Banana Classification: Botanical and Commercial Systems

Bananas are classified through two overlapping but distinct systems: a botanical system built around genome composition and a commercial system organized by how the fruit is used. The botanical system groups cultivars by how much of their DNA comes from two wild ancestor species, while the commercial system sorts them into categories like “dessert banana” and “plantain” based on starch content, cooking requirements, and market expectations. These two systems often align, but not always, and the gaps between them explain a surprising amount of confusion about what a banana actually is.

The Two Wild Ancestors Behind Nearly All Cultivated Bananas

Almost every banana you encounter traces back to two wild species in the genus Musa: Musa acuminata and Musa balbisiana. Wild M. acuminata originated in Southeast Asia and contributes sweetness, softer flesh, and thinner skin to the cultivars descended from it. Wild M. balbisiana, which spread across a range from India to southern China, contributes starchiness, hardiness, and tolerance to drought and other environmental stresses. Both wild species produce small, seedy, barely edible fruits. The edible bananas people actually grow and eat arose from hybridization events between and within these two species, combined with mutations that tripled their chromosome sets and made them seedless.

Genetic studies using molecular markers consistently separate banana accessions into clusters that correspond to their ancestry from these two species. Analyses using AFLP markers, for example, produce clusters that closely match genome composition, clearly distinguishing pure acuminata types from pure balbisiana types and their various hybrids.¹ Chloroplast and mitochondrial DNA work has further shown that a single M. acuminata cytotype dominates the maternal lineage of most hybrid cultivars, while a single M. balbisiana cytotype contributed to the formation of interspecific hybrids.²

The Genome Group System

Botanists classify banana cultivars using a letter-based shorthand where “A” represents a genome set from M. acuminata and “B” represents one from M. balbisiana. A diploid banana with two sets of acuminata chromosomes is labeled AA. A triploid with three sets of acuminata chromosomes is AAA. Mix in balbisiana contributions and you get groups like AAB (two acuminata sets, one balbisiana) or ABB (one acuminata, two balbisiana). Most cultivated bananas are triploids.³

This system matters because genome composition predicts a lot about how a banana behaves. AAA bananas tend to be sweet dessert types. AAB bananas include both dessert and cooking varieties depending on the subgroup. ABB bananas are typically starchy cooking types that also happen to be more drought-tolerant than those with only acuminata genomes.⁴ The presence of balbisiana genome sets generally pushes a cultivar toward higher starch, firmer texture, and greater environmental resilience.

Within each genome group, cultivars are further sorted into subgroups based on shared morphological and genetic traits. The AAB genome group alone contains subgroups as diverse as Plantain, Silk, Iholena, and Pisang Raja, each with distinct culinary profiles and regional significance.⁵ These subgroup names carry more meaning in practice than the genome group letters alone, because two AAB bananas from different subgroups can be as different in flavor and use as a Cavendish and a plantain.

Why Triploidy Matters

The dominance of triploid bananas in cultivation is not accidental. Having three sets of chromosomes instead of two creates a reproductive glitch: the chromosomes cannot pair up evenly during cell division for seed production, so most triploid bananas are functionally sterile. They produce fruit without pollination, through a process called parthenocarpy, and that fruit develops without seeds. Research into the molecular basis of parthenocarpy in Musa has identified a network of candidate genes likely involved in triggering seedless fruit development, including genes related to hormone signaling and cell expansion.⁶

Seedlessness is the trait that made bananas a viable commercial crop. Nobody wants to eat a fruit packed with hard, pea-sized seeds, which is what wild bananas look like. But it also means that every banana plant of a given cultivar is a genetic clone, propagated by cuttings or tissue culture rather than by seed. This uniformity is the source of both the banana’s commercial appeal and its greatest vulnerability.

When the Genome Letters Lie

The genome group system was originally built on morphological observation and simple molecular tests. Modern whole-genome sequencing has revealed that reality is messier than the neat letter codes suggest. A cultivar classified as “AB” based on traditional methods may actually carry small chromosomal segments that are AA rather than AB. A study using high-resolution genome sequencing found that the cultivar Kunnan, conventionally labeled AB, has a 2.4-megabase AA fragment on chromosome 9 and a roughly 6-megabase AA fragment on chromosome 8. The cultivar Pelipita, classified as ABB, carries entire chromosomes and large chromosome fragments that are BBB rather than ABB.⁷

These deviations happen because banana genomes have undergone extensive recombination and large-scale structural rearrangements over centuries of vegetative propagation. Ancient hybridization events swapped chunks of chromosomes between the A and B genomes, creating mosaic patterns that simple classification schemes cannot capture. For most growers and consumers this is an academic detail, but for breeders trying to introduce disease resistance or other traits, knowing the actual chromosome-level composition of a cultivar is critical.

