Principal Component Analysis and Genetic Association of Seed Related Traits in an Underutilized Pulse Crop, Ricebean (Vigna umbellata)

Legumes are the second most significant crop group after cereals which form the core component of human nutrition. For vast majority of the people, they are the chief source of protein. Despite the availability of a number of edible legumes, their rate of consumption is significantly higher than their rate of production. In this context, underutilized legumes have scope to address this challenge. Hence, efforts need to be undertaken to integrate the underutilized crops in to the mainstream cultivation. This could be effective in resolving the protein malnutrition. In India, ricebean [Vigna umbellata (Thunb.) Ohwi and Ohashi] remains as an underutilized legume species, despite having a good nutritional profile and many desirable traits such as high harvest index, resistance to bruchids, yellow mosaic virus, cercospora leaf spot, bacterial leaf spot, drought tolerance etc.
       
Among the biometrical traits, seed related traits viz., seed coat colour, seed size and hundred seed weight are important since the later two are directly related with single plant yield. These traits also draw consumer preference, thereby fetching high market value. But there are limited numbers of studies with reference to seed related traits. Hence, the present study was undertaken to study the genetic diversity and relationship among seed traits present among the set of ricebean germplasm.

A set of 109 ricebean germplasm with seven checks  were evaluated in Augmented design II during Rabi, 2022-23 at Department of Pulses, Centre for Plant Breeding and Genetics, Tamil Nadu Agricultural University, Coimbatore. The whole experimental area was divided in to three blocks. Each block encompassed a portion of the test entries and all the seven checks. Thus, the checks were replicated three times. The first, second and third block was allotted with 40, 40 and 29 test entries, respectively. The genotypes were planted in a single row of four meter length with a spacing of 45 × 15 cm. In order to establish a good crop, all the recommended package of practices was carried out at appropriate time.
       
All the hundred and nine ricebean germplasm accessions were characterized for 10 seed related traits (three qualitative and seven quantitative traits). For qualitative traits, viz., seed coat colour, seed coat pattern and type of seed coat pattern, the scaling adopted was as per the descriptor of ricebean available in genesys portal. For quantitative traits, five plants of each genotype were randomly selected; seeds were collected and used to record data on seed thickness, seed length, seed breadth, length breadth ratio, bulk density, seed volume and hundred seed weight.
       
A total of 25 seeds were selected randomly from five plants (5 random seeds /plant), of each genotype and used for measuring seed thickness, seed length and seed width. Seed thickness was measured using digital vernier calliper, whereas graphical method was adopted to measure seed length and seed breadth. Length breadth ratio was calculated by dividing the values of seed length with the corresponding seed breadth values.
       
Bulk density refers to the ratio of mass of seed sample to the total volume occupied by the seed sample. It was determined by recording the mass of known number of seeds (100 seeds) and dividing it by the total volume of the seeds. Seed volume refers to the rise in volume of water (Mohsenin, 1986). To determine seed volume, an initial quantity of 10 ml of water was taken in a graduated measuring cylinder. Then, known quantity of seeds (100 seeds) was added to the water. The rise in volume of water gives the seed volume.
       
The mean data was used for testing the statistical significance. Analysis of variance, genetic variability parameters, genetic association analysis and principal component analysis were done using the R software version 4.2.3. Analysis of variance for Augmented design II and genetic variability parameters were derived using augmented RCBD package (Aravind et al., 2021). Genetic association analysis and principal component analysis (Pearson, 1901) were carried out using R software with the packages, psych (Revelle, 2013) and factoextra (Kassambara and Mundt, 2020), respectively. Genetic variability parameters viz., genotypic coefficient of variation, phenotypic coefficient of variation, genetic advance as per cent of mean were determined as per the formula given by Jhonson et al., (1955). Heritability (broad sense) was calculated according to the formula given by Lush (1940). The estimates of genotypic and phenotypic correlation coefficients were worked as per the formula given by Jhonson et al., (1955).
 

