(CO)VARIANCE COMPONENT ESTIMATES FOR MILK-COMPOSITION TRAITS IN SARDADAIRY SHEEP USING A BIVARIATE ANIMAL-MODEL

A total of 7909 lactation records of 3071 Sarda ewes were used to esti mate (co)variance components for milk, fat and protein yields (MY, FY, PY) and contents (F%, P%). A derivative-free Restricted Maximum Likel ihood algorithm was used on ten bivariate analyses under an animal mod el with repeatability. Data, preadjusted for milking length, were anal ysed by a model including flock-year, parity, age at lambing, month of lambing, lambing interval and suckling period length as fixed effects . Additive genetic, permanent and temporary environmental influences w ere the random effects. Resulting heritabilities were 0.30, 0.25, 0.26 , 0.48 and 0.55 for MY, FY, PY, F% and P%, respectively. Repeatabiliti es ranged between 0.42 and 0.46 for yields and they were 0.61 and 0.69 for contents. Genetic correlations were highly positive between yield s (from 0.89 to 0.94) and between contents (0.69) and negative or null between MY, FY and PY and F% and P% (from 0.06 to -0.46). These param eters are consistent to those reported for other European dairy sheep breeds and provide basic information to address new selection objectiv es and criteria for the Sarda breed, according to the exclusive cheese -making destination of its milk. Further studies, at farm level, are n eeded to verify current estimates and to study simplified recording sc hemes for contents, cheaper than the monthly testing of two daily milk ings. (C) 1997 Elsevier Science B.V.