| Based on morphological varition 70 and 94 genotypes in two experiments were grouped
in 8 and 9 clusters, respectively. Nearly threefourth genotypes could be clustered in first two
groups. Thus majority of the genotypes appeared to be more or less similar for these two
characters. However, with in a cluster different grades of expression for remaining characters were also noted. Tocher method : Using generalised distance (square root of D2) between genotypes for a set of charcters, the 70 genotypes were grouped into nine clusters, whereas, 94 genotypes were grouped into eleven clusters. In both the experiments first two groups were largest and accounted for almost 73 and 64 percent genotypes, respectively. This envisaged that most of the genotypes grouped in these clusters were genetically close to each other and the apparent wide genetic diversity in both the experiments was due to 19 genotypes (exp. I) and 16 parents and 18 F1 hybrids (exp. II) scattered over remaining groups. There were 3 single genotype clusters in both experiments. These genotypes were extraordinary for one or more characters which made them so divergent. Canonical root analysis : -Clusters were formed using first two canonical variates (lambda1 and lambda2) which supplied best linear functions and contributed maximum to total variation. Both lambda1 and lambda2 accounted for 61.5 and 55.9 percent of the variability in experiment I and II, respectively. Eight clusters could be recognised in experiment I and nine in experiment II. Majority of the genotypes were grouped in first two groups. On over all basis, grouping pattern using canonical roots revealed 65.7 and 77.7 percent resemblence with that of Tocher's method in experiment I and II, respectively, as majority of the genotypes were grouped similarly in both cases. The most divergent groups in Tocher's method were apparent in canonical root analysis also. Discrepencies regarding grouping pattern by two methods are expected as two canonical vectors did not explain total variability. Contrarily, such resemblence between metroglyph analysis and Tocher method and metroglyph analysis and canonical root analysis was of the order of 28.6 and 51.1 percent and 27.1 and 47.9 percent in experiment I and II, respectively. Thus in general, metroglyph analysis showed less resemblence with other two methods, particularly in experiment I, whereas in experiment II such a change was not spectacular. Therefore, it seems that grouping pattern in first two groups is mutually exclusive. All these techniques suffer from one or more short comings. Metroglyph analysis is based on two characters explaining maximum variability and therefore subjective. So is true with canonical root analysis also. Although, D2 statistic between any pair of population amounts to a quantitative measure of genetic divergence, yet the grouping pattern is arbitrary, subjective and changeable under the influence of environment (SINGH & GUPTA 1979). Therefore, use of various methods to confirm grouping pattern more objectively has been advocated (VAIRAVAN et al., 1973 : Jain et al., 1978). If breeders requriment is fulfilled only by broad classification, metroglyph analysis being simpler offers a suitable alternative (CHANDRA 1976). In present study also, grouping following metroglyph analysis revealed sizeable similarity over two years as 15 out of 22 parents revealed almost same grouping pattern. References ANDERSON, E. 1957. Semigraphical methods for analysis of complex problems. Proc. Nat. Acad. Sci. Var., 43 : 923-927. ARUNACHALAM, V. 1981. Genetic distance in plant breeding. Indian. J. Genet., 41 : 226-236. CHANDRA, S. 1976. Comparision of Mahalanobis's method and Metroglyph technique in the study of divergence of Linum usitatissimum. L. germplasm collection. Euphytica., 26 : 141-148. JAIN, P.P., R.K. SINGH & S.C. SHARMA 1978. Comparative measure of genetic diversity in wheat. Proc. 5th Intern. Wheat Genet. Symp. (New Delhi)., 663-670. RAO, C.R., 1952. Advanced Statistical Methods in Biometric Research. John Wiley and Sons. Inc. New York. SINGH, S.P. & P.K. GUPTA, 1979. Genetic divergence in pearl millet. Indian. J. Genet., 39 : 210-215. VAIRAVAN, S., E.A. SIDDQUI, V. ARUNACHALAM & M.S. SWAMINATHAN 1973. A study on the nature of genetic divergence in rice from Assam and North East Himalayas. Theor. Appl. Genetics, 43 : 213-221. |
| <-- Back |