Red pericarp trait, common to weedy rice, contributes to freezing tolerance of seeds and their ability to overwinter in soil

TL;DR

This study shows that the red pericarp trait in rice, controlled by Rc and Rd genes, improves seed survival in cold conditions but does not cause deep dormancy.

Contribution

The combined effect of Rc and Rd on freezing tolerance and overwintering ability is newly demonstrated in rice.

Findings

NILs with both Rc and Rd showed higher freezing tolerance than other lines.

Red pericarp increases seed longevity through cold tolerance, not dormancy.

Lines with Rc or Rd alone failed to survive winter unlike weedy rice controls.

Abstract

The red pericarp trait is controlled by Rc and Rd, is prevalent in weedy rice, and is associated with seed dormancy and longevity. However, the individual and combined effects of these genes on seed adaptation remain unclear. We developed near-isogenic lines (NILs) carrying ‘Kasalath’ Rc and/or Rd alleles in the ‘Koshihikari’ background and evaluated seed dormancy, freezing tolerance, and overwintering ability under controlled and field conditions, using ‘Koshihikari’, ‘Kasalath’, and the Japanese weedy rice accession ‘JP_1177’ as controls. NILs carrying Rc, Rd, or both did not have the deep primary dormancy typical of weedy rice. Under field conditions, these lines germinated prematurely and failed to survive winter, unlike ‘JP_1177’. However, the NIL carrying both Rc and Rd had higher freezing tolerance than the other lines, suggesting an interaction between these genes. These findings indicate that the red pericarp trait does not confer deep dormancy but may increase seed longevity primarily by improving low-temperature tolerance and overwintering ability in soil. These results provide insights into the adaptive significance of the red pericarp trait in temperate environments, although our data are limited to a cultivated temperate japonica (Koshihikari) background.

This study shows that the red pericarp trait, controlled by Rc and Rd, does not by itself confer deep seed dormancy in rice. Using near isogenic lines in a temperate japonica background, we demonstrate that the combined action of Rc and Rd markedly enhances seed freezing tolerance and overwintering survival in soil. These findings suggest that red pericarp contributes to seed persistence primarily through improved cold tolerance rather than dormancy.

Figures

NILs tested. KH, ‘Koshihikari’; KH-Rd and KH-Rc, lines developed in this study; KH-RdRc, ‘Toyama-aka 71 gou’. (a) Graphical genotype of KH-RdRc according to Yamaguchi et al. (2015). Photographs show hulled and de-hulled rice. (b) Enlarged image of de-hulled rice. Only KH-RdRc had red pericarp. Alt text: Photographs and schematic diagrams of near-isogenic rice lines carrying different combinations of the Rc and Rd alleles in the Koshihikari background, showing pericarp colour differences among lines, with only the line carrying both Rc and Rd exhibiting a red pericarp.

Polymorphic sites in Rd and Rc genes. Green boxes, coding regions. White triangles, variants in ‘Koshihikari’ vs. ‘Nipponbare’ reference genome; red triangles, variants in ‘Kasalath’ vs. ‘Nipponbare’. Asterisks, mutation sites among weedy rice accessions. Black dots, mutation sites in wild rice. Black arrows, sites of mutations that result in loss of gene function (Furukawa et al. 2007). Alt text: Schematic comparison of nucleotide polymorphisms in the Rc and Rd genes among Koshihikari, Kasalath, weedy rice, and wild rice, indicating coding regions, variant positions, and mutations affecting gene function.

Trends in germination during storage. The proportion of viable seeds that germinated within 14 days was estimated using a generalized linear model (GLM) with a quasi-binomial distribution. Lines, maximum likelihood estimates of germination proportions; ribbons, 95% confidence intervals; dots, proportion of viable seeds that germinated within 14 days. Eight biological replicates were used. Red arrows highlight shallow dormancy in KH–RdRc at 30 °C one week after harvest, whereas blue arrows indicate deep dormancy in ‘JP_1177’ at 15 °C late in the storage period. Alt text: Line plots showing changes in seed germination proportion during dry storage after harvest at two incubation temperatures, highlighting shallow dormancy in near-isogenic lines and deep dormancy in weedy rice.

Freezing tolerance. Left panel, single freezing treatment; right panel, three consecutive freezing treatments for each cultivar and near-isogenic line (NIL). Seeds were soaked at 5 °C for 24 h, exposed to −10 °C for 24 h, returned to 5 °C for 24 h for once or three times, and then incubated at 30 °C to assess viability. Box plots show the median (central line) and interquartile range (box); whiskers indicate the 10th–90th percentiles. Jittered points represent individual biological replicates (n = 8; n = 7 for KH–Rd under three freezing cycles due to one missing replicate). Different letters above boxes indicate significant differences among cultivars/lines within each treatment (Tukey–Kramer test, P < 0.05). Alt text: Box plots comparing seed viability of rice cultivars and near-isogenic lines after one or three freezing treatments, showing enhanced freezing tolerance in the line carrying both Rc and Rd alleles.

Overwintering ability in soil under field conditions. Seed survival in April of the following year is shown for seeds buried in September (1 week after harvest; orange box plot) or in December (3 months after harvest; blue box plot). Box plots show the median (central line) and interquartile range (box); whiskers indicate the 10th–90th percentiles. Jittered points represent individual biological replicates (n = 8). Different letters indicate significant differences among cultivars/lines within each burial treatment (Tukey–Kramer test, P < 0.05). Alt text: Box plots showing overwintering seed survival of rice cultivars and near-isogenic lines buried in soil in autumn or winter, indicating higher overwintering survival in the line carrying both Rc and Rd alleles.

