An integrative taxonomic approach reveals a new species of Eranthis (Ranunculaceae) in North Asia

Abstract A new endemic species, Eranthis tanhoensissp. nov., is described from the Republic of Buryatia and Irkutsk Province, Russia. It belongs to Eranthis section Shibateranthis and is morphologically similar to E. sibirica and E. stellata. An integrative taxonomic approach, based on cytogenetical, molecular and biochemical analyses, along with morphological data, was used to delimit this new species.


Introduction
The genus Eranthis L. (Ranunculaceae) consists of eight to ten species distributed in southern Europe and temperate Asia (Lee et al. 2012;Park et al. 2019). Most species have narrow distributions and only one European species, E. hyemalis (L.) Salisb., has been widely cultivated in gardens and become naturalised in Britain (Boens 2014) and North America (Parfitt 1997). Eranthis are perennial herbs with tuberous rhizomes, basal long-petiolate leaves with the blades divided into several or many palmate segments (leaflets) that are entire or lobate; unbranched scapes carrying a solitary, bisexual and actinomorphic flower supported by three verticillate leaf-like bracts forming an involucre; (4-)5-8 yellow, white or pink, caducous sepals; 5-10(-15) yellow or white, bifid petals shorter than sepals; nectaries located at the middle or upper part of the petals; > 10 stamens; and 3-10 follicles with several smooth seeds in each fruitlet (Parfitt 1997). All species are early-blooming plants, with anthesis from March to May (depending on the altitude), but E. hyemalis has been found at full anthesis in mid-January in gardens (Sukhorukov, pers. obs. in Mainz, Germany, 2019 andLeiden, Netherlands, 2020).
On the basis of morphology, the genus has been divided into two sections: E. sect. Eranthis and E. sect. Shibateranthis (Nakai) Tamura (Tamura 1987). The type section is characterised by annual tubers, yellow sepals and emarginate or slightly bilobate upper petal margins without swellings (nectaries), whereas the members of section Shibateranthis have long-lived tubers, white sepals and bilobate or forked petal margins with swellings (Tamura 1995). Molecular phylogenetic analysis. based on nrITS and chloroplast trnL-trnF interspacer region, supports the subdivision of the genus into these sections (Park et al. 2019 Park et al. 2019). Two additional species with yellow sepals, E. bulgarica (Stef.) Stef. (Stefanoff 1963) and E. iranica Rukšāns & Zetterl. (Rukšāns and Zetterlund 2018), have been described from Bulgaria and Iran, respectively, but have not yet been included in molecular analysis.
Recent studies have revealed the genetic diversity, phylogeny and presumed origin of some narrowly distributed Korean and Japanese species with further conclusions about their taxonomic status (Lee et al. 2012;Oh and Oh 2019). The taxonomic and genetic diversity of Eranthis in the Asiatic part of Russia is insufficiently studied. To date, only two species have been found in Russia: E. sibirica and E. stellata (both belonging to sect. Shibateranthis) from South Siberia and Far East Russia (Malyshev 2005). High genetic polymorphism of E. sibirica across populations near Baikal Lake was discovered only recently (Protopopova et al. 2015) and this fact has inspired us to conduct a new study of Eranthis in the Asiatic part of Russia.
The aim of the present study was to investigate the morphological, molecular, biochemical and cytogenetic heterogeneity of the Baikal populations to determine wheth-er any undescribed species were present there. The relationship between E. sibirica, E. stellata and a new species, described and named below as Eranthis tanhoensis Erst, sp. nov. is explored here.

Plant material
More than 300 herbarium specimens were collected during field investigations in the Republics of Khakassia and Buryatia and the Irkutsk Province during 2018 and 2019. Fieldwork was conducted during different seasons to observe the species in both their flowering and fruiting stages. The specimens were deposited in the E and NS herbaria (herbarium abbreviations according to Thiers 2019+). Revision of herbarium materials was undertaken in the herbaria at IRK, LE, MW, NS, NSK, PE, VBGI and VLA. Drawings of the new species, Eranthis tanhoensis, are based on images of the type specimen (NS-0000948!) and paratype (NS-0000949!). The flowering and fruiting times and habitats are provided as cited on the collectors' labels. Maps of records were made with SimpleMappr (http://www.simplemappr.net). Conservation analysis was performed using criteria from the International Union for the Conservation of Nature (IUCN 2019). The Extent of Occurrence (EOO) and Area of Occupancy (AOO) of each species were estimated using GeoCat (Bachman et al. 2011).

