Taxonomic Composition of Iris Subser. Chrysographes (Iridaceae) Inferred from Chloroplast DNA and Morphological Analyses

The species of Iris subser. Chrysographes are herbaceous perennials found mainly in southwestern and central China and also in the Eastern Himalayas. To date, six species have been recognized in this group. In the framework of its taxonomic revision, we have carried out molecular and morphological studies. For this, we have sequenced four chloroplast DNA regions (trnS–trnG, trnL–trnF, rps4–trnSGGA, and psbA–trnH) for 25 samples across the major distribution ranges of the six species. Our phylogenetic analyses evidence that I. subser. Chrysographes is indeed a monophyletic group, which is sister to I. subser. Sibiricae. Within I. subser. Chrysographes, we have recovered four divergent lineages further supported by diagnosable morphological traits and geographical distributions. In this context, our data confirm the recognition of I. clarkei, I. delavayi, and I. wilsonii in their traditional concepts. Furthermore, both molecular and morphological data support the close affinities and similar distribution ranges of I. bulleyana, I. chrysographes, and I. forrestii, which suggests including I. chrysographes and I. forrestii as color forms in I. bulleyana. A revised taxonomic treatment for the group, including the notes on the species distributions and habitats, and also an identification key to the species are provided.


Introduction
While revising the series Sibiricae (Diels) G.H.M.Lawr. of the section Limniris Tausch of the genus Iris L., we found that the taxonomy of this group remains unclear [1]. Iris ser. Sibiricae includes rhizomatous herbaceous perennials morphologically distinguished mainly by their hollow flowering stems (except for I. clarkei Baker ex Hook.f.). The infrageneric taxon Sibiricae was proposed by Diels [2] as a subsection including eight species. Simonet [3] subdivided this subsection into two groups on the basis of their chromosome numbers. Nevertheless, the distinctness of the two subseries, recognized on morphological ground and from their distributions [4,5], gained support from molecular studies [6,7].
Iris sibirica L., I. sanguinea Hornem., and I. typhifolia Kitag., with their 2n = 28 chromosomes [3,8,9], have been recognized in the autonymic subseries of I. ser. Sibiricae [10,11] known to horticulturists under the common name "Siberian irises" [12]. In a recent study [1], we confirmed that I. subser. Sibiricae includes only the highly variable I. sibirica (Figure 1b,c). It is the most widespread Eurasian species of Iridaceae, occurring from central and eastern Europe to the Russian Far East. Iris subser. Chrysographes (Simonet) L.W.Lenz species are distributed from northeastern India and southwestern China, Nepal, and Bhutan to northern Myanmar, and are found at mid-to high elevations [11,13,14]. The name Chrysographes was proposed by Simonet [3] as a subsection for the species with the chromosome number 2n = 40. Alternatively, Chrysographes was referred to the genus Limniris (Tausch) Rchb. as a series [15], or a section [16], and is commonly known as "Sino-Siberians" among horticulturists [12]. Morphologically (see Figure 1), the I. subser. Chrysographes species is characterized by having longer bracts and perianth tube, and flowers without the venation characteristic of I. sibirica [5,11,17].
As is recognized, I. subser. Chrysographes comprises eight species, I. bulleyana Dykes, I. chrysographes Dykes, I. clarkei, I. delavayi Micheli, I. dykesii Stapf, I. forrestii Dykes, I. phragmitetorum Hand.-Mazz., and I. wilsonii C.H.Wright; however, the taxonomy of some of them was considered controversial [4,10,11,13,14,18]. As a result of a preliminary taxonomic study based on an analysis of the original material [19], six species were attributed to I. subser. Chrysographes (Figure 1d-i). Among them, I. delavayi, I. wilsonii, I. bulleyana, and I. forrestii were described from cultivated plants. The examination of the original material has shown that I. delavayi, as well as I. bulleyana and I. forrestii (both from Lijiang), originated from the northwestern Yunnan Province, China, while I. wilsonii originated from the northwestern Hubei Province, China [19]. Iris clarkei and I. chrysographes were described from plants collected in the wild, from eastern India and the Sichuan Province (China), respectively.
Nevertheless, it was indicated that there remains a great deal of confusion about the I. subser. Chrysographes species from southwestern China, and a thorough revision is required to resolve the taxonomy of this group [10,20,21]. A doubt was expressed about whether I. bulleyana is an independent species. For years, I. bulleyana was suggested to be a natural hybrid between I. forrestii and I. chrysographes since, in the wild, these species grow in the same regions, and their natural hybridization is possible [4,22]. However, fieldwork in the Yunnan Province (China) showed that I. bulleyana is a true species and is sometimes found in associations with I. chrysographes or I. forrestii [11,14]. According to another point of view [21], I. chrysographes and I. bulleyana are considered extreme forms of a single species. In particular, Noltie [10] (p. 300) noted that no differentiation can be made between the two yellow-flowered I. forrestii and I. wilsonii when based solely on herbarium material, I. bulleyana is merely a purple-flowered form of the same species, and that I. chrysographes and I. delavayi are probably distinct species, though it is very difficult to distinguish between them in a herbarium.
Accordingly, in the framework of taxonomic studies carried out on I. ser. Sibiricae [1, 19,35], the aims of the present study are as follows: (1) to resolve the phylogenetic relationships of the I. ser. Sibiricae species and some other series of I. sect. Limniris using four cpDNA regions; (2) to determine a possible number of putative chloroplast lineages within I. subser. Chrysographes; (3) to study the morphological characters of the I. ser. Sibiricae species; and (4) to compare the results of molecular and morphological studies to determining the taxonomic composition of I. subser. Chrysographes.