Tables

Table 1. Plant materials used in this study and their pericarp phenotypes. Pericarp colour is determined by the functional or non-functional state of Rc and Rd: red (both functional), brown (Rc only), white (rc and/or rd).

Plant material	Genotype	Phenotype (pericarp color)	Description
Koshihikari (KH)	rdrd/rcrc	White	japonica cultivar, genetic background of NILs
Toyama-aka 71 gou (KH-RdRc)	RdRd/RcRc	Red	japonica cultivar (NIL; Yamaguchi et al. 2015)
KH-Rd	RdRd/rcrc	Brown	NIL developed in this study
KH-Rc	rdrd/RcRc	White	NIL developed in this study
Kasalath	RdRd/RcRc	Red	indica cultivar, landrace, Rd and Rc donor of NILs
JP_1177	RdRd/RcRc	Red	Weedy rice accession from Japan (Imaizumi et al. 2021)

Table 2. Primers and PCR conditions used for marker genotyping in this study. Primers were designed to identify the genomic segments containing Rc and Rd derived from ‘Kasalath’.

Gene	Marker	Primer	Sequence (5′−3′)	Marker type	Position	PCR condition
Rc	Rc_mix	Rc_7835fw	CATTACAACACGCTGCGACA	ARMS		(1)
		Rc_8070fw	GAACGCGAAAAGTCGGTG	ARMS	Chr07_6068071
		Rc_8082rv	CTTGGATGGCATCCACTT	ARMS	∼6068471
		Rc_8472rv	CTTCCAATGTTCGTTAGAGGCT	ARMS
Rd	RM8129	RM8129_Primer-U	CTCAACCCGGCTTTCCATCTCG	Indel	Chr01_25071214	(2)
		RM8129_Primer-L	GCTGCAGAGTCTCGCACGTTCC	Indel	∼25071400
	RM3143	RM3143_Primer-U	AAAGCCTGGATAAGATGGTTCG	Indel	Chr01_26821629	(2)
		RM3143_Primer-L	CTGTAGTTGCTGTTTGCCTGTCC	Indel	∼26821752
	CO1_25.6	NKas_indel_C01_25.6(129)Fw	ACCACGAGCGTTCCATGTGAAGTTT	Indel	Chr01_25597988	(2)
		NKas_indel_C01_25.6(129)Rv	TACGTGCTGACGAGACGACTGAACA	Indel	∼25598116
(1) 35 cycles of 98°C for 10 s, 55°C for 5 s, 68°C for 5 s.
(2) 35 cycles of 98°C for 10 s, 55°C for 30 s, 72°C for 30 s.

Table 3. Pre-winter emergence of seedlings from seeds buried in September.

Cultivar/Line	2023		2024
	Maximum likelihood estimate	(95% confidence interval)	Maximum likelihood estimate	(95% confidence interval)
KH	0.64	(0.55–0.72)	0.79	(0.74279627–0.83)
KH-Rd	0.79	(0.72–0.86)	0.86	(0.82544874–0.90)
KH-Rc	0.57	(0.48–0.65)	0.79	(0.74113202–0.83)
KH-RdRc	0.68	(0.59–0.75)	0.88	(0.83751942–0.90)
Kasalath	0.32	(0.24–0.40)	0.08	(0.05818743–0.12)
JP_1177	0.01	(0.00–0.05)	0.03	(0.01596227–0.05)

Full text

Introduction

Weedy rice (Oryza spp.) belongs to the same genus and often the same species as cultivated rice (O. sativa) (Roma-Burgos et al. 2021) and is characterized by tall height, high seed shattering, and deep seed dormancy (Nadir et al. 2017). Globally, weedy rice populations have arisen through independent de-domestication events from cultivated rice and, in some cases, hybridization with wild relatives, resulting in diverse genetic backgrounds (Kanapeckas et al. 2016). In Japan, weedy rice originated from japonica cultivated rice through de-domestication and acquired wild-like traits via hybridization (Imaizumi et al. 2021). Weedy rice occurs in rice-growing regions worldwide, including the USA, Asia, and Southern Europe, and reduces rice quality and yield (Chin 2001, Karim et al. 2006, Shivrain et al. 2009). Weedy rice competition results in up to 50% yield losses (Vidotto and Ferrero 2009).

Seed shattering facilitates the dispersal of weedy rice seeds before and during rice harvest, while deep dormancy prevents germination under unfavourable conditions, allowing seeds to persist in the soil seed bank until the following spring. Seed dormancy and longevity are widely recognized as adaptive traits in plants, as they enable seeds to survive unfavourable environmental conditions and persist in the soil seed bank (Bewley 1997, Saatkamp et al. 2019, Pipatpongpinyo et al. 2020). Unlike cultivated rice, the seed longevity of weedy rice is particularly important because it allows seeds to remain viable in the soil seed bank across seasons, ensuring population persistence even when environmental conditions are unfavourable for immediate germination (Baek and Chung 2012). Weedy rice can compete with cultivated rice because it is taller, grows faster, and has more tillers (Kanapeckas et al. 2016, Fogliatto et al. 2020, Wang et al. 2023). The red pericarp trait is a common characteristic of weedy rice in any geographic location (Kanapeckas et al. 2016, Fogliatto et al. 2020, Wang et al. 2023). A recent study suggests that seed pigment traits, including the red pericarp, likely influence plant adaptation through pleiotropic effects—that is, single genetic changes that simultaneously affect multiple traits, such as seed coat structure, dormancy, and germination speed (Tang et al. 2025).

