﻿Taxonomy and distribution of Taraxacum sect. Erythrosperma (Asteraceae) in Poland

﻿Abstract The dandelions from Taraxacumsect.Erythrosperma are taxonomically well distinguished and ecologically restricted to warm and sunlit habitats of steppes, dry and sandy grasslands, and distributed in temperate regions of Europe and Central Asia, with some being introduced to North America. Despite the long tradition of botanical research, the taxonomy and distribution of dandelions of T.sect.Erythrosperma is still underexplored in central Europe. In this paper, by combining traditional taxonomic studies supported by micromorphological, molecular and flow cytometry analyses as well as potential distribution modelling we shed light on taxonomical and phylogenetical relationships between members of T.sect.Erythrosperma in Poland. We also provide an identification key, species-checklist, detailed descriptions of morphology and occupated habitats as well as distribution maps for 14 Polish erythrosperms (T.bellicum, T.brachyglossum, T.cristatum, T.danubium, T.disseminatum, T.dissimile, T.lacistophyllum, T.parnassicum, T.plumbeum, T.proximum, T.sandomiriense, T.scanicum, T.tenuilobum, T.tortilobum). Finally, conservation assessments performed using the IUCN method and threat categories for all the examined species are proposed.


Introduction
Human activity, climate warming and biological invasions have recently become the main factors threatening natural biodiversity in many regions of the world (Robinson et al. 2020;Wallingford et al. 2020). Environmental changes, to cite some examples, the deforestation of tropical forests, drying of bogs and meadows and their conversion to arable fields as well as expansive urban development, have a strong negative impact on the diversity of many groups of organisms. Many species become extinct even before they are discovered and described (Costello et al. 2013;Lees and Pimm 2015). This phenomenon is clearly visible in tropical regions, and the most prominent example is orchids, one of the most diversified and still underexplored groups of species (Szlachetko and Kolanowska 2021;Kolanowska et al. 2022). However, the problem also concerns species within taxonomically problematic and species-rich genera with morphologically similar or cryptic species, such as Rubus, Hieracium, Campanula, Orobanche, Stipa, Oxytropis, etc. (Wolanin et al. 2016, 2020Aleksić et al. 2018;Piwowarczyk et al. 2019;Chen et al. 2020;Nobis et al. 2020a;Baiakhmetov et al. 2021;Kosiński et al. 2021;Trávníček et al. 2021;Havlíček et al. 2022;Szeląg 2022;Vintsek et al. 2022;Wang et al. 2022;Nobis et al. 2023). Globally distributed dandelions (Taraxacum, Asteraceae) can also be included in this group (Uhlemann 2002(Uhlemann , 2016Scott and Rich 2013;Štěpánek and Kirschner 2017;Marciniuk et al. 2018;Kirschner et al. 2020Kirschner et al. , 2022. The genus Taraxacum Wigg. comprises ca. 3000 species classified into 60 sections distributed globally throughout the temperate zone (Vašut 2003;Reisch 2004;Kirschner et al. 2015;Vašut and Majeský 2015). Most of them are apomicts, sexually reproducing diploids being rare (Richards 1970; Mogie and Ford 1988;Van Dijk 2003). Diploids coexist with polyploidy apomicts in most sections. Only a few of them are considered primitive (T. sect. Piesis (DC.) Kirschner & Štěpánek, T. sect. Dioszegia (Heuffel) Heuffel, T. sect. Biennia R. Doll, T. sect. Glacialia Handel-Mazzetti, T. sect. Wendelboa (Soest) R. Doll), containing exclusively diploid species (Kirschner and Štěpánek 1996). The majority of the young European and Asian sections (e.g. T. sect. Taraxacum, T. sect. Palustria, T. sect. Erythrosperma) originate from hybrid taxa most likely formed during the Pleistocene glaciation as a result of multiple contacts of southern and northern species ranges (Kirschner and Štěpánek 1996;Marciniuk et al. 2010).

Species
Tacik 1980  with deeply lobed leaves, narrow lobes and petioles, small outer bracts, and mostly red or straw-coloured, strongly spinulose achenes with long cylindrical cone. Dandelions belonging to this group are adapted to warm and sunlit habitats, such as sandy grasslands, pseudosteppes, steppes, xerothermic swards, and ruderal communities (Dudman and Richards 1997;Vašut 2003;Wendt and Øllgaard 2015).
Due to enormous species-richness in the genus Taraxacum, the presence of multiple hybridisation events, frequent polyploidy and apomictic reproduction, as well as the limited number of studies related to the diversity and distribution of its species (Kirschner and Štěpánek 1996;Marciniuk et al. 2010;Kirschner et al. 2016;Jafari et al. 2018;Lee et al. 2021), the biogeography, phylogeny and genetic diversity of dandelions unfortunately still remain poorly explored. Most of the research so far focused on establishing general intrageneric phylogenetic relationships by using representatives belonging to different sections ( Van der Hulst et al. 2003;Kirschner et al. 2016) or between selected species occurring in a given area of interest (Majeský et al. 2012;Marciniuk et al. 2020;Lee et al. 2021). Some more detailed studies concerned population genetics (Jafari et al. 2018) or variation between complete chloroplast genomes (Salih et al. 2017;Lee et al. 2021) regarding selected individual species. Research on phylogenetic relationships between species within particular sections is also relatively sparse . The studies of Reisch (2004) and Majeský et al. (2015) are, to the best of our knowledge, the only ones to concentrate exclusively on species from T. sect. Erythrosperma, and the authors used various analyses such as random amplified polymorphic DNA (RAPD), nuclear Simple Sequence Repeats (SSR), amplified fragment length polymorphism (AFLP) DNA markers and a selected sequenced region of chloroplast DNA to study phylogenetic relationships and genetic differentiation respectively between and within the examined taxa.
It is assumed that evolutionary processes within different sections of Taraxacum are linked to the appearance of new habitats or habitat specialisation within a group of hybrids (Kirschner and Štěpánek 1996), and microspecies commonly show some different eco-geographical patterns Macháčková et al. 2018). Sexually reproducing diploids and asexual triploids from the same sections have been proven to differ in terms of geographic ranges and occupied niches (Meirmans 2021). Triploid plants from T. sect. Erythrosperma are characterised by a wider geographic range and a much more extensive ecological niche compared to the diploids (Meirmans 2021), which may suggest noticeable differences between triploid taxa within the section. Due to the vegetation period suitable for proper collection of dandelions being restricted to early spring and some of the localities having been found accidentally during excursions, the distribution maps of Taraxacum species may be incomplete. Thus, species distribution modelling (SDM) can contribute important information for the studied dandelions.
In this paper, by combining traditional taxonomic studies supported by micromorphological, molecular and flow cytometry analyses as well as potential distribution modelling, we shed light on taxonomical relationships between members of T. sect. Erythrosperma in Poland. In particular we would like to answer the following questions: I) which species in T. sect. Erythrosperma occur in Poland; II) in which regions and types of habitat do the species occur; III) which morphological characters are species-specific and helpful in species identification; IV) could micromorphological characters of achenes be useful in species identification; V) does the molecular analysis confirm the distinctiveness of the taxa identified on the basis of morphological characters, and what are their phylogenetic relationships? This work also contains an easy-to-use identification key, morphological descriptions and photos of representative specimens that significantly facilitate their determination.

