The field experiment was at the Experimental Farm of the Department of Genetics and Plant Breeding, School of Agriculture, Lovely Professional University, Phagwara, Punjab, India, during February – May 2018. The accessions were sourced from the Indian Council of Agricultural Research – National Bureau of Plant Genetic Resources, New Delhi and their details are represented in Table 1.

The soil of the experimental site was loamy in origin and was ploughed before sowing. The climate of the area represents a tropical condition with semi-arid, hot and subtropical monsoon types. Before sowing, farmyard manure and urea were applied as basal doses. The recommended package of practices and plant protection measures to raise a good crop were timely and uniformly applied.

The experiment was carried out in a completely randomized block design with three replications. Each accession was soaked in water for 8 hours and sown at a spacing of 45×30 cm of 5 m length. The following observations were recorded during plant growth and development stages viz., days to 50% flowering [DF], days to 80% maturity [DM], plant height [PH (cm)], first flowering node [FFN], fruit diameter [FD (cm)], fruit length [FL (cm)], number of fruit per plant [FP], number of seed per fruit [SF], 1000 seed weight [TSW (g)], yield per plant [YP (g)] and yield per hectare [YH (t/ha)] from 5 representative plants of each genotype.

Data collected were subjected to ANOVA (Analysis of Variance) to evaluate the presence of statistically significant differences among genotypes for the traits studied (Panse and Sukhatme, 1954). Genotypic coefficient of variation (GCV) and phenotypic coefficient of variation (PCV) was calculated as per the formula suggested by Burton, 1952. Genetic advance and heritability were calculated by using the formula of Lush, (1949) and Allard (1960). Heritability of more than 80% is considered high. Genotypic correlation coefficients, path analysis, principal component analysis (PCA) and cluster analysis were calculated using OP-STAT (Sheoran et al., 1998) and PAST (Hammer et al., 2001). Yield per hectare was taken as a dependent variable whereas, all other traits were considered as independent variables.

Table 1
Fifteen okra Abelmoschus esculentus L Moench genotypes repatriated from NBPGR used in the study

Diversity analysis of phenotypic traits in okra

The genotypes showed high positive and significant variations for all the traits (Table S1). Yield per hectare obtained the highest positive and significant variation (73394588.940**) and the lowest was obtained for fruit diameter (0.280**). Gondane and Lal (1994) and Alam and Hossain (2008) also obtained similar results in okra.

Yield per hectare ranged from 5313.870-33714.430 kg/hectare with a mean of 20056.945 grams per genotype. Profitable yield i.e., production of 25000- 30000 kg per hectare was achieved by EC305615, EC305740, IC013356 and IC0104018. In the genetic variability studies (Table 2), the phenotypic coefficient of variation (PCV) was higher than the comparing genotypic coefficient of variation (GCV) for every trait with the close relationship between them, therefore, the environment has low impact and subsequently, the phenotypic performance of traits ought to be utilized for selection. Moderate and high GCV values were observed for most of the traits except fruit diameter, days to 80% maturity and days to 50% flowering which exhibited the presence of a high magnitude of genetic diversity in the population examined. The previous workers also observed a similar trend of greater magnitude of PCV and GCV (Ranga et al., 2021; Shanthakumar and Salimath, 2010; Prakash et al., 2011). Narrow differences between the phenotypic and genotypic coefficient of variation in most of the traits indicated that they were comparatively stable to environmental variation (Majumdar et al., 1969). However, fruit diameter and yield per hectare registered wider variation between PCV and GCV.

Heritability is a good index of transmission of traits from parents to their off-springs (Falconer, 1981). Among ten traits, nine traits displayed high heritability (low <30%, moderate 31% to 60% and high >60%) coupled with high genetic advance (low<10%, moderate 11% to 20% and high >20%) as percent over mean. This focuses on the predominance of additive gene effects for these traits; thus, crop improvement through selection based on these traits would be beneficial. Fruit girth showed low heritability accompanied with moderate genetic advance over mean portraying the role of non-additive effects and hence selection based on this trait may not be rewarding. High estimates of heritability and genetic advance were also reported by Nwangburuka et al, (2012) and Hazra and Basu (2000).

