Updated Taxonomy of Iris scariosa (Iridaceae) Inferred from Morphological and Chloroplast DNA Sequence Data with Remarks on Classification of Iris subg. Iris

Iris scariosa is a rhizomatous perennial whose taxonomy and distribution range still remain unclear. The results of our examination of literature, specimens, and wild plants have shown that I. glaucescens, described from Kazakhstan, and I. timofejewii, considered to be endemic to the Republic of Dagestan, Russia, are very closely related to I. scariosa. We have carried out molecular phylogenetic analyses for the first time to clarify the taxonomy of I. scariosa. For this, we sequenced six chloroplast DNA regions of an extended sampling that comprised the accepted species I. glaucescens and I. timofejewii, which has revealed their strong affinity to the accession of I. scariosa from the vicinity of Astrakhan, Russia. A thorough revision of the morphological characters has confirmed the lack of evident differences between I. scariosa and I. timofejewii. Thus, the analyses support a broad species circumscription of I. scariosa. We here reduce I. timofejewii, as well as I. curvifolia, considered to be endemic to Xinjiang, western China, to synonymy of I. scariosa. Color illustrations, updated nomenclature, and data on distribution of I. scariosa are provided. A lectotype for I. astrachanica and a neotype for I. timofejewii are designated here. Also, the phylogenetic relationships within I. subg. Iris are outlined, and an updated classification of the subgenus is proposed. We have recovered six major lineages within four major clades which we recognize as sections. Here, we propose two new nomenclatural combinations, a revised taxonomic treatment, and a new identification key to I. subg. Iris.


Introduction
When carrying out studies in the framework of the taxonomic research on the genus Iris L. (Iridaceae) in Russia, we paid attention to the still unresolved taxonomy and distribution range of I. scariosa Willd.ex Link.It is a rhizomatous perennial distributed in Eurasia.The name I. scariosa was described by Link without indication of the collection locality [1].It was considered as originally described from specimens collected in Siberia [2,3] or near the Caspian Sea, from the Volga River estuary [4][5][6][7].
Iris scariosa is distinguished by its membranous bracts and extremely glaucous leaves (Figure 1a).The latter feature was noticed by Bunge when he gave the plant the name I. glaucescens Bunge [8].It was described from plants collected in the East Kazakhstan Region, Kazakhstan [8] (p.58).Due to their morphological similarity, I. glaucescens was considered a synonym of I. scariosa for almost a century [2,4,5,[9][10][11][12][13][14][15][16].Rodionenko [15] assumed that I. scariosa was a hybrid between I. pumila L. and I. glaucescens.Shevchenko [6] believed that there were some morphological differences between the two species, however, without specifying them, and resurrected I. glaucescens from I. scariosa.She also noted that I. scariosa had a small distribution range in the Eastern Ciscaucasia and the Lower Volga, while I. glaucescens had a wide distribution range, covering the southern part of the West Siberian Plain, the Kazakh Uplands, Altai, the Lake Balkhash area, Dzungaria, and northwestern Mongolia.Subsequent authors believed that I. glaucescens was common in Kazakhstan and Western Siberia, and the closely related I. scariosa was an endemic to the lower reaches of the Volga and Don rivers [7,17].Both species are currently accepted with their respective distributions [18][19][20][21][22][23][24][25][26][27][28][29][30].
The major aim of our study was to provide an updated molecular phylogeny of I. scariosa based on an extended sampling that included accessions from the recorded distribution ranges of I. glaucescens and I. timofejewii, i.e., from Kazakhstan and the Republic of Dagestan, Russia, respectively.In this study, we also aimed to examine the morphological characters of I. scariosa and related species, verify the pollen morphology using scanning electron microscopy, and propose a revised taxonomy of I. scariosa with both molecular and morphological evidence taken into account.The taxon sampling for the present molecular study also included representatives of all sections of I. subg.Iris according to references [16,36,37].In this regard, an additional aim was to resolve phylogenetic relationships between the sections in I. subg.Iris on the basis of a plastid sequence dataset for clarifying the classification of the subgenus.

Molecular Study 2.1.1. Taxonomic Sampling
The taxon sampling focused on the species of I. sect.Iris.We attempted to provide as extensive sampling as possible.One of us (E.V. Boltenkov) made botanical expeditions to Armenia in 2015, Kyrgyzstan in 2022, and the Republic of Dagestan, Rostov Oblast, and Stavropol Krai, Russia, in 2015, 2022, and 2023.Also, we used material collected from Bulgaria, Kazakhstan, Tajikistan, Uzbekistan, and Russia by our colleagues and ensured that all accessions were fully verified.The complete list of the examined taxa, including information on samples, is provided in Table 1.The main set of samples for the present study included 15 accessions representing I. timofejewii from the mountainous central and southern part of the Republic of Dagestan (S1-S7) where this species is distributed according to the literature and dedicated websites [22, 23,33], I. scariosa from Russia (S8), and I. glaucescens from Kazakhstan (S9-S14) and Altai Krai, Russia (S15).Two accessions (S1 and S2) were obtained from Botlikhsky Raion (formerly the Andiysky Okrug) of the Republic of Dagestan, Russia, the type locality of I. timofejewii [43].The accession S8 originally comes from the Volga River estuary in the vicinity of Astrakhan, Russia, the alleged type locality of I. scariosa according to references [6,7].The accession S14 was collected near the type locality of I. glaucescens in the vicinity of Ust-Kamenogorsk (or Oskemen), Kazakhstan.The sampling localities for accessions S1-S15 are shown in Figure 2.  The taxon sampling included the following species from all sections of I. subg.Iris according to references [16,36]