The red pericarp trait is regulated by two dominant genes, Rd and Rc (Furukawa et al. 2007). The Rd gene encodes dihydroflavonol-4-reductase (DFR), and Rc encodes a basic helix-loop-helix (bHLH) transcription factor. Rc is a key gene involved in pericarp pigmentation, and its non-functional allele, rc, results in a white pericarp (Furukawa et al. 2007). The change from the red pericarp of wild rice to the white pericarp of cultivated rice resulted from positive selection for a single frame-shift deletion (14 bp) within the Rc gene, which is found in 97.9% of white rice cultivars today (Sweeney et al. 2007). Rd is responsible for red pigment synthesis, and its non-functional allele, rd, results in the absence of red pigments, leading to the brown pericarp (Furukawa et al. 2007). The pericarp is red when both Rd and Rc are functional, white when both are non-functional, or only Rd is functional, and brown when only Rc is functional. In the flavonoid biosynthetic pathway, dihydroflavonols are reduced by DFR and subsequently converted by anthocyanidin synthase (ANS). The resulting intermediates are then polymerized by leucoanthocyanidin reductase (LAR) or anthocyanidin reductase (ANR) to form proanthocyanidins (condensed tannins) in the pericarp and aleurone layers (Tang et al. 2025). The Rc gene regulates the expression of LAR, ANR, and other downstream structural genes (Tang et al. 2025). In addition to their role in seed dormancy, red pericarp pigments such as proanthocyanidins have antioxidant properties and may modulate reactive oxygen species (ROS) levels, potentially contributing to tolerance against various environmental stresses (Oki et al. 2002, Dixon et al. 2005, Maeda et al. 2007, Corbineau 2022). Recent studies in sunflower have also shown that pigmented seed coats and pericarps can influence dormancy and stress responses through physical and physiological mechanisms (Dominguez et al. 2016, 2019).

Since seeds of all known accessions of the wild rice ancestor O. rufipogon have a red pericarp (Sweeney et al. 2007, Izawa 2017), the white pericarp of cultivated rice is considered a domestication trait (Sweeney et al. 2006, Furukawa et al. 2007). In Japan, no wild rice species are native, and it is considered that rice was introduced from the Asian continent around the 3rd century BCE (Sato 1992, D’Andrea et al. 1995). Although the pericarp is white in most modern and traditional rice cultivars in Japan, the prevalence of red pericarp among weedy rice suggests that this trait plays an important adaptive role. Genomic scans based on whole genome sequencing have detected positive selection in genomic regions containing Rc and Rd during de-domestication (Imaizumi et al. 2021).

The association between the red pericarp and seed dormancy in weedy rice has been suggested: quantitative trait locus (QTL) analysis of seed dormancy between the weedy rice accession ‘SS18-2’ and the cultivated indica accession ‘EM93-1’ identified locus qSD7-1 at the Rc gene position. However, the function of Rc may vary depending on the genetic background (Gu et al. 2011). Rc has also been shown to contribute to seed longevity in weedy rice, supporting its potential role in seed bank persistence (Pipatpongpinyo et al. 2020). However, since both ‘SS18-2’ and ‘EM93-1’ have Rd, its effect alone and its potential epistatic interaction with Rc have not been evaluated.

This study was designed to clarify the roles of the red pericarp trait, governed by the Rd and Rc genes, in the genetic background of a japonica cultivar. Specifically, we developed near-isogenic lines (NILs) carrying Rc or Rd alleles in ‘Koshihikari’ background and evaluated their effects on seed dormancy, freezing tolerance, and overwintering ability under field conditions. The implications for weedy rice are discussed as potential adaptive significance, rather than direct evidence.

Materials and methods

Plant materials

We used japonica ‘Koshihikari’, indica ‘Kasalath’ (donor of the Rd and Rc genes), and ‘Toyama-aka 71 gou’ (Yamaguchi et al. 2015, designated as KH-RdRc in this study), a NIL carrying the Rd and Rc alleles of ‘Kasalath’ in the genetic background of ‘Koshihikari’; ‘Toyama-aka 71 gou’ was obtained from the Toyama Agricultural Research Center. Two NILs, KH-Rd and KH-Rc, were developed in the ‘Koshihikari’ background (see ‘Developing NILs’ subsection). The weedy rice accession ‘JP_1177’ was collected in Nagano Prefecture, Japan (Imaizumi et al. 2021). Genotypes and phenotypes of all materials are summarized in Table 1.

Developing NILs

Using a simplified biotron breeding system (Tanaka et al. 2016), we developed KH-Rd and KH-Rc in the ‘Koshihikari’ background. F_1_ plants were obtained by crossing ‘Koshihikari’ and KH-RdRc. F_2_ seeds were obtained by self-pollinating the F_1_ individuals; 88 F_2_ individuals were genotyped at the seedling stage using DNA markers for Rc and Rd, and individuals with homozygous genotypes were selected (the DNA markers used for genotyping are described in ‘Development of DNA markers and genotyping’). One individual carrying RdRd/rcrc (designated KH-Rd) and one carrying rdrd/RcRc (KH-Rc) were selected and used as Rd and Rc NILs, respectively (Table 1). Seeds from the F_3_ generation and onwards were used in the subsequent experiment for seed multiplication. As illustrated in Fig. 1 (adapted from Yamaguchi et al. 2015 for KH-RdRc), the graphical representation indicates the size and chromosomal position of ‘Kasalath’ introgressed segments in each NIL. Non-isogenic regions outside the target loci are shown in Fig. 1.

