Principle Component Analysis (PCA) and Character Interrelationship of Irrigated Blackgram [Vigna mungo (L.) Hepper] Influenced by Liquid Organic Biostimulants in Western Zone of Tamil Nadu

Pulses are important crops in India because of their low cost and high quality protein. They play a major role in providing a balanced protein component in the diet of the people. Among pulses, blackgram [Vigna mungo (L.) Hepper], occupies a unique place. It is grown both as a pure and mixed crop along with maize, cotton, sorghum and other millets (Ajaykumar et al., 2022a).

The yield of blackgram is low due to various reasons including poor management practices, physiological, biochemical and inherent factors associated with the crop. Insufficient partitioning of assimilates, flower dropping and poor pod setting are mainly due to   lack of nutrients during critical crop growth resulting in poor yield. Fertilizer application is an important practice to increase the yield of blackgram (Ajaykumar et al., 2022b). Organic substances are known to influence a wide array of physiological parameters like alteration of plant architecture, assimilate partitioning, promotion of photosynthesis, uptake of nutrients (mineral ions), enhancing nitrogen metabolism, promotion of flowering, uniform pod formation, increased mobilization of assimilates to defined sinks, improved seed quality, induction of synchrony in flowering and delayed senescence of leaves (Pradeep and Elamathi 2007).

The role of foliar applied panchagavya and dhasagavya in the production of many plantation crops has been well documented in India (Selvaraj, 2003). The use of fermented, liquid organic fertilizers and effective microorganisms (EM) as foliar fertilizers have been introduced to modern agriculture in recent years to produce food with good quality and safety (Galindo et al., 2007).

Fish amino acid (FAA) is a liquid and great value to both plants and microorganisms in their growth. It has abundant amount of nutrients and various types of amino acids. Seaweed concentrates benefit plants as they contain growth-promoting hormones (IAA, IBA and cytokinins), trace elements, vitamins and amino acids (Khan et al., 2009). Integrated use of seaweed liquid fertilizer in combination with chemical fertilizer and their proper management for better growth and yield is very essential. In green gram, foliar application of liquid bio fertilizers during vegetative and flower bud initiation stages increased number of flowers, pods and seeds per plant and seed yield. Foliar application of organic substance increased the chlorophyll content and promoted epicotyls elongation of soybean, mungbean and pea (Senthil et al., 2003).

Exogenous application of pink-pigmented facultative methylotrophs (PPFM) produces some benefits in alleviating the adverse effects of drought stress and also improves germination, growth, development, quality and yield of crop plants (Hayat et al., 2010). Agronomists frequently assess a great diversity of characteristics for appraisal and characterization. In such situations, principal component analysis (PCA) is used to reduce massive data sets containing multiple variables into their principal components to acquire a deeper understanding of the data (Amy and Pritts, 1991). This statistical technique is frequently utilised for data compression, reduction and transformation (Mishra et al., 2017). Principle component analysis is a mathematical procedure that transforms a number of (possibly related) variables into a (smaller) number of principal component variables (García and García, 2010). The eigenvalue of a specific principal component represents the degree of variance in attributes that is explained by that principal component, which is extremely valuable for crop production trait selection (Singh et al., 2020). The PCA is widely utilized in examining elite growth and physiological features, it simultaneously analyses several parameters of each individual under investigation. PCA evaluates the significance and contribution of each variable to the overall variance (Leonard and Peter, 2009).

When evaluating materials based on a variety of characteristics, there are numerous crucial factors to consider. Consequently, PCA is used to study the relationships between traits and efficiently visualise the similarities between individuals or treatments in which various factors exert strong effects on growth, yield and physiological traits. Under these conditions, it is considerably more difficult to visually summarise a set of agronomic data by describing agronomic features; hence, multivariate methods should be employed. In recent years, numerous research efforts have been focused on agronomic physiological characters that influence yield. PCA is a useful technique for the reduction of large data set with many variables into important principal components for a better understanding of information. Keeping these points in view, the present investigation was conducted to assess the relationship among characters of blackgram under irrigated conditions.

