Polysomic Inheritance Research Articles

Defining autopolyploidy: Cytology, genetics, and taxonomy.

Autopolyploidy is taxonomically defined as the presence of more than two copies of each genome within an organism or species, where the genomes present must all originate within the same species. Alternatively, "genetic" or "cytological" autopolyploidy is defined by polysomic inheritance: random pairing and segregation of the four (or more) homologous chromosomes present, with no preferential pairing partners. In this review, we provide an overview of methods used to categorize species as taxonomic and cytological autopolyploids, including both modern and obsolete cytological methods, marker-segregation-based and genomics methods. Subsequently, we also investigated how frequently polysomic inheritance has been reliably documented in autopolyploids. Pure or predominantly polysomic inheritance was documented in 39 of 43 putative autopolyploid species where inheritance data was available (91%) and in seven of eight synthetic autopolyploids, with several cases of more mixed inheritance within species. We found no clear cases of autopolyploids with disomic inheritance, which was likely a function of our search methodology. Interestingly, we found seven species with purely polysomic inheritance and another five species with partial or predominant polysomic inheritance that appear to be taxonomic allopolyploids. Our results suggest that observations of polysomic inheritance can lead to relabeling of taxonomically allopolyploid species as autopolyploid and highlight the need for further cytogenetic and genomic investigation into polyploid origins and inheritance types.

Inference of Polyploid Origin and Inheritance Mode from Population Genomic Data.

Whole-genome duplications yield varied chromosomal pairing patterns, ranging from strictly bivalent to multivalent, resulting in disomic and polysomic inheritance modes. In the bivalent case, homeologous chromosomes form pairs, where in a multivalent pattern all copies are homologous and are therefore free to pair and recombine. As sufficient sequencing data is more readily available than high-quality cytological assessments of meiotic behavior or population genetic assessment of allelic segregation, especially for non-model organisms, bioinformatics approaches to infer origins and inheritance modes of polyploids using short-read sequencing data are attractive. Here we describe two such approaches, where the first is based on distributions of allelic read depth at heterozygous sites within an individual, as the expectations of such distributions are different for disomic and polysomic inheritance modes. The second approach is more laborious and based on a phylogenetic assessment of partially phased haplotypes of a polyploid in comparison to the closest diploid relatives. We discuss the sources of deviations from expected inheritance patterns, advantages and pitfalls of both methods, effects of mating types on the performance of the methods, and possible future developments.

Linkage disequilibrium under polysomic inheritance.

Linkage disequilibrium (LD) is the non-random association of alleles at different loci. Squared LD coefficients r2 (for phased genotypes) and [Formula: see text] (for unphased genotypes) will converge to constants that are determined by the sample size, the recombination frequency, the effective population size and the mating system. LD can therefore be used for gene mapping and the estimation of effective population size. However, current methods work only with diploids. To resolve this problem, we here extend the linkage disequilibrium measures to include polysomic inheritance. We derive the values of r2 and [Formula: see text] at equilibrium state for various mating systems and different ploidy levels. For unlinked loci, [Formula: see text] for monoecious and dioecious (with random pairing) mating systems or [Formula: see text] for dioecious mating systems (with lifetime pairing), where f is the number of females in a half-sib family and η is a constant related to the ploidy level. We simulate the application of estimating Ne using unphased genotypes. We find that estimating Ne in polyploids requires similar sample sizes and numbers of loci as in diploids, with the main source of bias due to using 0.5 as the recombination frequency.

Performing parentage analysis for polysomic inheritances based on allelic phenotypes.

Polyploidy poses several problems for parentage analysis. We present a new polysomic inheritance model for parentage analysis based on genotypes or allelic phenotypes to solve these problems. The effects of five factors are simultaneously accommodated in this model: (1) double-reduction, (2) null alleles, (3) negative amplification, (4) genotyping errors and (5) self-fertilization. To solve genotyping ambiguity (unknown allele dosage), we developed a new method to establish the likelihood formulas for allelic phenotype data and to simultaneously include the effects of our five chosen factors. We then evaluated and compared the performance of our new method with three established methods by using both simulated data and empirical data from the cultivated blueberry (Vaccinium corymbosum). We also developed and compared the performance of two additional estimators to estimate the genotyping error rate and the sample rate. We make our new methods freely available in the software package polygene, at http://github.com/huangkang1987/polygene.