Field studies, distribution and morphological analyses
Field studies were carried out in 2012-2019, from mid-April to mid-May, and supplemented in May 2021. Plants were initially determined in the field and collected from each population. Individuals that were causing problems with determination (juvenile plants or damaged plants from habitats under anthropopressure) were dug out, cultivated and observed for several seasons. The geographic coordinates of the localities were determined by GPS equipment. For a description of plant communities with a share of dandelions of the sect. Erythrosperma, floristic lists were prepared. The notes were complemented in mid-June, and the names of the species were given after Mirek et al. 2020. The herbarium collection is deposited in the Institute of Biology, University of Rzeszów (UR), with the exception of T. sandomiriense types, which were deposited earlier in the Herbarium of the Institute of Botany of the Jagiellonian University (KRA). The revision of plant collections was carried out in the following herbaria: KRA, KRAM, WRAB, KTU, UGDA, SZUB, POZNB, MPD and Herb. J&P Marciniuk. Maps of species distribution in Poland were prepared using the cartogram (ATPOL) method (Zając 1978) on a 10×10 km square grid. Morphological studies were conducted on both living and herbarium plants using a ruler and a stereo microscope equipped with an eyepiece reticle.

Macromorphological analyses of achenes
Achenes for macromorphological studies were collected from mature plants, at least 40 achenes per 3-5 plants from each population (Table 2). Five morphological characters were studied: achene length (incl. cone), cone length, achene body width, length of achene body spinose part, and length of beak. Samples are deposited in the Institute of Biology, University of Rzeszów.

Micromorphological analyses of achenes (SEM observations)
For the SEM observations, the achenes were attached to aluminium stubs using Pelco conductive liquid silver and sputtered with 20 nm of gold using a turbo-pumped Quorum Q 150T ES coater. Samples (Table 3) were observed using a scanning electron microscope (Hitachi High-Technologies Corporation, Tokyo, Japan) operated at 5 kV and 10 mm distance. Micromorphological structures of achenes were observed and photographs taken by means of the scanning electron microscope Hitachi SU 8010 at various magnifications (Figs 5-7). Achenes were studied from base to distal portions. The following qualitative characters were studied: the shape and arrangement of achene spines; details of surface ornamentation of the achene body, achene spines, the upper part of the achene body and the middle part of the cone. Samples are deposited in the Institute of Biology, University of Rzeszów.

DNA extraction
Isolation of genomic DNA was performed from dried leaf tissues, which were ground to a fine powder using a mixer mill MM400 (Retsch) and 3-5 mm glass beads. Isolation of genomic DNA was performed using a modified CTAB method (Doyle and Doyle 1987). The isolated DNA was purified using a gDNA Clean kit (Syngen, Poland) when necessary. The purity and concentration of extracted DNA were evaluated using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, USA). The quality of extracted DNA was roughly verified by electrophoresis on 1% agarose gels. The total number of samples used for further molecular analysis was 34. We decided to use T. jugiferum H. Øllg. and T. stridulum Trávn. ined as outgroups. The latter species is easy to distinguish although still not described (nomen provisorium; Trávníček pers. comm.) (Table 4).

SCoT-PCR amplifications
Start Codon Targeted (SCoT) polymorphism is a newly emerged DNA molecular marker developed based on the targeting start codon of the genes and their surrounding consensus sequences in a gene family (Collard and Mackill 2009). Due to their simplicity, relatively low cost requirements, high reproducibility, and considerable association with phenotypic data, SCoT markers has been applied to many genetic studies, including analysis of genetic diversity, detecting intra-and inter-genetic variation in different plant species, stability of in vitro derived plants, phylogenetic relationships, taxonomy, species/cultivar identification, quantitative trait loci (QTL) mapping and DNA fingerprinting in various plants (e.g. Zhang et al. 2015;Etminan et al. 2018;Jalilian et al. 2018;Jedrzejczyk 2020;Zarei and Erfani-Moghadam 2021;Rai 2023).
Of a set of 20 tested SCoT primers (Genomed, Poland), 19 generated stable band patterns were selected for further studies (Table 5). All the PCR reactions were carried out within a total volume of 12.5 µl, containing 30 ng of genomic DNA template, 0.1 U/µl Taq DNA polymerase, 4 mM MgCl 2 and 0.5 mM of each dNTPs (2xPCR Master Mix Plus; A&A Biotechnology, Poland), 10 µM of primer and sterile deionised water to the final volume. The DNA amplifications were performed using T100 Ther- T T. stridulum Taraxacum Błażowa 49°53'N, 22°06'E mal Cycler (BioRad, USA) under the following conditions: initial denaturation at 94 °C for 5 min., followed by 35 cycles of 94 °C for 1 min., annealing for 1 min., and extension at 72 °C for 2 min. The last cycle was followed by the final extension step of 7 min. at 72 °C. The annealing temperature for each primer was optimised, and varied from 49.5 °C to 63.5 °C (Table 5). Amplified PCR products were separated by electrophoresis using 1.5% (w/v) agarose gel made in 1.0× TBE buffer and stained with ethidium bromide (0.5 µg/mL). A DNA ladder of 3000 bp (Thermo Fisher Scientific, USA) was used to determine the size of the amplicons. The gels were visualised under UV light and photographed using GelDoc XR+ (BioRad, USA).

