INTRODUCTION Garden pea (Pisum sativum var. hortense L.), belongs to family leguminosae (Fabaceae), and is extensively grown and popular vegetable crop. It is the second most important food legume worldwide after Phaseolus vulgaris (Taran et al., 2005). It is a rich source of protein, amino acids and carbohydrates. Peas are highly nutritive and are rich source of digestible proteins (7%), along with carbohydrates and minerals. It is used as a fresh vegetable or in soup, canned, processed or dehydrated. Worldwide garden pea occupies an area of 2.66 million hectares, production of 20.67 million tonnes with a productivity of 7.75 t/ha. In India, area of 0.053 million hectares, production 5.345 million tonnes with a productivity of 10.08 t/ha. In Bihar, it occupies an area 10,510 ha with a production of 66,360 tonnes and the productivity is 6.31 t/ha (Anonymous, 2017). Although, it is cultivated in different regions of the country and is one of the preferred winter vegetables, but the average green pod yield is quite low in Bihar (6.31t/ha) when compared with National (10.08t/ha) and world average (7.5 t/ha). Hence the investigation was carried out to identify high yielding genotypes for Bihar condition. The variability between different cultivars of a crop species is known as genetic diversity. Variability differs from diversity in such a way that variability shows observable phenotypic differences whereas diversity may or may not having observable phenotypic differences, latter may or may not have such an expression. The method of surveying hereditary difference is the D2 measurement proposed by Mahalanobis (1936). In this method, forces of differentiation at two levels (intra and inter cluster levels) are screened out, and thus play an effective part in the selection of genetically divergent parents for utilization in any hybridization programme (Singh 1983). Keeping these facts in view, the present investigation was carried out with the objective to analyse genetic diversity among the genotypes of garden pea. MATERIALS AND METHODS The experimental material for the present study comprised of 28 genotypes of vegetable pea (Pisum sativum var. hortense L.). These genotypes were evaluated during rabi season at the Experimental farm of the Department of Horticulture (Vegetable and Floriculture), Bihar Agricultural College, BAU, Sabour, Bhagalpur, Bihar during 2018-19. The experiment was laid out in complete randomized block design with three replications. The pea seeds were sown at a spacing of 30 cm x 10 cm during the first week of October. Recommended package of practices were followed for healthy growth of the crop. The observations were recorded on randomly taken five plants of each genotype in each replication followed by computing their means for the horticultural and quality traits. The data were statistically analysed as per the standard procedure for analysis of variance (Panse and Sukhatme 1954). Using D2 values, different genotypes were grouped into various clusters following Tocher's method as suggested by Rao (1952). RESULT AND DISCUSSION The analysis of variance shown that mean sum of squares due to genotypes were significant for all the growth parameters, yield contributing traits, and quality characters, indicating the presence of sufficient genetic variability in the genotypes. On the basis of D2 values, 28 genotypes of garden pea were arranged into nine clusters following Tocher's procedure (Rao, 1952) and also represented in dendrogram (Fig. 1). Among different clusters, cluster I was the largest one. Singh and Mishra (2008); Katkani et al. (2022) also reported cluster I as the largest one. Out of the 9 clusters of 28 genotypes, cluster I comprised of maximum 15 genotypes (VM-10, Peas TSX-10, EC-507771, Pusa Prabhat, IC-552770, Muze-02, EC-412882, Azad P3, IC-269571, AP-3, VM-12, P-3771, Muze-01, P-3824 and Buxe-03) followed by cluster II with 6 genotypes (Badshah-10, Taj-C3, IC-109696, ADU-12, Same-04 and Arkel) and the remaining clusters namely III (EC-598559 ), IV (Punjab -89), V (NBR-Ruchi), VI (EC-269571), VII (VM-11), VIII (Haze-02), IX (Nirali) were monogenotypic i.e., containing one genotype (Table 1). Different clustering patterns in garden pea were also obtained by earlier workers Siddika et al. (2014), Georgieva et al. (2016); Katkani et al. (2022); Singh and Mishra (2008); Ahmed et al. (2021). The average intra-cluster distance ranged from 0 to 324.27 with the highest in cluster I (324.27) followed by cluster II (239.4). The clusters III, IV, V, VI, VII, VIII and IX were constituted by a single genotype each and hence, their intra-cluster distance was zero. The inter-cluster distance ranged from 118.51-3590.62. The maximum inter-cluster genetic divergence was observed between clusters II and IX (3590.62) followed by clusters III and clusters IX (2542.12), clusters II and VII (2193.85) and clusters II and VI (1902.16) (Table 2). This clearly suggests the presence of sufficient amount of genetic diversity among the garden pea genotypes. Since the intra-cluster distance was low, the chances of getting good recombinants by hybridization between parents within cluster would be low. Therefore, it is necessary to attempt hybridization between genotypes falling under different clusters based on inter-cluster distance. A wide range of inter-cluster genetic distance among the different clusters of pea genotypes have also been reported by Singh et al. (2008); Georgieva et al. (2016); Khan et al. (2017); Muthuselvi and Shanthi (2013); Kumar and Kumar (2016) also reported. Maximum number of transgressive segregants could be obtained in a hybridization programme involving genotypes of cluster II and IX as parents. Cluster means for different traits showed substantial differences among the clusters for each trait (Table 3). The cluster IV superior for lower node number at which at which first flower appeared, seeds per pod and shelling (%). Cluster V found superior for early maturity manifested by days to first picking and also for pod length and high TSS. Cluster VI Showed maximum mean values for number of primary branches. Cluster VII exhibited highest mean values for five traits namely days to first flower, days to 50% flowering, minimum internodal length, ascorbic acid and total sugar, whereas cluster VIII for maximum plant height, pods per plant and pod yield per plant. Cluster IX showed maximum mean values for nodes per plant and protein. These findings are in accordance with Brahmaiah et al. (2014). The contribution of individual characters to divergence has been worked out in terms of number of times it appeared first (Table 4). Protein (38.62%) contributed maximum towards genetic divergence followed by ascorbic acid (38.10%), total sugar (12.17%) and no. of primary branches (3.97%). Variable contribution of different plant growth and yield characters to genetic distance have also been reported by Georgieva et al. (2016); Gupta et al. (2017) in garden pea.