Introduction
Some of the recent research on the International Rice Genebank Collection (IRGC) at the International Rice Research Institute (IRRI) has focused on understanding and improving the longevity of rice seeds in storage, to ultimately increase the efficiency and effectiveness of genebank operations. Some innovations include automated seed sorting to get seeds into the genebank cold rooms sooner after harvest; changed drying conditions from immediate cool drying (15°C) to high-temperature drying (40°C) prior to cool drying, which improved the longevity of seeds, particularly when harvested with high moisture content (MC) (Whitehouse et al., Reference Whitehouse, Hay and Ellis2015, Reference Whitehouse, Hay and Ellis2017, Reference Whitehouse, Hay and Ellis2018; Timple and Hay, Reference Timple and Hay2018); confirmation of 5-year monitoring intervals for seeds stored at 5°C (medium-term) (Hay et al., Reference Hay, de Guzman and Hamilton2015); confirmed correlation between seed longevity and vitamin E homologue ratios, rather than total vitamin E content (Lee et al., Reference Lee, Kwak, Yoon, Lee and Hay2017, Reference Lee, Kwak, Cho, Chebotarov, Yoon, Lee, Hamilton and Hay2019a) and through a genome-wide association (GWA) study of diverse Indica rice varieties, identification of eight loci that are related to longevity parameters for seeds of Indica rice varieties stored at 45°C and 10.9% MC (Hay et al., Reference Hay, Valdez, Lee and Sta. Cruz2019; Lee et al., Reference Lee, Valdez, Punzalan, Pacleb, Kretzschmar, McNally, Ismail, Sta Cruz, Hamilton and Hay2019b). Among these aspects, attention has been given to vitamin E α-, β-, γ- and δ-tocopherol/tocotrienol homologues, due to their antioxidant role. During seed storage under hot and humid conditions, oxidative stress and in particular lipid peroxidation occurs and results in membrane breakdown and cell ageing (Sattler et al., Reference Sattler, Cahoon, Coughlan and DellaPenna2004). Vitamin E effectively scavenges lipid peroxy radicals, thus prolonging seed longevity. Indica is perhaps the largest variety group of cultivated rice, certainly the most represented within the IRGC, and widely grown in the lowlands of tropical Asia as well as some areas in the USA and Latin America (Londo et al., Reference Londo, Chiang, Hung, Chiang and Schaal2006; Wang et al., Reference Wang, Mauleon, Hu, Chebotarov, Tai, Wu, Li, Zheng, Fuentes, Zhang, Mansueto, Copetti, Sanciangco, Palis, Xu, Chen, Fu, Zhang, Gao, Zhao, Shen, Cui, Yu, Li, Chen, Detras, Zhou, Zhang, Zhao, Kudrna, Wang, Li, Jia, Lu, He, Dong, Xu, Li, Wang, Shi, Li, Zhang, Lee, Hu, Poliakov, Dubchak, Ulat, Borja, Mendoza, Ali, Li, Gao, Niu, Yue, Naredo, Talag, Wang, Li, Fang, Yin, Glaszmann, Zhang, Li, Hamilton, Wing, Ruan, Zhang, Wei, Alexandrov, McNally, Li and Leung2018). In contrast, Aus is a smaller variety group, mainly grown in Bangladesh and India during the March–August season (Khush, Reference Khush1997). Aus varieties are considered more closely related to wild rice species (Londo et al., Reference Londo, Chiang, Hung, Chiang and Schaal2006) with unfavourable characteristics such as excessively tall plant height, small grain size and high shattering (Lee et al., Reference Lee, Wissuwa, Zamora and Ismail2018). However, they are a potentially important source of genes conferring tolerance to abiotic stresses such as drought and zinc deficiency (Bin Rahman and Zhang, Reference Bin Rahman and Zhang2018; Lee et al., Reference Lee, Wissuwa, Zamora and Ismail2018). The Aus group is less well described than the Indica and Japonica variety groups, and there is little information about the relative longevity of seeds from Aus varieties.