DNA Extraction, Amplification, and Sequencing
Total genomic DNA was isolated as previously described [44] from silica-dried leaf materials collected in the field, obtained from living collections or taken from herbarium specimens deposited at LE and TASH (herbarium codes according to reference [38]).
Six plastid markers were used.Four intergenic spacers (IGS) of chloroplast DNA (cpDNA), trnH-psbA, rps4-trnS GGA , trnS-trnG, and trnL-trnF were sequenced for samples from all the sections of I. subg.Iris, except for the species of I. sect.Psammiris and I. sect.Pseudoregelia, for which these regions had already been sequenced previously [39].Amplification and sequencing of these IGS were carried out according to reference [45].In addition to these markers, partial sequences of the plastid genes ndhF (ca.2150 bp) and ycf1 (ca.1030 bp) were amplified and sequenced following the protocols described by Wilson [46,47].The ndhF and ycf1 genes were shown to be useful in resolving phylogenetic relationships between species of the genus Iris [37,46,48].
The cycle sequencing reactions were performed on both strands as described in references [45][46][47], and sequencing products were separated on an ABI 3130 genetic analyzer (Applied Biosystems, Bedford, MA, USA) at the Joint-Use Center "Biotechnology and Genetic Engineering", Federal Scientific Center of the East Asia Terrestrial Biodiver-  The taxon sampling included the following species from all sections of I. subg.Iris according to references [16,36]

DNA Extraction, Amplification, and Sequencing
Total genomic DNA was isolated as previously described [44] from silica-dried leaf materials collected in the field, obtained from living collections or taken from herbarium specimens deposited at LE and TASH (herbarium codes according to reference [38]).
Six plastid markers were used.Four intergenic spacers (IGS) of chloroplast DNA (cpDNA), trnH-psbA, rps4-trnS GGA , trnS-trnG, and trnL-trnF were sequenced for samples from all the sections of I. subg.Iris, except for the species of I. sect.Psammiris and I. sect.Pseudoregelia, for which these regions had already been sequenced previously [39].Amplification and sequencing of these IGS were carried out according to reference [45].In addition to these markers, partial sequences of the plastid genes ndhF (ca.2150 bp) and ycf1 (ca.1030 bp) were amplified and sequenced following the protocols described by Wilson [46,47].The ndhF and ycf1 genes were shown to be useful in resolving phylogenetic relationships between species of the genus Iris [37,46,48].
The cycle sequencing reactions were performed on both strands as described in references [45][46][47], and sequencing products were separated on an ABI 3130 genetic analyzer (Applied Biosystems, Bedford, MA, USA) at the Joint-Use Center "Biotechnology and Genetic Engineering", Federal Scientific Center of the East Asia Terrestrial Biodiversity, Far Eastern Branch, Russian Academy of Sciences (FSC EATB FEB RAS), Vladivostok, Russia.Forward and reverse sequences for each region were assembled using Staden Package version 1.4 [49].All newly generated sequences were deposited in the GenBank database (Table 1).The accession numbers for all the sequences used in the study are listed in Table 1.The sequence data of the ycf1 and/or ndhF plastid genes for I. dichotoma, I. imbricata, and the I. sect.Pseudoregelia and I. ser.Laevigatae (Diels) G.H.M.Lawr.species were accessed from GenBank (Table S1).In addition, sequences of six cpDNA regions were retrieved from the complete plastid genomes of the following I. subg.Iris species available in GenBank (https://www.ncbi.nlm.nih.gov/,accessed on 15 March 2024): I. germanica L. and I. lutescens Lam. from I. sect.Iris, I. sprengeri Siehe, I. lycotis Woronow, I. haynei Baker, and I. gatesii Foster from I. sect.Oncocyclus, I. afghanica Wendelbo and I. hoogiana Dykes from I. sect Regelia, and I. sichuanensis Y.T.Zhao (is a synonym of I. leptophylla Lingelsh.)from I. sect.Pseudoregelia (Table S1).

Sequence Alignment and Phylogenetic Analysis
The sequences of each cpDNA region obtained in this study, together with those retrieved from the chloroplast genomes of nine species belonging to I. sect.Iris, I. sect.Oncocyclus, I. sect.Regelia, and I. sect.Pseudoregelia (Table S1), were aligned manually in SeaView version 4 [50] using the CLUSTAL algorithm and concatenated for each accession.We included indels and length variation in mononucleotide repeats in the dataset because the repeatability tests allowed for removing of PCR errors.Based on this dataset, which also included the I. dichotoma sequences as outgroups, haplotypes were identified using DnaSP version 5 [51].This program was also used to calculate the degree of cpDNA sequence divergence within and between sections of I. subg.Iris based on nucleotide substitutions.
The haplotype network was constructed by the median-joining method (MJ) with default settings as applied in Network version 4.6 [52].An 8 bp inversion within the trnH-psbA spacer and each deletion/insertion, regardless of size, were treated as a single mutational event.
Phylogenetic relationships among irises were assessed by the maximum parsimony (MP) and maximum likelihood (ML) methods as implemented in PAUP version 4.0 b10 [53] and also by the Bayesian inference method (BI) in MrBayes version 3 [54] via the CIPRES portal [55].The dataset for these analyses included the sequences of 47 accessions from six recognized sections of I. subg.Iris and the sequences of 12 species from four series of I. subg.Limniris (I.ser.Lacteae Doronkin, I. ser.Laevigatae, I. ser.Sibiricae (Diels) G.H.M.Lawr., and I. ser.Ruthenicae (Diels) G.H.M.Lawr.), and two species from I. subg.Pardanthopsis (I.dichotoma and I. domestica) as outgroups.For the MP method, optimal trees were found using a heuristic search with 1000 random addition sequence replicates, starting trees obtained via stepwise addition, TBR branch swapping, and the MUL-Trees option in effect.For the ML and BI methods, the GTR + I + G model was selected according to the Akaike information criterion using MODELTEST version 3.6 [56].ML heuristic searches were performed with the resulting model settings, 100 replicates of random sequence addition, TBR branch swapping, and the MUL-Trees option.For the BI method, using the default prior settings, two parallel MCMC runs were carried out for 10 million generations, with sampling every 1000 generations for a total of 10,000 samples.The convergence of the two chains was assessed, and posterior probabilities (PP) were calculated from the trees sampled during the stationary phase.The robustness of nodes in the ML and MP trees was tested using bootstrap with 1000 replicates (bootstrap percentage, BP).

