Speciation and biogeographical history in the European Alpine System: the radiation of <fi>Ranunculus</fi> sect. <fi>Euromontani</fi> (Ranunculaceae)
Biogeographical analysis of species of Ranunculus sect. Euromontani and related taxa.
Biogeographical analysis of species of Ranunculus sect. Euromontani and related taxa.

High mountain systems are ecosystems with a high biodiversity and species richness. Mountain biota are shaped by complex climatic and geological histories and provide diverse environments (Nagy & Grabherr, 2009). Plant speciation in mountain systems is enhanced by various processes, often resulting in radiations. Geographical isolation through topographical but also climatic barriers can result in allopatric speciation within and between mountain chains (Kadereit et al., 2004; Smycka et al., 2022). Allopatric speciation can be vicariant, by emerging geographical barriers, or peripatric, by dispersal over existing barriers (Coyne & Orr, 2004). Ecological speciation can happen along altitudinal gradients, as well as through adaptation to different bedrock types (Smycka et al., 2022). The dynamics of climatic oscillations, for example of the Pleistocene glaciations, however, have also caused geographical range fluctuations of species. While ice shields forced plants to retreat into marginal glacial refugia, retreat of glaciers allowed for re-colonization and secondary contact of previously isolated lineages (Hewitt, 2004; Pittet et al., 2025). Secondary contact hybridization in plants may result in hybrid zones or in rapid speciation, especially if combined with polyploidy (allopolyploidy; Abbott et al., 2013). Polyploidy also allows for sympatric speciation by establishing crossing barriers between cytotypes (Abbott et al., 2013; Wagner et al., 2021), and is often accompanied also by different niche dynamics, with a tendency of polyploids to be found in colder climates (Rice et al., 2019). The many possibilities for in situ speciation often result in high numbers of endemic species (Tribsch & Schönswetter, 2003; Vargas, 2003; Mráz & Ronikier, 2016; Steinbauer et al., 2016; Kadereit, 2017; Muellner-Riehl, 2019).
The European Alpine System (EAS; Ozenda, 2009) is of great interest for studying diverse evolutionary processes and plant speciation mechanisms (Comes & Kadereit, 2003). It is part of the Alpidic-Himalayan orogenic belt that had its uplift in several waves from the Mesozoic to the late Cenozoic (Ager, 1975). In Europe, the mountain chains range from the Iberian mountains (Cantabrian mountains, Pyrenees) to the European Alps, the Apennines, the Carpathians, and the mountains of the Balkan Peninsula; geologically these chains continue eastward to Anatolia. Southward also the Sierra Nevada and the North African Atlas belong to the system (Fig. 1). All these mountain chains reach the alpine vegetation belt, with the highest peaks reaching at least 2500 m above sea level, and show a pronounced vegetation zonation. The mountain chains of the EAS share a long-term geological history but were largely established as high mountain system mostly before the diversification of plant groups, which happened in most plant genera from the Miocene to the Pleistocene (Comes & Kadereit, 2003; Vargas, 2003; Kadereit et al., 2004; Hörandl & Emadzade, 2011; Boucher et al., 2016; Kadereit & Abbott, 2021; Smycka et al., 2022; Španiel & Resetnik, 2022). However, each of the mountain chains has a distinct topological and climatic profile according to the geographical position and climatic history. The Alps and Carpathians are situated within the continent in the temperate zone and form a mountain chain without major geographical gaps (Janiczek et al., 2025). The Carpathians have a pronounced geographical structure and were glaciated only in the highest regions (Ronikier, 2011). The Alps were most heavily influenced by Pleistocene glaciations and covered by ice shields during cold periods, with marginal and central refugial areas (Stehlik, 2000; Tribsch & Schönswetter, 2003). The southern mountain chains are separated from each other by the Mediterranean Sea and represent high alpine islands in the otherwise Mediterranean climate regime of the surrounding lowlands (Vargas, 2003). They were, to a lesser extent, glaciated in the Pleistocene and are often regarded as glacial refugia for Central European plants (Vargas, 2003; Španiel et al., 2023). With these features, the EAS provides manifold preconditions for allopatric and sympatric speciation processes and also for adaptive, geographical, or climatic radiations (Simoes et al., 2016).

Distribution map of samples of Ranunculus sect. Euromontani used in this study. The numbers correspond to sampling information in Supporting Information Table S1. Distribution areas of species (except for the outgroups) are indicated by hatched areas in the respective color and line type, after data in Jalas & Suominen (1989) and Davis (1965). The names follow the taxonomic conclusions of this paper. The map was created with QGIS (Moyroud & Portet, 2018).
Ranunculus sect. Euromontani Hörandl is an example of such an EAS radiation. It represents a distinct clade within the cosmopolitan genus Ranunculus L. (Paun et al., 2005; Emadzade et al., 2011; Emadzade & Hörandl, 2011; Hörandl & Emadzade, 2012). The group was long recognized informally as ‘Ranunculus montanus group’ for the Alps and Pyrenees (Landolt, 1954, 1956), and later on expanded in Flora Europaea to include c. 14 species (Tutin & Akeroyd, 1993). Molecular phylogenetic analyses rendered the section monophyletic but refined its circumscription by excluding R. sect. Oreophili Tschuradze, a separate clade of mountain plants that is mostly distributed in the Caucasus and Irano-Turanian mountains (Hörandl & Emadzade, 2012). Ranunculus sect. Euromontani appeared phylogenetically most closely related to Mediterranean lowland taxa (Emadzade et al., 2011) and diversified in the whole EAS, with species occurring in all mountain chains (Jalas & Suominen, 1989). Landolt (1956) speculated that two species described from North Africa (Ranunculus aurasiacus Pomel and Ranunculus dyris (Maire) H.Lindb.) also should belong to the group but without detailed investigation. The biogeographical and evolutionary history of the section within the EAS, however, has not been analyzed so far.
The reason for this lack of information was largely due to unclear species delimitation. Species boundaries were unresolved in previous phylogenies because of restricted sampling and the use of only a few traditional plastid and nuclear markers (Paun et al., 2005; Emadzade et al., 2011; Hörandl & Emadzade, 2012). The morphological distinction of taxa is difficult, as all species have a relatively similar mountain plant habitus of c. 5–15(50) cm height, rhizomes with stringlike roots, three- to five-lobed basal leaves, narrow-segmented stem leaves, an apically hairy receptacle, and typical yellow buttercup flowers (Hörandl & Emadzade, 2012). Differences between taxa appear in leaf shape, indumentum, size and shape of achenes (Supporting Information Fig. S1), but these characters often exhibit a high and overlapping variation (Landolt, 1954, 1956). All species are sexual outcrossers with unspecific generalist flowers. They occur in similar habitats in subalpine and alpine grassland vegetation, with only some differentiation according to bedrock conditions (Landolt, 1954, 1956). For these reasons, different taxonomic opinions and species circumscriptions existed in various regional floras, and many authors classified the species just as subspecies (synonyms in Jalas & Suominen, 1989). The taxonomy and biogeography is, as in other alpine plants, not yet clear (Landolt, 2006). Both diploid (2n = 16) and tetraploid (2n = 32) cytotypes were recorded in the group (Tutin & Akeroyd, 1993), which raised the question of whether sympatric speciation via allopolyploidy had also played a role in this radiation.
