A study of genetic and environmental variation and covariation in productive traits of a flock of Perendale sheep : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Animal Science at Massey University

Performance records on 1724 two-year, 1160 three-year, 726 four-year and 449 five-year Perendale sheep from the Massey University flock were analysed. These records were collected between birth-years 1961 and 1974 inclusive, and therefore different numbers of years were involved at each age. The data comprised 97, 82, 65 and 56 sire groups for the respective ages. There were 403 ewes in 56 sire groups which had four consecutive records between two- and five-years of age. There were 922 two-year, 584 three-year, 296 four-year and 125 five-year dam-daughter pairs, but the last age group of records was considered too few to justify analysis. The dependent variables examined were quality number (QN), wool character (WC), staple length (SL), greasy fleece weight (GFW), fibre diameter (FD), and number of lambs weaned per ewe joined and present at lambing (NLW). Fibre diameter had been recorded in fewer years than the other traits and there were consequently less [i.e. fewer] observations available. The effects of the environmental factors of year, age of the dam of the ewe, rearing rank, NLW (wool traits only), and all interactions among them were assessed by computing the percentage of the total variance contributed by each factor for each variable. Quadratic components for these fixed effects were temporarily considered to be variance components for this purpose. Year effects were the most important source of enviromental variation in wool characteristics, but were less important for NLW. None of the other main effects were consistently important (i.e. contributed greater than two percent of the total variation) in explaining the observed variability. Year x NLW was the only interaction which contributed more than two percent of the total variance in any wool trait. The affected characteristics were two- and three-year QN, three-year FD and five-year SL. NLW (five-year-old ewes only) was affected by dam age x rearing rank and year x rearing rank interaction. Heritability estimates calculated by paternal half-sib method (daughter-dam regression estimates in brackets) were in the range of 0.22 - 0.28 (0.39 - 0.66), 0.16 - 0.39 (0.12 - 0.25), 0.32 - 0.71 (0.25 - 0.36), 0.24 - 0.53 (0.43 - 0.62), 0.38 - 0.68 (0.45 - 0.50) and 0.02 - 0.24 (-0.02 - 0.03) for QN, WC, SL, GFW, FD and NLW respectively. Differences between ages were seldom significant and the relevance of such a comparison was discussed. Paternal half-sib heritability estimates for the average of four mature (two- to five-years) production records were 0.50, 0.61, 0.84, 0.44, 0.49 and -0.01 for the same respective characteristics. Binomial analysis methods resulted in heritability estimates for fertility (twinning in brackets) of 0.06 (0.12), 0.02 (0.15), -0.01 (0.13), and -0.15 (0.01 for two-, three-, four- and five-year-old ewes respectively. Repeatability was estimated by averaging the regression of later on earlier record for all pairs of ages for wool traits and mature ages for reproductive characteristics and resulted in values of 0.46 (QN), 0.18 (WC), 0.50 (SL), 0.62 (GFW), 0.71 (FD), and 0.09 (NLW). Similar methods gave computed values for the repeatability of the difference between zero and one lamb weaned of 0.01, and between one and two lambs weaned of 0.13. Genetic and phenotypic correlations between all variables were calculated separately at each age by paternal half-sib and daughter-dam methods. The correlations were positive for GFW - FD, GFW - SL, GFW - WC, QN - WC and FD - SL, negative for GFW - QN, QN - FD and QN - SL, and mixed for FD - WC and SL - WC although the last of these were mostly positive. The phenotypic correlations between wool traits and reproduction were negative for NLW - GFW and NLW - SL and approximately zero for the remainder. Genetic correlations were negative for NLW - GFW, NLW - FD and NLW - SL, positive for NLW - QN and mainly positive for NLW - WC. Many of the genetic correlations involving NLW had large standard errors. Application of the results was discussed in terms of the selection indices which were constructed