Xanthium strumarium, L.

1.1 | Xanthium strumarium View in CoL

It is a thermophilic complex that is commonly found in sandy habitats, such as ditches and riverbanks ( Arcangeli 1882; Widder 1923). It is mainly distributed throughout Eurasia and Africa, and known to be native to these regions, with records in the literature dating back to the Dioscorides' De Materia Medica (First century BC) and fossil records from the last interglacial period ( Löve 1975; Chauhan 1991). However, the species has suffered a progressive decline in the last two centuries (parallel to the spreading of the non-native X. orientale ) and is nowadays difficult to find in nature in a great part of its native range ( Müller-Kiefer and Tomasello 2022). Differently from its congenerics, the species did not manage to spread outside its native distribution range. The only putative stable populations found outside the Old World were in southern Brazil, initially described by Vellozo (1881) as a different species ( X. brasilicum Vell. ) and presumably originated from burs arrived from the Mediterranean basin ( Widder 1923).

Xanthium strumarium View in CoL distribution range is unique when compared to the other species of the genus, which are all native to America ( Tomasello 2018). Even subtribe Ambrosiinae Less. and tribe Heliantheae Cass. to which Xanthium View in CoL belongs are well-known to have an American origin and distribution centre ( Baldwin 2009). The subtribe, for example, is composed by about thirteen genera and over 150 species ( Tomasello et al. 2019), and the only other taxon non-native to America apart from X. strumarium View in CoL is Ambrosia maritima L. View in CoL ( Martin et al. 2018). Ambrosia maritima View in CoL is supposed to be native to the Mediterranean basin and the coast of central-western Africa, although its distinctiveness from other taxa of the genus with an American origin is questionable based on phylogenetic data ( Martin et al. 2018). At the tribal level, only a few examples are known of genera including species with native ranges found outside America ( Baldwin 2009), all of them from subtribe Ecliptinae Less. ( Blainvillea Cass. View in CoL , Eclipta L., View in CoL Indocypraea Orchard View in CoL , Exomiocarpon Lawalrée View in CoL , Fenixia Merr. View in CoL , Hoffmanniella Schltr. ex Lawalrée View in CoL , Pentalepis F.Muell. View in CoL ).

The very distinctive distribution of X. strumarium View in CoL in the Old World may be the results of a relatively recent long-distance dispersal event, as hypothesized also for Ambrosia maritima ( Martin et al. 2018) View in CoL . This could be explained by means of the dispersal strategy of Xanthium View in CoL burs, that include zoochory (as above mentioned), but also hydrochory and thalassochory. Takakura and Fujii (2010) demonstrated that burs of Xanthium View in CoL can be soaked in salty water for weeks and still be viable, a feature that could have favoured the arrival of Xanthium View in CoL to the Old World via sea. Also, mammal and bird migration routes from America to Eurasia ( Backensto et al. 2016; Yong et al. 2021) might have facilitated the spread of Xanthium View in CoL diaspores in Eurasia.

On the other hand, it is widely acknowledged that from the Miocene to the present, numerous alternations of cold and warm periods have impacted the palaeogeography of continents, leading to the cyclical connections and separations of landmasses through ice sheet formation and sea-level changes ( Brikiatis 2014; Hosner et al. 2015). These shifts resulted in the formation of different land bridges that facilitated animal and plant dispersal ( Hosner et al. 2015; Maguilla et al. 2018). When concerning connections between America and the Old World, a few important events need to be mentioned. The De Geer Land Bridge appeared during the Late Cretaceous and Early Palaeocene in two time windows around 69 million years ago (Ma) and 65.5 Ma, connecting North America with Eurasia through Greenland and Fennoscandia ( Brikiatis 2014). The Thulean Route, present from the Early Palaeocene until the Early Eocene, also connected North America with Eurasia via the British Isles and Greenland ( Brikiatis 2014). The Japan Land Bridges appeared multiple times due to sea level changes, connecting the Japanese archipelago to modern Russia during the Pleistocene ( Millien-Parra and Jaeger 1999). Finally, The Beringia Land Bridge, perhaps the most well-known, connected North America to Eurasia through Russia, Alaska, and Canada and appeared multiple times during the Palaeocene and most recently during the Last Glacial Maximum circa 30–20 thousand years ago ( Brikiatis 2014; Hoffecker et al. 2016).

With the present study, we want to test if the arrival of the diaspores of the ancestor of X. strumarium in the new continent was caused by a long-distance dispersal event or if alternatively, it was the result of a progressive dispersal through a land bridge (e.g., the Beringian Bridge). Additionally, we aim at addressing a few main questions concerning the biogeographic history of Xanthium strumarium . (a) What is the temporal framework in which the ancestor of the species arrived in the Old World? (b) What have been the possible pathways followed by the species to spread all-over the Old World? To answer these questions, we used herbarium specimens and applied target enrichment of nuclear genes to infer the phylogeography of the species. We used Bayesian methods to estimate its divergence time and ancestral range reconstruction to test different possible scenarios of the pathways followed by the genus Xanthium to reach and colonise the Old World.