Table 2
Estimates of variability parameters for various traits of okra genotypes

(DF = Days to 50% flowering, DM = Days to 80% maturity, PH = Plant height, FFN = First flowering node, FD = Fruit diameter, FL = Fruit length, FP = Number of fruit per plant, SF = Number of seed per fruit, TSW = 1000 seed weight, YP = Yield per plant and YH = Yield per hectare, h2: heritability, GCV: Genotypic coefficient of variation, PCV: Phenotypic coefficient of variation, GA: Genetic advance)

The correlation is the overall or net impact of the segregating genes; few genes may increase both the traits leading to the positive correlation whereas, the others might increase the one and decrease the other causing the negative correlation (Falconer, 1981). Thus, to accumulate an optimum combination of yield contributing traits in a single genotype, it is essential to know the implication of the interrelationship of various traits (Ranga et al., 2019).

In the present investigation, the genotypic and phenotypic correlation coefficient analysis is represented in Table 3 and Fig.1. Days to 80% maturity showed a highly significant and positive correlation with days to 50% flowering (0.488**, 0.427**). Plant height showed a significant and positive correlation with days to 50% flowering (0.346*, 0.318*). Fruit diameter showed a highly significant and positive genotypic correlation with days to 80% maturity (0.517**, 0.314*) and plant height (0.714**, 0.393**), and only genotypic correlation was positive and significant for day to 50% maturity (0.707**). Number of fruits per plant showed a highly

Fig. 1
Correlation coefficient studies in okra genotypes

(DF = Days to 50% flowering, DM = Days to 80% maturity, PH = Plant height, FFN = First flowering node, FD = Fruit diameter, FL = Fruit length, FP = Number of fruit per plant, SF= Number of seed per fruit, TSW = 1000 seed weight, YP = Yield per plant and YH = Yield per hectare) significant and positive correlation with first flowering node (0.602**, 0.550**). Number of seeds per fruit showed a highly significant and positive correlation with plant height (0.693**, 0.684**) and fruit length (0.580**, 0.564**). 1000 seed weight showed a highly significant and positive correlation with fruit length (0.453**, 0.309*). Yield per plant showed a highly significant and positive correlation with first flowering node (0.459**, 0.442**), fruit length (0.357*, 0.355*), number of fruits per plant (0.417**, 0.395**) and number of seed per fruit (0.421**, 0.411**). Yield per hectare showed a highly significant and positive correlation with 1000 seed weight (0.391**, 0.340*). Comparable results for okra yield having a positive relationship were proposed by Ranga et al. (2021), Reddy et al. (2012), Raval et al. (2019) and Duggi et al. (2013). Yield per plant showed a highly significant and negative correlation with days to 80% maturity (-0.635**, -0.627**). Yield per hectare showed a significant and positive genotypic correlation with days to 50% flowering (-0.379*).

Path analysis provides information about the cause and effect in understanding the association between two variables. It allows the assessment of the direct effects of different traits on crop yield just as their indirect effects by means of other component traits. Hence, it gives a premise for the selection of predominant genotypes from diverse populations (Komolafe et al., 2021). The genotypic and phenotypic path coefficient analysis is represented in Table 4 and the data revealed that plant height (3.893, 0.239) had the highest direct positive effect towards the yield per hectare and other traits such as days to 80% maturity (0.811, 0.053), number of fruits per plant (2.871, 0.174) had direct effects. Traits such as days to 50% maturity (-0.575,-0.267), and yield per plant (-1.868, -0.197) had adirect effect with a negative sign. Residual effect (0.475, 0.800) indicated the effect of other possible independent traits, which were not included in the study, on the dependent variable i.e., yield per hectare. The results are in accordance with the findings of Ranga et al., (2021); Dwivedi and Sharma (2017); Das et al. (2012).

