Morphological characters associated with cultivation continues to segregate despite genomic admixture
Despite low levels of genetic differentiation, there are clear morphological separations between specimens identifiable as Mentha longifolia , M. suaveolens , and M. spicata (Figures 1 and 2 and Table 2). In particular, M. spicata shows a distinct morphology driven by a shift in indumentum toward fewer and shorter trichomes (Figures 1, S1, and S2). Mentha spicata is widely cultivated for culinary use and it is therefore possible that the reduction in characters of indumentum is associated with human selection for a more palatable plant (less hairy; Harley & Brighton, 1977; Munguía-Rosas, Jácome-Flores, Bello-Bedoy, Solís-Montero, & Ochoa-Estrada, 2019), but not all cultivated mints have smooth stems and leaves. The glands associated with trichomes are however, the main organs for production of the desired essential mint oils (Fahn, 1979; Jia, Liu, Gao, & Xin, 2013; McCaskill & Croteau, 1995; Mishra, Lal, Chanotiya, & Dhawan, 2017;Yu et al., 2018). It is therefore possible that the shift in trichome characters is a byproduct of selection for essential oil production in cultivation (Maffei, Gallino, & Sacco, 1986; Mishra et al., 2017; Jia et al., 2013, Szabó, Sárosi, Cserháti, & Ferenczy, 2010). The lack of previously identified genes associated with trichome function and morphology among F SToutlier scaffolds (Table S4) is likely a result of the polygenic nature of these characters (Chalvin, Drevensek, Dron, Bendahmane, & Boualem, 2020; Chopra et al., 2019, Figueroa-Pérez et al., 2019; Mishra et al., 2017) coupled with the large sharing of polymorphic sites betweenM. longifolia and M. spicata (Figure 3). Our sample size is therefore too small to detect significant genome wide associations (Hong & Park, 2012). However, three genes; Phospho-2-dehydro-3-deoxyheptonate aldolase 2 (DAHP2; Langer et al., 2014), Benzyl alcohol O-benzoyltransferase (BEBT1; Boatright et al., 2004; Orlova et al., 2006), and Tetrahydrocannabinolic acid synthase (THCAS; Sirikantaramas et al., 2004), involved in the production of volatile organic compounds, were found among theF ST-outlier scaffold (Table S4), indicating that there indeed is high sequence divergence between different mints for some genes potentially involved in the production of their characteristic aromatic oils.
Despite the overall genetic similarities between specimens of M. spicata and M. longifolia only a few individuals morphologically identified as one taxon genetically cluster with the opposite species (Figure 2a). However, these few cases do suggest that genetic variants associated with human desirable traits can be lost despite most of the genome otherwise being of a M. spicata origin. More importantly, mismatches between morphology and genomic clustering suggest that the variants associated with human desirable traits can enter the genome of other mints through hybridization similar to what has been observed in other systems (e.g. Fuchs et al., 2016; Karlsson et al., 2016). In addition, some specimens of M. × rotundifolia show admixture profiles and genomic clustering that resemble M. spicata (rot4 and spi1, Figure 2). The morphological clustering of these specimens suggest that they likely represent cases where gene flow between M. spicata and M. longifolia has resulted in the loss of genetic variants associated with the cultivated morphology and that their morphology resemble M. × rotundifolia due to a retention of polymorphisms from M. suaveolens . Introgression between M. spicata and M. longifolia has previously been suggested (Harley & Brighton, 1977), but never been evaluated using multiple specimens per species collected across large geographic ranges. Here we show the genetic admixture between these two species indeed is frequent. However, despite this it is clear that genetic variants associated with the cultivated morphology continue to segregate in both cultivated and wild populations.