Plant Samples, DNA Extraction, and Sequencing
A total of 25 fully verified samples representing I. subser. Chrysographes were used for the molecular analyses. All samples were taken from living collections or herbarium specimens, of which three were of unknown geographical origin and the others were from 22 localities in southwestern China (Yunnan, Sichuan, and Xizang provinces), India, and Nepal ( Figure 2). The complete list of samples, including their origin and voucher information, is provided in Table 1.  The methods for DNA extraction, amplification, and direct sequencing of four noncoding cpDNA regions (trnS-trnG, trnL-trnF, rps4-trnS GGA , and psbA-trnH) have been described elsewhere [32,37]. Forward and reverse sequences for each region were determined on a genetic analyzer ABI 3130 (Applied Biosystems, Bedford, MA, USA) at the Joint Center of Biotechnology and Gene Engineering, the Federal Scientific Center of the East Asia Terrestrial Biodiversity (Vladivostok, Russia), and assembled using the Staden Package, version 1.4 [38]. The sequences previously obtained for I. sibirica [1] were also included in the analyses. In phylogenetic analyses, we also used the sequences previously published for representatives of three series of I. sect. Limniris [33,34]. These are (1) I. laevigata Fisch., I. ensata Thunb., and I. pseudacorus L. from I. ser. Laevigatae  Table 1.