Chromosome inheritance and meiotic stability in allopolyploid Brassica napus.

Homoeologous recombination, aneuploidy, and other genetic changes are common in resynthesized allopolyploid Brassica napus. In contrast, the chromosomes of cultivars have long been considered to be meiotically stable. To gain a better understanding of the underlying mechanisms leading to stabilization in the allopolyploid, the behavior of chromosomes during meiosis can be compared by unambiguous chromosome identification between resynthesized and natural B. napus. Compared with natural B. napus, resynthesized lines show high rates of nonhomologous centromere association, homoeologous recombination leading to translocation, homoeologous chromosome replacement, and association and breakage of 45S rDNA loci. In both natural and resynthesized B. napus, we observed low rates of univalents, A–C bivalents, and early sister chromatid separations. Reciprocal homoeologous chromosome exchanges and double reductions were photographed for the first time in meiotic telophase I. Meiotic errors were non-uniformly distributed across the genome in resynthesized B. napus, and in particular homoeologs sharing synteny along their entire length exhibited multivalents at diakinesis and polysomic inheritance at telophase I. Natural B. napus appeared to resolve meiotic errors mainly by suppressing homoeologous pairing, resolving nonhomologous centromere associations and 45S rDNA associations before diakinesis, and reducing homoeologous cross-overs.

Polygene: Population genetics analyses for autopolyploids based on allelic phenotypes

Abstract Polyploidy has appeared in almost every ancestral plant lineage, and in extant species, occurs frequently. When present, polyploidy presents problems for genetic data analysis, which are caused by both genotypic ambiguities and double‐reduction. To address these problems, we developed a new software package, polygene, which enables the estimation of genotypic frequencies for a number of polysomic inheritance models. Specifically, polygene obtains posterior probabilities for genotypes hidden within allelic phenotypes. Comprehensive modes of genetic analyses are provided by polygene, which include genetic diversity analysis, tests for allelic phenotypic or genotypic distributions, linkage disequilibrium and genetic differentiation, genetic distance analysis, principal coordinates analysis, hierarchical clustering analysis, individual inbreeding coefficient estimation, individual heterozygosity index estimation, population assignment, pairwise relatedness estimation, parentage analysis, analysis of molecular variance and Bayesian clustering. polygene enables easy and convenient allelic phenotype‐ or genotype‐based analysis for both autopolyploids and diploids. polygene will thus facilitate molecular ecology research involving autopolyploids.

Inference of individual ploidy level using codominant markers.

A significant portion of plant species are polyploids, with ploidy levels sometimes varying among individuals and/or populations. Current techniques to determine the individual ploidy, e.g., flow cytometry, chromosome counting or genotyping-by-sequencing, are often cumbersome. Based on the genotypic probabilities for polysomic inheritance under double-reduction, we developed a model to estimate allele frequency and infer the ploidy status of individuals from the allelic phenotypes of codominant genetic markers. The allele frequencies are estimated by an expectation-maximization algorithm in the presence of null alleles, false alleles, negative amplifications and self-fertilization, and the posterior probabilities are used to assign individuals into different levels of ploidy. The accuracy of this method under different conditions is evaluated. Our methods are freely available in a new software package, ploidyinfer, for use by other researchers which can be downloaded from http://github.com/huangkang1987/ploidyinfer.

Quantifying the Power and Precision of QTL Analysis in Autopolyploids Under Bivalent and Multivalent Genetic Models.

New genotyping technologies, offering the possibility of high genetic resolution at low cost, have helped fuel a surge in interest in the genetic analysis of polyploid species. Nevertheless, autopolyploid species present extra challenges not encountered in diploids and allopolyploids, such as polysomic inheritance or double reduction. Here we investigate the power and precision of quantitative trait locus (QTL) analysis in outcrossing autopolyploids, comparing the results of a model that assumes random bivalent chromosomal pairing during meiosis to one that also allows for multivalents and double reduction. Through a series of simulation studies we found that marginal gains in QTL detection power are achieved using the double reduction model when multivalent pairing occurs. However, when exploring the effect of variable genotypic information across parental homologs, we found that both QTL detection power and precision require high and uniform genotypic information contents. This effect far outweighed considerations regarding bivalent or multivalent pairing (and double reduction) during meiosis. We propose that autopolyploid QTL studies be accompanied by both marker coverage information and per-homolog genotypic information coefficients (GIC). Application of these methods to an autotetraploid potato (Solanum tuberosum L.) mapping population confirmed our ability to locate and dissect QTL in highly heterozygous outcrossing autotetraploid populations.