2.5 | Xanthium strumarium Divergence Time View in CoL

We used the concatenated dataset with the 732 regions obtained in the previous step and 26 samples from all recognised species of the genus ( Table 1 View TABLE 1 ). Input files for Beast2 ( Bouckaert et al. 2019) were prepared using BEAUti v.2.7.6 ( Bouckaert et al. 2019) and the “beast” template. We used the GTR + G as sequence substitution model, letting Beast2 optimise model parameters. The “Random Local Clock” was selected as clock model. In order to obtain absolute divergence times, we followed the approach used in Tomasello et al. (2020) and gave both an informative prior on the clock rate and a calibration point. Accordingly, we gave the “clockrates” a uniform distribution (min: 5.0 e− 5, max: 5.0 e− 7) with an initial value of 5.0 e− 6. Assuming a standard substitution rate of 5 e− 9 in plants ( Wolfe et al. 1989), and since Xanthium plants are annuals, the clock rate will result in 5 e−6 mutations per site per thousand years (± an order of magnitude). As for the calibration point, we based it on the oldest Xanthium fossil, consisting of bur fragments found in Indiana ( USA) and dating back to the Upper Hemphillian/Blancan North American stages ( Farlow et al. 2001). Therefore, we applied to the root of the tree a lognormal prior distribution with mean 3.0 and standard deviation 2.9 (95% highest prior density ranging between 3000 and 8910 thousand years ago (ka)).

We ran two analyses for 100,000,000 generations, sampling every 10,000 iterations. Convergence between different analyses and effective sample size (ESS) were checked in Tracer v.1.7 ( Rambaut et al. 2018). The tree files from the two independent runs were combined using LogCombiner v.2.7 ( Bouckaert et al. 2019). Finally, a maximum clade credibility tree was calculated in TreeAnnotator v.2.6 ( Bouckaert et al. 2019), applying 10% burn-in, a posterior probability limit of 0.5, and “Mean Heights” for node heights. Additional sets of analyses were performed using only the clock rates prior or the fossil calibration. This was done to assess the effect of using both calibration and clock-rate priors in the same analysis.

2.7 | Phylogeny of the Xanthium strumarium View in CoL Species Complex

A total of 48 X. strumarium samples and six outgroups ( Table 1 View TABLE 1 ) were employed. Data processing and assembly was done in HybPhyloMaker using the aforementioned 2150 sequences as reference. Read mapping was achieved using BWA v0.7.16a ( Li and Durbin 2009), with the addition of a mismatch penalty of 8 to make reads more stringently map to the reference. Consensus sequences were generated in ConsensusFixer as it was done for the dating. BLAT was utilised for the alignment of the mapped reads with the targeted exons to generate PSLX files. The minimum sequence identity between the probe and the sample (termed ‘minident’) was set to 99. The subsequent steps were carried out as for the age estimation analyses. Further putative paralog sequences were excluded by employing the “HybPhyloMaker4a2_selectNonHet.sh” script and designating a maximum of 5 heterozygous sites per locus (“maxhet” in the HybPhyloMaker settings file). Missing data filtering was again performed as for the age estimation analyses.

A total of 744 loci ( Table S1b View TABLE 1 ) were retrieved, concatenated, and used to generate an ultrametric Bayesian phylogenetic tree. The analyses were conducted using the Beast2 software, employing the “Optimized Relaxed Clock”, the GTR + G as substitution model, and the “Birth-Death Model” tree model. To obtain absolute divergence times, the crown age of X. strumarium was calibrated using the age obtained in the above analysis (i.e., it was fixed to the estimate obtained in the age estimation analyses). Two independent analyses were run for 100,000,000 generations, sampling every 10,000 iterations. Convergence and effective sample size (ESS) were checked in Tracer v.1.7. The tree files from the two independent runs were combined using LogCombiner v.2.7 ( Bouckaert et al. 2019), and a maximum clade credibility tree was calculated in TreeAnnotator v.2.6 applying 10% burn-in, a posterior probability limit of 0.5, and “Mean Heights” for node heights.

3.3 | Phylogeny of X. strumarium View in CoL

The Bayesian analysis ( Figure 4A View FIGURE 4 ), revealed various well supported clades and showed a clear geographic pattern. The earliest diverging clade included samples from India and Sri Lanka, followed by a split in which samples from East Asia and Mauritius separated from other clades (posterior probability of 0.99). In the latter clades, sample X353, from Siberia, is found to be sister to the rest of the samples, which are divided into three other clades: a first including samples collected in the Horn of Africa (posterior probability of 1.0); a second counting specimens originated in the Mediterranean (along with the two Brazilian samples; posterior probability of 0.99) and a third including specimens from the continental Europe, and X355, collected in Ethiopia (posterior probability of 1.00).