INTRODUCTION Finger millet (Eleusine coracana (L.) Gaertn.) Sub-species coracana belongs to family Poaceae/Graminae and generally known as ragi, nachani and nagli. It is a tetraploid (2n = 36) with morphological resemblance to both E. indica (L.) Gaertn. having chromosome number 2n = 18 and E. Africana O. Byrne with chromosome number 2n = 36. It was domesticated from Ethiopia to Uganda about 5000 years ago. In India, it was introduced around 3000 years ago. Finger millet is a widely grown traditional and highly nutritious grain cereal crop (Sawardekar, 2016). It is cultivated in the semi-arid areas of Eastern and Southern Africa and South Asia, where it is a staple food for millions of poor people. It is also acknowledged for its health advantageous effects, like antidiabetic, anti-diarrheal, anti tumerogenic, anti-inflammatory, atherosclerogenic, antiulcer, antimicrobial and antioxidant properties. An enormous number of small farmers cultivate finger millet with limited water resources and in numerous nations this crop is frequently known as “poor people’s crop”. It is cultivated in the semi-arid areas of Eastern and Southern Africa and South Asia, where it is a staple food for millions of poor people. The long history of cultivation in Indian subcontinent under diverse agro ecological conditions and the associated natural and human selection has resulted in large diversity in the crop. India is often considered as secondary centre of origin for finger millet. In India, ragi is cultivated in an area of 1138.2 thousand-hectare with the production of 1821.9 mt and productivity of 1601 kg/ha majorly in Karnataka, Andhra Pradesh, Tamil Nadu, Maharashtra, Uttarakhand and Odisha (Anonymous, 2018). In India, area of finger millet stands sixth after wheat, rice, maize, sorghum and bajra (Chandra et al., 2016). As we know, population is still expanding while no significant increase in arable lands is foreseen. There is need to enhance the productivity of finger millet to feed the large population. Heterosis breeding is one of the important ways to achieve high productivity. The hybrids between to two divergent groups normally show high amount of heterosis than the hybrids between two genetically similar groups (Dwivedi et al., 1998; Melchinger, 1999). D2 analysis was generally used by many researchers in order to find more divergent lines that can be used to make hybrids. Muduli and Misra (2008) have done genetic divergence analysis in micro-mutant lines in finger millet and found divergent mutant lines. Similarly, Patel et al. (2019); Suryanarayana et al. (2019); Keerthana and Chitra (2020) also conducted D2analysis in finger millet and got divergent genotypes. In the present study, D2 analysis was used to identify divergent mutant line from M4 mutant lines of GPU 28 and KMR 204. MATERIALS AND METHODS The present study on induced mutations in finger millet was carried out at K block, GKVK, University of Agricultural Sciences; Bangalore represents Eastern Dry Agro Climatic Zone (Zone V) of Karnataka which is located at 12.9716°N latitude and 77.5946°E longitude and an altitude of 920 m above MSL. Material used for the present study comprised of M4 seeds selected for productivity per se traits form M3 generation derived from the gamma irradiation of finger millet varieties GPU 28 and KMR 204. These 58 selected mutants (31 M4 mutants of GPU 28 and 27 M4 mutants of KMR 204) were evaluated in Randomized complete block design with three replications with two checks (GPU 28 and KMR 204). Observation on days to 50 per cent flowering and days to panicle maturity were recorded on plot basis. Plant height (cm), productive tillers plant-1, fingers ear-1, finger length (cm), ear weight plant-1 (g), seed yield plant-1 (g) and 1000 - seed weight (g) were recorded on five randomly selected plants for each line in every replication. Estimation of genetic divergence was done by multivariate analysis using Mahalanobis’ (1936). D2 statistic calculated as described by Rao (1952). Contribution of each trait to the divergence, intra and inter cluster distance and cluster means were estimated in the present study. RESULTS AND DISCUSSION A. Distribution of M4 mutant lines into different clusters Diversity analysis is very crucial for plant breeders for identification of divergent genotypes which will help further in exploitation of heterosis. Mahalanobis’(1936) D2 analysis is one of the important tool to classify the genotypes into different clusters. In the present study, the M4 population of GPU 28 was classified into 7 clusters and m4 population of KMR 204 into six clusters (Table 1&2). This indicates the presence of the diversity among the mutant lines of both genotypes. In both the mutated population, cluster 1 showed maximum number of mutant lines. Cluster I of mutated population of GPU 28 has 8 lines and of KMR 204 has 12 mutant lines. Similar diversity analysis was done by Muduli and Misra (2008) in micro-mutants of finger millet and classified into different clusters. Inter- cluster and intra cluster D2 value indicates the amount of diversity among and within the clusters. Maximum inter cluster distance (392.01) was observed between cluster I and cluster IV and lowest inter cluster distance (52.73) was between cluster III and cluster IV in mutated population of GPU 28 (Table 3). In case of mutated population of KMR 204, maximum inter cluster distance (710.60) was observed between cluster II and cluster IV and lowest inter cluster distance (140.52) was observed between cluster I and cluster III (Table 4). Maximum intra cluster distance (40.78) was observed in cluster V and lowest (16.50) in cluster VI in mutated population of GPU 28 (Table 3). Maximum intra cluster distance (84.08) was found in cluster V and cluster VI has intra cluster distance is zero because of only one genotype in the cluster in the mutated population of KMR 204 (Table 4). Inter cluster distance was found to be maximum between cluster I and IV in mutated population of GPU 28 and between II and IV in case of mutated population of KMR 204. This indicates high amount of diversity between these clusters. Diversity is very important factor in exploitation of heterosis. Hybrid between diverse genotypes belongs to different cluster would be more effective than hybrid between less diverse genotypes. So, mutant lines from these diverse clusters can be used in hybridization programme in order to get high yielding hybrids or to get vast variability among the segregants. Similar results in finger millet were reported by Devaliya et al. (2017); Suryanarayana et al. (2014); Mahanthesha et al. (2017); Keerthana and Chitra (2020). B. Cluster means of different traits used under study Results indicated the significant difference among cluster means for most of the trait used under study in both the mutated population (Table 5 & 6). In case of mutated population of GPU 28, lowest value (101.17) for days to panicle maturity was observed in cluster I, highest value (8.46) was found for fingers ear-1 in cluster III, highest value for productive tillers plant-1 (5.14), highest value for finger length (8.74) and highest value for ear head weight (56.52), all were observed in cluster IV. Cluster VII had the highest mean value for seed yield plant-1 (36.39) and for 1000- seed weight (3.43). In case of mutated population of KMR 204, Cluster I had the highest mean value for the seed yield plant-1 (36.19), ear head weight plant-1 (55.98), fingers ear-1 (8.23) and productive tillers plant-1 (5.52). Cluster IV had the highest mean value for 1000- seed weight (3.43). Cluster V had the highest mean value for finger length (7.67). Mutant lines from these clusters can be used to develop high yielding cultivars. C. Trait contribution to the genetic divergence Trait contribution to the genetic divergence for both the mutated population were presented in Fig. 1 and 2. Maximum percentage of contribution to the genetic divergence was displayed by 1000- seed weight (58.40) followed by days to 50 % flowering (25.64), finger length (9.12) and days to panicle maturity (3.99) in mutated population of KMR 204. In mutated population of GPU 28, maximum percentage of contribution to the genetic divergence was displayed by days to 50 % flowering (47.10) followed by 1000- seed weight (40.43), finger ear-1 (5.16) and plant height (3.01). Similar results were reported by Patel et al. (2019) in finger millet.