Here we present a phylogenomic approach based on target enrichment of nuclear genes originally designed for Ranunculus (Tomasello et al., 2020) to infer phylogenetic relationships and species delimitation. Based on a robust phylogenetic framework, we reconstruct the biogeographical history of the clade to test for allopatric speciation processes and the time frame of diversification. By complementary flow cytometric analyses of ploidy levels, genetic structure, and genetic clustering analyses, we test hypotheses of allopolyploid or autopolyploid speciation processes. Since molecular data suggested novel species circumscriptions in the Carpathians and in the Pyrenees, we also performed morphometric analyses on these taxa to support a revised classification.
We collected a total of 78 samples from across the section's distribution range, along with four samples from closely related Ranunculus sections as outgroups (Fig. 1; Table 1a). Outgroup selection (Ranunculus millefoliatus Vahl and Ranunculus velutinus Ten. from R. sect. Ranunculastrum DC., and two Ranunculus breyninus Crantz samples from R. sect. Oreophili Tschuradze) followed the phylogeny of Hörandl & Emadzade (2012). Of the 82 samples, some were field-collected and silica-gel-dried, whereas others were obtained from herbarium collections (E, GOET, SAV, W, and WU). Two samples (WU0286443, 5904) were only used for morphometrics, one (11 444) only for flow cytometry. Samples with the IDs 13 258, 13 274, 13 277, 13 279, 13 300, 40 428, 40 436, 40 590, 40 758, 40 772, and 40 974, 40 975 are from Zeni et al. (2026). Six type specimens and original material were included to support taxonomic interpretation. Collection dates ranged from 1837 to 2025. We covered all taxa of the section according to Hörandl & Emadzade (2012). Because of an obvious similarity of the North African R. aurasiacus with R. demissus DC., a species recorded from S.E. Spain, the Balkan Peninsula, Turkey and Lebanon (Davis, 1965; Tutin & Akeroyd, 1993), we included also some samples of this taxon. Attempts to extract DNA from R. dyris samples from MPU failed due to historical chemical pretreatments of herbarium specimens. Since results suggested a more distant relationships of R. demissus and R. aurasiacus to the core sect. Euromontani, we refrained from including more North African samples in this paper.
| Species | FC (this study) | Chrom. number (2n) | Ploidy level (x) | Literature |
|---|---|---|---|---|
| R. aduncus Gren. & Godr. | 16 | 2 | Landolt (1954), Baltisberger & Hörandl (2016), Koutecký et al. (2026) | |
| R. apenninus (Chiov.) Pign. | 16 | 2 | Masci et al. (1987) | |
| R. carinthiacus Hoppe | 16 | 2 | Landolt, (1954), Lovka et al. (1971), Sopova & Sekovski (1981), Dickenmann (1982), Diosdado & Pastor (1992), Baltisberger & Hörandl (2016), Koutecký et al. (2026) | |
| R. carpaticus Herbich | 2/4 | 16/28 | 2/4 | Jalas & Suominen (1989)/Pashuk (1987) |
| R. gouanii Willd. (incl. R. ruscinonensis Landolt) | 2 | 16 | 2 | Landolt (1954, 1956); Goepfert (1974); Diosdado & Pastor (1992) |
| R. marschlinsii Steud. | 16 | 2 | Jalas & Suominen (1989); Baltisberger & Hörandl (2016) | |
| R. montanus Willd. s.str. | 32 | 4 | Landolt (1954), Goepfert (1974), Dickenmann (1982), Koutecký et al. (2026) | |
| R. pollinensis Chiov. | 2 | 16/32 | 2/4 | Baltisberger & Hörandl (2016)/Jalas & Suominen (1989) |
| R. gruinalis Schott et al. (= R. pseudomontanus Schur) | 2 | 16 & 32 | 2x & 4x | Ciocârlan (2000)?* |
| R. sartorianus Boiss. & Heldr. | 2 & 4 | 16 & 32 | 2 & 4 | Franzen & Gustavsson (1983); Baltisberger & Hörandl (2016) |
| R. tatrimontanus Manzo et al. (= R. pseudomontanus auct.) | 16/32 | 2/4 | Váchová (1976)/Skalińska et al. (1958) | |
| R. venetus Huter ex Landolt | 32 | 4 | Landolt (1954); Masci et al. (1987); Koutecký et al. (2026) | |
| R. villarsii DC. (= R. grenieranus Gren.) | 2 | 16 | 2 | Landolt (1954); Goepfert (1974); Dickenmann (1982) |
| R. demissus DC. | 16 | 2 | Diosdado & Pastor (1992) |
We used flow cytometry to estimate ploidy levels from silica-gel dried samples. Internal standards were selected according to known monoploid genome size (Leitch et al., 2019, see Methods S1). For each measurement, 0.5–1.0 cm2 of leaf tissue was placed in a 2-ml Eppendorf tube together with a steel bead and ground using a TissueLyser II (Qiagen, Hilden, Germany). Nuclei were isolated using Otto I buffer, and staining was performed with 4′,6-diamidino-2-phenylindole (DAPI) in an Otto II buffer (Otto, 1992; Doležel et al., 2007).
All measurements were conducted using a CyFlow Analyzer (Sysmex, Nordstedt, Germany) and analyzed with Cube16 v.1.6 (Sysmex). For each sample, the relative fluorescence intensity of DAPI-stained nuclei was recorded (Table S2). DNA content and ploidy levels were estimated based on the ratio between sample and standard fluorescence (Table S2); final interpretation was complemented by chromosome counts from the literature (Table 1). Results are presented in Table S2, and for species, in Table 1.
Leaf tissue for each sample was also used to extract genomic DNA. For each sample, 5 to 10 mg of leaf material was transferred into a 2-ml Eppendorf tube together with a sterilized steel ball and ground using a TissueLyser II (Qiagen).
For recently collected samples, DNA was extracted using the DNeasy Plant Mini Kit (Qiagen) following the manufacturer's instructions with slight modifications as described in Marinček et al. (2022). For herbarium specimens older than 100 yr, we conducted the extractions following an aDNA extraction method (Dabney et al., 2013; Gutaker & Burbano, 2017).
