Allium pallasii and A. caricifolium—Surprisingly Diverse Old Steppe Species, Showing a Clear Geographical Barrier in the Area of Lake Zaysan

Polymorph Allium pallasii s.l. from monotypic A. sect. Pallasia was studied using a wide spectrum of methods and divided into two clearly morphologically, geographically, cytologically and genetically isolated species: A. pallasii s. str.—North-East Kazakhstan, Western Siberia, and the Altai Mountains; A. caricifolium—Kyrgyzstan, Northwest China, South-East Kazakhstan until Zaysan Lake in the east. Despite serious genetic differences, both species are sisters and are related to species of the A. sect. Codonoprasum (Subg. Allium). Allium caricifolium differs from A. pallasii s. str. by taller stems, dense inflorescence, and with filaments longer than perianth. The possible phylogenetic reasons for the separation of these species are discussed. A nomenclature analysis of synonyms was carried out.


Introduction
Allium L. (Amaryllidaceae J.St.-Hil.: Allioideae Herb.) is one of the largest monocot genera with more than 1000 species [1] naturally distributed throughout the northern hemisphere [2][3][4][5][6][7]. The main centres of biodiversity are located in arid and sub-arid regions of Southwestern and Central Asia, and in the Mediterranean region. The significantly smaller centre is in western North America [5,[7][8][9]). The genus is characterized by bulbs (often formed on rhizomes) enclosed in membranous, fibrous, or reticulate tunics, free or basally connate sepals, and usually a subgynobasic style [7]. The overwhelming morphological diversity in the genus is mirrored by a complicated taxonomic structure consisting of 15 subgenera and 72 sections of three evolutionary lineages [4,7].
All subsequent phylogenetic studies [7,[9][10][11][12][13][14][15] confirmed the division of Allium into three major evolutionary lineages with the monophyletic origin of all subgenera included in the first and second evolutionary lineages. The phylogenetic relationships in the youngest third lineage are less clear. According to the latest studies, many subgenera are not monophyletic in the third evolution line. This mainly affects the subgenera Cepa (Mill.) Radić,     Figure S2). All A. pallasii s. l. accessions are divided into two sister groups and stand surprisingly as a sister group to the A. sect. Codonoprasum of subgenus Allium. Both species (A. pallasii and A. caricifolium) are sister groups also in the plastid tree (rpl32-trnL) with 115 accessions from most sections of the third evolutionary line and five accessions from A. pallasii s.l. Both clades of A. pallasii and A. caricifolium together are entitled as a sister clade to A. sect. Codonoprasum (See Figure S3). The generalized ITS tree with sections and subgenera names is shown in Figure 5. Some subgenera in the third evolution line after classification [7] are not monophyletic; this applies to subgenera Cepa, Reticulatobulbosa, Polyprason, Rhizirideum, and possible Allium. These results agree with previously published phylogenetic analyses [5,9,[11][12][13]27,28]. The phylogenetic consequences for the non-monophyletic subgenera should be made in the future with detailed analysis, but here is the most important finding for us, that the A. sect. Pallasii is a sister group to A. sect. Codonoprasum with strong support. The matching of the two sister clades A. pallasii and A. caricifolium is only moderately supported: Bayesian posterior probabilities (PP) = 0.86 and bootstrap support (BS) = 70.    Table 1. Figure 3. PCA analysis of morphological characters, shown in Table 1 Furthermore, we made a phylogenetic screening with 51 accessions A. pallasii s.l., carried out from the entire distribution area (31 accessions of A. pallasii and 20 accessions of A. caricifolium) with three representatives of the A. sect. Codonoprasum (A. flavum L., A. paniculatum L. [29,30] and A. praescissum Rchb. [31]) as outgroup with nuclear (ITS) fragments and two plastids (trnL-rpl32 and trnQ-rps16). ITS sequences within A. pallasii and A. caricifolium are monomorphic, with rare single-nucleotide swaps. Especially the accessions of A. pallasii s. str. have identical sequences. Only in the mountainous morphotype (A. caricifolium), are the accessions from the Alai Mountains in Kyrgyzstan grouped into a clade with relatively good support. Both species are divided into two sister groups with very high support because the sequences are very different (See Figure S3a). In the BLAST analysis, the nrITS sequences from A. pallasii s. str. were only 84.04% similar to A. caricifolium sequences from Northwest China (as A. pallasii in NCBI GenBank: GQ181077 China; KF693249 China: Xinjiang, Urumchi; KF693250 China, Xinjiang, Zhaosu), which correlates well with the group mean distance between A. pallasii and A. caricifolium ITS sequences (P = 0.188).
We obtained similar results with plastid sequences, where the polymorphism within morphotypes is significantly higher than with nrITS sequences. See the plastid tree in Figure S3b. There are only two discrepancies regarding the position of accession Am579 and Am606. In the ITS tree accession Am579 stays within A. pallasii s. str. clade and in the plastid tree clearly below the A. caricifolium clade. The situation at accession Am606 is reversed ( Figure S3). This is a clear indication of the hybrid origin of these accessions. In addition, their location in the border regions between both species supports the hybridogenic origin ( Figure 1). Except for these two cases, the topology of the trees is very similar, so we aligned and analysed all the sequences together ( Figure 6). Both hybrid accessions are expected to stand apart in the tree, but both sister clades are clearly monophyletic with very strong support. There are a few small groupings in A. pallasii clade with weak support; only one subclade with four accessions (Am189, Am482, Am574, Am575) has strong support (PP = 0.97; BS = 95). All these accessions are from the easternmost distribution. In the BLAST analysis, the trnL-rpl32 spacer sequences from A. pallasii s. str. were only 94.19% similar to A. caricifolium sequence from Northwest China (as A. pallasii in NCBI GenBank: MN648632 complete chloroplast genome).     Within A. caricifolium clade are two subclades with good support: accessions Am Am1275, Am1280 and 1281 form one, and two accessions from Alai valley in Kyrgyz are the second well-supported clade. In the first subclade, two accessions are from valley in Kyrgyzstan, one from Alay valley (Am1275), and one (Am580) is from western part of the Zaysan lowland.

