https://orcid.org/0009-0004-1055-6886
Figures
Abstract
Rice cultivation in the northern and northeastern districts of Bangladesh faces cold stress during the seedling and reproductive stages that drastically reduces yield. As a precursor to generate improved elite lines, a diverse panel of rice germplasm was screened to identify genotypes with resilience to low temperature at both key developmental phases. Seedling-stage tolerance was assessed using an artificial cold-water tank, while reproductive-stage cold tolerance was evaluated under both natural field conditions and controlled cold screening facilities. Two breeding lines - BR8907-B-1–2-CS1–4-CS2-P3-4 and BR8909-B-12–2-CS1–4-CS2-P2-3–2, demonstrating minimal leaf discoloration and the highest survival rates, and lacking the known QTL (qSCT1) for seedling stage cold tolerance indicated contributions from other QTLs or genetic factors. Five genotypes (Bhutan, IR83222-F11-173, Rata Boro, BRRI dhan74, BR11712-4R-227) showed tolerance at the reproductive stage, while three lines (Bhutan, BR11712-4R-227, and BR12266-44-11-32-5-1-1-HR10-B) showed moderate tolerance at both seedling and reproductive stages, indicating stage-specific responses. The genotypes BR10317-5R-25, IR18A1859, and BRRI dhan28 were consistently susceptible at both stages. Principal Component Analysis (PCA) indicated that seedling and reproductive traits contributed jointly to variation under field conditions but independent under controlled environments, reflecting stage-specific genetic regulation. These findings highlight the complex, stage-dependent nature of cold tolerance and the importance of stage-specific screening to inform effective breeding strategies for enhanced cold resilience in rice.
Citation: Akter F, Biswas PS, Islam AKMA, Raihan MS, Rahman MM, Iftekharuddaula KM, et al. (2026) Stage-specific screening reveals differential resilience response to cold stress in rice. PLoS One 21(4): e0338290. https://doi.org/10.1371/journal.pone.0338290
Editor: Md Ashrafuzzaman, Shahjalal University of Science and Technology, BANGLADESH
Received: November 20, 2025; Accepted: March 20, 2026; Published: April 15, 2026
Copyright: © 2026 Akter et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All codes used in this study have been uploaded into the public repository Zenodo with https://doi.org/10.5281/zenodo.19366327.
Funding: This was funded by Gates Foundation under contract ID: INV-002860.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Rice is the staple food for more than half of the world’s population, and is cultivated in over 100 countries. As the global population has already surpassed 7.5 billion and is projected to reach nearly 11 billion by 2100 [1], ensuring sufficient rice production has become increasingly important. In Bangladesh, rice contributes approximately 97% of total food grain production for a population of 172.92 million people [2]. By 2050, the population of Bangladesh is expected to reach around 215.4 million, requiring an estimated 44.6 million tons of milled rice to meet national demand [3]. Rice production, however, faces growing challenges due to declining arable land and increasing climate-related stresses, including drought, salinity, flooding, heat stress, cold stress and waterlogging. Among these constraints, low-temperature stress (LTS) is a major abiotic factor that limits growth, productivity, and yield of rice crop (S1 Fig in S1 File), affecting nearly 10% of the global rice-growing area (130 million ha) [4]. Rice cultivated during winter season in tropical and subtropical Asia is particularly sensitive to cold stress [5], especially during the seedling and reproductive stages [6,7]. In Bangladesh, approximately two million hectares of rice-growing areas are affected by low-temperature stress (LTS) each year [8].
In addition, the genetic makeup of rice has been optimized for warm environments through long-term domestication and selection process. Therefore, LTS is one of the primary abiotic stresses causing severe rice yield losses [9]. The optimum temperature range for normal growth and development of rice is between 25–30℃. Deviations beyond this critical range, rice plants are affected at variable degrees depending on the growth stage [10]. It can damage the plant during any developmental stage such as, germination, seedling, vegetative, reproductive and maturity [11]. Temperature below 20℃ significantly impair the germination process, reduce both the rate and percentage of germination, resulting in weak seedling establishment and poor crop stands [10]. During the winter season in northern Bangladesh, temperatures drop below 10°C at the seedling stage, causing leaf discoloration, yellowing, rolling, and stunted growth, often leading to seedling mortality and increased production costs [12]. Among all developmental phases, the reproductive stage is the most sensitive to cold stress and plays a major role in determining grain yield [13] Reproductive -stage cold stress thresholds are 20℃ and 15℃ for cold-sensitive and cold-tolerant varieties, respectively [14]. Reproductive development in rice comprises ten distinct stages, beginning with the initiation of panicle primordium and ending at flowering. Panicle initiation typically occurs approximately 30 days prior to flowering [10]. Cold stress during this period can have irreversible consequences. The most critical window of sensitivity occurs during the mid- reproductive phase, particularly at the young microspore stage, which takes place 10–12 days before heading. Exposure to low temperatures during this narrow developmental window induces male sterility [13], through malformed embryo sacs which inhibit pollen development and ultimately reduced spikelet fertility.
Cold stress during the booting stage disrupts pollen formation and leads to pollen abortion, resulting in no occurrence of anthesis. In contrast, when low temperature coincides with the flowering stage, pollen grain may form normally, but pollen tube germination is inhibited, leading to failure of fertilization. These cause panicle degeneration, incomplete panicle exsertion, and spikelet sterility, which ultimately lower the grain yield of rice [7,11,14–16]. These stage-specific disruptions ultimately cause high spikelet sterility and substantial yield losses [12]. Importantly, each reproductive sub-stage responds differently to cold stress, indicating that distinct physiological processes and regulatory mechanisms are involved [17].
The diverse and stage-dependent effects of cold stress reflect the involvement of complex and dynamic molecular pathways, including the differential expression of stress-responsive genes, hormonal signaling networks, and metabolic adjustments [9]. Because the physiological symptoms and underlying molecular mechanisms vary markedly from seedling to reproductive stages—and even among individual reproductive sub-stages, unique and stage-independent screening approach may fail to capture true genotypes having cold tolerance. Therefore, stage-specific cold stress screening is essential for accurately identifying tolerant genotypes capable of sustaining productivity under low-temperature stress conditions.
Screening of rice genotypes for cold tolerance under only natural field conditions may not be efficient because the natural climate is unpredictable in terms of its intensity, duration, and timing. The success rate of screening cold tolerance in natural field conditions is very low. This may be due to the complexity of cold stress tolerance traits, low genetic variance of yield components due to stress, and lack of proper selection criteria [18]. On the other hand, screening conducted under controlled (artificial) conditions may perform differently from actual field conditions. It’s, therefore, desirable to consider both field and artificial conditions when screening for tolerant genotypes [19]. This approach facilitates the large-scale identification and selection of cold-tolerant genotypes. Previous researchers used a reliable method of phenotyping for cold tolerance under both natural and artificial conditions [7]. The objective of this study was to screen for cold-tolerant genotypes by evaluating their performance at two key stages: the seedling stage, under controlled (artificial) conditions, and the reproductive stage, under both natural and artificial conditions.