Morphological Study 2.2.1. Plant Morphology
In order to clarify the differences between I. scariosa and I. timofejewii, we selected 31 qualitative and quantitative macromorphological characters: (1) rhizome shape; (2) rhizome diameter; (3) rosette leaf shape; (4) rosette leaf texture; (5) rosette leaf apex shape; (6) rosette leaf surface; (7) rosette leaf length (measured from the base to the apex of the longest rosette leaf); (8) rosette leaf width (when dry, measured at the broadest part of the broadest rosette leaf); (9) stem height (measured from the base of the flowering stem to the base of the outer bract); (10)  We obtained the scores of the characters for I. scariosa and I. timofejewii from our own observations of herbarium specimens at ALTB, LE, MHA, MW, NENU, NS, NSK, PALE, RWBG, TK, UBA, UBU, and VBGI, including the original material for the names studied, and through the database of specimens [57] (see Table S2 for more details).We measured rhizome diameter, fruit length and diameter, and seed length and diameter in the dry state with a digital Vernier caliper Series 532 (Mitutoyo, Aurora, IL, USA).We also examined living plants of I. timofejewii in the wild, including at the type locality.The terminology used in the morphological description follows reference [58].

Pollen Morphology
For describing the pollen morphology, we used mature anthers from the herbarium specimens collected in Russia: (1) Astrakhan Oblast, Astrakhan, s.d., S. Korzhinski s.n.Anthers and pollen grains were mounted on aluminum stubs and sputter coated with gold in a vacuum chamber Q150T ES (Quorum Technologies Ltd., Lewes, UK).The morphological features of dry pollen grains were studied by scanning electron microscopy (SEM).The SEM micrographs were taken with a high-resolution field emission scanning electron microscope Merlin TM (Carl Zeiss, Oberkochen, Germany) at the Joint-Use Center "Biotechnology and Genetic Engineering", FSC EATB FEB RAS.The accelerating voltage was set at 5 kV; the emission current was set at 80 pA.We focused primarily on the exine ornamentation of pollen grains.The height of the raised part of the sculpture elements was measured.The pollen terminology is based on reference [59].

Genetic Divergence and Phylogenetic Relationships within Iris subg. Iris
In the study, we used six cpDNA region sequences from 38 accessions of 20 I. subg.Iris species: 7 accessions of I. glaucescens; 7 accessions of I. timofejewii; 3 accessions of I. aphylla; 2 accessions of each of I. iberica, I. longiscapa, and I. pumila; and 1 accession of each of I. acutiloba, I. alberti, I. bloudowii, I. goniocarpa, I. humilis, I. imbricata, I. korolkowii, I. potaninii, I. reichenbachii, I. scariosa, I. stolonifera, I. thoroldii, I. tigridia, and I. vorobievii.Our major goal was elucidation of relationships among I. glaucescens, I. scariosa, and I. timofejewii to estimate the degree of genetic similarity among these three species and their taxonomic status.
We identified a total of 11 haplotypes among the 15 accessions of I. glaucescens, I. scariosa, and I. timofejewii based on polymorphic sites found across 6938 aligned positions of a combined dataset of six cpDNA regions.Of these, five haplotypes were each found in I. timofejewii (H1-H5) and I. glaucescens (H7-H11).Three haplotypes (H5, H8, and H10) were found at several localities, sometimes geographically very distant from each other, while the others were unique, i.e., found at a single locality.The accessions of I. glaucescens, I. scariosa, and I. timofejewii did not share any haplotypes.The sequence divergence (D XY ) of cpDNA between plants of these three species was low (Table 2), varying from 0.00052 (between I. scariosa and I. timofejewii) to 0.00099 (between I. timofejewii and I. glaucescens).The relationships between the haplotypes identified in I. glaucescens, I. scariosa, and I. timofejewii and other representatives of I. subg.Iris, including those retrieved from the complete plastid genomes of nine species, are shown in Figure 3.
The haplotypes of all the I. subg.Iris species were interconnected through multiple mutational steps, forming a single network (Figure 3), which was clearly divided into four haploclades.The latter were separated from each other and from I. dichotoma haplotype by 65 and more mutational steps.The maximum number of base pair changes was between each haploclade and I. dichotoma haplotype (more than 80).
Haploclade I contained all species of I. sect.Iris.In this haploclade, the I. glaucescens, I. scariosa, and I. timofejewii haplotypes (H1-H11) constituted a separate group that descended from an unsampled or extinct haplotype.In this group, haplotypes were separated from the neighboring ones by several (2-5) mutational steps.The haplotype of I. alberti was the closest to this group and connected with the same unsampled haplotype via seven mutational steps.Between two and five mutational steps also separated haplotypes A1-A4 of I. aphylla, as well as haplotypes P1 and P2 of I. pumila.Multiple mutational steps (more than nine) separated the haplotypes of the studied I. sect.Iris species from each other, except four steps between the haplotypes of I. lutescens and I. pumila.The cpDNA sequence divergence between I. lutescens and I. pumila was estimated at 0.00079, whereas D XY between other pairs of I. sect.Iris species ranged from 0.00159 to 0.00389 (Table 2).These values corresponded to D XY between each species of I. sect.Iris and the group that included I. glaucescens, I. scariosa, and I. timofejewii (0.00161-0.00320).The haplotypes of all the I. subg.Iris species were interconnected through multiple mutational steps, forming a single network (Figure 3), which was clearly divided into four haploclades.The latter were separated from each other and from I. dichotoma haplotype by 65 and more mutational steps.The maximum number of base pair changes was between each haploclade and I. dichotoma haplotype (more than 80).
Haploclade I contained all species of I. sect.Iris.In this haploclade, the I. glaucescens, I. scariosa, and I. timofejewii haplotypes (H1-H11) constituted a separate group that descended from an unsampled or extinct haplotype.In this group, haplotypes were separated from the neighboring ones by several (2-5) mutational steps.The haplotype of I. alberti was the closest to this group and connected with the same unsampled haplotype via seven mutational steps.Between two and five mutational steps also separated haplotypes A1-A4 of I. aphylla, as well as haplotypes P1 and P2 of I. pumila.Multiple muta- For haplotype codes, see Table 1.
The phylogenetic reconstruction methods (BI, ML, and MP) all resulted in basically similar topologies with few differences in statistical support (Figure 4) that were generally consistent with the network topology (Figure 3).As expected, the accessions of all species formed independent clades according to their subgeneric affiliations.Iris dichotoma and I. domestica formed a sister clade (PP 1.0, BP 100 and 100%) to the monophyletic clade comprising all the I. subg.Iris species (PP 1.0, BP 92 and 81%).In the I. subg.Iris clade, four highly supported clusters were resolved (Figure 4) that corresponded to haploclades revealed by the MJ method (Figure 3).Clusters I, III, and IV included species of I. sect.Iris, I. sect.Psammiris (PP 1.0, BP 100 and 100%), and I. sect.Pseudoregelia (PP 1.0, BP 100 and 100%), respectively.Cluster I (PP 1.0, BP 100 and 99%) included all the studied I. sect.Iris species which split into two sister groups (Figure 4, see arrows).One of them with moderate support in the ML and MP methods (PP 1.0, BP 86 and 78%) included I. aphylla, I. germanica, and I. reichenbachii.The accessions of I. glaucescens, I. scariosa, and I. timofejewii nested in a second group which also contained the accessions of I. alberti, I. imbricata, I. lutescens, and I. pumila and was robust in the BI analysis and weakly supported in the MP and ML methods (PP 0.92, BP 56 and 57%).The latter two species formed a branch with moderate support (PP 1.0, BP 78 and 75%).Another one branch in this group, weakly supported in the ML and MP methods (PP 1.0, BP 54 and 64%), included several accessions of I. glaucescens (S9-S11 and S15) and I. timofejewii (S2 and S3) together with a single I. scariosa (S8) accession (Figure 4).
Cluster II with high support (PP 1.0, BP 100 and 100%) corresponding to haploclade II included all the studied representatives of the three sections of I. subg.Iris: I. sect.Hexapogon, I. sect.Oncocyclus, and I. sect.Regelia.In this cluster, two sister groups were distinguished: the first group, supported only in the BI and MP methods (PP 0.94 and BP 61%), included species of I. sect.Regelia; the second group combined species of I. sect.Oncocyclus and I. sect.Hexapogon (PP 1.0, BP 93 and 89%).In this group, species of I. sect.Oncocyclus formed a monophyletic subgroup (PP 0.93, BP 65 and 59%) which was sister to I. longiscapa of I. sect.Hexapogon.

