Skip to main content
ARS Home » Research » Publications at this Location » Publication #72225

Title: APPLICATION OF A CANONICAL TRANSFORMATION WITH MISSING VALUES TO MULTITRAIT ANALYSIS OF U.S. JERSEY TYPE DATA

Author
item GENGLER, NICOLAS - FNRS (BELGIUM)
item Wiggans, George
item Wright, Janice
item Norman, H
item WOLFE, CARI - AJCA

Submitted to: Interbull Annual Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/23/1996
Publication Date: N/A
Citation: N/A

Interpretive Summary: The American Jersey Cattle Association currently scores 15 linear type traits (stature, strength, dairy form, foot angle, rear legs-side view, body depth, rump angle, thurl width, fore udder attachment, rear udder height, rear udder width, udder depth, udder cleft, front teat placement, and teat length) in addition to final score if requested by the breeder. In the past, genetic evaluations for type traits have used single-trait sire models. Recent advances in genetic evaluation methodology were used to develop a system for Jersey type traits that provides more accurate evaluations at a reasonable cost. The new system improves accuracy by considering correlations among traits and all relationships among animals. The analysis uses a multitrait animal model. Computational requirements are reduced by using a canonical transofrmation that allows for missing values. The effect of inbreeding also can be included to increase accuracy further. This methodology should give U.S. Jersey breeders the advanced tools they need to make more accurate genetic selections for final score and linear type traits.

Technical Abstract: Predicted transmitting abilities and associated reliabilities were estimated for final score and 15 linear type traits (stature, strength, dairy form, foot angle, rear legs-side view, body depth, rump angle, thurl width, fore udder attachment, rear udder height, rear udder width, udder depth, udder cleft, front teat placement, and teat length) for 225,632 U.S. Jersey cows using a multitrait animal model. Multiple diagonalization, an extension of canonical transformation to several random effects, was adapted for missing values and applied to 381,511 records with 10-15% of observations missing for final score, body depth, and teat length. Inbreeding was considered in the computations. The model contained a scoring year-parity-age effect to complement multiplicative adjustment of type scores for age and stage of lactation. Convergence was achieved in approximately 50 rounds. Correlations between multitrait animal model evaluations and those from the previous sire model ranged from .56 to .92, increased over time, and were higher for bulls with >40 progeny than for cows born during the same years. Traits with missing values generally had lower correlations than did other traits; correlations for final score decreased slightly for bulls born after 1984, and correlations for body depth and teat length were lower for bulls born before 1980. Recent annual trend in estimated breeding value was greater for dairy form (.77) than for final score (.55) or rear udder traits (.50-.52). Almost no trend was evident for feet and leg traits and fore udder attachment; trend for udder depth was negative (-.13). The new multitrait animal model methodology should improve accuracy of genetic evaluations for U.S. Jersey type traits.