Unraveling the Relationship between Fruit Yield and Yield Related Components in Snake Gourd Genotypes using Multivariate Analysis

Trichosanthes anguina L. (snake gourd) is a monoecious crop that prefers cross pollination. Because of its outcrossing features, variability is frequently generated. Genetic diversity is vital in any crop development programme. According to Ahmed et al., (2000), there is a lot of possibility for genetic improvement in snake gourd because the germplasm has a lot of variability. The nutritional value of green fruits is exceptional and they compare favorably with the nutritional value of any other type of vegetable. For every 100 g of edible fruit, the fruit has a respectable quantity of carbohydrates (3.3%), protein (0.5%), minerals (0.5%), fiber (0.5%) and fat (0.3%). Phosphorus (135 mg/100 g), potassium (121.6 mg/100 g), magnesium and zinc are the primary mineral elements that can be found in snake gourd. (Ojiako and Igwe, 2008). In general, snake gourd fetches a higher price per unit of land, but the average yield of the crop is low in India when compared to other neighbouring countries and its production is also restricted to only three to four months of the year  (Kumar et al., 2022). When genetically varied and geographically distant lines are united, a wide range of variation can be produced. The inherent low yield potential of varieties, a lack of genetic variation and inefficient plant types, weed competition, a lack of appropriate plant ideotypes and vulnerability to biotic and abiotic stress are the key roadblocks to bolstering yields. Several academicians have studied about genetic diversity, clustering patterns and the relative contributions of various features to divergence and selection effectiveness. Given the aforementioned, genetic divergence information in these genotypes would aid in the selection of potential parental materials for a successful breeding programme. Crop genetic improvement, at the very least, consists of three distinct aspects: genetic variability, genotype selection and evaluation. The evaluation of genetic diversity is the most important step since it is crucial in determining the breeding strategies. Furthermore, selecting targeted attributes, particularly those that are not apparent to the naked eye, may be troublesome.
              
Multivariate analysis, a statistical technique encompassing several variables, has been utilized to evaluate the genetic variation of plants and establish the relationship between their traits. Multivariate analysis is a powerful tool to assist initial stage of crop improvement as it allows many traits to be evaluated simultaneously (Barth et al., 2022). Numerous crop species have demonstrated the efficacy of multivariate analysis for evaluating genetic diversity and trait correlations. In this study, multivariate methods such as principal component analysis and cluster analysis were employed to assess the genetic diversity and association between important agronomic characteristics in snake gourd. The study is carried out to assess the degree of genetic variation among genotypes and to identify different parents for future genetic studies of snake gourd.

The current study was conducted at the Vegetable Research Fields, College Orchard, HC and RI, Coimbatore, during June to October 2021. Sixteen genotypes and two varieties of snake gourd (Table 1) were collected from various regions of Tamil Nadu and the experiment was conducted in a randomized block design with two replications. Each replication includes ten pits with two plants per pit. The spacing between pits and blocks was 2.0 m and 2.0 m, respectively. Multivariate analysis was performed for 18 accessions based on the characteristics of node order of male flower, node order of female flower, days to first male flowering, days to first female flowering, internodal length, days to first fruit harvest, fruit length, fruit girth, single fruit weight, number of fruits per plant and fruit yield per plant. Under multivariate analysis, R studio version 4.2.1 (R core team, 2021) was used to conduct Box plot analysis, Principal Component Analysis (PCA) and cluster analysis. For box plots, PCA and cluster analysis, the packages gplot, factoMineR and factoextra were used.
 

Boxplot analysis
 
Boxplots were constructed for eleven quantitative traits to know the phenotypic variation between and within the eighteen germplasm collected from different geographic origins. A box and whisker plot is a graph that displays a visual representation of a data set’s statistical five-number summary including sample minimum score, first (lower quartile), median and third (upper) quartile represents the 25^th,50^th and 75^th percentile of the accessions respectively (Krishna et al., 2022). The frequency distribution for eleven quantitative traits across 18 snake gourd accessions were presented in the form of box plots to highlight genetic diversity (Table 2, Fig 1). The population variation is represented by the vertical lines (whiskers). Outliers are drawn with dots. Similar results were also reported by Lakshmi et al., (2019) and Krishna et al., (2022) in rice.
 

 

 
PCA analysis
 
Principal Component Analysis is a well-known dimension reduction method that may be used to reduce a large set of interrelated variables to a small set that contains the majority of the information available in the large set (Singh et al., 2020). The result of the Principal Component Analysis has shown the genetic diversity of the snake gourd germplasm lines.
       
