Estimation of Breeding Values with Heterogeneous Residual Variances by Random Regression Models


TAKMA Ç. , AKBAŞ Y.

KAFKAS UNIVERSITESI VETERINER FAKULTESI DERGISI, cilt.22, ss.37-41, 2016 (SCI İndekslerine Giren Dergi) identifier identifier

  • Cilt numarası: 22 Konu: 1
  • Basım Tarihi: 2016
  • Doi Numarası: 10.9775/kvfd.2015.14364
  • Dergi Adı: KAFKAS UNIVERSITESI VETERINER FAKULTESI DERGISI
  • Sayfa Sayıları: ss.37-41

Özet

In this study the third order random regression models (RV1 and RV10) including the fixed, random additive genetic and permanent environmental effects were used. In the RV1 model residual variance was constant and in the RV10 model all test day records were taken as different groups. The predicted error variance was found 6.83 in RV1 model and this variance was changed between 5.30 and 9.19 in RV10 model. Heritability values were estimated 0.12-0.53 for RV1 and 0.04-0.18 for RV10 models. Spearman rank and Kendall rank correlations between estimated breeding values of test day milk yields estimated from RV1 and RV10 models within cows (0.79, 0.61) were found almost same with within sires (0.82 and 0.63). Consequently, these correlations indicate that breeding values estimated from RV1 and RV10 models were highly correlated. Although the association between two set of breeding values estimated from RV1 and RV10 models were high, ranking of cows and sires by breeding values were different for two models. The shift in rank of first 100 cows was found 22% same and 78% different, also the shift in rank of first 50 sires was found 30% same and 70% different on RV1 and RV10 models breeding values rank list. Overall, the RV10 model was found better than RV1 model in estimation of breeding value.