The Evolutionary History of the Arabidopsis arenosa Complex: Diverse Tetraploids Mask the Western Carpathian Center of Species and Genetic Diversity

The Arabidopsis arenosa complex is closely related to the model plant Arabidopsis thaliana. Species and subspecies in the complex are mainly biennial, predominantly outcrossing, herbaceous, and with a distribution range covering most parts of latitudes and the eastern reaches of Europe. In this study we present the first comprehensive evolutionary history of the A. arenosa species complex, covering its natural range, by using chromosome counts, nuclear AFLP data, and a maternally inherited marker from the chloroplast genome [trnL intron (trnL) and trnL/F intergenic spacer (trnL/F-IGS) of tRNALeu and tRNAPhe, respectively]. We unravel the broad-scale cytogeographic and phylogeographic patterns of diploids and tetraploids. Diploid cytotypes were exclusively found on the Balkan Peninsula and in the Carpathians while tetraploid cytotypes were found throughout the remaining distribution range of the A. arenosa complex. Three centers of genetic diversity were identified: the Balkan Peninsula, the Carpathians, and the unglaciated Eastern and Southeastern Alps. All three could have served as long-term refugia during Pleistocene climate oscillations. We hypothesize that the Western Carpathians were and still are the cradle of speciation within the A. arenosa complex due to the high species number and genetic diversity and the concurrence of both cytotypes there.


Introduction
Pleistocene climatic fluctuations strongly shape the evolutionary history of new species (e.g., grasshoppers [1]) and the distribution of genetic variants at the population level within species (e.g., Arabidopsis [2] and hominins [3]). In numerous studies of the European flora and fauna, these aspects of Pleistocene climatic oscillations have been investigated, but mainly confined to a single species. In this study we consider a species complex of wild relatives of the model plant Arabidopsis thaliana (L.) Heynh. with various taxa and cytotypes to investigate the hypothesis that past and ongoing gene flow between taxa and ploidal levels contributes to the ability of populations to adapt and survive in rapidly changing environments, particularly during the Pleistocene [4], [5].
The Arabidopsis arenosa species complex is one of three major species complexes within the genus Arabidopsis [6]- [10], formerly treated as Cardaminopsis [6], [7]. Arabidopsis arenosa and its segregates might not only represent the most ancestral species complex compared to the other two major species lineages, namely Arabidopsis lyrata (L.) O'Kane & Al-Shehbaz and A. halleri (L.) O'Kane & Al-Shehbaz [9], but the various taxa also harbor greater genetic diversity than any other Arabidopsis species and show a remarkable broad spectrum of ecological adaptations from high alpine regions in the High Tatras to sand dune vegetation in Scandinavia [9], [10]. A taxonomic overview of the various taxa of the A. arenosa species complex, including ploidal level and geographic distribution, is provided in Table 1. Arabidopsis arenosa is a colline, montane, and subalpine species complex with a mainly Central European distribution range including parts of the Alps and Carpathians. Only a few studies have been attempted to unravel the evolutionary history of the A. arenosa complex [9], [10]. Several studies focused on the natural hybrid A. suecica (Fr.) Norrl., which is of allopolyploid origin with the maternal parent A. thaliana and a paternal parent from the A. arenosa species complex [11], [12], confirmed also by artificial crosses [13]. Polyploidisation, mainly tetraploidisation, is frequent in several taxa of the A. arenosa complex [14], indicating repeated independent polyploidisation events. Introgression, the stable integration of genetic material from one species into another through repeated backcrossing, was observed between members of the A. arenosa and A. lyrata complexes [5], [15].
There is an increasing interest in A. arenosa as a model system for adaptation to calcareous versus siliceous bedrocks (Koch and Widmer, ongoing studies; Bomblies et al., ongoing studies), or character trait research such as shade-tolerance (Bomblies et al., ongoing studies). Additionally, A. arenosa is interesting in terms of hybrid speciation, as it is the paternal parent of the natural allopolyploid A. suecica. Studies on the genomic consequences of hybridization are underway, and a first assembly of the A. arenosa genome is available (wiki.bioinformatics.ucdavis.edu/index.php/ Arabidopsis_arenosa_whole_genome_assembly). Working with A. arenosa as a model system needs careful consideration of the evolutionary history of the taxa one is investigating, particularly the distribution of natural genetic variation within and among taxa.
The following two aspects are the focus of our research: Unravelling the broad-scale cytogeographic and phylogeographic patterns of diploids and tetraploids: We are particularly interested in contact zones of populations with different or mixed ploidal levels, as they can indicate ongoing species differentiation. The second is detecting centers of genetic diversity: In the northern hemisphere late Quaternary climate oscillations, especially the last glacial maximum (LGM), about 26,500 to 19,000-20,000 years ago, had the most severe influence on present-day distribution and diversity of plant taxa. Arabidopsis arenosa is distributed both in regions that remained largely unglaciated during Pleistocene climate oscillations and in areas formerly covered by glaciers, making it well-suited for comparative studies of evolution in changing environments.

