1. Introduction
Glutinous rice (sticky rice) is a type of rice found mainly in Southeast Asia countries (ASEAN), especially along the Me Kong subregion as Thailand, Lao PDR, Myanmar, Vietnam, and Cambodia, including the Southwestern area of China [
1]. The unique characteristics of glutinous rice are its opaque appearance (white color grain), low amylose content, and soft texture because of the high amylopectin content, which make it a common staple for food, desserts, and drinks [
2]. Amylose content is related to starch, and the loose network structures found in glutinous rice result in a low amylose content [
3]. Even though glutinous rice is of low economic value, it is a strategic crop that more than 20 million people consume on a daily basis [
1], and it is especially believed to provide basic nutritional security to rice farmers and their families [
4]. In Thailand, glutinous rice is omnipresent and mostly consumed in the Northeastern regions where approximately 17% [
5] of the land is affected by salt caused by underground salt rock [
6]. The glutinous rice variety most famous in Thailand is the RD6 cultivar due to greater cooking and eating quality, like softness, stickiness, and fragrant aroma [
7]. The RD6 variety is only one glutinous rice from three cultivars, including Khaw Dwak Mali 105 and RD15, which occupies lowland rain-fed areas the comprise over 70 percent of Northeastern Thailand. The RD6 rice variety has been approved since 1977, which, given its long time in use, is developed from KDML105 by a mutation method using gamma ray treatment. The mutant RD6 variety was changed to be glutinous rice with a low amylose content, but it still presents most of the agronomic traits of the parent, such as plant height, grain shape, and photo sensitivity, while disease susceptibility, including bacterial blight (BB) and blast disease, and saline soil influence plant growth and yield.
RD6 was susceptible to most Thailand local blast isolates collected from a diverse outbreak area [
8], which genotype is confirmed by microsatellite franking marker. In addition, blast susceptibility of the RD6 variety was demonstrated under field conditions by natural infection with mixed isolates during 2015–2018 at Khon Kaen province, Thailand. The result indicated that RD6 was susceptible to blast disease [
9]. Meanwhile, RD6 is also susceptible to BB disease infection in both greenhouse and field conditions. The greenhouse condition was confirmed during 2015–2016, which evaluated through disease scoring based on lesion length on symptomatic rice leaves at day 14 post-inoculation; the result clearly showed that RD6 varieties were more susceptible to BB disease infection of various isolates compared with resistance varieties [
10]. RD6 presented high susceptibility to salinity conditions, which affected plant growth at the seedling stage, confirmed under greenhouse conditions by validating twenty-one day seedlings on a salt solution and artificial soil salinity; the result showed that RD6 was highly susceptible for both conditions with a saline concentration of 12 dS m
−1 [
7]. After a breeding program, similar to each other, salinity evaluation under paddy field conditions at Northeastern Thailand, the yield of a salinity-susceptible rice variety decreased with increasing salinity levels above 8 dS m
−1 [
11].
To solve these constraints, Rice Project, Khon Kaen University located at Northeastern of Thailand has developed the glutinous rice RD6 for disease resistance specially focused on BB, blast resistance, and salt tolerance. The RD6 cultivar was bred for resistance to BB and blast disease with tolerance for saline soil by gene pyramiding into the RD6 parent. RD6 blast resistance was improved by pyramiding various resistance genes on chromosomes 1, 2, 11, and 12 into the RD6 cultivar through marker-assisted selection (MAS); the product has strength RD6 NILs [
8]. Continuously, RD6 development for various disease resistance was the destination for BB and blast resistance with gene pyramiding using MAS. RD6 NILs were produced from this research and possessed a total of five genes with four blast and one BB resistance genes; the NILs showed a greater resistance level for blast and BB diseases than its parent at the seedling stage [
12]. Furthermore, RD6 gene pyramiding was also developed for blast resistance plus saline soil tolerance, using the developed method as mentioned before. MAS was used for genotypic and phenotypic selection through the breeding steps to deliver promising lines. The experiments were conducted under greenhouse conditions by artificial inoculation for blast via the upland-short row and salt tolerance validated in a salt solution and artificial soil salinity. RD6 NILs were satisfied for both blast resistance and salt tolerance validation, and the NILs were greater than RD6, including other susceptible check varieties [
7]. After a breeding program, the plant breeder will investigate yield and its contributed performance of new improved genotypes across several environments
(multi-environment trials; MET
) to select and recommend a superior genotype for a target environment. This is because genotypes respond differently to changing environments, which causes yield instability [
13]. This yield instability is referred to as genotype by environment
(GE
) interaction. To define the presence of GE interaction in a MET, a data visualization tool, namely genotype by environment interaction
(GGE biplot
), is applied. GGE biplot is an effective tool to evaluate genotype evaluation and mega-environment identification of each genotype [
14]. GGE biplot was successfully used for GE interaction and yield stability in several crops [
15,
16,
17], including rice [
18]
. Therefore, in this study, the NILs were selected for field evaluation in Northeastern Thailand. Field evaluation was conducted at multiple locations in the vicinity of the area. The experiment was specially focused on the performance of NILs, including yield, agronomic traits, and field disease resistance by natural disease infection. The objective of this study was to evaluate RD6 NILs for yield performance under pressure of disease infections.