Equipment was sterilized using DNA Away (Thermo Fisher Scientific, Waltham, MA, USA) and pipettes were treated with the UVaClean UV Pipette Carousel (MTC Bio, Sayreville, NJ, USA) to minimize contamination. Fragment lengths were assessed on a 2% agarose gel. DNA concentration was measured using a DS-11 FX μVolume Spectrophotometer/Fluorometer (DeNovix, Wilgminton, DE, USA) with the DeNovix dsDNA High Sensitivity Kit.
Library preparations were implemented using the ‘NEBNext UltraExpress DNA Library Prep Kit’ for older or highly fragmented DNA, or the ‘NEBNext UltraExpress FS DNA Library Prep Kit’ (New England BioLabs, Ipswich, MA, USA) for recently collected samples requiring DNA fragmentation. A PCR was conducted after adapter-ligation for six to eight cycles, and the samples were barcoded using the ‘NEBNext Multipley Oligos for Illumina’ (96Unique Dual Index Primer Pairs; New England BioLabs) and purified with HighPrep magnetic beads (MagBio Genomics Inc., Gaithersburg, MD, USA).
Target enrichment was carried out using a Ranunculus custom bait kit (Tomasello et al., 2020). Six to eight indexed samples were pooled equimolarly and hybridized for 20 h at 65°C. Postcapture amplification was performed using P5 and P7 Illumina adapters and 2× KAPA HiFi HotStart ReadyMix (Roche, Basel, Switzerland).
Concentrations were measured using a DS-11 FX μVolume Spectrophotometer/Fluorometer and a DeNovix dsDNA High Sensitivity Kit, and quality control was performed using a QIAxcel (Qiagen) with a DNA High Resolution Kit (1200), a QX Size Marker (50 bp–800 bp, v.2.0), and a QX DNA Alignment Marker (15 bp–5 kb). The samples were equimolarly pooled and sequenced on a NovaSeq X Plus system (Illumina, San Diego, CA, USA) using a PE150 kit (2 × 150 bp), at Novogene GmbH (Munich, Germany).
The acquired sequences were first filtered for low quality and trimmed using the CAPTUS pipeline clean flag with standard settings (Ortiz et al., 2026). De novo assembly was performed using Megahit v.1.2.9 (Li et al., 2015) via the ‘captus_assemble’ module.
Nuclear contigs were mapped to the Ranunculus custom bait reference using the ‘captus_extract’ module, setting ‘--max_paralogs’ to 1, ‘--dna_min_identity’ to 92, and ‘--dna_depth_tolerance’ to 1.5. Alignments were generated using ‘captus_align’, setting ‘--max_paralogs’ to 0 and ‘--min_samples’ to 63 (c. 75% of samples), with the ‘--filter_method’ set to ‘informed’.
The obtained 1249 locus alignments were concatenated using AMAS (Borowiec, 2016). A maximum likelihood (ML) total-evidence tree was then constructed using IQ-Tree v.3.1.0 (Wong et al., 2025), using AIC to select the best model for each alignment partition, 1000 ultrafast bootstrap replicates (Minh et al., 2013; Hoang et al., 2018), and 1000 fast branch SH-aLRT tests (Anisimova et al., 2011).
A species tree was inferred under the multispecies coalescent using ASTRAL-III (Zhang et al., 2018). Individual gene trees for the 1249 loci were reconstructed in IQ-Tree using ModelFinder Plus (MFP, Kalyaanamoorthy et al., 2017), with branch support assessed by 1000 ultrafast bootstrap replicates and 1000 SH-aLRT tests. To reduce the influence of poorly supported relationships, branches with support values < 70 were collapsed using Newick Utilities (Junier & Zdobnov, 2010) before species-tree inference. The resulting gene trees were analyzed in ASTRAL-III with extensive branch annotations enabled and node support estimated as local posterior probabilities (LPP).
Since dichotomous trees cannot reflect well hybrid origins, we tested for putative allopolyploid origins of R. montanus Willd. and Ranunculus venetus Huter ex Landolt with PhyloNet analysis (Methods S2; Fig. S2; Table S3). To test for potential incongruence of maternally inherited plastomes with nuclear data, we used plastome reads to calculate a ML tree (Methods S3; Table S4) and present with IQ-Tree (Fig. S3).
Multispecies coalescent species delimitation analyses were performed using the speedemon v.1.1.0 (Douglas & Bouckaert, 2022) package in beast2 (Bouckaert et al., 2019). To reduce the computational burden, the analysis was performed using different datasets of 50 loci. In order to select the loci, gene trees were constructed for each locus using IQ-Tree with the same settings as for the total-evidence tree, and these were rooted using phyxrr (Brown et al., 2017). SortaDate (Smith et al., 2018) in Python 3 was then used to calculate the tip-to-root variation of each gene tree, the bipartition support, and the tree length, and five groups of 50 loci (250 in total) were selected for species delimitation estimation.
The input .xml files were then prepared using BEAUti v.2.7.6 (Bouckaert et al., 2019). The substitution models were set to GTR + G4, and a strict clock was enforced for all loci. We used the Yule-Skyline Collapse prior and set the ‘collapse weight’ to a gamma distribution with the default values (α = 2.0; β = 2.0). The Yule-Skyline Collapse birth rate was set to a normal distribution with a mean centered at 1.0e−4. The epsilon value was left at the default setting of 1.0e−4.
Two independent runs were performed for each locus set, each comprising 1 × e9 Markov chain Monte Carlo (MCMC) iterations and sampling every 50 000 generations. We checked the convergence of the MCMC in Tracer v.1.7 (Rambaut et al., 2018) to ensure that the two independent runs produced similar values for each parameter and that the effective sample size (ESS) was > 200.
The tree files were combined using LogCombiner v.2.7 (Bouckaert et al., 2019). Species delimitation analyses were then performed using both ClusterTreeSetAnalyser and Species Delimitation Analyzer (SDA; Jones et al., 2015). The analyses were initially conducted using the default collapse threshold (ε = 1.0e−4), and subsequently re-run under alternative ε values (1.0e−5–1.0e−3) to assess the sensitivity of species delimitation to this parameter. Cluster recovery was sensitive to the collapse threshold (ε). No stable clustering was observed under more conservative settings (ε = 1.0e−5 and ε = 1.0e−4), whereas a higher threshold (ε = 1.0e−3) resulted in the identification of eight clusters. This indicates that clustering outcomes depend strongly on the selected collapse threshold. For SDA, the flag ‘-collapse height’ was set accordingly, and a cutoff of 1.0 was applied. For each of the five independent datasets, a posterior similarity matrix was generated from the SDA output using a modified R script from Jones et al. (2015). Each matrix represents the posterior probability that pairs of individuals are assigned to the same cluster across sampled delimitations. To summarize the information, the five matrices were then averaged element-wise to obtain a consensus matrix, which was used to visualize clustering patterns and assess consistency of species assignments across loci.