Cytology, Flowcytometry
From A. pallasii, we examined the karyotypes of four accessions (Am457, Am Am776, Am781). All four accessions have similar chromosome morphology. Two mi pairs of chromosomes have very small dot satellites in the shorter arm. There are metacentric chromosomes in the karyotype of A. pallasii. Therefore, we calculate Within A. caricifolium clade are two subclades with good support: accessions Am580, Am1275, Am1280 and 1281 form one, and two accessions from Alai valley in Kyrgyzstan are the second well-supported clade. In the first subclade, two accessions are from Chu valley in Kyrgyzstan, one from Alay valley (Am1275), and one (Am580) is from the western part of the Zaysan lowland.

Cytology, Flowcytometry
From A. pallasii, we examined the karyotypes of four accessions (Am457, Am588, Am776, Am781). All four accessions have similar chromosome morphology. Two middle pairs of chromosomes have very small dot satellites in the shorter arm. There are only metacentric chromosomes in the karyotype of A. pallasii. Therefore, we calculated a combined idiogram of 36 metaphases (Figure 7a, Table 2. For A. caricifolium, we could study the chromosome morphology of the accession Am1246. The sixth pair of chromosomes are metacentric, and two satellite chromosomes are submetacentric. Compared to A. pallasii, the satellites are massive in A. caricifolium, between one and two µm ( Figure 7b, Table 3). Overall, the chromosomes in A. caricifolium are also slightly larger. Total karyotype diploid length (TKL) in A. pallasii = 87.14 µm and in A. caricifolium = 103.19 µm. This correlates well with the estimated genome size by flow cytometry in both species: A. pallasii 2C = 14.03 pg (Am607); A. caricifolium 2C = 20.37 pg (Am708). See the histograms of relative DNA content in Figure S5.

OR PEER REVIEW 11 of 23
pallasii, the satellites are massive in A. caricifolium, between one and two µm ( Figure 7b, Table 3). Overall, the chromosomes in A. caricifolium are also slightly larger. Total karyotype diploid length (TKL) in A. pallasii = 87.14 µm and in A. caricifolium = 103.19 µm. This correlates well with the estimated genome size by flow cytometry in both species: A. pallasii 2C = 14.03 pg (Am607); A. caricifolium 2C = 20.37 pg (Am708). See the histograms of relative DNA content in Figure S5.

