Flexible testing and use of milk-only records
P.M. VanRaden, G.C. Fok, L.R. Bacheller, G.B. Jansen, E.L. Nicolazzi, and J.A. Carrillo
Effective with the April 5, 2022 evaluations, edit programs were revised to begin using milk-only records in the multi-trait evaluation of yield traits. Other trait groups such as cow fertility and health also will have a greater number of usable records, because edits for those traits require usable milk yield records. Previously, national genetic evaluation software required that fat yield was recorded, and milk-only records were excluded. We estimate that this edit adds nearly 900,000 milk-only records for the relevant traits.
Cows with milk-only records will have evaluation code set to 5 in byte 110 of format 105.
Milking systems often can accurately measure and record milk volume, but inline estimation of milk components is more difficult. Since 1998 when data collection ratings (DCR) were introduced, milk and fat received the same weight calculated as an average of the DCR values for milk and components, instead of separate weights, because of software limitations. Most milk-only records are unsupervised and therefore get the same reduced weights and extra edits for percent milk shipped and percentage of valid sire ID as other owner-sampler herds.
Lactation weights for milk, fat, and protein now use three separate DCR based on the testing patterns and correlations among test days within lactations. Herd variance ratios had been estimated from one trait (milk yield since 1992 and then fat yield since 2007) and applied to adjust all three traits (milk, fat, and protein). To include milk-only records, separate trait-specific variance adjustments and weights were developed and applied to each trait. New and official genetic evaluations were compared from December 2020 data. Numbers of usable lactation records were 98,269,605 for milk, 97,393,419 for fat, and 78,044,073 for protein, indicating that 876,186 milk-only records were added. Correlations of new with previous Predicted Transmitting Abilities (PTAs) were > 0.9995 across all bulls for all three traits and were > 0.997 for bulls born since 2007 with > 50% reliability. The standard deviation (SD) of PTA increased slightly by 2.4% for milk, 0.1% for fat, and 0.4% for protein but reliability also increased slightly from the extra records.
Further research could help adapt to more flexible testing options and automated data collection that continue to increase in popularity.
Type composite update on non-Holstein breeds
By Gary Fok and Paul VanRaden
Type composite definitions for Body Weight Composite (BWC), Udder Composite (UDC) and Feet and Legs Composite (FLC) were updated in the CDCB system to match those of breed associations, per the table below. The standard deviation (SD) of BWC was revised upward to match the SD for Holstein. The genetic SD for rump angle (RAN) in Jersey was corrected according to an issue in the value being used previously in the multi-trait evaluation. The RAN SD update will have some impact on correlated traits like Rear Teat Rear View (RTR) and Rear Teat Side View (RTS) of old bulls, since these traits have been measured recently.
The change in weighting of traits on composites will affect NM$ ranking of animals, as the three composites have a combined 13.2% relative weighting on NM$. Previously BWC had less emphasis in the Jersey population, but the SD of BWC is now increased to make the emphasis more consistent with the emphasis already received in the Jersey Performance Index (JPI). On a test run based on the December 2021 evaluation, the average impact on Jersey AI bulls ranged from -10$ to -25$ on proven and young genomic bulls, respectively. However, it’s important to underline the reported NM$ variation included all the changes that will be implemented in April 2022 evaluation, and not just the type composite updates. A larger reduction in NM$ was observed in the crossbred Jersey population.
