Two upland rice germplasm lines from northern Laos, Kampeng and Non, were used in this study. Kampeng is a tropical japonica black rice domestically utilized as traditional medicine in rural villages of Houaphanh province. This landrace has recently been identified as a high rutin-accumulating variety with excellent grain appearance, particularly its strong black pigmentation, which is an important trait for black rice quality (Figures 1A,B) (Asai et al., 2020). In line with the national strategy for promoting value-added rice production, Kampeng has attracted increasing attention as a health-promoting rice; however, its yield performance is generally low (Asai et al., 2020). Non is a traditional indica white rice with strong shoot-branching characteristics (Figures 1A,B) (Asai and Soisouvanh, 2017). Due to its high yield potential in low-input upland systems, Non has been officially selected for seed distribution in northern Laos. Kampeng was crossed with Non, and the F₁ seeds were harvested. After growing the F₁ plants, self-pollinated F₂ seeds were harvested for QTL analysis. Based on the results of this analysis, four genotypes were selected from the F₂ individuals: (i) homozygous for Kampeng for detected QTLs qPA1 and Hd6; (ii) homozygous for Kampeng for qPA1 and Non for Hd6; (iii) homozygous for Non for qPA1 and Kampeng for Hd6; and (iv) homozygous for Non for qPA1 and Hd6. F₃ lines derived from the four genotypes were used to validate the effects of the QTLs.

Phenotyping was conducted at two locations with different climate conditions: Ishigaki, Japan, and Vientiane, Lao PDR. In 2021, more than 200 F₂ individuals and their parents were cultivated in a paddy field at the Japan International Research Center for Agricultural Sciences (JIRCAS), Ishigaki, Japan (24° 22′45.92” N, 124° 11′49.12″E, 30 m altitude). The trial was conducted under irrigated lowland conditions to ensure the growth potential in the environmental conditions of Japan. On 13 July 2021, the seeds were sown in seedling nursery boxes, and on 11 August 2021, the seedlings were transplanted into the experimental paddy field, one seedling per hill. The planting density was 18.5 hills m⁻², with 18 cm between each hill and 30 cm between the rows. Each row had 10 hills. At transplanting, nitrogen (N), phosphorus (P), and potassium (K) were applied as basal fertilizers at rates of 64 kg N ha⁻¹, 17.5 kg P ha⁻¹, and 33.2 kg K ha⁻¹, respectively. Thirty days after transplanting, 20 kg N ha⁻¹ was applied as a top dressing. A total of 158 F₂ individuals, excluding border plants, were used for phenotyping.

In 2022, 124 F₂ individuals and their parents were grown in pots at an experimental pool in the Rice and Cash Crop Research Center (RCCRC), Vientiane, Laos (18° 08′51.53” N, 102° 44′10.18″E, 170 m altitude). Pots were used to ensure precise data collection for genetic analysis under homogeneously simulated upland conditions. Soil (4 kg) was mixed with 1 kg of farmyard manure, and each pot (25 cm in diameter) was filled with the mixed soil. The pots had small holes at the bottom to allow water uptake, and all pots were placed in the experimental pool. On 15 June 2022, seeds were sown in pots (one per pot). The water level in the pool was maintained at 2–3 cm to simulate the upland conditions in the pots.

In 2023, the four F₃ genotypes selected from the QTL analysis of the F₂ populations were grown in a paddy field at JIRCAS, Ishigaki, Japan. On 11 July 2023, the seeds were sown in seedling nursery boxes, and on 8 August 2023, the seedlings were transplanted into an experimental paddy field, one seedling per hill. The planting density and fertilizer application were similar to those used in 2021.