In this study, we first screened a diverse Aus rice panel for seed longevity and conducted GWA analysis for QTL detection. Secondly, we tested previous conclusions (Lee et al., Reference Lee, Kwak, Cho, Chebotarov, Yoon, Lee, Hamilton and Hay2019a) of a positive correlation between rice seed longevity and the proportion of γ-type vitamin E, and a negative correlation between seed longevity and the proportion of α- and β-types of vitamin E.
Materials and methods
Plant materials
A total of 185 Aus rice accessions that were included in ‘The 3000 Rice Genomes Project’ (2014) were grown at the Zeigler Experiment Station of the International Rice Research Institute (IRRI), Philippines, in the 2016 wet season. Twenty-day-old seedlings were transplanted to the field with 200 mm spacing of plants within and between rows. Seed lots were harvested at 35 d after peak flowering. Harvest MC was determined by weighing three replicate samples of ground seeds before and after 2 h oven-drying at 130°C followed by 1 h over silica gel at room temperature (ISTA, 2018). Grain colour and pericarp colour were scored visually (Table 1). After blowing and sorting to discard immature, damaged, off-type or diseased seeds, seed lots were dried at 15°C, 15% relative humidity (RH) and then transferred to foil-laminate bags and stored at −20°C until the seed storage experiments commenced.
Phenotypic traits measured for 185 Aus rice accessions
Seed storage experiments
Foil bags were removed from −20°C storage and allowed to equilibrate to room temperature before opening and removal of seeds. Seed water content was adjusted at 60% RH, 25°C in a climate test chamber (Model VC3 0034-M; Vötschtechnik, Germany) to achieve a target MC of 10.9% (fresh weight basis) and then seeds were sealed inside foil-laminate bags and placed at 45°C. Samples were taken at 0, 7 and 14 d, then at 4 or 10 d intervals for up to 42 or 60 d of storage. For each sample, two replicates of 30 seeds were placed on two layers of Whatman No. 1 filter paper with 7 ml distilled water in 90 mm-diameter Petri dishes. Seeds were incubated at 30°C with 12 h light per day and germination was scored for up to 21 d after sowing (0 d-storage sample were scored daily and other storage treatments were scored at 5, 9, 14 and 21 d). MC was determined as above, at the outset, during and end of storage using the seeds from three further packets at each sampling time. Probit analysis of germination data was performed using GenStat version 18 (VSN International Ltd, Hemel Hempstead, UK). Seed longevity parameters from the Ellis and Roberts (Reference Ellis and Roberts1980) viability equation were estimated: K i, the initial viability in normal equivalent deviates (NED); and −1/σ, the slope of the transformed survival curve (i.e. change in viability in NED d−1). The p 50, the time (d) for viability to fall to 50% during storage at 10.9% MC and 45°C was also estimated.
Vitamin E analysis
Vitamin E in de-hulled (palea and lemma removed) seeds was analysed following Ko et al. (Reference Ko, Kim, Kim, Kim, Chung, Tae and Kim2003). Briefly, the ground sample (50 g) was mixed with 300 ml n-hexane for 2 h, then concentrated by evaporating hexane using nitrogen gas. The lipid extract (0.5 g) was mixed with 2 ml 5% pyrogallol solution in ethanol and 20 ml ethanol. After boiling at 70°C, 1 ml of 50% aqueous KOH was added for the 5 min saponification. The sample was extracted by 50 ml diethyl ether, washed with 20 ml distilled water, filtered through anhydrous sodium sulphate and evaporated at 30°C. The residue was diluted with 10 ml n-hexane and filtered through a Millipore 0.2 μm membrane. Individual vitamin E homologues were quantified by ultra-performance liquid chromatography (UPLC, H-Class System, Waters, Massachusetts, USA) at 298 nm excitation and 325 nm emission with a Lichrospher Si-60 column (250 × 4.6 mm i.d.; Merck Co., Gernsheim, Germany). Descriptive statistics of all traits and correlations were analysed using STAR v2.0.1 (International Rice Research Institute).
