Drought stress frequently reduces grain yield of rainfed lowland rice in Northeast Thailand. Development of drought resistant cultivars suitable for this region, so as to minimise yield loss, has long been an objective of the national breeding program, but little progress has been made over the last 25 years. The large genotype-by- environment (G X E) interactions for grain yield due to the highly variable nature of the drought characteristics in the region have contributed to the difficulty of developing drought resistant cultivars. The contributions of morpho-physiological traits that putatively confer drought resistance to grain yield have not been quantified. Thus, targeted incorporation of specific adaptive traits into breeding materials has not been a feasible breeding strategy. The objectives of this thesis were: (1) to examine genotypic variation for grain yield and to assess the magnitude of responses of genotypes for grain yield under various types of drought stress conditions in target drought prone areas in Northeast Thailand, (2) to investigate genotypic expression of putative drought resistant traits and their contributions to grain yield under drought stress conditions, and (3) to evaluate the use of the dry season screening for drought resistance to estimate grain yield under drought stress in the wet season. The goal of the present study is to utilise information achieved from the study to integrate into the modified breeding strategy that has been recently proposed by a project funded by the Australian Centre for International Agricultural Research (ACIAR), so that selection efficiency for drought resistant genotypes can be improved.
Four sets of ten field experiments were conducted under lowland conditions in the upper and lower parts of Northeast Thailand. The first three sets of experiments conducted in the wet season were used to investigate genotypic variation for grain yield and putative drought resistant traits in the wet season. To increase the chance of the drought stress, experiments were seeded later than the normal seeding time practiced in this region. Randomly sampled sets of 50 to 128 recombinant inbred lines derived from 4 biparental crosses were used for all experiments. Different types of drought stress were imposed in each set of experiments, i.e. mild drought stress during the grain filling period in experiment 3, severe drought stress just before flowering in experiment 5, prolonged severe drought stress during the reproductive to the grain filling stages in experiment 7, and prolonged mild drought stress during the vegetative and the grain filling stages in experiment 8. Grain yield of each genotype under the drought stress conditions in those four experiments was adjusted using potential grain yield and flowering time under well-watered conditions to determine a drought response index (DRI). The fourth set of two experiments were conducted over two dry seasons to examine genotypic variation for drought score (leaf death score) and investigate the reliability of using the drought score measured in the dry season to estimate grain yield measured under drought stress conditions in the wet season.
A large genotypic variation existed for grain yield under both irrigated and drought stress conditions. Depending on timing, duration and severity of plant water deficit and seeding time, grain yield of genotypes under drought stress, in relation to that under irrigated conditions, was reduced by 18% in experiment 3, 55% in experiment 5, 81% in experiment 7 and 52% in experiment 8. In each drought stress condition, the DRI described the magnitude of the response of genotypes for grain yield and identified drought resistant genotypes. The DRI of the genotypes was inconsistent across the four drought stress environments in experiments 3, 5, 7 and 8, indicating that genotypes respond differently with the change of the characteristics of the drought stress conditions.
Genotypes that were able to maintain high panicle water potential (PWP) during the drought stress period that developed just before flowering time (experiment 5) produced higher grain yield, harvest index, filled grain percentage, fertile panicle number per m2 and fertile panicle percentage and had higher DRI. Drought resistant genotypes maintained significantly higher PWP than susceptible genotypes (-1.83+0.016 vs. -1.97±0.043; P < 0.01), and produced more fertile panicles (74.3±3.7% vs. 34.3± 8.7%; P = P < 0.01), filled grains (58.7±2.5% vs. 31.8±6.3%; P = P < 0.01) and grain yield (2.40±0.10 vs. 1.02±0.08 t ha-1; P < 0.01). Genotypes with a larger amount of the above ground total dry matter at anthesis (TDMa), tall plants and those with larger root systems depleted soil water more rapidly than genotypes with smaller TDMa, shorter plants and those with smaller root systems and resulted in the larger decrease in PWP. Other drought-related traits in genotypes used in the study, e.g. root systems, canopy temperature and osmotic adjustment, have shown a small variation and their contributions to grain yield under drought stress were not significant. These may reflect similarity of such traits in their parents.
Drought stress delayed the flowering time of genotypes on average 3±0.22 days (experiment 5) and 10±6.3 days (experiment 7). In both experiments 5 and 7, a larger delay in flowering time was associated with greater reduction in grain yield (r = - 0.431**, r = -0.707**), harvest index (r = -0.336*, r = -0.653**), fertile panicle percentage (r = -0.399**) and filled grain percentage (r = -0.354*, r = -0.759**). In experiment 5, delay in flowering time was strongly associated with low PWP (r = -0.387**), hence genotypes with larger delay in flowering time suffered more drought stress since they flowered when available soil water was lower.
Larger drought score determined in the dry season was associated with lower grain yield under drought stress conditions in the wet season experiments, but the association was significant only in particular conditions, i.e. when patterns and severity of drought stress in both environments were similar. The dry season screening conditions for drought score screen should be managed to correspond to relevant types of drought conditions, i.e. mild or severe drought stress, in the wet season in the target areas.
Recommendations to breeding program for the Northeast Thailand were made based on the study in this thesis that selection for high grain yield under late season drought stress conditions should be conducted in the drought conditions relevant for the target areas. Selection of genotypes for high PWP, minimum flowering delay with high potential yield under drought stress conditions should be practised at the same time as for yield testing. Early generation breeding materials should be selected for low drought score in the dry season before high grain yield genotypes are selected under drought stress conditions in the wet season. Future research should be conducted to characterize and identify the key types of drought conditions in Northeast Thailand, evaluate the consistency and usefulness of panicle water potential, and identify factors and mechanisms controlling flowering time of genotypes under drought stress conditions.