Comparative analysis of two Korean irises (Iris ruthenica and I. uniflora, Iridaceae) based on plastome sequencing and micromorphology

Iris ruthenica Ker Gawl. and I. uniflora Pall. ex Link, which are rare and endangered species in Korea, possess considerable horticultural and medicinal value among Korean irises. However, discrimination of the species is hindered by extensive morphological similarity. Thus, the aim of the present study was to identify discriminating features by comparing the species’ complete plastid genome (i.e., plastome) sequences and micromorphological features, including leaf margins, stomatal complex distribution (hypostomatic vs. amphistomatic leaves), anther stomata density, and tepal epidermal cell patterns. Plastome comparison revealed slightly divergent regions within intergenic spacer regions, and the most variable sequences, which were distributed in non-coding regions, could be used as molecular markers for the discrimination of I. ruthenica and I. uniflora. Phylogenetic analysis of the Iris species revealed that I. ruthenica and I. uniflora formed a well-supported clade. The comparison of plastomes and micromorphological features performed in this study provides useful information for elucidating taxonomic, phylogenetic, and evolutionary relationships in Iridaceae. Further studies, including those based on molecular cytogenetic approaches using species specific markers, will offer insights into species delimitation of the two closely related Iris species.


Results
Leaf, flower, pollen, and orbicule morphological characters. The two species possessed similar floral traits (e.g., tepal color, shape, and size; Fig. 1 and Supplementary Fig. S1) and similar leaf epidermis, epidermal cell, and anticlinal wall shapes (Fig. 2). The leaf epidermis of both species consisted of long tabular cells, typically with elongated pavement cells (Fig. 2c,i), with undulate anticlinal walls (Fig. 2e,k) that were covered by either prominent (I. ruthenica; Fig. 2f) or weak (I. uniflora; Fig. 2l) epicuticular wax. However, the leaf margin of I. uniflora was clearly entire, whereas that of I. ruthenica was spiny with sharp stiff points (Fig. 2), and the leaves of I. ruthenica were hypostomatic (stomata are absent or extremely rare on the adaxial leaf surface while they are present on the abaxial leaf surface; Fig. 2c,d), whereas those of I. uniflora were amphistomatic (stomata are present on both adaxial and abaxial leaf epidermis; Fig. 2i-j). In addition, the width of guard cells ranged from 24.63-28.19 μm on the abaxial surfaces of I. ruthenica leaves and from 26.53-28.14 μm and 27.02-27.77 μm on the adaxial and abaxial surfaces of I. uniflora leaves, respectively (Tables 1, 2). SEM analysis revealed that stomata of both species were clearly sunken (Fig. 2f,l), and that they were all anomocytic lacking subsidiary cells, surrounded by four epidermal cells (Fig. 2d,i,j). Tepal epidermal cells had unicellular covered by a striated cuticle and slightly sunk stomata ( Supplementary Fig. S1). Furthermore, the outer anther epidermal cells were polygonal in shape and covered by a striated cuticle in both species, I. ruthenica and I. uniflora (Fig. 3g,h,q,r). Both species possessed anthers with anomocytic stomata in the middle section, despite differences in stomatal density (Fig. 3h,r). The guard cell surfaces of I. uniflora were weakly wrinkled (Fig. 3i), whereas those of I. uniflora were smooth (Fig. 3s).
The pollen grains of both species were monads of 47.54-56.31 μm in polar length and 39.18-53.55 μm in equatorial diameter (Table 2). While both species yielded fertile pollen grains (Fig. 3a,k), sterile pollen grains were only found extremely rarely or absent in I. ruthenica (Fig. 3a-c), and, in Iris uniflora, sterile pollen grains were frequently encountered ( Fig. 3j-l), which differed from the fertile pollen grains of both species in regard to equatorial width (Table 2), shape (subprolate to prolate, P/E = 1.19-1.49 vs. oblate-spheroidal to prolatespheroidal, P/E = 0.94-1.12; Table 2), and exine ornamentation (irregularly microreticulate exine ornamentation Figure 2. Stereo, light, and scanning micrographs of leaves of Iris ruthenica (a-f) and I. uniflora (g-l). Serrate (a,b) and entire (g,h) leaf margin. Hypostomatic (stomata is absent on adaxial surface but present on the abaxial surface of leaf) (c,d) and amphistomatic (stomata present on both adaxial and abaxial leaf epidermis) leaves (i,j) with anomocytic stomata. (e,f, k,l) Detailed leaf epidermal cell shape and sunken stomata on the leaf surface.    www.nature.com/scientificreports/ trnK-rps16 regions have shown consecutive nucleotide variations in the LSC region. The analysis of IR boundaries revealed that rpl22 was located in the LSC region of both species. The ycf1 and ndhA genes were located in the junction of SSC/IRb region. The rps19 gene, which was duplicated in the IR regions, was generally wellconserved. The ycf1 gene was located at the IRa/SSC and SSC/IRb junctions ( Supplementary Fig. S4).
