Multi-Trait Index-Based Characterization of Putative Drought-Heat Tolerant Gamma-Irradiated Rice Mutants Through Artificial Screening at the Seedling Stage

Abstract

Rice is highly vulnerable to concurrent drought-heat stress, which intensifies the disruption of plant growth and can cause high seedling mortality. Limited genetic diversity further constrains breeding for combined stress adaptation. Thus, mutation breeding has emerged as a solution for inducing beneficial variability in rice genomes. Based on this solution, this study aimed to identify rice mutants tolerant to drought and heat through artificial screening combined with multi-trait index analysis. Experiments were conducted from April to August 2025 at Hasanuddin University in two sequential phases: seedling screening in a controlled stress chamber and pot evaluation under semi-controlled conditions. Mutants were derived from the seeds of the double haploid (DH) rice line HS1-28-1-5, which was previously developed through anther culture and exhibited potential abiotic stress adaptability. DH seeds were exposed to gamma irradiation (200–1000 Gy) as the M1 population. Based on radiosensitivity evaluation, 200 and 400 Gy were selected and used in the present study as the M2 populations. Rice M2 evaluation results indicated that 200 Gy irradiation enhanced phenotypic variability with improved performance, whereas M2 400 Gy resulted in broader but less stable variation. Yield components were strongly associated with weight per clump, with panicle length being the primary direct contributor and weight per panicle mediating indirect effects. Principal component analysis accounted for 89.13% of the total variation, which was predominantly driven by yield traits. Integrated weighted average absolute score–tolerance score analysis classified genotypes into tolerance groups, identifying 33 putative tolerant mutants, with M2 200 Gy mutants demonstrating superior adaptability to drought-heat stress. These findings suggest that moderate irradiation coupled with multi-trait index selection provides a reproducible framework for identifying putative drought-heat-tolerant rice candidates at early stages while retaining relevance to subsequent yield recovery.

1. Introduction

Agricultural production systems are currently facing unprecedented challenges from global climate change, which alters environmental conditions. In response to these changes, phenomena such as rising temperatures, irregular precipitation patterns, and increased frequency of extreme weather events have intensified abiotic stresses that threaten crop productivity worldwide [1]. Drought and heat stress have emerged as the major abiotic stresses affecting plants. Both stresses critically impact fundamental plant processes by disrupting physiological functions, such as photosynthesis, stomatal regulation, and cellular homeostasis [2,3], which can lead to yield reductions of up to 50% [4], depending on crop type, developmental stage, and stress intensity. Under tropical conditions, drought-heat stress frequently occurs simultaneously rather than individually, often resulting in more severe impacts on plant growth, particularly in major staple crops, such as rice [5,6].

Despite its importance in providing carbohydrates to more than half of the global population, rice is highly sensitive to environmental stresses, particularly drought and heat stress [7,8]. The increasing occurrence of these stresses in major rice-growing regions has raised concerns about the stability of rice production throughout its life cycle. Among the different developmental stages, the seedling stage is particularly sensitive to abiotic stress, as it determines early plant vigor and successful crop growth [9,10]. Several studies have shown that exposure to drought and heat during this stage can impair root and shoot development, reduce biomass accumulation, and compromise plant survival, leading to the early death of seedlings [6,11]. Therefore, screening at the seedling stage is a crucial step for developing tolerant rice genotypes.

The development of tolerant rice genotypes has been widely pursued through conventional breeding and genomic approaches [12,13,14]. However, these methods are often time-consuming, costly, and labor-intensive, requiring multiple breeding cycles and extensive field evaluations [15]. Moreover, most breeding efforts have primarily focused on developing single-stress (mono-tolerant) rice, thus limiting their effectiveness under field conditions, where multiple stresses frequently occur simultaneously. Developing multi-tolerant varieties is inherently difficult because of narrow genetic diversity and limited variability within existing germplasm [15,16,17]. These constraints highlight the limitations of conventional and genomic breeding. Given these challenges, there is a pressing need for approaches that can effectively develop multi-tolerant rice.

Mutation breeding has emerged as a valuable approach for inducing diverse genetic variation within existing mutant populations without the introduction of external genes [18,19]. It offers rapid selection of new phenotypes and overcomes the limitations of naturally occurring variability [20,21,22]. However, mutation breeding faces multiple challenges, including the fact that most induced mutations are neutral or deleterious, making it difficult to identify beneficial changes. The frequency of desirable mutations is often low, requiring the evaluation of large populations [23,24]. Evaluating such populations under field conditions is time-consuming and labor-intensive [25,26]. Thus, artificial screening under controlled stress chambers provides a systematic approach that allows these populations to be scaled down in an environment similar to field conditions [27]. By creating artificial conditions in a controlled setting, this method efficiently narrows down large populations to a manageable set of candidate genotypes for further evaluation [28], while maintaining relevance to actual growing conditions for screening multi-tolerant rice.

Screening gamma-irradiation-induced rice mutants is often complicated by generational timing. Selection at the M1 generation is generally unreliable, as M1 plants have not yet undergone segregation and remain affected by chimerism [29,30]. For this reason, most mutation breeding programs conduct screening at the M2 generation once segregation begins to stabilize [31]. However, screening M2 populations using a single trait, such as yield alone, often results in poor selection efficiency, as yield performance does not necessarily reflect the genotype’s overall stress response [32], risking the exclusion of genetically promising mutants whose tolerance potential is not captured by yield alone.

To address these limitations, the present study applied a multi-trait index selection approach by combining principal component analysis (PCA) with weighted absolute scores (WASB) and tolerance scoring to evaluate multiple traits simultaneously in the M2 generations under a single-environment trial, instead of relying only on yield or one trait. This approach allows trait relationships and overall genotype adaptability to be captured collectively. Despite this demonstrated advantage, relatively few studies have applied multi-trait index approaches specifically to M2 mutant screening for combined drought and heat stress tolerance.

Thus, this study aimed to identify multi-tolerant rice mutant genotypes through the integration of tolerance scoring and multi-trait index approaches and establish a systematic and robust screening framework for the selection of multi-stress-tolerant mutant lines.

2. Materials and Methods

2.1. Experimental Site

The study was conducted at the Green House Experimental Farm, Hasanuddin University, Tamalanrea District, Makassar (7 m above sea level) during the dry season (April to August 2025). The experiment utilized two greenhouse facilities, a stress chamber, and a Center of Excellence (CoE) greenhouse, assigned to different screening stages. The heat chamber, used for seedling-phase screening (Figure S1), consisted of a 2 m (front) × 1.5 m (side) enclosed light steel frame structure fully covered with polyethylene plastic on all sides, with concrete block flooring (Figure S2), allowing maximum solar radiation and heat accumulation to induce high-temperature stress conditions.

Subsequently, selected genotypes from the seedling phase screening were evaluated in the CoE greenhouse at the vegetative to reproductive stages to assess growth performance and yield-related traits under semi-controlled conditions. At both facilities, rainfall was excluded by the greenhouse structures, whereas temperature and relative humidity were continuously monitored using digital thermohygrometer acquired from local agriculture store in Makassar, Indonesia.