2. Materials and Methods
2.1. Plant Materials
Rice RD6 NILs obtained from MAS were selected from two breeding programs, with entry numbers 1–4 selected from pyramiding genes of blast and bacterial blight resistance. The line numbers 1–3 contained blast resistance genes on 4 chromosomes [
qBLch1,
2,
11,
12] and
xa5 recessive gene for BB resistance, whereas line number 4 carried only 4 blast resistance genes [
qBLch1 and 2 are major, while
qBLch11 and 12 are minor QTLs]. Other groups with entry numbers 5–8 were selected from pyramiding genes of blast resistance plus salt tolerance; the NILs were common in containing salt tolerance gene
Salt1 but differed in the number of blast resistance genes, as number 5 contained four genes,
qBLch1,
2,
11, and
12; number 6 contained three genes,
qBLch1,
11, and
12; number 7 contained two genes,
qBLch11, and
12; and number 8 contained two genes,
qBLch1, and
11 [
7,
12]. The eight NILs were evaluated together with the standard check variety RD6. A total of 9 rice genotypes were evaluated in this experiment (
Table 1).
2.2. Experimental Sites, Experimental Design, and Field Management
The experiment was conducted in Northeastern Thailand under paddy field conditions in the wet season of the years 2019 and 2020. The experiment site was spread over three diverse locations in each year. The first location was Khon Kaen province (16°45′10″ N 102°37′58″), almost the heartland of the region, which reported natural blast disease infection [
9]. The second site was Nong Khai province (17°51′0″ N 102°35′6″ E), the uppermost area of the region close to the Mekong River. This area has high rainfall amounts. The last location was Roi Et province (15°36′33″ N 103°48′1″ E), the largest area for high quality jasmine rice production with an arid and saline territory [
19].
The experimental design was laid out by a randomized complete block (RCB) design with 3 replications of each individual site. Seedling preparation of individual sites followed the layout of each site; one thousand seeds per plot were prepared in a labeled zip-lock bag. Seeds were soaked for 24 h, drained, and incubated for another 24 h to induce germination. The nursery soil preparation was performed fifteen days before sowing; the soil was prepared by sieving and mixed well, weighed equally, and put into 30 cm diameter submerged pots and soaked until the sowing date. The 30-day seedlings were transplanted into the field plot with three seedlings per hill with a spacing of 25 × 25 cm between plant and row inside a plot size of 2 × 4 m (8 m2). Field management, including weeding, pest, and pathogen, was controlled manually. Fertilizer was applied following the recommendation of the Department of Agriculture of Thailand. The fertilizer formula used was 16-20-0 for clay soil at a rate of 50 kg ha–1:62.5 kg ha–1:0 kg ha–1 (N:P2O5:K2O) and 16-8-8 for sandy loam soil at a rate of 50 kg ha–1:25 kg ha–1:25 kg ha–1 (N:P2O5:K2O).
2.3. Soil Evaluation and Rainfall Measurements
Soil sampling was randomized following soil sampling guidelines [
20]. A soil core was used for sampling 30 cm of soil depth. The randomized area was nine points spread around the experimental area of each site. Soil samples were analyzed by the Soil Laboratory of Khon Kaen University; the soil analysis focused on soil texture, EC, and soil nutrients.
The rainfall data used were manually recorded by a rain gauge at each individual experimental site. The rainfall data were an accumulation of twenty-four hours of a daily record for individual rainy days; the rainfall record duration of individual sites started from the transplanting date and finished on the harvest date.
