DOTTORATO DI RICERCA

Rice production account for more than 50% of global irrigation water, thus, reducing water use is becoming a priority, mostly related to increase in surface air temperature and reduction of water availability due to the global warming. A solution would be to breed rice accessions adapted to aerobic water management system, which is characterized by periodic drying and re-flooding of rice field and allow reduction of water demand. However, since yield is penalized in most European temperate japonica rice when grown under aerobic conditions, it is important to maintain a sustainable level of yield while optimizing water use. To achieve this goal, genomic selection approaches were used to evaluate the feasibility of genome wide selection for the identification of rice breeding lines with tolerance to water scarcity and related genome wide-based selection tools. A training population of japonica rice composed of 283 rice accessions and a population of 97 F5-F7 progenies derived from 36 bi-tri-parental crosses between elite lines that belong to the training population were phenotyped in irrigated and aerobic conditions and subjected to genotyping by sequencing. The accuracies, as obtained from correlations between the Genome Estimated Breeding Values (GEBV) and True Breeding Values (TBVs) were evaluated for three agronomically relevant traits using cross validation in the training population panel and across generations using phenotypic and genotypic data of the progeny population. Moreover, also the effects on the accuracies of different levels of Minor allele frequency, linkage disequilibrium, prediction models and, for the evaluations across generations, of scenarios involving different set of the training panel, were evaluated. High levels of accuracies with both the procedures (cross validation and across generations) were achieved even for complex agronomic traits like panicle weight, flowering date and nitrogen index. Results permitted to assess the feasibility of genomic selection across generation in rice population and highlighted a group of progenies that can be exploited in the breeding for tolerance to water scarcity. Then, using phenotypic data obtained under two contrasted irrigation system, irrigated (I) and Aerobic (A), in a reference population (RP) and in a progeny population (PP) we tested two set of approaches for predicting response to aerobic system. The first approaches were based on response index and regression analysis, the seconds on explicit modelling of marker by environment interaction. Rank-correlation between the performances of the individual entries of the two population was high for flowering time and panicle weight traits, indicating rather limited level of GxE interactions. Accuracy of genomic predictions were much higher when GxE interactions were modelled explicitly. In the second part of the PhD Thesis, the economic performances of different water management systems, including the role of availability of rice varieties adapted to each management system, were evaluated and a multi-objective model was implemented to explore economic and water saving at the farm level. Results showed that the advantages of the aerobic method adoption depend on production costs, irrigation water cost and efficiency of selected rice varieties in different rice ecosystems.