Genetic Resources and Crop Evolution

, Volume 54, Issue 7, pp 1565–1572

Molecular phylogeny of banana cultivars from Thailand based on HAT-RAPD markers


    • Department of Biology, Faculty of SciencePrince of Songkla University
  • Klaus Eimert
    • Section of BotanyGeisenheim Research Center
  • Max-Bernhard Schröder
    • Section of BotanyGeisenheim Research Center
  • Benchamas Silayoi
    • Department of Horticulture, Faculty of AgricultureKasetsart University
  • Jessada Denduangboripant
    • Department of Biology, Faculty of ScienceChulalongkorn University
    • Department of Biology, Faculty of SciencePrince of Songkla University
Research Paper

DOI: 10.1007/s10722-006-9169-2

Cite this article as:
Ruangsuttapha, S., Eimert, K., Schröder, M. et al. Genet Resour Crop Evol (2007) 54: 1565. doi:10.1007/s10722-006-9169-2


Musa acuminata Colla (AA genomes) and Musa balbisiana Colla (BB genomes) are the wild progenitors of the cultivated banana, they are highly variable in Thailand. The genetic system is relatively unknown and complicated due to interspecific hybridization, heterozygosity and polyploidy, which are common in most clones. These factors make identification of closely related banana cultivars difficult, especially when sterile. The high annealing temperature-random amplified polymorphic DNA (RAPD) technique was used to estimate the genetic relationship between 22 selected banana cultivars, utilizing 14 random primers. Phylogenetic relationship was determined by unweighted pair group method with arithmetical averages cluster analysis. The dendrogram constructed from the similarity data showed that all the 22 cultivars analysed were closely related with a narrow genetic base. There were sufficient RAPD polymorphisms that were collectively useful in distinguishing the cultivars. The dendrogram grouped all the AA, BB, AAA, AAB and ABB genomes into a major cluster. Several subgroups are recognized within the major clade. As expected, Ensete glauca Roxb. (Musaceae) and Strelitzia reginae Banks (Strelitziaceae) were clearly differentiated from the analysed edible bananas. Our study showed that RAPD markers are sufficiently abundant to classify and readily dissect genetic differences between the closely related Musa germplasm and provide a basis for the selection of parents for improvement of this germplasm.


BananaMolecular phylogenyMusaHAT-RAPDThailand



High annealing temperature-random amplified polymorphic DNA


Base pair(s)


Unweighted pair group method with arithmetical averages


Banana and plantain (Musa L. spp.) are important as cash or subsistence crops in Africa, America and Asia. They originated mainly from intra- and interspecific hybridizations between two diploid wild species, M. acuminata Colla (‘A’ genome) and M. balbisiana Colla (‘B’ genome) (Simmonds and Shepherd 1955). Intercrossing among species and subspecies has resulted in the appearance of sterility, a trait that was selected for during domestication, together with parthenocarpy and vegetative propagation (Simmonds 1995). Most current cultivars are triploid, and sometimes diploid or tetraploid. Taxonomically, banana cultivars and hybrids are classified based on ploidy analysis and a set of 15 morphological descriptors into genomic groups, differing for genome constitution and ploidy (Simmonds and Shepherd 1955). The main genomic groups are AA, AAA, AAB and ABB, although AB, AAAB, AABB and ABBB are also possible (Stover and Simmonds 1987). Closely related clones or cultivars resulting from mutations in a single genotype are allocated to so-called subgroups, characterized by specific morphological and fruit quality attributes (Simmonds 1973). The total yield potential has not yet been realized in Musa because breeding has been neglected until recently. Thailand has been ranked the third largest producer of bananas in Asia. Almost all the production is destined for the local market. There is a large number of cultivars planted in Thailand but the occurrence of local names, synonymous and homonymous, plus the high occurrence of somaclonal variation for some cultivars have limited the full knowledge of the genetic resources available. Concomitantly, Musa breeders have been able to gain some insights into the Musa genome. A better knowledge of the available genetic diversity and of the origin and genome structure of current cultivars is needed to increase the efficiency of Musa breeding.