Materials and methods
Plant materials
A diverse panel of 46 rice genotypes, including advanced breeding lines, local landraces, and BRRI varieties, was used in this study (Table 1). The materials were evaluated for cold tolerance at the seedling and reproductive stages. Seedling stage screening was performed using cold-water tanks and reproductive stage screening was performed in the phytotron facilities at Plant Breeding Division of the Bangladesh Rice Research Institute (BRRI), Gazipur. Reproductive stage cold screening was also performed under natural cold condition in the experimental fields of BRRI. A total of 34 genotypes and four check varieties were tested for seedling stage cold tolerance. Based on the seedling-stage screening results, a subset of genotypes was selected for further evaluation under cold stress at the reproductive stage. This subset included five genotypes showing susceptible reactions, four entries showing moderately susceptible reaction, and five genotypes showing moderately tolerant reactions. These selected genotypes were evaluated under cold stress at reproductive stage along with two known susceptible genotypes (IR18A1859, BRRI dhan28), two known tolerant genotypes (IR83222-F11-173, BRRI dhan67), four untested advanced lines, and four untested BRRI varieties.
Screening of cold tolerance at seedling stage
A total of 34 breeding lines were evaluated under artificial cold stress conditions at 13°C, using cold-water irrigation in a tank setup [20]. BRRI dhan67 and HBJ B VI were used as cold-tolerant check varieties, while BR1 and BRRI dhan28 were included as susceptible check varieties [21]. Each genotype, including the checks, was represented by 10 seedlings planted in single-row plots with 3.0 cm spacing. The seedlings were grown in flat-bottomed plastic trays (dimensions: 60 cm × 30 cm × 2.5 cm) containing fertilized soil free of crop residues and gravel. A total of 34 breeding lines, along with four check varieties, were evaluated using a row column design. The experimental units were arranged in trays, each of which was subdivided into two blocks, serving as blocks within the replication (S2 Fig in S1 File).
The experiment was conducted in three batches (Batch-1, seed sown on 6/22/2022; Batch-2, seed sown on 7/30/2022; Batch-3, seed sown on 7/24/2024) with four replications in Batch-1 and Batch-2 and six replications in Batch-3. A thin layer of finely granulated soil was applied after sowing to cover the germinated seeds. The seedlings were allowed to grow until the 3-leaf stage (8−12 days), (S3 Fig in S1 File) and then the trays were placed into the cold-water tank [22] that was pre-set at 13°C (S4 Fig in S1 File). In Batch-1, the temperature in the cold-water tank was not stable throughout the tank. The water temperature at six positions (four points at the four corners and two points across the middle of the tank) within the tank ranged from 10.3 to 14℃. Then, an aquarium internal filter motor (model: SOBO WP-1200F) of AC 220V 50 Hz 15 watts was installed into the tank to circulate the movement of water to stabilize the temperature close to 13℃ all over the tank. The temperature was recorded five times daily with an interval of 3 hours (6 AM, 9 AM, 12 PM, 3 PM, 6 PM) within the period of cold treatment (S5 Fig in S1 File). The duration of cold treatment across three batches ranged from 8 to 14 days. The water level was maintained at 2 cm above the soil level to keep the trays submerged. Cold stress symptoms were assessed using a subjective leaf discoloration (LD) scale ranging from 1 to 9, following the protocol outlined by Shim et al. [23] and IRRI [24] as shown in Table 2. LD scoring was conducted at the point when the susceptible check variety, BR1, exhibited complete mortality; normally this is done within six to ten days after cold treatment starts (Batch 1: 6 days, Batch 2: 14 days, Batch 3: 10 days) (S6 and S7A Figs in S1 File) [21]. After recording the LD score, the trays were placed at ambient temperature in a sunny place for recovery from cold stress. The survival count was recorded from each plot after a seven-day recovery period (S7B Fig in S1 File). The survival rate in percentage was calculated as the percentage of recovered green plants to the total number of plants used in the cold treatment.
Screening of cold tolerance at the reproductive stage
Screening of 22 diverse rice genotypes was carried out for reproductive stage cold tolerance along with three standard check varieties: BRRI dhan28 as susceptible, and BRRI dhan67 and IR83222-F11-173 as a moderately tolerant check and tolerant check, respectively. Screening was conducted under two different conditions: natural field conditions during the Boro season and artificial cold screening facilities (Phytotron) under cold stress and non-stress conditions.
Field condition (natural condition)
To carry out field screening, one set of genotypes was sown on 21 October 2024 (34 days earlier than the regular sowing). This ensured the panicle initiation (PI) to booting stage of the crop would be exposed to low temperature from the end of January to the first week of February. This was the cold stress plot. A second set of the same genotypes was sown in the regular Boro season on 25 November and was treated as a non-stress plot. Flowering window and temperature profile of 25 genotypes were presented in S8 Fig and S9 Fig in S1 File respectively, for cold stress (A) and non-stress (B) conditions. Data on plant height, effective tiller number, days to heading, and panicle length were measured; vegetative stage score (VegS), panicle degeneration score (PDS), panicle exsertion score (PES), and spikelet fertility score (SFS) were recorded based on SES (Standard Evaluation System) presented in Table 3. The rate of reduction (RR) was calculated by the following method: RR = {(Mean value of nonstress – Mean value for cold stress)/mean value of non-stress} × 100%.
Cold screening in phytotron
The same set of genotypes were evaluated in a artificial cold screening facilities (phytotron) with controlled air (15.9-19.14℃, average 17℃) and water (16.66-20.24℃, average 18℃) conditions, considered as cold stress (S10A Fig in S1 File) following a RCB design with 3 replications. The phytotron setup consisted of two water tanks, wherein a total of 100 entries can be accommodated for cold screening at the reproductive stage. Thirty-day-old seedlings of each genotype were transplanted
into six plastic pots with a single seedling per pot. Three pots were treated as cold stress, and three pots were used as a non-stress control. Plants were grown in ambient temperature in the net house until the meiotic phase of the reproductive stage began. The reproductive stage was determined by the distance between the ligule of the flag leaf and that of the penultimate leaf [10], considering an interval of −3 (flag leaf ligule below the penultimate leaf ligule) to +10 cm (flag leaf ligule above the penultimate leaf ligule) as indicative of this stage [21]. In the reproductive stage, three pots treated as cold stress pots were labeled and placed in the cold-water tanks in the phytotron for 10 days (S11 and S12 Figs in S1 File); the other three pots were kept at ambient temperature as a non-stress control. The temperature of air and water in the phytotron (S10A Fig in S1 File) and ambient temperature (S10B Fig in S1 File) was recorded five times in a day at 3-hour intervals. After 10 days of cold treatment, the cold stress pots were further placed at ambient temperature until the plants matured. A similar set of data, as well as field conditions, was collected from the labeled tillers for treated pots and six tillers from non-treated pots, and each panicle was individually harvested.
