A total of 59 peanut (Arachis hypogaea L.) genotypes were evaluated in this study. These comprised 54 advanced breeding lines developed by the Department of Agronomy, Faculty of Agriculture, Kasetsart University, Thailand, and five standard check varieties recommended by national agricultural agencies. The check varieties were Tainan 9, Khon Kaen 5, Khon Kaen 9, KU Koh Kae 40, and KU ARDA 1, each representing widely cultivated cultivars with known performance for yield and adaptability under Thai agro-climatic conditions.

Yield trials were conducted during the 2020/21 dry season (December 2020-April 2021) at two research locations in Thailand: Chiang Mai Field Crops Research Center, San Sai District, Chiang Mai Province (18.9047˚N, 99.0045˚E) and Lopburi Research Station, Khok Charoen District, Lopburi Province (14.7995˚N, 100.6534˚E). The experiments were arranged in a randomized complete block design (RCBD) with three replications at each location. Each experimental plot consisted of two rows, each measuring 5 m in length, with 0.80 m row spacing and 20 cm plant spacing within rows. Two seeds were sown per hill to ensure proper establishment. Standard agronomic practices for peanut cultivation were followed uniformly across both locations, including land preparation, basal fertilization, pest and weed management, and supplemental irrigation where necessary to avoid moisture stress. The following agronomic and yield-related traits were recorded: pod yield (kg ha⁻¹), shelling percentage (%), harvest index, 100-seed weight (g), number of pods per plant, pod weight per plant (g), number of seeds per plant, seed weight per plant (g), and number of seeds per pod.

Data from the two locations were analyzed using a mixed model as follows:

where y { i j k } is the observed value of the i t h genotype in the k^thblock at the j t h environment; μ is the grand mean; a i is the random effect of genotype i; τ j is the fixed effect of environment j; ( a τ ) { i j } is the random genotype × environment interaction; b { j k } is the block effect within environment; and ε { i j k } is the residual error.

Variance components and broad-sense heritability were estimated. Best linear unbiased predictions (BLUPs) were generated using mixed linear models, with genotypes and environments considered as random effects to obtain precise estimates of genetic values across locations.

Multi-trait selection was carried out using the Factor Analysis and Ideotype-Design BLUP (FAI-BLUP) index, which integrates factor analysis with the BLUP methodology to facilitate simultaneous selection for multiple traits without the need to assign arbitrary economic weights. The FAI-BLUP index was computed according to the procedure described by Rocha et al. [21]. Genotypic values obtained from BLUP were first standardized, and factor analysis was employed to reduce data dimensionality and account for correlations among traits. An ideotype representing the optimal trait combination was then constructed based on the most desirable expression for each trait. The Euclidean distance of each genotype from the ideotype was calculated using the formula:

where P i j is the probability of genotype i matching the desired value j, and d i j is the distance between the observed and desired values. And then, genotypes were ranked according to their proximity to the ideotype, with smaller distances indicating superior performance. A selection intensity of 20% was adopted to identify superior genotypes. All statistical analyses were conducted using R software [27]. The metan package was employed for estimating variance components, BLUPs, and computing the FAI-BLUP index.

Analysis of variance using the mixed model method revealed significant (p < 0.05 or p < 0.01) differences among the 59 peanut genotypes for pod yield (PY), number of pods per plant (PdPt), seed weight per plant (SwPt), and number of seeds per plant (SdPt), indicating substantial genetic variability within the breeding population. Significant genotype × environment (G × E) interactions were detected for pod yield, 100-seed weight (SW100), and seed traits, confirming differential genotype performance across environments. Broad-sense heritability (h²) estimates ranged from low for the number of seeds per pod (0.06) to moderate for 100-seed weight (0.46) and shelling percentage (0.28). Traits with moderate heritability, particularly 100-seed weight, demonstrate a stable genetic component, suggesting that these traits are likely to respond reliably to selection across environments (Table 1).