Macromorphological Comparison
A morphological comparison of I. scariosa, including plants from the currently accepted distribution range of I. glaucescens, with I. timofejewii is listed in Table 4 (also see Table S2 and Figure 1).Both species are variable in rhizome diameter, rosette leaf length and width, flowering stem height, length of the cauline leaf, bract, pedicel, and perianth tube, flower diameter, fruit length and diameter, and also in seed length and diameter.In desert steppes in the Samur River valley (southern Republic of Dagestan) and in Kazakhstan, plants are found with a very dwarf habit (e.g., MW0816525 and MW0816526; see https: //plant.depo.msu.ru/module/itemsearchpublic#,accessed on 24 July 2024).Iris scariosa and I. timofejewii shared all the qualitative morphological characters (Figure 1).Both had a rhizome that was thick, tough, 0.6-3 cm in diameter, shortly creeping, brownish yellow, covered with short brownish fibers at top; adventitious roots were thickened, yellow-white, with upper and lower parts equal in thickness, and were up to 15 cm long or more.Rosette leaves were ensiform, usually falcate (Figure 1c), or straight under shaded conditions (Figure 1d), chartaceous, with an acute apex and a slightly wider middle, and with leaf sheaths enlarged at rosette base.Rosettes were surrounded by few old leaves, preserved in the form of thin fibers.The flowering stem bore a cauline leaf, 2-5 falcate basal leaves, two bracts, and one bracteole; the inflorescence was two-flowered.Bracts membranous (which refers to the specific epithet scariosa) were commonly lilac, broadly lanceolate, with a shortly acute apex.A cauline leaf was located slightly below the middle of the elongated flowering stem (e.g., LE01263915; see http://rr.herbariumle.ru/01263915,accessed on 24 July 2024) or usually at the base of the flowering stem (e.g., LE01263934; http://rr.herbariumle.ru/01263934,accessed on 24 July 2024).The rosette and flowering stem were covered with waxy coating, which was very glaucous (which refers to the specific epithet glaucescens), and, thus, grayish green in color.Flowers were variable and were sometimes extremely variable within the same locality, from reddish purple, purple, blue, and light blue, sometimes to almost white, and less often yellow in color (Figures 5 and 6), and were 3.5-5 cm in diameter, borne on short pedicels or sessile.Blade of falls obovate were ca.5.5 cm long and 2.2 cm wide, folding downwards, and gradually narrowed into a claw, having beard in the form of a central, longitudinal, linear band of hairs; standards were oblanceolate, as long as falls but slightly narrower than falls, and upright.Even beard hairs can be colored in different shades, from yellow to white and blue.Fruit was an oblong-ellipsoid capsule, light brown in color, and up to 8 cm long (e.g., LE01263920; see http://rr.herbariumle.ru/01263920,accessed on 24 July 2024), with a short beak up to 0.5 cm long; pericarp was glabrous and coriaceous, firm, with six distinct ribs, dehiscent very close to the apex (Figure 1e,f).Seeds were pyriform, reddish brown, ca. 8 mm long and 5.5 mm in diameter, with wrinkled surface and without aril (Figure 1g,h).

Pollen Morphology
The pollen morphological features of the specimens examined in this study by SEM are shown in Figure 7. Below is a general description.

Pollen Morphology
The pollen morphological features of the specimens examined in this study by SEM are shown in Figure 7. Below is a general description.The shape of the pollen grains was oblate spheroidal (Figure 7a,c,e,g).The polar axis ranged from 61 to 91 µm, slightly shorter than the equatorial diameter ranging from 65 to 98 µm.The pollen type was monosulcate (Figure 7a,c,e).The specimen was sulcus distal (anasulcate) and longer than polar axis of pollen grain because it extended over proximal face (Figure 7c), wide (30-70 µm), with sulcus membrane predominantly smooth.
The exine ornamentation of clavate-baculate-granulate type showed that the general surface ornamentation of exine granulate was composed of rounded elements (granula) less than 1 µm in diameter.Free-standing sexine elements were more or less The shape of the pollen grains was oblate spheroidal (Figure 7a,c,e,g).The polar axis ranged from 61 to 91 µm, slightly shorter than the equatorial diameter ranging from 65 to 98 µm.The pollen type was monosulcate (Figure 7a,c,e).The specimen was sulcus distal (anasulcate) and longer than polar axis of pollen grain because it extended over proximal face (Figure 7c), wide (30-70 µm), with sulcus membrane predominantly smooth.
The exine ornamentation of clavate-baculate-granulate type showed that the general surface ornamentation of exine granulate was composed of rounded elements (granula) less than 1 µm in diameter.Free-standing sexine elements were more or less regularly spread over the surface (Figure 7b,d,f) and were club-shaped, with a diameter smaller than the height and thicker at the apex than at the base (clavae), or rod-like, with rounded apices (bacula).The height of these elements in pollen grains of the specimens from Astrakhan, Republic of Dagestan, and Orenburg Oblast was 3.4-4.8,3.5-4.4,and 1.7-2.5 µm long, respectively.In the specimen from Orenburg Oblast, the surface of the pollen grains located in the anther at its base (Figure 7g,h) contained irregularly arranged, elongated, raised structures (muri) which were not anastomosed, and the pollen grains near the anther apex had a clavate surface (Figure 7e,f).