2.2. Rice Mutant Genotype Origin and Plant Materials

The rice mutant genotypes used in this study were derived from previous breeding work reported by Anshori et al. (2018; 2019) [33,34] developed through anther culture to produce double-haploid (DH) rice lines at IPB University, Indonesia. Specifically, the genetic material employed was one selected line among 56 DH rice lines, designated as HS1-28-1-5. These specific lines were selected based on their agronomic and abiotic stress adaptability. As a DH line, this genotype is fully homozygous, providing a uniform and stable genetic background that enables the direct expression of induced mutations without segregation effects. This genetic uniformity makes DH-derived materials particularly suitable for mutation breeding [35].

Seeds from this line were exposed to gamma irradiation at doses of 200, 400, 600, 800, and 1000 Gy to generate M1 mutant populations. Based on prior radiosensitivity evaluations, 200 and 400 Gy doses exhibited favorable agronomic performance, whereas higher doses (600–1000 Gy) resulted in deleterious effects [36]. Therefore, mutant lines from 200 and 400 Gy treatments in M1 mutant populations were advanced and used as M2 populations in the present study, where induced mutations were more stable and phenotypically expressed.

Four rice mutant genotypes, namely, M2 200 (6), M2 200 (15), M2 400 (5), and M2 400 (3), each with 91 seeds, were used. A non-irradiated line, such as M2 0, was also included as a wild-type (positive control), while IR29 was used because of its susceptibility to drought-heat stress (negative control) [37] to verify stress expression, with a total of 28 seeds per control.

2.3. Principle of the Controlled Stress Chamber Screening System

The controlled stress chamber in the present study was developed based on the principle of simultaneous drought-heat stress imposition. Drought stress was induced by withholding irrigation, whereas heat stress was applied through passive solar heat accumulation within a transparent polyethylene enclosure, which elevated air temperature while limiting convective heat loss. Therefore, through the combination of increasing air temperature, progressive soil water depletion, and elevated vapor pressure deficit (VPD), a compound drought-heat environment was created. Similar controlled-environment systems have been widely employed for rice stress screening because they provide uniform and reproducible stress conditions for genotype evaluation while minimizing environmental variation. Accordingly, the controlled-stress chamber used in this study was modified from [38] established screening principles to simulate combined drought-heat stress during seedling stage screening.

2.4. Exploratory Research Procedure

This study used an exploratory research approach to systematically evaluate mutant rice lines for their adaptability to combined drought–heat stress. Such a design is appropriate for the early-stage screening of large mutant populations, as it enables flexible, iterative evaluation under conditions of limited prior information regarding genotype performance. This research procedure facilitates the rapid identification of putative tolerant genotypes while optimizing resource allocation, particularly when dealing with high genetic heterogeneity induced by mutagenesis.

The screening phase was conducted in a sequential two-stage selection process. In Stage 1, seedling-phase screening was performed in seed trays under controlled stress conditions to identify genotypes that exhibited early tolerance. Mutant lines were assessed alongside non-irradiated controls and standard check varieties to establish comparative benchmarks for the phenotypic traits. All surviving genotypes were advanced to stage 2, which involved evaluation during the vegetative to reproductive phases under semi-controlled conditions. This stage focused on assessing growth dynamics and yield-related traits to determine whether seedling-stage tolerance was associated with multi-trait subsequent growth recovery and yield performance.

A separate non-stress treatment was not included because the primary objective of this study was not to partition the individual effect of mutation and stress on yield, nor to identify the highest-yielding genotype under optimum conditions after stress treatment. Instead, selection was primarily based on linking seedling stage tolerance score and the multi-trait index, which enabled comprehensive identification of putative drought-heat-tolerant mutants by simultaneously considering multiple stress-response traits. Therefore, agronomic and yield component traits at stage 2 were evaluated as indicators of post-stress recovery performance in relation to previous drought-heat screening during seedling stage screening.

2.4.1. Seedling Preparation

The experimental procedures involved a sequential workflow comprising growth medium preparation, seedling establishment, stress imposition, recovery evaluation, and subsequent growth assessment under pot conditions. The growth medium was prepared by mixing soil and rice husk at a ratio of 2:1 (v/v). The mixture was thoroughly homogenized and used for both seedling trays and pots. In trays, the medium was maintained under moist conditions to support germination and early growth, whereas in pots, it was brought to a saturated (puddled) state to simulate lowland rice conditions.

Seedling-stage screening was conducted using four seedling trays for each rice mutant genotype. The seedling trays comprised 105 planting cells arranged in a structured manner. The tray was divided into three sections: the left section containing 42 cells allocated for mutant lines, a central section comprising two columns of seven cells each assigned to the non-irradiated control (M2 0) and the susceptible check (IR 29), and the right section containing 49 cells for additional mutant lines. This configuration allowed direct within-tray comparison between the mutant genotypes and controls, as illustrated in Figure S3.

Prior to sowing, the seeds were soaked in water for 24 h to promote uniform germination. The pre-germinated seeds were then sown into prepared trays and grown for 16 days under controlled conditions. The seedlings were arranged in labeled rows to ensure accurate genotype identification. During this establishment phase, routine management practices were applied, including manual weeding, regular watering, and a single application of NPK fertilizer (3 g/L per tray at 10 days after sowing) to ensure uniform growth prior to stress imposition.

2.4.2. Multi-Stress Imposition Inside Controlled Stress Chamber

Multi-stress treatment combining drought and high-temperature stress was imposed at the seedling stage, starting 10 days after sowing (DAS). Seedling trays were transferred into a stress chamber consisting of a fully enclosed transparent plastic structure designed to maximize solar radiation and heat accumulation in the chamber. Drought stress was induced by withholding water, and high-temperature stress was controlled within the chamber. Stress was applied in two phases (15–17 and 21–23 DAS). Microclimatic conditions, including air temperature and relative humidity, and daily maximum and minimum air temperature and relative humidity, were recorded directly and monitored daily using a digital hygrothermometer. These measurements represent the daytime microclimatic conditions recorded between 05:00 A.M. and 18:00 P.M. Subsequently, microclimatic data were used to calculate heat degree days (HDD), saturation vapor pressure, actual vapor pressure, and vapor pressure deficit according to the FAO-56 Penman-Monteith methodology as presented in Table S1 [39].