Morphological characterization of 109 ricebean germplasm accessions revealed that green yellow seed coat colour was  predominant and was present in 50.00% of the genotypes, followed by yellow (25.86%), cream yellow (15.52%), dark green (5.17%), maroon (1.72%) and brown (1.72%) (Table 1). Seed coat pattern was absent in majority of the genotypes (83.62%), whereas the pattern was medium in 16.38% of the genotypes in which speckled type of seed coat pattern (100.00%) was present as the exclusive pattern type.
 

       
The mean performance of the genotypes for the seven seed traits is given in Table 2. Analysis of variance indicated that all the 109 ricebean germplasm accessions exhibited significant differences for the seven quantitative traits studied, signifying the prevalence of sufficient amount of variability in the experimental material (Table 3).
       

 

 
For all the seven traits studied, the phenotypic coefficient of variation (PCV) was marginally higher than the genotypic coefficient of variation (GCV) (Table 4). This revealed the lower influence of environment and preponderance of genetic factors in demonstrating variability on these traits. Bulk density, seed volume and hundred seed weight recorded moderate PCV and GCV (11-20%) which signifies a relatively better potential for further improvement through selection. Similar findings of moderate PCV and GCV were recorded for hundred seed weight (Adewale et al., 2012 in African yam bean; Latha et al., 2013 in horsegram; Devi et al., 2021 in pre-breeding lines of blackgram) and seed volume (Adewale et al., 2012). In the present study, the traits viz., seed thickness, seed length, seed breadth and length breadth ratio recorded low PCV and GCV (<10%) which implies a limited scope for further improvement of these traits. Similar results of low PCV and GCV were reported for seed thickness (Latha et al., 2013 and Devi et al., 2021), seed length (Adewale et al., 2012; Latha et al., 2013 and Devi et al., 2021), seed breadth (Adewale et al., 2012 and Devi et al., 2021) and length breadth ratio (Sulistyo et al., 2021 [low GCV alone] and Devi et al., 2021).
 

       
High heritability along with high genetic advance as per cent of mean (GAM) were recorded for the traits viz., bulk density, seed volume and hundred seed weight. In earlier investigations also, high heritability was noted for hundred seed weight in soybean by Sulistyo et al., (2021), whereas high heritability combined with high GAM was reported for seed volume in horsegram by Latha et al., (2013) and for hundred seed weight in horsegram and blackgram pre-breeding lines by Latha et al., (2013) and Devi et al., (2021), respectively. This indicates that the above traits are controlled by additive gene action and simple directional selection would be effective. High heritability along with moderate GAM were recorded for seed thickness, seed length and seed breadth. Similarly, high heritability was observed by Sulistyo et al., (2021) for seed breadth and seed thickness in soybean. This implies that the high heritability might be due to favourable environmental influence and moderate GAM might be due to additive gene action. Hence, through stringent selection, these traits could be improved in the successive breeding cycles. High heritability along with low GAM was recorded for length breadth ratio. This highlight there is no scope for improvement of these traits through direct selection.
       
