Comparison of Methods to Rank Bulls Across Countries

• Published in: Journal of dairy science 1992-09, Vol.75 (9), p.2560-2568
• Main Authors: Banos, G., Wiggans, G.R., Robinson, J.A.B.
• Format: Article
• Language: English
• Subjects: Animal productions; Biological and medical sciences; conversion; Fundamental and applied biological sciences. Psychology; international sire comparison; joint evaluation; linear model; Terrestrial animal productions; Vertebrates
• ISSN: 0022-0302; 1525-3198
• DOI: 10.3168/jds.S0022-0302(92)78018-7

Ayrshire (5149) and Jersey (10,525) bulls born in Canada and the US between 1950 and 1985 were ranked for genetic merit for milk and fat yields. Ranking methods included two joint evaluations (US and Canadian) using lactation records from both countries, a linear model combination of evaluations from each country's national evaluations, combined US national and converted Canadian evaluations, and combined Canadian national and converted US evaluations. Correlations of linear model combined evaluations for milk yield with joint evaluations were .96 for Ayrshires and .98 for Jerseys for US joint evaluations and .89 and .84 for Canadian joint evaluations. Correlations of combined national and converted evaluations with joint evaluations ranged from .88 to .94 for Ayrshires and from .83 to .98 for Jerseys. Variation and correlations between joint and linear model combined evaluations were consistent across birth year of bull. Sire-son correlations were similar for US joint and linear model combined evaluations but were lower for Canadian joint evaluations. Estimated genetic progress for bulls born from 1970 to 1985 was similar for Jerseys from US joint and linear model combined evaluations but for Ayrshires was over 25% higher from US joint evaluation than from linear model combined evaluation. For both breeds, progress estimated from Canadian joint evaluation was over 20% lower than from linear model combined evaluation. Results for milk and fat yields were similar. Joint evaluation provides the most accurate evaluations because it uses all relationships. However, because linear model combination evaluations are highly correlated with joint evaluations and usually are superior to national evaluations extended with converted foreign evaluations, they are recommended if joint evaluation is not feasible.