Changes to evaluation system (December 2014)
Genetic base change for all traits
By Paul VanRaden, Mel Tooker, Jan Wright, John Cole, Dan Null, and Tom Lawlor
Genetic bases for all traits were updated by 5 years. For most traits of most breeds, average predicted transmitting abilities (PTAs) decreased. For traits affected by herd-variance adjustments, standard deviations (SDs) were revised to reflect genetic SD for cows born in 2010 instead of cows born in 2005. Changes for each breed are reported in “Genetic Base Changes for December 2014.” The next base change is scheduled for 2020.
Net merit revised and grazing merit introduced
By John Cole and Paul VanRaden
Economic values in net merit (NM$) were updated, 2 more fertility traits were included, and a 4th index called grazing merit (GM$) was introduced to rank animals for grazing herds along with cheese merit (CM$) and fluid merit (FM$) indexes previously designed for herds with differing milk markets. GM$ will be provided in bytes 626–630 of format 38 and bytes 82–86 of format 105 and is recommended for herd owners desiring to improve fertility to maintain seasonal calving cycles. The indexes previously included daughter pregnancy rate (DPR) and now also include heifer conception rate (HCR) and cow conception rate (CCR). Benefits of fertility not already included in productive life (PL) are earlier age at first calving; decreased units of semen needed per pregnancy; decreased labor and supplies for heat detection, synchronization, inseminations, and pregnancy checks; additional calves produced; and higher yields because more ideal lactation lengths are achieved. Separating the economic benefits of CCR from DPR is not simple because the 2 cow fertility traits overlap. Both are major components of PL, but the benefits from more lactations are included in the PL economic value. Updated indexes promote balanced selection to ensure maximum economic progress. See “Net Merit as a Measure of Lifetime Profit: 2014 Revision” for more details on calculation of the indexes.
Delivery of preliminary genomic evaluations weekly
By George Wiggans, Leigh Walton, and Paul VanRaden
Recently received genotypes will be processed each week to generate approximate genomic evaluations for new animals. Results will be placed in the same directory as for monthly files in the file transfer protocol (FTP) accounts of the nominators, dairy records processing centers for animals from herds for which they process records, and breed associations for evaluations designated as “public.” Files will be posted at 8 AM each Tuesday. The files will include preliminary genomic evaluations for new animals and for animals with genotypes that became usable since the previous weekly evaluation (for example, because of corrected pedigrees). For the convenience of recipients, the CSV files will contain the same columns as the monthly files, but data for reliability and some other columns will be left blank. Processing time will be minimized by not calculating reliability and using an approximation in the step where genomic values are combined with traditional parent average or PTA. This difference along with month-to-month changes in imputation, pedigree, and estimates of effects for single nucleotide polymorphisms (SNPs) will result in some differences between the preliminary and following month’s genomic evaluations, but correlations are very high. The immediacy of the preliminary evaluations provides an incentive to smooth the workload by submitting genotypes throughout the month rather than submitting most at the monthly due date. Another major benefit is to allow earlier sale or culling of animals (or embryos) not needed for breeding purposes to minimize the expense of raising newborn calves.
New genetic evaluation software
By Paul VanRaden, Mel Tooker, Jan Wright, Chuanyu Sun, and Jana Hutchison
Flexible software was introduced to allow more efficient model changes and multi-trait processing. Genetic evaluations from the new and previous software were very similar for 7 traits of 30 million cows; correlations with the previous single-trait, all-breed evaluations were >0.995 for progeny-tested bulls. Multi-trait methods were nearly identical to single-trait for milk, fat, and SCS but improved for protein because of missing data in early years and high correlations with milk and fat. The new software was previously implemented for a 3-trait fertility model including HCR and CCR in December 2013 and will be implemented for yield, SCS, single-trait PL, and DPR in December 2014, replacing the animal model evaluation programs (Wiggans et al., 1988) used for official evaluations since 1989. Formats were not affected by the software change except that the “evaluation used for relatives” (EUFR) fields in format 105, bytes 299–323, are discontinued and filled with zeros because separate processing is no longer done for cows with missing first-lactation records. The genetic correlations computed by Interbull in the September test run mostly changed by 0.01 or less for U.S. with foreign bull evaluations, indicating little change in international rankings from implementing the new software.
Revised evaluation of DPR
By Paul VanRaden and Jan Wright
DPR is nonlinearly related to days open, whereas previously a linear approximation was used to convert from days open to DPR. Individual pregnancy rate observations are a series of “no” or “yes” events at 21-day intervals between 50 and 250 days in milk (DIM). The lactation record for pregnancy rate is the percentage of opportunity periods containing a yes, not including opportunity periods after the cow becomes pregnant. Cows still open at 250 DIM are no longer assumed pregnant at 250 DIM and instead have a pregnancy rate of 0. The lactation records are weighted by number of opportunities (n) using weights = n/[1 + (n−1)repeat], where repeat = 0.12 and is the repeatability of individual observations. Previously, DPR records received equal weights and had heritability = 4.0% per lactation, but the new heritability is 1.4% per 21 days because each individual observation contains less information than lactation averages. The evaluations and definitions of DPR and conception rate are now more similar, providing more stable multi-trait evaluations especially when early daughters are in mid-lactation. The new definition increased the SD of DPR by about 35% for Holsteins and other breeds but 70% for Jerseys, causing a more uniform genetic SD across breeds because the Jersey SD had been lower with the previous definition. Most genetic correlations computed by Interbull in the September test run changed by only 0.01 for U.S. with foreign bull evaluations. Correlations of previous and new DPR evaluations were 0.97 for domestic, progeny-tested bulls. The new, multi-trait DPR evaluation replaces the single-trait evaluation reported since 2003.