In all experiments, the heading date was recorded when the first panicle emerged on each plant. The number of days to heading was calculated as the number of days from the sowing date to the heading date. The number of panicles was counted at maturity, and the panicles were harvested and threshed. The grains were then dehulled, and the dehulled grains (brown rice) were used for grain color and phytic acid analysis. Grain color was assessed using the Grain Scanner instrument (Satake Engineering Co., Ltd., Tokyo, Japan). Brown rice was scanned, and the lightness parameter L* was calculated, with 100 representing white and 0 representing black, according to the Lab color system of Commission Internationale d’Eclairage (1976). PA analysis was conducted following the method reported by Matsuo et al. (2005), with some modifications. Brown rice (3 g) was placed in the grinding container and chilled in a freezer at −80°C before being ground using the automated TK-AM5 (Tokken Inc., Chiba, Japan). The ground powder samples were stored at −20°C until analysis. The powdered sample (0.3 g) was mixed with 1.2 mL of 0.8 N HCl in a 5-mL microcentrifuge tube and vortexed at 2,400 rpm for 1 h at 4°C using a Multi-Tube Vortexer BSR-MTV100 (Bio Medical Science, Tokyo, Japan), followed by centrifugation at 18,000 × g for 15 min at 4°C. The clear supernatant (15 μL) was mixed with 285 μL of sterilized reverse osmosis water in a 1.5 mL microcentrifuge tube and heated at 100°C for 20 min, followed by centrifugation at 18,000 × g for 15 min at 4°C. The clear supernatant was filtered through a 0.22 μm membrane filter and subjected to high-performance liquid chromatography (HPLC) using a binary HPLC system equipped with an anion-exchange column, TSKgel DEAE-5PW (7.5 mm φ × 75 mm, Tosoh Corporation, Tokyo, Japan). Phytic acid was eluted from an injected sample (100 μL) using a linear gradient of 0.01 M HNO₃ to 0.4 M NaNO₃ and 0.01 M HNO₃ over 40 min at a flow rate of 1.0 mL min⁻¹. Eluted phytic acid reacted with Wade reagent (0.06% FeCl₃∙6H₂O, 0.3% sulfosalicylic acid) in a post-column tube at a flow rate of 0.5 mLmin⁻¹, and the decrease in absorbance at 500 nm due to phytic acid was measured. The phytic acid concentration in an injected sample was calculated using standard solutions at concentrations of 0.1, 0.2, 0.3, and 0.5 mg mL⁻¹.

Daily air temperatures were recorded at JIRCAS, Ishigaki, Japan, and RCCRC, Vientiane, Laos.

The total DNA of each F₂ individual was extracted from the leaves using a conventional method with phenol-chloroform-isoamyl alcohol (Nippon Gene, Tokyo, Japan) (Gautam, 2022) and used for single-nucleotide polymorphism (SNP) analysis. 1K-Rice Custom Amplicon panel (1k-RiCA) (Arbelaez et al., 2019) was used to genotype F₂ individuals. The majority of the 1,024 SNP markers showed homozygosity with Kampeng and Non. Among the 1,024 SNP markers, 365 in the whole genome showed polymorphisms between Kampeng and Non and were used for QTL analysis.

Construction of the linkage map for the F₂ population and subsequent QTL analysis were conducted using R/qtl software (Broman et al., 2003). The chromosomal positions and effects of the putative QTLs were determined using composite interval mapping. The threshold for QTL detection was based on 1,000 permutation tests at a 5% significance level. The additive and dominant effects and the phenotypic variance (R²) explained by each QTL were estimated using the peak LOD score.

A Student’s t-test was conducted using SPSS software (version 23.0; IBM, Chicago, IL, United States) to compare the PA content and panicle number between Kampeng and Non. To compare PA content and days to heading among the four F₃ genotypes, a two-way analysis of variance (ANOVA) was performed to assess the effect of the two QTLs, qPA1 and Hd6, and their interaction. qPA, Hd6, and their interaction were considered as fixed effects, while replication was considered a random effect. The significance of the fixed effects was assessed using Tukey’s honest significant difference test at the 5% level. Normality and homogeneity of the variance were confirmed using the Shapiro–Wilk test and Levene’s test, respectively, before conducting the Student t-test and two-way ANOVA. Correlations between PA content, days to heading, panicle number, average daily mean temperature during grain filling, and PA content in F₂ individuals were also assessed using SPSS 23.0.

In both experiments in Japan and Laos, Kampeng (37.0 and 31.7 mg g⁻¹ grain, respectively) showed a significantly higher PA content than Non (33.8 and 25.0 mg g⁻¹ grain, respectively) (Figure 1C), whereas Non (12.8 and 11.5 plant⁻¹, respectively) produced significantly more panicles than Kampeng (8.5 and 6.0 plant⁻¹, respectively) (Figure 1D). The absolute values of the PA content and panicle number were lower in Laos than those in Japan.