Divergence times were estimated with beast2 (Bouckaert et al., 2019) using the concatenated alignment generated in the previous step. Input files for beast2 were prepared in BEAUti v2.7.6 using the standard template. The ML tree obtained from the phylogenomic analysis was used as the starting tree.
A GTR + G substitution model was applied. Three node calibrations were implemented based on previously published studies (Emadzade & Hörandl, 2011). Normally distributed priors were assigned to the three nodes, with the means corresponding to the reported mean ages and the SD being approximated from the 95% highest posterior density (HPD) intervals.
The root of the entire dataset, corresponding to the most recent common ancestor (MRCA) of all sampled taxa, was constrained using a normal prior with a mean of 7.87 million years ago (Ma) and a SD of 1.9; the stem age of R. breyninus was constrained using a normal prior with a mean of 7.15 Ma and a SD of 1.6; the crown age of R. section Euromontani was constrained using a normal prior with a mean of 2.32 Ma and a SD of 0.7. All calibrated nodes were constrained to be monophyletic (Emadzade & Hörandl, 2011). The tree is shown in Fig. S4.
Two independent MCMC runs were performed for 1 × 109 generations, sampling every 100 000 generations. Convergence and ESS were assessed in Tracer v.1.7 ensuring all parameters exceeded an ESS value of 200.
To infer the biogeographic history of R. sect. Euromontani, we used the biogeobears package (Matzke, 2018) implemented in R (R Core Team, 2021) to estimate ancestral geographic ranges based on the ultrametric phylogeny obtained from Beast2. We used here the R. demissus-aurasiacus clade as the closest relative of sect. Euromontani as outgroup, and the area coding shown in Fig. S5.
We evaluated multiple biogeographic models, including dispersal–extinction–cladogenesis (DEC; Ree & Smith, 2008), dispersal–vicariance analysis (DIVA; Ronquist, 1997), and Bayesian Inference of Historical Biogeography for Discrete Areas (BayArea; Landis et al., 2013), each tested both with and without a founder-event speciation parameter (‘+j’; Matzke, 2018).
Biogeographic areas were defined based on sample distributions and major mountain ranges. Model performance was compared using likelihood ratio tests (LRTs) for nested models (e.g. DEC vs DEC + J), and the best-fitting model was selected based on the Akaike information criterion (AIC).
Single-nucleotide polymorphisms (SNPs) were extracted from the locus alignments using a custom bash script. To ensure data quality, these alignments were manually inspected. Sites with excessive missing data, insertions/deletions, putative paralogous sequences, or singletons were removed before SNP selection, resulting in 1819 SNPs.
Following this filtering step, a random SNP selection procedure was implemented to obtain a final dataset of high-quality 942 unlinked SNPs in .str format using the script ‘Convert Fasta Set to Structure.py’ (available at https://github.com/dportik/Convert-fasta-alignments-to-Structure-format).
Population structure was inferred using the sparse non-negative matrix factorization (sNMF) function as implemented in the R package LEA (Frichot & Francois, 2015). Analyses were performed for K values ranging from 1 to 20, with 10 replicate runs for each K to assess consistency across runs. Cross-entropy values were calculated for each replicate and used to determine the optimal number of clusters. The best-supported K value was identified as the one minimizing cross-entropy. Individual ancestry coefficients were visualized using bar plots to illustrate genetic cluster assignment. Reticulate evolution and genetic clustering of species was further explored via NeighborNet networks in splitstree v.4.12.2 (Huson, 1998; Bryant & Moulton, 2004; Huson & Bryant, 2006) based on the SNPs (Methods S4).
In the Carpathians and in the Pyrenees, molecular data suggested different species delimitations than in Landolt (1956), Jalas & Suominen (1989), and Tutin & Akeroyd (1993), see Table 1(b). We conducted morphometric analyses for the suggested split of the Carpathian ‘R. pseudomontanus Schur’ into two species in the South and West Carpathians by using 10 putative diagnostic characters (Table S5) on 27 individuals from herbarium specimens of our molecular data, including also types and material from the locus classicus. For testing the differentiation of the two Pyrenean taxa R. gouanii Willd. and R. ruscinonensis Landolt, we used 25 individuals from herbarium specimens and 12 characters (Table S6). For both datasets, we performed principal coordinate analysis (PCoA, using Gower's coefficient for mixed datasets), principal component analysis (PCA), and tested the genetic grouping via discriminant analysis. We further tested for significant differences of groups for each character with nonparametric tests and showed variance of significantly different characters with jitter boxplots. All analyses were done using PAST vs 4 (Hammer, 2020).
Overall, all analyzed individuals could be consistently assigned to either diploid (2) or tetraploid (4) cytotypes based on their fluorescence ratios (Tables 1, 2). Ranunculus pollinensis Chiov., all samples of R. gouanii and R. gruinalis Schott et al., as well as most individuals of R. carpaticus Herb., were diploid, whereas a single individual of R. carpaticus was identified as tetraploid; for R. sartorianus Boiss. & Heldr. we found both 2 and 4 levels. For the species of the Alps, flow cytometry measurements confirmed previously known ploidy levels (Koutecký et al., 2026).
| Model | LnL | Num params | d | e | j | AIC | AIC_wt |
|---|---|---|---|---|---|---|---|
| DEC | −63.07 | 2 | 0.0070 | 1.0e−12 | 0 | 130.1 | 7.5e−06 |
| DEC + J | −51.01 | 3 | 1.0e−12 | 1.0e−12 | 0.0058 | 108 | 0.47 |
| DIVA | −65.57 | 2 | 0.012 | 1.0e−12 | 0 | 135.1 | 6.2e−07 |
| DIVA + J | −50.91 | 3 | 1.0e−12 | 1.0e−12 | 0.0067 | 107.8 | 0.52 |
| BayArea | −108.3 | 2 | 0.014 | 0.26 | 0 | 220.6 | 1.7e−25 |
| BayArea + J | −56.53 | 3 | 1.0e−07 | 1.0e−07 | 0.0089 | 119.1 | 0.0019 |
A total of c. 3.2 Gb of sequencing data were obtained. On average, 24 933 812 raw reads per sample were generated, ranging from 3515 860 to 102 209 064 reads. After quality control and adapter trimming, an average of 1.69% of reads was discarded, resulting in 24 525 409 reads per sample retained for mapping and downstream analyses. The CAPTUS pipeline recovered a total of 1249 single-copy nuclear loci, with a combined alignment length of 250 272 bp and 17 141 parsimony-informative sites, while 102 out of 130 loci were recovered for the chloroplast tree, for a concatenated length of 81 222 bp and 1704 parsimony-informative sites.
