Genetic and genomic evaluation for milk production and fatty acid traits predicted by mid-infrared (MIR) spectroscopy in Holstein dairy cattle

• Ano de defesa: 2021
• Autor(a) principal: Paiva, José Teodoro de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Viçosa
• Palavras-chave em Português: Bovinos de Leite; Melhoramento Genético; Parâmetros Genéticos; Leite - Qualidade; Análise de Regressão; Confiabilidade; Genética e Melhoramento dos Animais Domésticos
• Link de acesso: https://locus.ufv.br//handle/123456789/28126

Mid-infrared (MIR) spectroscopy is the current main tool that has been used to get access of rapid, cost-effective, and high-throughput data of milk composition. Over the last decade, milk fatty acids (FA) have been predicted by MIR, allowing to record milk quality data at the population level. Interest in milk FA profile has increasing given its nutritional value, technological properties, and its use as biomarker of the cow’s status. The availability of these phenotypes makes possible their inclusion in genomic evaluations, which brings unprecedent and substantial impacts to improve milk quality. Therefore, the general objective of this thesis was to perform genetic and genomic evaluations for milk production and FA traits predicted by MIR using random regression models (RRM) in dairy cattle. Firstly, it was investigated different Legendre polynomials orders to better modeling of random effects in first lactation cows. Third-order Legendre polynomials seem to be most parsimonious and sufficient to describe milk production and FA traits over days in milk. Lower Spearman correlations at the beginning of lactation suggest the optimal model appeared to be even more important in the case of selection in this period. In addition, optimal polynomial orders tend to show lower residual variation, which can help to avoid overestimation at beginning of lactation. The effects of permanent environment and herd-year of calving had a high impact on early lactation. Heritability curves indicated that as long as lactation progressed the proportion of genetic variance increased for all traits. In a second step, it was investigated the potential implications of selection for milk production traits about FA across the first lactation through bi-trait RRMs using pedigree and genomic information. Selection for higher milk yield would decrease FA. Improving the milk FA profile would seem to be an effective way by indirect selection of fat yield, fat and protein content. Subsequently, genomic predictions using the single-step genomic best linear unbiased prediction (ssGBLUP) approach were performed based on RRM. It was investigated the performance of genomic predictions (in terms of reliability and bias) using ssGBLUP approach and it was compared with the parent average (PA) method. Moreover, different scaling and weighting factors to be used in the construction of the H matrix were tested. The test-day ssGBLUP approach yielded higher validation reliability compared to PA method for young bulls, even when no scaling and weighting factors were used in the H matrix.In addition, choosing optimal parameters led to less biased prediction (regression coefficient close to 1) for genomic evaluation of milk production traits. Nonetheless, inflated GEBVs were still observed for milk FA. The findings reported in this thesis will contribute to advance on the modeling of milk production and milk FA traits in Walloon Holstein dairy cattle by inclusion of genomic information. Results from this research suggest that changes in milk FA profile can be achieved by the direct selection or indirect by selecting for fat yield and fat content. Moreover, this thesis provides the first results about the impact of different ssGBLUP methods (i.e., different scaling and weighting factors) based on RRM for genomic prediction of milk FA. In summary, our results demonstrated the superiority of ssGBLUP approach based on RRM in prediction performance of milk FA and supports further studies in order to improve reliabilities and reduce bias for milk FA. Keywords: Genetic parameters. Genomic prediction. Milk quality. Random regression model. Reliability. Single-step GBLUP.