Data analysis of SCoT-PCR products
The PCR-amplified SCoT products were detected on gels and scored as a binary data matrix, as the presence (1) or absence (0) of a band. Only clear, reproducible and well-defined bands were counted. The numbers of monomorphic and polymorphic amplification products generated by each SCoT primer were determined. Polymorphic information content (PIC) was calculated according to Ghislain et al. (1999) by the formula: PIC = 1 -p2 -q2, where p is band frequency, and q is no band frequency. Genetic distances were calculated for all Taraxacum accessions, according to Nei and Li (1979), followed by a dendrogram construction using the unweighted pair group method, with arithmetic average (UPGMA), using the Treecon ver. 3.1 software (Van de Peer and De Wachter 1994). Statistical support of the branches was tested with bootstrap analysis using 2000 replicates.

2C DNA content measurements
Genome size estimation was prepared according to the procedure of Jedrzejczyk and Sliwinska (2010) with minor modifications. The leaves of Vicia villosa 'Minikowska' (2C = 3.32 pg; Dzialuk et al. 2007) were used as an internal standard. Young and fresh leaves of 11 Taraxacum species (Table 6) and the internal standard were chopped simultaneously with a sharp razor blade in a plastic Petri dish with 1 ml of Galbraith's nucleus-isolation buffer (Galbraith et al. 1983) supplemented with an antioxidant of 1% (w/v) polyvinylpyrrolidone (PVP-10), propidium iodide (PI, 50 µg/mL) and ribonuclease A (50 µg/mL). The nuclei suspension was passed through a 50-µm mesh nylon and for each sample, 5000-7000 nuclei were measured using a CyFlow Ploidy Analyser (Sysmex Partec GmbH, Görlitz, Germany) equipped with a high-grade solid state laser with green light emission at 532 nm as well as with side (SSC) and forward (FSC) scatters. Histograms were evaluated using the CyFlow Cube software (Sysmex Partec GmbH, Görlitz, Germany). Genome size was estimated using the linear relationship between the ratio of Taraxacum and the internal standard 2C peak positions on the histogram. At least three replicates were analysed for each Taraxacum species. Mean coefficients of variation of the 2C DNA content were estimated for all samples and ranged from 5.00 to 6.20 (Table 7). The 2C DNA content (pg) was transformed to megabase pairs (Mbp) of nucleotides using the following conversion: 1 pg = 978 Mbp (Doležel and Bartoš 2005). The results were estimated using a one-way analysis of variance followed by Duncan's test (P < 0.05; Statistica v. 13, StatSoft, Poland).

Distribution data
Distribution data of species from the Taraxacum sect. Erythrosperma were obtained from herbarium collections, published taxonomic studies (Uhlemann 2003;Vašut 2003;Schmid et al. 2004;Vašut et al. 2005;Dudáš 2014Dudáš , 2018Zámečník 2016;Dudáš et al. 2020;Nobis et al. 2020b;Dudáš and Vašut 2022), and our herbarium materials collected in the field. Since the studied species are a group of critical, morphologically similar taxa, we decided not to use data from available online databases. We determined the geographical coordinates of records from herbaria and the literature that had only locality descriptions and entered all coordinates into the WGS84 coordinate system. To avoid statistical artefacts related to pseudoreplications, only one datum of the species was considered for each 1 km cell of the grid (in correspondence with the resolutions of environmental layers used in our study  (Hijmans et al. 2005; available online: http://www.worldclim.org/). Soil variables, including bulk density in tonnes per cubic-meter (bld), weight percentage of clay particles (< 0.0002 mm; cly), weight percentage of silt particles (0.0002-0.05 mm; slt), weight percentage of sand particles (0.05-2 mm; snd), soil organic carbon content in permilles (orc), volume percentage of coarse fragments (> 2 mm; crf ), cation exchange capacity in cmol+/kg (cec), soil water-holding capacities (AWCtS), and pH (soil pH × 10 in H2O), were derived from the ISRIC database (Hengl et al. 2017) (https://www.isric.org/). We used layers at a spatial resolution of 30 arcseconds for both bioclimatic and soil variables.

Ecological niche modelling
To model the potential distribution of species from Taraxacum sect. Erythrosperma, we used MaxEnt software version 3.3.3 k., a generative species distribution modelling tool recommended for applications involving presence-only datasets (Phillips et al. 2006;Phillips and Dudík 2008). We ran the model with default values (a maximum of 500 iterations, convergence threshold 0.00001, and five auto feature classes). We opted for a logistic format, as it is currently considered easier and potentially more accurate for interpretation than cumulative and raw approaches (Phillips and Dudík 2008;Baldwin 2009). Because some of the occurrence data were determined on the basis of descriptions of locations and maps, not coordinates obtained in the field, we decided to use 10 percentile training presence as a threshold rule, which eliminated the most outlying data. The model was calibrated using 75% of the occurrence records and tested on the remaining 25%. We performed 20 replicates using the subsample replicated run type and then averaged the results. To provide a different random test/ train partition in each replicate, we used the 'random seed' option. We evaluated the final model using the threshold-independent area under the curve (AUC) generated automatically by MaxEnt (Phillips et al. 2006).
To select a set of variables appropriate for all species from the studied section, the initial model was run using all the above-mentioned variables as well as all 711 localities from all species. To avoid overfitting the model, we built a correlation matrix (Pearson's correlation coefficient) and removed highly correlated variables (r > 0.7). To choose which of the strongly correlated variables to remove, we performed a jackknife test of variable importance and eliminated variables that showed low or negative gain values (Baldwin 2009). However, when selecting the variables, we also took into account whether the variable could be easily explained from a biological point of view. Finally, we ran 17 models, separately for each species. To make the niches of individual species comparable, all models were run on the same set of 11 variables, six bioclimatic (bio3, bio7, bio10, bio11, bio18, bio19), and five physical and chemical properties of soil (awcts, cly, crf, orc, pH) (for abbreviation see Environmental data chapter). To make the models easier to interpret, we divided the probability of occurrence into 5 categories: very low (<0.2), low (0.2-0.4), medium (0.4-0.6), high (0.6-0.8), very high (>0.8), by using ArcMap 10.5 software (ESRI Inc 2016).

