Abstract/Summary

Faba bean (Vicia faba L.) is a nitrogen-fixing, high protein grain legume crop with high yield potential in temperate maritime climates. Across Europe as a whole, the small size of the faba bean seed market supports relatively small numbers of commercial breeding programmes. The overarching aim of this thesis was to design and validate a deliberately small-scale, low cost faba bean breeding programme suited to the purpose of obtaining faba bean varieties tailored to the uniquely mild, high-rainfall Irish climate. A Modified Recurrent Selection Scheme (MRSS) was established in two sites (Reading, UK and Oakpark, Ireland) based on principles of extreme simplicity – after the creation of the founding pool (219 F1 hybrid combinations), outcrossing was left entirely to bee-assisted pollination and selection was based on one simple-to-measure trait – individual plant seed yield. In each selective generation the top c. 4% highest yielding plants was advanced to winter bulking. In the fourth year of the project, co-ordinated replicated yield plot trials were conducted at both sites to compare yield performance of the foundation bulk, the penultimate and latest selections from both sites together with the founding spring varieties and a newer additional spring variety ‘Lynx’ as checks. Selection on individual plant seed yield in Reading had in effect achieved a total gain in yield of 16.5% compared to the founding bulk, while the equivalent parallel selection at Oakpark was 9.6% higher yielding than the founders. The increase in yield was accompanied by a substantial increase in seed weight. Stochastic simulations were carried out using the R package AlphaSim in which the main breeding parameters – Population Size (PS) and Selection Intensity (SI) – were varied in factorial combinations with the main unknown parameter (effective outcrossing rate ‘OR’) to find the optimal and most cost-effective combinations. The highest simulated SI invariably produced higher initial rates of gain (e.g. <5 years), though in lower PS, the rapid elimination of variation through sharp SI limited the long-term gain. High OR diminished gain in the short term but gave substantially higher long-term gain in all scenarios and weakened the effect of increasing SI as well as delaying the crossover point where low SI overtook high SI in cumulative genetic gain. Since SI can be varied for free but cost grows almost linearly with PS, the most cost-effective scenarios involved low PS with high SI with the caveat that over the medium to long term, either bringing in new genetic material to replenish diversity or attempts to lift OR would be required to sustain high rates of gain. Finally, genomic responses to selection were investigated in the Reading selections by genome-wide genotyping of the founders, all selected individuals and a sample of unselected individuals from each selective generation. The total number of alleles present diminished in each selection cycle, as simulations predicted, by c.5% per cycle. In contrast, after an initial drop from the founding generation, heterozygosity trended upwards in all subsequent selections. Comparisons of selected and unselected cohorts in each generation showed a strong tendency for highly heterozygous individuals to be selected, suggesting that hybrid vigour contributes strongly to individual seed yield performance raising the concern that heterosis would therefore mask true additive variation. In order to look for signals of directional selection at particular loci, characteristic of underlying additive variation for the trait under selection, an FST scan was conducted, resulting in the detection of numerous loci that appeared to have undergone convincing directional selection.