Trait genotyping
Trait specific genotyping was performed with most cold-tolerant and most cold-susceptible rice genotypes identified from seedling and reproductive stage cold screening. Based on the screening results, a total of twelve genotypes at the seedling stage (two cold-tolerant, eight moderately tolerant, and two cold-susceptible) and eighteen genotypes at the reproductive stage (five tolerant, eight moderately tolerant, and five susceptible) were used for trait genotyping analysis. The response to cold stress of the most tolerant and susceptible genotypes at the seedling stage was validated using four SNP markers linked to the qSCT1 locus, a previously reported QTL for seedling-stage cold tolerance (snpOS00403, snpOS00404, snpOS00406, and snpOS00407) derived from the Korean donor Gymbubyo (Table 6). The favourable alleles C:C, A:A, T:T, C:C for snpOS00403, snpOS00404, snpOS00406, and snpOS00407, respectively. Alternative genotypes at each locus represented as unfavorable allele. Similarly, the presence of reproductive stage cold tolerance-specific known QTLs qPSST3 was monitored using SNP markers associated with qPSST3 (snpOS00815, snpOS00816, snpOS00817), qPSST6 (snpOS01157, snpOS01158), qPSST7 (snpOS00818, snpOS00819, snpOS00820, snpOS00821), and qPSST9 (snpOS00822, snpOS00823, snpOS00824, snpOS00825, snpOS00826). Favorable alleles were identified as G:G for qPSST3_1, qPSST3_2, and qPSST3_3; C:C for qPSST7_1 and qPSST7_4; G:G for qPSST7_2; T:T for qPSST7_3; A:A for qPSST9_1 and qPSST6_1; C:C for qPSST9_2, qPSST9_3, and qPSST9_5; and G:G for qPSST9_4 and qPSST6_2. Alternative genotypes at each locus were considered unfavorable alleles.
Data analysis
Descriptive statistics- including mean, range, standard deviation, and a two-stage analytical approach were employed in R version 3.2.1 to estimate LSD (least significant difference), heritability, BLUE (best linear unbiased estimation), and pBLUP (phenotypic best linear unbiased prediction) values. Grouping of genotypes by scatter plot was performed in Minitab.
Spearman’s rank correlation coefficient was analyzed using the cor.test () function. The formula for calculating the Spearman correlation is as follows:
where = Spearman correlation coefficient; di = the difference in the ranks given to the two variables’ values for each item of the data; n = total number of observations.
Principal component analysis (PCA) was performed using the prcomp() function. The grouping of germplasms was done by the hierarchical clustering method using Euclidean distance, and clustering was done by Ward’s minimum variance method (Ward.D2) via the hclust() function.
All analyses were performed using R statistical software (version 4.3.1) within the RStudio integrated development environment (IDE).
Results
Cold tolerance at the seedling stage
Leaf discoloration (LD) and survival percentage are the two important traits for discriminating cold-tolerant lines from susceptible lines under cold stress. The mean LD score recorded at 10 days of cold stress showed a wide range of variation among the genotypes (3.7 to 8.8), and the survival rate after seven days of recovery from cold stress also varied from 0 to 37% (Table 4).
The susceptible checks, BRRI dhan28 and BR1, exhibited high LD values (8.0 and 8.8, respectively) coupled with very low survival rates (0.9% and 0%). In contrast, the moderately tolerant check BRRI dhan67 and the tolerant check Hbj. B. VI recorded lower LD scores (5.0 and 5.7) and higher survival percentages (20.3% and 17.4%). Analysis of variance indicated a highly significant genotype effect for both traits (p = 3.02 × 10 ⁻ ⁸ for LD; p = 2.04 × 10 ⁻ ⁴ for Survival%). The coefficient of variation was 12.59% for LD, reflecting high experimental precision, whereas Survival% showed a CV of 29.97%, indicating substantial relative variability among genotypes.
A scatter plot constructed based on combined BLUE for LD score and survival rate of 38 breeding lines, including check varieties, across the three batches presented in Fig 1A, and detailed data across three batches were presented in S1 Table in S1 File. Out of 38 genotypes, two genotypes, viz., BR8907-B-1–2-CS1–4-CS2-P3-4 and BR8909-B-12–2-CS1–4-CS2-P2-3–2, obtained the lowest LD score (3.7 and 3.9) and highest survival rate (34.4-37%) in all three batches, indicating these are highly tolerant at the seedling stage. Eight genotypes (Hbj. B. VI, Bhutan, BR11318-5R-106, BR11712-4R-227, BR12266-44-11-32-5-1-1-HR10-B, BRRI dhan58, BRRI dhan89, and BRRI dhan67) formed a moderately tolerant group having LD scores of 4.9-6.5 with 17.4-31% survivability. Pictorial view of several seedling stage cold tolerant lines was shown in Fig 1B. Thirteen genotypes were found as moderately susceptible (LD score 5.4-7.5 with survival 6.4-15.8%), and 15 genotypes showed susceptible (LD score 5.7-8.8 with survival 0-5.7%) to cold stress at seedling stage.
The LD score exhibited the highest variability among the evaluated traits, with genotypes fitting an approximately normal distribution (Fig 2). Survival percentage, in contrast, showed a narrower distribution and was skewed towards lower values. A significant and negative correlation was observed between LD score and survival percentage (Spearman’s r = −0.708, p < 0.0001), indicating that genotypes with higher LD scores showed susceptibility to cold stress and tended to exhibit lower survival percentages.
An analysis of variance (ANOVA) was conducted to evaluate the effects of genotype, batch, and their interaction on leaf discoloration (LD) and survival percentage under cold screening conditions at the seedling stage (Table 5). A highly significant effect of genotype was observed both for LD and survival rate, indicating distinct LD and survival rate responses among genotypes. Batch and the genotype-by-batch interaction showed no significant interaction for LD and survival percentage, suggesting that the genotypes maintained their relative performance across different batches.
Seedling stage cold tolerance-specific genotyping for qSCT1 with snpOS00403, snpOS00404, snpOS00406, and snpOS00407 exhibited distinct allelic variation (Table 6). BR11712-4R-227 carried favorable alleles at all four loci and displayed a moderately tolerant response during seedling-stage, indicating a partial but consistent contribution of qSCT1. Similarly, Bhutan and Hbj. B. VI were moderately tolerant and possessed favorable alleles at snpOS00403 and snpOS00407, suggesting incomplete genetic fixation for seedling-stage cold tolerance. Notably, BR8907-B-1–2-CS1–4-CS2-P3-4 and BR8909-B-12–2-CS1–4-CS2-P2-3–2 exhibited strong seedling-stage cold tolerance despite lacking favorable qSCT1 alleles, indicating the involvement of additional QTLs or alternative genetic control. Conversely, susceptible genotype BRRI dhan28 predominantly carried alternative allelic combinations. Most importantly, the susceptible check BR1 possessed favorable alleles for all four SNPs of qSCT1, suggested that the qSCT1 haplotype alone is insufficient to confer cold tolerance under the climatic landscape of Bangladesh, and additional loci or favorable allelic interactions are necessary for functional expression of resilience to cold stress.