SCoT markers analysis
We performed Internal transcribed spacer (ITS) analysis in the initial phase of the studies. The total alignment across the 32 individuals sampled was 680 bp. Although the alignment revealed differences in sequence length between the samples of dandelions, the tree topologies from the Bayesian inference method contained many polytomies, and were inconsistent with the morphological variation of studied species. Thus, we decided to use highly variable SCoT markers to differentiate species and establish the taxonomic relationships within section Erythrosperma. In total, 34 Taraxacum samples were analysed using 19 SCoT primers, which revealed reproducible band patterns. The primers amplified 319 loci, with 301 polymorphic bands. The number of bands generated per primer varied from 10 (SCoT-7) to 26 (SCoT-33). The size of the amplified bands ranged between 170 and 3000 bp. The percentage of polymorphism ranged from 87 to 100%, with an average of 94%. The PIC value, which describes the informativeness of the primer, varied from 0.33 (SCoT-17) to 0.50 , with an average of 0.45 (Table 5). The genetic distances estimated between 34 accessions of Taraxacum ranged from 0.03 to 0.56 (Suppl. material 1). In contrast to ITS, the UPGMA analysis based on SCoT markers revealed that samples belonging to the same taxon were grouped together, within the same cluster, thus confirming their proper taxonomic identification. However, the ordination of clusters in the UPGMA dendrogram is partially in polytomy. The largest clade with the two sister subclades comprise eight species in total: T. bellicum, T. brachyglossum, T. danubium, T. cristatum, T. disseminatum and T. dissimile in the first and T. scanicum and T. plumbeum in the second (Fig. 1). It is in polytomy with the subsequent two clusters, which comprise specimens belonging to T. tenuilobum, T. parnassicum and T. tortilobum. The remaining species, i.e., T. sandomiriense, T. lacistophyllum as well as T. stridulum and T. jugiferum form the outermost branches of the three. Although the last two mentioned species, as representatives of the section Taraxacum, represent an outgroup, based on SCoT analysis, T. stridulum was located closer to T. sandomiriense than to T. jugiferum. Whereas T. jugiferum was the most distant and not clustered with any of the examined species (Fig. 1). Similar results in terms of clustering of particular samples were presented by Reisch (2004). The author studied six species of erythrosperms, of which five also occurred in Poland. However, compared to our results, the relation of species segregated into particular clusters was somewhat different.

Genome size of the examined Taraxacum species
The 2C DNA content of the 11 studied species ranged from 2.29 pg in T. cristatum and T. danubium to 2.76 pg in T. lacistophyllum, which corresponds to 2,240 and 2,699 Mbp, respectively (Table 7). All studied species possessed a very small genome size (Soltis et al. 2003), however, this is in line with the genomic size of other triploid representatives of the genus Taraxacum studied so far (Záveský et al. 2005;Macháčková et al. 2018). Statistical differences in genome size between species were detected, and two species (T. lacistophyllum and T. parnassicum) could be distinguished based on 2C DNA content. Within the species examined to date, similar minor variations or even no significant differences in genome size were observed, which may be explained by their asexual reproduction mode (Záveský et al. 2005;Macháčková et al. 2018). The differences in genome size (2.35 pg/2C vs. 2.62 pg/2C) in T. brachyglossum between our studies and the previous ones (Záveský et al. 2005) may result from both natural variance in DNA content as well as differences in the measurement methodology (Macháčková et al. 2018). To the best of our knowledge, the presented results provided new data on the genome size for 10 Taraxacum species.

Macro-and micromorphology of achenes
The shape and colour of achenes are important morphological features that greatly facilitate the identification of species representing the section Erythrosperma (Tacik 1980;Vašut 2003;Savadkoohi et al. 2012;Rewicz et al. 2020), (Figs 2, 3). During field work, we observed that, depending on the habitat conditions in which particular dandelions grow, the size of their achenes varies considerably, e.g. in the population of T. lacistophyllum from Roland pleasure ground in Gdańsk, the achenes of individuals growing in shadow (under the canopy of trees) were almost twice as long as compared to specimens growing in extremely dry conditions a few meters away. Preliminary analysis of five measurable achene features in Erythrosperma species showed very high similarity and a similar range of variability. All the examined taxa have rather similar achenes in terms of cone length, achene body width, length of the spinose part of the achene body, and beak length. Achene length (incl. cone) is one of the most species-specific morphological characters. Three species, T. tortilobum, T. proximum and T. dissimile, have the longest achenes as well as the longest cone, whereas the achenes of T. tenuilobum, T. plumbeum and T. danubium are the shortest (Fig. 4). However, there is also a group of the three species, T. scanicum, T. brachyglossum and T. cristatum, in which the length of achenes varies considerably. SEMs observations of achenes showed some morphological differences in the achenes' ornamentation (Figs 5-7), and in particular the spines' shape and extent of their fusion with the pericarp surface. For example in T. parnassicum and T. proximum (Figs 5-7H, J) the spines protrude only at the ends, while in T. tenuilobum the spines are slender and not fused in almost their entire

Distribution of Taraxacum sect. Erythrosperma in Poland
Of the 14 examined species of Taraxacum sect. Erythrosperma in Poland, 7 are definitely rare, known from 3 to 13 localities to date. Three of them (T. danubium, T. cristatum, T. sandomiriense) are distributed in south-central Poland in relatively small areas, the next 3 are known from the north-eastern part of the country (T. dissimile) and the Baltic Sea seashore (T. tortilobum, T. lacistophyllum). Another rare species, T. disseminatum, is known from 11 localities scattered over a relatively large area. The other species from the section are fairly frequent (>20 localities), although they are known from no more than 50 localities. The most common is T. scanicum, known from 42 localities. As for the concentration of T. sect. Erythrosperma species-localities per grid square, the highest is observed within the Kraków-Częstochowa Upland, the Gdańsk Coastland and the Wielkopolska Lowland (Fig. 8).

