The emergence of HTS methodologies has allowed the detection of cryptic species (Carstens and Satler, 2013), the resolution of complex phylogenetic trees (Bogarín et al., 2018; Frajman et al.,2019; Hassemer et al., 2019; Yang et al., 2019) and the reconstruction of evolutionary patterns in extinct species (Moreno-Aguilar et al., 2020). Thus, HTS methods provide new sources of useful data to clarify phylogenetic enigmas that classical molecular methods could not decipher (e.g., Urtubey et al.,2018). Among HTS methods, target capture is currently being used in a broad number of plant systematics and evolutionary studies due to its versatility to successfully sequence hundreds of loci from highly degraded DNA samples (Brewer et al., 2019, Viruel et al.,2019). Herbarium samples constitute a valuable and vast source of information for morphological and niche modelling approaches, and recently proved to be equally important for phylogenetic studies based on DNA sequence data obtained using HTS methods. In our study, we used herbarium material, with the oldest specimen sequenced collected in 1788, and a custom bait capture kit targeting 260 low copy nuclear genes (Soto Gomez et al., 2019), to reveal the evolutionary patterns and relationships between taxa belonging to the Tamus clade ofDioscorea (Figure 1). By sequencing 76 samples of theTamus clade, the phylogenomic and genetic clustering approaches revealed extensive infraspecific variability in D. communis sensu lato , clearly dividing it into three genetic groups, each showing a distinct geographic distribution across the Mediterranean and western Europe (Figure 6). Two of these genetic groups are congruent with the previously recognized Tamus edulis and T. cretica , which were recently placed within the large morphological variability and wide distribution of D. communis s.l . HTS methodologies applied to herbarium material allowed us to recognize the species rank for these genetic groups and to support the split of D. communis s.l. intoD. edulis , D. cretica and D. communis , and to maintain D. orientalis as a species.
Whole-genome duplication events (i.e., polyploidy) have been commonly reported across flowering plants and have been correlated with diversification of gene functions and new genetic architecture, which could be linked with adaptative traits (Wendel et al., 2018). Increased speciation events have been observed in some angiosperm lineages reported to have a high incidence of whole-genome duplication events (Wood et al., 2009; Zhan et al., 2016). Polyploidy is a common phenomenon, which has been frequently reported in severalDioscorea species (Viruel et al., 2008), although defining ploidy of the Tamus and Borderea clades has been challenging. The twoDioscorea species belonging to the Borderea clade, D. chouardii and D. pyrenaica , have chromosome counts of 2n= 24. Based on the discovery of allotetraploidy using microsatellite markers (Segarra-Moragues et al ., 2003), it was proposed that the chromosome base number for the Borderea clade was x = 6 (see also Viruel et al ., 2008). Extrapolating this find to the sister Tamus clade, the known chromosome counts reported for D. communiss.s. of 2n = 36 and 48 (Al-Shehbaz and Schubert, 1989; Viruel et al ., 2019) would therefore represent hexaploid and octoploid forms, respectively. Similarly, the Macaronesian D. edulis , with 2n = 96, would be 16-ploid assuming a base chromosome number of x = 6. Using flow cytometry to estimate ploidy in D. communis s.s. , multiple ploidies were observed (1C-values ranging from 0.41 to 1.36 pg; Viruel et al ., 2019). The chromosome number and genome size of D. orientalis andD. cretica remain unknown, but allelic ratios estimated for each SNP per sample using HTS data can be used as a proxy to distinguish between diploid and polyploid forms when multiple ploidies are expected in a group of plants (Viruel et al ., 2019). Median and mean values of allelic ratios based on the number of reads supporting each SNP were recently proposed to classify Dioscorea samples as diploid forms when the allelic ratio is <2, and polyploids when >2 (Viruel et al ., 2019). For example, all samples of D. edulis had mean and median allelic ratios >2 (Table 1), confirming the polyploid nature of this species based on chromosome data. In all cases, D. orientalissamples studied here showed allelic ratios >2 and would therefore be estimated to be a polyploid species (Table 1). For D. communis s.s ., all samples were estimated to be polyploids except for two samples of the clade DC3 from the eastern Mediterranean with mean and median allelic ratio values <2 (samples S67 and R12, Table 1). Samples estimated to be diploid based on allelic ratios were also observed in D. cretica , with half of the samples (eight) having average and median allelic ratios <2 (Table 1). The incidence of diploid forms, as estimated using allelic ratio values, in the eastern Mediterranean will require further investigation applying cytological and flow cytometry methodologies.