GWA analysis
GWA analysis was performed using TASSEL 5.2.7 (Bradbury et al., Reference Bradbury, Zhang, Kroon, Casstevens, Ramdoss and Buckler2007) based on the 446k (filtered for 20% missing data and minor allele frequencies <5%) single nucleotide polymorphisms (SNP) marker data of The 3000 Rice Genomes Project (Wang et al., Reference Wang, Mauleon, Hu, Chebotarov, Tai, Wu, Li, Zheng, Fuentes, Zhang, Mansueto, Copetti, Sanciangco, Palis, Xu, Chen, Fu, Zhang, Gao, Zhao, Shen, Cui, Yu, Li, Chen, Detras, Zhou, Zhang, Zhao, Kudrna, Wang, Li, Jia, Lu, He, Dong, Xu, Li, Wang, Shi, Li, Zhang, Lee, Hu, Poliakov, Dubchak, Ulat, Borja, Mendoza, Ali, Li, Gao, Niu, Yue, Naredo, Talag, Wang, Li, Fang, Yin, Glaszmann, Zhang, Li, Hamilton, Wing, Ruan, Zhang, Wei, Alexandrov, McNally, Li and Leung2018). A mixed linear model (MLM) with a kinship matrix was applied and plots above a significance threshold (P < 3.00 × 10−5) were considered as major QTL associated with the traits. For haplotype analysis, SNP markers in the region of major QTL were extracted from the Rice SNP-Seek Database (http://snp-seek.irri.org) containing genotype data of The 3000 Rice Genomes Project (Wang et al., Reference Wang, Mauleon, Hu, Chebotarov, Tai, Wu, Li, Zheng, Fuentes, Zhang, Mansueto, Copetti, Sanciangco, Palis, Xu, Chen, Fu, Zhang, Gao, Zhao, Shen, Cui, Yu, Li, Chen, Detras, Zhou, Zhang, Zhao, Kudrna, Wang, Li, Jia, Lu, He, Dong, Xu, Li, Wang, Shi, Li, Zhang, Lee, Hu, Poliakov, Dubchak, Ulat, Borja, Mendoza, Ali, Li, Gao, Niu, Yue, Naredo, Talag, Wang, Li, Fang, Yin, Glaszmann, Zhang, Li, Hamilton, Wing, Ruan, Zhang, Wei, Alexandrov, McNally, Li and Leung2018) and effects of haplotype on the phenotype were determined.
Results
Due to variation in flowering date (66–121 d from sowing to flowering), accessions were harvested on different dates (between 6 August and 16 October) under different weather conditions. The harvest MC (% fresh weight) ranged between 12.4 and 34.3%.
The mean MC of the seeds during storage at 45°C in foil bags was 11.09% (SD 0.39) (three replicate observations per accession on each of three occasions) and did not vary significantly depending on sampling time.
Many accessions showed evidence of dormancy at the start of storage, with germination increasing during the early phase of storage, before declining as the seeds aged and lost viability (Supplementary Figs S1–S13). Because there were, in general, insufficient data to characterize the breaking of dormancy, the probit analysis to describe the loss of viability only used data in which germination was declining. Some accessions lost viability relatively quickly, with little or no germination after 20 d in storage (e.g. IRGC 120876, 120915, 125757, 126123, 127170, 127236, 127528, 128297); other seed lots showed some germination after 50 d in storage (e.g. IRGC 127130, 127228, 127481, 127652, 128369, 132306). Thus, there was a wide variation in the results of the probit analysis for each accession.
The initial seed lot viability, K i, ranged between 0.84 (IRGC 127796) and 8.36 NED (IRGC 127652) with a mean of 3.28 NED (Table 1). The slope of the survival curves, −σ −1 ranged between −0.72 (IRGC 127253) and −0.06 (IRGC 127903) NED d−1 with a mean of −0.22 NED d−1. The estimate of p 50 ranged between 2.43 (IRGC 127454) and 54.20 (IRGC 127652) d with a mean of 16.90 d. A total of 90 accessions composed of 30 high, 30 intermediate and 30 poor seed longevity were selected for vitamin E analysis. Total vitamin E content in de-hulled seeds ranged between 0.29 and 0.66 mg g−1 oil (accessions IRGC 128442 and 127139, respectively), with a mean of 0.51 mg g−1 oil.