NILs tested. KH, ‘Koshihikari’; KH-Rd and KH-Rc, lines developed in this study; KH-RdRc, ‘Toyama-aka 71 gou’. (a) Graphical genotype of KH-RdRc according to Yamaguchi et al. (2015). Photographs show hulled and de-hulled rice. (b) Enlarged image of de-hulled rice. Only KH-RdRc had red pericarp. Alt text: Photographs and schematic diagrams of near-isogenic rice lines carrying different combinations of the Rc and Rd alleles in the Koshihikari background, showing pericarp colour differences among lines, with only the line carrying both Rc and Rd exhibiting a red pericarp.

Development of DNA markers and genotyping

DNA markers were designed to identify the genomic segments containing Rc and Rd derived from ‘Kasalath’. The data on rice genome polymorphism were obtained from TASUKE + for RAP-DB (https://agrigenome.dna.affrc.go.jp/tasuke/ricegenomes/) (Kumagai et al. 2013, 2019). DNA markers were designed in Primer 3 software (https://bioinfo.ut.ee/primer3-0.4.0/) for insertions and deletions in the target genomic regions. RM8129 and RM3143 were designed in the International Rice Genome Sequencing Project (2005) and were confirmed to distinguish between ‘Kasalath’ and ‘Koshihikari’ alleles (Yamaguchi et al. 2015). The DNA markers used for genotyping, including the specific primers and PCR conditions, are shown in Table 2.

Genomic DNA was extracted using the DNA Sui Sui S kit (Rizo Inc.). PCR amplification was performed using either EmeraldAmp MAX PCR Master Mix (Takara Bio Inc.) or KOD One PCR Master Mix (Toyobo Co. Ltd.), following the manufacturers’ protocols. PCR products were separated by electrophoresis in 2% or 3% agarose gels to detect polymorphisms.

Sequence comparison of target genes

To compare the Rd and Rc gene sequences of Japanese weedy rice (Imaizumi et al. 2021) and Chinese weedy rice (Qiu et al. 2017), sequencing reads were aligned to the ‘Nipponbare’ reference genome in BWA-MEM (International Rice Genome Sequencing Project 2005). Variants were called using GATK HaplotypeCaller (McKenna et al. 2010) to generate variant call format (VCF) files. These files were visualized in TASUKE + (Kumagai et al. 2013, 2019). ‘Kasalath’, ‘Koshihikari’ and wild rice relatives were compared on the TASUKE + website for RAP-DB (Kumagai et al. 2013, 2019, URL: https://agrigenome.dna.affrc.go.jp/tasuke/ricegenomes/).

Growth conditions and seed harvesting

The lines were grown in 2022, 2023, and 2024 in a paddy field (36°01′25.0″N, 140°06′25.5″E) in Kannondai, Tsukuba City, Ibaraki Prefecture, Japan. The experimental site is located in a temperate monsoon climate zone, characterized by hot, humid summers and mild winters. Climatic data for the 2022–2024 growing seasons are shown in Supplementary Figure S1. Seedlings were transplanted at 15-cm × 30-cm spacing with a base fertilizer of 50 kg N/ha. For the weedy rice accession ‘JP_1177’ and ‘Kasalath’, which have the seed shattering trait, a mesh net with 1 mm × 2 mm openings was placed over the plants 3 weeks after heading to collect both retained and naturally shed seeds. Seeds were harvested 6 weeks after heading, air-dried, and stored in a desiccator at 20°C with 30% relative humidity until used in experiments. For each cultivar/line and treatment, eight individual parent plants were used. Seeds harvested from each plant were pooled by accession and treatment. Each plant was considered a biological replicate for phenotypic measurements, while pooled seeds were used for germination, overwintering, and freezing tolerance tests. The days to heading, culm length, panicle length, and number of panicles per plant for each accession and year are presented in Supplementary Table S1. Tukey −Kramer’s honest significant difference test was performed using the multcomp package in R. All statistical analyses were conducted in R v. 4.2.1 software (R Core Team 2022).

Temporal changes in seed dormancy under controlled conditions

To evaluate the temporal dynamics of seed dormancy after harvest, we conducted germination assays in 2022–2024 at six time points: 1, 2 (2023 only), 4, 8, 12, and 18 weeks after harvest. Seeds were stored in a desiccator at 20°C with 30% relative humidity until used. Twenty seeds were placed in a 5.5-cm-diameter Petri dish with 5 ml of distilled water and incubated in the dark at 15 or 30°C. Germination assays were conducted at both 15°C and 30°C because germination can be strongly influenced by temperature. Lower temperatures (e.g. 15°C) suppress germination, whereas higher temperatures (e.g. 30°C) promote germination. Responses at both temperatures were compared to evaluate differences among lines. The number of seeds that germinated within 14 days was recorded. The viability of ungerminated seeds was assessed by cold stratification at 5°C for 14 days, followed by incubation at 30°C for 7 days. Ungerminated seeds were pressed with a pestle, and soft, easily crushed seeds were considered unviable. The germination proportion was calculated as the number of germinated seeds divided by the total number of viable seeds after 14 days of incubation.

