The evolutionary history of the Arabidopsis lyrata complex: a hybrid in the amphi-Beringian area closes a large distribution gap and builds up a genetic barrier

Abstract

Background: The genomes of higher plants are, on the majority, polyploid, and hybridisation is more frequent in plants than in animals. Both polyploidisation and hybridisation contribute to increased variability within species, and may transfer adaptations between species in a changing environment. Studying these aspects of evolution within a diversified species complex could help to clarify overall spatial and temporal patterns of plant speciation. The Arabidopsis lyrata complex, which is closely related to the model plant Arabidopsis thaliana, is a perennial, outcrossing, herbaceous species complex with a circumpolar distribution in the Northern Hemisphere as well as a disjunct Central European distribution in relictual habitats. This species complex comprises three species and four subspecies, mainly diploids but also several tetraploids, including one natural hybrid. The complex is ecologically, but not fully geographically, separated from members of the closely related species complex of Arabidopsis halleri, and the evolutionary histories of both species compexes have largely been influenced by Pleistocene climate oscillations.

Results: Using DNA sequence data from the nuclear encoded cytosolic phosphoglucoisomerase and Internal Transcribed Spacers 1 and 2 of the ribosomal DNA, as well as the trnL/F region from the chloroplast genome, we unravelled the phylogeography of the various taxonomic units of the A. lyrata complex. We demonstrate the existence of two major gene pools in Central Europe and Northern America. These two major gene pools are constructed from different taxonomic units. We also confirmed that A. kamchatica is the allotetraploid hybrid between A. lyrata and A. halleri, occupying the amphi-Beringian area in Eastern Asia and Northern America. This species closes the large distribution gap of the various other A. lyrata segregates. Furthermore, we revealed a threefold independent allopolyploid origin of this hybrid species in Japan, China, and Kamchatka.

Conclusions: Unglaciated parts of the Eastern Austrian Alps and arctic Eurasia, including Beringia, served as major glacial refugia of the Eurasian A. lyrata lineage, whereas A. halleri and its various subspecies probably survived in refuges in Central Europe and Eastern Asia with a large distribution gap in between. The North American A. lyrata lineage probably survived the glaciation in the southeast of North America. The dramatic climatic changes during glaciation and deglaciation cycles promoted not only secondary contact and formation of the allopolyploid hybrid A. kamchatica, but also provided the environment that allowed this species to fill a large geographic gap separating the two genetically different A. lyrata lineages from Eurasia and North America. With our example focusing on the evolutionary history of the A. lyrata species complex, we add substantial information to a broad evolutionary framework for future investigations within this emerging model system in molecular and evolutionary biology.

Figure 1

Distribution of Arabidopsis accessions investigated. Accessions of the Arabidopsis lyrata complex were analysed for nuclear ITS and PgiC regions, and cpDNA trnL/F. Accessions of the Arabidopsis halleri complex were analysed for PgiC only. Maximal glaciation of the LGM is drawn according to Ehlers and Gibbard [56].

Figure 2

Distribution of cpDNA trnL/F suprahaplotypes in the Arabidopsis lyrata complex. Data newly presented in this study were combined with previous results from Koch and Matschinger [27] and Schmickl et al. [28]. TrnL/F suprahaplotypes were characterised as trnL intron and trnLF-IGS excluding the pseudogene-rich region in the trnLF-IGS. Accessions from Austria are listed separately. Maximal glaciation of the LGM is drawn according to Ehlers and Gibbard [56].

Figure 3

CpDNA trnL/F suprahaplotype network of the Arabidopsis lyrata complex. Data newly presented in this study were combined with previous results from Koch and Matschinger [27] and Schmickl et al. [28]. TrnL/F suprahaplotypes were characterised as trnL intron and trnLF-IGS excluding the pseudogene-rich region in the trnLF-IGS. The sizes of the circles indicate the relative frequency of a suprahaplotype in the whole dataset.

Figure 4

Distribution of accessions with/without PgiC1 amplification in the Arabidopsis lyrata complex and A. halleri. Amplification was successful in Arabidopsis halleri ssp. gemmifera from China and Japan, and in Central European subspecies of the Arabidopsis halleri complex (ssp. dacica, ssp. halleri, and ssp. tatrica). Amplification was also successful in all A. kamchatica accessions. Maximal glaciation of the LGM is drawn according to Ehlers and Gibbard [56].

Figure 5

Distribution of ITS supratypes in the Arabidopsis lyrata complex. Data newly presented in this study were combined with previous results from Koch and Matschinger [27] and Schmickl et al. [28]. ITS supratypes were characterised as ITS1, 5.8 S rDNA, and ITS2 region, with ambiguous sites replaced by the bases with higher fluorescence intensity in the electropherogramm. Sequences with equal fluorescence intensity of the two bases at the ambiguous positions were only found between ITS supratypes b and e and labelled b/e ambiguous. The ITS supratypes are corresponding to those shown in Figure 6. Maximal glaciation of the LGM is drawn according to Ehlers and Gibbard [56].

Figure 6

Strict consensus tree from the maximum parsimony analysis of ITS supratypes in the Arabidopsis lyratacomplex and Arabidopsis halleri ssp. gemmifera. From altogether 10 ITS supratypes (with ambiguous sites replaced by the bases with higher fluorescence intensity in the electropherogramm) of both the A. lyrata complex and A. halleri ssp. gemmifera, a strict consensus tree (length = 43) was constructed using maximum parsimony with MEGA version 4.1 [60]. Heuristic searches were performed with 10 random addition sequences and Closest Neighbour Interchange (CNI) branch swapping. Bootstrap values were calculated based on 500 replicates. Consistency index (CI) = 0.89, retention index (RI) = 0.96. Arabidopsis thaliana was used as outgroup.