Search

The genomic breeding value accuracy of scarcely recorded traits is low because of the limited number of phenotypic observations. One solution to increase the breeding value accuracy is to use predictor traits. This study investigated the impact of recording additional phenotypic observations for predictor traits on reference and evaluated animals on the genomic breeding value accuracy for a scarcely recorded trait. The scarcely recorded trait was dry matter intake (DMI, n = 869) and the predictor traits were fat-protein-corrected milk (FPCM, n = 1520) and live weight (LW, n = 1309). All phenotyped animals were genotyped and originated from research farms in Ireland, the United Kingdom and the Netherlands. Multi-trait REML was used to simultaneously estimate variance components and breeding values for DMI using available predictors. In addition, analyses using only pedigree relationships were performed. Breeding value accuracy was assessed through cross-validation (CV) and prediction error variance (PEV). CV groups (n = 7) were defined by splitting animals across genetic lines and management groups within country. With no additional traits recorded for the evaluated animals, both CV- and PEV-based accuracies for DMI were substantially higher for genomic than for pedigree analyses (CV: max. 0.26 for pedigree and 0.33 for genomic analyses; PEV: max. 0.45 and 0.52, respectively). With additional traits available, the differences between pedigree and genomic accuracies diminished. With additional recording for FPCM, pedigree accuracies increased from 0.26 to 0.47 for CV and from 0.45 to 0.48 for PEV. Genomic accuracies increased from 0.33 to 0.50 for CV and from 0.52 to 0.53 for PEV. With additional recording for LW instead of FPCM, pedigree accuracies increased to 0.54 for CV and to 0.61 for PEV. Genomic accuracies increased to 0.57 for CV and to 0.60 for PEV. With both FPCM and LW available for evaluated animals, accuracy was highest (0.62 for CV and 0.61 for PEV in pedigree, and 0.63 for CV and 0.61 for PEV in genomic analyses). Recording predictor traits for only the reference population did not increase DMI breeding value accuracy. Recording predictor traits for both reference and evaluated animals significantly increased DMI breeding value accuracy and removed the bias observed when only reference animals had records. The benefit of using genomic instead of pedigree relationships was reduced when more predictor traits were used. Using predictor traits may be an inexpensive way to significantly increase the accuracy and remove the bias of (genomic) breeding values of scarcely recorded traits such as feed intake.

This study set out to demonstrate the feasibility of merging data from different experimental resource dairy populations for joint genetic analyses. Data from four experimental herds located in three different countries (Scotland, Ireland and the Netherlands) were used for this purpose. Animals were first lactation Holstein cows that participated in ongoing or previously completed selection and feeding experiments. Data included a total of 60 058 weekly records from 1630 cows across the four herds; number of cows per herd ranged from 90 to 563. Weekly records were extracted from the individual herd databases and included seven traits: milk, fat and protein yield, milk somatic cell count, liveweight, dry matter intake and energy intake. Missing records were predicted with the use of random regression models, so that at the end there were 44 weekly records, corresponding to the typical 305-day lactation, for each cow. A total of 23 different lactation traits were derived from these records: total milk, fat and protein yield, average fat and protein percentage, average fat-to-protein ratio, total dry matter and energy intake and average dry matter intake-to-milk yield ratio in lactation weeks 1 to 44 and 1 to 15; average milk somatic cell count in lactation weeks 1 to 15 and 16 to 44; average liveweight in lactation weeks 1 to 44; and average energy balance in lactation weeks 1 to 44 and 1 to 15. Data were subsequently merged across the four herds into a single dataset, which was analysed with mixed linear models. Genetic variance and heritability estimates were greater (P < 0.05) than zero for all traits except for average milk somatic cell count in weeks 16 to 44. Proportion of total phenotypic variance due to genotype-by-environment (sire-by-herd) interaction was not different (P > 0.05) from zero. When estimable, the genetic correlation between herds ranged from 0.85 to 0.99. Results suggested that merging experimental herd data into a single dataset is both feasible and sensible, despite potential differences in management and recording of the animals in the four herds. Merging experimental data will increase power of detection in a genetic analysis and augment the potential reference population in genome-wide association studies, especially of difficult-to-record traits.