Phylogenic relationships among Iris species. The ML and BI topologies were highly congruent for the whole plastome and CDS datasets, and all but one lineage was strongly supported (ML > 95%, BI = 1.0). More specifically, the topologies were clearly divided into three major clades, which corresponded to the subgenera Limniris, Pardanthopsis, and Iris (Fig. 7). Most of the Iris species included in the present study were assigned to section Limniris, which is consistent with the Angiosperm Phylogeny Group (APG) IV classification system, except for I. tectorum 53 . Furthermore, a clade containing I. domestica and I. gatesii was clustered as a sister group to I. tectorum; I. ruthenica and I. uniflora formed a monophyletic clade (Fig. 7).

Discussion
Micromorphology. Iris ruthenica and I. uniflora are sister taxa within ser. Ruthenicae and, as such, share a variety of morphological characters, including creeping rhizomes, leaf shape and size, flower color, and fruit shape 37 . Indeed, due to their morphological similarity, the taxonomic status of the species has been controversial. For example, Zhao et al. 37 suggested that I. uniflora be considered a subgroup of I. ruthenica, and Zheng et al. 54  www.nature.com/scientificreports/ argued that morphological differences between the species are the result of environmental conditions (i.e., different habitats). However, in the present study, I. ruthenica and I. uniflora could be differentiated using a variety of micromorphological features, including stomatal complex distribution (hypostomatic leaves vs. amphistomatic leaves), slight differences in the protruded conical cells of leaf surfaces, pollen grain fertility (all fertile vs. both fertile and sterile), and anther stomata density. The micromorphological analysis performed in the present study revealed that I. ruthenica and I. uniflora possess leaves with an irregular shape and sinuate anticlinal striation, as previously reported 14,20,55 . Because both leaf surface micromorphology and stomata have shown considerable variation in papillae, they are not taxonomically significant, as suggested by Wu and Cutler 14 and other monocot species 56,57 . The papillae that cover stomata likely contribute to defense against unfavorable environmental conditions 58 or attack by pathogens 59 . Papillae can also play an important role in light reflection, by preventing overheating 60 , and, as such, may be more correlated with environmental conditions than taxonomic status. Regardless of stomatal distribution, both species possessed anomocytic stomata (Table 2), which is likely the ancestral (plesiomorphic) condition among monocots 61 . The guard cell and genome sizes of the two Iris species included in the present study were not significantly distinct ( Table 2; Choi et al. 46 ), although it is well known that stomata size is positively correlated with genome size and that ploidy level changes in plants [62][63][64] .
With the exception of clear differences in pollen viability, the morphology (e.g., shape, size, exine pattern, and orbicule presence) of pollen in I. ruthenica and I. uniflora were not significantly different (Fig. 3). As I. uniflora was only found under low-temperature stress conditions, whereas I. ruthenica was widely distributed 46-48 , the   65,66 . Orbicules, also known as Ubisch bodies, are small sporopollenin particles that can be produced on the inner locule walls of anthers 67 and that are, here, reported in I. ruthenica and I. uniflora for the first time. Even though orbicule morphology has been widely used to elucidate systematic and evolutionary relationships 12,68,69 , the occurrence of orbicules has, until now, only been reported in I. pallida Lam. 70 . Thus, further investigation of orbicule occurrence in the Iridaceae might yield a significant phylogenetic trait, as reported in other plant groups [67][68][69][70][71][72] . However, as with other morphological features of Iris taxa, such as tepal structure 17  The mVISTA results indicated that the Iris plastomes contained little variation and that genic regions were more conserved than IGS regions, which is consistent with angiosperm plastomes in general [77][78][79] . More specifically, the psbA, trnK-matK, matK-trnK, and trnK-rps16 regions were hotspots for genetic variation (Fig. 5), which indicated underlying evolution [80][81][82][83] and value as molecular markers 84,85 . In terms of nucleotide diversity (Pi), most of the divergent regions were non-coding, which is consistent with previous reports [86][87][88] . Other plastomes were highly variable for the non-coding regions at psbA, trnK-matK, matK-trnK, and trnK-rps16 in the present study (Fig. 6). mVISTA and Pi analysis showed difference between I. ruthenica and I. uniflora at the plastome level. Furthermore, these regions will play an important role in the discrimination of I. ruthenica and I. uniflora, as well as other species within the Iridaceae. IR contraction and expansion causes variation in the size of angiosperm plastomes 89 . Previous studies have