Evolutionary patterns of the Tamus clade of Dioscorea in the Mediterranean
Overall evolutionary patterns of Dioscorea lineages have been thoroughly studied using plastid markers (e.g., Wilkin et al.,2005; Hsu et al., 2013; Maurin et al., 2016; Viruelet al., 2016) and low copy nuclear genes (e.g. Viruel et al., 2018, Soto Gomez et al., 2019). Previous studies determined that yams diverged and expanded since the Late Cretaceous, probably from Laurasia, and that a split of the African-Mediterranean lineage, which includes the Tamus clade, likely occurred in the Oligocene, following a westward migration ca. 33 MY (Viruel et al ., 2016). Fossil records indicate that Dioscorea ancestors persisted in Europe during the Oligocene (Andreànzky, 1959). Based on data from four plastid markers, the split between the two Mediterranean clades, Borderea and Tamus, was estimated to have occurred during the late Oligocene (ca. 25.7 MY) (Viruel et al ., 2016), a similar divergence time to the one we obtained with our analyses based on 260 nuclear genes, which indicate that this divergence took place in the early Oligocene (28.2–28.4 MY); Supplementary Data Figure S4). Two narrowly endemic species of the Borderea clade survived in refugia in the Pyrenean mountains, D. chouardii , a critically endangered species with only one known locality (Martínez & Otano, 2011), and D. pyrenaica , with a slightly wider distribution in the central Pyrenees and Pre-Pyrenees (Segarra & Catalán, 2005; Catalan et al ., 2006; Segarra-Moragues & Catalan, 2008; García et al., 2012). A previous study (Viruel et al., 2016) estimated a split between these two species during the early Pliocene (ca. 4.3 MY), whereas our results indicated that this divergence likely occurred during the late Miocene (8.1–10.6 MY).
The differences observed in divergence times for the Tamus clade in comparison with previous studies are a consequence of the newly recognized species (D. cretica ). In Viruel et al . (2016), the crown node of the Tamus clade was estimated to be 15.3 MY (D. edulis (D. communis , D. orientalis )), and the subsequent split between D. orientalis and D. communis at 10.4 MY. However, the samples of D. orientalis included in Viruel et al . (2016) have now been reidentified as D. cretica , and thus the older age estimates herein for the crown node of the Tamus clade (20.6–18.2 MY) demonstrate an early split of D. orientalis in the eastern Mediterranean, followed by a split of D. edulis ca. 16.0-13.5 MY, and the divergence of D. communis s.s. andD. cretica ca. 6.6–5.6 MY. Given the findings presented here, with a sampling representative of the whole distribution range of the Tamus clade across the circum-Mediterranean region, we conclude that the divergence times estimated here are more robust and taxonomically more representative, which allowed us to reassess the species delimitation in this group.
The Mediterranean region is considered one of the major biodiversity hotspots of the world (Médail and Quézel, 1997). Fossil records and evolutionary studies have confirmed that the ancestors of several plant lineages were part of a tropical flora that occupied the Mediterranean region during the Miocene and early Pliocene (Suc et al., 2018). The drastic subsequent climatic changes that came during the Pliocene (3.5–2.4 MY), with a significant drop in temperature and a marked seasonality in thermal and rainfall regimes, impacted the diversification patterns of plant lineages and resulted in narrow endemics in the margins of the distribution range of their sister species (e.g., Ceratonia oreothauma Hillc., G.P.Lewis & Verdc.; Viruel et al., 2020). The diversification of species in the Tamus clade likely occurred during the Miocene when subtropical climatic conditions were present across the Mediterranean (Suc et al ., 2018). The most recent common ancestor of all Tamus clade taxa likely diversified during the early Miocene (20.6–18.2 MY), when the lineages that gave rise to the current D. orientalis and the clade comprising the three lineages of D. communis s.l. likely split. This was followed by a subsequent split of the Macaronesian D. edulis that would have taken place in the mid-Miocene (16.0–13.5 MY), after the formation of some of the Canary Islands, which has been estimated to start around 23 MY (Sanmartín et al., 2008; Florencio et al ., 2021). The most recent split between D. communis s.s. and D. cretica is estimated to have occurred during the Messinian (Miocene, 6.6–5.6 MY). During this period, the significant and rapid lowering of the sea level of the Mediterranean also resulted in new terrestrial biogeographical connections allowed by the formation of land-bridges.
Several phylogeographic studies have attempted to explain the biodiversity patterns and processes that shaped the Mediterranean region and its development into one of the world’s biodiversity hotspot (e.g., Nieto Feliner, 2014; Thompson, 2021). Two main areas of high plant endemism were identified in the western (Iberian Peninsula and Morocco) and eastern Mediterranean (including Turkey and Greece) (Médail and Quézel, 1997). In both these areas, Quaternary glaciations likely played a major role shaping the distribution of species and left a footprint in the genetic structure of many Mediterranean species, particularly in refugia (Médail and Diadema, 2009). Western and eastern genetic groups have been identified in phylogeographic patterns of several Mediterranean plants, leading to disjunct distributions in some cases, such as in Microcnemum Ung.-Sternb (Amaranthaceae) andMandragora L. (Solanaceae) (Kadereit and Yaprak, 2008; Voliset al., 2018), or by differentiating morphotypes that later hybridized in intermediate zones (e.g., Quercus ilex L., Lumaretet al., 2002). The strong geographic influence in the genetic structuring of D. communis across the Mediterranean may have also been slightly influenced by bird dispersal. Bird dispersals have contributed to shaping the postglacial recolonization of the Mediterranean, such as seen in Frangula alnus Mill. (Hampeet al ., 2003). The birds that consume berries produced by species of the Tamus clade, mainly blackbirds (Turdus merula L.), robins (Erithacus rubecula L.) and blackcaps (Sylvia communisLatham) (Chiscano, 1983; Herrera, 1984), are predominantly sedentary birds or have modern migratory routes that do not strictly coincide with the past and current distribution patterns estimated in this study (Adriaensen, 1988; Burfield and Van Bommel, 2004). It would thus be useful to analyse the patterns of genetic structure at the population level of D. communis s.s. in more detail and in connection with the possible magnitude of ornithochory, which has never been studied in detail to our knowledge.
Changes in ploidy, morphological differences and introgression between the central Mediterranean and western European populations have been shown to have occurred between the D. communis s.s. and D. cretica lineages (Figures 2, 5). The central-eastern Mediterranean area constitutes the contact region between these two species and is congruent with the introgression patterns found in our study (Figure 2). Five out of 16 samples studied of D. cretica exhibited <20% of admixture index with D. communis s.s. , and all individuals of D. communis belonging to the eastern clade ofD. communis s.s. showed <20% of admixture index withD. cretica (Figure 2). However, only four individuals from the central Mediterranean and western European subclades of D. communis were detected as introgressing with a sister species, and one sample of D. cretica was placed in a clade of D. communis s.s. in the phylogenetic tree based on plastid data (R32, Figure 1). These results are congruent with their potential distribution in disjunct refugia followed by secondary contact through recolonization, and by maintaining some capacity of interspecific gene flow between closely related species (see Viruel et al., 2021). The topological incongruencies found between the nuclear and plastid phylogenetic trees, indicative of plastid capture events (Figure 1), are congruent with these hypothesized introgression patterns: plastid clades I and II are found in the central and eastern Mediterranean lineages without a clear geographic separation (Figure 6), whereas clade III is uniquely found in the western part of the Mediterranean where lower introgression events have been inferred.