Published online by Cambridge University Press: 27 March 2009
Random F3 families from six crosses typical of those being made in a spring barley breeding programme at the Scottish Crop Research Institute were used to rank the crosses on the basis of their predicted probabilities of exceeding a given standard for a number of characters. The predicted rankings were then compared with the observed rankings of the crosses based on observations of random inbred lines from the same crosses. The agreement between predicted and observed rankings was very good for single characters and good for bivariate combinations of the characters. The agreement for the bivariate combination of height and single plant yield was exact and the combination also appeared to be capable of accurately ranking crosses on the basis of height and plot yield.
The practical considerations of using such a cross prediction technique in a cereal breeding programme are discussed. Results from this experiment suggest that single plant measurement could be replaced by measurements on progeny rows, thereby saving considerably on labour. Predictions could just be made on the most important characters in the breeder's model for the early identification of the best crosses but this is still an advance on the approach of making many crosses and assessing relatively few selections from each cross in yield trials.