reported extremely short IRs or the loss of IR regions and genes 90,91 . Compared to I. ruthenica, I. uniflora had a highly conserved IR length and gene positions. However, the rps19 genes of I. ruthenica and I. uniflora were located in the IRa region, and ycf1 was located in the IRa/SSC, region which overlapped with ndhF. Thus, the Iris plastome possessed an extended IR, as reported previously 32,33 . Phylogenetic relationships in the Iridaceae. Plastid genome (i.e., plastome) sequences are valuable genomic resources for estimating phylogenetic relationships, particularly among closely related species and unresolved taxa 4,27,31 . The systematics of Korean irises have been widely discussed, and several molecular studies based on single molecular markers (e.g., psbA-trnH, trnL-F) and plastome structure have been performed in the past 33,92 . The topologies of phylogenetic trees based on whole plastome sequences in the present study are similar to those reported previously (Fig. 7, Supplementary Figs. S5-S7) 33,41,45,52,92 . In agreement with earlier results, I. ruthenica and I. uniflora were nested deeply within section Limniris ser. Ruthenicae. However, in the present study, the monophyly of section Limniris was compromised by the inclusion of I. tectorum (ser. Lophris), which was deeply nested with subgenera Pardanthopsis and Iris in both the ML and BI trees (Fig. 7). Due to insufficient taxonomic sampling, the results of the present study are not suitable for the discussion of inter-subgeneric relationships within Iris s.l. 44 . Nevertheless, the present study provides important genetic resources for further www.nature.com/scientificreports/ studies within the genus, as well as micromorphological comparisons between two closely related species in a taxonomic context. The results of the present study confirm that Iris section Limniris ser. Ruthenicae (formerly section Ioniris sensu Zhao et al. 37 ) contains the two species, I. ruthenica and I. uniflora. Morphologically, I. uniflora is distinguished from I. ruthenica because of the presence of narrow leaves 54 , but the flowers of the species are difficult to distinguish. In Korea, I. ruthenica is widespread, while I. uniflora is restricted to alpine areas [46][47][48] . Based on the comparison of morphological and chloroplast genome data of the two species, it is doubtful whether I. ruthenica and I. uniflora can be recognized as independent species-level taxa, as suggested by Zheng et al. 54 . Thus, further studies on molecular and morphological analyses of the two species at population level are required to clarify the taxonomic status of the two taxa.

Conclusions
The present study provides detailed insights into the leaf and flower micromorphologies and plastome structures of the two closely related species I. ruthenica and I. uniflora. Micromorphological features, including leaf margins, stomatal complex distribution (hypostomatic vs. amphistomatic leaves), anther stomata density, and floral epidermis cell patterns, are somehow useful for distinguishing the taxa, despite that the drastic influences of environmental variation, especially climate factors (e.g., temperature and light intensity), may also contribute to the morphological variations. The plastome sequences of the two related species possessed similar genome lengths, gene numbers, and gene orientations. Most of the variable sequences, which were found in non-coding regions, could be used as molecular markers for the differentiation of I. ruthenica and I. uniflora, as well as other Iris taxa. Given the economic and ecological importance of Korean Iris species, the molecular phylogenetic studies can now prompt a search for diagnostic characters, such as general morphological and micromorphological traits, a necessary prerequisite for any systematic and taxonomic context. Further integrative analyses of plastome sequences and morphological data of the two species at population level as well as employing molecular cytogenetic approaches using species-specific satellite DNA as probes may also offer insights into the species delimitation of the two closely related Iris species.

Materials and methods
Taxon sampling. All plant materials were collected from natural populations in Korea (Table 1) Table 1). To evaluate the consistency of morphological and micromorphological characters, living specimens were collected from multiple populations (three I. ruthenica specimens from three populations and three I. uniflora specimens from one population), and cultivated at Chungnam National University (Table 1). Meanwhile, for plastome sequencing, representative fresh leaves were collected from I. ruthenica and I. uniflora specimens (accession numbers: BKC939 and JCK2019-77, respectively; Table 1). Specific locality information including GPS coordinates, latitude, and longitude cannot be provided due to the endangered/rare status of the species in Korea. All voucher specimens were deposited in the Chungnam National University Herbarium (CNUK).

Micromorphological analysis.