To further ensure that the imposed drought conditions were effectively achieved, plant response to soil water deficit was calculated utilizing the fraction of transpirable soil water (FTSW), which represents the proportion of soil water available for transpiration between field capacity and a lower limit approximately 10% of its maximum, approaching conditions near the permanent wilting point. In this research, field capacity was defined as the soil water content after free gravitational drainage had ceased. FTSW was determined gravimetrically from pot mass during the screening phase using a weighing scale, where $W_i$ is the pot weight on the day of measurement, $W_{Fc}$ is the pot weight at field capacity, and $W_{Final}$ is the pot weight at the end of the drying cycle:

$${FTSW}_i= {\displaystyle \frac{(W_i-W_{Final})}{(W_{Fc}-W_{Final})}}$$

(1)

At the onset of each screening phase, the soil moisture was adjusted to field capacity (FC) to ensure a uniform initial water status across the screening phases. Following stress exposure, the seedlings were removed from the chamber after screening in phase 1 and intensively rewatered to initiate a 3-day recovery (readjustment) period. After screening in phase 2, the plants were rewatered and maintained under optimal conditions, allowing normal growth to resume until 38 DAS.

Selected seedlings were transplanted into pots under controlled, puddled soil conditions at 38 DAS to simulate lowland conditions. All pots were standardized to have equal soil mass and composition to ensure uniform growth conditions, with water maintained through regular irrigation. Plants were grown to maturity under standard agronomic practices, and harvesting was conducted when approximately two-thirds of the panicles had reached physiological maturity. The experimental timeline for the imposition of multi-stress is shown in Figure 1.

2.5. Data Acquisition and Multi-Trait Index Analysis Pipeline

The collected data included plant height (PH), number of total tillers (NTT), number of productive tillers (NPT), day of flowering (DoF), day of maturity (DoM), flag leaf length (FLL), flag leaf width (FLW), number of total panicles (NP), panicle length (PL), weight per panicle (WPP), weight of 1000 grains (W1000G), number of total grains (NTG), number of filled grains (NFG), number of empty grains (NEG), percentage of filled grains (PFG), percentage of empty grains (PEG), and weight per clump (WPC). Tolerance scoring based on the Standard Evaluation System (SES) of the International Rice Research Institute (IRRI) [40] was also performed, including leaf rolling scoring (LRS), leaf drying scoring (LDS), heat tolerance (HTol) [38], and spikelet fertility (SF).

Prior to the analyses, all measured variables were subjected to appropriate scaling procedures to ensure comparability and prevent bias arising from differences in measurement units and value ranges [41]. Agronomic traits were standardized using Z-score normalization, a process that transforms each observation based on its deviation from the mean relative to the standard deviation, resulting in variables with a mean of zero and a variance of one. In contrast, tolerance-related variables derived from the Standard Evaluation System (SES) were normalized using min–max normalization, which adjusts values to fit within a specified range, typically between 0 and 1 [42].

Spearman’s rank correlation analysis quantified the strength and direction of monotonic relationships among agronomic traits [43], particularly to identify traits significantly associated with WPC under stress conditions. Traits significantly correlated with WPC were then subjected to path coefficient analysis following Wright’s method, enabling decomposition of correlation coefficients into direct and indirect effects [44,45]. This distinguished traits with true causal influence on yield from those contributing indirectly. Traits with substantial direct effects on WPC were incorporated into PCA in Rstudio IDE version 2026.04.0 Build 526 using R version 4.6.0. PCA were utilized to address multicollinearity and reduce dimensionality. Principal components were retained using the Kaiser criterion (eigenvalues > 1) [46,47], facilitating identification of key trait combinations underlying genotypic variation and performance.

Subsequently, the Weighted Absolute Score (WASB) was calculated using the principal component scores, where each genotype (i) was evaluated based on its position in the reduced multivariate space [48]. This approach integrates the contributions of multiple traits into a single composite index, enabling more robust genotype ranking. Overall, the integration of correlation analysis, path analysis, PCA, and WASB establishes a systematic and hierarchical selection framework that allows the identification of superior genotypes based on both individual trait effects and their complex interrelationships. WASB calculation is as follows:

$${WASB}_{PC1}=\left| {PC1}_i\right| \times {\displaystyle \frac{\left({Var}_{PC1}\right)}{\left({Var}_{PC1}+\left({Var}_{PC2}\right)\right)}}$$

(2)

$${WASB}_{PC2}=\left| {PC2}_i\right|\times {\displaystyle \frac{\left({Var}_{PC2}\right)}{\left({Var}_{PC1}+\left({Var}_{PC2}\right)\right)}}$$

(3)

WASB value was further combined with tolerance scoring (TScore), which was averaged from four tolerance scores and normalized using min-max normalization. TScore value was obtained from four tolerance parameters. The thresholds were 0.00–0.40 (Tolerant), 0.40–0.70 (Moderate), and 0.70–1.00 (Susceptible). The stricter threshold for the tolerant class was applied to increase selection intensity by assigning only genotypes exhibiting consistently favorable performance across all SES scoring.

3. Results

3.1. Rice Mutant Seedling Survival and Tolerance Scores Under Combined Drought-Heat Stress

The temporal dynamics of the fraction of transpirable soil water (FTSW) confirmed the efficiency of the stress chamber in simulating drought-heat stress in screening rice mutant seedlings (Figure 2). During screening phase 1, FTSW declined from 100% to 8% by 17 DAS, whereas in screening phase 2, it decreased from 100% to 12% by 23 DAS. Collectively, these environmental conditions indicate that the artificial screening concept utilizing stress chamber infrastructure has the potential for generating sufficiently severe and reproducible methods for screening rice under combined drought-heat stress.

The effectiveness of the screening protocol was reflected in the survival responses of the evaluated rice populations (

Table 1. Seedling survival of mutant rice M2 under drought-heat stress.

Group	Early Seedling Population	Survive	Survival (%)	Mortality	Mortality (%)
M2 0 (Wild Type)	28.00	6.00	21.43	22.00	78.57
M2 200 (15)	91.00	18.00	19.78	73.00	80.22
M2 200 (6)	91.00	17.00	18.68	74.00	81.32
M2 400 (3)	91.00	6.00	6.59	85.00	93.41
M2 400 (5)	91.00	6.00	6.59	85.00	93.41
IR29	28.00	0.00	0.00	28.00	100.00
Total	420.00	53.00	12.62	367.00	87.38

). 420 seedlings were screened, of which only 53 survived the combined drought-heat stress, corresponding to an overall survival rate of 12.62%, whereas 87.38% of the seedlings failed to recover following stress exposure. The susceptible check variety, IR29, exhibited complete mortality (100%), confirming the severity of the imposed stress. Among the evaluated populations, the wild type (M2 0) recorded the highest survival rate (21.43%), followed by two mutant populations derived from 200 Gy irradiation, M2 200 (15) and M2 200 (6), with survival rates of 19.78% and 18.68%, respectively. In contrast, both 400 Gy mutant populations, M2 400 (3) and M2 400 (5), exhibited substantially lower survival (6.59%), indicating reduced capacity to withstand the combined drought-heat stress treatment.

To further characterize surviving seedlings’ response, tolerance potential was evaluated using the SES, including LDS, LRS, Htol, and SF. The tolerance score results, showing a distribution pattern across parameters. Notably, M2 200 exhibited more individuals at low-medium scores than M2 0 and M2 400 (

Table 2. Tolerance scores of mutant rice M2 under drought-heat stress.

