Stochastic Simulation of Genetic Advancement of Multiple Traits Using Sequential Culling

A software tool was developed for the stochastic simulation of multiple traits and uses sequential culling for the different traits. It selects superior progeny and, at present, can calculate genetic gain from the backward selection of parent clones based on progeny performance, crossing those parent clones, and forwards selection among the progeny. Key information input to the program includes the base population size, number of parents selected, economic weights for different traits, genetic correlations between traits, mating design and number of crosses, number of progeny planted and selected, and proportion culled within the progeny for each trait. An application of the software, with particular relevance to the Pinus radiata D.Don breeding population in New Zealand, deals with the development of breeds for specific products and for specific site types. Gains obtained at age 8 from different selection emphases were compared, for an unspecialised main population of 300 parents and four specialised breeds of 24 parents each. The parents came from a base population of 2000 progeny-tested parents. The simulation covered forwards selection in the progeny obtained by crossing the selected parents. The breeds were designated Structural Timber, Growth and Form Elite, Long Internode, and Fibre, and reflected different end-product objectives. Six traits were involved—namely dbh, stem straightness score, needle retention score, branch cluster frequency score, spiral-grain angle, and wood density. The simulation provided support for differentiating a Structural Timber breed (with a strongly multinodal branching habit, higher wood density, and reduced spiral grain angle). As well, we were able to simulate breeds similar to the existing Long Internode breed and a Growth and Form Elite population. The software can also be used to study the effect of varying population size.