author(s)
Dr. BV Reddy S, Dr. Punya Kumari B, Dr. Sudhakar K, Dr. Jagadeeswara Rao S and Dr. Ramani Pushpa RN

abstract
Breeding values were estimated for body weights at birth, 3, 6, 9 and 12 months of age in Nellore Jodipi sheep maintained at Net Work Project on Sheep Improvement (NWPSI), Livestock Research Station, Palamaner of Chittoor district, Andhra Pradesh, India. Records on production traits of 4969 Nellore Jodipi lambs from1993 to 2016 (23 years) descended from 188 sires were included in the analysis. Analyses were carried out by least-squares (LS), best linear unbiased prediction (BLUP) and derivative free restricted maximum likelihood (DFREML) methods. BLUP method had the lower error variance (kg2) for all the production traits as 0.18 for BW, 8.90 for 3M, 13.73 for 6M, 16.19 for 9M and 16.13 for 12M. Whereas, in LS and DFREML methods the error variances (kg2) for these traits were 0.18, 0.18 for BW; 8.90, 9.15 for 3M; 13.74, 14.07 for 6M; 16.22, 16.51 for 9M; 16.19, 16.30 for 12M body weights. The coefficient of R2 (%) values obtained for the traits were 22 (BW), 31 (3M), 34 (6M), 37 (9M) and 45 (12M) by LSM, followed by BLUP as 18 (BW), 27 (3M), 30 (6M), 31 (9M) and 37 (12M). For DFREML 9 (BW), 22 (3M), 22 (6M), 18 (9M) and 17 (12M).Spearman’s rank correlation for DFREML with BLUP and LS estimates were highly significant (P<0.01). The rank correlations between BLUP and DFREML were greater than 0.9 for all the traits indicating that any of these two methods of sire evaluation could select the same set of sires with almost similar ranking for unbiased estimates of breeding values for production traits. The BLUP method seems to be the most efficient (lower error variance) and LSM as the most accurate (higher coefficient of determination) method for the evaluation of production traits in Nellore Jodipi sheep, but the accuracy of BLUP was much closer to LSM.