Cold tolerance at the reproductive stage
Under cold stress conditions, all the genotypes flowered between 19 February and 28 March 2024 (S8A Fig in S1 File), indicating that panicle initiation (PI) to booting occurred from the third week of January to the third week of February. During this period, temperatures ranged from 8.7°C to 31.4°C, with an average of 19.8°C (S9A Fig in S1 File). Notably, between 15 and 30 January, daily averages remained below 20°C, with maximum between 18.4°C and 26.9°C, suggesting exposure to natural low temperatures during critical reproductive stages.
Genotypes flowered between 15 and 30 February, including BR12266-44-11-32-5-1-1-HR10-B, BR11303-5R-156, BR11318-5R-106, BR10317-5R-25, Bhutan, BRRI dhan74, BRRI dhan28, BRRI dhan88, IR18A1859, and TP16199, likely encountered severe cold stress at PI and booting stages. Additionally, entries flowering in early March (e.g., Rata Boro, IR83222-F11-173, Hbj. B. VI, Tepi Boro, Mineasahi, and several BR11894 derivatives) experienced moderate cold stress (19.1°C–21.3°C), particularly during booting.
In contrast, all breeding lines in non-stress plots flowered between 17 March and 10 April 2024 (S7B Fig in S1 File). The non-stress plots recorded temperatures between 21.1°C and 25.4°C (max: 26.3°C–31.4°C) from 15 February to 10 March (S9B Fig in S1 File), indicating favorable thermal conditions during reproductive development (PI to booting stage).
Under natural field conditions, all the phenotypic traits except vegetative stage score (VegS) showed a significant response to cold stress (Table 7). Although effective tiller number was significantly affected under both cold stress and non-stress conditions, it was excluded from further analysis. Specifically, in the cold stress plots, young seedlings were damaged by night birds shortly after transplanting, creating vacant spaces in the field. These gaps allowed the remaining plants to compensate by producing more tillers, resulting in a higher effective tiller number under cold stress relative to non-stress conditions. Therefore, traits like PDS, PES, SFS, plant height, days to heading, and panicle length were used for further analysis. Under phytotron conditions, these traits have a significant effect on cold stress, whereas plant height trait didn’t exhibit any significant effect on cold stress (Table 7). Thus, the traits that had a significant effect on cold stress were used for further analysis.
The variation in phenotypic performance of the diverse genotypes based on measured agronomic traits from non-stress to stress conditions under natural (field) and artificial cold screening facilities (phytotron) was represented by the reduction rate in Table 8. The lowest plant height reduction (PHR) was observed in TP16199 (G13) (−10.88% in field). BRRI dhan74 (G21) and IR100723-B-B-B-B-61 (G15) exhibited the least delay in heading date (DHD) (3.50 and 0.57 days) in field and phytotron conditions, respectively. Minimum panicle length reduction (PLR) was observed in BRRI dhan92 (G26) (−4.26% in field) and BR11712-4R-227 (G8) (−16.71% in phytotron). Whereas, the highest PHR was shown in BR10317-5R-25 (G7) (21.85% in the field). The longest DHD was observed in Mineasahi (G5) (25.50 days in the field) and BR11318-5R-106 (G10) (10.07 days in the phytotron). IR100722-B-B-B-B-11 (G14) and BR11894-R-R-R-R-329 (G17) showed the highest PLR (21.96 and 15.49%) under field and phytotron conditions, respectively.
The variation of genotypes in cold-related traits like PDS, PES, and SFS under stress and non-stress conditions in the field and phytotron was represented as paired dot plots, comparing where non-stress (blue circle) and cold stress (red triangle) scores for a given entry (Fig 3). A longer line indicates a greater difference between the two conditions, suggesting higher susceptibility to cold stress. Conversely, a shorter line indicates a smaller difference, implying better tolerance to cold stress. In the field, the lowest cold trait scores (indicating tolerance) were observed in Bhutan, BRRI dhan67, and BRRI dhan92 (1, PDS); BR11894-R-R-R-R-110, Mineasahi, and BRRI dhan92 (1, PES); and IR83222-F11-173 (1, SFS) (Fig 3A).
PDS: Panicle degeneration score, PES: Panicle exsertion score, and SFS: Spikelet fertility score. Where, 1 = Highly tolerant, 3 = Tolerant, 5 = Moderately tolerant, 7 = Moderately susceptible, 9 = Susceptible.
In the phytotron, most entries exhibited low degeneration; IR83222-F11-173 and Bhutan performed the lowest PES (1), and Bhutan also showed the lowest SFS score (1; Fig 3B). While BRRI dhan92 demonstrated cold tolerance in the field, this might be attributed to its longer growth duration, avoiding stress conditions. Conversely, Rata Boro and IR83222-F11-173 showed reverse results in the phytotron. SFS was higher under non-stress conditions than cold stress in the phytotron due to high-temperature stress at the reproductive stage in non-stress pots. The ambient temperature for non-stress pots at the flowering stage was 27.6-38.1℃, i.e., higher than 35℃ (S10B Fig in S1 File). So, nonstress plots suffered from heat stress at the reproductive stage, leading to spikelet sterility.
Spearman correlation analysis was performed to elucidate the relationships among the different phenotypic variables at two different levels of cold stress (field and phytotron conditions), and the results are presented in Table 9. Under field, SFS exhibited a significant positive correlation with PDS and PES, indicating that spikelet fertility was associated with panicle degeneration and exsertion. A significant positive correlation was also found between plant height and panicle length. Days to heading was negatively correlated with SFS, suggesting that longer growth duration was associated with lower spikelet fertility. In the case of phytotron conditions, the traits of plant height, days to heading, and panicle length were significantly and positively correlated with each other. Days to heading showed a significant negative correlation with SFS, consistent with field observations.
Principal Component Analysis (PCA) was conducted to explore the phenotypic variation and reduce the dimension among 25 rice genotypes under cold stress at the reproductive stage under the two environments (field and phytotron conditions). Under field conditions, the first two principal components (PC1 and PC2) explained 57.8% of the total variance, where PC1 accounted for 39.1% of the variability (Fig 4A). Whereas, under controlled phytotron conditions, PC1 and PC2 explained 31.5% and 24.6% of the total variance, summing to 56.1% (Fig 4B). Across both environments, G17 (BRRI dhan28), a known susceptible cultivar, consistently clustered with other susceptible genotypes on the positive side of PC1 for relevant cold traits (such as PDS, PES, SFS, PHR, and PLR in the field and PES, SFS, PLR, and DHD in the phytotron).
G1: Rata Boro, G2: Tepi Boro, G3: Mineasahi, G4: Bhutan, G5: IR83222-F11-173 (Tol.ck), G6: Hbj. B. VI, G7: IR100722-B-B-B-B-11, G8: IR100723-B-B-B-B-61, G9: BR10317-5R-25, G10: BR11303-5R-156, G11: BR11318-5R-106, G12: BR11712-4R-227, G13: BR12266-44-11-32-5-1-1-HR10-B, G14: BR8899-14-4-1-2-2-1, G15: TP16199, G16: IR18A1859, G17: BRRI dhan28 (Sus. Ck.), G18: BRRI dhan67 (Mod. tol.Ck.), G19: BR11894-R-R-R-R-169, G20: BR11894-R-R-R-R-329, G21: BR11894-R-R-R-R-105, G22:BR11894-R-R-R-R-110, G23: BRRI dhan74, G24: BRRI dhan88, G25: BRRI dhan81, G26: BRRI dhan92.