The Domestication Trail Out of New Guinea

Banana domestication was not a single event. Recent genomic analyses have mapped the ancestry mosaics of cultivated bananas and traced the earliest steps to New Guinea, where wild M. acuminata subspecies were first crossed, both with each other and, unexpectedly, with Musa schizocarpa, a separate New Guinean species. Plants bearing these initial introgressions were then transported westward into Southeast Asia, where further hybridization with additional M. acuminata subspecies created the enormous diversity of cultivars found across the region today.⁸

That research identified at least eleven genetic pools that contributed to modern cultivars, including two contributors whose wild progenitor populations have not yet been definitively matched to known subspecies. One of these may be M. acuminata ssp. halabanensis, and another is still uncharacterized. Field studies in mainland Southeast Asia have documented the full spectrum from strictly wild forms, like M. yunnanensis and M. acuminata ssp. burmannica, through semi-managed plants to fully domesticated cultivars, illustrating how domestication unfolded in stages over thousands of years.⁹

The Commercial Split Between Dessert and Cooking Bananas

The commercial world cares less about genome letters and more about what you do with the fruit. The two broadest categories are dessert bananas, eaten raw when ripe, and cooking bananas, which include plantains and other starchy types that are fried, boiled, or roasted. This is the classification most consumers interact with, and it maps roughly but imperfectly onto genome groups.

What drives the difference is starch. Plantains and cooking bananas carry more starch on a dry-weight basis than dessert bananas do. In one analysis of Colombian varieties, plantain flour averaged about 86.5% starch on a dry basis, while dessert types averaged around 82%.¹⁰ That gap may sound small in percentage terms, but it translates into a very different eating experience. Dessert bananas convert their starch to sugar rapidly as they ripen, becoming soft and sweet. Plantain starch granules are structurally more resistant to enzymatic breakdown, which is why even a ripe plantain remains firmer and less sweet than a ripe dessert banana.¹¹

Ripening also proceeds differently. In cooking banana hybrids, starch content can drop from around 750 grams per kilogram to below 100 at advanced ripening stages, but the sugar that accumulates tends to be non-reducing sugars rather than the fructose and glucose that make dessert bananas taste sweet.¹² In practical terms, this means a very ripe cooking banana still does not taste like a Cavendish. It is softer than when green, but the flavor profile remains different.

Cavendish and the Monoculture Problem

When people outside the tropics think of “a banana,” they are almost always picturing a Cavendish. This subgroup within the AAA genome group dominates the international trade, accounting for roughly half of global banana production and approximately 95% of banana exports.¹³ Cavendish replaced the Gros Michel cultivar after Panama disease (Fusarium wilt Race 1) devastated Gros Michel plantations in the mid-twentieth century. Cavendish happened to be resistant to that particular race of the fungus, so the global industry simply switched.

Now history is repeating itself. A new strain, Fusarium oxysporum f. sp. cubense Tropical Race 4 (TR4), attacks Cavendish. There is no commercially viable Cavendish cultivar resistant to TR4, and because every Cavendish plant is genetically identical, the entire export banana industry is at risk. Researchers have been screening diverse genome groups for resistance, finding that wild AA diploids harbor the highest proportion of resistant genotypes (about 45%), followed by AAA triploids (21%) and AAB triploids (18%).¹⁴ Wild relatives like Calcutta 4 and DH Pahang have emerged as especially promising sources of resistance genes. Most plantain cultivars within the AAB group have also shown high resistance to TR4, with only a few French-type cultivars proving susceptible.⁵

The Cavendish monoculture illustrates a tension between commercial classification and botanical diversity. The commercial system optimized for one cultivar that shipped well, ripened uniformly, and looked the same in every grocery store. The botanical classification system, with its hundreds of genome groups and subgroups, contains the diversity that could solve the disease problem, but that diversity has been commercially sidelined for decades.

The Broader Family Tree

Bananas belong to the family Musaceae, which contains three genera: Musa, Ensete, and Musella. Phylogenetic analyses consistently show the family is monophyletic (all members share a single common ancestor), with Musella as sister to Ensete rather than to Musa.¹⁵ Musella is a single-species genus found in southwestern China, and its close relationship to the African genus Ensete suggests it is a highly specialized member of that lineage rather than a distinct group warranting its own genus.