Molecular analysis
We sampled 15 individuals of E. tanhoensis and six of E. sibirica. Two individuals of E. stellata and one each of E. pinnatifida and E. longistipitata were also included. The details of the samples are presented in Suppl. material 1: Table S1. Six nuclear and plastid DNA regions (ITS,rps16,matK and rbcL) were included in the molecular analysis. Total genomic DNA was extracted from silica gel-dried leaves or herbarium specimens using DNeasy Mini Plant Kits (Tiangen Biotech, Beijing, China) following the protocol specified by the manufacturer. Sequencing reactions were conducted using BigDyeTM Terminators (Applied Biosystems Inc., Foster City, CA, USA). Sequences were read using an automated ABI 3730xl DNA Analyzer. Geneious v8.0.4 (Kearse et al. 2012) was used to evaluate the chromatograms for base confirmation and to edit contiguous sequences. We first used the Maximum Likelihood (ML) method to perform non-parametric bootstrap analyses for each DNA region in RAxML v7.0.4 (Stamatakis 2006). No significant bootstrap support for conflicting nodes was evident amongst individual DNA regions (here considered to exceed 70%) and the six-locus datasets were therefore combined for subsequent analyses. Phylogenetic analyses of the combined dataset were conducted using ML and Bayesian Inference (BI) methods. RAxML was conducted with the GTR + Γ substitution model for each region with the fast bootstrap option using 1000 replicates. BI analysis was conducted in MrBayes v3.2.1 (Ronquist et al. 2012). Data partitioning and nucleotide substitution models were determined using PartitionFinder 2.1.1 (Lanfear et al. 2016). Two independent analyses, consisting of four Markov Chain Monte Carlo chains were run, sampling one tree every 1000 generations for 10 million generations. Runs were completed when the average standard deviation of split frequencies reached 0.01. The stationarity of the runs was assessed using Tracer v1.6 (Rambaut et al. 2014). After removing the burn-in period samples (the first 25% of sampled trees), a majority rule (> 50%) consensus tree was constructed.

Morphological analysis
The morphology of vegetative and reproductive structures was examined on well-developed specimens. For numerical analysis, 25 specimens at flowering and 25 specimens at fruiting stages were examined for each species (more than 150 specimens altogether). For each species, we studied different populations from across the range, including populations from the type localities of E. stellata and E. sibirica. As E. stellata often does not produce basal leaves at flowering, we studied this character in a limited number of samples. The morphological characters were measured using AxioVision 4.8 software (Carl Zeiss, Munich, Germany).
The missing values in the original data table were restored using multidimensional linear regression, in accordance with recommendations of Myers (2000) and Lee and Carlin (2010). A one-way analysis of variance (ANOVA), according to Chambers et al. (1992), was used to identify the distinguishing morphometric features of each species. The differences were considered significant at P-value < 0.05. As multiple statistical testing was performed, the calculated P-value was adjusted using the procedure proposed by Benjamini and Hochberg (1995). The principal component analysis was used to visualise the distribution of the analysed individuals over the space of morphometric characters. This method was employed only for those characters that displayed significant intergroup differences, according to the results of the ANOVA. For scale adjustment, the logarithmic transformation of data was used. The results of the principal component analysis were visualised using the Factoextra package (Kassambara and Mundt 2017).

Cytogenetic analysis
Somatic chromosomes were studied in root tip cells. Tubers were germinated in wet moss at ~15 °C for 2-4 weeks. Newly formed 1-2 cm long roots were excised and pretreated in a 0.5% colchicine solution for 2-3 h at 15 °C. Roots were fixed in a mixture of 96% ethanol and glacial acetic acid (3:1). Root tips were stained with 1% aceto-haematoxylin and the squash method was employed for investigation of the karyotype (Smirnov 1968).
Chromosomes were counted in 50-100 mitotic cells for each population. Mitotic metaphase chromosome plates were observed using an Axio Star microscope (Carl Zeiss, Munich, Germany) and photographed using an Axio Imager A.1 microscope (Carl Zeiss, Germany) with AxioVision 4.7 software (Carl Zeiss, Germany) and Ax-ioCam MRc5 CCD-camera (Carl Zeiss, Germany) at 1000× magnification in the Laboratory for Ecology, Genetics and Environmental Protection (Ecogene) of the National Research Tomsk State University. KaryoType software (Altinordu et al. 2016) was used for karyotyping, whereas Adobe Photoshop CS5 (Adobe Systems, USA) and Inkscape 0.92 (USA) were used for image editing. Karyotype formulae were based on measurements of mitotic metaphase chromosomes taken from photographs. The measurements were performed on 5-10 metaphase plates. The symbols used to describe the karyotypes followed those of Levan et al. (1964): m = median centromeric chromosome with arm ratio of 1.0-1.7 (metacentric chromosome); sm = submedian centromeric chromosome with arm ratio of 1.7-3.0 (submetacentric chromosome); st = subterminal centromeric chromosome with arm ratio of 3.0-7.0 (subtelocentric chromosome); t = terminal centromeric chromosome with arm ratio of 7.0-∞ (acrocentric chromosome); T = chromosome without obvious short arm, i.e. with arm ratio of ∞.

