Evaluation of different animal models to predict breeding value for growth traits in Lori Bakhtiari sheep

Accurate prediction of breeding value of animals is influenced by several factors such as selection of the best genetic analysis model. For this purpose, pedigree and phenotypic information of birth weight (BW0), weaning weight (BW3) and six-month weight (BW6) of Lori-Bakhtiari sheep from 1989 to 2011 were used. (Co) variance components were estimated by Restricted Maximum Likelihood (REML) with different univariate and multivariate animal model. The results showed that heritability of traits is constant in different two and three trait models (0.47, 0.27 and 0.12 for BW0, BW3 and BW6, respectively), but absolute difference between the genetic and residual correlations changed up to %11 in different models (0.23, 0.32 and 0.65 in two-trait vs. 0.19, 0.31 and 0.54 in three-trait for BW3-BW6, BW0-BW3 and BW0-BW6 analyses, respectively). The correlation coefficients between animal breeding values were principally upper 0.99 for BW0 in different models, whereas these coefficients for BW3 in total population were 0.43-0.97, in ewe population 0.89-0.98 and in rams population 0.92-0.99, and for BW6 in total population 0.53-0.93, in ewes population 0.52-0.94 and in rams population 0.58-0.93. Rank correlations were also high for BW0 in three studied populations, but BW3 rank correlations between single with two and three-trait models in three populations were 0.42-0.96, 0.90-0.98 and 0.92-0.97, and for BW6 0.52-0.60, 0.54-0.56 and 0.61-0.67, respectively. The results showed that there was no difference between different models to predict breeding values of birth weight, but differences were significant for other two traits.