Genotyping results for the reproductive stage cold screening revealed clear differentiation between the most cold-tolerant and most cold-susceptible genotypes based on the allelic composition of qPSST3, qPSST7, and qPSST9 (Table 10). The genotypes Rata Boro, Bhutan, and BR11712-4R-227 possessed partial alleles of qPSST3, qPSST7, qPSST9, and qPSST6, whereas IR83222-F11-173 carried partial alleles of qPSST3, qPSST7, and qPSST9. Although these genotypes exhibited tolerance to cold stress at the reproductive stage, genotypic analysis indicated that none of them possessed the complete set of alleles associated with qPSST3, qPSST7, and qPSST9. This differences between phenotypic performance and allelic composition suggests that additional QTLs, alleles, or specific SNP positions beyond qPSST3, qPSST7, and qPSST9 contributed to reproductive-stage cold tolerance.
Under both field and phytotron conditions, PC1 is the primary descriptor of sensitivity, with sensitivity associated with positive values of SFS, PES and possibly PDS. PC2 was explained mostly by height-related parameters PHR and PLR, which may or may not be directly linked to cold stress. These traits could represent growth characteristics that are either indirectly affected by cold or simply vary independently.
However, a notable difference was observed that PDS and DHD were associated with PC1 under field conditions, but contributed more strongly to PC2 under phytotron conditions. This suggests that PDS and DHD may reflect different physiological responses depending on environmental condition. Importantly, another key discrepancy between field and phytotron conditions was DHD contribute to PC1 in the field, while PC2 under phytotron conditions. This divergence highlights the influence of environmental condition on trait expression and underscores the need to interpret PCA results within the framework of specific growing conditions.
To support and validate the PCA-based grouping of genotypes, hierarchical clustering was performed using Euclidean distance and Ward’s method under both field and phytotron conditions. The resulting clusters showed strong concordance with PCA, particularly in the consistent placement of cold-tolerant genotypes in Cluster I and the susceptible check BRRI dhan28 (G23) in Cluster IV across both environments (S13A and S13B Figs in S1 File).
Cluster mean values (S2 Table in S1 File) were used to rank genotypes based on trait performance under field and phytotron conditions (S3 Table in S1 File). The summed scores ranged from 18 to 38, with genotypes with lower summed scores across traits were classified as tolerant (18–22), while those with higher scores were considered as susceptible (33–38). The tolerant genotypes identified were Bhutan, IR83222-F11-173, BR11712-4R-227, BRRI dhan74, and Rata Boro. Genotypes which were grouped into moderately tolerant included, BR11894-R-R-R-R-110, BRRI dhan67 (Mod. tol. ck.), BR11894-R-R-R-R-105, BR11303-5R-156, TP16199, IR100723-B-B-B-B-61, BR12266-44-11-32-5-1-1-HR10-B, and BR11894-R-R-R-R-169. Susceptible genotypes were BRRI dhan88, BRRI dhan28 (Sus. ck.), IR18A1859, BR11318-5R-106, BRRI dhan81, and BR10317-5R-25.
Performance of genotypes evaluated at both seedling and reproductive stage
The performance of 18 common rice genotypes evaluated under both seedling and reproductive stages revealed distinct patterns of cold stress tolerance. The full spectrum of observed responses for each genotype at both developmental stages is presented in Fig 5. Genotypes such as Bhutan, BR11712-4R-227, BR12266-44-11-32-5-1-1-HR10-B, and BRRI dhan67 consistently exhibited moderately tolerant to tolerant phenotypes at both stages, indicating stable performance under cold stress.
Bar heights represent the cold tolerance status for each genotype. Status was defined as: 1 = Tolerant, 2 = Moderately tolerant, 3 = Moderately susceptible, and 4 = Susceptible.
A few genotypes, including Rata Boro, BR100723-B-B-B-B-61, BR11303-5R-156, TP16199, and IR83222-F11-173, showed tolerance specifically at the reproductive stage but were moderately susceptible to susceptible at the seedling stage. In contrast, BR11318-5R-106 was moderately tolerant only at the seedling stage but susceptible at the reproductive stage. Genotypes like Mineasahi, BR100722-B-B-B-B-11, BR8899-14-4-1-2-2-1, BR10317-5R-25, IR18A1859, Tepi Boro, and the susceptible check BRRI dhan28 showed poor performance across both stages.
Spearman correlation analysis between seedling and reproductive stage results (Table 11) revealed a weakly positive but non-significant correlation between cold tolerance scores at the seedling and reproductive stages (r2 = 0.18, P = 0.479), indicating no statistically meaningful association between tolerance levels across these two developmental phases. Thus, seedling and reproductive stage cold tolerance should be considered as separate traits and quite likely under distinct genetic control.
A PCA was employed to investigate the relationship among cold stress-related traits at both the seedling (LD, survival percentage) and reproductive stages (PDS, PES, SFS, PHR, DHD, PLR) across 18 common rice genotypes, under field and phytotron conditions (Fig 6).
G1: Rata Boro, G2: Tepi Boro, G3: Mineasahi, G4: Bhutan, G5: IR83222-F11-173 (Tol.ck), G6: Hbj. B. VI, G7: IR100722-B-B-B-B-11, G8: IR100723-B-B-B-B-61, G9: BR10317-5R-25, G10: BR11303-5R-156, G11: BR11318-5R-106, G12: BR11712-4R-227, G13: BR12266-44-11-32-5-1-1-HR10-B, G14: BR8899-14-4-1-2-2-1, G15: TP16199, G16: IR18A1859, G17: BRRI dhan28 (Sus. Ck.), G18: BRRI dhan67 (Mod. tol.Ck.).
Under field conditions (Fig 6A), the first two principal components (PCs) accounted for 62.1% of the total variation, with PC1 contributing 40.6%. PC1 had a greater contribution of LD, PDS, and SFS, higher values of these traits corresponding to cold susceptibility. On the other hand, PLR, PHR, DHD and survival percentage contributed to PC2. Thus, a combination of seedling and reproductive traits contributed to each of the first two principal components.
Under phytotron conditions (Fig 6B), PC1 and PC2 explained 50.4% of the total variation (PC1: 26.2%, PC2: 24.2%). PC1 was mostly dominated by seedling traits LD and survival%, with some contribution from SFS, while PC2 was mostly controlled by reproductive traits DHD, PES and SFS. Thus, phytotron data shows distinct partitioning of PC1 and PC2 between seedling and reproductive-stage traits, supporting the hypothesis that seedling and reproductive-stage tolerance are separate traits having different genetic control.