| Composite | Linear | Ayrshire | Brown Swiss |
Guernsey | Holstein | Jersey | Milking Shorthorn |
| BWC | stature | 0.23 | 0.23 | 0.23 | 0.23 | 0.42 | 0.23 |
| BWC | strength | 0.72 | 0.72 | 0.72 | 0.72 | 0.42 | 0.72 |
| BWC | dairy form | -0.47 | -0.47 | -0.47 | -0.47 | -0.21 | -0.47 |
| BWC | body depth | 0.08 | 0.08 | 0.08 | 0.08 | 0.00 | 0.08 |
| BWC | thurl width | 0.17 | 0.17 | 0.17 | 0.17 | 0.35 | 0.17 |
| UDC | stature | 0.00 | 0.00 | 0.00 | -0.20 | 0.00 | 0.00 |
| UDC | fore udder | 0.07 | 0.22 | 0.15 | 0.16 | 0.18 | 0.22 |
| UDC | rear udder ht | 0.33 | 0.18 | 0.10 | 0.23 | 0.27 | 0.18 |
| UDC | rear udder wd | 0.19 | 0.22 | 0.10 | 0.19 | 0.01 | 0.22 |
| UDC | udder depth | 0.31 | 0.10 | 0.33 | 0.20 | 0.27 | 0.10 |
| UDC | udder cleft | 0.01 | 0.18 | 0.15 | 0.08 | 0.02 | 0.18 |
| UDC | teat placement | 0.04 | 0.10 | 0.15 | 0.04 | 0.02 | 0.10 |
| UDC | teat length | 0.04 | 0.00 | -0.02 | 0.05 | -0.06 | 0.00 |
| UDC | rear teat rear | 0.00 | 0.00 | 0.00 | 0.05 | 0.07 | 0.00 |
| UDC | rear teat side | 0.00 | 0.00 | 0.00 | 0.00 | 0.09 | 0.00 |
| FLC | stature | 0.00 | 0.00 | 0.00 | -0.20 | 0.00 | 0.00 |
| FLC | foot angle | 0.70 | 0.00 | 0.48 | 0.09 | 0.50 | 0.00 |
| FLC | legs side view | -0.30 | 0.00 | -0.16 | 0.00 | -0.50 | 0.00 |
| FLC | legs rear view | 0.00 | 0.00 | 0.36 | 0.21 | 0.00 | 0.00 |
| FLC | foot leg score | 0.00 | 1.00 | 0.00 | 0.70 | 0.00 | 1.00 |
New phenotypic editing on fertility evaluations
By Asha Miles, Jana Hutchinson, and Paul VanRaden
The use of embryo transfer (ET) technology in U.S. herds has sharply increased over the years. The U.S. fertility trait evaluations have accounted for reported embryo donors or recipients since 2010, but most embryo events are not yet reported. Unreported ET reproductive events remain difficult to identify, and their systematic accumulation can lead to biased fertility evaluations of cows and bulls. New edits introduced in April 2022 now exclude fertility records from herds with large amounts of ET births but a low number of ET reproductive events. The new edits will directly impact evaluations for Sire Conception Rate (SCR), Heifer Conception Rate (HCR), and Cow Conception Rate (CCR), and also indirectly affect Daughter Pregnancy Rate (DPR). The PTA correlation of proven bulls after the implementation of this edit is over 99%. Individual animals – especially elite young bulls with early daughter records – might be highly impacted, especially by the new edit on HCR/CCR.
International evaluations processing change in policy for Heifer and Cow Conception Rate
By Ezequiel Nicolazzi and Paul VanRaden
When the Intergenomics exchange was established in 2012, a reduction in prediction accuracy for Heifer Conception Rate (HCR) and Cow Conception Rate (CCR) was observed in Brown Swiss, because of low Multiple Across Country Evaluations (MACE) correlations with some countries. As a conservative measure, only evaluations 15% more reliable than U.S. predictions were retained as phenotypes for genomic evaluations. Foreign evaluations below that threshold were discarded and the Predicted Transmitting Abilities (PTAs) were replaced for their parent averages (PAs).
Although the 15% reliability threshold is still considered valid, many of those bulls now have (direct or indirect) domestic evaluations that could be used instead of their less reliable PA. Therefore, effective with the April 2022 evaluation, PTAs for animals with foreign evaluations below the 15% threshold will be replaced with domestic PTAs where available. In a test run based on the December 2021 evaluation, only a small fraction of the population was affected directly. Since the reference population is affected, there is some impact on the overall evaluation, both as an average and as single-animal variation.
International evaluations processing change in policy for DPR
By Ezequiel Nicolazzi and Paul VanRaden
In the international evaluations for Daughter Pregnancy Rate (DPR), a determination is made whether direct traits or indirect traits will be used for foreign animals. With the April 2022 evaluations, this decision will be streamlined to consider only two trait reliabilities, compared to the previous method of comparing reliabilities of several fertility traits. Our test run shows this change had the desired effect to include many more animals with the direct DPR trait in the evaluation.
Previously, the international evaluations for DPR have compared reliabilities of several fertility traits to decide if direct or indirect traits with higher reliability should be used for foreign animals. Note that DPR was initially used as a substitute for another fertility trait until 2012 when U.S. bulls calving to first insemination PTAs were estimated and provided to Interbull: https://aipl.arsusda.gov/reference/changes/eval1212.htm
In a test run based on the December 2021 evaluation, this change in policy had the desired effect to include many more animals with the direct DPR trait in the evaluation. Most of the animals changing evaluation were old-to-very old animals, with Ayrshire being the breed most affected. Nearly 2,000 over a total of 15,970 bulls got their DPR evaluation modified, with an average change of 0.1, but with many old bulls now receiving an evaluation vs. getting a PA in December. In Holstein and Jersey, the animals affected were 8,626/151,189 and 584/13,388, with an average variation for the affected bulls of -0.066 (range: -2.3/2.0) and -0.064 (range:-1.8/2.1), respectively.
Source: uscdcb.com
