In this thesis opportunities for prediction of cross performance in a plant breeding program are investigated. For this research 20 SSD-line populations from crosses between European two-row spring barley lines were evaluated for four quantitative agronomic traits in seven environments, divided over two years.
The midparent value appeared to be a good predictor of average offspring performance and useful in practical breeding. However, for most crosses the midparent values for grain yield overestimated the offspring average. The relatedness between parents was expected to predict the variance among the offspring. However, predictions using genetic distances based on pedigree data, morphological trait data, and AFLP-marker data, performed poorly. A genetic distance based on AFLP-markers associated with the trait variation among the parents gave a somewhat better prediction.
The correlation between the parental responses to different environments appeared to be a reasonable predictor of grain yield among the offspring. Other variance predictors based on parental differences for agronomic trait data or early generation (F4) variance among the offspring, mainly predicted variance resulting from segregating major genes. These genes are often fixed in practical breeding programmes and therefore not very relevant. Grain yield data from small three-row plots in an early generation evaluation did not correspond with large plot yield data due to interplot competition.
|Qualification||Doctor of Philosophy|
|Award date||10 Nov 1998|
|Place of Publication||S.l.|
|Publication status||Published - 1998|
- hordeum vulgare
- genetic variation
- yield forecasting
- plant breeding