Molecular markers that provide useful information and new insight into the classification have been available for several years and have been employed in the characterization and evaluation of genetic diversity in Musa species, including diversity for anthocyanin content and profile (Horry and Jay 1988), enzyme polymorphisms (Bonner et al. 1974; Bhat et al. 1992), rRNA spacer-length heterogenity (Lannaud et al. 1992), chloroplast DNA polymorphisms (Gawel et al. 1992), restriction fragment length polymorphisms (Jarret et al. 1992; Gawel et al. 1992; Fauré et al. 1994; Bhat et al. 1995; Carreel et al. 2002), random amplified polymorphic DNA (RAPD) (Jarret et al. 1993; Bhat and Jarret 1995; Crouch et al. 2000a, b; Pillay et al. 2000, 2001), variable number tandem repeats (Kaemmer et al. 1993), amplified fragment length polymorphism (Loh et al. 2000; Ude et al. 2002) and microsatellites or simple sequence repeats (Lagoda et al. 1998; Crouch et al. 1998; Kaemmer et al. 1997; Grapin et al. 1998).

In this paper we demonstrate the possible relationship of several important Thai banana cultivars based on a stringent RAPD technique, now called high annealing temperature (HAT)-RAPD (Sitthipron et al. 2005). We report on the use of this technique in investigating the classification and establishment of genetic relationship between the cultivars and discuss the potential of the molecular data. A better knowledge of the relationship is of great importance for the application of Musa germplasm collections, evaluation programmes and plant breeding projects.

Materials and methods

Plant material

A total of 22 Musa cultivars (listed in Table 1) were examined. Ensete glauca Roxb. ‘Kluai Naun’ and Strelitzia reginae were included as outgroup taxa, for comparison with Musa. The samples were harvested from the collection at Pak Chong Research Station of Kasetsart University, Pak Chong, Nakhon Ratchasima, Thailand, with the exception of the S. reginae leaf, which was taken from Section of Botany, Geisenheim Research Center, Geisenheim, Germany.
Table 1

The Musa accessions studied (excl. Ensete and Strelitzia) and their assumed genomic compositiona



1. Musa acuminata ‘Kluai Pa Pli Som’


2. Musa acuminata ‘Kluai Hom Jampa’


3. Musa acuminata ‘Kluai Khai’


4. Musa acuminata ‘Kluai Namthai’


5. Musa acuminata ‘Kluai Hom Thong’


6. Musa acuminata ‘Kluai Krang’


7. Musa × paradisiaca ‘Kluai Niu Mu Nang’


8. Musa × paradisiaca ‘Kluai Chin’


9. Musa × paradisiaca ‘Kluai Nam’


10. Musa × paradisiaca ‘Kluai Tamnuan’


11. Musa × paradisiaca ‘Kluai Roi Wi’


12. Musa × paradisiaca ‘Kluai Nam Kabdum’


13. Musa × paradisiaca ‘Kluai Klai’


14. Musa × paradisiaca ‘Kluai Nam Wa Saidang’


15. Musa × paradisiaca ‘Kluai Nom Mi’


16. Musa × paradisiaca ‘Kluai Som’


17. Musa × paradisiaca ‘Kluai Hak Muk Khieo’


18. Musa × paradisiaca ‘Kluai Namwa Chantaburi’


19. Musa balbisiana ‘Kluai Hin’


20. Musa × paradisiaca ‘Kluai Tip Kham’


21. Musa × paradisiaca ‘Kluai Theparod’


22. Musa balbisiana ‘Kluai Tani (Eisan)’


aMorphological classification according to Silayoi (2002)

DNA extraction

The genomic DNA of young banana leaves was extracted by using a modified microscale protocol of Wolf et al. (1999). Briefly, leaf tissue was rinsed with deionized water and blotted dry prior to be placed into an Eppendorf tube with 10–20 mg insoluble polyvinylpolypyrrolidone (SIGMA) and a total of 900 μl UEB. Then it was directly homogenized in the tube, vortexed hard and incubated for 30 min at 65°C. Debris was pelleted by centrifugation and the supernatant was extracted twice with phenol–chloroform–isoamylalcohol (25:24:1). DNA was precipitated with ethanol in a high salt concentration (5.0 M NaCl). The pellet was washed twice with 70% ethanol, dried and re-suspended in 50 μl of TE buffer. On average, 100 mg plant tissue yielded 6–27 μg clean (as judged by the OD260/280 ratio) DNA.