Scoring	Evaluation Criteria	Tolerance Score	Category	M2 0	M2 200	M2 400
LDS	No leaves are drying	0	Highly tolerant	0	0	0
	The tip of the leaf is slightly drying	1	Tolerant	2	5	2
	Leaves dry up to 1/4 of each leaf	3	Slightly tolerant	1	8	7
	1/4–1/2 portion drying on all leaves	5	Moderate	2	14	2
	More than 2/3 of all dried leaves	7	Slightly susceptible	1	8	1
	All parts of the leaf have dried out	9	Susceptible	0	0	0
LRS	There are no curled leaves	0	Highly tolerant	0	0	0
	The leaf begins to curl	1	Tolerant	3	11	8
	Leaves curling to form a V shape	3	Slightly tolerant	1	8	2
	Leaves curl forming the letter U	5	Moderate	1	8	1
	The leaf curls to form the number 0, with the tip and base of the leaf touching each other	7	Slightly susceptible	1	8	1
	The leaf is fully curled up	9	Susceptible	0	0	0
HTol	Normal growth, slight curling at the leaf tips	1	Highly tolerant	2	11	4
	Almost all the leaves are curling, turning yellow, and the number of leaves is decreasing	2	Moderately tolerant	3	16	6
	All the leaves have dried up, and some have undergone necrosis	3	Susceptible	1	8	2
	All the plants are dying or dead	4	Highly susceptible	0	0	0
SF	>80% Spikelet Fertility	1	Tolerant	3	13	7
	61–80% Spikelet Fertility	3	Slightly tolerant	2	14	3
	41–60% Spikelet Fertility	5	Moderate	1	8	1
	20–39% Spikelet Fertility	7	Slightly susceptible	0	0	1
	0–19% Spikelet Fertility	9	Susceptible	0	0	0

LDS: leaf drying score, LRS: leaf rolling score, HTol: heat tolerance, SF: Spikelet fertility.

). For leaf drying, the distribution shifted from M2 0, which had few individuals across scores, to M2 200, concentrated at medium scores (3–7) with a peak at score 5, then declined in M2 400, with lower frequency centered at score 3. A similar pattern appeared in leaf rolling, with M2 200 dominant at low to medium scores (1–5), whereas M2 0 and M2 400 showed narrower distributions with fewer individuals. For heat tolerance, individuals increased from M2 0 to M2 200, peaking at score 2, then decreased at M2 400, forming an optimum-like response where M2 200 had the most individuals in the main score category. Spikelet fertility showed a similar trend, where M2 200 was concentrated at scores 1 and 3 with the highest frequency, while M2 0 and M2 400 had lower, more dispersed distributions, including higher scores in M2 400.

3.2. Phenotype Variation and Trait Relationships of Gy-Irradiation Induced Mutant in Rice Under Drought-Heat Stresses

The distribution of agronomic traits differed between wild-type and mutant populations (200 and 400 Gy). Mutant lines showed broader distributions for most traits, indicating increased phenotypic variability induced by gamma irradiation (Table S2). PH was higher in the 200 Gy treatment, whereas the 400 Gy treatment showed greater dispersion, suggesting variability rather than improvement. Similarly, tillering and yield traits, including NTT, NPT, PL, NTG, and WPP, showed increased spread in mutant populations versus the wild type. At times, 200 Gy showed higher values, whereas 400 Gy showed more extreme values (Figure S4). These patterns indicate gamma irradiation generates variability, with moderate doses enhancing trait performance and higher doses increasing heterogeneity.

Figure 3 illustrates the Spearman correlation analysis, showing relationships among traits in rice mutants under drought–heat stress, with WPC as the key variable. WPC was positively correlated with vegetative traits PH (0.56), NTT (0.32), and FLW (0.32), and more strongly with yield components WPP (0.63), W1000G (0.50), NTG (0.60), and NFG (0.60). WPC was also positively correlated with PL (0.45), while correlations with physiological parameters LDS, LRS, and HToI were low to moderate. Yield components generally showed strong positive correlations: NFG with WPP (0.95) and NTG (0.82), NTG with WPP (0.78) and W1000G (0.55), and W1000G with WPP (0.70). Among vegetative traits, PH correlated with NTG (0.47), WPP (0.38), and NFG (0.34), whereas NPT correlated with PH (0.38) and strongly with NP (1.00). Conversely, PEG showed negative correlations with PFG (−1.00), WPP (−0.44), and NTG (−0.29); PFG was also negatively correlated with other yield variables. LDS, LRS, HTol, and SF were strongly positively correlated (>0.88).

3.3. Identification of Yield-Contributing Traits Through Path Coefficient Analysis and PCA in M2 Rice Mutant Population

The results (

Table 3. Path coefficient analysis of direct, indirect, and residual effects of agronomic traits on weight per clump in the rice mutant M2 population under combined drought-heat stress.

Character	Direct Effect	Individual Effect								Indirect Effect	Residual
		PH	NTT	FLW	WPP	PL	W1000G	NTG	NFG
PH (cm)	0.27	0.27	0.04	0.02	0.03	0.18	0.02	0.04	−0.05	0.56	0.42
NTT (tiller)	0.22	0.05	0.22	0.01	0.00	0.04	0.01	0.02	−0.02	0.32
FLW (cm)	0.08	0.08	0.02	0.08	0.01	0.13	0.01	0.02	−0.03	0.32
WPP (g)	0.07	0.12	−0.01	0.01	0.07	0.24	0.03	0.06	−0.06	0.45
PL (cm)	0.47	0.10	0.02	0.02	0.04	0.47	0.04	0.07	−0.13	0.63
W1000G (g)	0.06	0.08	0.02	0.02	0.03	0.33	0.06	0.05	−0.09	0.50
NTG (grain)	0.09	0.13	0.04	0.02	0.04	0.36	0.03	0.09	−0.11	0.60
NFG (grain)	−0.14	0.09	0.03	0.02	0.03	0.45	0.04	0.08	−0.14	0.60

Note: PH: plant height; NTT: number of total tillers; FLW: flag leaf width; WPP: weight per panicle; PL: panicle length; W1000G: weight 1000 grain; NTG: number of total grain; NFG: number of filled grain.

) indicate that panicle length (PL) exerted the highest positive direct effect (0.47) on WPC, highlighting its primary contribution to yield formation. Plant height (PH) and the number of total tillers (NTT) also showed positive direct effects of moderate magnitude (0.27 and 0.22, respectively), with PH further contributing indirectly through weight per panicle (WPP) (0.18), resulting in a relatively strong total association (0.56). Flag leaf width (FLW) exhibited a comparatively low direct effect (0.08), and its total contribution (0.32) was primarily supported by indirect effects, particularly via WPP (0.13).