Principal component analysis (PCA) reflects the importance of the largest contributor to the total variations at each axis of differentiation (Sharma, 1998). PCA (Table 5 and Table S2) was performed to provide partial visualization of the data set in a reduced dimension and first three principal components have Eigenvalues>1 and contributed to 80.517 percent variation. From the loading of the variables in PC I, it was found that days to 50% flowering, days to 80% maturity and plant height were the dominant features that contributed to 36.662 percent of the total variation. In PCA II, plant height, fruit length, number of seed per fruit, yield per plant and yield per hectare exerted a maximum influence which accounts for 27.862 percent of the total variation. Days to 50 percent 1998). PCA (Table 5 and Table S2) was performed to provide partial visualization of the data set in a reduced dimension and first three principal components have Eigenvalues>1 and contributed to 80.517 percent variation. From the loading of the variables in PC I, it was found that days to 50% flowering, days to 80% maturity and plant height were the dominant features that contributed to 36.662 percent of the total variation. In PCA II, plant height, fruit length, number of seed per fruit, yield per plant and yield per hectare exerted a maximum influence which accounts for 27.862 percent of the total variation. Days to 50 percent flowering, days to 80% maturity, first flowering node and fruit diameter were the dominant features in PCA III which accounted for 15.993 percent of the total variation. Ranga et al. (2021), Ahiakpa et al. (2013) and Amoatey et al. (2015) also indicated high genetic diversity using PCA. Few traits viz., fruit length, 1000 seed weight, number of seed per fruit and fruit yield per plant offered more towards variation as accounted for by different scientists in okra (Bhardwaj et al., 2021; Ahiakpa et al., 2013; Amoatey et al., 2015; Nwangburuka et al., 2012).

Table 3
Genotypic and phenotypic correlation coefficient studies in okra genotypes

(Bold numbers indicated positive and significant correlation between two traits)

(DF = Days to 50% flowering, DM = Days to 80% maturity, PH = Plant height, FFN = First flowering node, FD = Fruit diameter, FL = Fruit length, FP = Number of fruit per plant, SF = Number of seed per fruit, TSW = 1000 seed weight, YP = Yield per plant and YH = Yield per hectare; rg = genotypic correlation coefficient, rp = phenotypic correlation coefficient)

Table 4
Genotypic and phenotypic path coefficient analysis for various okra genotypes

gp residual effect: 0.475 pp residual effect: 0.800

(DF = Days to 50% flowering, DM = Days to 80% maturity, PH = Plant height, FFN = First flowering node, FD = Fruit diameter, FL = Fruit length, FP = Number of fruit per plant, SF = Number of seed per fruit, TSW = 1000 seed weight, YP= Yield per plant and YH = Yield per hectare); gp = genotypic path coefficient, pp = phenotypic path coefficient)

Table 5
Eigen value and percent variation explained by first 5 principal components and correlations between PC scores and quantitative traits

(DF = Days to 50% flowering, DM = Days to 80% maturity, PH = Plant height, FFN = First flowering node, FD = Fruit diameter, FL = Fruit length, FP = Number of fruit per plant, SF = Number of seed per fruit, TSW = 1000 seed weight, YP = Yield per plant and YH = Yield per hectare)

A biplot was drawn using the values of PC I and PC II (Figure 2). The greater the angle between the traits, the lesser the association between them. Placement of genotypes in quadrants signifies variability. Accessions are placed in quadrants using vector scores of PC I and PC II. However, no obvious grouping of genotypes was observed, and some overlapping occurred among groups the relatedness of the genotypes across the collection. In the biplot graph of

Fig. 2
Biplot between PC1 and PC2 showing contribution of various traits responsible for variability in okra.

PCA, quadrant I (+,+) consists of zero accessions formed the cluster 1 which were highly influenced by three traits characters viz. fruit length, yield per plant and yield per hectare through genotypes spread towards midway through X and Y-planes of quadrant- I. The cluster II corresponding to the quadrant II (-,+) contained 5 genotypes, which were influenced by number of seeds per plant, plant height, fruit diameter and days to 50% maturity. Similarly, the cluster III corresponding to quadrant III (-,-) consisted also of 3 genotypes which were influenced by days to 80% maturity only whereas the cluster IV corresponding to quadrant IV (+,-) also consisted of 7 genotypes which were influenced by first flowering node, number of fruits per plant and 1000 seed weight, respectively.