The ML analysis recovered a well-resolved phylogeny, with maximal support values (100% bootstrap) across the backbone (Fig. 2a). The earliest diverging clade (following the outgroups) comprised individuals of R. demissus and R. aurasiacus, forming a well-supported lineage (100% bootstrap) clearly distinct from all other samples, which represent the core of R. sect. Euromontani.

Phylogenetic trees and genetic structure of Ranunculus sect. Euromontani and related taxa. (a) Maximum likelihood tree of species of Ranunculus sect. Euromontani (clade with black dot) and related taxa based on 1249 nuclear gene loci. Asterisks denote type specimens; (b) Coalescent based-ASTRAL species tree; samples that changed the position in the tree topology compared with (a) are indicated by dotted lines. Node support values for both trees are indicated by squares: black = 100% bootstrap support, gray = 80–99%, and white = 70–79%. Nodes without squares had support values < 70%. (c) Genetic structure analysis based on 942 unlinked SNPs (rows corresponding to samples in a).
A second fully supported clade followed including samples from the Carpathians, with individuals from the West Carpathians (R. tatrimontanus Manzo et al.) as sister to the clade of R. carpaticus from the Southern and East Carpathians. The remaining samples formed a third major clade (85% bootstrap), showing a clear geographic structure. Within this clade, individuals from the Alps (R. venetus, R. aduncus Gren. & Godr., and R. villarsii DC.) and Pyrenees (R. gouanii) formed a strongly supported clade (100% bootstrap), which is sister to the clade, including R. montanus and R. carinthiacus Hoppe, Corsica (R. marschlinsii Steud.), Apennines (R. pollinensis, R. apenninus (Chiov.) Pign.), South Carpathians (R. gruinalis), and Balkans (R. sartorianus; 86% bootstrap; Fig. 2a). Altogether, 12 well-supported clades corresponding to described species were observed.
The ASTRAL-III species tree (Fig. 2b) recovered the same major lineages identified in the concatenated ML analysis. Ranunculus demissus and R. aurasiacus formed a strongly supported lineage that diverged early from the remaining members of the core group of R. sect. Euromontani (pp = 1). Likewise, Carpathian R. tatrimontanus and R. carpaticus were recovered as a distinct and fairly supported clade (pp = 0.76). All species formed well-supported clades (pp > 0.70). Although relationships among lineages were less well resolved, the ASTRAL analysis largely corroborated the topology obtained from the concatenated nuclear dataset.
The PhyloNet analyses revealed for R. montanus a hybrid origin from R. venetus and R. carinthiacus, for R. venetus from R. montanus and R. villarsii (Fig. S2).
The chloroplast ML tree was generally less resolved than the nuclear phylogeny, while retaining a fully supported backbone which did not show incongruence to the nuclear trees (Fig. S3). As in the nuclear analyses, R. demissus and R. aurasiacus formed a well-supported lineage clearly separated from R. sect. Euromontani. Among the remaining samples, only the R. gruinalis clade (90% bootstrap) was confirmed, while all other species were intermingled in clades on very short branches and with low support.
The SDA results in the ε = 1.0e−3 scenario, identified nine well-defined genetic clusters (within cluster similarity higher than between clusters), largely corresponding to the clades of predominantly diploid species (Fig. 3). Among the most strongly supported lineages, R. carpaticus and R. tatrimontanus from the Carpathians were resolved as two distinct and well-separated species. Ranunculus apenninus (Chiov.) Pignatti and R. pollinensis formed a single cluster. Ranunculus aduncus Gren. & Godr., R. villarsii DC., and R. carinthiacus Hoppe each formed well-supported and distinct groups. R. gouanii, which included samples previously identified as R. ruscinonensis, showed a cluster with decreasing similarity from east to west, similar to phylogenetic and genetic structure analysis. Geographical patterns of similarity appear also within the R. demissus-aurasiacus cluster and within R. sartorianus. By contrast, the tetraploid species R. montanus and R. venetus did not form well-supported clusters and showed affinities with multiple groups. This result is likely due to sampling effects of using a subset of 250 genes, which might not fully cover the complex structure and relationships of polyploid genomes.

Results of coalescent based species delimitation analyses in Ranunculus sect. Euromontani, based on a consensus matrix of five sets of 50 genes each. Pairwise posterior probabilities of individuals belonging to the same species are visualized as a similarity matrix, with black indicating 100% and white indicating 0.0% similarity. Well-supported clusters are indicated by a black frame.
The BEAST2 analysis recovered a topology largely congruent with the ML phylogeny, with most nodes receiving full posterior probability support, except for a few terminal branches (Fig. 4a). No major topological conflicts were observed between the two approaches. Divergence time estimates revealed the crown age of the ingroup estimated at c. 2.69 Ma (95% HPD: 1.34–3.55 Ma). The R. demissus-aurasiacus-clade, however, separated c. 2 Ma, long before the radiation of R. sect. Euromontani (crown age 1.39 Ma; 95% HPD: 0.93–1.85 Ma; dates for all nodes are given in Fig. S4).

Biogeographical history of Ranunculus sect. Euromontani and related taxa. (a) Biogeographical analysis of species of Ranunculus sect. Euromontani and related taxa according to the DIVA+J model. Asterisks denote type specimens. Node I represents the ingroup node (R. sect. Euromontani). (b) Visualization of further events, starting from the well-supported ancestral areas II (separation into two taxa) and III (dispersal to other areas).
Among the six biogeographical models tested, those including the founder-event speciation parameter (‘+J’) consistently showed a better fit to the data. The best-supported model was DIVA+J, closely followed by DEC + J, with very similar AIC values (107.8 vs 108) and AIC weights (0.52 vs 0.47; Table 2). By contrast, BayArea+J received substantially lower support (AIC = 119.1; AIC_wt = 0.0019). The most likely area of origin of the ingroup is equivocal under the DIVA+J model, including several areas; for the ancestor of R. sect. Euromontani the ancestral area is Alps/Southeast Carpathians/West Carpathians, in combination with other areas (node I) in Fig. 4(a).
The reconstruction further suggested a pattern consistent with range expansion from a widespread ancestor toward other mountain systems. A subsequent split c. 1.4 Ma separated the Carpathians (areas F and I; Fig. 4a) and the Alps (area A), as main ancestral areas II and III for the subsequent dispersals (Fig. 4b). In the Carpathians, the split between West and South/East Carpathians was followed by speciation. From the Alps, several lineages subsequently expanded to Corsica, to the Central Apennines, to the Balkan Peninsula and from there to the South Carpathians, to the Pyrenees, and from there to the Cordillera Cantábrica. Within the Alps, dispersal to marginal parts of the Alps and polyploidization of two lineages resulted finally in five species.