Table 1. Estimate of variance component and genetic parameters for pod yield (PY), shelling percentage (SH), 100-seed weight (SW100), harvest index (HI), number of pod per plant (PdPt), pod weight per plant (PwPt), seed weight per plant (SwPt), number of seed per plant (SdPt), number of seed per pod (SdPd) for 54 peanut lines and 5 standard varieties evaluated across 2 locations in the dry crop season.

Note:^* and ^**: Significant difference at p < 0.05 and p < 0.01, respectively; ^ns: Non-significance; h 2 : Broad-sense heritability; σ g 2 : Genetic variance; σ g e 2 : G × E interaction variance; σ r e s 2 : Residual error variance; CV: Coefficient of variation.

Factor analysis with varimax rotation classified the nine evaluated traits into three principal factors, cumulatively accounting for most of the total genetic variance. Factor 1, which comprised pod yield, number of pods per plant, pod weight per plant, seed weight per plant, and number of seeds per plant, explained the largest proportion of variation, demonstrating that these traits were the primary contributors to yield. Factor 2 was dominated by harvest index, while Factor 3 was associated with seed quality traits such as shelling percentage and 100-seed weight. High communality values (>0.69) for all traits confirmed that the extracted factors reliably captured the underlying structure of genetic variation.

Application of the FAI-BLUP index under a 20% selection intensity produced positive selection differentials for all evaluated traits (Table 2). The highest percentage gains were recorded for seed weight per plant (+15.39%), number of seeds per plant (+12.28%), and harvest index (+6.49%), followed by pod yield (+5.29%). These results demonstrate that the multi-trait selection index effectively captured genetic gains in both productivity-related and physiological traits. This confirms the efficiency of the FAI-BLUP approach in identifying genotypes that simultaneously excel in multiple desirable characteristics.

Table 2. Factorial loading after varimax rotation and communalities, overall mean (Xo), selected genotype mean (Xs), selection differential (SD) and percentage of SD (%SD) for the traits based on FAI-BLUP index.

Based on the FAI-BLUP ranking index, twelve genotypes were selected as superior performers (Table 3). Among them, line 12BS018 recorded the highest BLUP-estimated pod yield of 2,278 kg·ha⁻¹, which exceeded the overall population mean (1,868 kg·ha⁻¹) by 410 kg·ha⁻¹. Similarly, lines 12BS022 (2,247 kg·ha⁻¹), 13W089 (2,035 kg·ha⁻¹), and 11231-3 (1,938 kg·ha⁻¹) also demonstrated outstanding yield potential along with favorable values for shelling percentage, hundred-seed weight, and seed weight per plant. Importantly, several selected breeding lines outperformed the highest-yielding check variety, Khonkaen 9, indicating their strong potential for advancement in breeding pipelines and possible varietal release.

Table 3. BLUPs mean for pod yield (PY), shelling percentage (SH), 100-seed weight (SW100), harvest index (HI), number of pods per plant (PdPt), pod weight per plant (PwPt), seed weight per plant (SwPt), number of seeds per plant (SdPt), number of seeds per pod (SdPd) of selected genotypes at selection intensity of 20 % and check varieties.

Narrative interpretation of BLUP values (Figure 1) revealed a wide distribution of pod yield across genotypes, reflecting strong genetic divergence and the presence of high-yielding genotypes within the breeding population. Several breeding lines surpassed the overall mean as well as the majority of the check varieties, confirming the presence of superior genetic potential for yield improvement. The dispersion pattern further highlighted the importance of BLUP in accurately estimating genotypic performance under multi-environment conditions.

Figure 1. Best linear unbiased prediction for pod yield (PY) of 59 peanut genotypes.

The ranking of all genotypes based on the FAI-BLUP index is presented in Figure 2. The index clearly discriminated the superior group of genotypes from the remainder of the population. Twelve genotypes were identified under a 20% selection intensity, each showing a higher probability of similarity to the ideotype compared with checks and the population mean. Among these, lines 12BS018, 12BS022, 13W089, and 11231-3 consistently ranked in the top positions, reflecting their strong alignment with the predefined ideotype. In contrast, several check varieties, such as Tainan 9 and KU Koh Kae 40, clustered in the lower ranks, indicating limited conformity to the ideotype. These results highlight the usefulness of the FAI-BLUP index as an effective tool for ranking and selecting peanut genotypes with superior multi-trait performance.

