Heat stress during the reproductive and grain-filling stages leads to severe yield losses in rice, especially in irrigated areas during the dry season in Thailand. Thus, breeding for heat-tolerant rice is one of the strategies for developing rice crops adapted to the effects of high temperature in major growing regions. M9962, a fast neutron-induced mutant, was identified as a heat-tolerant line that was then used as a donor parent and crossed with the Phisanulok 2 (PSL2) cultivar (the recurrent parent) in a backcross breeding programme. Heat tolerance scoring under controlled greenhouse and field conditions was based on the percentage of spikelet fertility. Yield trials for the BC2F7 and BC3F6 lines were subsequently conducted under field conditions at three locations and under controlled greenhouse conditions during the dry season of 2019 (DS2019). In the early generation, 210 plants of the BC1F1 population and 440 plants of BC2F1 population were evaluated for their heat tolerance, and the spikelet fertility distribution within the BC1F1, BC2F1 and BC3F1 populations showed a skewed-right pattern. In addition, most rice plants in every generation were heat sensitive. Finally, compared with PSL2, the four-candidate heat-tolerant lines exhibited a high percentage of spikelet fertility and high yield. In addition, the one candidate line, BC2F7-6-5-4-1-1-21 was most similar to PSL2 in terms of genetic background, plant type and grain quality. Thus, backcross breeding programmes via the spikelet fertility can be used as an indirect trait to select for heat tolerance.

The reproductive stage of rice is the most sensitive to heat stress, which can lead to spikelet sterility. Thus, heat-tolerant and heat-susceptible genotypes were used to investigate their differences in terms of phenotypic responses and expression changes of Hsf genes at the pre-flowering stage under heat stress. Results clearly showed that panicles had the highest temperature compared with other plant parts under both natural and heated conditions. However, the temperatures of tolerant and susceptible genotypes were not significantly different. In terms of spikelet fertility, the tolerant lines M9962 and M7988 had high seed set because their anther dehiscence, pollen viability and pollen germination were only slightly affected. In contrast, the susceptible line Sinlek showed severe effects at all steps of fertilization, and the pollen viability of M7766 was slightly affected under heat stress but was more affected in terms of anther dehiscence and pollen germination. Both susceptible lines showed dramatically decreased seed set. In addition, the expression of six HsfA genes in the flag leaves and spikelets at the R2 stage of plants under heat stress showed different responses. Notably, expression of the HsfA2a gene was predominantly upregulated in the flag leaf and spikelets under heat stress in M9962. Therefore, it can be concluded that heat stress has severe effects on the stamen, and that different genotypes have different susceptibilities to heat stress.

To maximize heterosis, it is important to understand the genetic diversity of germplasm and associate useful phenotypic traits such as fertility restoration for hybrid rice breeding. The objectives of the present study were to characterize genetic diversity within a set of rice germplasm groups using coefficient of parentage (COP) values and 58 simple sequence repeat (SSR) markers for 124 genotypes having different attributes such as resistance/tolerance to various biotic and abiotic stresses. These lines were also used for identifying prospective restorers and maintainers for wild abortive-cytoplasmic male sterile (CMS) line. The mean COP value for all the lines was 0.11, indicating that the genotypes do not share common ancestry. The SSR analysis generated a total of 268 alleles with an average of 4.62 alleles per locus. The mean polymorphism information content value was 0.53, indicating that the markers selected were highly polymorphic. Grouping based on COP analysis revealed three major clusters pertaining to the indica, tropical japonica and japonica lines. A similar grouping pattern with some variation was also observed for the SSR markers. Fertility restoration phenotype based on the test cross of the 124 genotypes with a CMS line helped identify 23 maintainers, 58 restorers and 43 genotypes as either partial maintainers or partial restorers. This study demonstrates that COP analysis along with molecular marker analysis might encourage better organization of germplasm diversity and its use in hybrid rice breeding. Potential restorers identified in the study can be used for breeding high-yielding stress-tolerant medium-duration rice hybrids, while maintainers would prove useful for developing new rice CMS lines.