The analysis of population structure supported K = 10 genetic clusters indicated by the cross-entropy values. These clusters largely corresponded with the phylogeny, aligning with both the taxonomy and distribution of the samples (Fig. 2c). Most taxa formed well-defined genetic clusters, characterized by generally low levels of admixture. All samples of R. demissus and R. aurasiacus shared a consistent cluster assignment despite their broad geographic distribution. High levels of admixture were observed in R. marschlinsii and in the tetraploid R. montanus individuals, which showed genetic contributions from multiple clusters, particularly the R. sartorianus/venetus one, and R. carinthiacus. A subset of the R. gouanii individuals (6479, 11 431, 11 429, 11 421, and 10 323) also exhibited high admixture levels, mainly involving R. aduncus and R. carpaticus. Interestingly, the tetraploid R. carpaticus sample 11 475 showed high admixture with the cluster shared by R. sartorianus and tetraploid R. venetus. The second supported model from the cross-entropy (K = 11; Fig. S6) showed a mostly consistent structure, with the exception of two R. carinthiacus individuals (10 514 and 10 523) from Cordillera Cantábrica which formed a distinct cluster. In general, clustering patterns and admixture levels remained stable across both K values.
The NeighborNet analysis based on 1819 SNPs revealed a reticulate evolutionary structure, with no fully discrete clusters but several well-supported groupings (bootstrap 81.5–100), broadly consistent with the phylogenetic results (Fig. S7a). The taxa of R. sect. Euromontani formed distinct clusters of species, corresponding to clades of the phylogenetic analysis. The second analysis based on unlinked SNPs largely revealed the same groupings, but with lower bootstrap support, ranging from 77.7 to 100 (Fig. S7b).
In the Carpathians, PCoA (Fig. 5a) and PCA (Fig. S8a) of previous ‘R. pseudomontanus’ samples revealed two separate, only slightly overlapping morphological clusters, corresponding to the molecular grouping into West and South Carpathians. Discriminant analysis confirmed this grouping with 92.95% correctly classified samples (Table S5). Seven characters showed significant differences, whereby the plants from the South Carpathians have smaller basal leaves with fewer segments, narrower stem leaf middle segments, larger achenes, but proportionally shorter beaks than the plants in the West Carpathians (Table S6; Fig. S9). In the Pyrenees, we observed a clinal variation from the eastern Pyrenees (‘R. ruscinonensis’) to the western Pyrenees (R. gouanii), with a broad zone of overlap in the central parts (mostly in Andorra). The three geographical groups strongly overlapped in PCoA (Fig. 5b) and PCA (Fig. S8b), and discriminant analysis revealed only 50.0% of cases correctly classified (Table S6). An alternative grouping of two species, following Landolt (1956), revealed 54.17% of cases correctly classified. Basal and stem leaf length gradually increased from the east to the west, whereas stem leaf incision and achene length led to two contradicting groupings (Fig. S10).

Principal coordinate analyses of taxa in Ranunculus sect. Euromontani based on morphometric data, shown with convex hulls. (a) Ranunculus gruinalis and Ranunculus tatrimontanus in the Carpathians. (b) Ranunculus gouanii in the Pyrenees, subdivided into three geographical groups.
We base our taxonomy on the molecular data, following an evolutionary lineage concept (e.g. Coyne & Orr, 2004; Hörandl, 2022). Our genetic data confirmed the species delimitation in the Alps, which was based on the comprehensive revision by Landolt (1954) and is nowadays broadly accepted (Aeschimann et al., 2004). We could confirm the taxonomy of R. aduncus and the highly variable R. sartorianus (Strid, 1986) by including type material.
In the Carpathians, however, we found that a species previously recorded as R. pseudomontanus (Fröhner, 1968; Jalas & Suominen, 1989; Tutin & Akeroyd, 1993) is split into two genetically and morphologically distinct species, to be named R. gruinalis (South Carpathians) and R. tatrimontanus (West Carpathians). In the South Carpathians, we revealed R. pseudomontanus Schur (Schur, 1877) conspecific with R. gruinalis described by Schott et al. (1854), including material from the locus classicus of R. pseudomontanus (Götzenberg = Vârful Măgura near Cisnădie) and a type specimen from R. gruinalis (peak Surul in the Făgăraș Mountains; Table S1). For this species, the name R. gruinalis Schott, Nyman & Kotschy has priority (Turland et al., 2025), see detailed comments and synonyms in Note S1. All samples in the South Carpathians were diploid.
Ranunculus gruinalis Schott, Nyman & Kotschy, Analect. Bot. 33 (1854).
= R. pseudomontanus Schur, 1877: 42 ≡ R. montanus b. multicaulis Schur 1866: 19 ≡ R. montanus subsp. pseudomontanus (Schur) Ciocârlan, 2000: 168.
Lectotype designated here: ‘Plantae Transsilvaniae Herbarii Schott. 340. In graminosis Surul loco Gropa Turculuj dicto. 5200 ped. Legit Th. Kotschy, Die 17. Jun. 1850’. GOET009804!. Isotype in MEL2424193! Both available online via https://plants.jstor.org/.
Both genetic and morphometric data revealed the West Carpathian taxa as distinct. Since R. pseudomontanus is a synonym of R. gruinalis, the West Carpathian species needs a new name and thus is described here as new species:
Ranunculus tatrimontanus Manzo, Lorberg, Bremer, Tomasello, Schönsw., Hörandl, spec. nov.
Holotype: Slovakia, mountains. ‘Flora Slovakiae Septentrionalis. Ranunculus montanus Willd. Vysoké Tatry. Biely žľab pod Sklavkovským štítom. 2100 m. Leg. O. Grebenščikov 9.8.54’, SAV0021655! (online via https://ibot.sav.sk/herbarium/object/SAV0021655a; https://ibot.sav.sk/herbarium/object/SAV0021655b).
Diagnosis: Plants perennial, rhizomatous, 10–40 cm high; basal leaf blades in the outline reniform, pubescent to glabrous, 1.8–3.2 cm long, with 5–9 segments (often secondarily lobed), middle incision 69–92% of length; lowest stem leaves with three to nine segments, the middle one 1.7–9.0 mm wide (lanceolate to linear, often secondarily lobed); petals (= nectary leaves) yellow, 1–1.3 cm long; achenes 2.2–2.7 mm long, the beak 22–32% of its length. The strongly segmented basal leaves, mostly seen in robust individuals, are typical for the taxon. Diploid and tetraploid (Table 1; records from the Western Carpathians).
In the Pyrenees, both genetic and morphometric data suggested one single clade/cluster with a clinal variation from the eastern populations (described as R. ruscinonensis by Landolt, 1956) to the central and western parts (described as R. gouanii by Willdenow, 1799). The transition from small, delicate plants in the dry conditions in the east to more robust, stout plants in the wetter western parts could be due to climatic gradients. A supposed relationship of ‘R. ruscinonensis’ to R. sartorianus (Jalas & Suominen, 1989) can be clearly rejected. We conclude to merge the two Pyrenean species into one, to be named R. gouanii Willd. (Willdenow, 1799); see details in Notes S1.