In contrast, several traits displayed relatively small direct effects but substantial total effects because of strong indirect contributions. Weight per panicle (WPP) had a low direct effect (0.07); however, its overall association (0.45) was enhanced by indirect pathways through other yield components, including the total number of grains (NTG) and filled grains (NFG). Similarly, 1000-grain weight (W1000G) showed a low direct effect (0.06), but a notable indirect contribution via WPP (0.33) resulted in a higher total effect (0.50). NTG also followed this pattern, with a low direct effect (0.09) but a substantial total association (0.60), largely mediated through WPP (0.36).

Notably, NFG exhibited a negative direct effect (−0.14) on WPC; however, this was offset by relatively strong positive indirect effects, particularly via WPP (0.45), leading to a positive overall association (0.60). These findings indicate that while certain traits, such as PL and PH, contribute directly to WPC, others, including W1000G, NTG, and NFG, primarily influence WPC indirectly through their interactions with key intermediary traits, especially WPP.

As shown in

Table 4. Principal component loadings and explained variance of selected characters influencing weight per clump in rice M2 mutants under combined drought-heat stress.

Variable	PC1	PC2	PC3	PC4	PC5
PH	0.67	0.36	0.31	0.33	−0.31
NTT	0.23	0.75	−0.55	0.09	0.26
FLW	0.41	0.4	0.66	−0.38	0.27
WPP	0.92	−0.19	−0.08	−0.23	−0.05
PL	0.55	−0.29	0.18	0.65	0.37
W1000G	0.74	−0.21	−0.16	−0.22	0.29
NTG	0.86	−0.11	−0.1	0.11	−0.09
NFG	0.88	−0.24	−0.16	−0.25	−0.08
WPC	0.83	0.25	−0.04	0.06	−0.23
Eigenvalue	4.56	1.14	0.93	0.86	0.52
Proportion	50.68	12.7	10.38	9.58	5.8
Cumulative	50.68	63.37	73.75	83.33	89.13

Note: Orange-colored and bolded values indicate the highest loading for each PC. PH: plant height; NTT: number of total tillers; FLW: flag leaf width; WPP: weight per panicle; PL: panicle length; W1000G: weight of 1000 grains; NTG: number of total grains; NFG: number of filled grains.

, the principal component analysis (PCA) results reveal that five principal components account for the majority of the variation in mutant rice characteristics under drought–heat stress conditions. The first principal component (PC1) has an eigenvalue of 4.56 and contributes the most to the variation (50.68%), followed by PC2 (1.14; 12.70%), PC3 (0.93; 10.38%), PC4 (0.86; 9.58%), and PC5 (0.52; 5.80%), culminating in a total cumulative variation explained of 89.13%. Within PC1, most characteristics exhibit high positive loadings, notably WPP (0.92), NFG (0.88), NTG (0.86), WPC (0.83), and W1000G (0.74), followed by PH (0.67) and PL (0.55), whereas FLW (0.41) and NTT (0.23) contribute less significantly. In PC2, NTT provides the largest contribution (0.75), followed by FLW (0.40), PH (0.36), and WPC (0.25), while other characteristics display negative loadings, such as PL (−0.29), NFG (−0.24), and WPP (−0.19). In PC3, FLW has the highest loading (0.66), followed by PH (0.31) and PL (0.18), whereas NTT exhibits a substantial negative loading (−0.55). In PC4, PL contributes the most (0.65), followed by PH (0.33), while other characteristics generally have low to negative loadings. In PC5, all variables demonstrate relatively minor contributions, with the highest values for PL (0.37) and W1000G (0.29), while PH (−0.31) and WPC (−0.23) indicate a negative direction. Overall, PC1 predominates in data variation with a contribution exceeding 50%, with most characteristics, particularly yield components and WPC, exhibiting high positive loadings, whereas the subsequent components offer additional contributions with more varied directions and magnitudes of loadings.

3.4. Multi-Trait Index Selection of Putative Multi-Tolerant M2 Rice Mutant Genotypes to Drought-Heat Stress

Multi-trait index analysis was further explored by visualizing the distribution of mutant genotypes across TScore, WASB-PC1, and WASB-PC2 dimensions. As shown in Figure S5 and Figure 4, and

Table 5. Rice Mutant M2 Genotypes TScore, PCA, and WASB Values under Drought-Heat Stress.

Genotype	Genotype Code	TScore	PC1	PC2	WASB PC1	WASB PC2
M2 0 (10) 04	G1	0.00	0.42	−1.69	0.33	0.34
M2 0 (10) 10	G2	0.46	1.47	−0.23	1.17	0.05
M2 0 (10) 11	G3	0.00	−0.36	1.02	0.28	0.21
M2 0 (10) 18	G4	0.13	−2.13	−0.05	1.70	0.01
M2 0 (10) 22	G5	0.38	−3.22	0.12	2.57	0.03
M2 0 (10) 24	G6	0.75	−1.95	0.49	1.56	0.10
M2 200 (15) 09	G7	0.75	1.57	0.06	1.25	0.01
M2 200 (15) 10	G8	0.38	1.00	1.46	0.80	0.29
M2 200 (15) 12	G9	0.75	1.76	2.13	1.41	0.43
M2 200 (15) 17	G10	0.46	2.04	−1.34	1.63	0.27
M2 200 (15) 18	G11	0.21	4.37	0.29	3.49	0.06
M2 200 (15) 19	G12	0.21	1.28	−1.74	1.02	0.35
M2 200 (15) 20	G13	0.38	5.63	1.68	4.50	0.34
M2 200 (15) 22	G14	0.38	0.01	−0.86	0.01	0.17
M2 200 (15) 23	G15	0.38	0.38	0.90	0.30	0.18
M2 200 (15) 26	G16	0.75	−2.64	0.05	2.11	0.01
M2 200 (15) 27	G17	0.75	−1.85	−1.12	1.48	0.22
M2 200 (15) 28	G18	0.46	2.00	−1.02	1.60	0.20
M2 200 (15) 29	G19	0.75	3.47	0.83	2.77	0.17
M2 200 (15) 30	G20	0.46	2.45	0.60	1.96	0.12
M2 200 (15) 31	G21	0.46	1.56	−0.62	1.25	0.12
M2 200 (15) 33	G22	0.38	0.09	3.07	0.07	0.62
M2 200 (15) 36	G23	0.00	0.31	1.45	0.25	0.29
M2 200 (15) 39	G24	0.38	1.31	−1.00	1.05	0.20
M2 200 (6) 04	G25	0.00	1.04	−1.87	0.83	0.38
M2 200 (6) 13	G26	0.00	0.80	−0.97	0.64	0.20
M2 200 (6) 22	G27	0.08	1.72	−0.88	1.37	0.18
M2 200 (6) 23	G28	0.00	1.72	0.01	1.38	0.00
M2 200 (6) 24	G29	0.46	0.86	0.07	0.69	0.01
M2 200 (6) 25	G30	0.46	0.02	0.63	0.01	0.13
M2 200 (6) 30	G31	0.46	−1.23	−0.55	0.99	0.11
M2 200 (6) 32	G32	0.08	−0.77	0.05	0.62	0.01
M2 200 (6) 38	G33	0.75	−2.78	0.93	2.22	0.19
M2 200 (6) 39	G34	0.75	0.05	−0.92	0.04	0.18
M2 200 (6) 40	G35	0.75	−3.19	1.19	2.55	0.24
M2 200 (6) 41	G36	0.08	−4.46	−0.72	3.56	0.15
M2 200 (6) 43	G37	0.08	−2.34	0.09	1.87	0.02
M2 200 (6) 44	G38	0.08	0.46	−0.15	0.37	0.03
M2 200 (6) 45	G39	0.00	1.34	−0.62	1.07	0.12
M2 200 (6) 47	G40	0.08	−0.80	−1.39	0.64	0.28
M2 200 (6) 49	G41	0.46	−1.64	−1.37	1.31	0.27
M2 400 (3) 01	G42	0.13	0.83	−0.91	0.66	0.18
M2 400 (3) 06	G43	0.21	−2.16	0.02	1.73	0.00
M2 400 (3) 15	G44	0.83	−2.47	−0.41	1.98	0.08
M2 400 (3) 23	G45	0.13	1.26	0.11	1.01	0.02
M2 400 (3) 25	G46	0.08	2.10	−0.36	1.68	0.07
M2 400 (3) 30	G47	0.00	2.24	0.82	1.79	0.16
M2 400 (5) 08	G48	0.08	−3.08	1.78	2.46	0.36
M2 400 (5) 12	G49	0.13	−4.75	0.03	3.80	0.01
M2 400 (5) 13	G50	0.58	−2.20	1.98	1.76	0.40
M2 400 (5) 14	G51	0.38	−0.14	0.25	0.11	0.05
M2 400 (5) 16	G52	0.29	0.10	−0.17	0.08	0.03
M2 400 (5) 29	G53	0.00	−1.47	−1.16	1.18	0.23

