Applicability of best linear unbiased prediction (BLUP) for the selection of ortets in Eucalyptus hybrid populations

Abstract

The prediction of the performance of hybrid plants as production material is complicated by the genetic structure of the hybrid population. Fundamental quantitative genetic concepts are defined with respect to disomic, randomly fertilized populations which are in linkage and Hardy-Weinberg equilibrium. Hybrid populations of crosses are, however, in disequilibrium and do not meet these assumptions of quantitative genetic theory. The theory of inheritance in hybrid populations has not been fully developed and the complex models that have been developed have not reached a stage where they have been implemented to adapt selection practices for hybrid populations. Therefore in practice hybrids are often selected using selection methodology that has been developed for pure species despite large differences in the underlying genetic properties of the populations. It is not known to which extent such selections are able to predict the performance of progeny or clones in hybrid populations. This study was based on historical data and investigated the efficiency of BLUP in hybrid populations. As a case study the selection of ortets in three hybrid Eucalyptus populations was investigated. Ortet selection in E. grandis × E. saligna, E. grandis × E. urophyllaand E. grandis × E. camaldulensis populations was compared with selection in E. grandis populations. Clonal performance was predicted from a series of hybrid family trials whereas realised clonal performance was assessed in clonal trials. The predicted and realised clonal performances were correlated to assess the efficiency of the predictions. A series of constructed indices was created that placed a range of weights on family versus individual tree performances to obtain an indication of the range of correlations that could be expected from the data. Different BLUP methods were compared including selection for breeding and various forward selections for clonal forestry (“clonal forward selection”). The clonal forward selections were based on different estimates of the proportion of total genetic variance that is ascribed to non-additive genetic variance. The results of the study indicated that there were no observable differences in the efficiency of the BLUP predictions in the studied hybrids in comparison to prediction in E. grandis There was, however, a decrease in the correlations between predicted and realised clonal performances with increasing genetic distance between the parents. Furthermore, the genetic values were better predictors of clonal performance than breeding values and genetic values which were based on higher assumptions of non-additive genetic variances (37% of total genetic variance) were more efficient than those based on assumptions of lower proportions (20%). This study indicates that BLUP methods may be efficient in predicting the clonal performance in the three hybrid populations that were investigated. Copyright