Potential distribution modelling
The distribution model was performed for 11 species of dandelions. All the models show a high value of AUC (0.98 up to 1.00), which confirmed their reliability (Table 8). Variables with a relatively higher percentage of contribution in the MaxEnt models for the greatest number of species were bio11, bio3 and bio7 and, among soil factors, crf (Table 9). For most species, the area of high and very high probability of occurrence is quite large, indicating that these species may be much more common in Central Europe than previously thought, and their poorly recognised distribution is an effect of insufficient study. Such species include T. bellicum, T. cristatum, T. danubium, T. parnassicum or T. plumbeum. All of these species are characterised by a similar pattern of potential distribution, covering steppe regions of Central Europe, from southern (Bavaria) and north-eastern Germany (areas on the middle and lower Elbe river valley), through central and southern Czech Republic (including Moravia), northern and central Slovakia, north-eastern Austria (on the Danube), the highlands of Silesia and Central Poland, the valleys of the lower Odra and Vistula rivers, to eastern Poland, and in the case of some species also south-western Ukraine ( the general pattern of distribution is similar, but the area of high probability of occurrence is much smaller (Fig. 9B). Three species (T. lacistophyllum, T. proximum, T. scanicum), are characterised by a potentially more concentrated range in the north-western part of Poland (mostly Pomerania and Silesia) and eastern Germany (the middle Elbe river valley, Saxony and Brandenburg), (Fig. 9F, I, J). This is especially noticeable in the case of T. lacistophyllum, which in Poland probably occurs only in Pomerania, but is likely much more common in Germany (Fig. 9F). Two other species, T. disseminatum and T. tenuilobum, show rather scattered potential distribution patterns, with high and very high probability of occurrence in central and northern Czech Republic, eastern Germany (the middle Elbe river valley, Saxony and Brandenburg) to Pomerania (both in Germany and Poland), the valley of the lower Odra and the Vistula, highlands in Silesia, central, southern, and eastern Poland, as well as western Ukraine (Fig. 9E, K). In the case of T. dissimile, T. sandomiriense, and T. tortilobum, we were unable to construct reliable models due to an extremely small number of known localities (Figs 10F, 23E, 36B).

Remarks on plant collection and species identification
Determination of species representing the section Erythrosperma can be difficult for beginners, who are not familiar with the morphological variability that is observed in Figure 8. Collective distribution of Taraxacum sect. Erythrosperma species in Poland; 1 -5-6 species per 10 km × 10 km square, 2 -3-4 species per 10 km × 10 km square, 3 -1-2 species per 10 km × 10 km square. the field. Except for some features within the inflorescence, most of the measurable features are characterised by a very wide range of variability. During determination, it is extremely important to carefully analyse qualitative features, such as the absence or presence of pollen; the shape, colour and arrangement of the outer bracts; the shape of the capitulum, the shape of the terminal lobe, side lobes, and interlobes; the presence or absence of teeth on lobes and in the interlobes; the colour and hairiness of the leaves and peduncles; the colour of the flowers and achenes. Some quantitative features are also important, e.g. the number of side lobes and outer bracts. Species- Figure 9. Models of the potential distribution of selected species of Taraxacum sect. Erythrosperma probability of occurrence: very low (grey), low (green), medium (yellow), high (orange), very high (red).
specific features are best visible in the field, in numerous populations, preferably in full flowering/beginning of fruiting time (in Poland, this period begins in the second half of April in the Uplands and in the first week of May in the north and in the mountains; in Poland this period overlaps with the flowering of Prunus spinosa).
Rainless and warm springs are favourable for field research. If spring is rainy and cold, small plants from the Erythrosperma section are usually overgrown by grass and other perennials; they then lose their diagnostic features and are hardly noticeable from a greater distance. In the field, it is worth noting features such as the arrangement and colour of outer bracts, the colour of petioles, and the colour and shape of the capitulum. It is crucial to carefully dry collected plants after the harvest; this makes later determination much easier. All data should be taken into consideration during determination, and the specimen should be compared both with the identification key and the description.