Note = Genotype is formatted as M2 X (Y) Z, where X denotes the gamma irradiation intensity, Y specifies the screening population group, and Z represents the individual genotype number within each group.

, mutant rice genotypes exhibited substantial variation in TScore and WASB components, resulting in distinct distribution patterns among tolerance categories under combined drought-heat stress. In this research, lower TScore values indicate greater drought-heat tolerance. Therefore, a clear separation of genotypes according to tolerance classification was observed in both Figure 4, where putative tolerant mutant genotypes (G1, G3, G4, G23, G25, G26, G28, G39, G47, G49, and G53) were predominantly located on the left side of the plots and susceptible genotypes were concentrated on the right side (G6, G7, G9, G16, G17, G19, G33, G34, G35 and G44), while moderately tolerant genotypes occupied intermediate positions (G2, G10, G13, G18, G20, G21, G22, G29, G30, G41, and G50), forming a transitional region between tolerant and susceptible groups.

In TScore-WASB-PC 1 (Figure 5A), genotype separation occurred along the TScore axis, while WASB-PC 1 differentiated genotypes within tolerance categories. Putative tolerant genotypes showed a broad distribution across the WASB-PC 1 axis, indicating heterogeneous multi-trait responses despite comparable tolerance classification. In contrast, susceptible genotypes were more compactly distributed with moderate WASB-PC 1 values. A few genotypes deviated from the main distribution, particularly G13 (WASB-PC 1 = 4.50), G49 (3.80), and G36 (3.56), with high WASB-PC 1 values. A similar pattern was observed in TScore-WASB-PC 2 (Figure 5B), where most genotypes clustered within a narrow WASB-PC 2 range, but G22 (WASB-PC 2 + 0.62), G9 (0.43), and G50 (0.40) were separated. Distinct patterns were also observed among the Gy irradiation dose groups, particularly at 200 Gy, which generated substantial phenotypic diversity in multi-trait responses, as visualized in the TScore-WASB-PC value. A similar trend with less variability was observed in the 400 Gy and 0 Gy (wild type) groups. These distribution patterns indicate that gamma irradiation induced diverse multi-trait responses rather than a uniform response. Collectively, these findings highlight the utility of the multi-trait index approach as one of the approaches for identifying putative drought-heat-tolerant rice mutant genotypes.

4. Discussion

4.1. M2 Rice Mutant Seedling-Screening Response Under Combined Drought-Heat Stress and Their Implications for Spikelet Fertility

Controlled stress chambers were used to generate a reproducible and uniform stress environment for screening rice mutant seedlings under drought-heat stress. Despite the relatively short multi-stress screening cycle, the imposed conditions were sufficient to induce physiologically relevant combined drought-heat stress. As shown in Table S1, the microclimatic conditions inside the stress chamber were characterized by elevated vapor pressure deficit (VPD) and reduced relative humidity. These conditions increased the vapor pressure gradient between the leaf and the surrounding air. Combined with water withholding, this resulted in a rapid decline in plant water status [49,50], effectively simulating acute drought-heat stress within a limited screening period in the seedling stage in rice.

Seedlings represent one of the most vulnerable developmental stages in rice, during which exposure to combined drought and heat stress can substantially impair subsequent plant growth. In the present study, Figure 6 reveals distinct responses among mutant genotypes. These contrasting responses suggest differences in the capacity of each genotype to withstand stress during early vegetative growth. Although the underlying physiological mechanisms were not directly investigated in this study, previous studies have suggested that maintenance of leaf greenness and reduced leaf rolling are commonly associated with improved leaf water status, delayed chlorophyll degradation, sustained photosynthetic activity, and enhanced osmotic adjustment under drought and heat stress [51,52,53]. Therefore, it is hypothesized that the superior performance of the putatively tolerant mutant may be attributed to more efficient regulation of these protective processes. To validate this hypothesis, further physiological and molecular analyses are required in future studies.

Apart from its immediate effects on seedling survival, damage incurred during the seedling phase may also exert carry-over effects on subsequent plant development, particularly during the reproductive stage. In the present study, this potential carry-over effect was reflected by the contrasting spikelet fertility among the M2 rice mutant population and the wild type (Figure 7). This phenomenon has been well documented in previous studies, where early exposure to combined drought and heat stress in rice was reported to disrupt normal vegetative growth by reducing biomass accumulation [6,54], limiting tiller development [55,56], impairing photosynthetic capacity [57,58], and restricting carbohydrate production [59]. These hypothesized effects may subsequently reduce the assimilate supply required for panicle development and grain filling during the reproductive stage, thus leading to increased spikelet sterility and yield loss.

The wild-type line retained moderate spikelet fertility following stress despite exhibiting seedling-stage stress injury, compared with most M2 rice mutants. This suggests that the wild type already possesses a capacity for recovery following combined drought-heat stress. The superior reproductive performance in the mutants may represent not the acquisition of a new stress response, but enhancement of the adaptive capacity in the wild type through mutation-induced genetic variation. Gamma irradiation induces random genetic variation throughout the genome, which may modify the function or regulation of genes involved in abiotic stress responses in the wild type. Consequently, induced mutagenesis increases available genetic variations within irradiated populations, enabling selection of individuals with enhanced adaptation to drought and heat stress.