Plant material
The accession list is provided in Table S1. Geographic distribution of single accessions is shown in Fig. 1A. Ploidy information was obtained from 214 accessions (126 populations, one to eleven individuals with five flowers each), AFLP data from 356 accessions (275 populations, one to seven individuals each), and plastid trnL/F sequence data from 365 accessions (260 populations, one to eleven individuals each) (Table S1).

Mitotic chromosome preparations
Ploidy was determined both in this and a previous publication [5] and additionally from herbarium vouchers from the Herbarium of the Natural History Museum Vienna on which Polatschek had indicated chromosome numbers. The respective source of ploidy determination is recorded in Table S1. Cytological methods and light microscopy were applied according to Schmickl et al. [5].

DNA isolation and amplified fragment length polymorphisms (AFLPs)
Total DNA was obtained from dried leaf material and extracted according to a CTAB protocol [20] with modifications according to previous studies [4].
AFLP analysis was performed using a standard protocol [21] with the following modifications: Approximately 200-500 ng DNA was digested and ligated in a 15 ml reaction mix containing T4 ligase buffer and ATP solution (Bioline, USA), 50 mM NaCl, 0.75 mg BSA, 1.5 U T4 ligase (Bioline, USA), 1 U MseI and 5 U EcoRI (New England Biolabs, USA), 0.37 mM EcoRI adapter and 3.67 mM MseI adapter. The reaction mix was incubated for 3 h at 37uC, followed by an inactivation step for 10 minutes at 65uC. The restriction-ligation product was subsequently diluted tenfold.
In the pre-selective PCR 2.5 ml of the diluted restriction-ligation product was used in a total reaction volume of 12. . The reactions were held at 95uC for 5 min followed by 13 cycles of 94uC for 30 s, 65uCR56uC (20.7uC per cycle) for 1 min, and 72uC for 1 min, followed by 23 cycles of 94uC for 30 s, 56uC for 1 min, and 72uC for 1.5 min with a final 8 min extension at 72uC. Three differentially fluorescence labelled PCR products of the same sample were multiplexed and diluted, and the fragments were electrophoretically separated on a MegaBase 500 sequencer together with an ET-ROX 550 size standard (Amersham Biosciences, USA). For each run a total of 48 samples were analyzed, including one standard sample, one negative control, one repeat within the runs, and several other replicates (altogether 6.5%), as recommended by Bonin et al. [22]. Raw data were visualized and the fragments in the range of 60-513 bp manually scored using GeneMarker version 1.9 (SoftGenetics, USA). Processed data were exported as a presence/absence matrix.

TrnL/F amplification and sequencing
For the cpDNA markers trnL intron and trnL/F intergenic spacer (trnL/F-IGS), primers, PCR cycling scheme, purification of the amplified fragment, cycle sequencing, and sequencing on a MegaBace 500 sequencer followed the protocol of Schmickl et al. [4]. Amplified sequences of trnL/F-IGS included the complete trnL/F-IGS and the first 18 bases of the trnF gene.