Flow cytometry
Flow cytometry with propidium iodide (PI) staining was used to determine the absolute DNA content. The relative DNA content in the nucleus (C-value) in representatives of three Eranthis species -E. stellata, E. sibirica and E. tanhoensis from different populations, was determined in this study. In total, more than 70 samples from 15 populations were studied (see Suppl. material 1: Table S1). Silica gel-dried leaf material (0.5-1.0 cm 2 ) was chopped with a sharp razor blade in a 1 ml cold nuclei extraction buffer composed of 50 mM Hepes, 10 mM sodium metabisulphite, 10 mM MgCl 2 , 0.5% polyvinylpyrrolidone, 0.1% bovine serum albumin, 0.3% Tween20, 0.2% Triton X-100, 50 μg/ml RNase, 1 μg/ml β-mercaptoethanol and 50 μg/ml propidium iodide (PI). The samples were filtered through 50 μm nylon membranes into sample tubes and incubated in the dark at 4 °C for 15 min. Samples were measured using a Partec CyFlow PA flow cytometer equipped with a green laser, at 532 nm wavelength. The absolute nuclear DNA content, the 2C-value according to Greilhuber et al. (2005), was calculated as the ratio of the mean fluorescence intensity of the nuclei of the sample to that of an external standard multiplied by the total nuclear DNA content of the standard. The possible effect of secondary metabolites on the binding of the intercalating dye was evaluated by measuring the fluorescence of Allium fistulosum L. leaf samples prepared as described above, but with the addition of the supernatant from Eranthis samples, centrifuged without PI. The samples were measured three times at 10 min intervals. If no variation in the average values of the detection channels was observed for the A. fistulosum peak, the effect of secondary metabolites was considered negligible.
The 1Cx-value (monoploid DNA content sensu Greilhuber et al. 2005) was calculated by dividing the 2C-value by the ploidy level of the species. The species, used as external standards, were Zamioculcas zamiifolia Engl., 2С = 48.35 pg and Vicia faba L. 'Inovec' 2С = 26.90 pg (Doležel et al. 1992;Skaptsov et al. 2016). We used the Statistica 8.0 software (StatSoft, Inc.), Flowing Software 2.5.1 (Turku Centre for Biotechnology) and CyView software (Partec, GmbH) for data analyses. Flow cytometry was performed at the Laboratory for Bioengineering of the South-Siberian Botanical Garden, Altai State University (Barnaul, Russia).

High-performance liquid chromatography (HPLC) analysis of individual phenolic compounds in ethanol leaf extracts
In order to determine the composition of phenolic compounds, air-dried plant material was mechanically ground to obtain a homogenous powder and then samples of ~0.2 g were extracted three times using 70% aqueous ethanol solution for 30 min in a water bath at 72 °C. Next, the combined extract was concentrated in porcelain dishes to 5 ml. The solutions were filtered and stored at 4 °C until analysis. Analysis of phenolic components was performed using an Agilent 1200 HPLC system equipped with a diode array detector and a ChemStation system for the collection and processing of chromatographic data (Agilent Technology, Palo Alto, CA, USA). The separation was performed on a Zorbax SB-C18 column (5 μm, 4.6 × 150 mm) at 25 °C. The methanol content of the mobile phase in an aqueous solution of phosphoric acid (0.1%) varied from 50-52% over 56 min (van Beek 2002). The eluent flow rate was 1 ml/min. Detection wavelengths were 255, 270, 340 and 360 nm. Groups of phenolic substances were identified by their spectral characteristics (Bate-Smith 1962; Mabry et al. 1970). For identification of the phenolic components in plant extracts, standard samples of salicylic and chlorogenic acids, quercetin, kaempferol, orientin (Sigma-Aldrich Chemie GmbH, Munich, Germany), gentisic and caffeic acids (Serva Heidelberg, Germany), hyperoside and vitexin (Fluka Chemie AG, Buchs, Switzerland) were used. The samples were analysed twice.