As our data on germination are proportional, rather than continuous measurements, differences in seed dormancy among accessions were evaluated using a generalized linear model (GLM) assuming a quasi-binomial distribution. Maximum likelihood estimates and 95% confidence intervals were used for statistical inference. Germination proportion was used as the response variable. Accession, incubation temperature, and weeks of dry storage after harvest were included as explanatory variables. Germination proportion was used as an indicator of temporal changes in seed dormancy. To identify any differences in the germination proportion between the lines, analysis of deviance was conducted using the car package in R (version 4.2.1; R Core Team 2022).

Seed freezing tolerance test

Seeds harvested in 2023 were tested in 2024. Seeds were kept in a desiccator at 20°C with 30% relative humidity for about 5 months and then in a refrigerator at 5°C. To minimize the effects of seed dormancy, seeds stored for more than one year were used. Twenty seeds per replicate were placed in a 5.5-cm-diameter Petri dish with 5 ml of distilled water and soaked at 5°C for 24 h. Excess surface moisture was removed with paper towels. The freezing treatment was conducted following the method of Baek and Chung (2012): the seeds were incubated at −10°C for 24 h and then at 5°C for 24 h. This treatment was applied one or three times. Then, seeds were incubated at 30°C for 10 days, and the number of germinated seeds was recorded to assess seed viability. Ungerminated seeds were cut in half with a razor blade so that the embryo was visible, soaked in 0.1% 23,5-triphenyltetrazolium chloride, wrapped in aluminium foil, and kept in the dark at 30°C for 1 day. Seeds with red embryos were considered ungerminated viable, and seeds with uncoloured embryos were considered dead. The percentage of viable seeds was calculated as (number of germinated seeds + number of ungerminated viable seeds)/(number of germinated seeds + number of ungerminated viable seeds + number of dead seeds) × 100%. Differences among accessions were evaluated using a GLM assuming a quasi-binomial distribution, followed by Tukey–Kramer multiple comparisons using the multcomp package in R (version 4.2.1; R Core Team 2022).

Seed overwintering test

To evaluate the effects of Rc and Rd on overwintering ability under field conditions, we buried seeds in the field during autumn and winter and assessed the proportion of surviving seeds in the following spring. The test was applied to all lines, including ‘Koshihikari’ (KH; control), KH-Rd, KH-Rc, KH-RdRc, ‘Kasalath’, and ‘JP_1177’. Seeds were kept in a desiccator at 20°C with 30% relative humidity until the experiments were conducted. To address the issue that seeds with shallow dormancy may germinate before winter and be lost from the soil seed bank, seeds were also buried during winter, when low temperatures naturally prevent emergence, allowing evaluation of overwintering ability (see Supplementary Figure S3).

The test was conducted in 2022–2024 in a paddy field in Kannondai and characterized by Andosol (volcanic ash soil). As soil temperature at any depth can be estimated from air temperature (Hirota et al. 2002), the latter was used as an indicator. The lowest daily average air temperature during the burial period was −7.79°C in 2022–23, −6.39°C in 2023–24, and −6.49°C in 2024–2025; the trends of daily minimum temperatures are shown in Supplementary Figure S2.

A conceptual diagram of the autumn and winter seed burial experiments is shown in Supplementary Figure S3, and the burial setup is shown in Supplementary Figure S4. To simulate seed scattering and tillage burial in the field, 100 seeds were mixed with 750 ml of sterilized soil and placed in a stainless-steel strainer (16 cm diameter, 8 cm high, 0.42 mm wire diameter, 14 mesh), which was buried so that the top edge of mesh was flush with the soil surface, at a depth of approximately 8 cm. To prevent damage by birds, the strainer was covered with another stainless-steel strainer (15 cm diameter, 7 cm high, 0.42 mm wire diameter, 14 mesh). Autumn burial was conducted 1 week after harvest in September. Winter burial was conducted in early December, when outdoor temperatures had decreased sufficiently to prevent emergence.

Seedlings that emerged from the soil surface were recorded at 1-week intervals until March and were removed, whereas seeds that did not germinate remained inside the strainers throughout the experiment. In April, the buried strainers were retrieved, and all seeds within them were collected. Seeds that had not germinated at the time of collection were incubated at 30°C in the dark for 2–3 days, and those that germinated were counted. Seeds that remained ungerminated were left under the same conditions for an additional 2–3 days and re-examined. Seeds that did not germinate were cut in half with a razor blade to reveal the embryo, treated with 0.1% 23,5-triphenyltetrazolium chloride, and processed as described in the seed freezing tolerance test section to distinguish between viable and dead seeds. Seedlings that emerged before 3 March 2023, before 20 February 2024, or before 6 March 2025 (the dates after which no new germination was observed) were counted as seedlings that emerged during overwintering, those that emerged or remained viable after strainer retrieval were counted as seeds that survived overwintering, and those that were not viable were counted as dead seeds.

Pre-winter emergence proportion was defined as the proportion of seedlings that emerged before the onset of winter and died. Post-winter survival proportion was defined as the proportion of seeds that germinated or remained viable after strainer retrieval in spring. As both were proportional data, differences among accessions were evaluated using a GLM assuming a quasi-binomial distribution, followed by Tukey–Kramer multiple comparisons using the multcomp package in R (version 4.2.1; R Core Team 2022).