A field experiment was conducted during 2021 in Vanavarayar Institute of Agriculture, Pollachi. The experiment comprised of seven treatments viz., 100% RDF along with the foliar application of dhasagavya at 3% (T₁), 100% RDF along with foliar application of liquid rhizobium at 1% (T₂). 100% RDF along with foliar application of fish amino acid at 1% (T3), 100% RDF along with foliar application of panchagavya at 3% (T₄), 100% RDF along with the foliar application of PPFM at 1% (T₅), 100% RDF along with the foliar application of seaweed extract at 3% (T₆), control (T₇) and was laid out in randomized block design with in three replications. Liquid biofertilizers and organic bio-stimulants were purchased from Tamil Nadu Agricultural University, Coimbatore. Blackgram variety ‘VBN 8’ was used for the study.

The recommended doses of N, P₂O₅, K₂O were 25, 50, 25 kg ha^-1, respectively. Full dose of nitrogen, phosphorus, potassium in the form of urea, SSP, MOP were applied basal as per treatments. In addition to this, gypsum 20 kg ha^-1 and soil application of 25 kg ZnSO₄ ha^-1 were applied. Liquid bio fertilizers and organic bio stimulants were given as foliar spray at 30 and 45 days after sowing of blackgram. All other agronomic practices were adopted as per the need of the crop.

The growth characters viz., plant height, number of branches per plant, leaf area index (LAI) and dry matter production (DMP) were recorded. The maximum plant height was measured from the base of the stem to the tip of the longest trifoliate leaf. Numbers of branches were counted by manual and LAI was measured by using leaf area meter (LICOR 3000). DMP of various plant parts was arrived at by taking the sum of all the plant parts after keeping the sample in oven at 80°C for 48 hours. The physiological biochemical parameters viz., crop growth rate, chlorophyll content, soluble protein and Nitrate reductase activity were estimated. The CGR was computed using the formula suggested by Watson (1958). Chlorophyll content of leaves was recorded as described by Yoshida et al., (1976). Soluble protein content of the leaf was estimated at by using folinciocalteau reagent by adopting the procedure described by Lowry et al., (1950). Nitrate reductase activity (NRase activity) (Nicholas et al., 1976) were also estimated. Yield attributes viz., number of pods per plant, number of seeds per pod, pod weight per plant, pod length and grain yield were recorded during harvest stage. These thirteen quantitative characteristics were considered for the study.

The primary data were subjected to PCA using XLSTAT software. It reduces the dimensions of multivariate data to a small number of principal axes, generates an eigenvector for each principal axis and produces component scores for the characters. The PCA analysis was done to investigate the link between attributes and establish selection criteria and the optimum factors for yield estimation. In agronomy, the forms of PCA visualisation for improved decision-making are typically a cloud of dots or a circle of correlations between variables. They reflect the directions in which the data exhibits the greatest variance and the greatest dispersion. PCA was employed to the replicated data to sort out the large amount of quantitative data. 

The descriptive statistics viz., mean, maximum, minimum, standard deviation (SD) and coefficient of variation (CV) were measured for 13 characters recorded under irrigated conditions. The results revealed that the highest variation was found in crop growth rate with a CV of 36.76 per cent (Table 1). The pod length had the lowest coefficient of variation (11.89 per cent). It was noticed that most of the variables were highly correlated among themselves (Table 2 and Fig 1). Strong correlations were found between the grain yield and plant height (0.72), number of branches per plant (0.84), DMP (0.82), crop growth rate (0.74), total chlorophyll content (0.75), nitrate reductase activity (0.85); plant height and total chlorophyll content (0.72); number of pods per plant and crop growth rate (0.71); DMP and crop growth rate (0.79), total chlorophyll content (0.76), nitrate reductase activity (0.88); crop growth rate and total chlorophyll content (0.76); total chlorophyll content and nitrate reductase activity (0.78). Since most of the correlation coefficients between the variables were greater than 0.3, all the variables included in the correlation analysis were subjected to the principal component analysis (Rymuza et al., 2012).







