Principal component analysis
Principal components and their correlation coefficients (Eigen vectors) for twenty-one traits are presented in Table 1. The principal component analysis (PCA) grouped the traits under six principal components (PCs), which accounted for 88.97% of the total variation. The number of stamens per flower (0.308) and seed weight (g) per receptacle (0.340) were correlated positively with PC1, which accounted for 37.543% of total variation. Number of seeds per receptacle and the number of petals per flower were positively correlated and length of petiole and length of flower stalk was correlated negatively with PC2, which accounted for 19.089% of total variation. Rhizome length, rhizome weight, weight of flower and diameter of flower bud were positively correlated with PC3, which accounted for 14.193% of total variation. Diameter of flower stalk and leaf width were correlated negatively and leaf length was correlated positively with PC4, which accounted for 7.798% of total variation. Number of leaf venation was correlated positively with PC5, accounting for 6.159% of total variation. Diameter of flower and petal length was correlated negatively with PC6, which accounted for 4.189% of the total variation. PCA is a useful technique for evaluating genetic variation among accessions which pinpoint the major variables causing variation and highlight the significant variation in genotypes
(Verma et al., 2018). The use of principal component analysis (PCA) has proven useful in identifying the genotypes with the most desirable traits for breeding programs in crop plants
(Sarkar et al., 2024). Similarly,
Guo et al., (2010) reported four principal components with 77.33% of total variation in sixty-eight rhizome lotus (
Nelumbo nucifera Gaertn) genotypes from China. Six principal components with 84.75% of total variation based on the different physicochemical parameters of the
Garcinia pedunculata Roxb. accessions were reported from Manipur, India
(Hazarika et al., 2023). Su et al., (2019) also reported seven principal components with a cumulative contribution of 81.86% of total variation under a comprehensive evaluation of 49 waterlily germplasm. Six principal components with a contribution of 84.75 % of total variation were estimated in the bottle gourd
(Kumar et al., 2023) and 94.05% of total variation was estimated in the snake gourd
(Fathima et al., 2023).
Clustering pattern of the genotypes
Cluster analysis using D
2 statistics is shown in Table 2. On the basis of D
2 values, 33 genotypes were categorized into six different clusters. Cluster II was the largest with twenty-three genotypes followed by cluster I with four genotypes. The clusters IV, V and VI possessed only one genotype. The given method of analysis may help to select diverse parents and broaden the local germplasm base for future crop improvement programs. These findings are in close conformity with the results of
Guo et al., (2010) and
Su et al., (2019), who reported five clusters based on phenological characteristics of rhizome lotus genotypes and six clusters in waterlily, respectively.
Intra- and inter-cluster distance
Intra- and inter cluster distances are the indicators of genetic diversity between the clusters. Table 3 shows the average intra and inter-cluster distance based on D
2 analysis. Inter-cluster distances were higher in magnitude than intra-cluster distances in the present investigation, indicating a significant genetic variation among the genotypes under investigation. The intra-cluster distance ranged from 0.00 (cluster IV, V and VI) to 184.40 (cluster II). The minimum intra-cluster distance was zero as only one genotype was present in these clusters. It is clear from table 3 that the highest inter-cluster distance was obtained between cluster III and VI (5098.07) followed by cluster I and VI (4971.24). The diversity between the genotypes might be due to the differences in cultivation history, climatic adaptation and cross-breeding among the genotypes, which causes gene flow between the populations
(Ghorbani et al., 2020). The result showed crop improvement programs could employ the distant cluster’s genotypes as a viable source to get a broad range of diversity among the segregates and create populations with high yielding transgressive segregates
(Singh et al., 2014).
Cluster mean for twenty one characters
Table 4 (a,b) shows the cluster means of different clusters based on twenty-one morphological characters under the study. In clusters I, cluster mean was highest for most of the economic traits
viz., number of petals per flower (115.48) and weight of flower (50.46 g). Whereas in cluster III, cluster mean was highest for length of petiole (223.86 cm) and length of flower stalk (230.11cm). Maximum mean values for rhizome length, rhizome diameter and rhizome weight (g) per 50 cm were recorded in clusters VI [Table 4(b)]. Genotypes with significant mean performance of cluster and inter-cluster distance could be utilized as potential parents for further breeding programs (
Sushil et al., 2023).
a
b