Sequence Alignment and Phylogenetic Analyses
The sequences of each cpDNA region were aligned manually in SeaView, version 4 [39], and concatenated for each specimen. We included indels and length variation in mononucleotide repeats in the dataset because repeatability tests allowed excluding PCR errors. The haplotypes were identified based on combined DNA sequences using DnaSP, version 5 [40]. A network of haplotypes was constructed using Network, version 4.6 [41], with treating each deletion/insertion, regardless of size, as a single mutational event and using the MJ algorithm with default settings.
Phylogenetic analyses were conducted using the ML and MP methods as implemented in PAUP, version 4.0 b10 [42]. Bayesian analysis was performed using MrBayes, version 3.2.2 [43] via the CIPRES portal [44]. The dataset was composed of sequences from the I. subser. Chrysographes specimens, haplotypes H1-H8 of I. subser. Sibiricae [1], and sequences of species from three other series of I. sect. Limniris (i.e., Laevigatae, Lacteae, and Ruthenicae) and I. dichotoma as outgroup. For the MP analysis, gaps were coded according to the simple indel coding procedure [45] as implemented in FastGap, version 1.2 [46]. Optimal trees were found using a heuristic search with TBR branch swapping and the MulTrees option in effect. For ML and BI analyses, the GTR + I + G model was selected according to the Akaike information criterion using Modeltest, version 3.6 [47]. ML heuristic searches were done using the resulting model settings, 100 replicates of random sequence addition, TBR branch swapping, and MULTrees option on. In BI, using the default prior settings, two parallel MCMC runs were carried out for 10 million generations, sampling every 1000 generations for a total of 10,000 samples. Convergence of the two chains was assessed, and PP were calculated from the trees sampled during the stationary phase. The robustness of nodes in ML and MP trees was tested using bootstrap with 1000 replicates (BP).
The degrees of divergence between the groups identified in the MJ and phylogenetic analyses were calculated based on nucleotide substitutions using DnaSP. The distribution of genetic variation within and among these groups and F ST among them was determined by AMOVA as implemented in Arlequin, version 3.5 [48]. The significance of the variance components and genetic distances were tested using 1000 random permutations.

Morphological Data
The main taxonomic works dealing with I. subser. Chrysographes were consulted [4,10,11,13,14,18]. In order to clarify morphological characters of the species and compile the morphological key, the herbarium specimens deposited at BM, E, K, and LE (herbarium codes according to Thiers [36]), including the type material of the names studied, were examined personally by the authors [19]. In addition, the specimens have  (Table 2 and Table S1, Figure 1a-c). Since the collected data were used to identify morphological distinctions in the entire series studied, we, hence, incorporated the detailed data on I. sibirica from recent work [1]. In a total, 540 specimens (see Annex 1) of well-developed plants in flowering and fruiting stages, collected from Bhutan, China, India, and Nepal, we examined based on the qualitative and quantitative morphological characters useful to distinguish species. The herbarium specimens were identified on the basis of our own experience in dealing with this group. The quantitative characters were measured using AxioVision, version 4.8 (Carl Zeiss, Germany).

Morphometric Analysis
Our morphometric analysis of I. bulleyana, I. chrysographes, I. delavayi, and I. forrestii was based on eight quantitative (BL, CL, LL, LW, NC, NF, PL, and SH) and one qualitative (IS) characters ( Table 2). As the dataset of I. delavayi was limited to 13 individuals, for statistical analysis, we randomly selected 13 samples for each species from the initial dataset by using the built-in resample function of the R free software for statistical analysis [49], version 4.1 [50]. The dataset was analyzed by using one-way ANOVA. Differences were considered statistically significant at a p-value < 0.05. After a multiple statistical testing was performed, the calculated p-values were adjusted using the procedure proposed by Benjamini and Hochberg [51]. To test one-way ANOVA assumptions, the Shapiro-Wilk's test for normality of distribution [50] and Levene's test for equality of variances [52] were used. If an ANOVA showed a statistically significant difference among species, then subsequent pairwise comparisons were made using the Tukey's post-hoc test [53]. Inequality of variance was taken into account by using the heteroscedastic consistent covariance estimation provided in the R add-on package "sandwich", version 2.3.0 [54,55]. Analysis of the countable characters (NF and NC) was done by the Poisson regression using the respective R built-in function [50].

Taxonomy and Distribution
Here, the conservative taxonomy of Iris is used [2,6,7,10,13,14,22,24]. The types of the I. subser. Chrysographes names were selected in a recent nomenclatural study [19]. For the taxonomy, the Shenzhen Code [59] was consulted. In the case of disagreement on the infraspecific rank at which a name should be accepted, we followed Brummitt [60]. In the Taxonomic treatment section (see below), we extracted the information on distribution of the accepted taxa from the herbarium specimens. We also consulted the information provided in References [4,10,11,13,14,18,20,22,61], which are commonly recognized as taxonomically reliable sources.

