| Crown rust is the most devastating and widespread disease on oats, causing losses in grain yield and grain quality. Race non-specific partial resistance, characterized by a low amount of disease with a susceptible infection type, is considered to be a more durable form of resistance, compared to race specific resistance. The objectives of this thesis were to evaluate the efficiency of recurrent selection for partial resistance to oat crown rust, and to determine indirect changes associated with selection in two recurrent selection populations derived from the same base population. The closed system-rust population and the three-way cross system population were selected during six and four recurrent selection cycles, respectively. The closed system-rust population was maintained as a closed population, while in the three-way cross system four oat cultivars were introgressed. In the closed system-rust population selection was effective only from cycles zero to three, resulting in an average selection gain of 6% per cycle, measured as area under the disease progress curve (AUDPC). The three-way cross system produced an average selection gain of 5.4% per cycle, and selection was effective from cycles two to four. Indirect selection of later and taller plant was found in both populations. Path coefficient analysis revealed that later heading dates were associated with greater partial resistance, especially in the closed system, mainly because of an indirect effect via longer time to reach 33% of maximum severity (T33-maxsev), which tended to reduce the maximum severity. In the three-way cross system later heading dates also resulted in longer epidemic durations, which leaded to an increase in the amount of disease in the end of the epidemic. Plant height seems not to be associated with partial resistance. Broad sense heritabilities were estimated for all traits and selection cycles, but they were not appropriate indicators of selection efficiency. Selection of partial resistant genotypes should avoid very late flowering genotypes, and selection should be based on both AUDPC and maximum severity values. Path analysis proved to be a very useful tool to understand how recurrent selection for partial resistance to crown rust operated in the studied populations. |
