Empowering breeding programs with new approaches to overcome constraints for selecting superior quality traits of rice

<strong>Empowering breeding programs with new approaches to overcome constraints for selecting superior quality traits of rice</strong> Mariafe N. Calingacion Most rice breeding programs have focused on improving agronomic traits such as yield, while enhancing grain quality traits such as flavour and aroma, especially of non-fragrant rices, has not been given high priority. In this study, we utilised a multi-disciplinary approach to understand better quality traits of aroma and flavour in rice grains, and to determine whether good flavour in the grain could be combined with stress tolerant genotypes. To understand what factors drive rice preferences, an extensive survey among members of the International Network for Quality Rice who are local experts in grain quality evaluation programs in 25 countries was conducted (Chapter 2). The objective was to identify the grain quality characteristics of the popular rice varieties in each region. Eighteen combinations of size and shape of the grain, amylose content (AC), gelatinisation temperature (GT) and fragrance were identified. These trait combinations reveal the complexity of consumer preferences. The two most popular combinations both have long and slender grains, while one has low amylose, low GT and is aromatic, and the other has intermediate AC and intermediate GT and is non-aromatic. Further evaluation of varieties having the same combination of grain quality traits showed that consumers readily identify differences between these varieties. For example, BRS Primavera and IR64 that are popular in Brazil and in the Philippines, respectively, have the same combination of all 18 traits, however, panellists of sensory evaluation can easily perceive differences in aroma and flavour of BRS Primavera and IR64. This emphasises that the current tools we have available to assess rice quality are unable to capture all the quality traits consumers are looking for in rice. In Chapter 3, a novel multiplatform metabolomic and ionomic approach with genome-wide genotyping was utilised to investigate the effect of different nitrogen fertiliser regimes on the biochemical profile of three premium waxy rice varieties, Hom Nang Nouane (HNN), Kai Noi Leuanag (KNL) and Tha Sa No (TSN) from Lao PDR. The current tools used to phenotype grain quality such as GT, values from viscosity curves, and hardness and stickiness, were unable to differentiate between HNN, KNL and TSN either on the basis of nitrogen treatment nor genotype. However, metabolite profiling of metabolites and minerals followed by multivariate statistical methods readily separated the genotypes on each platform, and discriminatory compounds that were identified were relevant to consumers in terms of flavour, taste and nutrition. However, despite yield differences, nitrogen treatment did not significantly affect the overall metabolite and mineral profiles of the samples. Using 1536 single nucleotide polymorphism (SNP) loci, the Euclidean distance between each variety was calculated and compared to the distance between each variety for each metabolomic platform. Procrustes analysis was used to rotate and scale the variety mean scores on the metabolite principal components to give the best fit to the genetic principal coordinates. Comparing the triangles whereby each vertex of the triangle is a variety and the length of each side is equal to the scaled Euclidean distance, mineral elements, polar metabolites and volatile compounds all associate very well with the genetic distance between each variety. This study highlights that multiple metabolomic platforms are potential phenotyping tools to characterise rice quality in a comprehensive and efficient way, and in a way that provides data that is relevant to consumers. To gain insights on the influence of water availability to the metabolomic profile of drought tolerant rice, two contrasting varieties, Apo and IR64, and a mapping population derived from them were extensively characterised in Chapters 4 and 5. Apo is drought tolerant but has unacceptable grain quality while IR64 is drought susceptible with premium grain quality. Apo and IR64 were grown under irrigated and drought conditions. Yield of Apo from both water conditions was higher than yield of IR64 under the same conditions. Moreover, metabolite profiling and sensory analysis showed that grains of Apo were not affected by drought conditions i.e. panellists perceived no difference in the aroma of Apo from both conditions and Principal Components Analysis (PCA) of the volatiles showed one cluster of Apo from both conditions. However, grains of IR64 formed two clusters based on water condition in the PCA and panellists were able to perceive ‘water-like metallic’ aroma in IR64 that was grown under drought conditions but this was not detected in grains from the irrigated treatment. This suggests that response to water stress in the metabolomic profile of the grain is variety dependent. In Chapter 5, a mapping population derived from Apo and IR64 was grown, with the parents, under irrigated and drought conditions. The yield of more than half of the population was higher than the yield of Apo and IR64 under both irrigated and drought conditions; this indicates significant transgressive segregation. Using a dense linkage map based on genotyping by sequencing data, quantitative trait loci (QTL) analysis of drought stress identified one major QTL on chromosome 3 that is likely to be qDTY3.1 which was previously detected in a population derived from Apo as the drought tolerant parent. All the lines of the population carrying this QTL showed significantly higher yield under drought than those without it, indicating the potential importance of this QTL in drought tolerance. Metabolite profiling and sensory analysis were also conducted in the grains of the population. More than a hundred volatiles were detected in the headspace of rice samples and PC1 and PC2 explained 55.6% of the variation in the metabolite profiles with many of the lines clustering in between the Apo and IR64 parent values. Six novel metabolite QTLs for volatile compounds were identified - 1 QTL was detected in chromosome 1 for 3,7-dimethyl-octen-1-ol, 1 QTL for hexanol in chromosome 2, and 4 QTLs for pentanol, hexanol, hexanal, and heptanone in chromosome 3. Interestingly, three lines were observed by the panellists to have similar aroma as IR64 while four lines were observed to have similar aroma as that perceived in Apo. Lines 20, 164 and 28 were perceived by the panellists to have high levels of corn, dairy and sweet aromatic features. Moreover, the yield of these 3 lines under both irrigated and drought conditions was similar to that of the Apo parent under the same conditions with Line 28 yielding the highest under drought and has the QTL associated with yield under drought on chromosome 3. Finally, the potential of metabolomics as a phenotyping tool in characterising grain quality is further highlighted in Chapter 6. Combining metabolomics with high throughput genotyping and sensory analysis offers new breadth of approach in understanding grain quality of rice. Three lines identified that carry IR64 quality along with high yield in both irrigation and drought, are recommended to enter a rice breeding program at the stage of advanced replicated and multi-location testing. By using advanced tools of phenotyping and genotyping, with validation by sensory panels, these three advanced lines have been selected in just three years.