INTRODUCTION Greengram (Vigna radiata (L.) Wilczek) is the 3rdmost important pulse crop in India next to chickpea and redgram. This self-pollinating legume belongs to family fabaceae. It isa diploid crop having chromosome number 2n=22 (Karpechenko, 1925) and originated from the Indo-Burma region of the Hindustan center (Vavilov, 1926). Nutritionally, greengram is a protein-rich food containing around 23% protein. The protein is rich in lysine, minerals, and vitaminshence meeting the dietary needs of the vegetarian population of the country. The genetics of the protein or its constituents may give some valuable signs for further enhancement in protein level and quality of protein. Greengram is cultivated widely because of its adaption to short growth duration, low water requirement, restoring of soil fertility through biological nitrogen fixation, tendency to reduce greenhouse gases, capacity to increase carbon sequestration, and thus, plays an important role in sustainable agriculture. It is grown all over the country in several seasons as a sole crop as well as an intercrop or mixed crop with cereals forming an important constituent of crop rotation. Yield is a complex trait and highly influenced by the environment due to polygenic effects (Allard, 1960). To accumulate optimum yield, it is necessary to know the presence of genetic variability among the yield contributing characters for effective selection. Genetic diversity is the basic requirement for any breeding programme which aims at genetic amelioration of yield. Effective hybridization programmes between genetically diverse parents will cause a substantial amount of heterotic response in F1 hybrid and therefore the broad spectrum of variability in segregating generations. Therefore, an attempt was made to study the extent of the genetic divergence among 66 genotypes of greengram through D2 statistics (Mahalanobis, 1936) to choose genetically diverse parents. Further protein analysis was done on the seed protein content of germplasms. MATERIALS AND METHODS The present experiment was conducted at EB-II section, Department Plant Breeding and Genetics, College of Agriculture, Odisha University of Agriculture and Technology (OUAT), Bhubaneswar, during rabi 2019-20.A set of 66 greengram germplasm lines including standard ruling varieties, important pre-released varieties and popularly adopted local landraces were laid out in randomized block design with 3 replications. The crop was raised following recommended agronomic packages & practices and observations were recorded on five randomly selected plants per replication for ten quantitative characters viz., days to 50 % flowering (DF), days to maturity (DM), plant height (PH) in cm, number of branches/plant (B/P), number of clusters/plant (C/P), number of pods/plant (P/P), pod length (PL) in cm, number of seeds/pod (S/P), hundred seed weight (HSW) in gram and seed yield/plant (SY/P) in gram. The mean for all the characters of five plants in each plot was subjected to analysis of variance (ANOVA) as per the method suggested by Panse and Sukhathme (1967). Estimation of genetic parameters like phenotypic and genotypic coefficients of variation, heritability (h2, broad sense in %) and genetic advance (GA as % of mean) were worked out according to the formulas given by Singh and Chaudhary (1977). On the basis of heritability values characters were classified to high (>60), moderate (31-60) and low (0-30) heritable (Johnson et al., 1955). Likewise, characters were classified as having high (>20), moderate (10-20) and low (0-10) genetic advance as per the method of Johnson et al. (1955).Genetic diversity analysis was conducted using Mahalanobis D2 statistic (1936) and the genotypes were grouped into clusters by Tocher’s method (Rao, 1952). For protein analysis, matured seeds from each genotype from each replication were harvested, dried up to a safe moisture level (12-13%) and powered to estimate protein content. Nitrogen content present in seed was estimated in CHNS (O) analyzer. Then it was multiplied with a factor 6.25 to arrive at the percentage protein content of seeds (Jones, 1941). RESULTS AND DISCUSSION Genetic variability: Result of ANOVA revealed that all the genotypes were significantly different for all the characters studied suggesting the presence of substantial variability among the germplasm under investigation (Table 1), which will be much beneficial for the selection of breeding material. This wide range of variation may be due to diverse origins and geographical adaptation of genotypes. Estimates of phenotypic coefficient of variations (PCV) were higher than genotypic coefficient of variations (GCV) for all the characters under study (Table 2), which indicates that there was some environmental influence on these characters (Rahim et al., 2010). Seed yield per plant exhibited the highest value of PCV and GCV indicating a relatively higher contribution of this character towards genotypic variability. Hence direct selection based on this trait would be effective. This result was in conformity with the findings of Srivastava and Singh (2012); Patel et al. (2014). It was noticed that number of branches/plant had maximum difference between GCV and PCV which indicate the environmental effect is more in this character, hence much care should be taken up while selecting this character (Varma et al., 2018). Characters like plant height, number of clusters/plant, number of pods/plant, hundred seed weight and seed yield exhibited high heritability (broad sense in %) accompanied with high genetic advance (G.A. in % of mean) resembling