Changes to evaluation system (August 2016)

Cow livability evaluation

By Jan Wright, Paul VanRaden, Gary Fok, and Mel Tooker

Predicted transmitting abilities (PTA), reliabilities, and parent averages for cow livability (LIV) are now reported in bytes 414-425 of format 105 for cows and bytes 631-653 of format 38 for bulls, along with numbers of daughters and herds with data. LIV measures a cow's ability to stay alive while on the farm, whereas PTA for productive life (PL) measures a cow's ability to avoid either dying on the farm or being culled. Cows that die are a large expense compared to those sold for beef. About 20% of cows die instead of being sold, with death losses averaging 7% / lactation, but higher in later than in earlier lactations. Reasons for disposal including death loss have been reported and stored in DHI records since about 1970.

LIV records for 69,710,392 lactations of 25,514,760 cows were evaluated with an all-breed animal model, using edits and methods of evaluation similar to a previous study of U.S. cow mortality (Miller et al., 2008). The scale is now reversed to report cow livability instead of mortality so that positive PTAs are favorable (0 = died, 100 = lived for each lactation). A multi-trait model was developed to increase the reliability of livability, using individual lactation records for PL instead of one lifetime PL record that is used officially. The LIV PTA from this 2-trait model will be reported, but official PL PTA will continue to use the lifetime record because further research is needed on the proposed PL lactation trait definition. The heritability of LIV per lactation on the observed scale was 1.3% (Miller et al., 2008) compared to a heritability of 3.0% for overall culling rate per lactation.

Genomic PTA (GPTA) for LIV of young bulls were computed from 4-year truncated data and had squared correlations with future data about twice as high as parent averages (PAs). Genomic reliability was 56% compared to 30% from PA. Livability is a good trait to add because so many historical records are already available for free in the national database, allowing accurate GPTA to be computed. The United States in 1994 was the first country to evaluate PL, and in 2016 is the first country to add cow mortality or livability as a specific economic trait.

The SD of true transmitting ability for LIV is 0.82% / lactation or 2.3% / lifetime using an average of 2.8 lactations / cow. Reporting cow livability on a lifetime instead of per lactation basis should improve understanding because LIV PTAs will express mortality differences as a percentage of all cows leaving the herd. For recent bulls with > 80% reliability, LIV PTAs are correlated favorably by about 0.70 to PL, 0.45 to daughter pregnancy rate (DPR), and -0.25 to somatic cell score (SCS) PTAs, with low correlations to yield trait PTAs. The 0.70 correlation with PL seems sufficiently below 1 to add value from selecting for both LIV and PL in an index.

Economic values for LIV and PL were derived using input and output prices assumed in the current version of net merit (NM$; VanRaden and Cole, 2014). The relative emphasis on LIV was estimated to be 7.4% of total emphasis, and the relative emphasis on PL value will decline from 19.1% relative value to 13.5% because the PL economic value will include only the losses other than death. Thus, total emphasis on cow longevity and livability should increase from 19.1% to 20.9% but will be split across 2 correlated traits instead of just 1 trait. The proposed change to include LIV in NM$ will not be implemented immediately so that breeders and producers can first get some experience with the PTA for this new trait. Further information is available on use of livability (Norman et al., 2016) and methods of calculation (Form GE).

Future evaluations might also include calf livability. About 5% of heifers have termination codes reported, and >20,000 calf deaths were reported to the CDCB database each year for about 10 years, but heritability for calf loss is very low.

Genomic mating file format

By Gary Fok, Leigh Walton, and Paul VanRaden

Sizes of the genomic mating files were reduced by revising the format and excluding cows that are likely to be dead. The Holstein file had increased to over 28 Gbytes and to over 6 Gbytes compressed after zipping into 4 separate subfiles. The new format has the genomic relationship multiplied by 100 to remove the 0 and decimal point and also has 1 less digit of accuracy. The format change reduces the uncompressed and compressed file sizes to about half of their previous sizes, but users must then divide the relationship values by 100 to obtain the previous scale. The other change removed USA cows with a termination code of 3-7 (culled), or a last calving date of at least 27 months since last fresh date and not still a record in progress. Those edits were applied to females at least 3 years old at the start of the evaluation, but not to foreign females because no direct reports of alive/dead status are available for those. This edit removed only about 20% of USA cows from the file, but will limit file sizes much more in the future as more genotyped females become older and leave their herd. Previously all females were included in the mating file for each breed.

Genotype exchanges with Japan and Switzerland

By George Wiggans

Genotypes for 3,000 Holstein bulls from Japan were added to the North American reference population in May 2016, with American and Canadian proven bull genotypes provided to Japan in exchange. This agreement between the National Livestock Breeding Center of Japan and the Cooperative Dairy DNA Repository (CDDR) included only daughter proven bulls. Because the Holsteins of Japan and North America are highly related and the genetic correlations are also high for most traits, reliabilities of U.S. genomic predictions are expected to increase slightly. After exchanges with both Japan and Switzerland in recent months, the Holstein reference population has increased to 34,016 bulls plus 232,919 cows.

Genotypes for Holstein and Red Holstein bulls from Switzerland were provided by Qualitas AG (Zug, Switzerland) pursuant to an exchange agreement between the Holstein Association of Switzerland, SwissHerdbook Cooperative Zollikofen, and the CDDR. After removing duplicate genotypes already available in both databases, 4,284 bulls were added to the CDCB database. Of the 4,284 bulls, 1,537 have a progeny-based evaluation and add to the North American reference population. About 120 of the bulls have Simmental dams; they were identified as crossbreds and, as such, did not have genomic evaluations calculated. The genotypes were added in the weekly evaluations released on February 23, 2016, and in the March 2016 monthly evaluation. This agreement with Switzerland is similar to previous exchange agreements with Italy, the United Kingdom, and Canada. Bull genotypes are shared by those 5 countries.

Mutation in HH5

By Paul VanRaden and Dan Null

Holstein haplotype 5 (HH5) carrier status was improved by using the exact location of the mutation. Schütz et al. (2016) reported that the mutation on chromosome 9 is a large deletion at location 93,223,651 – 93,370,998 (UMD3) containing the whole gene TFB1M. Previously 46,784 animals were determined to be HH5 carriers using a fine-mapped region of about 1 Mbase, and only 69 additional animals became carriers using the known location of the mutation. Direct test results for the mutation could be included in the future to further improve accuracy, as is done with several other recessive haplotypes.