Milk progesterone measures to improve genomic selection for fertility in dairy cows

Abstract: Improved reproductive performance has a substantial benefit for the overall profitability of dairy cattle farming by decreasing insemination and veterinary treatment costs, shortening calving intervals, and lowering the rate of involuntary culling. Unfortunately, the low heritability of classical fertility traits derived from calving and insemination data makes genetic improvement by traditional animal breeding slow. Therefore, there is an interest in finding novel measures of fertility that have a higher heritability or using genomic information to aid genetic selection for fertility. The overall objective of this thesis was to explore the use of milk progesterone (P4) records and genomic information to improve selection for fertility in dairy cows. In a first step, the use of in-line milk progesterone records to define endocrine fertility traits was investigated, and genetic parameters estimated. Several defined endocrine fertility traits were heritable, and showed a reasonable repeatability. Also, the genetic correlation of milk production traits with endocrine fertility traits were considerably lower than the correlations of milk production with classical fertility traits. In the next step 17 quantitative trait loci (QTL) associated with endocrine fertility traits, were identified on Bos taurus autosomes (BTA) 2, 3, 8, 12, 15, 17, 23, and 25 in a genome-wide association study with single nucleotide polymorphisms. Further, fine-mapping of target regions on BTA 2 and 3, identified several associated variants and potential candidate genes underlying endocrine fertility traits. Subsequently, the optimal use of endocrine fertility traits in genomic evaluations was investigated; using empirical and theoretical predictions for single-trait models, I showed that endocrine fertility traits have more predictive ability than classical fertility traits. The accuracy of genomic prediction was also substantially improved when endocrine and classical fertility traits were combined in multi-trait genomic prediction. Finally, using deterministic predictions, the potential accuracy of multi-trait genomic selection when combining a cow training population measured for the endocrine trait commencement of luteal activity (C-LA), with a training population of bulls with daughter observations for a classical fertility trait was investigated. Results showed that for prediction of fertility, there is no benefit of investing in a cow training population when the breeding goal is based on classical fertility traits. However, when considering a more biological breeding goal for fertility like C-LA, accuracy is substantially improved when endocrine traits are available from a limited number of farms.