Genetic Divergence of Chloroplast Non-Coding Sequences within Iris Subser. Chrysographes
Among the 25 specimens of I. subser. Chrysographes, 12 haplotypes (C1-C12; see Table 1) were identified on the basis of polymorphic sites detected across 3704 aligned positions of four cpDNA regions. Four of these haplotypes (C1, C2, C7, and C9) were found in several localities, sometimes geographically very distant from each other (e.g., C9 in SLK and TNS from India and Nepal, respectively), while the others were unique, i.e., found in a single locality.
The relationships between haplotypes found in representatives of I. ser. Sibiricae are shown in Figure  Haplogroup B includes one haplotype C7, which was found in samples from different localities near the Cang Mountains (Yunnan Province, China), and is separated from haplogroup A by 7 mutational steps. Haplogroups C and D are separated from each other by 11 mutational steps and from other two haplogroups (A and B) by 12-14 mutational steps including 9-bp insertion in the trnL-trnF region. Haplotypes of C group were found in the specimens from the Xizang Province, China (C8 in SXG and C9 in TNS), and the Eastern Himalayas (C9 and C10 in SLK and NTP, respectively). Haplotypes C11 and C12 of haplogroup D were found in the I. wilsonii specimens: YLZ from Shangri-La (formerly known as Zhongdian; Yunnan, China) and DAL from the Daliang Mountains (Sichuan, China). The pairwise F ST values between four haplogroups varied from 0.609 to 0.929 (p < 0.05), and K S varied from 0.00374 to 0.00447.
In all the phylogenetic analyses, Iris accessions were distributed with a robust support (PP 1.0, BP 100 and 100%) in accordance with their affiliation to the corresponding series of I. sect. Limniris (Figure 4). Iris ser. Sibiricae was resolved as a monophyletic group (PP 1.00, BP 100 and 100%) consisting of two strongly supported sister clades corresponding to I. subser. Sibiricae (PP 1.0, BP 100 and 100%) and I. subser. Chrysographes (PP 1.0, BP 100 and 100%) that are recognized in this group. The pairwise F ST between these subseries was 0.825 (p = 0.00001), and K S between them was 0.00892. Within clade I. subser. Chrysographes, there was a polytomy of three monophyletic clusters, with the divergence between them varying from 0.00408 to 0.00440 (Table S3). In cluster I (PP 1.0, BP 99 and 98%), two sister groups, A and B, were resolved with a support of PP 1.0, BP 97 and 95% and PP 0.93, BP 86 and 85%, respectively. These groups corresponded to haplogroups A and B revealed by the MJ-network analysis (Figure 3). The pairwise F ST between these haplogroups was 0.86 (p = 0.00001), and K S between them was 0.00153. The nucleotide divergence between the sequences of species comprising haplogroup A (I. bulleyana, I. forrestii, and I. chrysographes, hereinafter referred to as the bulleyana group) varied from 0.00023 to 0.00063 (Table S3). The pairwise F ST values between them varied from 0.385 to 0.419 and were not significant (p > 0.05). No nucleotide substitutions or indels differentiating these species were revealed. The sequence divergence between each of these species and I. delavayi (from haplogroup B, Cluster I) was higher and varied from 0.00133 to 0.00173. Clusters II and III were consistent with haplogroups C and D, which was revealed by the MJ-network analysis (Figure 3).