AFLP genetic diversity statistics and Principal Component Analysis (PCA)
Several statistical parameters were computed using the R script AFLPdat [23], R 2.9.2 environment [24] for geographic and taxonomic groups: proportion of variable markers (FP) and Nei's gene diversity (H E [25]). The following seven geographic regions were considered: (1) Balkan Peninsula (Balk), (2) Carpathians (Carp), (3) unglaciated Eastern and Southeastern Alps (UnglaE-SEAlps), (4) glaciated Eastern Alps (GlaEAlps), (5) glaciated Western Alps (GlaWAlps), (6) unglaciated Central Europe (UnglaCentrEur), and (7) glaciated northern Europe (GlaNEur). These regions are illustrated in Figure 1. Regarding taxonomy, six entities were distinguished following Měsíček [14], [18] and Kolník [19]: A. arenosa subsp. arenosa, A. arenosa subsp. borbasii, A. carpatica, A. neglecta, A nitida, and A. petrogena. In order to analyze and display the similarity among the AFLP genotypes, a Principal Component Analysis (PCA) was performed using MVSP version 3.1 (Kovach Computing Services, UK). Pairwise Euclidean distance was applied as distance measure and, alternatively, both Jaccard and simple match coefficient [26].  [32]. The borders of the seven geographic regions are indicated (see also Table S1, where the affiliation of each accession to one of these regions is listed). A: Visualization according to taxonomy. Seven entities are distinguished: A. arenosa subsp. arenosa, A. arenosa subsp. borbasii, A. carpatica, A. neglecta, A. nitida, and A. petrogena, following Měsíček [14], [18] and Kolník [19], and Arabidopsis arenosa var. intermedia from the Alps. B: Visualization according to ploidal level (diploids and tetraploids). Ploidal level estimates were only available for a subset of accessions. C: Visualization according to chloroplast DNA suprahaplotypes. doi:10.1371/journal.pone.0042691.g001 Plastid trnL/F sequence definition, network analysis, genetic diversity statistics, and map reconstruction Plastid trnL/F sequences were defined as i) haplotypes and ii) suprahaplotypes following our previous studies (e.g., [4]). Haplotypes (i) are characterized by varying (in sequence and structure) trnF pseudogenes in the 39-region of the trnL/F-IGS close to the functional trnF gene: Haplotypes belonging to one suprahaplotype (ii) share the same base order throughout the whole sequence except for the pseudogene-rich region, where they vary in both length and base content. Mutation rate within the pseudogene-rich region is about 10 to 20 times higher than within the non-coding spacer and intron regions [27]. Therefore, our cpDNA dataset is based on trnL/F suprahaplotypes only. Suprahaplotypes differ from each other by single point mutations and/or indels. Newly defined trnL/F haplotypes were assigned to GenBank numbers [FJ477684-FJ477690, FJ477705-FJ477716] ( Table S1). The network was constructed using TCS version 1.21 [28] using the statistical parsimony algorithm [29]. Single gaps (except polyT stretches) were coded as single additional binary characters. Genetic diversity statistics were performed with Arlequin version 3.11 [30]: Genetic diversity was estimated as effective genetic diversity according to Gregorius (V a ) [31], nucleotide diversity p and Nei's unbiased gene diversity H E .
In order to visualize the geographical data, ArcView version 8.2 (ESRI, USA) was used. The maximum extent of the ice sheets during the LGM was taken from Ehlers and Gibbard [32].