2.4. Morphological Evaluation of Rice Traits
Morphological recording was focused on flowering date, plant height, and tiller number. Flowering date was recorded when the plant of each plot would start flowering; half the amount of flowered plants of each plot were recorded at days to 50% flowering. Plant height of each plot was measured from 4 plants randomly from the plot center at ripening stage, measuring from the ground surface along to the plant tip. Tiller number was recorded at the maximum tiller stage, which averaged 4 hills randomized from the plot center.
2.5. Yield and Yield Components
Yield evaluation data were collected from the data from the harvesting area of the plot inside 4 m2 or 64 hills without border effects. The harvest date of the individual plot was based on 30 days after the 50% flowering date of each plot. Rice grains were collected together with a label inside a net bag, and seed moisture was decreased by sundry under a translucent air flow greenhouse; fourteen percent seed moistures of individual plots were threshed and processed, and perfectly filled seed of each plot were measured by a digital weighing scale, which converted into weight per hectare.
Yield components were recorded for panicle number, harvest index, seed per panicle, sample seed weight, and seed size. Panicle number was recorded by the average from 4 hills randomized on the center of each plot without border effects; record time was conducted on grain filling stage by manual numerate observation. The harvest index was the harvest of 4 whole plants randomized from the harvest area of each plot collected in a net bag together with a label inside; biological yield and economic yield after sundry were measured by a digital weighing scale (HI = Biological yield/Economic yield; Dry grain yield/Total dry weight). Seed per panicle was collected from 4 perfect panicles randomized from each plot after harvest. The yield, harvest index, and number of seeds per panicle were averaged from filled seed of four panicles collected. Sample seed weight was conducted after yield weighing by preparing 1000 perfectly filled seeds per plot from seed yield, which was measured by a digital weighing scale. Seed size was averaged from the measurement of 10 paddy seeds randomized from the seed yield and measured by a vernier caliper.
2.6. Disease Screening/Index
Prior to conducting NILs evaluation, all improved lines were tested for disease resistance via artificial inoculation against some isolate pathogens. This research was in stage of multilocational trial to confirm their resistance to different combinations of natural pathogens in individual sites. The improved lines were preliminarily evaluated by artificial inoculation and then evaluated under natural infection in this study. Disease evaluation followed the severity scale of the Standard Evaluation System for rice (SES) [
21,
22]. Disease severity scores were recorded from slight symptom to higher severity from ten hills of each plot. The lowest severity scores were based to analyze and categorize resistance levels (
Table 2 and
Table 3) [
23].
2.7. Statistical Analysis
Individual data were analyzed via analysis of variance (ANOVA) following a randomized complete block (RCB) design. Prior to combined analysis, the normality and homogeneity of the data from all environments were verified. A combined analysis was used to analyze the interaction of genotype and environment (GxE interaction). A least significance difference (LSD) at
p < 0.05 [
24] was used for mean comparison for quantitative data. Yield stability and yield response were analyzed consistent with the AMMI model (cluster dendrogram and GGE-biplot); yield stability and yield response analyses were carried out using R software Version 4.0.2 [
25,
26].
3. Results
3.1. Environment Characterizations
Environmental data were collected for climate data and disease occurrence. The climate data collected were rainfall with distribution, soil properties, and altitude; the data were used together from other experiments, such as the field evaluation of KDML105 introgression lines [
27] (
Figure 1). Disease occurrence of blast and bacterial blight (BB) was found and recorded.
The altitudes of the experimental area ranged from 120 to 220 m above sea level (MASL); the Khon Kaen site was higher than Nong Khai and Roi Et, respectively. Soil textures were different among the locations; Khon Kaen and Roi Et had sandy loam soil, and Nong Khai had clay soil. Rainfall data among the environments were different for the number of rainy days, number of rainy weeks, and total rainfall value (
Table 4).
Rainfall of all environments had an equivocal difference among the sites. The rain accumulates ranged from 501.1 to 740.5 mm; the higher rainfall accumulate environment is Roi Et 2019, followed by Nong Khai 2019, Nong Khai 2020, Roi Et 2020, Khon Kaen 2019, and Khon Kaen 2020, respectively. Overall, rain accumulates for all three locations in 2019 were higher than those in 2020. However, rainfall distributions were not related to rainfall accumulations. In 2020, there was a wider rain distribution than in 2019. The number of rainy days for Khon Kaen 2020 was higher than 2019 with 41 and 28 days, respectively; the number of rainy days for Nong Khai location in 2020 was higher than 2019 with 34 and 29 days, respectively; but for the Roi Et location, there was no statistically significant difference, with 33 rainy days in both years. In addition, the number of rainy weeks was evidently concerned, which, in 2020, all locations had a wider rain distribution than 2019. The result indicated, in year 2019, there was a short rainfall distribution with heavy rain; the rain accumulation per week showed the highest peak at above 200 mm/week, with less rainfall after the maximum tiller (MT) stage. In year 2020, rainfall was substantially widely distributed, with rainfall accumulation per week showing a wide distribution from the start of planting (TP) until the flowering-grain filling stage.
