Abstract The variance components needed to calculate genetic and phenotypic parameters for selection to improve carcass composition in white-faced sheep were estimated from a sire model using REML on data from 1602 Romney and Romneycross lambs. The sires were either Romney (n = 60), Border Leicester (n = 18), Poll Dorset (n = 14), or Coopworth (n = 10) with approximately 16 progeny per sire and all dams were Romney. Lambs had an average liveweight of 31.0 kg at an average age of 205 days. Carcasses were dissected into lean, subcutaneous fat, intermuscular fat, bone, and waste. The multivariate mixed linear model included birth date of the lamb as a linear covariate, year-rearing group, breed of sire, sex of lamb, birth rank, and slaughter group as fixed effects, and sire as a random effect. Traits were categorised as either lean, fat, or skeletal dimension indicators. Heritability estimates for lean indicator traits obtained by dissection averaged 0.36. The lean indicator trait with the highest heritability was longissimus muscle width (h 2 = 0.63). Heritability estimates for fat depth measurements averaged 0.23. Heritability of skeletal dimension traits averaged 0.43. Expected responses in a selection objective, comprising increased lean weight and decreased fat weight, were compared for selection indices consisting of hot-carcass weight and one of 26 carcass measurements, In general, both lean and fat indicator traits contributed to a reduction in fat response. The index comprising hot-carcass weight and weight of lean tissue, obtained by dissection, from the leg joint, produced the greatest economic response in the selection objective. The economic response from a nondissection trait was from selection on the index which comprised hotcarcass weight and longissimus muscle width, A. The increase in lean tissue response over that predicted from single trait selection on HCW was small (0.04 kg) relative to the decrease in FAT (0.20 kg). None of the dimension traits produced a negative predicted response in fat. In the range of predicted responses from the two trait indices, there was greater variation in the fat responses than in the lean responses.
Read full abstract