Chromosome counts identify diploids exclusive to the Balkan Peninsula and the Carpathian Mountains
Two ploidal levels, diploid and tetraploid, were observed within our sampling (Fig. 1B). Diploids were exclusively found in southeastern and eastern Europe on the Balkan Peninsula, in northern Hungary, and in the Carpathians. In contrast, tetraploids have a large distribution range and occur from the Julian Alps (Slovenia) in the south, the Western Carpathians (Slovakia) in the east, France and Belgium in the west, and Scandinavia in the north. Several regions were reported as areas of recent, mainly anthropogenically influenced colonisation after 1890 [33] (Belgium, Finland, France, Great Britain, Greenland), frequently along railway tracks. The Balkan Peninsula and the Western Carpathians were the only regions where both diploid and tetraploid populations were found. Populations of mixed ploidal levels were not observed, but can not be completely ruled out, as only a limited number of populations from the Eastern Alps (n = 28) and the Western Carpathians (n = 9) were analyzed with more than one individual per population.
AFLP data indicate similar values for gene diversity throughout Europe and demonstrate high gene diversity of a widespread tetraploid Diversity statistics, based on AFLP data, showed similar values for gene diversity throughout the whole distribution range of the A. arenosa species complex (Table 2), ranging from H E = 0.133 (GlaNEur) to H E = 0.159 (UnglaESEAlps). The proportion of variable markers differed more strongly between geographic regions, ranging from FP = 0.411 (Balk) to FP = 0.906 (UnglaE-SEAlps). However, the proportion of variable markers is biased with respect to sample size (Balk: n = 9, UnglaESEAlps: n = 114) and, therefore, not a valid measurement of genetic diversity. Genetic diversity patterns in the A. arenosa complex need additional consideration in terms of taxonomy, as for certain regions, e.g., Carp, numerous taxa are described, and in other regions, e.g., GlaNEur, only one taxon is found. Gene diversity of the different taxa ranged from H E = 0.157 (A. arenosa subsp. arenosa) and H E = 0.155 (A. arenosa subsp. borbasii) to H E = 0.138 (A. carpatica and A. petrogena) and H E = 0.125 (A. neglecta). The proportion of variable markers is, again, highly correlated with sample size (A. neglecta: n = 6, FP = 0.296; A. arenosa subsp. arenosa: n = 277, FP = 0.985) and so should be treated with caution.
PCA according to regions ( Fig. 2A) resulted in overlapping groups of AFLP genotypes from nearly all regions, except for accessions from parts of Carp and UnglaESEAlps. Groups of AFLP genotypes according to taxonomy also largely overlapped (Fig. 2B). Widespread A. arenosa subsp. arenosa but also subsp. borbasii formed large groups of AFLP genotypes in comparison to A. carpatica, A. neglecta, and A. petrogena (A. nitida was omitted, as it was represented by one accession only). This finding underlines the high genetic plasticity of the tetraploids (A. arenosa subsp. arenosa and subsp. borbasii) in contrast to the mainly diploids [A. carpatica (exclusively diploid), A. neglecta (predominantly diploid), A. petrogena (partially diploid)]. PCA according to ploidal levels (Fig. 2C) revealed two partly overlapping clusters of diploids and tetraploids, but due to many accessions without ploidal level estimates the two clusters could actually be more strongly intermingled.
The Balkan Peninsula, the Carpathians and the unglaciated Eastern and Southeastern Alps are the three centers of chloroplast sequence diversity of the A. arenosa species complex Based on trnL/F sequence data, we detected three centers of genetic diversity of the A. arenosa complex ( Table 2). The unglaciated Eastern and Southeastern Alps displayed highest effective diversity according to Gregorius (V a = 4.14), highest nucleotide diversity (p = 0.483%), and highest Nei's gene diversity (H E = 0.764) within the whole dataset. The Balkan Peninsula was detected as the second center of genetic diversity (V a = 2.94, p = 0.395%, H E = 0.733). The third center of genetic diversity, the Carpathians, was characterized by gene diversity values similar to those of the Balkan Peninsula (V a = 3.49, p = 0.352%, H E = 0.719). In contrast to these three genetically highly diverse regions, which remained largely unglaciated during Pleistocene climate oscillations, formerly glaciated regions showed reduced values of effective genetic diversity according to Gregorius, nucleotide and Nei's gene diversity: The part of the Eastern Alps formerly covered by glaciers was characterized by V a = 2.23, p = 0.310%, and H E = 0.561. The formerly glaciated Western Alps showed strongly reduced values (V a = 1.24, p = 0.109%, H E = 0.205). Glaciated northern Europe was characterized by V a = 1.81, p = 0.315%, and H E = 0.491. Although Central Europe remained largely unglaciated during Pleistocene climate oscillations, genetic diversity was also reduced (V a = 1.64, p = 0.241%, H E = 0.403). In contrast to AFLP data, effective genetic diversity according to Gregorius, nucleotide and Nei's gene diversity of diploid A. carpatica (V a = 3.71, p = 0.514%, H E = 0.751) exceeded that of tetraploid A. arenosa subsp. arenosa (V a = 3.48, p = 0.410%, H E = 0.714).

