The analysis of variation revealed highly significant differences among the accessions for all the characters studied, including days to 50% flowering, days to maturity, plant height, number of branches per plant, pods per plant, 100-seed weight (g), Cu (mg/kg), Zn (mg/kg), Fe (mg/kg), Mn (mg/kg), protein content (%) and single plant yield (g), indicating the existence of considerable genetic variation in the experimental material. Examination of the variance components revealed that the phenotypic coefficient of variation (PCV) was higher than the genotypic coefficient of variation (GCV) for all characters studied, indicating the role of environmental variance in the total variance (Table 1).
Variability and heritability
The genotypic coefficient of variation for various characters ranged from 5.154% to 46.681%, while the phenotypic coefficient of variation ranged from 5.228% to 46.681%. High genotypic and phenotypic coefficients of variation were observed for single plant yield (g), Fe (mg/kg), and number of pods per plant. Moderate GCV and PCV were observed for protein content (%), number of branches per plant, 100-seed weight (g), and Cu (mg/kg). Low GCV and PCV values were observed for days to 50% flowering, plant height, Zn (mg/kg), Mn (mg/kg) and days to maturity. Heritability estimates are crucial for determining the heritable portion due to genetic variation. High heritability estimates, indicating that the characters are least influenced by environmental factors and can be transmitted to subsequent generations, were observed for almost all the characters studied. Moderate heritability estimates were observed for the number of branches per plant. The success of genetic advance under selection depends on the magnitude of genetic variability in the base population and the heritability of the character under consideration. Genetic advance is usually expressed as a percentage of the mean. High genetic advance as a percentage of the mean was observed for Fe (mg/kg), single plant yield (g), and pods per plant. Moderate genetic advance as a percentage of the mean was observed for plant height, number of branches per plant, 100-seed weight (g), Cu (mg/kg), Zn (mg/kg) and Mn (mg/kg). Similar findings reported by
Edukondalu et al., (2023); Vanniarajan et al., (2023); Anuj et al., (2018); Pushpavalli et al., (2018); Mallesh et al., (2017); Ram et al., (2016); Shunyu et al., (2013); Sharma et al., (2012) and
Hamid et al., (2011).
According to
Johnson et al., (1955), heritability along with genetic advance is mostly useful and reliable in predicting the resultant effects of selection. Selection can only be achieved when high heritability is accompanied by high genetic advance (
Burton, 1952). In the present study, high estimates of heritability coupled with high genetic advance as a percentage of the mean were observed for single plant yield (g), pods per plant and Fe (mg/kg), indicating better scope for these traits for direct selection.
Correlation of attributing characters with grain yield
Genotypic and phenotypic correlations between single plant yield and other traits were depicted in (Fig 1). Single plant yield had positive and highly significant associations with plant height (r
g=0.246*, r
p=0.225*), number of branches per plant (r
g=0.540**, r
p=0.377**), number of pods per plant (r
g=0.748**, r
p=0.720**) and 100-seed weight (g) (r
g=0.231*, r
p=0.218*) at both genotypic and phenotypic levels, Similar results reported by
Vanniarajan et al., (2023) and
Sharma et al., (2023). Conversely, significant negative relationship was observed with Mn content (mg/kg) (r
g=-0.279**, r
p=-0.259*) and Cu (mg/kg) (r
g=-0.228*, r
p=-0.181*) at both genotypic and phenotypic levels. Days to maturity (r
g=, r
p=0.011) and Fe (mg/kg) (r
g=, r
p=0.005) had slightly positive but non-significant associations at both genotypic and phenotypic levels with single plant yield.
In contrast, Days to 50% flowering (r
g=-0.006, r
p=-0.002), Zn (mg/kg) (r
g=-0.077, r
p=-0.073) and protein content (%) (r
g=-0.009, r
p=-0.009) exhibited negative and non-significant associations with single plant yield at both genotypic and phenotypic levels. These results align with findings reported by
Edukondalu et al., (2023), Vanniarajan et al., (2023); Sharma et al., (2023); Ramasamy et al., (2021); Kandarkar et al., (2020); Narayanan et al., (2018); Anuj et al., (2018); Pandey et al., (2016); Hemavathy et al., (2019); Chaudhary et al., (2023).
Direct and indirect effects of attributes on grain yield
Path coefficient analysis, which splits the correlation coefficient into direct and indirect effects, was performed to gain a clear picture of the interrelationships among various component traits with yield (Table 2). In the present study, the number of pods per plant (genotypic path coefficient, 1.0918; phenotypic path coefficient, 0.8192) exhibited the maximum positive direct effect on single plant yield, followed by 100-seed weight (g) (0.2924, 0.2658), Zn content (mg/kg) (0.2053, 0.1189), protein content (%) (0.2018, 0.1319), and days to 50% flowering (0.1813, 0.1092). In contrast, the number of branches per plant (-0.3307, -0.0296), Cu content (mg/kg) (-0.3545, -0.2465), and days to maturity (-0.062, -0.0179) had negative direct effects on single plant yield.