Nomenclatural Remarks
Murray [17] described A. pallasii on the plants grown in the botanical garden of the University of Göttingen from seeds, sent by P.S. Pallas, without geographical origin. According to the description and analysis of the picture [17] (Table 1) and type material, the name A. pallasii belongs to the plain steppe morphotypes of plants. The species named A. tenue G.Don [34] is also based on the samples from the Herbarium of Pallas. The description staminibus perianthio aequalibus" clearly refers to it as a synonym of A. pallasii s. str. The species named A. lepidum for the plain steppe morphotypes was donated by Ledebour [35] that included in the protologue a short description and an illustration (Table CCCLV)  This name is a nomen nudum cited by Ledebour [36] as "A. nitidulum Fisch. in herb. reg. berol." as synonym of A. pallasii. That means Ledebour, in both cases (A. lepidum and A. nitidulum), did not recognize the species status for Altai plants, but ultimately included them in synonyms to A. pallasii. Only Regel [37] first validated the name "A. nitidulum" as a variety of A. pallasii. The name of A. saxatile Hohen. ex Boiss. (nom.illeg.) was also incorrectly cited as synonym to A. pallasii [26]. It is a synonym of A. kunthianum Vved. [21].
All other names regarded as synonyms of A. pallasii s. l. (i.e., A. caricifolium, A. alberti, and A. semiretschenskianum) belong to the southern mountain morphotype with priority name A. caricifolium. Allium caricifolium Kar. & Kir. is described on the plants collected in the Mountains near the Ajagus settlement. The morphological character in the description " . . . staminibus simplicibus, basi subuîatis, perigonium subdupio excedentibus" unequivocally refers to the mountainous morphotype from south-east Kazakhstan and Kyrgyzstan [38], and is typified by [39] (lectotype MW0591659). Regel had correctly identified the differences between A. pallasii plants and the plants from southern Kazakhstan and described the plants from the Almaty region (formerly Vernoe) as A. semiretschenskianum [40]. The name A. caricifolium Regel had been unfortunately placed as a synonym for A. pallasii [41]. The situation with the name A. alberti Regel [41] is a bit complicated. This species was described from plants grown in the garden from bulbs collected by Albert Regel in the Chinese part of the Ili River in 1876 [42]. The lectotype of A. alberti (LE01010227, designated by [43]) shows a bulbless plant, and the morphological characters show extreme similarity with A. caricifolium and A. semiretschenskianum (Lectotype of A. semiretschenskoanum LE000518202 designated here, LE00052546). But Regel's original description of A. alberti is slightly confusing. He gives a detailed description of the slender reticulate-fibrous outer tunics of the bulbs "Bulbi ovati tunicis exterioribus tenuibus totis reticulate-fibrosis, . . . " and gives A. moschatum L. and A. sindjarense Boiss. & Hausskn. ex Regel (A. sect. Scorodon) as related species. All other characteristics in the description match A. caricifolium very well. We can only guess whether, or not, this is a mix-up with another bulb that his son Albert Regel [42] collected during his trip to China. It was Vvedensky [21] who put both species A. semiretschenskianum and A. alberti as synonyms to A. pallasii, and we put these two names as synonyms to A. caricifolium.
Section Description-Bulb ovoid, 12 mm diam., with outer gray, almost leathery shells. Shells with clear parallel veins. Stems 120-200 (290) mm high, covered by leaf sheaths up to 1/3-1/2 of its length. Leaves 2-3, filiform, semi-cylindrical, smooth, shorter than stem. Spathe 2 (3) times shorter than umbel, shortly pointed. Inflorescence is hemispherical or more often spherical, many-flowered, loose. Pedicels are almost equal between themselves, 2-3 (4) times longer than perianth. The tepals are pink, with a purple vein, shiny, 3 mm long, equal in length, lanceolate, and acute. The filaments of the stamens do not exceed the length of the perianth, subulate, and are slightly widened internally at the base. The style of the pistil is equal to or slightly longer than the perianth. Description-Bulb ovoid, 10-20 mm thick, outer shells gray, papery, without veins. Stem 20-65 cm high, 1/3 or almost up to 1/2 covered with smooth leaf sheaths. Leaves 3-4, filiform or narrowly linear, 1.5 (2.5) mm wide, shorter than the stem. Spathe 2-3 times shorter than an inflorescence, shortly pointed. Inflorescence spherical, many-flowered, dense. Pedicels are almost equal, 2-3 times longer than perianth. Tepals are pink with a purple vein, shiny, 3-4 mm long, equal, lanceolate or oblong-lanceolate, and acuminate. Filaments of stamens up to 1.5 times as long as tepals, subulate from a triangular base, inner base wider than the outer ones. The style of the pistil is slightly longer than the perianth.