Discussion
The Western Carpathian Mountains are the cradle of speciation within the A. arenosa complex Speciation is often accompanied by polyploidisation [34]. Within the A. arenosa complex two ploidal levels, diploid and tetraploid, were observed. Contact zones of these two cytotypes were localized in the northwestern part of the Balkan Peninsula and in the Western Carpathians, and, consequently, at least one independent polyploidisation event can be assumed for each of these regions. Regarding the Western Carpathians, we observed a mosaic pattern of diploids and tetraploids, which is in congruence with Měsíček [14] (see also [19]). In the contact zones of diploids and tetraploids in the Western Carpathians we found no populations with mixed ploidal levels. This finding can be discussed with respect to the origin of the tetraploids: Populations of mixed ploidal levels could either be the result of a relatively recent autopolyploidisation event or the result of secondary contact of formerly allopatric populations with different ploidal levels. The genus Melampodium L., in particular the white-rayed species complex with mainly diploids and tetraploids, is an example for postglacial formation of polyploids via autopolyploidisation. Intrapopulational cytotype mixture was reported, but emphasized as rare [35]. In contrast, secondary contact of formerly allopatric populations with different ploidy is the explanation for populations of mixed ploidal levels (diploids and tetraploids or tetraploids and hexaploids) in the Knautia arvensis agg. [36]. Therefore, we conclude, that the lack of mixed ploidy populations and the lack of uneven ploidy and aneuploidy in the A. arenosa complex excludes recent polyploidisation events as well as secondary contact zones. We assume ancient polyploidisation, probably dated several glacial cycles ago. Slightly different ecological adaptations of diploids and tetraploids might have favored such a distinct pattern of diploid and tetraploid populations. Diploid A. neglecta subsp. neglecta, for example, is found on siliceous substrates, mostly in high alpine habitats above the tree line, where it grows along mountain streams. The tetraploid subspecies robusta (previously recognized as Cardaminopsis neglecta subsp. robusta [18], but never validated by publication), corresponds to a taxon, which, although it also occurs on siliceous substrates, is found in different mountain ranges and at lower altitudes compared to typical subspecies (mostly around or below the tree line [Kolník, unpubl. data]). This is comparable to Senecio  carniolicus Willd. with an altitudinal, ecological gradient composed of mainly diploid and hexaploid populations in the Eastern Alps [37].
Additionally, the high genetic diversity, based on plastid trnL/F, in the Western Carpathians is an indicator for past and also ongoing speciation within the A. arenosa complex. Not only the high genetic diversity but also the high number of species and subspecies underline the Western Carpathians as a cradle of A. arenosa speciation. Numerous proposed taxa are unique for the Carpathians: A. neglecta is considered to comprise two subspecies, one diploid (subsp. neglecta) and one tetraploid (subsp. robusta) [18;Měsíček,unpubl. data]. Arabidopsis petrogena is exclusively described from the Carpathians as diploid (subsp. petrogena) and as tetraploid (subsp. exoleta [18;Měsíček,unpubl. data]). Arabidopsis nitida is the third unique taxon of the Carpathians. And within broadly defined A. borbasii a Carpathian subspecies (subsp. carpatica, referred to as A. carpatica, nom. prov. [19], and as C. borbasii subsp. carpatica [18]) is discussed. However, taxonomy of this highly diverse species complex needs to be revised in the near future.

Long-term evolution in two glacial refugia: the Carpathians and the unglaciated Eastern/Southeastern Alps
In numerous studies of both plant and animal species three classical LGM refugia were reported, based on the fossil record [38] and species and genetic diversity [39]: the Balkan Peninsula, the Iberian Peninsula, and the Appenin. Out of these three Pleistocene refugia, the Balkan Peninsula was emphasized as the most important refugium, especially for tree species [40], but also for upper and lower montane taxa of especially eastern European distribution [41], [42]. The Balkan Peninsula probably served as one of the major refugial areas for the A. arenosa complex, which will be discussed in more detail in the following section. Although the Carpathians remained largely unglaciated during Pleistocene climate oscillations, except in the south, they have been reported as a refugium for only a few plant species so far, including temperate trees [39], [43] and herbs, e.g., Campanula alpina Jacq. [44]. For the A. arenosa complex, the Western Carpathians were already described as a cradle of speciation, and they probably served as a second major refugial area. In contrast, the unglaciated eastern Austrian Alps have long been assumed as a glacial refugium. Along the eastern border of the Austrian Alps a cryptic refugium for tree species was already suggested [45], according to palynological data and supported by phylogeographic analyses [46]. Additionally, the northeastern Limestone Alps have already been described as rich in subalpine endemics [47], and this view was further supported by Tribsch and Schönswetter [48]: Based on species diversity and the number of endemics, they suggested refugia for numerous calcicolous and silicicolous plants. As A. arenosa is both a colline-montane and subalpine species complex, only parts of the subalpine refugia of the northeastern Limestone Alps overlapped with the A. arenosa refuge area of the unglaciated Eastern and Southeastern Alps. This is in congruence with the view that the eastern edge of the northeastern Limestone Alps is an area of periglacial survival for montane plant species [49], [50].