Discussion
Cold stress adversely affects rice productivity, with its varying effects depending on growth stage as well as the severity and duration of exposure. Rice plants are susceptible to cold damage at any developmental phase – including germination, seedling, vegetative, reproductive, and maturity stages [11,21]. This study investigates a diverse set of rice genotypes under cold conditions during the seedling and reproductive stages, aiming to elucidate stage-specific vulnerabilities and to reinforce the importance of comprehensive cold screening at all growth stages to unravel the complex resilience pathways to cold stress.
Cold damage at the seedling stage manifests primarily as leaf yellowing due to inhibited chlorophyll synthesis, stunted growth, and eventual seedling mortality. These physiological disruptions often reduce photosynthetic efficiency and elevate respiration, leading to assimilate shortages that impair tissue development and halt growth. LD is widely used to differentiate cold-tolerant genotypes from susceptible ones; however, it alone does not fully capture the complexity of seedling-stage cold tolerance [19].
Therefore, survival rate is also considered a critical metric for identifying promising donor genotypes [21]. In this study, 38 diverse rice genotypes were evaluated for seedling stage cold tolerance across three batches (B1, B2, B3) using both LD and survival percentage. Two genotypes, BR8907-B-1–2-CS1–4-CS2-P3-4 (G36) and BR8909-B-12–2-CS1–4-CS2-P2-3–2 (G37) consistently demonstrated superior tolerance across all batches, with the lowest LD scores (3.7 and 3.9) and highest survival rates (34.4−37%). These lines, derived from three-way crosses involving donor parents 89010-TR1232-4-1 and CUNJING15 under BRRI dhan29 and BRRI dhan28 backgrounds, respectively, represent valuable resources for breeding programs. The performance of the tolerant and susceptible varieties aligned with their expected tolerance categories (Fig 3), justifying the tolerance levels of G36 and G37. Also, six genotypes- BR11318-5R-106 (G11), BR11712-4R-227 (G12), BR12266-44-11-32-5-1-1-HR10-B (G13), Bhutan (G4), BRRI dhan58 (G43), and BRRI dhan89 (G45) were classified as moderately tolerant, showing LD scores of 4.9-6.6 and survival rates of 14.1−31%. Among these, Bhutan and BR12266-44-11-32-5-1-1-HR10-B showed relatively high survival (24.3-24.7) despite moderate LD, indicating possible recovery or stress avoidance mechanisms. This partial independence of LD and survival percentage suggests distinct physiological pathways governing cold tolerance for these two traits in some genotypes. Suh et al. (2013) reported LD scores ranging from 3 to 8 (mean value of 5) among 23 elite cultivars grown under cold-water irrigation, reinforce LD as a visual assessment of damage, while survival percentage offers direct measure of resilience. The strong negative correlation observed between LD score and survival percentage (Spearman’s r = −0.708, p < 0.0001) indicates these two parameters are closely related and can, at least partially, substitute for each other in assessing seedling-stage cold tolerance. Genotypes with low LD and high survival are reliably tolerant, while outliers may possess unique genetic and physiological mechanisms worth further investigation.
Seedling-stage cold tolerance in rice is a quantitatively inherited trait governed by multiple QTLs, including qSCT1, with individual loci contributing partial effects to overall tolerance [5,7,25] (Table 6). The occurrence of cold-tolerant genotypes lacking favorable alleles of qSCT1 further supports the involvement of additional loci underlying seedling-stage cold tolerance. Genome-wide association studies have also revealed multiple QTLs and favorable haplotypes contributing to seedling cold tolerance and functional QTL fine-mapping has led to the discovery of specific cold tolerance genes such as OsbZIP72 [26] and cold stress response loci including COLD11 and OsCTS11 [27]. These findings reinforce the polygenic architecture of seedling-stage cold tolerance and support the idea that, in addition to major QTLs like qSCT1, other genetic components contributed to phenotypic variation observed among tolerant genotypes.
Reproductive-stage cold stress in rice was evaluated under field and phytotron conditions to assess genotype performance across environments. Screening at young microspore stage using cold air, water and natural winter conditions enabled consistent evaluation [28]. The trait values were normalized as reduction rates (e.g., PHR, PLR, DHD) relative to controls [21,29], along with visual injury scores (PDS, PES, and SFS), to quantify sterility and panicle development failure (Table 8; Fig 3A and 3B). These remain standard indicators for reproductive stage cold tolerance [21,30].
Significant phenotypic variation among the genotypes across environments revealed diverse cold response mechanisms. Genotypes like Bhutan, IR83222-F11-173, Rata Boro, and BRRI dhan74 showed consistently low reduction rates and injury scores, suggesting physiological resilience to cold stress. By contrast, the genotypes TP16199, BR11303-5R-156, and BR12266-44-11-32-5-1-1-HR10-B showed moderate cold tolerance scoring and moderate level of reduction in the studied traits, while several genotypes (BR11894-R-R-R-R-110, BR11303-5R-156, Rata Boro, IR83222-F11-173, and BRRI dhan67) exhibited delayed heading (12-20.50 days) but maintained low PHR, PLR, PDS, PES, and SFS, suggesting escape of cold stress in these genotypes due to the delayed heading. However, the opposite result was reported by Suh et al. (2010) [7,31]. Thus, DHD alone may not be a reliable measure for screening reproductive-stage cold-tolerant rice [32].
Correlation analysis revealed that growth reduction traits (PHR, PLR, DHD) were positively associated, as were traits related to injury scores (PDS, PES, SFS) (Table 9), while negative correlations were observed between injury score and growth reduction traits, indicating that greater reduction in growth traits aligns with higher injury. This observation is consistent with the findings [21,31] that cold stress reduces culm length and panicle development, ultimately impairing spikelet fertility. PCA distinguished the genotype responses, with DHD contributing to PC1 in field condition implying delayed flowering as a cold escape strategy. Under phytotron condition, DHD explained by PC2. This distinct shift in DHD’s contribution between the two environments highlights a significant Genotype×Environment interaction, suggesting that the phenotypic traits governing cold stress variability differ based on growth conditions (Fig 4).
Comparable studies [29] have validated the Average Tolerance Index (ATI) for rice germplasm assessment under cold stress at early growth stages emphasizing genotype × environment interactions in trait expression.
Hierarchical clustering and trait-based ranking demonstrated stable performance of genotypes such as IR83222-F11-173, Bhutan, Rata Boro, and BRRI dhan74 across both environments, suggesting robust cold tolerance (S12 Fig, S3 Table in S1 File). This validates previous claims that cultivars tolerant under controlled cold conditions also perform well under field stress [30], supporting phytotron screening as a reliable proxy. Eighteen genotypes evaluated at both seedling and reproductive stages showed a weak and non-significant correlation between the two stages indicating different genetic mechanisms governing cold tolerance at each stage. This is supported by studies showing distinct QTLs and physiological responses control tolerance at seedling (e.g., qCTS12.1) and reproductive stages (e.g., qPSR10) [5,12]. For example, early-stage traits such as germination percentage, shoot and root length, and coleoptile length are strongly intercorrelated, but non-significant and negligible positive association between the germination and early seedling establishment [29]. Together these results emphasize the need to screen germplasm for cold tolerance at both stages, rather than assuming tolerance at one stage will predict tolerance for another.