There was no significant correlation between total vitamin E content and seed longevity in this (data not shown) and previous studies (Lee et al., Reference Lee, Kwak, Yoon, Lee and Hay2017, Reference Lee, Valdez, Punzalan, Pacleb, Kretzschmar, McNally, Ismail, Sta Cruz, Hamilton and Hay2019b). We further tested correlations between proportions of each vitamin E homologue and seed longevity. K i was positively correlated with γ- and δ-tocopherols (correlation coefficient, r = 0.254 (P < 0.05) and 0.282 (P < 0.01), respectively) and negatively correlated with α-tocotrienol (r = −0.245; P < 0.01) (Table 2). The parameter p 50 was positively correlated with δ-tocopherol (r = 0.233; P < 0.05). The parameter −σ−1 did not correlate with any vitamin E homologues. Pericarp colour (scores between 0 and 3.5 from pure white to dark red) was positively correlated with all seed longevity values (P < 0.001 for K i and −σ−1 or P < 0.0001 for p 50). Seed harvest MC (% fresh weight) was strongly negatively correlated with all seed longevity values, especially p 50 (r = −0.600; P < 0.0001) (Table 2 and Fig. 1A). Seed harvest MC was also negatively correlated with days to flowering (r = −0.334; P < 0.0001), which indicates the tendency of decline in harvest MC in later-maturing seeds.
(A) Correlation between harvest moisture content (%) and p 50 (d) for 185 Aus rice accessions. Correlation coefficient (r) was significant at ****P < 0.0001; (B) GWA analysis (MLM) of seed longevity value (p 50) without considering harvest moisture content as an environmental factor. A large false positive peak appeared on chromosome 6; (C) GWA analysis (MLM) of seed longevity value (p 50) with including harvest moisture content as a covariate of p 50.
Correlation coefficients (Spearman's) for the traits potentially associated with seed longevity in a diverse Aus rice panel
Harvest moisture content (italics) was considered as an environmental factor for the GWAS.
Significant at *P < 0.05; **P < 0.01, ***P < 0.001; ****P < 0.0001; ns, not significant (P > 0.05)
GWA analysis was conducted for seed longevity and other associated traits such as α-tocotrienol, δ-tocopherol and pericarp colour (Fig. 2; Supplementary Fig. S14). The initial Manhattan plot on p 50 showed one significant single marker on chromosome 5 and a large consistent peak on chromosome 6 (Fig. 1B). As harvest MC was an environmental factor strongly affecting seed longevity, we took this factor into account in the analysis to correct false positives (Figs 1C and 2). Three of the most significant single markers (P-value < 3.00 × 10−5) on chromosomes 4 and 5 were associated with K i (Fig. 2 and Table 3). Eight significant single markers on chromosomes 2, 3, 5, 6, 7, 9 and 11 were associated with p 50. Two single markers on chromosome 2 and a consistent major peak on chromosome 4 were associated with −σ−1. These significant markers were located in genes or in the range of 3549 basepair (bp) downstream and 6505 bp upstream of the nearest genes (Table 3). Major loci associated with each trait differed (Fig. 2). However, a number of minor loci (P-value > 3.00 × 10−5) were associated with multi-traits (Table 4). Seven markers on chromosomes 2, 6, 8, 10, 11 and 12 associated with both K i and p 50, with P-values between 7.78 × 10−4 and 1.71 × 10−5 (Table 4). These markers were located in the genes or in the range of 7740 and 289 bp downstream of the nearest gene. Three markers on chromosomes 1, 3 and 4 were associated with both δ-tocopherol and K i (Fig. 3). Two markers on chromosomes 1 and 8 were associated with both δ-tocopherol and p 50. The allelic effects of those markers on phenotype values were estimated through haplotype mapping. The presence of favourable haplotypes (green shaded) was associated with enhancements of δ-tocopherol and K i. Genotype no. 1 with favourable haplotypes on all markers had higher δ-tocopherol (120%) and K i (67%) compared with genotype no. 5 which lacked favourable haplotypes (Fig. 3A). The favourable haplotypes with the two markers on chromosomes 1 and 8 (genotype no. 1) enhanced both δ-tocopherol and p 50, by 100 and 116%, respectively, when compared with the negative group (genotype no. 3) (Fig. 3B).