Overall description of section Erythrosperma
Plants mostly small to middle-sized, often forming a tunic of dried leaf leftovers. Leaves usually deeply lobed with narrow lobes and petioles. Scapes often slender, thin. Outer bracts usually small, often with cornicules. Capitulum mostly small, flowers often light yellow, sometimes golden yellow. Achenes mostly red, less often straw-coloured, strongly spinulose with a cylindrical or narrowly conical cone, narrow at the base. Plants bloom in early spring (from mid-April). Related to warm and sunlit habitats. Description. Plants small to middle-sized, 5-12(-15) cm tall. Leaves greenish, almost glabrous, (5-)7-15(-20) cm long and 1.5-3 cm wide, generally 3-6 times longer than wide, blades broadest in middle, with 3-5(-6) pairs of lateral lobes; lateral lobes of the inner leaves patent or slightly recurved, narrowly triangular, acute, with an entire or slightly dentate distal margin, proximal margin usually entire or with a few small teeth; lateral lobes of the outer leaves triangular, proximal margin usually entire, distal margin entire or slightly dentate; interlobes often toothed, sometimes blackish rimmed; terminal lobe of the inner leaves mostly with lingulate apex; terminal lobe of the outer and medial leaves triangular or slightly lingulate, usually packed lateral lobes below; petioles unwinged, pale purple to pale brown-purple. Scapes as long as or longer than leaves, green suffused with pale purple, hairy below capitulum. Capitulum slightly convex, 2.5-3.5 cm in diameter, dark yellow, outer strips grey brown; inner bracts greyish-green, corniculate, outer bracts usually 10-14, lanceolate, usually 6.5-9 mm long, 1-3 mm broad, usually red-violet, hyaline margin inconspicuous (up to 0.1 mm broad), regularly recurved, usually with small cornicules; stigmas yellow-greenish, yellow-green-blackish after drying, pollen present. Achenes greyish purple-brown, sparsely spinulose at the top, 3.5-4.0 mm long (incl. the 1.0-1.3 mm long, narrowly conical cone), rostrum 6.0-7.0 mm long, pappus white.
Habitat. In the Polish lowlands this species occurs in a wide spectrum of habitats; mostly in dry, sandy semiruderal locations exposed to the sun, e.g. roadsides, paths in dry pine forests, sandy embankments, dry pastures, sandy paths in cemeteries (especially in Wielkopolska Lowland); plant communities with its participation are dominated by species characteristic to the Molinio-Arrhenatheretea and Sedo-Scleranthetea classes. In Podlchia (Klimaszewnica) (Uhlemann 2003;Horn et al. 2004;Schmid et al. 2004;Nobis et al. 2020b). Populations from Finland are most likely of anthropogenic origin (EURO+MED 2006-onwards).
Distribution in Poland. Scattered localities, quite frequent in Podlachia, the western part of Lesser Poland and Greater Poland (Fig. 10A). Notes. The species shows high morphological variability within leaf shape and the position and colour of outer phyllaries. This variability is evident among the populations from the Polish lowlands, often found in semi-shaded semi-ruderal and ruderal habitats such as sandy and gravely roadsides, backyards, sandy roads and paths in the forests and thickets. Features typical of the species, such as regularly recurved, redpurple, narrowly-edged outer phyllaries, or the distinct lingously elongated apex of the inner leaves terminal lobe, are well visible in specimens growing in stable, dry and full sun habitats, e.g. in sandy grasslands and rock grasslands in the south of Poland. Due to high morphological plasticity, the species can sometimes be confused with T. scanicum, which differs from T. bellicum, e.g. outer phyllaries are distinctly bordered (0.1-0.2 mm), mostly green, and the leaves' side lobes are regularly incised (Figs 11, 12).  Taraxacum erythrospermum subsp. brachyglossum Dahlst., Bot. Not., 1905: 170. 1905 Basionym.
Distribution in Poland. Scattered localities in S Poland, quite frequent in W part of Lesser Poland (Fig. 10B).   Notes. Species distinguished by dark green leaves with side lobes narrowly triangle and bent downwards, outer phyllaries relatively wide, greyish-purple, narrowly bordered, and often a fully flowering capitulum partly-opening and dark yellow. Species morphologically variable; in specimens found in very dry, rocky habitats, the side lobes of the tripartite terminal lobe are very often positioned upwards, which often helps in their identification (Figs 13, 14). Description. Plants usually small, 5-10 cm tall. Leaves (pale) green, almost glabrous, approximately (3-)5-10 cm long and (1-)2-2.5(-3.5) cm wide, usually 4-5 times longer than wide, blades eliptical or oblanceolate, with 3-4 pairs of lateral lobes; lateral lobes mostly opposite; lateral lobes of the inner leaves narrowly triangular, falcate, with a dentate, convex distal margin, proximal margin entire or with a few teeth; lateral lobes of the outer leaves triangular, entire or somewhat denticulate at the distal margin; interlobes narrow and long, undulate or denticulate, often dark maculate; terminal lobe of the inner leaves prolate, lingulate and denticulate at the base; terminal lobe of the outer leaves triangular, prolate, undulate at the base; petioles unwinged, reddish-purple, almost glabrous. Scapes as long as or slightly lon ger than leaves, almost glabrous or with few barely visible hairs. Capitulum convex, 2-2.5 cm in diameter, yellow, outer strips greyish-brown-purple; inner bracts greyish-green, often suffused with purple at the ends, corniculate; outer bracts usually 9-11, lanceolate, usually 6-8 mm long, 1.5-2 mm broad, pale green, suffused pale red-purple, with a white hyaline margin (0.05-0.1 mm broad), recurved and corniculate; stigmas olive-greyish, pollen present. Achenes purplish-brown, with thin spinules in the upper part, 3.5-4.0 mm long (incl. the 0.8-1.1(-1.3) mm long, narrowly conical pyramid), rostrum 5.5-7.1 mm long, pappus white.