Genotypic Frequencies at Equilibrium for Polysomic Inheritance Under Double-Reduction.

Polyploids are organisms whose genomes consist of more than two complete sets of chromosomes. Both autopolyploids and allopolyploids may display polysomic inheritance. A peculiarity of polysomic inheritance is multivalent formation during meiosis resulting in double-reduction, which occurs when sister chromatid fragments segregate into the same gamete. Double-reduction can result in gametes carrying identical-by-descent alleles and slightly increasing homozygosity. This will cause the genotypic frequencies to deviate from expected values and will thus bias the results of standard population genetic analytical methods used in molecular ecology and selective breeding. In this study, we extend existing double-reduction models to account for any even level of ploidy, and derive the symbolic expressions for genotypic frequencies via two methods. Inbreeding coefficients and heterozygosity under double-reduction and inbreeding are also calculated. Numerical solutions obtained by computer simulations are compared with analytical solutions predicted by the model to validate the model.

Analysis of genetic variation using ISSR and the development of SCAR marker in synthetic autotetraploids of Vigna mungo (L.) Hepper

Autopolyploidy is one of the foremost mechanisms by which additional numbers of genome copies are acquired resulting in genetic and evolutionary novelty to the organisms. As such, it is considered to be a pathway for adaptation and has played a significant role in plant speciation. Autopolyploids are characterized by increased genome flexibility, allowing them to adapt and persist across heterogeneous landscapes in the long run. Genomic redundancy and polysomic inheritance are the hallmarks of doubled genome(s). In view of the significance of autopolyploidy in plant evolution, plant breeding and crop improvement, we sought to investigate the genetic variation/s associated with genome multiplication by exploiting artificial somatic autopolyploids in Vigna mungo. ISSR was employed to detect and identify any genomic DNA variation in colchitetraploids of V. mungo of three different generations, viz. C0, C1 and C2, in comparison to their putative diploids plants. Our data suggests a random, ‘short-term’ rapid change of the genome that may have occurred immediately after polyploidization. The population of autopolyploids demonstrated some degree of genetic heterozygosity in comparison to their diploid counterparts. Further, a polymorphic ISSR DNA loci was identified among the colchitetraploids which was translated into SCAR marker for identification of colchitetraploids versus the putative diploids of V. mungo.

Analysis of Molecular Variance (AMOVA) for Autopolyploids

Autopolyploids present several challenges to researchers studying population genetics, since almost all population genetics theory, and the expectations derived from this theory, has been developed for haploids and diploids. Also many statistical tools for the analysis of genetic data, such as AMOVA and genome scans, are available only for haploids and diploids. In this paper, we show how the Analysis of Molecular Variance (AMOVA) framework can be extended to include autopolyploid data, which will allow calculating several genetic summary statistics for estimating the strength of genetic differentiation among autopolyploid populations (FST, φST, or RST.). We show how this can be done by adjusting the equations for calculating the Sums of Squares, degrees of freedom and covariance components. The method can be applied to a dataset containing a single ploidy level, but also to datasets with a mixture of ploidy levels. In addition, we show how AMOVA can be used to estimate the summary statistic ρ, which was developed especially for polyploid data, but unfortunately has seen very little use. The ρ-statistic can be calculated in an AMOVA by first calculating a matrix of squared Euclidean distances for all pairs of individuals, based on the within-individual allele frequencies. The ρ-statistic is well suited for polyploid data since its expected value is independent of the ploidy level, the rate of double reduction, the frequency of polysomic inheritance, and the mating system. We tested the method using data simulated under a hierarchical island model: the results of the analyses of the simulated data closely matched the values derived from theoretical expectations. The problem of missing dosage information cannot be taken into account directly into the analysis, but can be remedied effectively by imputation of the allele frequencies. We hope that the development of AMOVA for autopolyploids will help to narrow the gap in availability of statistical tools for diploids and polyploids. We also hope that this research will increase the adoption of the ploidy-independent ρ-statistic, which has many qualities that makes it better suited for comparisons among species than the standard FST, both for diploids and for polyploids.

Insights into population genetics and evolution of polyploids and their ancestors.