The known QTLs qPSST3, qPSST7, and qPSST9 were reported to confer reproductive-stage cold tolerance [7]. However, genotyping results (Table 10) indicated that these loci alone did not fully account for the observed tolerance in the evaluated 12 genotypes. Many tolerant lines possessed only partial alleles, reflecting a weak association between these markers and phenotypic performance. This supports the view that reproductive-stage cold tolerance is a polygenic trait, influenced by multiple minor-effect QTLs and stage-specific regulatory mechanisms [33]. Consequently, reliance on a limited set of qPSST-linked markers is insufficient for effective marker-assisted selection. Comprehensive approaches, including high resolution QTL mapping, fine mapping, and identification of key SNPs, are necessary to uncover additional genomic regions and improve the precision of breeding for cold-tolerant rice adapted to low-temperature environments.
PCA combining both developmental stages under field conditions indicated seedling and reproductive traits may share similar physiological pathways, supporting the hypothesis that certain genotypes possess integrated tolerance mechanisms operating across the developmental stages. This is consistent with recent studies suggesting that cold stress responses involve shared signaling pathways – such as ABA-mediated regulation, ROS detoxification, and membrane stabilization – active during both vegetative and reproductive phases [34]. However, under phytotron conditions, seedling and reproductive traits contributed to different components, supporting partial independence and stage-specific genetic control. These observations align with previous reports showing that different QTLs contribute to cold tolerance at seedling and reproductive stages, highlighting the complex nature of cold tolerance at different stages [21]. Supporting this finding, a comprehensive study was carried out by Kim et al. (2025), integrated QTL mapping with transcriptome analysis at the budding and seedling stages, revealed distinct differences in cold tolerance mechanisms between developmental stages [35]. Together, these findings suggest that while certain cold-responsive pathways may be shared, independent regulatory mechanisms act at specific growth stages, emphasizing the need for stage-specific screening approaches for accurate classification of genotypes and effective breeding for cold tolerance.
Conclusions
In this study, BR8907-B-1–2-CS1–4-CS2-P3-4 and BR8909-B-12–2-CS1–4-CS2-P2-3–2 exhibited consistent cold tolerance at seedling stage, Rata Boro, BR100723-B-B-B-B-61, BR11303-5R-156, TP16199, and IR83222-F11-173 demonstrated tolerance to cold stress at reproductive-stage and Bhutan, BR11712-4R-227, and BR12266-44-11-32-5-1-1-HR10-B exhibited moderate to strong tolerance across both stages, indicating presence of stage-specific differential cold-stress response among these genotypes. Genotyping results and phenotypes under cold stress also indicated genetic control in these genotypes is independent and tolerance expressed at one stage does not necessarily confer tolerance at another, underscoring the importance of stage-specific cold screening. Integrating stage-specific tolerant donors into a single genetic background is recommended to develop rice varieties resilient to cold stress across multiple developmental phases, thereby stabilizing yields under low-temperature conditions.
Supporting information
S1 File.
S1 Fig. Growth stage-specific low temperature response associated with physiological and morphological impairment during seedling to flowering stage in rice. S2 Fig. Seeding of pregerminated seeds in seedling trays for cold screening at the seedling stage. S3 Fig. Seedlings at three-leaf stage prepared for placing in the cold water tank for artificial cold screening at the seedling stage. S4 Fig. Seedlings in the cold water tank adjusted at 13℃ for artificial cold screening at the seedling stage. S5 Fig. Temperature profile of batch 1, 2 and 3 in cold water tank. S6 Fig. (A) Onset of cold stress symptoms in seedlings; (B) Seedlings at the scoring stage (8th day), when the susceptible check died. S7 Fig. (A) Leaf discoloration (LD) scoring based on SES score, (B) Cold water treated trays after 7 days of cold stress. S8 Fig. Dot plot showing flowering window of the diverse panel under A) early sown cold stress plot (CS) and B) regular Boro season (NS) plot in field. S9 Fig. Maximum and minimum temperature of field conditions under A) cold stress, B) Non stress conditions. S10 Fig. Temperature profile of air and water in artificial cold screening facilities (CSF) of A) cold stress, and B) ambient air temperature. S11 Fig. Cold treated plants at PI to booting stage in phytotron for reproductive stage cold screening. S12 Fig. Tiller which was at meiotic stage was labeled and treated under cold stress condition. S13 Fig. Dendrogram showing the hierarchical clustering of rice genotypes based on phenotypic traits under (A) field and (B) phytotron conditions with cold stress at the reproductive stage. S1 Table. Phenotypic performance in term of LD score and survival rate (%) of 38 rice genotypes after cold stress (13°C) at seedling stage. S2 Table. Cluster mean of different cold traits in the clusters formed at different cold levels using Euclidian distance matrix. S3 Table. Ranking of diverse rice germplasm for their relative cold related traits at reproductive stage in a cluster analysis using Euclidian distance matrix.
https://doi.org/10.1371/journal.pone.0338290.s001
(RAR)
Acknowledgments
The authors wish to thank Dr. Md. Abu Syed, MFRK Prince, Dr. Md. Rafiqul Islam, Dr. Sharmistha Ghosal, Dr. Ripon Roy, Dr. Ratna Rani Majumdar, and Dr. Md. Adil for technical assistance and support in various aspects of this work. The authors also deeply acknowledge BRRI management for providing spaces in the experimental field, phytotron and laboratory facilities and Gates foundation for granting funds for the study.
References
- 1.
World Food and Agriculture – Statistical Yearbook 2023. World Food and Agriculture Statistical Yearbook 2023. FAO; 2023. https://doi.org/10.4060/cc8166en
- 2. Lee Y, Jang S-G, Jin B, Jo S-M, Kang J-W, Lee S-M, et al. Integrating molecular markers and phenotypic analysis to assess cold tolerance in rice germplasm. Mol Breed. 2025;45(2):23. pmid:39927325
- 3.
Kabir MS, Salam MU, Chowdhury A, Rahman MF, Iftekharuddaula KM, Rahman MS, et al. Rice Vision for Bangladesh: 2050 and Beyond. Bangladesh Rice J. 2015.
- 4. Lv Y, Guo Z, Li X, Ye H, Li X, Xiong L. New insights into the genetic basis of natural chilling and cold shock tolerance in rice by genome-wide association analysis. Plant Cell Environ. 2016;39(3):556–70. pmid:26381647
- 5.
Zhang Q, Chen Q, Wang S, Hong Y, Wang Z. Rice and cold stress: methods for its evaluation and summary of cold tolerance-related quantitative trait loci. 2014. Available: http://www.thericejournal.com/content/7/1/24
- 6. Cruz RP da, Milach SCK. Cold tolerance at the germination stage of rice: methods of evaluation and characterization of genotypes. Sci agric (Piracicaba, Braz). 2004;61(1):1–8.
