Independent genomic histories despite admixture from the same gene pools
Although both M. spicata and M. × rotundifolia are generally believed to stem from the same parental gene pools, our admixture analyses suggest they have independent histories (Figure 2b). While chromosome counts for M. spicata suggest that it is a polyploid, reports of chromosome counts for M. × rotundifolia are rare but do point to a diploid status (Chambers & Hummer, 1994; Harley & Brighton, 1977). The usage of herbarium material prevented cytological evaluations of our samples and we therefore bioinformatically estimated ploidy using our sequencing data. Our estimates are not conclusive but in combination with our admixture analyses they do support that M. spicata is an allopolyploid and that M. longifolia is likely cytomictic with both diploid and polyploid populations (Figure 2 and Table 1). However, the ploidy level of M. x rotundifolia is more uncertain as most were assigned as “likely polyploid” (triploids; Table 1). The existence of triploid individuals among the analyzed specimens is not unlikely. However, it is unclear to what extent HMMploidy is able to distinguish between triploids and tetraploids. For example, tetraploids formed from the merging of the very closely related genomes of M. longifoliaand M. suaveolens could potentially appear as triploids due to extensive allele sharing between them (Figure 3). But it is unclear why this would more often be the case for specimens morphologically assigned as M. x rotundifolia (Table 1). It is possible that the distinct genomic histories of M. spicata and M. xrotundifolia are linked to differences in ploidy but no firm conclusions can be made based on our data.
Differences in frequencies of sexual reproduction can also drive the genomic distinction between M. × rotundifolia and M. spicata . While M. × rotundifolia exhibits a reduction in fertility, M. spicata is usually completely fertile (Table S3; Harley & Brighton, 1977). It is therefore possible that M. ×rotundifolia to a higher degree propagates clonally via rhizomes. However, such a scenario is not completely supported by our plastome analysis that suggests that most of the analyzed M. ×rotundifolia stem from independent hybridizations while the plastomes of M. spicata are largely homogeneous suggesting a single origin (Figure S3). It is however likely that human imposed selection on M. spicata in cultivation coupled with clonal propagation of plants with desired traits (Harley & Brighton, 1977) have caused shifts in allele frequencies in M. spicata and that the distinct genomic histories of these taxa, in addition to potentially be associated with ploidy differences, also are coupled to the (partly) cultivation status of M. spicata .