Figure 2.Peanut genotypes ranking and selected genotypes using the FAI-BLUP index.

The significant genotypic variation observed among the 59 peanut genotypes across two environments confirms a broad genetic base suitable for selection and improvement. Traits such as pod yield, seed weight per plant, and pod number showed highly significant genotypic effects, supporting their use as primary selection targets. The detected genotype × environment (G × E) interaction indicates differential responses across environments. The findings highlight the need to select superior genotypes from diverse germplasm and confirm their stability through multi-environment evaluation [13][28][29]. Heritability estimates ranged from low to moderate, suggesting variable genetic control among traits. Moderate heritability for 100-seed weight and shelling percentage indicates that these traits can be improved via phenotypic selection, whereas low heritability for harvest index and seeds per pod highlights environmental sensitivity and the value of mixed-model BLUP approaches for accurate genetic value estimation [13][20][26].

Factor analysis efficiently grouped correlated traits into three principal components, simplifying interpretation and enhancing selection accuracy. Factor 1 encompassed yield and yield-related traits (pod yield, pod number, seed weight, and seeds per plant), confirming that these jointly define yield potential and can be improved simultaneously. Factor 2 was dominated by harvest index (assimilate partitioning), and Factor 3 captured seed quality traits (shelling percentage and 100-seed weight). This pattern aligns with prior applications of multi-trait indices and factor analysis in cereals and legumes, which reduce dimensionality and mitigate redundancy among correlated traits [24][25]. High communality values (>0.69) across traits further indicate that the retained components adequately captured underlying variability, validating factor analysis for multi-trait selection frameworks [24][25][28].

The FAI-BLUP index provided consistent, positive selection differentials across all traits under a 20% selection intensity, with the largest predicted gains for seed weight per plant (+15.39%), seeds per plant (+12.28%), and harvest index (+6.49%). By integrating factor analysis and ideotype design with BLUP, FAI-BLUP avoids subjective economic weights and improves selection precision [21][22]. Similar advantages have been reported in maize, wheat, and sorghum, where FAI-BLUP outperformed traditional indices by balancing gains across correlated traits and enhancing selection response [22][23][24]. The present results, therefore, confirm the suitability of FAI-BLUP for peanut breeding, especially when improvement targets multiple yield-determining traits under variable environments [21][23][24][28].

Integrating BLUP with FAI-BLUP identified twelve elite lines with consistent superiority across environments. In particular, 12BS018, 12BS022, 13W089, and 11231-3 ranked among the top positions and combined high pod yield with favorable values for seed and physiological traits. Line 12BS018 recorded the highest BLUP-estimated pod yield (2,278 kg·ha⁻¹), exceeding the population mean by ~22%. Several selected lines outperformed the best check (Khonkaen 9), indicating tangible breeding progress and strong potential for advancement to regional trials and potential release. Comparable successes using multi-trait indices and BLUP-based rankings have been reported in peanut and other crops, reinforcing the robustness of this integrated approach for discriminating high-potential genotypes [13][21][23][24][26].

The presence of high-yielding lines that also exhibited a favorable harvest index under dry-season testing suggests that these genotypes possess intrinsic physiological traits that enhance their adaptation to the dry-season environment. Breeding for drought and heat resilience is essential for sustaining peanut productivity in Southeast Asia’s variable rainfall regimes, and the present results align with reports emphasizing physiological efficiency and stress adaptation as core breeding targets [10][12]. Combining BLUP with FAI-BLUP strengthens prediction under G × E and provides a rigorous framework for simultaneous improvement in yield and physiological traits. Looking ahead, integrating genomic information and high-throughput phenotyping with FAI-BLUP could further accelerate breeding progress and improve the accuracy of selecting elite lines. Genomic selection enables early prediction of breeding values using genome-wide markers, reducing the need for multi-year field evaluations. This approach is especially advantageous for complex, low-to-moderate heritability traits influenced by climate stress. Together, these tools offer an efficient, modern framework for developing climate-resilient peanut cultivars [28][30].