Predicting the properties of recombinant inbred lines derived by single seed descent

Abstract

The genetical components of family means and variances, simply and reliably estimated in the first few generations of any breeding programme initiated from a cross between two inbred lines, are all that is required to predict the distribution of the inbred lines that can be derived by single seed descent from the F2 of such a cross. We can, therefore, determine the probability of obtaining inbreds that fall outside of the parental range or of exceeding the F1 if it shows heterosis, by any amount we care to specify. While epistasis, genotype × environmental interactions and linkage may at first sight appear to seriously disturb any such prediction they can all be accommodated. In practice only epistasis is expected and found to produce disturbances of a sufficient magnitude to justify making allowance for its presence in making these predictions. This is because it leads to asymmetry in the distribution of derived inbreds relative to the initial inbred parental means. The predictive power of this new approach is demonstrated on two complete breeding cycles, each initiated by a cross between two pure-breeding varieties of Nicotiana rustica and culminating in the production of over 100 recombinant, pure-breeding lines. The ability of the analysis to discriminate between two crosses on the basis of their probabilities of producing inbred lines that fall outside of their parental range is also demonstrated. By using analyses of this kind there is no reason why we need ever go beyond the F2 of an inbreeding programme without a fairly clear idea of the outcome.