Taxonomy of Iris scariosa
The protologue of I. scariosa is dedicated to the Willdenow Herbarium at Berlin-Dahlem, B [1].The original material of the name is represented at B by a single specimen (B-W00959010), which is indicated as the lectotype in reference [72].This specimen is kept in a folder accompanied by a label on which Willdenow handwrote the diagnostic phrase name "Iris scariosa . .." followed by the synonym "Iris biflora Pall." with a note on geographical origin: "Habitat in Sibiria".It should be mentioned that, in some cases, Willdenow erroneously indicated the origin on his original labels [73].
The above data of phylogenetic analyses (Figures 3 and 4) and comparisons of nucleotide divergence levels (Table 2) confirm that I. scariosa from Astrakhan; the plants from Altai Krai, Russia, and Kazakhstan, here referred to as I. glaucescens; and the plants from the Republic of Dagestan, Russia, here referred to as I. timofejewii, are best recognized as a single species for which I. scariosa has priority.
In the present study, we did not find diagnostic features to distinguish I. timofejewii from I. scariosa.Both species demonstrated similar patterns of morphological variability and shared all the qualitative characters (Table 4).The comparative study of plants from the mountainous part of the Republic of Dagestan and I. scariosa showed that the foliage and the organs of the flowering stem were very variable, with their morphology depending, to a very large extent, on habitat conditions.The field observations in the Republic of Dagestan clearly showed that in the Samur River valley, on clayey dry soils, plants are dwarf, while in the central, mountainous Dagestan, on loose soils, plants have a habit typical of I. scariosa.We found that the shape of rosette leaves depends on light intensity within the same locality: the rosette leaves were falcate in sunlit areas (Figure 1c) and straight under shaded conditions (Figure 1d).It was also noted that, under cultivation, leaves of I. timofejewii are usually straight rather than sickle-shaped [15] (p.257).
Rodionenko [15] noted that I. scariosa differs from I. timofejewii and I. astachanica by the reticulate ornamentation of pollen grains, while in the latter two species, the exine ornamentation is papillate.Nonetheless, the results of the present study clearly show the clavate-baculate-granulate type of exine ornamentation in all pollen samples (Figure 7).Hence, from a morphological viewpoint, all the plants may, in fact, be regarded as a single variable species.We have confirmed that I. glaucescens is a synonym of I. scariosa and consider I. timofejewii to be a synonym of the latter.The same exine pattern is characteristic of I. adriatica Trinajstić ex Mitić, I. attica Boiss.& Heldr., and I. pumila [79][80][81].These species belong to the group of dwarf bearded irises treated as I. ser.Pumilae G.H.M.Lawr.[60].
Iris scariosa was occasionally in cultivation under the name I. eulefeldi Regel [11].The original material of I. eulefeldi was collected along the Talki River in Chinese Dzungaria (see below).It was regarded as a robust variety of I. scariosa [10].Grubov [5] noted that I. eulefeldi and typical plants of I. scariosa were found growing together in the Tien Shan Mountains (Yining County-level city, or Ghulja) and the Dzungarian Alatau mountain range.Iris eulefeldi has long been recognized as a taxonomic synonym of I. scariosa [4,5,11,12,[14][15][16]19,24,27,29,30].To the best of our knowledge, these two taxa are identical.
According to the diagnosis, I. curvifolia differs from I. scariosa by having yellow flowers and ovoid, shortly beaked fruit [68].However, in the diagnosis of I. curvifolia, the features of the flower and fruit completely match those of I. scariosa (Figures 1 and 5, Table 4).A comparison of the available herbarium specimens of I. curvifolia, including illustrations [68][69][70][71] and the relevant species descriptions available in the literature [3,62,68], to the specimens of I. scariosa from China (e.g., HNWP No. 19365, KUN0360536, NAS00555345, PE01013381, and PE01013382; see https://www.cvh.ac.cn/index.php,accessed on 24 July 2024) and from other parts of its distribution range (Table 4) has shown that their features of the rhizome, roots, rosette leaves, flowering stem, bracts, flowers, fruit, and seeds are identical.
The original material of I. curvifolia (NENU00014010-NENU00014012) is represented by three plants.Rhizome of I. curvifolia is thick, tough, 0.7-1.5 cm in diameter, shortly creeping, brownish yellow, covered at top with short brownish fibers; adventitious roots are thickened, yellow-white, with their upper and lower parts similar in thickness.Rosette leaves are ensiform, falcate or straight, with acute apex and slightly wider middle and with leaf sheaths at the rosette base enlarged, up to 20 cm long and 0.9-1.3cm wide.Rosette is surrounded by few old leaves.The flowering stem bears the cauline leaf and falcate basal leaves, 8-18 cm tall; two bracts and one bracteole, membranous, bracts broadly lanceolate, apex shortly acute, and outer bract 4-5 cm long; inflorescence two-flowered; perianth tube 2-2.2 cm long.According to references [3,62,68], leaves of I. curvifolia are glaucous green; flowers are 4.5-6 cm in diameter; blade of falls are obovate, ca.4.5 cm long and 1.5 cm wide; standards are oblanceolate, ca. 4 cm long and 1.3 cm wide; capsules are oblong-ellipsoid, with six distinct ribs, ca. 4 cm long and 2 cm wide, shortly beaked at the apex; and seeds are pyriform, reddish brown, ca.0.7 cm long.Since there are no consistent morphological differences between these species, we suggest that I. curvifolia should be considered a synonym of I. scariosa.