In this study, two disease occurrences, blast and BB, were encountered. Blast disease infections were found at four environments: Khon Kaen 2019, Khon Kaen 2020, Nong Khai 2020, and Roi Et 2020. BB disease infections were found at three environments conducted in 2020: Khon Kaen 2020, Nong Khai 2020, and Roi Et 2020. The comparison among experiment years showed that the year 2020 was found to be greater than the year 2019 in disease occurrence, which was related to longer rainfall distributions, an important factor for both disease spread and occurrence.
3.2. Blast and Bacterial Blight Disease Virulence
Blast disease score of virulence was found at four environments: Khon Kaen 2019, Khon Kaen 2020, Nong Khai 2020, and Roi Et 2020. The grand mean of severity score was found at Khon Kaen 2020, with an average score of 2.69, followed by Khon Kaen 2019, with an average score of 1.89, while Nong Khai 2020 and Roi Et 2020 had average scores of 1.03 and 0.81, respectively. The effect of experimental site on severity score was highly significant (
p ≤ 0.01). The RD6 variety was found to be blast susceptible, as a higher disease score than that of the NILs was observed. All of the blast resistance NILs were great in gene expression for blast resistance, with scores lower than RD6. At the Khon Kaen 2019 site, the NILs were strong in resistance to blast, with a slight score of 1.12–1.50, which is still significant among the NILs due to its difference in genetic introgressed. Meanwhile, the RD6 variety appeared as highly susceptible, with a blast score of 6.42. For the Khon Kaen 2020 site, the rice NILs had scores that ranged from 1.97 to 3.01, which was not different among the lines, while the RD6 variety showed a higher blast score than the NILs at 4.29. At the Nong Khai 2020 site, slight blast disease symptom was shown on the NILs, with scores ranging from 0.57 to 1.17, which slightly differed among the lines, and the RD6 variety showed a higher blast score of 3.13. The last site of blast observation was Roi Et 2020, also having slight blast symptom; the rice NILs had scores ranging from 0.50 to 0.80, which is not different among the lines, and the RD6 variety showed a higher blast score of 2.90 (
Table 5).
BB virulence was found at only three sites evaluated in 2020. The BB scoring comparison of each site showed a significant difference (
p ≤ 0.05) for Khon Kaen 2020 and highly significant difference (
p ≤ 0.01) for Nong Khai 2020 and Roi Et 2020. The highest average BB score was found at Nong Khai, with an average score of 4.44, whereas at the inferior sites, Khon Kaen and Roi Et had average scores of 4.17 and 3.30, respectively. Interestingly, the NIL BC
2F
5 2-8-2-26-1 was the lowest in severity score over the other lines at all three sites. In contrast, RD6 was the most susceptible over all entries at all three sites. Moreover, NIL BC
2F
5 2-8-2-26-1 was greater than the other NILs carrying the
xa5 gene for BB resistance. It demonstrated the expression of horizontal BB resistance (
Table 5).
3.3. Yield Evaluation on Multi-Environments
Yield evaluation over six environments showed a high significance among sites (
p ≤ 0.01) but not significant between genotypes. Moreover, an interaction between environment and genotype (GxE effect) was detected at
p ≤ 0.05 (
Table 6).
Yield comparison among the environments showed that Khon Kaen 2019 had a higher yield, followed by Nong Khai 2020, Khon Kaen 2020, Roi Et 2020, Roi Et 2019, and Nong Khai 2019, with grain yields of 3535, 3484, 3195, and 3087 kg/ha
−1, respectively, which are not different among the five lines group (
Table 7). Khon Kaen 2019 is, thus, the greatest in grain yield compared with the other environments.