GWA analysis (MLM) of seed longevity traits (K i, −σ−1 and p 50) and other traits (α-tocotrienol and δ-tocopherol) significantly correlated with seed longevity.
Haplotype effects on δ-tocopherol and seed longevity traits in a diverse Aus rice panel. The accessions with the absence of genotype data on the regions were not included in this figure.
SNP markers most significantly (P-value < 3.00 × 10−5) associated with seed longevity traits
SNP markers associated with multi-traits
Discussion
Seed longevity is a complex trait, with high phenotypic plasticity (Leprince et al., Reference Leprince, Pellizzaro, Berriri and Buitink2017). This makes it difficult to identify the molecular basis of this trait. For example, our recent study using seeds of diverse Indica rice varieties produced during the 2015 dry season at IRRI identified that harvest MC, which depends on the environmental conditions at the time of harvest once seeds are in the late maturation phase, was more strongly correlated with p 50 (r-value −0.344; P < 0.0001) than any genetic trait (Lee et al., Reference Lee, Valdez, Punzalan, Pacleb, Kretzschmar, McNally, Ismail, Sta Cruz, Hamilton and Hay2019b). The current study, using seeds of Aus varieties produced in the 2016 wet season, showed a stronger correlation (r = −0.600; P < 0.0001) between harvest MC and p 50 (Table 2). This led to the identification of loci which were a consequence of harvest MC and hence progression through the late maturation phase, rather than being associated with longevity per se (Fig. 1B). Another factor that significantly affects seed longevity is the post-harvest drying conditions. Whitehouse et al. (Reference Whitehouse, Hay and Ellis2018) reported that subsequent seed longevity at 10.9% MC and 45°C is improved for seeds that are harvested with a high MC, if seeds are initially dried at a higher temperature (45–60°C) prior to routine cool (15°C) drying. Therefore, the effect of seed drying condition on subsequent seed longevity may also interfere with the genetic analysis of this trait. For accurate QTL analysis, such environmental factors must be controlled or well addressed in genetic analysis. Thus, while we have attempted to identify genes with a role in Aus seed longevity, for gene validation purposes, it would be important to screen accessions produced in different seasons and perhaps with different drying treatments, depending on harvest MC.
Longevity parameters (K i, −σ−1 and p 50) observed in this study for seeds of Aus varieties were similar to those observed for seeds of Indica varieties stored in the same way (Lee et al., Reference Lee, Valdez, Punzalan, Pacleb, Kretzschmar, McNally, Ismail, Sta Cruz, Hamilton and Hay2019b). For example, p 50 ranged between 5.41 and 59.12 d for Indica accessions (maximum for each accession across sequential harvests) with a mean of 24.24 and coefficient of variation (CV) of 44.32%, compared with a range of 2.43 to 54.20 for Aus accessions, with a mean of 16.90 and CV of 52.90% (Table 1). The variation in the slope, −σ−1, in both studies, again demonstrates that the Ellis and Roberts (Reference Ellis and Roberts1980) viability equation should be used with some caution to make predictions of longevity in storage, at least across very diverse rice accessions (Whitehouse et al., Reference Whitehouse, Hay and Ellis2018).
