Modelling Plant Breeding Programs: Applications to Forage Crops

Abstract

Breeding programs operate to bring about changes in the combinations of genes within individuals and populations. The ease with which this is achieved depends on the complexity of the genetic control of variation for the target traits. Molecular biology has introduced new tools to facilitate this process. Quantitative genetics theory provides a framework for modelling the effects of genes within populations. This often requires a number of simplifying assumptions (e.g. no epistasis, many genes each with relatively small and similar effects, linkage equilibrium in the reference population). Using this approach, prediction equations of expected response to selection have been constructed for a range of genetic models and breeding strategies. Molecular tools have progressed our understanding to the stage where we are now able to develop specific genetic models to describe the architecture of some target traits. In some cases this suggests a need to relax a number of the simplifying assumptions we have previously made in the theoretical models. Removing these assumptions complicates the construction of prediction equations. To complement the classical quantitative modelling and prediction approaches used in genetics, computer simulation models of plant breeding programs can be developed to capture in silico many of the features that are specific to a breeding program and the genetic variation with which they work. With the progress that has been made in computer hardware and software, these simulation-based models enable quantification of the efficiency of breeding strategies for genetic models that range from simple to complex. An example evaluating the efficiency of marker assisted selection strategies for a half-sib recurrent selection breeding strategy is discussed.