Distribution-To the west of the Zaisas basin, Tarbagatai, Dzungarian Alatau and Central and Eastern Tian Shan
Habitat-On fine earth, gravelly and rocky slopes, outcrops of variegated rocks in the mountainous and subalpine belt

Discussion
All of our results (morphological, geographical, cytological, and molecular) quite clearly confirm the presence of two very well separated species in the formerly monotypic A. sect. Pallasia: A. pallasii s. str., and A. caricifolium. Li et al. [9] erroneously introduced several other Allium species into the section, mostly belonging to A. sect. Caerulea [46]. Complete chloroplast genome analysis of seven Chinese species (A. delicatulum, A. schoenoprasoides, A. songpanicum, A. tanguticum, A. caeruleum and A. teretifolium, including A. pallasii (A. caricifolium) from northwest China [27]), supports the isolated position of section Pallasia. The plastid genome of Chinese A. pallasii (MN648632) and nrITS sequences (GQ181077, KF693249, KF693250) belong to the A. caricifolium. It is possible that A. pallasii s. str. also occurs in the border region east of the Black Irtysh River (see Figure 1). So far, we have seen no evidence of this.
We confirmed 2 n = 16 for both morphotypes as expected from earlier studies [23] for both species and for A. caricifolium [23,24]. Vakhtina & Kudryashova [23] studied the morphology of the chromosomes of both morphotypes (A. pallasii from North East Kazakhstan and A. caricifolium from Transili Alatau) and found that both karyotypes differ in the position and the size of the satellites in the satellite chromosomes. Our data confirm these differences. Differences in plant morphology were also well recognized, but unfortunately, no consequent conclusions were made [23]. Simply A. pallasii was declared as very polymorphic.
It is also very surprising that the sequences of both species are so different (only 84% similarities in ITS sequences) and still grouped as a sister subclade. The closest relationship to A. sect. Codonoprasum cannot be explained morphologically either. Morphologically A. pallasii and A. caricifolium are more like species from the A. sect. Caerulea (e.g., with A. delicatulum, A. caesium, and others), which explains the inclusion of some species by Li et al. [9] in A. sect. Pallasia. In the nrITS and plastid trees, the species from A. sect. Caerulea are relatively distant from A. sect. Pallasia (Figure 1, Figures S1 and S2). When comparing the genetic differences between A. pallasii and A. caricifolium with other Allium species where times of evolutionary splits were estimated [16,28], we hypothesize an Oligocene split between A. sect. Codonoprasum and A. sect. Pallasii, and between A. pallasii and A. caricifolium Myocene split. These splits can be explained by the vegetation/landscape history of the Zaysan Depression.
The Oligocene in extratropical Eurasia is marked by the expansion of the Boreal vegetation zone (warm and humid) and the formation of temperate deciduous mesophyllous coniferous-broadleaved forests (Turgai Flora) [47,48]. In East Kazakhstan, the Turgai Flora became dominant during the Oligocene and the first half of the middle Miocene [49,50]. During the Miocene, large depressions in the hilly zone of the present-day Altai and northern Tien Shan were formed, and an inland lake has been proved for the Zaysan Depression [49]. It is suggested that a paleolake existed here since the Cretaceous period and that the Zaysan Basin was never dried [51].
Present-day Altai and northern Tien Shan mountains are believed to be of relatively recent origin (Neogene) and started to develop from the Miocene onwards as a direct result of the far-field effects of the Himalayan collision [52]. With the rising mountains, the relief energy increased and had consequences for the drainage pattern. It is hypothesized that the Altai-draining rivers flew southwards into the Zaysan and adjacent Junggar Basin, and the Tien Shan-draining rivers northwards also into the Junggar Basin [53] filling the Zaysan paleolake and creating paleolakes in the Junggar Basin. The filling of the paleolakes culminated in a united Zaysan-Junggar Basin Paleolake, which in the Late Pliocene-Pleistocene cut through the northern end of the Zaysan Basin triggering the birth and the formation of the course of the Irtysh River [53].
This scenario has consequences for the vegetation history in the Zaysan Depression. Forest vegetation (Turgai Flora) and paleolakes prevented the establishment of modern steppes for a long time, and it would appear that the steppe occurred only recently. Unfortunately, there are no Pliocene and younger paleo records from the Zaysan Basin itself but several studies from neighboring regions such as the area near Semei on the Irtysh River and the Kulunda Steppe point to a late Pleistocene/early Holocene steppe vegetation [54,55].
Based on the climate/landscape history outlined above, we suggest the following scenario of the evolutionary history of our vicarious species Allium pallasii and A. caricifolium: The original distribution area of the ancestral species was separated with the emergence of the Altai orogeny into two disjunct areas, leading to allopatric speciation. Allium pallasii s.str. survived in the Altai mountains (Kurai Steppe) and A. caricifolium in the Tian Shan and Tarbagatai mountains. With floods after the breaching of the dams of Chuya and Kurai lakes in Altai after the Ice Age [56,57], the seeds of A. pallasii were spread to the Kulunda Steppe and from there dispersed very quickly in the steppe of northern Kazakhstan. This could explain why the ITS sequences of all accessions of A. pallasii are identical. Allium caricifolium may have persisted in several places in the Tian Shan Mountains and spread north and east after the Ice Age, where it met with A. pallasii at Lake Zaysan. Similar splits between northern Kazakhstan, western Siberia, including the right bank of the Irtysh River up to the Altai mountains in one site, and mountainous regions in south-eastern Kazakhstan, west of Zaysan Lake in the second, have recently been discovered and molecularly confirmed in other taxa of the genus Allium: sect. Oreiprason [58]; Allium tulipifolium Ledeb. and A. robustum Kar. et Kir. [28]; A. obliquum L. [59], and also in other plant groups: genera Krascheninnikovia (family Amaranthaceae) [60] and Goniolimon (family Plantaginaceae) [61]. All this confirms the complex phylogenetic history of the steppe flora [62,63].