Within Musa itself, plastome (chloroplast genome) analyses divide the genus into two major clades corresponding to different base chromosome numbers: one group with a base number of 11 and another with base numbers of 10, 9, or 7.¹⁶ Most edible bananas fall in the chromosome-11 clade, but knowing that Musa contains these structurally distinct lineages matters for breeding programs trying to cross different species.

Fe’i Bananas and the Pacific Outlier

Not all cultivated bananas fit the acuminata/balbisiana framework. Fe’i bananas, found in Papua New Guinea and other Pacific islands, were domesticated independently from a completely separate lineage within Musa. The best-studied Fe’i banana, Karat (Musa troglodytarum), is an autotriploid, meaning all three of its chromosome sets derive from the same species group rather than from a cross between acuminata and balbisiana. Karat fruits are immediately recognizable by their deep yellow-orange flesh and the pink sap they produce when cut, traits linked to exceptionally high levels of beta-carotene.¹⁷

Fe’i bananas also differ physiologically from mainstream cultivars in that their fruit is non-climacteric, meaning it does not undergo the post-harvest ripening burst driven by ethylene that Cavendish and other commercial bananas rely on. This makes Fe’i bananas poorly suited to the global supply chain, which depends on picking green and ripening in controlled-atmosphere rooms. But their nutritional profile, particularly the beta-carotene content, has made them a focus of interest for addressing vitamin A deficiency in Pacific island communities.

The Ripening Scale Used in Trade

Commercial banana handling uses a standardized color-based ripening scale, most commonly applied to Cavendish (Grand Naine). This scale typically runs from stage 1 (fully green) through intermediate stages tracking the shift from green to yellow, up to stage 7 (yellow with brown speckling) and stage 8 (skin turning black).¹⁸ Retailers usually want to receive bananas at stage 3 or 4 and sell them at stage 5 or 6. The scale was developed for Cavendish and does not translate perfectly to other cultivars. A red banana or a Burro banana may not follow the same color progression at all, which creates logistical headaches when diverse banana types enter the same distribution chain.

Molecular Tools and the Classification Headaches They Reveal

Classifying banana cultivars accurately has long been difficult. Morphological descriptors, the traditional method, are influenced by growing conditions and overlap between groups. Molecular markers have improved matters but also revealed how tangled banana genetics really are. Studies using ISSR and SRAP markers, for instance, have found that placing cultivars into the correct genome group remains problematic even with molecular approaches, because hybrid cultivars may show marker profiles that do not cleanly match any single expected group.¹⁹

Whole-genome profiling technologies like DArT (Diversity Arrays Technology) have provided finer resolution, and clustering results from these methods are generally consistent with accepted taxonomy at the subgroup level.²⁰ But the chromosome-level mosaics described earlier mean that any classification system assigning a single genome label to a cultivar is, at best, an approximation. Genebank databases like the Musa Germplasm Information System (MGIS) now store passport data, morphotaxonomic descriptors, ploidy information, and digital images for thousands of accessions, providing researchers with a more complete picture than any single classification scheme can offer.²¹

How Farmers Classify Bananas on the Ground

Neither the genome group system nor the international trade categories fully capture how bananas are classified by the people who grow and eat the widest diversity of them. In Papua New Guinea, where banana diversity is enormous, farmers use their own classification systems based on use, growth habit, and environmental requirements. Research in the PNG lowlands documented three locally recognized banana subgroups at study sites: a “kalapua” subgroup of cooking bananas, a separate group of other cooking bananas, and a dessert banana group. Farmers plant kalapua cultivars and other cultivars in separate gardens because they differ in growth rate and climate tolerance. Nutritional analysis showed that kalapua cultivars are comparatively low in nutrients other than carbohydrates, reinforcing the practical logic of the local classification.²²

These local systems sometimes capture functional distinctions that formal botanical classification misses. Two cultivars in the same genome group might end up in different local categories because one tolerates waterlogging and the other does not, or because one produces fruit year-round while another is seasonal. Conversely, cultivars from different genome groups may sit side by side in a local “cooking banana” category because, from the farmer’s perspective, they serve the same purpose. The tension between these perspectives is not a flaw. It reflects the fact that classification systems are tools built for specific purposes, and a system designed to track evolutionary relationships will not always align with one designed to organize a kitchen or a farm.