Taxonomic Treatment of Iris scariosa
There are some comments concerning the type citation of the names under study.Below are the details that should be clarified: (i) On the herbarium sheet at B (B-W00959010; see https://herbarium.bgbm.org/object/BW00959010, accessed on 24 July 2024), which is the current lectotype of I. scariosa [72], there are two notes handwritten by Von Schlechtendal: "Ir.scariosa 1" and "[collection history]: Pallas.[to] W.[illdenow Herbarium]".According to reference [73] (p.344), "if a folder contains more than one sheet, the individual sheets are also sequentially numbered".Thus, the number "1" means that the folder of I. scariosa contained more than one sheet, as in the cases of I. caricifolia Pall.ex Link [85] and I. setosa Pall.ex Link [86].It has been established that the label of HAL0109666 was erroneously replaced by that of the specimen HAL0109667 [85] (p.288), which is the original material of the name I. scariosa (see below).The specimen HAL0109666, erroneously reported as an isolectotype of I. scariosa [72], is actually an isolectotype of I. oxypetala Bunge [85].
(ii) Iris timofejewii was described from plants cultivated at the Tiflis Botanical Garden, Tbilisi, Georgia [43].These plants were raised from the rhizomes collected by Alexander Alfonsovich Grossheim in Andiyskiy Okrug (now Botlikhsky Raion, western Republic of Dagestan, Russia) in 1915.Fedtschenko [2] (p.549) noted that no original material for I. timofejewii was known.Our attempts to find the original material for this name in the framework of the present study have not been successful as well.Consequently, neotypification is required according to the Art.19.11 of the ICN.For this purpose, LE01268154 is here designated as a neotype since it is accompanied by a printed label with the note "G.Woronow.Notae criticae", on which Jurij Nikolaewitch Woronow, the author of the name, handwrote "Iris timofejewii m.March (19)24".It was collected in the vicinity of Khadzhalmakhi Village, Levashinsky Raion, Republic of Dagestan, Russia.
(iii) The name I. astrachanica was first validly published by Rodionenko in 1977 [15].In the protologue of I. astrachanica, a single gathering was designated as the type.At least five undated specimens at LE (LE01015406-LE01015410!), accompanied by labels with the note "Iris astrachanica Rodion.",handwritten by Rodionenko before 1977, belong to the original material of I. astrachanica and are syntypes (see Arts. 9.6 and 40, Note 1 of the ICN).These specimens were collected by Sergei Ivanovitsch Korshinsky, a Russian botanist, near the Volga River estuary in the vicinity of Astrakhan, Russia, in approximately 1880-1883.One of the syntypes (LE01015406) is designated here as the lectotype of I. astrachanica.It corresponds to the protologue of the name and contains a label on which Rodionenko handwrote "Iris astrachanica Rodion.27 February 1958", and at the bottom of the sheet, he handwrote in Russian "pyl'tsa s borodavchatoi ekzinoi" (translated = "pollen with a papillate exine").
In the extended circumscription presented here, I. scariosa includes five synonyms.The information on all the names, with full nomenclature citations and the main findings on the distribution and habitat of I. scariosa, is provided below.
Iris scariosa Willd.ex Link, Jahrb.Gewächsk.Krylov [13] noted that I. scariosa was distributed in the western and southern foothills of the Altai Mountains and was not distributed eastward.Meanwhile, we found a specimen at NSK (NSK0069117, sub I. glaucescens) collected in the north of Uymonskaya Steppe, Russia ("Altai, Ust-Koksinsky Raion, Terektinskiy mountain range, in the vicinity of Terekta Village, southern slope, stony steppe, 13 August 1984, M. Lomonosova s.n.[originally in Russian]"; see http://herb.csbg.nsc.ru:8081/,accessed on 24 July 2024).The record of I. scariosa from this locality was confirmed by the collector (M.Lomonosova, pers.comm.).This is probably the first record of I. scariosa from the Altai Republic, since neither it nor I. glaucescens are listed for this territory in the literature (e.g., [17,89]).Unfortunately, we do not know about any other collections of I. scariosa from this locality over the past 40 years.The finding of I. scariosa in the Altai Republic in May 2024 (P.Kosachev, pers.comm.) was not confirmed, and, therefore, needs further clarification.
According to references (sub I. glaucescens) [6,17,90], I. scariosa is found in Uvs Aimag, northwestern Mongolia.However, we did not encounter any herbarium specimens that would confirm the distribution of I. scariosa in Mongolia.
Iris scariosa can be found at elevations ranging from below sea level to 2700 m a.s.l.It occurs in stony, sandy, or gravelly habitats; on saline, clayey or limestone soils in dry steppes; grasslands on sunny hillsides, slopes, or terraces of low mountains; or beside ditches.The flowering period is from late April to mid-May, and the fruiting period is from July to August.Mature seeds have a sticky and sweet-tasting surface (A.Grebenjuk, pers.comm.)that, in our opinion, attracts ants and can be involved in the seed dispersal (i.e., myrmecochory).
Iris sect.Iris is monophyletic with the type species I. germanica nesting in its clade (Figure 4).In this section, two subclades are resolved (Figure 4, see arrows) that do not correspond to frequently described subgroups such as I. ser.Pumilae (plants dwarf) and I. ser.Elatae G.H.M.Lawr.(plants medium to tall), as was reported in reference [37].
In Iris, the epithet Hexapogon was first used by Bunge [92] (p.329) in the name of an unranked subdivision (Art.37.1 of the ICN) of the genus, comprising I. falcifolia Bunge and I. filifolia Bunge (nom illeg., Art.53.1 of the ICN), which are taxonomic synonyms of I. longiscapa [93].Bunge noted that these plants had beards on both the inner and outer perianth segments as follows: "laciniis perigonii omnibus barbatis".The taxon Hexapogon was assigned a sectional rank by Baker [94] as I. sect.Hexapogon, etc. (see below).Rodionenko [95] resurrected I. sect.Hexapogon that comprised bearded irises with arillate seeds in it.Subsequently, he combined all species of the genus Iris with seeds containing an aril into I. subg.Arillosae Rodion.[96], including the five sections of bearded irises according to references [16,36].However, I. subg.Arillosae cannot be considered monophyletic [37].According to the presented molecular data, I. sect.Hexapogon comprises species of three previously recognized sections, i.e., I. sect.Hexapogon, I. sect.Oncocyclus, and I. sect.Regelia, treated here at a serial rank.We believe that the main diagnostic feature of the I. sect.Hexapogon species is the presence of hairs on the adaxial side not only of the falls but also of the standards.In the species of I. ser.Hexapogon and I. ser.Regelia, the standards have a conspicuous, more or less linear beard of hairs down the claw, whereas in the I. ser.Oncocyclus species the standards have occasional hairs at the base of the claw.

