|Title||Opportunities for genomic prediction for fertility using endocrine and classical fertility traits in dairy cattle|
|Author(s)||Tenghe, A.M.M.; Berglund, B.; Wall, E.; Veerkamp, R.F.; Koning, D.J. de|
|Source||Journal of Animal Science 94 (2016)9. - ISSN 0021-8812 - p. 3645 - 3654.|
Animal Breeding and Genetics
LR - Animal Breeding & Genomics
|Publication type||Refereed Article in a scientific journal|
|Keyword(s)||Dairy cattle - Endocrine fertility traits - Fertility - Milk progesterone - Multitrait genomic prediction - Validation|
Endocrine fertility traits, defined from progesterone concentration levels in milk, have been suggested as alternative indicators for fertility in dairy cows because they are less biased by farm management decisions and more directly reflect a cow’s reproductive physiology than classical traits derived from insemination and calving data. To determine the potential use of endocrine fertility traits in genomic evaluations, the improvement in accuracy from using endocrine fertility traits concurrent with classical traits in the genomic prediction of fertility was quantified. The impact of recording all traits on all training animals was also investigated. Endocrine and classical fertility records were available on 5,339 lactations from 2,447 Holstein cows in Ireland, the Netherlands, Sweden, and the United Kingdom. The endocrine traits were commencement of luteal activity (C-LA]) and proportion of samples with luteal activity (PLA); the classical trait was the interval from calving to first service (CFS). The interval from C-LA to first service (C-LAFS), which is a combination of an endocrine trait and a classical trait, was also investigated. The target (breeding goal) trait for fertility was CFS or C-LAFS, whereas C-LA and PLA served as predictor traits. Genomic EBV (GEBV) for fertility were derived using genomic BLUP in bivariate models with 85,485 SNP. Genomic EBV for the separate fertility traits were also computed, in univariate models. The accuracy of GEBV was evaluated by 5-fold cross-validation. The highest accuracy of GEBV was achieved using bivariate predictions, where both an endocrine fertility trait and the classical fertility trait were used. Accuracy of GEBV for predicting adjusted phenotypes for CFS in the univariate model was 0.04, but when predicting CFS using a bivariate model with C-LA, the accuracy increased to 0.14 when all training animals were phenotyped for C-LA and (or not) for CFS. On phenotyping all training animals for both C-LA and CFS, accuracy for CFS increased to 0.18; however, when validation animals were also phenotyped for C-LA, there was no substantial increase in accuracy. When predicting CFS in bivariate analysis with PLA, accuracy ranged from 0.07 to 0.14. This first study on genomic predictions for fertility using endocrine traits suggests some improvement in the accuracy of prediction over using only the classical traits. Further studies with larger training populations may show greater improvements.