We have developed the first comprehensive simulator for polyploid genomes (PolySim) and demonstrated its value by performing large-scale simulations to examine the effect of different population parameters on the evolution of polyploids. PolySim is unlimited in terms of ploidy, population size or number of simulated loci. Our process considered the evolution of polyploids from diploid ancestors, polysomic inheritance, inbreeding, recombination rate change in polyploids and gene flow from lower to higher ploidies. We compared the number of segregating single nucleotide polymorphisms, minor allele frequency, heterozygosity, R2 and average kinship relatedness between different simulated scenarios, and to real data from polyploid species. As expected, allotetraploid populations showed no difference from their ancestral diploids when population size remained constant and there was no gene flow or multivalent (MV) pairing between subgenomes. Autotetraploid populations showed significant differences from their ancestors for most parameters and diverged from their ancestral populations faster than allotetraploids. Autotetraploids can have significantly higher heterozygosity, relatedness and extended linkage disequilibrium compared with allotetraploids. Interestingly, autotetraploids were more sensitive to increasing selfing rate and decreasing population size. MV formation can homogenize allotetraploid subgenomes, but this homogenization requires a higher MV rate than previously proposed. Our results can be considered as the first building block to understand polyploid population evolutionary dynamics. PolySim can be used to simulate a wide variety of polyploid organisms that mimic empirical populations, which, in combination with quantitative genetics tools, can be used to investigate the power of genomewide association, genomic selection or breeding programme designs in these species.

Population assignment in autopolyploids.

Understanding the patterns of contemporary gene dispersal within and among populations is of critical importance to population genetics and in managing populations for conservation. In contrast to diploids, there are few studies of gene dispersal in autopolyploids, in part due to complex polysomic inheritance and genotype ambiguity. Here we develop a novel approach for population assignment for codominant markers for autotetraploids and autohexaploids. This method accounts for polysomic inheritance, unreduced gametes and unknown allele dosage. It can also utilise information regarding the origin and genotype of one parent for population assignment of maternal or paternal parents. Using simulations, we demonstrate that our approach achieves high levels of accuracy for assignment even when population divergence is low (FST~0.06) and with only 12 microsatellite loci. We also show that substantially higher accuracy is achieved when known maternal information is utilised, regardless of whether allele dosage is known. Although this novel method exhibited near identical levels of accuracy to Structure when population divergence was high, it performed substantially better for most parameters at moderate (FST=0.06) to low levels of divergence (FST=0.03). These methods fill an important gap in the toolset for autopolyploids and pave the way for investigating contemporary gene dispersal in a widespread group of organisms.

An ultra-dense integrated linkage map for hexaploid chrysanthemum enables multi-allelic QTL analysis

Key messageWe constructed the first integrated genetic linkage map in a polysomic hexaploid. This enabled us to estimate inheritance of parental haplotypes in the offspring and detect multi-allelic QTL.Construction and use of linkage maps are challenging in hexaploids with polysomic inheritance. Full map integration requires calculations of recombination frequency between markers with complex segregation types. In addition, detection of QTL in hexaploids requires information on all six alleles at one locus for each individual. We describe a method that we used to construct a fully integrated linkage map for chrysanthemum (Chrysanthemum × morifolium, 2n = 6x = 54). A bi-parental F1 population of 406 individuals was genotyped with an 183,000 SNP genotyping array. The resulting linkage map consisted of 30,312 segregating SNP markers of all possible marker dosage types, representing nine chromosomal linkage groups and 107 out of 108 expected homologues. Synteny with lettuce (Lactuca sativa) showed local colinearity. Overall, it was high enough to number the chrysanthemum chromosomal linkage groups according to those in lettuce. We used the integrated and phased linkage map to reconstruct inheritance of parental haplotypes in the F1 population. Estimated probabilities for the parental haplotypes were used for multi-allelic QTL analyses on four traits with different underlying genetic architectures. This resulted in the identification of major QTL that were affected by multiple alleles having a differential effect on the phenotype. The presented linkage map sets a standard for future genetic mapping analyses in chrysanthemum and closely related species. Moreover, the described methods are a major step forward for linkage mapping and QTL analysis in hexaploids.

Autopolyploidy genome duplication preserves other ancient genome duplications in Atlantic salmon (Salmo salar).