On the basis of data of seven treatments, principal component analysis was performed to find the most important growth and physiological characteristics. PCA revealed that, out of thirteen principal components, only five had eigen values greater than 0.5 and accounted for 85.35 percent of the analysed traits’ variability, while the remaining eight principal components contributed just 14.65 per cent. A scree plot (Fig 2) depicted the proportion of variance linked with eigen values and principal components for each graphed principal component (PC). PC1 produced the most variance, 57.95 per cent, followed by PC2 (9.57 per cent), PC3 (6.44 per cent), PC4 (6.31 per cent) and PC5 (6.31 per cent) (5.07 per cent). Principal component approach permitted the reduction of thirteen core traits to eight new variables from five principal components while retaining a substantial portion of the variation of primary data. Jeberson et al., (2018) assessed the principal components of twenty-five blackgram genotypes and reported that the first three components accounted for 84.52 per cent of the overall variation, while the remaining four components accounted for just 15.48 per cent.



On the basis of factor loadings (Table 3), the first principal component increases with grain yield, DMP and nitrate reductase activity. This implies that increase in one variable viz., grain yield would increase the other variables like DMP and nitrate reductase activity. Meanwhile, the second component is represented by the number of pods per plant and leaf area index. The third, fourth and fifth components carried the information related to pod length, soluble protein content and leaf area index, respectively. Overall, PCA was able to identify the crucial agronomic characteristics responsible for population variability. Jeberson et al., (2018); Sridhar et al., (2020); Girgel (2021); Beyzi et al. (2019); Mohi-Ud-Din et al. (2021); Singh et al. (2020); Qaseem et al. (2019); Zafar et al. (2021) and Kakar et al. (2021) conducted similar studies related to agronomic trait selection.



Several researchers utilised the PCA biplot to investigate the link between traits in various crops (Mohanlal et al., 2020; Aslam et al., 2017 and Maqbool et al., 2016). The length of the vector was determined by the character’s contribution to the primary component (Fig 3). In addition, the angle of the character vectors reflected the relationship between variables. A positive correlation existed if the angle between two trait vectors was less than 90° degrees (acute angle). The biplot of the PCA showed that all vectors were concentrated in the first and fourth quadrants. It implied that most of the traits were closer to each other and highly correlated among themselves which was earlier evident in the above-mentioned correlation results. If the angle between the two traits exceed 90° degrees (an obtuse angle), it shows a negative correlation. In this study, all the traits had acute angles between themselves, which indicated the absence of negative correlation. The variables that exhibited lesser or zero relations were pod length and DMP; number of seeds per plant, nitrate reductase activity, pod weight per plant and number of branches per plant.



A scatter plot drawn between the first and second principal components depicted clear projections of the cases on a factor plane (Fig 4). The distribution of the active observations based on PC1 and PC2 illustrated that the treatment variations within the population and explained how they were widely dispersed along both axes. PC1 and PC2 demonstrated a very distinct separation between the seven treatments with minimal overlap, indicating that the treatments were distinct.



The analysis allowed the reduction of thirteen primary traits to five new variables which illustrated 67.52 per cent of the variability from the first two principal components. According to Sivaprakash et al., (2004), the features that comprise the first, second and subsequent principal components have the greatest discriminating power and  they differentiate the studied treatments. On the whole, traits such as grain yield, DMP and nitrate reductase activity had a high influence on PC1. Meanwhile, numbers of pods per plant and leaf area index were the characters that exhibited higher influences on PC2.

Principal component analysis (PCA) was performed to determine the most advantageous traits and to determine the relationship between characters in various treatments in blackgram. The primary components that contributed  most to the quantitative characters based on the interaction between treatments and vector were grain yield, DMP, nitrate reductase activity, number of pods per plant  and leaf area index. In the study, the reduction of thirteen investigated features to two principal components explains approximately 67.52 per cent of the total input data variability. The generated non-correlated variables may be used in further analyses when the co-linearity of variables is not assumed.