List of Taxa
The species of I. subg.Iris are distributed in the north temperate zone of Eurasia.The subgenus comprises four sections as circumscribed below.The composition of I. sect.Hexapogon is restored in the present study.It combines three groups, the autonymic series and two series proposed here, I. ser.[60] (p.354)): Iris falcifolia Bunge (a taxonomic synonym of I. longiscapa [93]).
(1) Iris ser.Hexapogon It is considered unispecific, including only I. longiscapa that is distributed mainly in desert and semi-desert areas of Central Asia (in Uzbekistan, Kazakhstan, and Turkmenistan, Tajikistan, and Afghanistan) and also in Iran and southwestern Pakistan [93].Iris longiscapa shows a chromosome number of 2n = 18 [31,98], which is unique in I. subg.Iris.
( It comprises about six species occurring in Central Asia (southern Kazakhstan, Afghanistan, Tajikistan, Uzbekistan, and Turkmenistan).Two species, I. stolonifera and I. hoogiana, with 2n = 44, are amphidiploids, characterized by a wide-spreading rhizome bearing slender stolons, whereas the other species, with 2n = 22, are considered as diploids, characterized by a comparatively compact rhizome [99].Among the latter species, I. afghanica is very characteristic, presumably holding a separate position (Figures 3 and 4).
( This is the broadest and very variable group in which many species have been described on the basis of weak morphological differences [100][101][102].It comprises about 40 species, all with 2n = 20 [103], well adapted to arid conditions of the Middle East (Egypt, Israel, Palestine, Jordan, Lebanon, Syria, Turkey, Iraq, and Iran).They also grow in southern Turkmenistan, Transcaucasus, and in the Republic of Dagestan, Russia.
Currently, the section has one of the best elaborated systematics [39].It comprises only five species and is subdivided into an autonymic series (with I. bloudowii, I. humilis, and I. vorobievii) and two unispecific series: I. ser.Potaninia Doronkin with I. potaninii and I. ser.

Conclusions
To consider the taxonomy of I. scariosa more in detail, we compared morphological characteristics and conducted molecular phylogenetic analyzes using sequence data for six chloroplast DNA regions.These are the most comprehensive phylogenetic analyses to date for the species.Iris scariosa is distinguished by its high variability in the morphological characters, especially in the flower color.We could not find any discontinuities of variation independent of environmental influences or any geographical pattern for these variations.Our major results are as follows: (1) the molecular data confirm that I. glaucescens is a synonym of I. scariosa; (2) the molecular data and the thorough examination of living plants confirm that I. timofejewii, recognized on the basis of morphology and considered as endemic to the Republic of Dagestan, Russia, is a synonym of I. scariosa; (3) a critical evaluation of the original material and literature have shown that I. curvifolia and I. scariosa are the same taxon; and (4) the exine ornamentation in I. scariosa is of clavate-baculategranulate type.As a consequence, our findings have clarified the composition of I. subg.Iris in Russia, where it is represented by nine species.Currently, these include five species of I. sect.Psammiris [39], I. acutiloba, and also three species of I. sect.Iris.Of the latter, I. aphylla and I. pumila are distributed in the European part of Russia and the North Caucasus, while I. scariosa grows in the south of the European part of Russia, in the east of the North Caucasus, and in the south of the Western Siberia.
This study also offers a path forward to a revised infrageneric classification of the genus Iris based on molecular data.All species presented in this study are divided into three clades that correspond to three subgenera: I. subg.Iris, I. subg.Pardanthopsis, and I. subg.Limniris.The sister clade to I. subg.Iris is I. subg.Pardanthopsis with I. dichotoma and I. domestica, which should be treated as legitimate species of the genus Iris (also see [37,61,63,64]).As the first step, the classification of I. subg.Iris is revised here.The data that we present have several important implications for the taxonomy of I. subg.Iris: (1) four monophyletic clades correspond to the sections, i.e., I. sect.Iris, I. sect.Hexapogon, I. sect.Psammiris, and I. sect.Pseudoregelia, which are morphologically quite clearly distinguished; (2) I. sect.Hexapogon comprises three series, i.e., I. ser.Hexapogon, I. ser.Oncocyclus, and I. ser.Regelia, previously recognized as sections; and (3) our analyses confirm the split of monophyletic I. sect.Iris into two groups, but no pronounced morphological differences or geographic patterns have been found to explain this division.We believe that this classification provides a foundation for future endeavors.Nevertheless, more phylogenetic analyses within the genus are required.Also, further revision of the current subsectional system in I. sect.Iris and its taxonomic composition, as well as of I. ser.Oncocyclus, I. sect.Pseudoregelia, and I. ser.Regelia, needs substantial efforts.

