An application of Bayesian techniques to the genetic evaluation of growth traits in Eucalyptus globulus

A Bayesian procedure coupled with Gibbs sampling was implemented to obtain inferences about genetic parameters and breeding values for height and diam eter of 7-year-old Eucalyptus globulus Labill, is described. The data set c onsisted of 21 708 trees from 260 open-pollinated families taken from 10 di fferent Australian provenances, from one Spanish population, and from two c lones. The trees are distributed over eight sites in the south of Spain, wi th 20 blocks per site. Data were corrected for heterogeneity of phenotypic Variances between blocks. In the analysis, a self-pollination rate of 30% f or the open-pollinated families is assumed in the relationship matrix. The posterior means (and standard deviations) of the heritabilities of height a nd diameter and the genetic and phenotypic correlation were 0.217 (0.014), 0.128 (0.084), 0.768 (0.028), and 0.799 (0.003). Results from the standard restricted maximum likelihood method were 0.173, 0.113, 0.759, and 0.798, r espectively. Most of the discrepancy in heritability estimates from both me thods can be attributed to the adjustement of residual maximum likelihood e stimates to the assumed self-pollination rate, which ignores the presence o f clones in the trial. The effect of the method of prediction of breeding v alues (best linear unbiased prediction or Bayesian techniques) on the genet ic superiority of the selected trees was not important. Differences in bree ding value among provenances and among families were evidenced for both tra its.