Molecular phylogenetic analysis
Bayesian and ML analyses of the combined dataset produced highly consistent topologies. Eranthis sibirica and the new species E. tanhoensis formed a sister clade of that of E. pinnatifida. The monophyly of each species, E. tanhoensis sp. nova, E. sibirica and E. stellata, was strongly supported (Fig. 1).

Morphological analysis
The morphological analysis revealed that E. sibirica was not homogeneous across its distribution area. We compared 41 characters to distinguish E. sibirica, E. tanhoensis and E. stellata (Suppl. material 1: Table S2). The basal and involucral leaves in Eranthis spp. undergo changes at fruiting and, for this reason, the lengths of all leaves, their segments and segment lobes were measured both at the flowering and fruiting stages. In Suppl. material 1: Table S2, an asterisk (*) indicates the characters used in the numerical analysis. An ANOVA was conducted only for quantitative characteristics. As basal leaves are often absent at the time of flowering and there were no samples with basal leaves in herbarium collections, there were limited data on these characteristics of E. stellata.
The ANOVA of morphometric characters showed significant differences amongst the studied species in characters (1) Tables S3, S4). In total, significant differences amongst the species were found in 10 out of 15 morphometric parameters measured at flowering and in 9 out of 13 parameters at fruiting. The principal component analysis revealed that the first two main components accounted for 83.1% and 81.8% of the variance in the entire data array of the parameters measured at flowering and fruiting, respectively and showed the best species discrimination. The highest variability of morphometric characters was found at flowering in E. sibirica ( Fig. 2A) and at fruiting in E. tanhoensis (Fig. 2B). As signified by the directions of the vectors indicating the gradients in the character values, at flowering, E. sibirica differed from E. tanhoensis by having lower values for characters (18), (22), (24) and (31) and a higher value for character (9). At fruiting, E. sibirica was characterised by having lower values for parameters (19), (17), (23) and (25) and higher values for parameters (10), (40) and (29), in comparison with those of E. tanhoensis. E. sibirica differed from E. stellata by having higher values for characters (1), (16), (29), (30) and (32) at flowering and (10) and (14) at fruiting. The pattern of overlap between the species differed between flowering and fruiting plants. For instance, E. tanhoensis was reliably distinguished from E. sibirica only at fruiting (the ellipses enclosing the samples did not overlap; Fig. 2B). In addition to numerical parameters, the new species was also distinguished by qualitative characters.
Eranthis stellata. In all four studied populations of E. stellata, the basic chromosome number was x = 8. This species was diploid with 2n = 2x = 16, which is typical of the genus (Table 1; Fig. 3F). Five pairs of chromosomes were metacentric, two pairs were submetacentric and one pair was acrocentric (Fig. 3F). The karyotype formula of E. stellata was 2n = 10m + 4sm (2sat) + 2t. No B chromosomes were observed in this species.
The basic chromosome number x = 8 has been reported for the entire genus Eranthis (Langlet 1932;Kurita 1955;Tak and Wafai 1996;Gömürgen 1997;Yuan and Yang 2006;Kim et al. 2011;Marhold et al. 2019). Our results are consistent with previously published data (Yuan and Yang 2006), with insignificant differences in the karyotype formula. However, we showed, for the first time, that E. sibirica and E. tanhoensis are distinguished from other species of the genus by the basic chromosome number x = 7. Such differences in basic chromosome numbers (x = 7 and x = 8) have been found in some other genera of Ranunculaceae, for example, Anemone L. and Ranunculus L. (Rice et al. 2015). Our results regarding the chromosome numbers in E. sibirica (2n = 28 and 2n = 42) differed from the data reported by other researchers for this species (2n = 32: Krogulevich (1976) or 2n = 16: Gnutikov et al. (2016Gnutikov et al. ( , 2017). Eranthis tanhoensis was found to have 2n = 14. Based on the incongruence of the chromosome data with previous and recent analyses, we assume that some populations of E. sibirica and E. tanhoensis may have diverse cytotypes. Both species clearly differed from E. stellata by the absence of acrocentrics. All three species were characterised by five metacentrics and two submetacentrics per monoploid chromosome set.
Tetraploids and hexaploids of E. sibirica exhibited insignificant differences in DNA content (9.25 pg for 6x and 9.55 for 4x), whereas diploids of E. tanhoensis showed a higher 1Cx level (12.49 pg), which may indicate a relatively ancient diversification of these species. Data on the 1Cx level of E. stellata (15.77 pg) indicated the independent or parallel evolution of genome size in this species. According to flow cytometry, variations in 1Cx levels between diploid samples of E. tanhoensis and hexaploids and tetraploids of E. sibirica were in accordance with the hypothesis of genome downsizing in polyploid flowering plants (Leitch and Bennett 2004).
The chromatographic profile of E. sibirica differed from that of E. tanhoensis in the presence of caffeic acid, orientin, vitexin and flavone (peak 6: t R , min = 9.4, λ max , nm = 270, 310) in 70% ethanol leaf extracts (Fig. 4). Caffeic acid, orientin and flavone (peak 6) were generally absent from leaves of E. tanhoensis, whereas vitexin was found in some samples in trace amounts. The leaves of E. tanhoensis from almost all the studied populations contained quercetin, which was not detected in E. sibirica. Distinguishing compounds in leaf extracts of E. stellata were gentisic acid, phenolic acid (Fig. 4, peak 4: t R , min = 7.1, λ max , nm = 250, 300) and flavone (Fig. 4, peak 21: t R , min = 42,2; λ max , nm = 210, 310), which were absent from the two other species. Vitexin, hyperoside and salicylic acids were not found in E. stellata leaves. All samples of E. stellata contained orientin and caffeic acid, which were characteristic of E. sibirica and quercetin, which was typical of E. tanhoensis.

