Abstract
Four experiments were conducted for collection. evaluation, characterization and selection of superior accessions of banana. A survey covering Narsingdi, Moulvibazar, Comilla, Rangamati and Bogra areas was conducted. Field experiments were carried out at the Fruit Research Farm of Horticulture Research Centre, BARI, Gazipur, and molecular study was done at the Molecular Biology and Genetic Engineering Laboratory of Biotechnology Division of BARI during 2006-2008. It was evident from the survey that the cultivars Rangin Sagar, Champa, Amritasagar, Banglakola, Sabri, Nepali and Chini Champa were commercially cultivated in Bogra, Rangamati and Narsingdi regions. The survey also indicated that 33 types of banana were under cultivation both in homestead and commercial plots. In the hilly regions, especially in Rangamati, Champa and Banglakola were found under cultivation both in homestead and commercial plots particularly due to less infestation of diseases and insect pest. Ranking of the banana accessions based on performance under BARI farm conditions is as followes: MS028, MS031 and MS011 could be selected for earliness, MS015, MS027 and MS021 for dwarfness and MS029, MS003 and MS028 for size of bunch and yield of fruit. Based on edible portion, number of seeds per fruit, pulp softness, shelf life and firmness of finger with hand, the banana accessions MS006, MS010, MS024 and MS023 appeared better under the experimental conditions. Yield per plant had highly significant positive association, both at genotypic and phenotypic levels, with number of hands/bunch (0.584 and 0.564), number of fingers per bunch (0.549 and 0.544), length of fruit (0.680 and 0.652) and individual fruit weight. (0.652 and 0.636). The characters had high degree of significant positive association with yield per plant resulting maximum yield per unit area. Considering path analysis, the number of hands per bunch had the highest positive direct effect (3.578) on yield per plant followed by number of fingers par hand (1.758). Based on multivariate analysis, the banana accessions were grouped into 8 clusters. The cluster VIII had the highest number of accessions (8) followed by cluster VI (6) and cluster 1 (5). The accessions collected from the same location did not necessarily fall into the same cluster. The highest inter- cluster distance was found between cluster IV and II (22.55), while it was the lowest between cluster VI and I (4.77). Amplification with 5 decamer random primers generated 49 bands, of which 100% were polymorphic. The UPGMA dendrogram based on the Nei's genetic distances segregated the accessions into 11 clusters. Comparison between clustering based on morphological traits and molecular markers indicated that grouping of accessions according to the morphological traits did not match with the grouping based on RAPD markers.