Salmonids (e.g. Atlantic salmon, Pacific salmon, and trouts) have a long legacy of genome duplication. In addition to three ancient genome duplications that all teleosts are thought to share, salmonids have had one additional genome duplication. We explored a methodology for untangling these duplications from each other to better understand them in Atlantic salmon. In this methodology, homeologous regions (paralogous/duplicated genomic regions originating from a whole genome duplication) from the most recent genome duplication were assumed to have duplicated genes at greater density and have greater sequence similarity. This assumption was used to differentiate duplicated gene pairs in Atlantic salmon that are either from the most recent genome duplication or from earlier duplications. From a comparison with multiple vertebrate species, it is clear that Atlantic salmon have retained more duplicated genes from ancient genome duplications than other vertebrates--often at higher density in the genome and containing fewer synonymous mutations. It may be that polysomic inheritance is the mechanism responsible for maintaining ancient gene duplicates in salmonids. Polysomic inheritance (when multiple chromosomes pair during meiosis) is thought to be relatively common in salmonids compared to other vertebrate species. These findings illuminate how genome duplications may not only increase the number of duplicated genes, but may also be involved in the maintenance of them from previous genome duplications as well.

SSR-enriched genetic linkage maps of bermudagrass (Cynodon dactylon×transvaalensis), and their comparison with allied plant genomes.

We report SSR-enriched genetic maps of bermudagrass that: (1) reveal partial residual polysomic inheritance in the tetraploid species, and (2) provide insights into the evolution of chloridoid genomes. This study describes genetic linkage maps of two bermudagrass species, Cynodon dactylon (T89) and Cynodon transvaalensis (T574), that integrate heterologous microsatellite markers from sugarcane into frameworks built with single-dose restriction fragments (SDRFs). A maximum likelihood approach was used to construct two separate parental maps from a population of 110 F1 progeny of a cross between the two parents. The T89 map is based on 291 loci on 34 cosegregating groups (CGs), with an average marker spacing of 12.5cM. The T574 map is based on 125 loci on 14 CGs, with an average marker spacing of 10.7cM. Six T89 and one T574 CG(s) deviated from disomic inheritance. Furthermore, marker segregation data and linkage phase analysis revealed partial residual polysomic inheritance in T89, suggesting that common bermudagrass is undergoing diploidization following whole genome duplication (WGD). Twenty-six T89 CGs were coalesced into 9 homo(eo)logous linkage groups (LGs), while 12 T574 CGs were assembled into 9 LGs, both putatively representing the basic chromosome complement (x = 9) of the species. Eight T89 and two T574 CGs remain unassigned. The marker composition of bermudagrass ancestral chromosomes was inferred by aligning T89 and T574 homologs, and used in comparisons to sorghum and rice genome sequences based on 108 and 91 significant blast hits, respectively. Two nested chromosome fusions (NCFs) shared by two other chloridoids (i.e., zoysiagrass and finger millet) and at least three independent translocation events were evident during chromosome number reduction from 14 in the polyploid common ancestor of Poaceae to 9 in Cynodon.

Population genetics of autopolyploids under a mixed mating model and the estimation of selfing rate.

Nowadays, the population genetics analysis of autopolyploid species faces many difficulties due to (i) limited development of population genetics tools under polysomic inheritance, (ii) difficulties to assess allelic dosage when genotyping individuals and (iii) a form of inbreeding resulting from the mechanism of 'double reduction'. Consequently, few data analysis computer programs are applicable to autopolyploids. To contribute bridging this gap, this article first derives theoretical expectations for the inbreeding and identity disequilibrium coefficients under polysomic inheritance in a mixed mating model. Moment estimators of these coefficients are proposed when exact genotypes or just markers phenotypes (i.e. allelic dosage unknown) are available. This led to the development of estimators of the selfing rate based on adult genotypes or phenotypes and applicable to any even-ploidy level. Their statistical performances and robustness were assessed by numerical simulations. Contrary to inbreeding-based estimators, the identity disequilibrium-based estimator using phenotypes is robust (absolute bias generally < 0.05), even in the presence of double reduction, null alleles or biparental inbreeding due to isolation by distance. A fairly good precision of the selfing rate estimates (root mean squared error < 0.1) is already achievable using a sample of 30-50 individuals phenotyped at 10 loci bearing 5-10 alleles each, conditions reachable using microsatellite markers. Diallelic markers (e.g. SNP) can also perform satisfactorily in diploids and tetraploids but more polymorphic markers are necessary for higher ploidy levels. The method is implemented in the software SPAGeDi and should contribute to reduce the lack of population genetics tools applicable to autopolyploids.