Continuous variations were observed in days to heading, grain color, PA content, and panicle number among F₂ individuals in both experiments (Figure 2). The days to heading were transgressively segregated, ranging from 67 to 100 days (85.5 days in Kampeng and 75.3 days in Non) in Japan and from 61 to 117 days (84.8 days in Kampeng and 88 days in Non) in Laos (Figure 2A). The values for grain color, represented as L*, ranged from 23.0–77.0 (25.9 in Kampeng and 74.3 in Non) in Japan and from 34.7 to 77.6 (41.8 in Kampeng and 74.0 in Non) in Laos (Figure 2B). PA content also showed transgressive segregation, ranging from 25.3–54.5 mg g⁻¹ grain in Japan and from 23.9–54.4 mg g⁻¹ grain in Laos (Figure 2C). The panicle numbers ranged from 4 to 27 plant⁻¹ in Japan and from 7 to 19 plant⁻¹ in Laos (Figure 2D).

Since days to heading can affect agronomic traits, the relationship between days to heading and grain color, PA content, and panicle number was investigated. Days to heading were negatively correlated with grain color only in Laos (Figure 3A). Days to heading were also negatively correlated with the PA content in both Japan and Laos (Figure 3B). The absolute value of the correlation coefficient was higher in Japan (0.47) compared to Laos (0.22). No correlation was observed between days to heading and the panicle number (Figure 3C).

The QTL analysis of F₂ individuals detected three QTLs for days to heading in both Japan and Laos, among which two were commonly detected on the short and long arms of chromosome 3 (Figure 4A). The QTL on the long arm of chromosome 3 had a considerable effect, explaining 48.2–48.8% of the total variance in this trait (Table 1). The additive effect of the Kampeng allele was 7.3–12.0 days, indicating that the Kampeng allele increased the days to heading. Based on its location, the QTL was suggested to be Hd6 (Takahashi et al., 2001). One and three QTLs were detected for grain color in Japan and Laos, respectively. In both regions, one QTL was detected on the long arm of chromosome 4 (Figure 4B). The QTL showed a strong effect, explaining 48.3–36.4% of the total variance in this trait (Table 1). Based on its location and decreased additive effect with the Kampeng allele (blackening brown rice), the QTL was suggested to be Kala4, a gene from black rice that regulates anthocyanin pigmentation (Oikawa et al., 2015). Two and one QTLs were detected for PA content in Japan and Laos, respectively, in which one QTL was commonly detected at the end of the short arm of chromosome 1 (Figure 4C). The QTL explained 11.1–15.6% of the total variance in the trait. The additive effect of the Kampeng allele was 2.6 to 3.0 mg g⁻¹ grain, showing that the Kampeng allele increased the PA content (Table 1). Another QTL detected only in Japan was located near Hd6, and the Kampeng allele decreased the PA content. Two and one QTLs were detected for panicle numbers in Japan and Laos, respectively (Figure 4D). Each of these QTLs was detected in different chromosomal regions. Although Non produced more panicles than Kampeng in both experiments, the Kampeng allele of the three QTLs increased the panicle number (Table 1).

We named the QTL for PA content located at the end of the short arm of chromosome 1 qPA1 and attempted to validate the effects of qPA1 and Hd6 on days to heading and PA content using four different F₃ genotypes in Japan. Days to heading were significantly influenced by Hd6, whereas the effects of qPA1 and its interaction were not significant (Figure 5A). The mean number of days to heading in Kampeng homozygous for Hd6 was 15.2 days later than that of Non homozygous. The PA content was affected by both qPA1 and Hd6, but the effect of the interaction was not significant (Figure 5B), suggesting that qPA1 contributed to the enhancement of PA content, irrespective of Hd6. The mean PA content in genotypes of Kampeng homozygous for qPA1 was 7.1 mg g⁻¹ grain higher than that of Non homozygous. In contrast, the mean PA content in genotypes of Kampeng homozygous for Hd6 was 3.4 mg g⁻¹ grain lower than that of Non homozygous.

Finally, the effect of air temperature on the PA content during grain filling was investigated because the PA content in grains is typically determined 30 days after heading (Perera et al., 2019). The daily mean temperature during grain filling among F₂ individuals decreased by 10.5°C from 28.6 to 18.1°C in Japan and 7.3°C from 30.2 to 22.9°C in Laos (Figure 6A). Therefore, we calculated the average daily mean temperature for 30 days after heading for each F₂ individual. This ranged between 23.1 and 27.3°C in Japan and 24.8 and 27.8°C in Laos and was positively correlated with PA content in both Japan and Laos (Figure 6B). The absolute value of the correlation coefficient was higher in Japan (0.44) than that in Laos (0.22).