Fresh leaf and flower materials from all four Iris populations were preserved using a formalin-acetic acid-alcohol solution and dehydrated by soaking in an acetone series (50, 70, and 90%) for 30 min and absolute acetone for 1 h. The dehydrated materials were then immersed in carbon dioxide for critical point drying (EMCPD300, Leica Microsystems, Germany), coated using an ion-sputtering device (E-1010, Hitachi, Japan), and analyzed using scanning electron microscopy (SEM; S3000N, Hitachi, Japan), with an accelerating voltage of 20 kV and working distance of 9-15 mm, following Choi et al. 5 , as well as light microscopy (BX53F, Olympus, Japan), which was used to analyze the structure of leaf epidermal pavement cells and stomatal complexes, following Kim et al. 7 . At least 20 guard cells were examined on both adaxial and abaxial leaf surfaces from each sample as described by Choi et al. 5 . To measure the pollen viability of each species, 10 randomly selected anthers from each individual plant were placed in an aniline blue dye solution to distinguish fertile pollen grains as described by Jang et al. 93 . At least 20 sterile and fertile pollen grains from each sample were randomly selected for size measurements. Both leaf and pollen micromorphological characters were measured using MicroMeasure ver. 3.3 program following Jang et al. 94 . For SEM, imaging was performed for both the abaxial and adaxial leaf surfaces, as well as for both the inner and outer surfaces of the outer tepals, stamens, and pollen grains. Orbicules, which are cellular structures of sporopollenin particles produced by the secretory tapetum, were investigated in this study for the first time using SEM in the genus Iris.
Plastome sequencing and assembly. Total genomic DNA was extracted from freshly collected samples using the modified CTAB method 95 , and Illumina short-insert paired-end sequencing libraries (TruSeq DNA Nano kit, Illumina, San Diego, CA, USA) were constructed and sequenced using the Illumina MiSeq platform. For de novo plastome assembly, sequencing reads were trimmed and filtered using FastQC v0.11.7 96 , and the resulting trimmed paired-end reads (Phred score ≥ 20) were assembled using Velvet 1.2.10 97 , with kmer values of 71, 91 101, and 111 to form large contigs. The Velvet contigs were then assembled into complete plastomes using the de novo assembly option in Geneious prime (https:// www. genei ous. com) and ordered using reference plastome sequences from I. gatesii (NC_024936), I. sanguinea (NC_029227), and I. missouriensis (NC_042827). Finally, the LSC/IR, IR/SSC, SSC/IR, and IR/LSC regions of the complete plastomes were validated using PCR- Plastome annotation and repeat sequence analysis. The I. ruthenica and I. uniflora plastomes were annotated using GeSeq 98 . Protein-coding sequences were manually curated and confirmed using Artemis 99 and then checked against the NCBI protein database. tRNA genes were confirmed using tRNAscan-SE 1.21 100 , and IR region sequences were confirmed using IR finder and RepEx 101 . Finally, circular maps of the I. ruthenica and I. uniflora plastomes were generated using OGDRAW 102 . For plastome comparison, GC content and relative synonymous codon usage (RSCU) were calculated using MEGA6 103 . The two plastomes were also compared using mVISTA in Shuffle-LAGAN mode, with I. ruthenica plastome as a reference, and nucleotide variation (Pi) among the plastomes, excluding regions of < 200 bp, was calculated using DnaSP version 6.1 104 . Each plastome was divided into genes, introns, and intergenic regions.
Plastome phylogenetic analysis. The plastome sequences of eight taxa, including six other taxa from the Iridaceae and the outgroup taxa Crocus cartwrightianus (NC_041459) and C. sativus (NC_041459), were obtained from NCBI GenBank (Supplementary Table S3), and two matrices, one of whole plastome sequences and another of 78 conserved protein-coding sequences (CDS), which excluded duplicate genes in the IR region, were generated using MAFFT ver. 7 105 ; all ten plastome sequences were then manually adjusted using Bioedit 106 . For CDS analysis, the aligned CDS were extracted and concatenated using Geneious (https:// www. genei ous. com) and filtered to remove ambiguously aligned regions using GBLOCKS ver. 0.91b 107 . The best-fitting model for nucleotide substitution was determined using the Akaike Information Criterion (AIC) in JModeltest V2.1.10 108 (Supplementary Table S4). Maximum likelihood (ML) analysis was performed using RaxML v8.0.5 109 , with 1000 bootstrap replicates and the GTR + I + G model, and Bayesian Inference (BI) analysis was performed using MrBayes 3.2.2 110 , with two independent runs and four simultaneous Markov Chain Monte Carlo runs of 5,000,000 generations each. The resulting trees were sampled every 100,000 generations, with the first 25% discarded as burn-in, and the 50% majority-rule consensus tree was visualized using Figtree V.1.4.2 111 , with posterior probability (PP) values estimated from trees sampled after the burn-in fraction was discarded.
Ethics approval. The experimental research and field studies on plants, including the collection of plant material, complied with relevant institutional, national, and international guidelines and legislation. The appropriate permissions and/or licenses for collection of plant were obtained for the study.