Here, we present the first fully resolved phylogeny of R. sect. Euromontani, based on target enriched nuclear genes. These data serve well for reconstructing the evolutionary history and biogeography of recently diversified species groups with shallow phylogenetic divergence, polyploidy and reticulate evolution. These markers can well delimit species and provide robust phylogenomic frameworks for understanding radiations in high mountain systems (McKain et al., 2018; Tomasello et al., 2020; Kandziora et al., 2022; Karbstein et al., 2022; Španiel et al., 2023; Aguirre-Santoro et al., 2024). Similar to other studies on Ranunculus (Emadzade et al., 2015; Karbstein et al., 2022), the plastome phylogeny confirms major subclades only, but does not resolve relationships at the species level. We therefore base our conclusions on the nuclear data.
In R. sect. Euromontani, a phylogenetic backbone with short branches, the SDA results, and a reticulate central structure in NeighborNet analyses point at evolutionarily young origins and rapid speciation events, nevertheless resulting in distinct and well-supported clades or clusters. Genetic structure showed a predominant genetic partition for most species, with low levels of admixture. Admixture should be interpreted with caution, as the analysis was based on a rather conservative, small SNP data set derived from genes. Admixture might reflect incomplete lineage sorting or introgression events before speciation, rather than recent processes on population level. However, only few genes appear to be affected by admixture as visualized in Fig. 2(c). Polyploid samples showed high levels of admixture, which points at a more complex evolutionary history, with different subgenome contributions and postorigin genome evolution (Wagner et al., 2020; Karbstein et al., 2022; Španiel et al., 2023).
Our study presents R. sect. Euromontani as a new case for the few well-studied Quaternary radiations of high alpine plants in the EAS. Kadereit & Abbott (2021) list only four comparative groups with radiations in the EAS from the genera Gentiana L., Primula L., Saxifraga L., and Soldanella L. Other radiations in the EAS are pre-Quaternary, for example within Androsace sect. Aretia (Schneeweiss et al., 2004; Roquet et al., 2013) and in Phyteuma sect. Capitata (Schneeweiss et al., 2013). So far, climatic oscillations causing cycles of geographical spread and isolation during the Pleistocene were thought to be the major drivers for diversification in this time period (Kadereit & Abbott, 2021). Allopatric vicariant speciation via an emerging barrier – in this case, ice shields – were regarded the main mechanism for geographical isolation and divergence. This scenario appears most reasonable for the Alps, where ice shields forced plants to retreat in peripheral or central refugia (Tribsch & Schönswetter, 2003). The diploid species of the Alps could have originated via allopatry in classical peripheral refugia (Tribsch & Schönswetter, 2003). Speciation could have been accompanied by specialization on siliceous (R. villarsii) vs calcareous bedrock (R. carinthiacus, R. aduncus; Dickenmann, 1982; Aeschimann et al., 2004). Diversification according to bedrock conditions was also an important process in other alpine genera (Smycka et al., 2022). However, this scenario does not apply to the EAS as a whole. Most other mountain chains were only covered by ice at highest peaks. Since all the mountain chains of the EAS already existed in the Quaternary, geographical isolation by orogeny can be ruled out as an isolation mechanism for allopatric speciation (Vargas, 2003).
Our results suggest the prevalence of peripatric speciation in the EAS, that is dispersal over existing barriers, colonization of the new area by few individuals, and founder-event speciation in the new area (Coyne & Orr, 2004). This scenario was best supported by the DIVAlike+J and DEC + J models (Table 2). Peripatric speciation is usually observed in island archipelagos (Coyne & Orr, 2004), but mountain systems do represent island-like topographies as well (Nagy & Grabherr, 2009). Hence, we assume that after orogeny, speciation followed similar mechanisms after dispersal events. Most dispersal events in R. sect. Euromontani were directed southward from the Alps, thereby escaping from the heavily glaciated areas. The Mediterranean mountain chains became island-like after the onset of the summer-dry Mediterranean climate and the establishment of evergreen, xerophytic Mediterranean vegetation 2.3 Ma in the lowlands (Suc, 1984). The lowland areas between the major southern mountain chains represented evergreen forest barriers for high alpine plants during warmer time periods, and deciduous and conifer forests during cold periods (Büdel, 2021). Our results suggest jump dispersal of buttercups over lowland areas, and founder-event speciation in suitable habitats at higher altitudes of the Pyrenees, the Apennines, and the Balkan Peninsula, where summer-wet conditions and open grassland remained available. After speciation, the lowland conditions eventually led to infraspecific vicariant patterns and disjunct distributions (e.g. in R. carinthiacus), as it was found by phylogeographic studies in other alpine species in southwestern Europe; however, also cases of long-distance dispersal were inferred (Kropf et al., 2006). Dispersal and vicariance scenarios are not mutually exclusive (Sanmartín, 2009) and may act at different time levels. The lowlands between the Alps and Carpathians were covered by steppe vegetation during cold periods, eventually allowing also continued dispersal (Hurka et al., 2019; Büdel, 2021).
Sympatric speciation via allopolyploidy, resulting in stable tetraploid lineages, happened only in two species of the Alps (R. montanus and R. venetus). Both PhyloNet and genetic structure suggest one shared parent of both tetraploids, probably from a diploid, extinct lineage. The other diploid parent could have been R. carinthiacus for R. montanus, as already hypothesized by Landolt (1954), and R. villarsii for R. venetus. The high morphological similarity of R. montanus and R. venetus supports the hypothesis of one shared parent, whereas genetic data indicate genome contributions from an additional parent. This interpretation, however, must be taken with caution, considering the limitations of reduced datasets and samples and should be confirmed by whole genome sequencing. In all other investigated species, polyploids emerge in single locations without apparent changes of phenotypes and probably just represent products of recurrent autopolyploidization without further speciation or range expansions. For outcrossing polyploids, minority cytotype disadvantages could have hindered establishment of new populations (Levin, 1975; Kirchheimer et al., 2018).
Polyploidy allowed sympatric speciation of R. montanus and R. venetus in the Alps by providing crossing barriers to the sympatric diploid species. Crossing experiments between diploid and tetraploid species revealed drastically reduced seed set, and no triploids were found in the chromosome data from wild, partly sympatric populations (Landolt, 1954; Dickenmann, 1982). Whether polyploidy is associated with different ecological niche dynamics, as observed in other alpine species (e.g. Kirchheimer et al., 2016, 2018), remains an open question. Tetraploid R. montanus has a broad ecological niche with respect to bedrock type, but prefers locally calcareous substrates when in sympatry with R. villarsii which is restricted to siliceous bedrock (Dickenmann, 1982). Tetraploid R. venetus prefers calcareous bedrock (Aeschimann et al., 2004). To which extent bedrock type could play a role for sympatric speciation in the whole EAS, requires further studies.