Results

Development of NILs KH-Rd and KH-Rc and their agronomic phenotypes

KH-Rd, with the functional Rd allele, and KH-Rc, with the functional Rc allele, were developed in the genetic background of ‘Koshihikari’; both alleles were derived from ‘Kasalath’. The graphical genotypes of these lines are shown in Fig. 1 (KH-RdRc adapted from Yamaguchi et al. 2015). No significant differences in major agronomic traits, including heading date, were observed among KH-Rd, KH-Rc, KH-RdRc, and ‘Koshihikari’, except that culm length was significantly greater in KH-RdRc than in ‘Koshihikari’ and KH-Rc (P < 0.05 by Tukey–Kramer test, Supplementary Table S1).

Sequence comparison of Rc and Rd alleles ‘Kasalath’ and rice accessions

Sequence comparison is shown in Fig. 2. The sequence of the ‘Kasalath’ Rd allele with its promoter region was identical to that of the weedy rice accessions tested. The ‘Kasalath’ Rc allele differed in nucleotide sequence from the weedy rice alleles tested, however, these variations in the ‘Kasalath’ Rc allele were also found in wild and weedy rice. Several SNPs and small indels were detected in the Rc region among ‘Kasalath’, weedy rice, and wild rice accessions. None of these polymorphisms resulted in loss of Rc function; all alleles encoded a functional bHLH transcription factor.

Polymorphic sites in Rd and Rc genes. Green boxes, coding regions. White triangles, variants in ‘Koshihikari’ vs. ‘Nipponbare’ reference genome; red triangles, variants in ‘Kasalath’ vs. ‘Nipponbare’. Asterisks, mutation sites among weedy rice accessions. Black dots, mutation sites in wild rice. Black arrows, sites of mutations that result in loss of gene function (Furukawa et al. 2007). Alt text: Schematic comparison of nucleotide polymorphisms in the Rc and Rd genes among Koshihikari, Kasalath, weedy rice, and wild rice, indicating coding regions, variant positions, and mutations affecting gene function.

Temporal dynamics of seed dormancy under controlled conditions

To compare the dormancy status among lines, we used germination proportion at each time point after harvest; its rapid increase indicated faster dormancy release. Significant differences were observed among each cultivar/line, temperature and cultivar/line:temperature verified by analysis of deviance (P < 0.01). At 15 °C, germination proportions of ‘JP_1177’ and ‘Kasalath’ were around 10% even 18 weeks after harvest in both years; at 30 °C, that of ‘Kasalath’ varied between the 2 years, and that of ‘JP_1177’ was the lowest among the lines in both years (Fig. 3). Clear varietal and line differences were observed between ‘JP_1177’ and lines with the KH genetic background. In comparison with ‘JP_1177’, KH-RdRc had a distinct germination pattern; its germination proportions increased more rapidly at both temperature in both years and reached over 50% in 2023% and 55% in 2024 at 30 °C 1 week after harvest. At both temperatures and in both years, germination proportion decreased in the order of KH-RdRc ≥KH-Rd > KH = KH-Rc. KH-RdRc also germinated earlier than KH from seeds harvested in 2022 (Supplementary Figure S5), whereas KH-Rc showed lower germination than KH-Rd and KH-RdRc shortly after harvest at 30 °C in the dark.

Trends in germination during storage. The proportion of viable seeds that germinated within 14 days was estimated using a generalized linear model (GLM) with a quasi-binomial distribution. Lines, maximum likelihood estimates of germination proportions; ribbons, 95% confidence intervals; dots, proportion of viable seeds that germinated within 14 days. Eight biological replicates were used. Red arrows highlight shallow dormancy in KH–RdRc at 30 °C one week after harvest, whereas blue arrows indicate deep dormancy in ‘JP_1177’ at 15 °C late in the storage period. Alt text: Line plots showing changes in seed germination proportion during dry storage after harvest at two incubation temperatures, highlighting shallow dormancy in near-isogenic lines and deep dormancy in weedy rice.

Seed freezing tolerance

After one freezing cycle, the survival proportions of all lines except ‘Kasalath’ were high (Fig. 4). After three freezing cycles, ‘Kasalath’ showed reduced survival, whereas the weedy rice accession ‘JP_1177’ maintained high survival, and ‘Koshihikari’ showed intermediate survival. Seed viability of the near-isogenic lines differed among lines; KH-Rd and KH-Rc showed lower survival, whereas KH-RdRc exhibited higher survival (Fig. 4). Overall, Rd or Rc alone did not enhance freezing tolerance, but together they improved freezing tolerance in the triple freezing treatment.

Freezing tolerance. Left panel, single freezing treatment; right panel, three consecutive freezing treatments for each cultivar and near-isogenic line (NIL). Seeds were soaked at 5 °C for 24 h, exposed to −10 °C for 24 h, returned to 5 °C for 24 h for once or three times, and then incubated at 30 °C to assess viability. Box plots show the median (central line) and interquartile range (box); whiskers indicate the 10th–90th percentiles. Jittered points represent individual biological replicates (n = 8; n = 7 for KH–Rd under three freezing cycles due to one missing replicate). Different letters above boxes indicate significant differences among cultivars/lines within each treatment (Tukey–Kramer test, P < 0.05). Alt text: Box plots comparing seed viability of rice cultivars and near-isogenic lines after one or three freezing treatments, showing enhanced freezing tolerance in the line carrying both Rc and Rd alleles.