Random amplified polymorphic DNA-PCR

A pre-screening of 114 RAPD primers (University of British Columbia, Vancouver, Canada; RAPD Primer Sets No. 1 and No. 4) was performed using 24 samples, only 14 RAPD primers (UBC 70, 77, 81, 82, 83, 86, 87, 91, 96, 302, 303, 304, 308, 335) showing strong amplification and good reproducibility were selected.

The protocol for the HAT-RAPD analysis was adapted from Eimert et al. (2003). PCR was performed in a volume of 15 μl containing 1 ng/μl DNA, 1.5 mM MgCl2, 0.2 mM 4 deoxynucleotide triphosphates, 1 μM primer and 0.1 U/μl Taq DNA polymerase (MBI FERMENTAS GmbH). The amplification protocol was as follows: an initial denaturation step at 95°C for 5 min, followed by 38 cycles of 1 min at 95°C, 1 min at 45°C, 2 min at 72°C and 5 min at 72°C. Amplifications were performed in a Primus HTD thermocycler (MWG AG Biotech).

At least two PCR amplifications were performed for each sample with RAPD primers to evaluate the reproducibility of the bands obtained. DNA amplification fragments were separated in a 1.3% agarose gel using 1 ×  TBE buffer, and stained with ethidium bromide.

RAPD analysis

Only distinct, reproducible, well-resolved fragments were scored as present or absent band for each of the RAPD primers with the 24 accessions. The dendrogram was constructed by cluster analysis based upon the unweighted pair group method with arithmetical averages (UPGMA) of the BioNumerics software package, version 3.0 (Applied Maths BVBA, Sint-Martens-Latem, Belgium).


The aim of the present study was to produce RAPD markers for identification of several important edible banana cultivars in Thailand. We were able to successfully establish the HAT-RAPD protocol for Thai banana accessions belonging to the section Eumusa in the genus Musa. Clear amplified polymorphic DNA products were obtained from the screening of six genotypes (of AA, BB, ABB, AAB and AAA composition) with 114 RAPD primers (data not shown). These allowed for a selection of 14 primers. Reproducibility was checked by amplifications of each individual DNA sample and only highly reproducible markers were analysed (data not shown). The reproducibility of the amplification pattern was very consistent; identical RAPD patterns were obtained under the same amplification conditions for at least two replicates with each primer. A list of primers yielding reproducible patterns is available on request.

Random amplified polymorphic DNA technique was optimized for Thai bananas and produced amplified fragments varying from 400 to 2,000 bp in size for the different primers to be scored across 22 cultivars. Each primer yielded a wide array of strong and weak bands (Fig. 1). However, only the data of the 14 primers that gave reproducible product formation were included in the statistical analysis. Figure 1 illustrates the typical level of polymorphisms observed among the Thai banana cultivars, which was generally low.
Fig. 1

Random amplified polymorphic DNA profiles of DNA from 22 banana cultivars (+ two outgroups) using primer UBC 82. Lanes are: 1 K. Chin, 2 K. Roi Wi, 3 K. Tamnuan, 4 K. Hak Muk Khieo, 5 K. Pa Pli Som, 6 K. Som, 7 K. Tani (Eisan), 8 K. Nam, 9 K. Hom Thong, 10 K. Nam Wa Saidang, 11 Ensete glauca Roxb., 12 K. Klai, 13 K. Krang, 14 K. Nom Mi, 15 Strelitzia reginae, 16 K. Hom Jampa, 17 K. Nam Kabdam, 18 K. Namwa Chantaburi, 19 K. Khai, 20 K. Hin, 21 K. Theparod, 22 K. Namthai, 23 K. Niu Mu Nang, 24 K. Tip Kham, M molecular-weight markers