Flowering period. April (May).
Notes. Species belonging to the Scanicum group, similar to the rest of the species from this group, with an asymmetrically incised terminal lobe. However, compared to T. bellicum and T. scanicum, the terminal lobe in T. cristatum is much more denticulate, as is the distal margin of the side lobes. T. cristatum may sometimes closely resemble T. plumbeum (especially individuals of T. plumbeum growing in extremely dry, rocky habitats), but it differs from it in its purple-brown achenes and lower number of pairs of side lobes (3-4) (Figs 15, 16). Description. Plants usually small, up to 10(-12) cm tall. Leaves greyish-green, dull, sparsely hairy, approximately 3-5(-7) cm long and (1-)1.5-2.5 cm wide, usually 3-4 times longer than wide, blades oblanceolate, usually broadest in upper 1/3, with 3-4 pairs of lateral lobes; lateral lobes opposite to remote; lateral lobes of the inner leaves patent, with a wide abruptly narrowed base and generally slightly widening at the apex, entire or with a few small teeth at the margin; lateral lobes of the outer leaves recurved and obtuse at the apex, entire or occasionally with a few small teeth at the margin; interlobes often with teeth; terminal lobe of the inner leaves triangular, often with a distinct short and obtuse tip; terminal lobe of the outer leaves triangular, obtuse; petioles narrowly winged, pale purplish. Scapes as long as or slightly longer than leaves, reddish-purplish, sparsely hairy in the upper part. Capitulum convex, yellow, 2-3 cm in diameter, ligules with greyish brown-red stripes; inner bracts greyish-green, corniculate; outer bracts usually 10-14, lanceolate, usually 4-6 mm long, 1.5-2.5 mm broad, greyish-green, quite often suffused purple, with a white hyaline margin 0.1(-0.2) mm broad, regularly spreading to quite regularly arranged and recurved, 4-6 mm long, 1.5-2.5 mm broad, corniculate; stigmas greyish-green, pollen present. Achenes dark brown-red, achene body densely spinulose above, 3.3-3.8 mm long (incl. the 0.7-1.0 mm long, narrowly conical cone).
Distribution in Poland. Very rare, so far only found in the western part of Lesser Poland (Fig. 10D). Notes. Species quite small, with sparsely hairy and dull leaves, usually narrow interlobes and side lobes patent or slightly bent, often a little bloated near the ends. Juvenile specimens of T. danubium often have poorly split leaves in the upper part, which makes them similar to T. parnassicum, but due to the presence of pollen, the leaves of T. danubium are hairy and its outer phyllaries longer and wider, and therefore these two species can be easily distinguished. In the populations of T. danubium observed in Poland, the vast majority were young individuals with small rosettes with leaves shaped similarly to the external leaves of several-year-old specimens (Figs 17, 18). Type. Sweden, Göteborg, 9 May 1943, T. A. Borgvall (holotype in S).
Description. Plants middle to quite large-sized, 5-15(-20) cm tall. Leaves greyish-green, sparsely hairy on the upper side, approximately (5-)7-12(-15) cm long and (1.5)2-3(-4.0) cm wide, usually 3-4 times longer than wide, blades elliptical to oblanceolate, with 3-4(-6) pairs of lateral lobes; lateral lobes opposite to remote; lateral lobes of the inner leaves triangular, broad at the base, with a convex, strongly dentate and often incised distal margin, proximal margin usually entire or with a few teeth; lateral lobes of the outer leaves triangular, uniform, broad and short, with strong teeth at often incised and convex distal margin, proximal margin usually entire and slightly concave; interlobes narrow; terminal lobe of the inner leaves triangular, somewhat elongate, sometimes lingulate, denticulate on the upper margins; terminal lobe of the outer leaves triangular, subacute, entire or with a large tooth on the upper margins; petioles unwinged, purple. Scapes as long as or longer than leaves, sparsely hairy, especially under the capitulum, their lower parts usually purple in colour. Capitulum convex, 2.5-4.0 cm in diameter, yellow, medium dense, outer strips grey-purple; inner bracts dark grey-green, pruinose; outer bracts usually 9-12, lanceolate, usually 6-10 mm long, 2-3.5 mm broad, grey-green, with a distinct white hyaline margin (0.1-0.3 mm broad), arcuate-reflexed, without or with a small corniculum; stigmas dark, greyish-green, pollen present. Achenes red-brown, with thin and long spinules in the upper part, 3.5-4.2 mm long (incl. the 1.0-1.4 mm long, cylindrical pyramid), rostrum 7-9 mm long, pappus white.
Distribution in Poland. Scattered localities, rare (Fig. 10E). Notes. Plant quite large with a medium dense capitulum (particularly visible in the peripheral part of the inflorescence) up to 4 cm in diameter. Leaves broad with a rather large triangular terminal lobe. The terminal lobe edge is strongly lobed and serrated in the base part. Outer bracts with significant wide hyaline margin. Species distinct and easy to recognise, although not very common, and usually populations are not numerous (Figs 19, 20). Type. Sweden, Gothenburg archipelago, Branno, seashore, 19 May 1910, Th. Lange (lectotype in TURA [sheet No. I, middle specimen], designated by Lundevall and Øllgaard 1999: 78;isolectotype in TURA [sheets No. 2 and 3]).
Habitat. Species most often found in semiruderal locations, such as sandy and sunny edges of pine forests, paths, cliffs; less often in ruderal habitats (concrete promenades, walls). On the coast of the Baltic Sea (Gdańsk) we noted this species on the edge of a sandy forest road, accompanied by Achillea millefolium, Agrostis capillaris, Alliaria petiolata, Anthriscus sylvestris, Artemisia vulgaris, Berteroa incana, Hypericum maculatum, Melandrium album, Plantago major, Potentilla argentea, Tanacetum vulgare, Tragopogon pratensis.
Distribution in Poland. Species noted only in Pomerania, chiefly on the coast of the Baltic Sea (Fig. 23A).    Notes. Plant charming, gentle, with tasteful capitulum up to 4 cm in diameter, light yellow ligules, outer bracts spreading-arcuate, greyish-green/violet, pruinose. Leaves regularly lobed, side lobes most often falcate and interlobes often crisped. Species easy to recognise (morphological features of the leaves are highly visible, even for specimens growing in unusual places) (Figs 24, 25).  Lundevall and Øllgaard 1999: 125; isolectotypes in S and BM).
Habitat. In the south of Poland, T. parnassicum usually grows in thermophilic grasslands on limestone rocks (most often in trampled or eroded places) and in rock crevices. In the north, this species was recorded in sandy grasslands and on a dry lawn. Notes. Plant usually small. Leaves with 4-7 pairs of uniform lateral lobes and narrow interlobes, side lobe distal margin often convex and entire. Capitulum small, with light yellow ligules, no pollen or only rarely a few poorly developed grains present. Fruit with relatively short spinules. Species not very morphologically variable, easy to recognise, charming (Figs 26,27).  = Taraxacum franconicum Sahlin, Ber. Bayer. Bot. Ges. 55: 49. 1984 Description. Plants small to middle-sized, 5-10(-15) cm tall. Leaves dark green, dull, sparsely hairy, approximately 5-12 cm long and 1.5-2.5(-3.0) cm wide, usually 5-7 times longer than wide, blades narrowly elliptical to narrowly oblanceolate, with 4-6 pairs of lateral lobes; lateral lobes opposite to remote; lateral lobes of the inner leaves narrowly triangular, usually falcate, acute, with a somewhat convex, often denticulate distal margin, proximal margin usually entire, concave; lateral lobes of the outer leaves triangular, distinctly falcate, with an entire or denticulate distal margin; interlobes often long and narrow, plicate and denticulate, blackish rimmed; terminal lobe of the inner leaves with lingulate apex, denticulate margins and/or in-were relatively homogeneous in their morphological features, such as: leaves with quite wide, entire or slightly toothed side lobes and suberect or patent outer phyllaries. In upland areas, in rock grasslands, the species show greater variability of morphology compared to lowland populations. In general, side lobes are narrower, more numerous and slightly more serrated, interlobia are wider and incised, and outer phyllaries are narrower and more recurved. Specimens growing in extremely dry and rocky habitats, usually with strongly dissected leaves, may resemble T. tenuilobum, however the terminal lobe in T. plumbeum has a slightly different shape, with a tongue-shaped apex slightly incised on both sides in the base, and much smaller teeth and lobules in the interlobes in relation to the side lobes. The yellowish light red-brown hue of the achenes is a very useful diagnostic feature typical of T. plumbeum (Figs 28, 29).  Taraxacum erythrospermum subsp. proximum Dahlst., Bot. Not. 1905: 165. 1905. Basionym. = Taraxacum attenuatum Brenner, Meddeland. Soc. Fauna Fl. Fenn. 32: 114. 1906. Type: Finland, Nylandia, Ingå (Inkoo), Svartbäck, dry hill, 17 August 1905, M. Brenner (lectotype in H 660607, designated by Lundevall and Øllgaard 1999. longer than wide; leaf blade elliptical, regularly lobate, with 4-8 pairs of lateral lobes; lateral lobes opposite to remote; lateral lobes of the inner leaves triangular, acute, patent, with a regularly dentate, slightly convex distal margin, proximal margin usually entire; lateral lobes of the outer leaves triangular, usually toothed at the distal margin; interlobes often short, blade often toothed in lower part of leaf; terminal lobe subacute or subsaginate, quite often with elongate apex; petioles unwinged, purplish. Scapes as long as or longer than leaves, somewhat hairy. Capitulum convex, 2.5-3.0 cm in diameter, greenish-yellow, with numerous tubular inner flowers, outer strips purple-brownish; inner bracts greyish-green, with lumps or small cornicules; outer bracts usually 11-14, lanceolate, usually 7-9 mm long, 2-3.0 mm broad, bright greyish-green suffused with purple, narrowly bordered (up to 0.05 broad), recurved, with lumps or small cornicules; stigmas greyish-green, pollen absent or very poorly developed (up to a few grains on the stigma). Achenes reddish-brown, narrow, with erect thin spinules in the upper part, 3.5-4.1(-4.5) mm long (incl. the 0.8-1.1(-1.4) mm long, cylindrical cone), rostrum (6-)7-8(-8.5) mm long, pappus white.
Distribution in Poland. Scattered localities mainly in northern Poland, quite frequent (Fig. 23F). Notes. Plants with narrow leaves. Lobes tend to be dissected and petioles suffused purple. Outer bracts grey-green with white hyaline margin, often lightly suffused red-violet, recurved or patent. Fully flowering capitulum convex, yellow, sigmas dark. Fruits brown-red, deeply coloured (Figs 34, 35).
Habitat. In the northern part of Poland, this species grows most often in dry and sandy habitats, such as sandy grasslands, roadsides of forest roads, edges of pine fo rests, paths, cliffs, dunes, and lawns. In southern Poland, we noticed this species most often in rock grasslands (in eroding and trampled areas). Plant communities with the participation of T. tenuilobum were predominated by species typical to the Festuco-Brometea (especially in S Poland) and Sedo-Scleranthetea classes. In Świętokrzyskie Mts (Miedzianka place) we noted this species in rocky grassland growing together with Allium montanum, Arenaria serpyllifolia, Artemisia Tutin et al. 1976;Tacik 1980;Fedorov 1989;Lundevall and Øllgaard 1999;Mosyakin and Fedoronchuk 1999;Uhlemann 2003;Wendt and Øllgaard 2015). Notes. Species included in the Dissimilia group, easily identified by a combination of pale grey-brown achenes, leaves strongly crisped, lateral lobes often toothed and curled, outer phyllaries loosely adpressed to obliquely spreading (Figs 39, 40).