Overwintering ability in soil under field conditions

The pre-winter emergence proportions of seedlings from seeds buried in September (1 week after harvest) are shown in Table 3. The survival proportions of seeds buried in September or December (3 months after harvest) are shown in Fig. 5 and Supplementary Figure S6. The pre-winter emergence was low in JP_1177, whereas it was high in KH-background lines, resulting in minimal overwinter survival (Table 3). Weekly seedling emergence dynamics are shown in Supplementary Figure S7. The post-winter survival proportion of ‘JP_1177’ was high, regardless of burial timing (Fig. 5). Survival of ‘Kasalath’ varied among years, and NILs showed low survival after September burial. In the winter burial treatment, post-winter survival proportion of KH-RdRc was comparable to that of ‘JP_1177’ (Fig. 5, Supplementary Figure S6). KH-Rd exhibited lower survival, whereas KH-Rc showed intermediate survival, indicating overwintering ability comparable to or slightly higher than that of ‘Koshihikari’ (Fig. 5).

Overwintering ability in soil under field conditions. Seed survival in April of the following year is shown for seeds buried in September (1 week after harvest; orange box plot) or in December (3 months after harvest; blue box plot). Box plots show the median (central line) and interquartile range (box); whiskers indicate the 10th–90th percentiles. Jittered points represent individual biological replicates (n = 8). Different letters indicate significant differences among cultivars/lines within each burial treatment (Tukey–Kramer test, P < 0.05). Alt text: Box plots showing overwintering seed survival of rice cultivars and near-isogenic lines buried in soil in autumn or winter, indicating higher overwintering survival in the line carrying both Rc and Rd alleles.

In summary, NILs sown in September germinated prematurely. Among NILs sown in December, only KH-RdRc had post-winter seed survival proportions comparable to those of ‘JP_1177’.

Discussion

This study demonstrates that, in a japonica (Koshihikari) background, the red pericarp trait conferred by Rc and Rd does not induce the deep seed dormancy characteristic of weedy rice, but does enhance seed freezing tolerance and overwintering ability. Our results are consistent with previous studies in indica backgrounds (e.g. Gu et al. 2011, Pipatpongpinyo et al. 2020), and further highlight the importance of genetic background in determining the function of these genes.

While seed dormancy and longevity are critical adaptive traits for weedy rice, cultivated rice typically has reduced dormancy and seed longevity due to selection for uniform germination and crop management (Nadir et al. 2017, Pipatpongpinyo et al. 2020). Our results demonstrated that the red pericarp may enhance seed survival under cold stress by improving freezing tolerance and overwintering ability.

The NILs developed in this study–KH-Rd and KH-Rc–provide a valuable resource for dissecting the functional roles of these genes in seed adaptation. The culm was significantly longer in KH-RdRc than in ‘Koshihikari’, as reported for ‘Toyama-aka71 gou’ by Yamaguchi et al. (2015). This difference may be due to linkage drag associated with the introgression of Rd and Rc alleles or with pleiotropic effects. Although most agronomic traits were similar among the lines, this difference should be considered when interpreting the adaptive significance of the red pericarp trait. Residual non-isogenic regions may remain outside the target loci, a limitation of the NIL approach.

These findings underscore the importance of considering both genetic background and trait interactions when evaluating the adaptive significance of seed dormancy and longevity in weedy rice. Future studies should include a wider range of genetic backgrounds and environmental conditions to validate these findings and clarify the molecular mechanisms involved.

The red pericarp trait is insufficient to explain deep seed dormancy in weedy rice

Genetic analyses of weedy rice have suggested a relationship between the red pericarp trait and seed dormancy (Cohn and Hughes 1981, Gu et al. 2011). In our study, none of the three NILs had the seed dormancy characteristic of weedy rice and no epistasis was detected for this trait. One possible explanation is the potential influence of other genes that are not involved in seed coat coloration, linked to the two target genes. The seed dormancy of KH*-Rd* and KH*-RdRc* was shallower than that of ‘Koshihikari’ (Fig. 4, Table 3), but the difference between all three NILs and ‘Koshihikari’ was smaller than that between ‘Koshihikari’ and ‘JP_1177’. Therefore, the red pericarp trait alone cannot fully explain the deep seed dormancy observed in weedy rice. Our conclusion was consistent with previous reports (Zhang et al. 2017, Pipatpongpinyo et al. 2020) that other QTLs in weedy rice are important in regulating seed dormancy, and that a single gene, including Rc, is insufficient to confer a sufficient level of dormancy. Seed dormancy QTLs from weedy rice (Gu et al. 2004) and an association between red seed coat and deep seed dormancy in wheat (Nilsson-Ehle 1914, Freed et al. 1976) have been reported. However, Mares and Himi (2021) noted that Tamyb10-A1, which controls red seed coat pigmentation in wheat, accounts for only part of the variation in seed dormancy and suggested possible interactions with other seed dormancy loci.

The functions of the red pericarp genes in ‘Kasalath’ may differ from those in wild or weedy rice. Using publicly available whole-genome data, we found no polymorphisms in ‘Kasalath’ Rd, including its promoter region. Polymorphisms in the ‘Kasalath’ Rc allele were also found in wild and weedy rice populations, suggesting that Rc in ‘Kasalath’ is not functionally distinct from those in wild or weedy rice, consistent with gene complementation analysis in previous studies confirming that Rc in ‘Kasalath’ is functional for red pericarp (Furukawa et al. 2007) and seed dormancy (Gu et al. 2011). The ‘Kasalath’ germination proportion was low at 15°C but good at 30°C. This may be explained by the fact that ‘Kasalath’ is an indica landrace originating from India, a region with a warm climate. Indica rice cultivars are generally adapted to higher temperatures and their germination or growth are often reduced at lower temperatures than japonica cultivars (Lv et al. 2016, Zhang et al. 2025). Therefore, ‘Kasalath’ may not have evolved cold tolerance, resulting in lower germination proportions at 15°C.