The genetic similarity coefficients for the 22 clones ranged from 19 to 98 (data available upon request). The UPGMA analysis made it possible to cluster all of the genotypes of this study. The dendrogram (Fig. 2) showed a clear distinction into a major and a minor cluster. The major cluster could be divided into three subgroups. The first subgroup with about 56% similarity was composed of Kluai Tamnuan, Kluai Nam, Kluai Hom Thong, Kluai Chin, Kluai Roi Wi and Kluai Pa Pli Som. Within this subgroup, the closest relationship at approximately 81.71% similarity occurred between Kluai Chin and Kluai Roi Wi. Kluai Nam Wa Sai Dang, Kluai Nom Mi, Kluai Som, Kluai Tani (Eisan), Kluai Hak Muk Khieo and Kluai Klai were composing the second subgroup with approximately 64% similarity. Within this subgroup, the hybrid cultivars named Kluai Tani (Eisan) paired with Kluai Som at the highest percentage of similarity value of 95.21. The third subgroup was composed of all AA genomes, which originated from M. acuminata, Kluai Krang, Kluai Hom Jampa and Kluai Khai. Kluai Krang clustered with Kluai Hom Jampa at 78.55% similarity. The remainder: Kluai Hin, Kluai Theparod, Kluai Nam Kabdum, Kluai Namwa Chantaburi, Kluai Nam Thai, Kluai Niu Mu Nang and Kluai Tip Kham were categorized into the minor cluster. Kluai Nam Kabdum clustered with Kluai Namwa Chantaburi at the closest relationship of 98% similarity. Outgroups Kluai Naun and S. reginae were clearly distinct from the major and minor clusters and each other.
Fig. 2

Dendrogram constructed by cluster analysis (UPGMA) of RAPD data of the 24 accessions with 14 primers


This study was conducted to verify the classification and to establish the genetic relationship between the cultivars in a sample of 22 Thai banana cultivar germplasm, based on the RAPD marker system that makes use of 14 arbitrary primers. The outgroup species (E. glauca Roxb. and S. reginae) clearly formed distinct clusters separate from the cultivated bananas, most of which are of hybrid origin between M. acuminata and M. balbisiana.

Until recently, only morphological and agronomic characters were extensively used to assess genetic relationship among Musa (Ortiz 1997; Ortiz et al. 1998; Karamura 1998). Morphological markers can be monitored visually without specialized biochemical or molecular techniques. Morphological traits that are controlled by a single locus can be used as genetic markers, provided their expression is reproducible over a range of environments. Besides the environment, the expression of such markers can also be altered by epistatic and pleiotropic interactions. The number of morphological markers is very limited; their alleles interact in a dominant–recessive manner, thereby making it impossible to distinguish the heterozygous individuals from homozygous individuals. Often these do not reflect genetic relationship because of interaction with the environment, epistasis and the largely unknown genetic control of the traits (Smith and Smith 1989). This has also been observed and shortly discussed as a possible reason for the discrepancy of the systematics in banana based on morphology vs. molecular markers (De Langhe et al. 2005). DNA markers have provided valuable tools in various analyses, ranging from phylogenetic analysis to the positional cloning of genes and predicting hybrid performance (Tenkouano et al. 1999a, b). DNA markers are not influenced by the environment or developmental stage of a plant, making them ideal for genetic relationship studies.

Although RAPD-PCR has been successfully used for identification of varieties or cultivars in various plant species (Hu and Quiros 1991; Koller et al. 1993; Schnell et al. 1995; Wolf et al. 1999), problems have been reported concerning the sensitivity to experimental conditions and the reproducibility of results (Penner et al. 1993; Williams et al. 1993; Paul et al. 1997). Most of these problems seem to be due to mis-priming of the short decamers at the relatively low temperatures common in RAPDs and competition between different DNA fragments for amplification (Williams et al. 1993; Halldén et al. 1996). In most cases an annealing temperature of about 35°C is used. At these temperatures even slight changes of the reaction conditions may result in different PCR pattern. In this present study, primers with a higher G/C content and HAT-RAPD technique were used to further minimize the possibility of mis-priming (Eimert et al. 2003).

