The first Musa description and classification systems were based on morphological observations. However, morphological characterization similar to all phenotypic characterization suffers from the infl uence of environmental factors on the morphological/phenotypic descriptors (e.g., pseudostem color is highly affected by local fi eld conditions). However careful and purposeful selection of morphological markers may result in very precise analysis of the raised question. For example, the morphological characterization system developed by Simmonds and Shepherd (1955) for characterizing the genome composition of M. acuminata/M. balbisiana hybrids was based on 15 descriptors clearly discriminating M. acuminata and M. balbisiana wild types and assuming that all these phenotypic markers show a proportional manifestation in the hybrids. In this system 13 diagnostic characters were confi ned to the phenotype of the infl orescence. Seven markers were related to the bract of the male bud (shoulder, curling,

shape, apex, color, color fading, and scars), two were related to the male fl ower (free tepal and color of the fl ower), two were related to the female fl ower (stigma color, arrangement of ovules), two related to the peduncle and pedicels of the infl orescence, while another two characters were related to the pseudostem (color) and the petiolar canal of the leaves. Based on this scoring system the various hybrid genotypes representing different ploidy levels could be well characterized (Stover and Simmonds 1960) and fi ve main genetic groups (AA, AB, AAA, AAB, ABB) have been described for cultivated bananas (Simmonds and Shepherd 1955; Simmonds 1962). Later detailed morphological analysis of wild types revealed little variation in M. balbisiana, while nine different subspecies (ssp. banksii, ssp. burmannica, ssp. Burmannicoides, ssp. errans, ssp. malaccensis, ssp. microcarpa, ssp. siamea, ssp. truncata and ssp. zebrina) with distinct geographic distribution area is at present recognized in M. acuminata (Fig. 3-1). The drawback of morphological characterization is that the quantitative/precise measurement of the descriptors is diffi cult. Additionally most of the morphological characters possibly have polygenic inheritance and consequently minor changes in the genotype will not necessarily manifest in the phenotype of the plant,

Figure 3-1 Distribution of chloroplast and mitochondrial haplotypes in Musa. A: Distribution of the chloroplast haplotypes in M. acuminata subspecies. B: Distribution of the mitochondrial haplotypes in M. acuminata subspecies. C: Distribution of the 45S ribosomal genotypes of M. acuminata. D: Combined genotypes of the morphological subspecies listed as: chloroplast/ mitochondrial/45S ribosomal. I-V.: suggested grouping of the morphological subspecies according to their combined genotypes. The distribution area of M. acuminata ssp. truncata is within the ssp. malaccensis area (not shown).