Eranthis tanhoensis
Notes. Turczaninow (1842) described the species E. uncinata Turcz., growing at higher altitudes and distinguished from E. sibirica by the number of petals (5-6, not strictly 5), by the shape of the stylodium (recurved rather than straight), smaller flowers and more dissected leaf blades. However, our studies have shown that these morphological characters are variable and all variations can be found both in the foothill and alpine plants. Shipchinskiy (1937) merged E. uncinata with E. sibirica. However, he described two varieties: E. sibirica DC. var. nuda Schipcz. with glabrous pedicels (= E. sibirica var. sibirica) and E. sibirica DC. var. glandulosa Schipcz. with glandularpubescent pedicels. These varieties were not validly published under ICN Article 39.1 (Turland et al 2018). Nakai (1937) attributed E. sibirica and E. uncinata to the genus Schibateranthis Nakai (≡ Eranthis sect. Schibateranthis (Nakai) Tamura).
Affinity. The new species belongs to E. sect. Shibateranthis (Nakai) Tamura and it is sister to E. sibirica, according to the results of molecular phylogenetic analysis (Fig.  1). E. tanhoensis is morphologically similar to E. sibirica and E. stellata  in having white sepals, tubular two-lipped petals with bilobate or forked lips, apically acute lobes with abaxial lip and globular yellow swellings (nectaries) at the top or in the central part. The differences amongst the three species are presented in Table 2.
The new species differs from other related species by dense glandular pubescence of the flower stems, rounded or widely rhombic (not obovate or lanceolate) leaf blade segments, acute, rather than rounded teeth apices of the basal and stem leaves, a large number of teeth and width of the segments of the basal and stem leaves (see also 2). Additionally, all three species growing in Russia have different distribution patterns (Figs 10, 11).
Habitat and ecology. Eranthis tanhoensis can be found at 350-2400 m a.s.l., where it grows in fir, Siberian pine, spruce and birch forests, on riverbanks, beside streams (up to 1500 m a.s.l.) and in subalpine meadows (at higher altitudes).
Etymology. The specific epithet of the new species is derived from the type locality, Tanhoi        Preliminary conservation status. Although the species seems to have a small distribution area in southern Baikal Lake, the populations observed in 2018 and 2019 consisted of numerous individuals producing viable fruits and no threats to the habitats were observed in the field studies. The EOO of E. tanhoensis was estimated for an area of more than 1372 km 2 , while the AOO was 72 km 2 . Preliminary conservation status, according to IUCN's Extent of Occurrence criteria indicates the species as Endangered (EN) (IUCN 2019).