Morphological and Distribution Analyses
We compiled distribution maps from literature and online databases and analysed herbarium collections, including field collections. A total of 20 individuals of A. caricifolium from 4 Herbarium sheets were analyzed for the morphological analysis [Am705, Am715, Am1192 (see the origin in Appendix A), Am1281 OSBU-24372 (47 • (Table 1) using the Pearson correlation coefficient. In addition, a Kolmogorov-Smirnov test for normal distribution was carried out beforehand.

Taxon Sampling
Bulbs and leaf samples of more than 50 accessions of A. pallasii s. l. for DNA isolation were collected in the course of several collecting trips in Russia (Altai), Mongolia, and Kazakhstan from 2010 and growing in the Botanical Gardens in Osnabrück (Germany) and Barnaul (Russia). Some accessions of DNA were isolated from Herbarium sheets.
Newly sequenced accessions are marked with Am number in the trees, and their origin is shown in Appendix A. To determine the position of the A. pallasii in the genus, we took the available nuclear ITS sequences and rpl32-trnL (UAG) plastid fragment of accessions of the species with representatives from all sections of the third evolution line while some accessions from the first and second evolution lines were selected as the outgroup [7]. Sequences from NCBI GenBank (https://www.ncbi.nlm.nih.gov/nucleotide/ accessed on 3 December 2021) are marked with GenBank accession numbers on the trees.

DNA Extraction, Amplification and Sequencing
Total genomic DNA was isolated from leaves in silica gel using the InnuPREPP Plant DNA Kit (Analytic Jena AG) according to the manufacturer's instructions and used directly in PCR amplification. The complete nuclear ribosomal ITS region (ITS1, 5.8S and ITS2) was amplified using the primers ITS-A [67] and ITS-4 [68]. The PCR conditions for ITS followed ref. [7]. PCR conditions and primers for the chloroplast regions trnL-rpl32 and trnQ-rps16 were described in [69]. PCR products were sent to Microsynth SeqLab (Balgach, Switzerland for sequencing. The sequences from all the individuals were manually edited in Chromas Lite 2.1 (Technelysium Pty Ltd.South Brisbane, Australia) and aligned with ClustalX [70], the alignment was manually corrected using MEGA 7 [71].