In order to investigate the presence of any potential association between the seven seed related traits, Pearson correlation analysis was carried out. In the present investigation, seed volume (0.81), seed thickness (0.45), bulk density (0.41), seed length (0.41) and seed breadth (0.36) exhibited highly significant and positive correlation with hundred seed weight (Fig 1). Similarly, significant positive association of hundred seed weight with seed thickness, seed length and seed volume was reported by Latha et al., (2013) in horsegram; with seed length and seed breadth was recorded by Khamassi et al., (2021) in faba bean. Hence, these traits could be considered during indirect selection in order to improve the hundred seed weight. In addition, seed thickness evidenced highly significant and positive correlation with seed breadth (0.57) followed by seed volume (0.42) and seed length (0.37). Akin to the present study, Latha et al., (2013) also reported that the trait seed thickness exhibited positive association with seed breadth. However, seed thickness also exhibited highly significant and negative correlation with length breadth ratio (-0.33). From the results, it could be ascertained that, in ricebean, the traits viz., seed breadth, seed volume and seed length have a positive influence on seed thickness, whereas length breadth ratio has a negative influence on seed thickness. Hence, seed thickness could be increased by selecting genotypes with increased seed length, but only up to a certain limit. Therefore, in order to increase seed thickness, selection of genotypes with high seed breadth value and moderate seed length value might be rewarding. Further, the trait seed length expressed highly significant and positive correlation with seed breadth (0.61), seed volume (0.38) and length breadth ratio (0.22). In line with the present finding, significant positive correlation of seed length with seed breadth and seed volume was reported by Latha et al., (2013); with seed breadth by Khamassi et al., (2021). This indicates that increase in seed length increases seed breadth, length breadth ratio and ultimately, volume of the seeds. Besides, seed breadth showed highly significant and positive correlation with seed volume (0.41), whereas highly significant and negative correlation with length breadth ratio (-0.64). Latha et al., (2013) also reported significant positive association between seed breadth and seed volume in horsegram accessions. Moreover, significant and positive correlation (0.21) was also observed between seed volume and bulk density.
 

       
Principal component analysis revealed that the first three principal components were found with Eigen value greater than one (Table 5). Eigen values of the first three principal components were 3.0319, 1.5742 and 1.1062, respectively. The first principal component accounted for 43.31% of the variance to the total variance, whereas the second and third principal components attributed 22.49% and 15.80% of the variance, respectively and the above three components cumulatively accounted 81.60% to the total variance.
       

 
The variation in the first principal component was ascribed due to positive loading of length breadth ratio (0.2078), whereas the variation in the second principal component was due to positive loading of the following traits viz., length breadth ratio (0.5832), bulk density (0.5167), hundred seed weight (0.3688), seed volume (0.2440) and seed length (0.1130) (Table 6). The traits viz., bulk density (0.4940), hundred seed weight (0.1606) and seed volume (0.1178) contributed variation to the third principal component.
 

       
In the present study, the angle between the trait vectors, hundred seed weight and seed volume; seed volume and seed length; seed length and seed thickness; seed thickness and seed breadth were found to be less than 90° (Fig 2) indicating a strong positive correlation between the traits pairs. Besides, a strong negative correlation was observed between seed breadth and length breadth ratio as well as between seed thickness and length breadth ratio since the angle between the trait pairs were greater than 90°.
 

       
The genotypes in the biplot were dispersed among all the four quadrants indicating the presence of significant amount of genetic diversity among the accessions. However, the genotypes (excluding the outliers) could be grouped in to four groups in which group 1, 2, 3 and 4 comprised of 12,15,9 and 18 genotypes, respectively (Table 7). The genotypes within each group, as compared to other groups, exhibited relatively similar range of values for the seed related traits.
 

       
The interrelationships among the accessions and seed related traits based on PC1 and PC2 were delineated in the trait by genotype biplot (Fig 2). In a biplot, the location of a genotype with reference to the direction of arrow reveals the performance of that genotype for a specific trait. The biplot revealed that seed thickness was maximal in the genotype IC361364 and IC520750, while the seed length was greater in the genotype IC369282. The values of seed breadth were determined to be higher in the genotypes IC469191 and IC342374. The genotype IC444157was detected with higher length breadth ratio. Bulk density was high in the genotypes viz., IC469192 and IC520941, whereas seed volume was observed to be high in the genotypes IC369282 and IC521004. Hundred seed weight was maximum in the genotypes IC521004 and LRB324.

The best performing genotypes identified in the present study for the seed traits, could be further tested for yield. In order to develop a ricebean cultivar with one or more desired seed characteristics, hybridization could be effected across the corresponding groups and selection could be made among the progenies using the associated traits established in the present study.

The authors are grateful to National Bureau of Plant Genetic Resources (NBPGR) for providing the seed materials.