Abstract

In New Zealand genetic evaluation for cow longevitywas based on a single trait model where the phenotypic record was actual or predicted days survived in the milking herd. Cows which were yet to express their actual longevity received a predicted phenotypic record based on the milk production and liveweight production values, traits other than production (TOP), and fertility breeding values. The prediction equation was a proxy for a multiple trait model. At the time of implementation a full multiple trait model was not computationally feasible. A study was undertaken to investigate the feasibility of developing a full multiple trait model. The multiple trait model has the advantage of incorporating the appropriate genetic relationships among the predictor traits and the longevity traits, as well as allowing different genetic correlations among the individual longevity traits. The first stage of the study was to estimate the genetic correlations among all traits under genetic evaluation and two longevity traits; survival from first to second lactation and first to fifth lactation. Data used for genetic correlation estimation was from first lactation records from the Sire Proving Scheme herds for seasons 1987 to 2005. The two fertility traits, 42 day calving rate and percent mated in the first 21 days, had the highest genetic correlations with the longevity traits, ranging from 0.84 to 0.27. Body condition score, management TOP and the milk production traits had moderate genetic correlations with the longevity traits, ranging from 0.42 to 0.24. Somatic cell score, udder support, udder overall, legs, capacity and dairy conformation had low genetic correlations with the longevity traits but were significantly greater than zero. Four longevity traits, and 9 predictor traits were included in a national genetic evaluation for cow longevity. The 9 predictor traits were chosen from the traits that had significant genetic correlations with cow longevity and contained genetic information that was not already included in other predictor traits. Comparisons between sire breeding values from the single trait model and the multiple trait model show good agreement with a correlation of 0.95 for Animal Evaluation enrolled sires. Investigation of the sire breeding values shows that the multiple trait model puts a greater emphasis on cow fertility than the single trait model.

BL, Harris, and WA Montgomerie

Proceedings of the New Zealand Society of Animal Production, Volume 67, Wanaka, 377-381, 2007