the action of additive gene in controlling these characters and selection based on these characters would be rewarding for enhancement of yield. Evidently, high heritability coupled with high genetic advance suggesting the role of additive gene action for seed yield per plant was reported earlier by Patel et al. (2014), Raturi et al. (2015); Mariyammal et al. (2019). Diversity analysis: Composition of clusters: Sixty-six genotypes were grouped into 6 clusters using Tocher’s method (Table 3). Cluster I containing 51 genotypes was the largest one followed by cluster III (7), cluster II (5), while the rest three clusters viz., cluster IV, V and VI represented one genotype each suggesting presence of heterogeneity among the genotypes. It was observed from the results that genotypes having different sources of origin grouped into same cluster. Also, materials of the same source have been grouped in different clusters indicating eco-geographic diversity is not a reliable index for genetic diversity (Vijaya and Shekhawat, 2012). Therefore, it is advised that the selection of parents during crossing programme should be based on genetic divergence rather than their geographic diversity. Intra- and inter-cluster distance: Average intra- and inter-cluster distances (D2) values among the six clusters were calculated using divergence analysis (Table 4). The intra-cluster distance varied from 0 (cluster IV, V and VI) to 1023.36 (cluster III) and the inter-cluster distance varied from 654.34 (cluster V & VI) to 6154.40 (cluster IV & VI). Maximum intra-cluster distance indicated the presence of greater diversity among the genotypes allocated in those respective clusters. Minimum inter-cluster distance was revealed between cluster V & VI (654.34) showing that the genotypes present in them were closely related. The highest inter-cluster distance was observed between cluster IV & VI (6154.40) indicating the genotypes included in them were diverse. Greater the distance between two clusters, wider the genetic diversity among genotypes. So, hybridization between the genotypes of the cluster IV & VI having maximum inter-cluster distance would result in high heterotic combination. Similar type of findings were earlier observed by Singh et al. (2014); Rekha et al. (2015). Cluster means: Considerable variations in mean performances of various traits among the clusters suggest that quantitative traits can reveal the existing diversity (Walle et al., 2019).Characterization of clusters was done as per the D2 distances (Table 5). Cluster III recording the highest intra-cluster distance was characterized by highest pod length (6.76cm), more number of seeds per pod (11.32) and maximum 100 seed weight (3.59gm) with comparatively larger seeds and was found to be superior among all the 6 clusters studied. The traits like plant height (41.00 cm) and seed yield per plant (3.38gm) had the highest mean value in cluster IV. Cluster V had genotypes with more number of branches/plant (2.13) and more number of pods/plant (11.73). Cluster VI was characterized by less number of days to attain 50% flowering (34.33), less number of days to attain maturity (60.67) and maximum number of pods/plant (11.73). These type of findings have been earlier confirmed by Abna et al. (2012); Razzaque et al. (2016). The results indicate that for creation of variability cross between genotypes with high cluster mean values followed by selection is required. Relative contribution of component characters to genetic divergence: According to Gupta et al. (2019) characters giving the highest contribution towards the diversity could be used further in the identification of the parents for the hybridization programme. Relative contribution of individual characters to overall genetic divergence among the genotypes was assessed by rank average method (Table 6). Among all the characters studied plant height (18.70%), number of seeds/pod (16.22%), days to 50% flowering (15.71%), number of pods/plant (15.48%) and seed yield/plant (10.02%) contributed maximum (60.65%) towards genetic divergence. Rest of the traits exhibiting divergence in order were number of clusters/plant (8.86%), hundred seed weight (8.11%), days to maturity (3.73%), pod length (1.68%) and number of branches/plant (1.49%).So characters like days to 50% flowering, plant height, number of seeds per pod, number of pods per plant and seed yield per plant should form criteria for selection of parents from distantly placed clusters for hybridization programme. These results have been in conformity with the findings of Garje et al. (2013); Jeeva and Saravanan (2017). Protein analysis: Analysis of variance for this character showed highly significant difference among all the genotypes suggesting presence of substantial variability in the material under investigation (Table 7) which will be much beneficial for the selection of breeding material (Reddy et al., 2011). The seed protein content was ranged from 16.56% in PDM-154 to 35.68% in HUM-10 (Table 8). Wide range of variation in seed protein content was previously observed by Raturi et al. (2014). Coefficient of variation was found to be low for this character (1.74). Estimates of PCV (11.53) and GCV (11.40) were moderate. High heritability (97.73%) and high expected genetic advance as a percentage of the population mean (12.63%) indicated the role of additive gene action, hence providing the scope for enhancement of this trait through selection (Jayalakshmi et al., 2019 in chickpea). Further, it showed negative correlation with seed yield (-0.018), which was earlier reported by Kumar et al. (2013).