A pairwise relatedness estimator for polyploids

Studies in genetics and ecology often require estimates of relatedness coefficients based on genetic marker data. Many diploid estimators have been developed using either method-of-moments or maximum-likelihood estimates. However, there are no relatedness estimators for polyploids. The development of a moment estimator for polyploids with polysomic inheritance, which simultaneously incorporates the two-gene relatedness coefficient and various 'higher-order' coefficients, is described here. The performance of the estimator is compared to other estimators under a variety of conditions. When using a small number of loci, the estimator is biased because of an increase in ill-conditioned matrices. However, the estimator becomes asymptotically unbiased with large numbers of loci. The ambiguity of polyploid heterozygotes (when balanced heterozygotes cannot be distinguished from unbalanced heterozygotes) is also considered; as with low numbers of loci, genotype ambiguity leads to bias. A software, PolyRelatedness, implementing this method and supporting a maximum ploidy of 8 is provided.

Meiotic chromosome pairing in Actinidia chinensis var. deliciosa

Polyploids are defined as either autopolyploids or allopolyploids, depending on their mode of origin and/or chromosome pairing behaviour. Autopolyploids have chromosome sets that are the result of the duplication or combination of related genomes (e.g., AAAA), while allopolyploids result from the combination of sets of chromosomes from two or more different taxa (e.g., AABB, AABBCC). Allopolyploids are expected to show preferential pairing of homologous chromosomes from within each parental sub-genome, leading to disomic inheritance. In contrast, autopolyploids are expected to show random pairing of chromosomes (non-preferential pairing), potentially leading to polysomic inheritance. The two main cultivated taxa of Actinidia (kiwifruit) are A. chinensis (2x and 4x) and A. chinensis var. deliciosa (6x). There is debate whether A. chinensis var. deliciosa is an autopolyploid derived solely from A. chinensis or whether it is an allopolyploid derived from A. chinensis and one or two other Actinidia taxa. To investigate whether preferential or non-preferential chromosome pairing occurs in A. chinensis var. deliciosa, the inheritance of microsatellite alleles was analysed in the tetraploid progeny of a cross between A. chinensis var. deliciosa and the distantly related Actinidia eriantha Benth. (2x). The frequencies of inherited microsatellite allelic combinations in the hybrids suggested that non-preferential chromosome pairing had occurred in the A. chinensis var. deliciosa parent. Meiotic chromosome analysis showed predominantly bivalent formation in A. chinensis var. deliciosa, but a low frequency of quadrivalent chromosome formations was observed (1 observed in 20 pollen mother cells).

Cytogenetic evidence of mixed disomic and polysomic inheritance in an allotetraploid (AABB) Musa genotype

Edible bananas originated mainly from two wild species, Musa acuminata Colla (AA) and Musa balbisiana Colla (BB), and triploid cultivars with an AAA, AAB or ABB genome are the most widely used. In the present study, chromosome pairing affinities are investigated in a sterile AB Indian variety and in its fertile colchicine-induced allotetraploid (AABB) derivative to determine the inheritance pattern of the tetraploid genotype. The potential implications of interspecific recombination and chromosomal composition of diploid gametes for Musa improvement are presented. The pairing of different chromosome sets at diploid and tetraploid levels was investigated through a combination of conventional cytogenetic and genomic in-situ hybridization (GISH) analyses of meiotic chromosomes, leading to a likelihood model of the pairing behaviour. GISH analysis of mitotic chromosomes was also conducted to reveal the chromosome constitution of hybrids derived from crosses involving the allotetraploid genotype. Analysis of chromosome associations at both ploidy levels suggested that the newly formed allotetraploid behaves as a 'segmental allotetraploid' with three chromosome sets in a tetrasomic pattern, three sets in a likely disomic pattern and the five remaining sets in an intermediate pattern. Balanced and unbalanced diploid gametes were detected in progenies, with the chromosome constitution appearing to be more homogenous in pollen than in ovules. Colchicine-induced allotetraploids in Musa provide access to the genetic background of natural AB varieties. The segmental inheritance pattern exhibited by the AABB allotetraploid genotype implies chromosome exchanges between M. acuminata and M. balbisiana species and opens new horizons for reciprocal transfer of valuable alleles.