Environment (E) and genotype (G) interaction for yield of the nine rice genotypes under six evaluation sites was significant (
p ≤ 0.05). Yield comparison among genotype under a particular environment was different. At Khon Kaen 2019, the rice yield among genotype was not significantly different, with an average of 4651 kg ha
−1 and a yield ranging from 4174 to 4913 kg ha
−1. Whereas, Nong Khai 2019, Roi Et 2019, Khon Kaen 2020, Roi Et 2020, and Nong Khai 2020 had location mean yield and yield ranges of 3087, 2490–3419 kg ha
−1; 3149 kg ha
−1, 2599–3737 kg ha
−1; 3484 kg ha
−1, 3206–3745 kg ha
−1; 3195 kg ha
−1 2669–3467 kg ha
−1, and 3535 kg ha
−1, 2842–4771 kg ha
−1, respectively. Yield among genotype in these environments was not significantly different. Except for the Nong Khai 2020 environment, rice yield among genotype was significantly different at
p ≤ 0.05 level. The NIL BC
2F
5 2-8-2-26-1 gave a yield of 4771 kg ha
−1 that was significantly different from the other genotypes. Those genotypes were not significantly different among themselves. In addition, the environment Nong Khai 2020 was found to have a considerable yield range value by yield range of 1929 kg ha
−1, and its range was higher than the other five environments (
Table 8).
3.4. Yield Stability and Adoption
Yield stability analyzed by GGE-biplot showed stability of the genotype G1-G8 compared with RD6 under environments E1-E6 (arrow lines). GGE-biplot explained for PC1 with 46.4% and PC2 with 25.9%. The stability map shows that the yields of NILs G1 (Blast + BLB-36), G3 (BC
2F
5 2-8-2-52), G4 (BC
2F
5 9-1/15-1-27), G5 (BC
4F
5 132-12-61), G6 (BC
4F
5 132-25-18), and G7 (BC
4F
5 132-98-87) were of greater stability than RD6, and NILs G2 (BC
2F
5 2-8-2-26-1) and G8 (BC
4F
5 132-174-54) were of low stability. However, specific adaptation of crop yield was showing for NILs adaptation to various environments: G2 (BC
2F
5 2-8-2-26-1) was high yield for environment E2 (Nong Khai 2019), E4 (Khon Kaen 2020), and E5 (Nong Khai 2020); G8 (BC
4F
5 132-174-54) had greater yield for environment E1 (Khon Kaen 2019) and E3 (Roi Et 2019); and RD6 had greater yield for environment E6 (Roi Et 2020) (
Figure 2).
3.5. Yield Components and Agronomic Traits Evaluation
Yield components and agronomic traits means of individual genotypes for comparison were averaged over six environments. Plant height (PH) average among genotype was highly significant (
p ≤ 0.01). RD6 was the highest with 161 cm of plant height, and G2 (BC
2F
5 2-8-2-26-1) was the lowest plant height with 142 cm; the plant heights of the other NILs were similar among genotypes. Days to flowering (DTF) was highly significant (
p ≤ 0.01); almost all genotypes were not significant, with the range of 106–110 days, but only one of its G2 (BC
2F
5 2-8-2-26-1) was shorter in day to flowering with an average 99 days. The tiller number (TN) and panicle number (PN) were both not significant among genotypes, with means of 10.4 and 10.2 for tiller number and panicle number, respectively. Seed per panicle (SP) was not significantly different, with a mean of 164 seeds per panicle. Thousand seed weight (1000 SW) was significant at 95%; G6 (BC
4F
5 132-25-18) was higher in seed weight, but the other NILs did not have a different seed weight than that of RD6. Harvest index (HI) was highly significant at
p ≤ 0.01; all of the NILs had a higher harvest index than that of RD6. The NILs’ harvest index ranged between 0.37 and 0.40, whereas RD6 was 0.33 in harvest index (
Table 9).
5. Conclusions
Rice RD6 NILs evaluation was performed at three locations replicated by two years from 2019 to 2020. NILs promising lines selection for utilization on Northeastern of Thailand was the focus for agronomic traits and yield adaptability under pressure of natural disease infections. The high yield stability genotypes G1, G3, G4, G5, G6, and G7 had greater yield stability than RD6 and are recommended for wider utilization. On the other hand, two genotypes, G2 and G8, had low yield stability; G2 had a greater yield maintained under severe BB disease infection that was appropriate for a severe BB spread area. In contrast, G8 had higher yield under low disease pressure. In addition, G2 expressed segregation traits in non-photosensitivity and lower plant height that could be recommended for an irrigation area or to replace RD6 for early maturity when subjected to drought and flooding avoidance. Furthermore, all of the rice genotypes need more pyramiding BB resistance genes to increase the durable resistance.