Using a scree plot, the proportion of variance related with eigenvalues and principal components was depicted for each principal component (PC) derived from a graph. PC 1 demonstrated the greatest variation, 33.76 per cent, with an eigenvalue of 3.71, which steadily decreased in the remaining principal components. Up until the third Principal component, a semi-curved line was noticed; beyond that, a straight line with minimal variation in each Principal Component was detected. PC1 through PC11 have eigen values ranging from 3.71 to 0.009 According to Kaiser’s criterion (eigen value >1), conditions are regarded satisfied (Kaiser, 1958). PC1 explains 33% of the variance, followed by PC2 (25.25%), PC3 (13.41%) and PC4 (10.24%). Table 3 revealed that, out of eleven main components, five displayed >0.5 eigenvalues and approximately 90.05 percent variability, whereas four exhibited >1 eigenvalues and around 82.68 percent variability. The variability among genotypes using principal component analysis has been reported by Karunakar et al., (2022) in moringa. It is evident from the graph (Fig 2) that PC1 exhibited the greatest variation compared to the other eleven PCs; consequently, the selection of lines for characters under PC1 may be desired. Verma et al., (2017) also reported a similar curve line in pointed gourd accessions.
       

 

 
The first four PCs in the rotated component matrix showed to have the highest variability (82.68%), suggesting that traits falling within these PCs may be prioritised in snake gourd breeding. It was shown that the internodal length, fruit girth and single fruit weight were the main factors contributing to the first principal component (PC1), which explained the majority of variation. Therefore, PC 1 permits simultaneous selection of that particular phenological trait, but other PCs permit simultaneous selection of other related traits. Each component’s PC scores included both positive and negative values (Table 4). These scores could be used to construct precise selection indices based on the variability that each primary component can explain. (Rahevar et al., 2021) PC1, PC2 and PC4 contained the majority of the yields contributing traits, according to PCA. Fig 3 and Fig 4 shows how each trait and variable contributed to overall variability in the genotyping studies. The results from the present study are equivalent to those from Belay et al., (2019) and Rahevar et al., (2021). The squared cosine variables based on eleven quantitative characters and five major principal components are depicted in Fig 5 which infers that high cosine square values for traits viz., Days to first male flowering, Days to first female flowering, Days to first harvest in fruit component indicates good representation of such traits in the first component. This is in accordance with the results of Vijayakumar et al., (2020) who have explained the squared cosine variables on five major principal components in Indian cowpea.
 

 

 

 

       
To ascertain how the evaluated genotypes and characteristics were categorised, biplot analysis was utilised preceding PCA (Fig 6). Commercial varieties CO₂ and PKM1 were discovered to cluster with other genotypes using the biplot technique. The results of correlation analysis may be validated using biplot analysis, which also revealed the relationships between the characters. The biplot’s narrow angle between the qualities reveals a high correlation between them (Cai et al., 2018). Days to first female flowering, Days to first male flowering, Days to first fruit harvest and number of fruits per plant were exhibited the oppositional narrow angular direction of traits, which suggested a strong negative association between them. Widyavan et al., (2020) presented a similar biplot analysis in the Yardlong bean that explained the correlation between traits.
 

 
Cluster analysis
 
Woodyard’s Hammock method or wards method of clustering has been used to do cluster analysis of snake gourd genotypes among the traits. Wards method of clustering uses the agglomerative clustering algorithm which forms cluster based on analysis of variance instead of distance matrix or measures of association. The cluster sum of squares was 24.8% which in turn indicates variability of traits within a cluster. The cluster means of traits for the 2 cluster have been presented in Table 5 in which days taken for first female flowering and days taken to first fruit harvest exhibited maximum cluster means in cluster 1 and cluster 2. Cluster analysis grouped 16 genotypes and 2 varieties into two clusters based on traits as shown in Fig 7. Cluster I comprised of genotypes TA15 and TA16 followed by genotypes TA04, TA05, PKM1,TA07,TA08,TA10, TA03, TA09, TA02, TA12, TA06, TA13,TA11, TA14 in cluster II. Similar Ward D2 method of clustering has been used for genetic relationships among yam accessions by Agre et al., (2019).

The frequency distribution of 18 snake gourd accession displayed the data sets statistical five set summary viz., sample minimum score, first (lower quartile), median and third (upper) quartile represents the 25^th,50^th and 75^th percentile of the accessions respectively. The results of principal component analysis showed that first four principal components contributed to the highest variability. Ward D2 method of clustering clustered the 18 genotypes into two clusters. Cluster II comprised of maximum number of genotypes followed by cluster I. The cluster mean sum of squares was 24.8 %.

None.