Unravelling genomic drivers of speciation in Musa through genome assemblies of wild banana ancestors

Abstract

Hybridization between wild Musa species and subspecies from Southeast Asia is at the origin of cultivated bananas. The genomes of these cultivars are complex mosaics involving nine genetic groups, including two previously unknown contributors. This study provides continuous genome assemblies for six wild genetic groups, one of which represents one of the unknown ancestor, identified as M.acuminata ssp. halabanensis. The second unknown ancestor partially present in a seventh assembly appears related to M. a. ssp. zebrina. These assemblies provide key resources for banana genetics and for improving cultivar assemblies, including that of the emblematic triploid Cavendish. Comparative and phylogenetic analyses reveal an ongoing speciation process within Musa, characterised by large chromosome rearrangements and centromere differentiation through the integration of different types of repeated sequences, including rDNA tandem repeats. This speciation process may have been favoured by reproductive isolation related to the particular context of climate and land connectivity fluctuations in the Southeast Asian region.

Fig. 1. Distribution of the main wild genetic groups that contributed to banana cultivars.

Adapted from De Langhe et al. and Ahmad.

Fig. 2. In silico chromosome ancestry painting of the Pisang Madu assembly and phylogenetic analysis.

In silico chromosome ancestry painting of the haplotype 1 a and haplotype 2 b assembly of Pisang Madu. Colour codes indicate the ancestors involved, as defined by Martin et al.. ‘Unknown’ is for the remaining unknown ancestry and ‘NA’ corresponds to regions in which no origin could be attributed. Maximum likelihood phylogenetic analysis (GTR + Γ substitution model) of ADH c and GBSS d haplotypes. The trees were obtained with MEGAX; branch lengths are measured according to the number of substitutions per site, and the percentage of trees in which the associated taxa clustered together is shown alongside the branches. Sequences that are grouped in polytomy together with M. a. ssp. halabanensis are enframed in purple in c and d. Black arrows locate the ADH (chr06) and GBSS (chr09) on each Pisang Madu haplotypes assemblies. Source data are provided as a Source Data file.

Fig. 3. Global synteny comparison and phylogenetic analysis of the assemblies.

Assemblies are colored based on their origin. Syntenic regions are indicated by gray ribbons, while colored ribbons indicate translocated regions. Translocation events are indicated by stars on the maximum likelihood phylogenetic tree. This tree was constructed using only the genes in regions assigned to the unknown ancestor in the P. Madu H2 assembly and their orthologs in the other assemblies. Branch support, calculated as the proportion of CDS gene phylogenies supporting the branch, are indicated. Divergence times, in million years (My), are indicated in bold italics close to nodes. Source data are provided as a Source Data file.

Fig. 4. Structure of the 1/4, 3/8 and 1/7 chromosome rearrangements compared to the M. a. malaccensis reference structure.

The reference structure corresponds to the M. a. ssp. malaccensis DH-Pahang (M. a. malaccensis (reference)) assembly compared to a 1/4 rearranged structure of the M. a. ssp. malaccensis h1 (M. a. malaccensis h1 chr1T4 and chr4T1) assembly; b 3/8 rearranged structure of the M. a. ssp. zebrina assembly (M. a. zebrina chr3T8 and chr8T3); c 1/7 rearranged structure of the Pisang Madu haplotype 2 assembly (P. Madu H2 chr1T7 and chr7T1). Syntenic regions are indicated with ribbons. Coloured ribbons correspond to translocated fragments. Chromosome and translocated fragments were named according to the nomenclature of Martin et al.. Curves within the chromosomes correspond to (peri)centromeric repeats in grey and Nanica in red. Source data are provided as a Source Data file.

Fig. 5. Transposable elements and tandem repeats.

Stacked curves representing the density of SIRE/Maximus transposable elements, centromeric sequences and tandem repeats along (a) the M. a. ssp. zebrina and (b) M. schizocarpa assemblies. The pink asterisk indicates that the region contains a stretch of N impacting the full representation of the rDNA cluster. Typical results of fluorescence in situ hybridization (FISH) on metaphase chromosomes of M. schizocarpa with (c) Nanica transposable elements detected in red, (d) 45S rDNA detected in pink and (e) Nanica transposable element detected in red and 5S rDNA detected in blue. White horizontal bars in (c), (d) and (e) represent 5.1 µm. Arrows in (d) indicate 45S rDNA detected with lower signal intensity. The inserts represent a chromosome that was offset from the presented field of view. Source data are provided as a Source Data file.

Fig. 6. In silico chromosome ancestry painting of a triploid Cavendish cultivar genome assembly and comparison with ancestral contributor assemblies.

Circular representation of the Cavendish haplotype assembly (Baxijiao accession) with: a the distribution of tandem repeats and SIRE/Maximus transposable elements along chromosomes (color code from Fig. 5), b In silico chromosome painting according to the ancestral contributors (color code from Fig. 2), c In silico chromosome painting according to suggested parental gametes with black for the 1x gamete and orange and brown for the two haplotypes of the 2x gamete. d Comparison of the assembled Cavendish chromosomes that should have the 1/4, 1/7 and 3/8 rearrangements on the basis of their ancestral origin, with the reference rearranged structures found in ancestral contributor assemblies. The reference structure used for the 1/4 reciprocal translocation corresponds to M. a. ssp. malaccensis (M. a. malaccensis h1 chr1T4 and chr4T1), that used for the 1/7 reciprocal translocation corresponds to Pisang Madu H2 (P. Madu H2 chr1T7 and chr7T1) and that used for the 3/8 reciprocal translocation corresponds to M. a. ssp. zebrina (M. a. zebrina chr3T8 and chr8T3). Dashed lines indicate chromosomes from the Cavendish assembly that were compared. Translocated segments on the reference translocated structure are indicated by vertical black bars. Gray shapes locate the syntenic regions. Stars indicate the translocation breakpoints that are present in the assembly. Green, red and blue dots on d locate the BACs used for BAC-FISH analysis on metaphase chromosomes of Cavendish (Grande Naine accession) to validate the presence of the 1/4 and 1/7 e and 3/8 f translocated structures in Cavendish. White horizontal bars in e and f represent 5.1 µm. White dashed line separated chromosomes from another cell. Source data are provided as a Source Data file.