Changes to evaluation system (December 2016)


Revised conformation evaluation for non-Holstein breeds

By Jan Wright and Paul VanRaden

Conformation trait evaluations for Jerseys, Brown Swiss, Ayrshires, Guernseys, and Milking Shorthorn were revised in several ways. The new genetic evaluation software implemented in 2014 for most other traits is now also applied to linear type traits, final score, and also to Brown Swiss milking speed and mobility. All variance component estimates (heritabilities and trait correlations) were updated by sire model REML analysis. All traits are now solved together instead of performing separate analyses for traits introduced more recently, and that is a main advantage of the new software. A disadvantage is that pre-adjustments for variance are used instead of simultaneous adjustment for variance during iteration. The authors thank Nicolas Gengler for use of the previous software since Feb. 1998 (Gengler et al., 1999). Most genetic standard deviations changed by < 0.1, and only 3 breed-trait combinations changed by > 0.2: 0.24 for Brown Swiss angularity, -0.23 for Guernsey rear legs rear view, and 0.20 for Ayrshire stature. Correlations of new with previous predicted transmitting abilities (PTAs) for progeny tested bulls born since 1995 for 14 traditionally scored linear type traits averaged 0.985 in Jersey, 0.969 in Brown Swiss, 0.966 in Ayrshire, 0.985 in Guernsey, and 0.892 in Milking Shorthorn. Respective correlations for final score were 0.993, 0.974, 0.988, 0.982, and 0.929. Traits introduced more recently such as rear legs (rear view) and mobility had much lower correlations because of the new multi-trait model. The genetic correlations with other countries estimated by Interbull were nearly the same before and after these changes.

Rear teat placement rear view (RTPR) and rear teat placement side view (RTPS) were added as new traits for Jerseys. Heritabilities were estimated to be 21% and 13%. The revised formats 40 and 41 incorporating the new changes are available at:

https://www.uscdcb.com/CF-RCS/GetRCS.cfm?DocType=formats&DocName=fmt40.html
https://www.uscdcb.com/CF-RCS/GetRCS.cfm?DocType=formats&DocName=fmt41.html


Variance adjustments and heritability for cow livability

By Paul VanRaden and Jan Wright

Cow livability (CL) evaluations were revised slightly with variance adjustments that now include different heritabilities by parity. The CL records were pre-adjusted for parity-year variance and weighted based on heritability to account for the changes in variance with the mean. Heritabilities for each parity were estimated by sire model REML from > 5 million daughters of 4,976 Holstein sires. Heritabilities of CL from the first 5 lactations were 0.43, 0.70, 0.73, 0.74, and 0.74%, respectively, and average only 0.6% when weighted by number of lactations. The new heritabilities are equivalent to about 3% on the underlying scale but are lower than the previous estimate of 1.3% for heritability per lactation on the observed scale from Miller et al. (2008) which was used in the initial August 2016 evaluation. The new, lower estimates of lactation specific heritability will be introduced in December. Livability deviations from group mean are adjusted by multiplying by square root of the ratio of base group mean to lactation-year group mean and by square root of ratio of base heritability to lactation heritability.