These variations presented in the study suggest that stress responses expressed during the seedling stage may influence subsequent reproductive performance and yield-related traits. However, this relationship was not quantitatively evaluated in the present study. Therefore, future studies should investigate the direct mathematical relationship between seedling-stage tolerance and reproductive-stage performance.

4.2. Hypothesized Gene Mechanism Underlying Seedling Tolerance and Reproductive Resilience Under Combined Drought-Heat Stress in M2 Rice Mutant

Abiotic stress tolerance in rice is a trait regulated by interconnected physiological and molecular pathways coordinating plant adaptation to drought and heat stress throughout the life cycle. Rather than functioning independently at specific stages, many stress-responsive genes activated during early vegetative growth may contribute to maintaining growth, recovery, and reproductive development under prolonged or recurrent stress [60,61]. Although these mechanisms were not investigated in the present study, the contrasting responses between tolerant and susceptible mutants may reflect differential regulation of stress-responsive pathways reported in rice (

Table 6. Physiological and molecular mechanisms underlying rice response to combined drought and heat stress reported in recent studies.

Plant Response	Hypothesized Mechanism	Candidate Genes, Pathways	Supporting Reference
Maintained leaf greenness	Delayed chlorophyll degradation and sustained photosynthesis	OsSGR, NYC1, chlorophyll metabolism genes	[62,63,64,65]
Reduced leaf rolling	Improved leaf water status through osmotic adjustment and ABA-mediated stomatal regulation	OsDREB2A, OsbZIP23, SNAC1, ABA signaling	[66,67,68,69]
Reduced leaf desiccation	Enhanced ROS scavenging and membrane stability	OsAPX2, CAT, SOD, OsHSPs	[70,71,72]
Seedling survival	Greater cellular protection and stress-responsive signaling	OsHsfA2e, HSP70, HSP101, LEA proteins	[73,74,75,76]
Higher spikelet fertility	Maintenance of pollen viability, assimilate transport and reproductive development	OsHTAS, TT1, sucrose transport genes	[77,78,79]

). The candidate genes presented in the table represent regulatory pathways for water status maintenance, protection against oxidative and heat-induced damage, chlorophyll preservation, and reproductive resilience under abiotic stress. Although these mechanisms remain hypothetical in the present study, they provide a plausible physiological-molecular framework linking seedling tolerance with variation in yield-related traits. Gene expression analysis of these candidate genes would be valuable to elucidate the molecular basis of tolerance to drought-heat stress in rice.

4.3. Differential Sensitivity of Yield and Vegetative Traits to Gamma Irradiation in Rice M2 Mutants Under Drought–Heat Stress

The findings indicate that gamma ray irradiation enhances agronomic trait diversity within the line population, evidenced by broader trait distribution in the 200 Gy and 400 Gy lines than in the 0 (wild type). The expanded distribution suggests individuals with extreme superior and inferior performance due to induced random mutations. This pattern was most evident in yield component traits, including WPC, WPP, NTG, NFG, and W1000G, which showed greater variation in the mutant population than in the wild type. However, the increase was less pronounced in vegetative traits. This trend underscores the sensitivity of yield components to mutation-induced genetic changes. A similar pattern was reported by Anshori et al. (2025) [27], indicating pronounced diversity increases in yield components compared to vegetative traits.

The characteristics of yield components are predominantly governed by multiple genes, a condition known as polygenic control [80,81]. Random mutations induced by gamma rays have the potential to modify one or several genes within these regulatory networks [82], thereby generating increased variation in yield traits. Furthermore, yield traits represent the culmination of various physiological processes, meaning that minor genetic alterations can be magnified into significant phenotypic differences. In contrast, vegetative traits exhibit greater stability against mutational effects. Although they are generally under polygenic control, they are often predominantly influenced by major genes, resulting in genetic changes that do not necessarily lead to extreme variation.

The median distribution showed patterns among M2 0 (wild-type), M2 200, and M2 400 lines. In M2 0 lines, the distribution was narrow and centered on the median, indicating low variation in untreated populations. In contrast, M2 200 and M2 400 lines showed broader distributions than the wild type. However, in M2 400 lines, character value distribution expansion did not coincide with a median shift. By contrast, M2 200 lines showed broad variation and a median shift toward higher levels. This suggests that high dosage may induce extreme variation but does not significantly increase the proportion of individuals with optimal performance [27]. Therefore, moderate-dose gamma-ray induction was more effective in generating lines with enhanced diversity and improved agronomic performance.

4.4. Adaptability of M2 Generation Rice Mutant Lines to Various Doses of Gamma-Ray Irradiation Under Drought and High-Temperature Stress Based on Tolerance Scoring

The tolerance scoring method revealed variability among mutant lines generated through gamma-ray irradiation. The 200 line showed adaptability to drought and elevated temperatures over the wild-type and 400 lines. Evidenced by the concentration of individuals in the low-to-medium score range, the 200 group mostly comprised individuals with mild-to-moderate damage. This indicates that irradiation at 200 Gy induced beneficial mutations in the plant’s defense mechanisms against drought-heat stress. Previous studies suggest that tolerant rice genotypes exhibit stomatal regulation, cell turgor maintenance, and delayed leaf rolling under water deficit conditions [6,58,83]. These effects are reflected in tolerant-to-moderate LDS and LRS values, facilitating photosynthesis under drought-stress conditions.

The HTol and SF scoring values suggest that the M2 200 line exhibits potential as a genotype adaptable to high-temperature stress conditions. This adaptability is hypothesized to be associated with the line’s capacity to maintain cell membrane stability [69] and the integrity of the photosynthetic apparatus [84]. An HTol score within the tolerant-to-moderate category indicates minimal thermal damage to structural proteins, whereas an elevated spikelet fertility signifies successful reproductive processes that remain optimal despite exposure to extreme drought-heat stress [85]. According to [86], rice plants exposed to high-temperature stress accumulate compatible osmolytes, such as proline, soluble sugars, and protective proteins. Studies from [87] also show that an increase in the accumulation of these compounds occurs under drought stress treatment rather than well-watered rice. These compounds serve as osmoregulatory agents to maintain cell turgor and provide protection for enzymes and cell membranes against thermal denaturation.

In contrast, the M2 400 line exhibited a dispersed distribution pattern of scores, characterized by a low frequency of individuals within the tolerant category. This pattern indicates limited adaptability to multiple stressors. It was hypothesized that a high irradiation dose induces excessive genetic variation mutations, thereby disrupting the biochemical pathways responsible for neutralizing reactive oxygen species (ROS) [84]. Consequently, plants in the M2 400 line were prone to symptoms of chlorosis and leaf necrosis; previous studies have shown that these symptoms were attributed to chlorophyll degradation and lipid peroxidation in the cell membrane [52]. These observations were also aligned with the findings of Yanting et al. (2024) [88], who reported that gamma ray irradiation doses exceeding the optimal threshold result in an increase in ROS that cannot be neutralized by the antioxidant enzyme system, thus may lead to inhibited growth and tissue death in leaves due to biochemical imbalances in the plant’s leaf tissues [69,89].

