Characterising the quantity and quality of data used in Merino sheep genetic evaluation systems

Abstract

<p>Estimated Breeding Values (EBVs) published by Sheep Genetics Australia have an accuracy estimated with them. While the EBVs, their accuracy, and errors of genetic parameter estimates are all influenced by both data quantity and quality, these calculations do not explicitly take into account all aspects of data quality. To encourage increased genetic gains, Sheep Genetics provides participating breeders with data quantity and quality metrics in a 'RAMping Up Genetic gains' report. This paper demonstrates the considerable variation in these metrics for Merino flocks, and proposes additional descriptors metrics to characterise the quantity and quality of sheep genetic evaluation data. Current results show that there are opportunities to improve the completeness of pedigree and reproduction trait recording. Flocks had on average 46.6 ± 36.1% (mean ± SD) of animals with full pedigree, and 4.1 ± 6.9% of animals within each flock with reproduction trait records. The average proportion of effective progeny was 64.3 ± 19.1%. Flocks had on average 40.2 ± 37.3% of animals in contemporary groups that had variation in birth date recording. Since variation in age within contemporary groups is expected, this highlights potential issues with accurate recording of birth dates. Additional metrics describing lambing date distributions and deviations from the expected dates were derived, and emphasise potential issues of birth date accuracy, with some flocks recording birth dates on a non-random proportion of days of the week. Feedback on the quantity and quality of their current data should help ram breeders target improvements on their recording program. However, the optimum or reasonable level of quantity and quality to maximise genetic gains is currently undefined.</p>