Phylogenetic Analyses
Both data sets (nrITS and the cpDNA trnL-rpl32 markers) for identifying the position of A. sect Pallasia in the third evolution line and to find the closest relatives of A. pallasii were analysed separately through Fitch parsimony with the heuristic search option in PAUP version 4.0 b10 [72]) with MULTREES, TBR branch swapping and 100 replicates of random addition sequence. Gaps were treated as missing data. The consistency index (CI) [73] was calculated to estimate the amount of homoplasy in the character set. The most parsimonious trees returned by the analysis were summarized in one consensus tree using the strict consensus method. Bootstrap analyses (BS) using 1000 pseudoreplicates were performed to assess the support of the clades [74]. Bayesian phylogenetic analyses were also performed using MrBayes 3.1.23 [75]. The sequence evolution model was chosen following the Akaike Information Criterion (AIC) obtained from jModelTest2 [76]. Two independent analyses with four Markov chains were run for 10 million generations, sampling trees every 100 generations. The first 25% of trees were discarded as burn-in. The remaining 150,000 trees were combined into a single data set, and a majority-rule consensus tree was obtained along with posterior probabilities (PP). To determine molecular variability throughout the range, more than 50 accessions of A. pallasii s.l. and three species from the A. sect. Codonoprasum as outgroup, nrITS, and two noncoding regions plastid DNA (trnL-rpl32, trnQ-rps16) were sequenced and analysed as above. The group mean distance (P) was estimated with MEGA7.

Cytology, Flowcytometry
Bulbs were planted in pots, and growing roots were used for the karyotype analysis. Root tips were excised from the bulbs and kept overnight in distilled water on ice. They were then transferred to room temperature for 20 min and pre-treated for 3 h at room temperature in an aqueous solution of 0.1% colchicine. Roots were then fixed in a freshly prepared mixture of 96% ethanol and glacial acetic acid (3:1 v/v). Root tips were stained using hematoxylin according to the protocol reported by Smirnov [77]. Well-spread metaphase plates were electronically documented (digitally photographed), and finally, the chromosomes of the best plates were measured and pairwise arranged using the KaryoType software [78]. For A. caricifolium, 5 metaphase plates from one individual were evaluated (Am1246, Appendix A), for A. pallasii, 4 individuals were used (Am776, Am781, Am588, Am457, Appendix A), which provided 3-12 usable metaphase plates. The measurements from all metaphase plates were combined here, a total of 36 metaphase plates. Because the idiograms automatically assembled by the software were not satisfactory, we manually ordered the chromosome pairs according to their length and shape. The idiograms were designed using the bar graph function implemented in MS Excel ® . The terminology of [32,33] was applied.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/plants11111465/s1, Figure S1. Phylogenetic tree of third evolutionary lineage of the genus Allium, based on ITS sequences from NCBI GenBank. Figure S2. Phylogenetic tree of third evolutionary lineage of the genus Allium, based on CP DNA sequences (trnL-rpl32) from NCBI GenBank. Figure S3. A-Phylogenetic tree of A. sect. Pallasia accessions, based on ITS sequences; B-Phylogenetic tree of A. sect. Pallasia accessions, based on two combined fragments of plastid DNA (trnL-rpl32, trnQ-rps16). Figure S4. Histograms of relative DNA content were obtained after analysis of nuclei isolated from young leaf tissues of A. pallasii, accession Am607 (A) and A. caricifolium, accession Am708 (B). Acknowledgments: We thank the curators and staff members of the following herbaria: AA, ALTB, BRNO, HAL, GAT, FRU, LE, M, MHA, TK, MW, NS, NSK, OSBU, XJA, and W for their valuable help. Special thanks go to Alexander Naumenko (Nova Zahrada s.r.o., Czech Republic) for bulbs of A. caricifolium from Kyrgyzstan (Am1246). We would like to thank the editor and three anonymous reviewers whose input improved the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest. Appendix A.