A comparison of the adaptability responses between 400 Gy and the moderate dose of 200 Gy illustrates the hormesis phenomenon, in which a moderate dose has the potential to enhance plant adaptability to abiotic stress, whereas a high dose proves toxic to plant physiological processes, whereas higher doses produce inhibitory or deleterious effects. Under moderate irradiation, transient oxidative signals induced by gamma rays are hypothesized to function as signaling molecules that activate antioxidant defense systems, osmotic adjustment, DNA repair mechanisms, and stress-responsive metabolic pathways, thereby increasing plant resilience to subsequent abiotic stress. Similar biphasic dose–response patterns have been reported in rice [90,91] and other cereal crops [92,93], where intermediate gamma irradiation doses promoted stress tolerance and optimal growth performance, whereas higher doses exceeding the optimal threshold resulted in excessive ROS production that overwhelmed the antioxidant capacity, thereby promoting cellular injury and progressive reductions in physiological performance. These findings support the interpretation that the superior adaptability of the M2 200 line reflects a beneficial hormetic response, whereas the reduced performance of the M2 400 line represents the detrimental consequences of irradiation beyond the hormetic optimum.

In the M2 0 line (wild type), the score distribution appeared to be homogeneous across each assessment. This suggests a limited variation in the natural response of wild-type individuals when subjected to extreme environmental stress, as the entire population exhibited similar symptoms of damage. This observation serves as an indicator of the effective application of gamma-ray irradiation in inducing genetic variation within mutant lines, which may be considered potential candidates for selection as lines with enhanced adaptability to drought and high-temperature stress.

4.5. Multivariate Approach in Assessing Diversity and Mutation Effectiveness in M2 Generation Rice Populations Resulting from Gamma Irradiation Under Drought-High Stress

A multivariate approach integrating correlation, path analysis, PCA, and WASB elucidated relationships among rice mutant traits. Spearman’s correlation and path coefficient analyses identified NTG, NFG, WPP, and W1000G as primary yield determinants, with WPP mediating NTG and WPC. This supports the view that cereal crop yield is governed by interactions among traits [94]. Vegetative characters PH and NTT contribute indirectly by enhancing source capacity and supporting photosynthesis production and allocation. Yield stability under stress is determined by source–sink relationship efficiency, reflected in causal interactions among traits.

Path coefficient analysis highlighted PL as an indicator of sink capacity, exerting the strongest direct impact on WPC and contributing indirectly via WPP. Similar interpretations were reported by Parida et al. (2022) [95] and Anshori et al. (2025) [27], where panicle-related traits emerge as key determinants of sink strength and yield. Although NTG, NFG, and W1000G have minor effects, their overall impact is considerable through WPP, a central node in relationships. This structure reflects the hierarchical nature of yield formation, in which upstream traits indirectly influence yield through components. Vegetative traits, PH and NTT, also act through indirect pathways associated with photosynthate distribution. This structure is consistent with PCA findings distinguishing productivity (PC1) and vegetative (PC2) dimensions, supporting plant potential differentiation.

The combination of tolerance scores, PCA, and the WASB selection index enables comprehensive genotype identification by considering both tolerance and productivity together. Using stability indices such as WASB with multivariate techniques was reported previously by [96,97] to improve selection in variable, especially stress-prone, environments. This approach avoids bias from focusing on one trait group, ensuring genotypes have balanced characteristics. The relationship between genotypes’ PCA positions and selection index values also shows that high-performing genotypes generally occupy regions of strong productivity supported by adequate vegetative vigor, providing the basis for evaluating mutation treatment effectiveness.

The differences in genotype responses to gamma irradiation doses indicate that diversity is influenced by treatment intensity. A 200 Gy dose generates broader, more targeted variation in yield traits, whereas higher doses produce variation less efficient for agronomic performance. This pattern is consistent with mutation breeding theory, in which moderate irradiation doses are optimal for inducing useful variability without excessive deleterious effects. Genotypes without irradiation still show limitations in exploring diversity, reinforcing mutation as a source of variability. Studies by [27,98,99] on induced mutagenesis in rice showed that gamma irradiation enhances genetic variability, expanding selection for desirable traits. Thus, selection success depends not only on the extent of variation but also on its relevance to target traits.

Although this approach provides a framework for selecting putative multi-tolerant mutants to drought-heat stress, high residual values in path analysis indicate factors remain unaccounted for in the present study. High residuals suggest missing explanatory variables [100], particularly unmeasured physiological or molecular traits. Parameters such as photosynthetic rate, chlorophyll content, and molecular aspects likely influence seed filling efficiency and yield stability. Evidence from Chaturvedi et al. (2017) [101], Yang et al. (2024) [102], and Sharma et al. (2025) [103] has shown that physiological traits, including photosynthetic efficiency and assimilate partitioning, are determinants of stress tolerance in rice. In this study, drought and heat stress were imposed simultaneously, which may obscure each stressor’s contribution and interaction on trait expression and yield formation. The study lacked a separate non-stress control group; future studies should include non-stress, separate, and combined stress treatments, especially using stabilized lines, to disentangle individual and synergistic effects of drought and heat stresses.

Thus, future research that combines agronomic approaches with physiological and molecular dimensions and incorporates a non-stress control group, individual and combined drought and heat treatments is a logical step to improve selection accuracy and pave the way for the development of more precise screening methods to identify adaptive mutant rice lines to multiple stresses.

5. Conclusions

The findings demonstrated that the controlled stress chamber effectively simulated combined drought-heat stress during the rice seedling stage, as confirmed by the progressive decline in the FTSW, complete mortality of the susceptible check (IR29), and clear phenotypic differentiation among the evaluated mutant populations. Among the irradiation treatments, the 200 Gy populations exhibited higher seedling survival, more favorable tolerance score distribution, and a greater proportion of putative multi-tolerant mutants than the 400 Gy population, indicating that the moderate irradiation dose was more effective for inducing beneficial phenotypic variation. Multivariate analysis through correlation, path coefficient, and principal component analysis identified PL, WPC, NTG, NFG and W1000G as the principal traits associated with post-stress growth performance, with PL exerting the strongest direct effect on WPC, acting as a key mediator of indirect trait contributions. Furthermore, the integration of TScore, PCA, and the WASB provided a systematic multi-trait selection framework that was able to distinguish putative tolerant genotypes by simultaneously considering stress tolerance potential and multi-trait performance. These findings demonstrate that the proposed screening strategy is an effective approach for the early identification of putative multi-tolerant rice mutants and provides promising candidate genetic materials for subsequent validation in future analysis through physiological and molecular analyses.