The shallow seed dormancy of KH-Rc and KH-RdRc indicates that the red pericarp trait alone does not confer sufficient dormancy to prevent germination before winter. The deep seed dormancy in weedy rice results from the combined action of multiple dormancy-enhancing genes (Gu et al. 2004, 2011, Pipatpongpinyo et al. 2020). Our results suggest that red pericarp may enhance seed longevity by improving low-temperature tolerance and overwintering ability in soil. While Rc may contribute to dormancy, it alone is not sufficient to provide the adaptive advantage of preventing premature germination in temperate regions. Genomic studies have identified regions under positive selection that overlap with loci associated with weediness traits, such as seed shattering, heading date, and emergence date, suggesting selection of multiple weediness traits during de-domestication (Li et al. 2017, 2023). Thus, our findings further support the notion that the red pericarp trait must act in concert with other weediness traits to enable the establishment and survival of weedy rice in temperate regions.

The red pericarp trait enhances seed freezing tolerance

Baek and Chung (2012) compared weedy rice with red pericarp and white rice cultivars and suggested a possible association between the red pericarp trait and freezing tolerance. We confirmed that the red pericarp trait enhances freezing tolerance and thereby contribute to overwintering ability in soil, specifically seeds with japonica-derived genetic background.

After a single freezing, the viability of ‘Kasalath’ seeds was the lowest among the lines tested. After three cycles of freezing, tolerance of KH-Rd or KH-Rc was slightly lower than that of ‘Koshihikari’, which may be attributable to the introduction of ‘Kasalath’ genomic regions linked to these genes. Freezing tolerance was markedly higher in KH-RdRc than in ‘Koshihikari’ (Fig. 5). The interaction between these two substituted genome fragments is strongly associated with the red pericarp trait, at least through epistasis between Rd and Rc, and this strongly supports the hypothesis that the red pericarp trait pleiotropically enhances freezing tolerance.

However, our analysis did not include weedy rice with the white pericarp, such as aus- or indica-derived biotypes. Therefore, these findings should not be generalized to all types of weedy rice. Our analysis may help explain why the red pericarp trait is a common characteristic of weedy rice (Kanapeckas et al. 2016, Fogliatto et al. 2020, Wang et al. 2023). Further studies comparing weedy rice with red and white pericarp from diverse genetic backgrounds are needed to determine whether the association between red pericarp and freezing tolerance holds more broadly.

The red pericarp trait enhances seed persistence in soil by improving overwintering ability

When seeds were buried in September, more than half of the seeds of ‘Koshihikari’ and its NILs germinated before winter and could not survive the low winter temperatures (Supplementary Figure S3). When seeds were buried in December, the post-winter seed survival of KH-RdRc was comparable to that of weedy rice ‘JP_1177’ (Fig. 5, Supplementary Figure S6). The overwintering ability of lines with shallow seed dormancy could be evaluated by burying their seeds in the ground during the winter, when environmentally induced dormancy occurs. Using this method, we found that the red pericarp trait enhances seed overwintering ability, thereby promoting seed persistence in the field. Curiously, although ‘Kasalath’ is originally a tropical crop, its alleles contributed to overwintering ability.

Using NILs of four seed dormancy QTLs including Rc, Pipatpongpinyo et al. (2020) found that these QTLs additively regulate seed longevity in the field. The red pericarp had a major effect on seed longevity in the soil when interacting with the other seed dormancy genes. However, since all their materials had the Rd allele, they did not investigate the interaction between the functional alleles of Rc and Rd. Our results suggest this interaction increases freezing tolerance and overwintering ability. We propose that the red pericarp trait improves overwintering ability, highlighting the adaptive significance of Rd in addition to that of Rc.

Adaptive significance of the red pericarp trait

While the red pericarp trait may contribute to tolerance to various stresses by conferring antioxidant properties, as discussed above, our study primarily highlights its potential adaptive significance in enhancing freezing tolerance and overwintering survival. Our experiments mainly addressed freezing tolerance and overwintering ability; further research is needed to clarify whether the red pericarp also contributes to tolerance against other environmental stresses via antioxidant activity and ROS modulation, as suggested by previous studies (Oki et al. 2002, Dixon et al. 2005, Maeda et al. 2007, Dominguez et al. 2016, 2019, Corbineau 2022). Our findings provide evidence that the red pericarp trait, governed by the Rc and Rd genes, may confer an adaptive advantage to weedy rice by enhancing seed freezing tolerance and overwintering survival in temperate environments. However, these findings were obtained in a ‘Koshihikari’ background and should not be interpreted as direct evidence for weedy rice. The function of Rc may vary depending on the genetic background (Gu et al. 2011). This limitation should be considered when extrapolating our results to weedy rice populations. This trait may contribute to the persistence and spread of weedy rice populations in regions where they are subjected to seasonal cold stress, providing a potential explanation for the reported prevalence of the red pericarp among weedy rice (Gross et al. 2010, Li et al. 2022, Wang et al. 2023). To validate our findings in natural weedy populations, further studies using weedy rice materials are required. By elucidating the functional basis of this adaptive trait, our study contributes to the understanding of weed evolution and may offer new perspectives for the management of weedy rice in agricultural systems.

Supplementary Material

plag014_Supplementary_Data