Species not confirmed in Poland
Based on the literature data, 25 Taraxacum species from the section Erythrosperma have been reported from Poland to date (Table 1). However, the presence of 10 species listed by Tacik (1980) (Mirek et al. 1995(Mirek et al. , 2002(Mirek et al. , 2020, were not confirmed during the revision of the herbarium materials. All of the above-mentioned species had been misidentified with other species. Moreover, the occurrence of the species was also not confirmed during the field studies. It is worth menti oning that, with the exception of T. erythrospermum, all of these species are also absent in the neighbouring countries/regions (eastern Germany, Czech Republic) (Uhlemann 2003;Trávníček et al. 2010). The specimens of T. erythrospermum collected in the vicinity of Kraków (Tacik 1980) were misidentified and actually belonged to T. danubium . In 2005, Vašut et al. (2005 reported the presence of T. maricum in Solec, however, their determination of the specimens was uncertain. During field studies, we could not confirm the presence of T. maricum in Solec, despite intensive field penetration at the locality. Thus, in the face of the lack of additional herbarium material of this species from Poland, we decided to exclude it from the list.

Threats
Using the IUCN (2022) threat categories, most of the investigated species (T. bellicum, T. brachyglossum, T. lacistophyllum, T. plumbeum, T. proximum, T. scanicum, T. tenuilobum) should be considered as of least concern (LC) in Poland. The species listed above are rare in Poland, growing in large dispersion (T. bellicum, T. brachyglossum, T. plumbeum, T. scanicum, T. tenuilobum), the others (T. lacistophyllum, T. proximum) occur more regionally, often abundantly, in semi-ruderal habitats ( Fig. 41A-C). One gets the impression that human activity causes them more benefits than harm, for example, in Wielkopolska Lowland, T. scanicum, T. proximum, T. plumbeum, T. bellicum grow on intensively trampled paths, on roadsides or even in cemeteries. We also included the rarer plants (T. disseminatum, T. dissimile) in this category, due to the relatively large area of occurrence and the lack of noticeable factors that could threaten them at present. T. parnassicum is considered a species near to threat (NT) due to its close relation to specific, rare habitats and observed unfavourable habitat transformations that clearly threaten this species, e.g. xerothermic rock grasslands overgrowing due to lack of grazing (Fig. 41D). A significant part of the private property limestone rocks in the Kraków-Częstochowa Upland is also successively fenced, which definitely accelerates the overgrowing by shrubs due to the lack of touristic exploration. In the Lower Silesia, many localities of T. parnassicum reported in the nineteenth and second half of the twentieth century are now most likely historical, which may have been caused by the intensification of agriculture in this area and secondary succession in closed sand mines. T. tortilobum is classified as a vulnerable species (VU) due to very small and limited populations (the species was found only in Gdańsk). T. danubium should be recognised as an endangered species (EN) due to its limited range in Poland, a very low number of sites and very significant fluctuations observed in the number of mature individuals. It seems that a significant factor preserving this species in the largest localities (Olsztyn, Skały Twardowskiego in Kraków) is their being re creational destinations (Fig. 41E). T. cristatum should be considered critically endangered (CR) in Poland due to the extremely low number of its localities, small populations and the observed decline in the number of individuals. The same risk category (CR) was assigned to T. sandomiriense due to the extremely small number of sites and mature plants within them as well as the tendency for its habitats to decline (very rare and specific to this species; Fig. 41F).
manuscript. Special thanks are given by the first author to his wife Magdalena Wolanin for her invaluable help during field trips.
Open Access of this paper was founded by University of Rzeszów.