Parallel evolution in the Eastern Alps and the Western Carpathians
According to trnL/F sequence data, the unglaciated Eastern/ Southeastern Alps and the Carpathians formed two distinct genetic groups: the Alps characterized by suprahaplotype B and the Carpathians by suprahaplotype L. Although other suprahaplotypes, e.g., A and E, were shared between these two mountain ranges, we assume strong barriers to gene flow between the Alps and Carpathians, due to the Pannonian Basin, which constituted a lowland barrier for montane to subalpine taxa since the Holocene warming. Long-term genetic isolation between the Alps and Carpathians was also proposed for several other plant species, such as Campanula alpina [44]. Additional studies from Ranunculus glacialis L. [51] and Rosa pendulina L. [52] support this view.
However, an alternative explanation for genetic differentiation between A. arenosa populations from the Alps and Carpathians has to be considered. Based on our karyologic data, independent colonisation of both mountain ranges from the Balkan Peninsula can be hypothesized: Compared to all other regions, exclusively diploid cytotypes (except for one accession in the northern part) were found on the Balkan Peninsula. Diploid A. arenosa could have migrated northwards and experienced multiple polyploidisation events, especially in the Carpathians, where taxa of both diploid and tetraploid cytotype are described. The Alps were probably colonised by (a) tetraploid cytotype(s) of A. arenosa, which could have originated in the Julian Alps, where only tetraploids were found. The Balkan Peninsula as the ancient refuge area for the A. arenosa species complex is additionally supported by the occurrence of an Arabidopsis species endemic to the Balkan Peninsula, Arabidopsis croatica. This endemic is closely related to the A. arenosa complex, based on ITS sequence data (internal transcribed spacer region of nuclear encoded ribosomal DNA) [9], [10].
Taxa in refuge areas like the Eastern Alps and the Western Carpathians probably underwent long-term adaptational processes, which could result in adaptation to the same ecological niche in the two mountain ranges in parallel. Indeed, we found one example for parallel evolution in the Eastern Alps and the Western Carpathians: Diploid A. neglecta grows on siliceous bedrock along mountain streams in alpine habitats of the High Tatras (Slovakia), a similar ecological niche in the Eastern Alps (Wölzer Tauern, Styria) is occupied by a tetraploid taxon corresponding to Cardaminopsis arenosa var. intermedia (Kovats) Hayek [17]. Besides sharing similar ecological demands, the two taxa are discussed to be morphologically more similar to each other than to any other member of the A. arenosa complex. However, AFLP data did not confirm that the vicariant populations from the Eastern Alps and the High Tatras represent a single species (Fig. 2B). This needs further experimental confirmation based on a broader sampling of both taxa. Interestingly, two other species from the Brassicaceae co-occur in the same or neighbouring alpine regions on siliceous bedrock along mountain streams: hexaploid Cochlearia tatrae Borbás in the High Tatras (Slovakia) and diploid Cochlearia excelsa Zahlbr. ex Fritsch in the Eastern Alps (Seckauer Tauern, Styria). In parallel to A. neglecta and C. arenosa var. intermedia, these two Cochlearia species evolved independently and most likely within the last approximately 100,000 years [53], [54]. Parallel evolution of species pairs implies effective reproductive isolation between the two mountain ranges and limited multiple immigration.
Tetraploid A. arenosa subsp. arenosa is a highly genetically diverse taxon Based on both AFLP and chloroplast sequence data, genetic diversity was not strongly reduced in regions formerly covered by glaciers (GlaEAlps, GlaWAlps, GlaNEur) in comparison to formerly unglaciated regions (UnglaCentrEur). In all these four regions, exclusively tetraploid A. arenosa subsp. arenosa and subsp. borbasii were found. Local, periglacial survival could serve as one explanation for the high genetic diversity of especially A. arenosa subsp. arenosa. In Central Europe, numerous populations are restricted to relict habitats on exposed rocks in low mountain ranges, e.g., the Black Forest, the Eifel, the Elbe Sandstone Mountains, the Harz Mountains, and the Swabian Mountains, where they often co-occur with Pleistocene relict species such as Dianthus gratianopolitanus Vill. [Koch, unpubl. data]. A second explanation could be the lack of genetic bottlenecks and the maintenance of large effective population sizes during postglacial migration into formerly glaciated regions, probably enhanced by the plant's biennial life cycle. In a third alternative, gene flow between different taxa and/or ploidal levels could account for the high genetic diversity in especially tetraploid A. arenosa subsp. arenosa, probably before this taxon migrated from the Eastern Alps and Western Carpathians, its putative refuge areas, to Central Europe. Gene flow between different taxa of the A. arenosa species complex is documented by numerous hybrids, including triploid ones, reported from the Western Carpathians [19]. Gene flow between different ploidal levels was recently described for A. lyrata [15]: Based on an isolation with migration model analysis [55], [56], gene flow from diploids to tetraploids and vice versa was hypothesized. We assume that all three factors could have contributed to the high genetic diversity of tetraploid A. arenosa, which could have resulted in its ability to be a successful coloniser on various different substrate types (e.g., limestone, sandstone, granite, basalt) in various different habitat types (''natural'' sites on rocks, gravel, and sand; anthropogenically influenced sites such as railway tracks and rural areas).