**4. Discussion**

Rice is one of the most important staple food crops globally, being consumed by approximately 50% of the global population [33]. Low-temperature germinability (LTG) is one of the major factors influencing stable crop establishment in the direct seeding method of rice cultivation in tropical and subtropical regions of the world. Along with rapid low-temperature germination, vigorous coleoptile growth is essential in the direct-seeding method of rice when rice seeds are sown in flooded paddy fields and watered with cold irrigation water [14]. Rapid coleoptile elongation after germination is necessary to improve seedling establishment rate [6].

In the present study, we analyzed LTG and coleoptile length in rice cultivated under 13 ◦C conditions. Using the F2 population derived from a cross between two introgression lines, CR1517 and CR1518, we detected a total of five and three QTLs for LTG and coleoptile length, respectively, over two scoring dates. QTLs associated with LTG were detected on chromosomes 1, 3, 8, and 10. Among them, two major QTLs, *qLTG1* and *qLTG3*, were detected on chromosomes 1 and 3, respectively. One minor QTL for coleoptile elongation, *qLTG8*, together with *qCCL8*, was detected between RM72 and RM22705 markers on chromosome 8. We failed to detect the QTL using 96 introgression lines in a previous study [27], which could be due to a masking e ffect of the major QTLs (*qLTG1* and *qLTG3*) and the genetic structure of the population [27]. Several studies have reported the presence of QTLs for LTG and coleoptile elongation on chromosome 8 [5,6,18,34]. For example, Najeeb et al. (2020) detected an LTG QTL, *qLTG(III)8* (497SNP\_8\_8509144), which is located near *qLTG8* in the present study [34]. The results sugges<sup>t</sup> that the region on chromosome 8 participates in the regulation of germination and coleoptile elongation under low-temperature conditions. Identification of such genes from diverse varieties could enhance our understanding of the roles of the region in germination activities.

Low-temperature germination QTLs are distributed widely throughout the rice genome, including on chromosome 10 [12,17]. We detected two linked QTLs, *qLTG10.1* and *qLTG10.2* for LTG, with an LTG-increasing allele originating from *O. rufipogon* at *qLTG10.1* and Hwaseong at *qLTG10.2*. The results were confirmed in the near-isogenic background in the F3 population. Notably, the locations of *qLTG10.1* and *qLTG10.2* (20.0~22.9 Mb region) are similar to that of *qGR-10* for low-temperature germination ability detected between C809 and C797 on chromosome 10 (21.0~22.0 Mb region) in a study by Ji et al. (2009) [17]. They mapped *qGR10* using a recombinant inbred line population derived from a cross between Asominori and IR24. At *qGR-10*, the IR24 allele increased germination rate at two scoring dates (8 and 9 DAI), whereas the Asominori allele increased germination rate at later scoring dates (10 and 14 DAI). Similar results have been observed, where the beneficial alleles for LTG originate from two parents, WTR-1 and Haoannong, at two linked loci, *qLTG(I1)*11 and *qLTG(III)*11, on chromosome 11 [34]. Li et al. (2019) also observed that Dongxiang wild rice (*O. rufipogon* Gri ff.) introgressions at five detected QTLs in 94 BC1F7 population delayed germination rates under 15 ◦C conditions in the background of *indica* variety, Xieqingzao. Among them, two QTLs, *qLTG10-1* and *qLTG10-2*, were identified on chromosome 10 [25]. The results indicate that two linked QTLs potentially act in opposite directions in such QTL regions. Such genetic linkage is common in rice [35]. A linkage between two desirable genes would be advantageous in the selection of improved lines. For example, the tight linkage of two QTLs (*qSPP5* for spikelet no. and *qTGW5* for grain weight) could be valuable for improving rice yield [36]. However, linkage between desirable and undesirable genes is complex in terms of its application in rice breeding [37]. Since two LTG QTLs, *qLTG10.1* and *qLTG10.2*, acting in opposite directions are linked and have minor e ffects, selection of the high LTG lines with *qLTG10.1* from *O. rufipogon* and *qLTG10.2* could be accomplished using DNA markers. In a previous study, we detected *qLTG10.1* and *not qLTG10.2*, possibly due to the bu ffering e ffect of two QTLs and interactions among QTLs in the population [27]. Overall, it appears that the region carries gene(s) with a strong e ffect on germination performance and represent additional genetic targets for MAS directed development of rice varieties with improved LTG.

The interactions among the four QTLs, *qLTG1*, *qLTG3*, *qLTG8*, *qLTG10.1*, and *qLTG10.2* were examined using general regression models. The plants that harbor the *O. rufipogon* alleles at *qLTG1*, *qLTG3*, *qLTG8*, and *qLTG10.1* exhibited the highest germination rates at 13 ◦C in the nine groups, and the five QTLs cumulatively explained 42.0% of the phenotypic variance in LTG. The results imply that five QTLs control the LTG in an additive manner. Pyramiding the four QTLs from the *O. rufipogon* into cultivated rice with *qLTG10.2* would facilitate breeding programs aimed at enhancing LTG for direct-seeding production systems. It is also notable that the plants with four *O. rufipogon* alleles at *qLTG1*, *qLTG3*, *qLTG8*, and *qLTG10.1* exhibited lower LTG than *O. rufipogon*, the donor parent at 5–7 DAI, indicating the presence of additional QTLs for LTG in *O. rufipogon*. Further experiments are underway to detect and characterize such unknown QTLs in *O. rufipogon*.

Also, three QTLs for coleoptile length detected on chromosomes 1, 3, and 8 shared their locations with three LTG QTLs, *qLTG1*, *qLTG3*, and *qLTG8,* respectively, and the *O. rufipogon* alleles at all three loci increased the coleoptile length, suggesting a pleiotropy of a single QTL at each locus.