- 7. Suh JP, Jeung JU, Lee JI, Choi YH, Yea JD, Virk PS, et al. Identification and analysis of QTLs controlling cold tolerance at the reproductive stage and validation of effective QTLs in cold-tolerant genotypes of rice (Oryza sativa L.). Theor Appl Genet. 2010;120(5):985–95. pmid:20012263
- 8. Rashid M, Yasmeen R. Cold Injury and Flash Flood Damage in Boro Rice Cultivation in Bangladesh: A Review. Bangladesh Rice J. 2018;21(1):13–25.
- 9. Xiao P, Xiong M, Hou K, Guo X, Li H, Liu Y. Cold stress in rice (Oryza sativa L.): Molecular mechanisms of sensing, signaling, transcriptional regulation, membrane lipid remodeling, and hormonal modulation. Front Plant Sci. 2026;16.
- 10.
Yoshida S. Fundamental of rice crop science. International Rice Research Institute, Los Baños, Laguna, Philippines. 1981. http://books.irri.org/9711040522_content.pdf
- 11. Ye C, Fukai S, Godwin I, Reinke R, Snell P, Schiller J, et al. Cold tolerance in rice varieties at different growth stages. Crop & Pasture Science. 2009;60(4):328–38.
- 12. Biswas PS, Rashid MdM, Khatun H, Yasmeen R, Biswas JK. Scope and Progress of Rice Research Harnessing Cold Tolerance. Advances in Rice Research for Abiotic Stress Tolerance. Elsevier. 2019. p. 225–64.
- 13. Sato R, Sugawara N, Zewdu Z, Takasago H, Abe A, Matsunami M. Prescreening rice for cold tolerance at the reproductive stage using anther length (prescreening rice reproductive-stage cold tolerance). Field Crops Research. 2025;333.
- 14. Rice Improvement. Springer International Publishing. 2021.
- 15. Shakiba E, Edwards JD, Jodari F, Duke SE, Baldo AM, Korniliev P, et al. Genetic architecture of cold tolerance in rice (Oryza sativa) determined through high resolution genome-wide analysis. PLoS One. 2017;12(3):e0172133. pmid:28282385
- 16. da Cruz RP, Sperotto RA, Cargnelutti D, Adamski JM, de Freitas Terra T, Fett JP. Avoiding damage and achieving cold tolerance in rice plants. Food and Energy Security. Blackwell Publishing Ltd. 2013. p. 96–119.
- 17. Roychowdhury R, Das SP, Sarkar P, Khan Z, Kumar A, Sarker U, et al. Physiological, biochemical and molecular signaling basis of cold stress tolerance in plants. Front Plant Sci. 2025;16:1707204. pmid:41487347
- 18. Neelima P, Sandeep Varma V. Review on Cold Tolerance in Rice Genotypes at Various Growth Stages. International Journal of Genetics. 2020;12. Available: https://www.bioinfopublication.org/jouarchive.php?opt=&jouid=BPJ0000226
- 19. Kumar Singh B, Sutradhar M, Lahkar C, Kumar Singh A, Tombisana Meetei NG, Mandal N. Screening of rice germplasms (Oryza sativa L.) for seedling stage cold tolerance utilizing morphological and molecular markers. Ecol Genet Genom. 2022;24.
- 20. Khatun H, Biswas PS, Hwang HG, Kim K-M. A Quick and Simple In-house Screening Protocol for Cold-Tolerance at Seedling Stage in Rice. Plant Breed Biotech. 2016;4(3):373–8.
- 21. Biswas P, Khatun H, Anisuzzaman M. Molecular and Phenotypic Characterization for Cold Tolerance in Rice (Oryza sativa L.). Bangladesh Rice J. 2020;23(2):1–15.
- 22. Jo S, Jang S-G, Lee S-B, Lee J-Y, Cho J-H, Kang J-W, et al. Analysis of quantitative trait loci and candidate gene exploration associated with cold tolerance in rice (Oryza sativa L.) during the seedling stage. Front Plant Sci. 2025;15:1508333. pmid:39840352
- 23. Shim K-C, Yun Y-T, Kang J-W, Ahn S-N. Identification of QTLs for Cold Tolerance at Seedling Stage Using a Population Derived from an Inter-specific Cross in Rice. Plant Breed Biotech. 2022;10(4):282–9.
- 24.
International Rice Research Institute. Standard Evaluation System. 5th Edition. 2013. http://www.knowledgebank.irri.org/images/docs/rice-standard-evaluation-system.pdf
- 25. Kim S-M, Suh J-P, Lee C-K, Lee J-H, Kim Y-G, Jena KK. QTL mapping and development of candidate gene-derived DNA markers associated with seedling cold tolerance in rice (Oryza sativa L.). Mol Genet Genomics. 2014;289(3):333–43. pmid:24464311
- 26. Gu Y, Guo H, Li H, Su R, Khan NU, Li J, et al. QTL mapping by GWAS and functional analysis of OsbZIP72 for cold tolerance at rice seedling stage. The Crop Journal. 2024;12(6):1697–708.
- 27. Wu B, Fu M, Du J, Wang M, Zhang S, Li S, et al. Identification of the Cold-Related Genes COLD11 and OsCTS11 via BSA-seq and Fine Mapping at the Rice Seedling Stage. Rice (N Y). 2024;17(1):72. pmid:39576378
- 28. Zhang Z, Li J, Pan Y, Li J, Zhou L, Shi H, et al. Natural variation in CTB4a enhances rice adaptation to cold habitats. Nat Commun. 2017;8:14788. pmid:28332574
- 29. Monzón DLR, Cantero J, Danielowski R, da Luz VK, Venske E, Mota MS, et al. An optimized index for cold tolerance assessment in rice during germination and early seedling stage. J Crop Sci Biotechnol. 2022;26(2):243–53.
- 30. Farrell TC, Fox KM, Williams RL, Fukai S. Genotypic variation for cold tolerance during reproductive development in rice: Screening with cold air and cold water. Field Crops Research. 2006;98(2–3):178–94.
- 31. Jena KK, Kim SM, Suh JP, Yang CI, Kim YG. Identification of Cold‐Tolerant Breeding Lines by Quantitative Trait Loci Associated with Cold Tolerance in Rice. Crop Science. 2012;52(2):517–23.
- 32. Horai K, Ishii A, Mae T, Shimono H. Effects of early planting on growth and yield of rice cultivars under a cool climate. Field Crops Research. 2013;144:11–8.
- 33. Jeong BY, Lee Y, Kwon Y, Kim JH, Ham T-H, Kwon S-W, et al. Genome-Wide Association Study Reveals the Genetic Basis of Chilling Tolerance in Rice at the Reproductive Stage. Plants (Basel). 2021;10(8):1722. pmid:34451767
- 34. Shahzad N, Nabi HG, Qiao L, Li W. The molecular mechanism of cold-stress tolerance: Cold responsive genes and their mechanisms in rice (Oryza sativa L.). Biology. 2024.
- 35. Kim CA, Chen W, Zhu S, Zhang G, Shen C, Chen K, et al. Combining QTL mapping and transcriptomics to identify candidate genes for cold tolerance during the budding and seedling stages in rice. BMC Genomics. 2025;26(1):756. pmid:40830828