The number of pods per plant, 100-seed weight (g), Zn content (mg/kg), and protein content (%) had significant positive direct effects on grain yield. Similar findings were observed by
Edukondalu et al., (2023); Kandarkar et al., (2020) and
Pandey et al., (2016). Conversely, the number of branches per plant, Cu content (mg/kg), and days to maturity showed low negative direct effects on single plant yield. Such direct effects were also reported by
Sharma et al. (2023);
Ramasamy et al., (2021); Narayanan et al., (2018); Anuj et al., (2018); Verma et al., (2018); Sharma et al., (2023); Bhadru et al., (2010).
Cluster analysis
The D2 analysis classified the genotypes into relatively homogeneous groups to minimize within-cluster diversity and maximize between-cluster diversity. The respective genotypes from diverse clusters can be utilized in breeding programs depending on the breeding objectives.
A set of 45 indigenous germplasm of pigeon pea was subjected to D2 analysis for twelve characters. Based on D2 values, four clusters were formed (Table 3, Fig 2). This indicated substantial diversity in the available gene pool of pigeon pea. Cluster analysis results revealed that Cluster III was the largest, consisting of 14 accessions, followed by Cluster I with 12 accessions, Cluster II with 11 accessions and Cluster IV with 8 accessions, similar results were reported by
Naing et al., (2022). The clustering pattern demonstrated that the pigeon pea germplasm accessions collected from different locations in Telangana were genetically diverse. Hence, the genotypes studied are reliable for hybridization and selection.
The mean values for different characters were compared across the clusters and are presented in (Table 4). Results revealed that Cluster I was better for the early days to 50% flowering and days to maturity, whereas Cluster II exhibited the highest values for 100-seed weight (g), Cu (mg/kg) and Zn (mg/kg). Similarly, Cluster III had better genotypes for plant height, number of branches per plant, pods per plant and single plant yield (g), while Cluster IV exhibited the highest values for Fe (mg/kg), Mn (mg/kg), and protein content (%). These findings align with previous studies by
Edukondalu et al., (2024); Ranjani et al., (2023); Nyirenda et al., (2020); Ranjani et al., (2021) and
Reddy et al., (2015). Similar patterns have been reported in studies
Bhatt et al., (2024); Kalyan et al., (2025); Kaur et al., (2023).
Principle component analysis
The scree plot illustrated the percentage of variance for each principal component, with PC1 showing 23.22% variability and an eigenvalue of 2.786, which then declined gradually (Table 5, Fig 3). PCA on agro-morphological, yield, and nutritional components of pigeon pea indicated that four principal components (PCs) with eigenvalues greater than 1 accounted for 69.16% of the total variability. PC1 contributed 23.22%, PC2 18.58%, PC3 15.11% and PC4 12.24% (Fig 4).
The rotated component matrix (Table 6, Fig 5) revealed that PC1 was dominated by yield traits like plant height, number of branches per plant, number of pods per plant, and single plant yield (g), while PC2 was associated with days to 50% flowering and days to maturity. PC3 and PC4 were primarily related to mineral and protein content traits such as Cu, Zn, Mn and protein content percentage (Table 7).
PC scores of the genotypes
Genotypes were selected based on PC scores, which can propose precise selection indices. High PC scores indicate high values for specific traits in those genotypes (
Singh and Chaudhary, 1977). PC scores showed positive and negative values (Table 8) across components:
In PC1, the positive scores ranged from 0.0950 (IC-0634391) to 3.741 (IC-0634411), while negative values ranged from -0.0255 (IC-0634419) to -2.3433 (IC-063443). In PC2, the positive values ranged from 0.0057 (IC-0634412) to 1.9786 (IC-0634435) and negative values ranged from -0.0182 (IC-0634398) to -1.7855 (IC-0634414). In PC3, the positive values ranged from 0.0131 (IC-0634385) to 2.0743 (IC-0634424) and negative values ranged from -0.0169 (IC-0634395) to -2.3487 (IC-0634392). In PC4, the positive values ranged from 0.0243 (IC-0634426) to 1.9577 (IC-0634411), while negative values ranged from -0.0585 (IC-0634407) to -4.2902 (IC-0634400).
Top PC scores of positive values in each PC were selected in four principal components. Collection IC-0634411 performed well across all components. IC-0634401 and IC-0634398 excelled in yield-related traits in PC1 and PC2, while IC-0634403, IC-0634416, and IC-0634435 showed strong performance in protein and mineral content traits in PC3 and PC4. These results align with findings by
Edukondalu et al., (2024); Dhanushasree and Hemavathy (2022);
Hemavathy et al., (2017) and
Rekha et al. (2013). Similar patterns of genotype performance across principal components have been reported in studies
Jain et al., (2023); Mohanlal et al., (2023); Gupta et al., (2023) and
Kumar et al., (2023), highlighting the effectiveness of PCA in identifying superior genotypes for breeding programs.