Seed longevity loci of Indica (Lee et al., Reference Lee, Valdez, Punzalan, Pacleb, Kretzschmar, McNally, Ismail, Sta Cruz, Hamilton and Hay2019b) and Aus (current) rice panels did not coincide. This was possibly due to the intraspecific variation in genomic structure (Wang et al., Reference Wang, Mauleon, Hu, Chebotarov, Tai, Wu, Li, Zheng, Fuentes, Zhang, Mansueto, Copetti, Sanciangco, Palis, Xu, Chen, Fu, Zhang, Gao, Zhao, Shen, Cui, Yu, Li, Chen, Detras, Zhou, Zhang, Zhao, Kudrna, Wang, Li, Jia, Lu, He, Dong, Xu, Li, Wang, Shi, Li, Zhang, Lee, Hu, Poliakov, Dubchak, Ulat, Borja, Mendoza, Ali, Li, Gao, Niu, Yue, Naredo, Talag, Wang, Li, Fang, Yin, Glaszmann, Zhang, Li, Hamilton, Wing, Ruan, Zhang, Wei, Alexandrov, McNally, Li and Leung2018). Among 19 sub-groups of 3010 diverse rice accessions, there was a large gap in genomic structures between Indica and Aus groups. McCouch et al. (Reference McCouch, Wright, Tung, Maron, McNally, Fitzgerald, Singh, DeClerck, Agosto-Perez, Korniliev, Greenberg, Naredo, Mercado, Harrington, Shi, Branchini, Kuser-Falcaõ, Leung, Ebana, Yano, Eizenga, McClung and Mezey2016) also discovered subgroup-specific alleles associated with grain traits in 1953 diverse rice accessions. Famoso et al. (Reference Famoso, Zhao, Clark, Tung, Wright, Bustamante, Kochian and McCouch2011) identified clear SNP variation in three exon regions of the Nrat1 Al tolerance gene between Aus and Indica varieties. With this evidence, we speculate that different rice groups may have different mechanisms conferring high seed longevity. TPP7, a trehalose-6-phosphate phosphatase played an important role in seed longevity extension in near-isogenic lines (NIL) derived from the cross between temperate Japonica (short survival time) and Aus (long survival time) parents (Sasaki et al., Reference Sasaki, Takeuchi, Miura, Yamaguchi, Ando, Ebitani, Higashitani, Yamaya, Yano and Sato2015). However, in many Indica accessions, there was a large deletion in the TPP7-region, thus this gene is not functional in Indica group (Kretzschmar et al., Reference Kretzschmar, Pelayo and Trijatmiko2015). In the Indica panel, a major mechanism conferring high seed longevity was related to DNA repair and transcription rather than TPP7 (Lee et al., Reference Lee, Valdez, Punzalan, Pacleb, Kretzschmar, McNally, Ismail, Sta Cruz, Hamilton and Hay2019b).
Among 26 candidate genes where the most significant SNP markers located on or nearby, 13 genes are expressed genes, hypothetical genes or (retro-)transposon proteins (Table 3). The function of these genes has not been examined in detail, mainly due to the lack of technologies elucidating their expressions and functions. However, recent biological studies revealed that those genes are potentially important in biological processes as well as stress tolerance (Ijaq et al., Reference Ijaq, Malik, Kumar, Das, Meena, Bethi, Sundararajan and Suravajhala2019; Yang et al., Reference Yang, Zeng and Tsui2019). Among seven SNP markers associated with multi-traits (K i and p 50), two, one and two markers were on or nearby expressed genes, hypothetical genes and retrotransposon proteins, respectively; the other two were located on annotated genes (Table 4). Further integrative approaches combining systems-genetics and networking functions will be key to unravelling the role of those genes.
• LOC_Os05g04950, associated with K i, was annotated to protein-binding proteins (Rice Genome Annotation Project; accessed on 22 September 2019). Based on gene ontology with networking genes (RiceNet v2, accessed on 22 September 2019; Lee et al., Reference Lee, Oh, Yan, Shin, Hwang, Kim, Kim, Shim, Shim, Ronald and Lee2015), this gene involved protein modification process, lipid metabolic process and RNA modification.