Figure 3 .
Figure 3. Median-joining network based on cpDNA haplotypes of the Iris subg.Iris species and I. dichotoma as outgroup.Each circle indicates a haplotype, with the size of the circle proportional to the number of localities where this haplotype was found.The 11 haplotypes derived from 15 accessions of I. scariosa, I. glaucescens, and I. timofejewii are indicated by colored circles: red, I. scariosa; yellow, I. glaucescens; violet, I. timofejewii.Each line between two haplotypes indicates a mutational step, and dashed lines indicate alternative connections of haplotypes.Numerals near the lines connecting haplotypes indicate the number of mutational steps interconnecting two haplotypes (no numeral = one mutation).Small black circles indicate median vectors, inferred by Network version 4.6.Elliptic lines outline the haplotypes representing haploclades I-IV within I. subg.Iris.For haplotype codes, see Table 1.
Figure 3. Median-joining network based on cpDNA haplotypes of the Iris subg.Iris species and I. dichotoma as outgroup.Each circle indicates a haplotype, with the size of the circle proportional to the number of localities where this haplotype was found.The 11 haplotypes derived from 15 accessions of I. scariosa, I. glaucescens, and I. timofejewii are indicated by colored circles: red, I. scariosa; yellow, I. glaucescens; violet, I. timofejewii.Each line between two haplotypes indicates a mutational step, and dashed lines indicate alternative connections of haplotypes.Numerals near the lines connecting haplotypes indicate the number of mutational steps interconnecting two haplotypes (no numeral = one mutation).Small black circles indicate median vectors, inferred by Network version 4.6.Elliptic lines outline the haplotypes representing haploclades I-IV within I. subg.Iris.For haplotype codes, see Table 1.

Figure 3 .
Figure 3. Median-joining network based on cpDNA haplotypes of the Iris subg.Iris species and I. dichotoma as outgroup.Each circle indicates a haplotype, with the size of the circle proportional to the number of localities where this haplotype was found.The 11 haplotypes derived from 15 accessions of I. scariosa, I. glaucescens, and I. timofejewii are indicated by colored circles: red, I. scariosa; yellow, I. glaucescens; violet, I. timofejewii.Each line between two haplotypes indicates a mutational step, and dashed lines indicate alternative connections of haplotypes.Numerals near the lines connecting haplotypes indicate the number of mutational steps interconnecting two haplotypes (no numeral = one mutation).Small black circles indicate median vectors, inferred by Network version 4.6.Elliptic lines outline the haplotypes representing haploclades I-IV within I. subg.Iris.For haplotype codes, see Table1.

Figure 4 .
Figure 4. Bayesian majority rule consensus tree for species of the genus Iris inferred from combined trnH-psbA, rps4-trnS GGA , trnS-trnG, trnL-trnF, ndhF, and ycf1 chloroplast data.Asterisks (*) indicate species for which sequences of six cpDNA regions were accessed from GenBank (see TableS1).Numerals above branches are Bayesian posterior probabilities (PP > 0.9) and bootstrap values for the ML and MP methods (BP > 50%).Bold lines indicate branches to four sections of I. subg.Iris, and arrows indicate two subclades resolved in I. sect.Iris.The haplotype codes correspond to those listed in Table1.

Figure 4 .
Figure 4. Bayesian majority rule consensus tree for species of the genus Iris inferred from combined trnH-psbA, rps4-trnS GGA , trnS-trnG, trnL-trnF, ndhF, and ycf1 chloroplast data.Asterisks (*) indicate species for which sequences of six cpDNA regions were accessed from GenBank (see TableS1).Numerals above branches are Bayesian posterior probabilities (PP > 0.9) and bootstrap values for the ML and MP methods (BP > 50%).Bold lines indicate branches to four sections of I. subg.Iris, and arrows indicate two subclades resolved in I. sect.Iris.The haplotype codes correspond to those listed in Table1.

4. 1 . 2 .
Distribution and Ecology of Iris scariosaIn Russia, it is distributed in the eastern North Caucasus (western Republic of North Ossetia-Alania, northern Chechen Republic, and Republic of Dagestan), in the south of the European part (northeastern Stavropol Krai, southeastern Rostov Oblast, Republic of Kalmykia, Astrakhan Oblast, Volgograd Oblast, and southeastern Republic of Bashkortostan), and in the southern Western Siberia (Orenburg Oblast, in the south of the Chelyabinsk Oblast and Omsk Oblast, southwestern Novosibirsk Oblast, Altai Krai, and probably in the Altai Republic).Also, it occurs in Kazakhstan (Abai, Akmola, Aktobe, Almaty, East Kazakhstan, Jetisu, Karaganda, North Kazakhstan, Pavlodar, Ulytau,

4. 1 . 2 .
Distribution and Ecology of Iris scariosaIn Russia, it is distributed in the eastern North Caucasus (western Republic of North Ossetia-Alania, northern Chechen Republic, and Republic of Dagestan), in the south of the European part (northeastern Stavropol Krai, southeastern Rostov Oblast, Republic of Kalmykia, Astrakhan Oblast, Volgograd Oblast, and southeastern Republic of Bashkortostan), and in the southern Western Siberia (Orenburg Oblast, in the south of the Chelyabinsk Oblast and Omsk Oblast, southwestern Novosibirsk Oblast, Altai Krai, and probably in the Altai Republic).Also, it occurs in Kazakhstan (Abai, Akmola, Aktobe, Almaty, East Kazakhstan, Jetisu, Karaganda, North Kazakhstan, Pavlodar, Ulytau, Kostanay, and West Kazakhstan regions) and China (western and northern Xinjiang Uygur Autonomous Region).

Table 1 .
Sampled Iris taxa with voucher information and GenBank accession numbers.

Table 2 .
Nucleotide divergence between species within Iris sect.Iris: below the diagonal, average number of nucleotide substitutions per site (D XY ); above the diagonal, average number of nucleotide differences (in brackets, the number of fixed differences).

Table S1
). Numerals above branches are Bayesian posterior probabilities (PP > 0.9) and bootstrap values for the ML and MP methods (BP > 50%).Bold lines indicate branches to four sections of I. subg.Iris, and arrows indicate two subclades resolved in I. sect.Iris.The haplotype codes correspond to those listed in Table1.

Table S1
). Numerals above branches are Bayesian posterior probabilities (PP > 0.9) and bootstrap values for the ML and MP methods (BP > 50%).Bold lines indicate branches to four sections of I. subg.Iris, and arrows indicate two subclades resolved in I. sect.Iris.The haplotype codes correspond to those listed in Table1.

Table 3 .
Nucleotide divergence between sections within Iris subg.Iris: below the diagonal, average number of nucleotide substitutions per site (D XY ); above the diagonal, average number of nucleotide differences (in brackets, the number of fixed differences).
All measurements are in centimeters, except for seeds.Data are presented as range (minimum and maximum values).See supplementary raw data in TableS2for more details; for illustrations, seeFigures 1, 5 and 6.
Funding: This research received no external funding.