discussion of this thesis (Chapter 6), I discuss the genomic prediction of heterosis, focusing on possible reasons for the lack of a consensus on the approach to predict heterosis, even after decades of research. I also discuss new opportunities for the genomic prediction of heterosis, considering the advancements in genotyping and computation methods. Lastly, I give an example of the application of results from this thesis in crossbreeding programs.