• LOC_Os04g35650 associated with −σ−1 was annotated to pentatric opeptide which is a kind of RNA binding protein mediating gene expression in the nucleus as well as organelles (Manna, Reference Manna2015). This gene was connected with LOC_Os06g30380 and LOC_Os08g29110 which are related to RNA modification and oxidation reduction, respectively.
• LOC_Os04g35930 and LOC_Os04g35990, associated with −σ−1, were annotated to OsFBX134 - F-box domain-containing protein which is known to mediate protein–protein interactions (Ho et al., Reference Ho, Tsai and Chien2006).
• LOC_Os02g39480, associated with p 50, was annotated to protein phosphatase 2C which is a central component in abscisic acid signal transduction (Rodriguez, Reference Rodriguez1998).
• LOC_Os11g45620, associated with p 50, was annotated to rust-resistance protein Lr21. There was no other network gene, but since this gene was highly expressed in mature seeds (Rice Genome Annotation Project), we speculate that it may play a role in the seed defence mechanism, i.e. oxidative stress occurring during seed storage.
Further gene validation studies are needed to confirm the roles of these highlighted genes.
Previous (Lee et al., Reference Lee, Kwak, Yoon, Lee and Hay2017, Reference Lee, Kwak, Cho, Chebotarov, Yoon, Lee, Hamilton and Hay2019a) and current studies consistently showed that high proportions of γ- and δ-types of vitamin E homologues were positively correlated with seed longevity (Table 2). This could be due to more efficient and synergistic antioxidant activity of γ- and δ-homologues when compared with α- and β-types (Kadoma et al., Reference Kadoma, Ishihara, Okada and Fujisawa2006; Jiang, Reference Jiang2014; Kim, Reference Kim2014). Hence, we recommend that breeding programmes for improving seed longevity of elite rice varieties focus on increase in proportions of γ- and δ-vitamin E homologues rather than total vitamin E content. Pericarp colour was positively correlated with p 50, which indicated stronger correlations than that of γ- and δ-vitamin E homologues (Table 2). Pericarp colour is known to contain various antioxidants such as flavonoids (Goufo and Trindade, Reference Goufo and Trindade2013). Marker 315490397 on chromosome 10 at position 22005127 associated with both K i and p 50 located at 7740 bp downstream of the flavonol synthase gene (Table 4). With this evidence, there is a possibility that in Aus rice, other types of antioxidants, besides vitamin E, that are abundant in pericarp colour can play an important role in extending seed shelf life. Our complementary studies on seed metabolite profiling also suggest the specific types of flavonoids are relevant to high germination rate after seed storage (not published).
Conclusion
In line with previous studies, we confirmed that a high proportion of γ- and δ-type homologues rather than total vitamin E content was associated with seed longevity in a large rice panel. This confirmation revisited a theory of prolonged seed longevity via an antioxidant mechanism. Red pericarp colour containing other types of antioxidants was also correlated with high seed longevity.
Based on the comparative genetic analysis, we observed totally different QTL regions between GWA mapping with and without taking into account harvest MC. This environmental factor, which depends on ambient humidity of the harvest date, strongly interfered with genetic analysis. We, therefore, concluded that along with accurate phenotyping methods, interpretation of data in the context of environmental factors is extremely important in discovering the correct genes associated with the trait. Identified gene functions such as RNA modification and oxidation reduction, and/or SNP markers associated with both high δ-tocopherol and seed longevity traits (K i and p 50) would be useful information in further vitamin E-seed longevity studies.
Supplementary material
To view supplementary material for this article, please visit: https://doi.org/10.1017/S0960258520000173.
Acknowledgements
We would like to thank IRRI colleagues: Renato Reaño for seed multiplication, Nora Kuroda for seed storage experiments, Dmytro Chebotarov for technical advice on genotype data and Ruaraidh Sackville Hamilton for supporting this research.
Financial support
This research was funded by grants from the Rural Development Administration, Korea (RDA) as RDA-IRRI cooperative research project (Grant Ref. PJ012723) and Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH. BMZ Funding of Genebank Uplift (Grant Ref. 17.7860.4-001.00).
Conflicts of interest
None declared.