The radiation of R. sect. Euromontani in the EAS appears to be mainly due to geographical isolation (sensu Simoes et al., 2016), accompanied by some instances of sympatric speciation. The genus Ranunculus probably has pre-adaptations to high alpine habitats, as several independent colonizations of the Alps and of other high mountain systems world-wide have been documented (Hörandl & Emadzade, 2011). Ranunculus sect. Euromontani has no obvious key innovation to colonize alpine habitats, as it would be typical for adaptive radiations; long rhizomes, stringlike roots, and a small habitus (Paun et al., 2005) appear in other high alpine Ranunculi as well. All species (except R. carpaticus, to be described later) grow preferentially in subalpine to alpine grassland habitats, sometimes extending to screes, snowbeds, and rocks; R. montanus regularly occurs also in montane forests. We conclude that the radiation is not primarily adaptive, but rather a result of geographical isolation, in line with Smycka et al. (2022) on other diversification scenarios in the EAS. Compared with other radiations in Ranunculus, R. sect. Euromontani resembles the Himalayan radiation with shallow relationships and low ecological differentiation (Emadzade et al., 2015). In the New Zealand Alps, however, the radiation of Ranunculus sect. Pseudadonis is connected to habitat shifts (Lockhart et al., 2001).
Our biogeographical analysis revealed an origin of the R. sect. Euromontani clade from a widespread ancestor, likely from the Mediterranean area (Paun et al., 2005; Emadzade et al., 2011; Emadzade & Hörandl, 2011). The southern Mediterranean R. demissus-aurasiacus clade split off long before the diversification of R. sect. Euromontani and is therefore not regarded as part of the EAS radiation, but represents a separate lineage that needs to be explored further. The long branch suggests that extinction could have played a role before the ancestor of R. sect. Euromontani established, whose area was in the Alps or in the Carpathians. A clear picture emerges c. 1.4 Ma, with two well-supported ancestral areas, one in the Carpathians, the other in the Alps (Fig. 4). The Carpathian branch splits into R. carpaticus, a morphologically very distinct widespread species of montane to subalpine tall-herb communities, and into the high alpine species R. tatrimontanus in the West Carpathians. From the Alps, several jump dispersal events are indicated toward the southern mountain chains of the EAS, probably driven by the onset of the Pleistocene glaciations in Central Europe. Interestingly, from the Balkan Peninsula, where R. sartorianus evolved and became widespread, a dispersal to the Southern Carpathians resulted in speciation of R. gruinalis. These independent speciation events in the Western Carpathians and in the South-eastern Carpathians underline the biogeographical boundaries between these two mountain systems (Ronikier, 2011) and could be further driven by distinct climatic conditions (Spinoni et al., 2015). In the Apennines, on Corsica and in the Pyrenees, jump dispersal resulted in the evolution of endemic species. The only widespread species is R. carinthiacus which colonized not only the Alps but also the Dinarids (not sampled) and the Iberian Peninsula, and reached (probably via the Pyrenees) the Cordillera Cantábrica, similar to, for example, alpine Salix breviserrata (Pittet et al., 2025).
In Ranunculus, dispersal of achenes, possessing hooked beaks, is quite efficient via epizoochory on the fur of mammals or on feathers of birds, but also by anemochory (Kirchheimer et al., 2018). This may enable also long-distance dispersal to islands, such as Corsica. Similar patterns of dispersal between western Alps and Corsica can be observed even in taxa lacking adaptations for dispersal over long distances, as in Bupleurum stellatum L. (Schönswetter & Tribsch, 2005) or Leucanthemopsis alpina (L.) Heywood (Tomasello & Oberprieler, 2017). Lowland buttercups have even undergone transoceanic dispersal between continents (Emadzade et al., 2011). Establishment of populations in new areas might be facilitated by efficient generalist flowers of buttercups (Steinbach & Gottsberger, 1995), allowing shifts to novel insect pollinators in a newly colonized area.
Our study confirms that alpine floras still need taxonomic revisions and biogeographical investigations (Landolt, 2006). In our case, morphology-based species delimitation was always difficult due to convergent features in habitus and the high variation of diagnostic characters (Table S5). We base our classification on a phylogenomic approach and confirm that the use of herbarium material can improve the sampling and ascertain taxonomic conclusions by including type material (Manzo & Tomasello, 2025). Our phylogenomic approach recognized allopatric genetic lineages in the different mountain systems on the diploid level and confirmed two polyploid species. The clinal east–west pattern in R. gouanii favored inclusion of R. ruscinonensis (see Note S1). Considering the moderate genealogical distinctiveness, along with morphological and ecological differences, a recognition of the two taxa as subspecies could be justified (see also Oberprieler, 2023), but would require more detailed population genetic and ecological studies. Clinal east–west admixture patterns have been found within other Pyrenean species as well, for example in Salix pyrenaica (Carnicero et al., 2022) and in Leucanthemopsis alpina (Tomasello & Oberprieler, 2017).
Other results suggest a need for taxonomic revisions in the Apennines, as R. pollinensis and R. apenninus always clustered together and are largely sympatric (Pignatti, 1982), and for the more distantly related R. demissus–R. aurasiacus clade. Even if our results remain inconclusive due to the limited sampling for these very large distribution areas, they provide useful frameworks for further studies.
We thank L. Hodač, V. Margreiter, A. Wylie, T. Zeni, and E. Nitz for help with sampling in the field, the herbaria SAV, WU, W, E, and GOET for providing materials, and Melanie Heinemann for technical assistance. Funding is acknowledged from the Ursula Hofmann Stiftung of the University of Göttingen, the Deutsche Forschungsgemeinschaft DFG Project ‘Evolutionary genomics: consequenzes biodiverse reproductive systems (EvoRest) GRK 2984’ to EH, and the Austrian Science Fund (FWF) grant DOI: 10.55776/P34092 to PS. Open Access funding enabled and organized by Projekt DEAL.
None declared.
EH contributed to the design of the research. EH and PS contributed to the sampling of material. EM, ESL, LB, ST and EH contributed to the data collection, analysis, and interpretation. EM and EH contributed to the writing the manuscript with contributions from all coauthors.
Raw sequencing reads and chloroplast loci alignments are available through the EMBL European Nucleotide Archive under accession no.: PRJEB113281, and R and SNP mining scripts are available via the GRO data repository at doi: 10.25625/MXONJM.
References
The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.
Texto de New Phytologist tal como lo publicó, leído de su página y mostrado aquí con el formato de El Autómata. El original, en nph.onlinelibrary.wiley.com · Volver a la noticia de El Autómata