The conventional classification of banana genotypes into distinct genome combinations by Simmonds and Shepherd (1955) is based on their morphological similarity to M. acuminata Colla or M. balbisiana Colla. The cultivars examined in this study did not cluster accordingly to their hypothetical genetic homologies. For example, cultivars designated as having AAA genomic constitution (Kluai Hom Thong, Kluai Pa Pli Som, Kluai Krang, Kluai Hom Jampa, Kluai Khai and Kluai Namthai) fell in separate clades, rather than clustering together. However, cultivars with two different putative genotypes, e.g. Kluai Tamnuan (AAB) and Kluai Hom Thong (AAA), clustered together (Fig. 2). A close examination of the study by Simmonds and Shepherd (1955) resulted in several observations. First, their method of classifying Musa cultivars did not have a genetic basis due to the difficulty in carrying out conventional breeding. Since the cultivated bananas are highly sterile, crossing experiments are severely impaired. Secondly, the history of banana cultivars is mostly unknown, since many have been cultivated since ancient times. Moreover, it is not known whether a cultivar recognized by the same morphological characters has arisen more than once or whether cultivars in a different region have the same genetic origin, nor what the effect of random genetic drift is over the long period of cultivation. The picture is further complicated as bananas have been exported outside their place of origin since ancient times. Current classification based on morphological characters and deduced genotypes based on certain agronomic characters are therefore not always reliable. Phenotypic characters, such as sugary or starchy fruits, may not be due to simple euploidic difference (AAA, AAB or ABB). Different phenotypes could result from allelic differences in single or multiple genes. The difficulty involved in the identification of banana cultivars, which are mostly sterile, therefore highlights the need for a DNA marker system for classification (Loh et al. 2000).

In the present study, all banana cultivars analysed were collected from Thailand and planted in the same field. The fact that banana cultivars in the different communities are referred to different names, may suggest that they are not genetically uniform and that there is a need for a comprehensive analysis of genotypes using molecular methods to put the nomenclature on a firm footing. This is because being clonal in nature and highly sterile due to triploidy, a cultivar should be genetically identical. This work has provided new information regarding the genetic relationship among taxa of the section Eumusa by using RAPD analysis. RAPD data showed that the edible Thai bananas used in this study are closely related with a narrow genetic base. Additionally, the low level of DNA diversity contrasts with the high level of morphological variability present in these plants. A probable reason for this discordance could be the failure of the RAPD primers to anneal to areas of the genome responsible for the morphological variation, resulting in non-random sampling of the genome and an insufficient number of polymorphisms (Pillay et al. 2001). Possibly, the close genetic similarity and the large number of shared fragments could also suggest a common origin and subsequent divergence by mutations. In any case, the RAPD polymorphisms are collectively useful in distinguishing the cultivars and specific SCAR primers could be developed if necessary.

In conclusion, this research demonstrated RAPD markers to be a useful tool to detect DNA polymorphisms to examine genetic relationship in Thai banana germplasm. The advantages of the RAPD technique include its speed, low DNA template requirements and technical simplicity. It makes a convenient tool for detecting genetic variation within this germplasm. Furthermore, it helps in the identification of duplications among accessions in the field and in tissue culture germplasm banks. It can also be used in monitoring of genetic stability of tissue culture material (i.e. somaclonal variation) and identification of trait markers for use in cross and mutation breeding programs. DNA profiling would also be important for policing plant patents and for legal protection of new bred cultivars in asexual Musa crops (Kaemmer et al. 1997).


The authors are most grateful to the Royal Golden Jubilee Ph.D. Programme and Graduate School of Prince of Songkla University, Thailand for financial support for this study. We also thank the Section of Botany, Geisenheim Research Center, Geisenheim, Germany for the use of facilities, the Department of Horticulture and Pak Chong Research Station of Kasetsart University, Thailand for sources of plant materials.

Copyright information

© Springer Science+Business Media, Inc. 2007