Genetic parameters and response in conception, litter size and ewe rearing ability following 35 years of divergent selection for number of lambs weaned

Context Selection for a composite trait, such as number of lambs weaned per ewe joined (NLWEJ), aims to simultaneously improve the underlying components, such as fertility (ELEJ), litter size (NLBEL) or ewe rearing ability (ERAEL). However, a concern with selection on NLWEJ is the higher risk of mortality as litter sizes increase. Aim This study estimated genetic parameters, (co)variance components and the long-term genetic changes in the component traits in the Elsenburg Merino flock that has been divergently selected in favour (H-Line) or against (L-Line) performance in NLWEJ. Methods The dataset included up to 6779 reproduction phenotypes recorded on 1836 ewes linked to a pedigree containing 10,903 identities. The data were analysed with a repeated measures linear mixed model using ASREML V4.2 software. Genetic trends were derived by regressing breeding values on year within each line. Results Heritability estimates were low, with the highest value observed for NLBEL (0.12 ± 0.02). All traits were affected by an animal permanent environmental effect from 0.05 for ERAEL to 0.12 for ELEJ. All traits diverged between the H- and L-Lines, with an upward rate of annual gain of nearly 1% for the composite trait NLWEJ, and positive but smaller gains of 0.3% for ELEJ, 0.45% for NLBEL and 0.15% for ERAEL. It was reassuring to see that the trend for ERAEL was stable despite the increase in NLBEL for H-Line ewes and the unfavourable genetic correlation (rg = −0.57 ± 0.14) between these two traits. Conclusions In a well designed breeding program, the use of NLWEJ as selection criteria can be effective to breed ewes with a high output at weaning. Use of a composite trait might not deliver optimal genetic gains for the components ELEJ and ERAEL, however, and could favour NLBEL, which might not be desirable in all environments. Implications For commercial flocks, favourable results will depend on complete recording and correct processing of information from all ewes mated and lambs born. Otherwise, NLWEJ only reflects NLBEL, where the predicted response in ERAEL is unfavourable.