Abstract
In total 98 genotypes (40 genotypes of Balam, 27 of Jesso-Balam TAPL, 21 of Kartiksail and 10 of Dhaliboro groups, respectively) of similar and duplicate named rice germplasm of Bangladesh were studied for agro-morphological, physico-chemical and molecular characters at BRRI during 2009-12. Analysis of variance of 43 morpho-physicochemical characters showed highly significant differences among the genotypes in different seasons, as well as among the genotypes of different groups for all the studied characters. The cluster analysis using Mahalanobis' D2 statistics, grouped the genotypes into 10 clusters, where no duplicate germplasm was found for 25 quantitative morpho-physicochemical characters. The highest number of genotypes (14) were constellated in clusters VI and X and the lowest (5) in cluster II. The intra- and inter-cluster distances were maximum (2.24 and 25.33) in cluster VIII and between clusters II and X and minimum (0.77 and 2.85) in cluster IX and between clusters XI and X, respectively. Again, hybridization between the genotypes from the clusters II and IX may give useful transgressive segregants with maximum heterotic vigour. Besides, the genotypes from clusters II and X, as well as II and I may be used as parents in hybridization program. It appeared from the canonical analysis that effective tiller number per hill, coleoptile length, 1000-grain weight, grain yield per panicle, secondary branch number and amylose content contributed maximum to the divergence. The UPGMA clustering method based on Dice coefficient, generated seven clusters, where (DB6 and DB2), (JBPL25 and JBPL17), (KS4 and KS3), (KS18, JBPL14 and JBPL7), (BRRI dhan50 and BR7) and (JBPL27, KS9, JBPL 19, B23 and B7), respectively were found duplicates for 19 qualitative agro-morphological traits. On the other hand, total of 514.2 alleles were detected across the 45 SSR loci for 98 genotypes. The number of alleles per locus varied from 4 to 20, with an average of 11.2 per locus. The highest number of alleles (20) and gene diversity (0.92) were found at loci RM302 and RM206, respectively. The highest PIC value (0.91) was obtained in loci RM21, RM206 and RM224. RM21, RM206, RM224 and RM584 were selected as the best markers to identify and distinguish the studied rice. The UPGMA clustering method based on Nei's genetic distance generated seven clusters, where no duplicate was found across the 45 SSR loci. As a result, it can be said that similar or even duplicate named rice germplasm were not duplicate. Besides, the Shannon's information index was highest in Balam (1.728) and lowest in Dhaliboro (1.003) groups of landraces. Analysis of molecular variance showed that the maximum percentage of variation was present among individuals within populations (85%), followed by among populations (15%) and no variation was observed within individuals. The spearman's rank correlation coefficient value was found highly significant (r=0.510 and t=2.98) for quantitative morphological and SSR diversity analysis, indicating strong association between them. The different rank values revealed that SSR diversity analysis is the most powerful method for grouping genotypes. But, the combination of morphological and molecular analyses may be the best method for studying genetic divergence. Moreover, a total of 22, 19, 11 and 5 genotypes were selected as core collections for Balam, Jesso-Balam TAPL, Kartiksail and Dhaliboro groups, respectively. The genotype B7 and B18 of Aus, genotype B8, B17 and B39 of T. Aman and genotype B31 of B. Aman for Balam, the genotypes JBPL16, JBPL21 and JBPL23 for Jesso- Balam TAPL, genotype KS19 and KS9 for Kartiksail and genotype DB7 for Dhaliboro groups, respectively may be used as parents in hybridization programs for improving respective group or developing new varieties. Finally, it can be concluded that the exclusive variability and unique feature of the traditional rice germplasm can offer a valuable gene pool for utilization in future breeding programs.