Morphological Study
To evaluate the taxonomic significance of the molecular results, we performed a morphological study of the I. ser. Sibiricae species. The main diagnostic characters that allow distinguishing between the I. ser.Sibiricae species are summarized in Table 3. Morphologically, the I. subser. Chrysographes species are distinguished from I. sibirica by their longer bracts (most commonly more than 6 cm in length), which are also green at blooming (dry in I. sibirica), by the 1-2.2 cm long perianth tube (not longer than 0.5 cm in I. sibirica), and by the slightly veined ornamentation of the falls (strongly veined in I. sibirica; see Figure 1).   Table S1 for more details. Descriptions of the characters and their codes are provided in Table 2; for illustrations, see Figure 1.
Eigenvalues of the measurable morphological characters (principal components), which were essential to the observed variations between the taxa, are given in Table S2. Variables CL, LL, LW, NC, NF, and SH had the largest share in the separation of the species of the bulleyana group and I. delavayi, the closely allied species according to our molecular data. The cumulative percentage of the explained variance was 68.74%. A further analysis showed a statistically significant difference in BL and PL between I. delavayi on the one hand, and between I. bulleyana and I. forrestii on the other (p = 0.0012 and p < 0.0001, respectively). Between the species of the bulleyana group, no statistically significant difference was found in BL, CL, LL, NC, NF, and PL. A significant difference in LW and SH was found in the pair I. forrestii and I. chrysographes; the other pairs of species in the bulleyana group did not show any difference.
The PCA of the quantitative characters of the estimated variance component for all the samples gave values of 46.2% and 11.5%, respectively, for the first two principal components ( Figure 5). Two characters, NF and PL, displayed the highest correlations with the first (NF, r = 0.85) and the second axis (PL, r = 0.60); the third one, CL, highly influenced the third axis (CL, r = −0.83). In the PCA scatter-plot of all the individuals in the plane defined by the first two principal components (Figure 5), the studied specimens grouped together in accordance with their taxonomic affiliation, creating two separate groups. The first one included the species of the bulleyana group, and the second one included I. delavayi ( Figure 5). An examination of the biplot ( Figure 5) from first two principal components revealed a partial overlapping of I. bulleyana, I. chrysographes, and I. forrestii and also their significant morphological similarity. The separation of I. delavayi from the other three taxa was defined by the first two principal components. Thus, having likely different average values of some morphometric characters caused by environmental conditions and interspecific trait variability, I. bulleyana, I. chrysographes, and I. forrestii can still be considered as indistinguishable in a generalized (PCA) factor space. Therefore, the result of PCA proved to be the same as that of the molecular study, and it was sufficient to supplement molecular evidence.

Discussion
Based on the sequencing of cpDNA regions for the samples from different localities within the I. ser. Sibiricae distribution range, our study confirms the monophyly of two main divergent lineages and a sister relationship between them. Such a pattern is generally consistent with results of several phylogenetic studies [6,24,62] and supports the splitting of I. ser. Sibiricae into two subseries, as it was previously suggested on the basis of chromosome numbers, morphology, and distribution (e.g., References [5,11]). The first lineage contains haplotypes of all samples from the northern part of the range, where one species (I. sibirica) is distributed [1], and the second one contains haplotypes of samples from the southern part of the range, where species of I. ser. Sibiricae with 2n = 40 chromosomes are distributed [11,13,14]. The high levels of genetic differentiation and nucleotide divergence of cpDNA between these lineages indicate a deep genetic split between them, which may suggest a long independent evolutionary history of species from the two subseries.
Recently, six taxa at the species rank have been assigned to I. subser. Chrysographes [19]. Based on the cpDNA region sequence analyses, we revealed four distinct genetic lineages (Figures 3 and 4). The values of nucleotide sequence divergence between these lineages (Table S3) are comparable with the divergence between other species in Iris [32,34] and between closely related species in other genera (e.g., References [63,64]). The lowest value was found between haplogroups A and B (Table S3) forming a single cluster (I) in the phylogenetic analyses (Figure 4), which is indicative of close genetic relationships between the species constituting them. In our study, we could not distinguish genetically between these three taxa with minimally divergent haplotypes lacking species-specific markers and forming a star-like structure in the network, which indicates no deep phylogenetic split between them and is consistent with the rapid range expansion. The species forming haplogroup A, i.e., the bulleyana group, can easily interbreed with each other during cultivation and in the wild and, thus, lose their identity [4,14].
Grey-Wilson [4,11] believed that I. clarkei might perhaps best be placed in a separate group of its own. On the contrary, our molecular data showed that both I. clarkei and I. wilsonii are placed in the monophyletic clade of I. subser. Chrysographes as distinct lineages. Dykes [13] suggested that I. forrestii may roughly be described as a dwarf I. wilsonii. It should be noted that, despite the fact that the overall distributions of I. forrestii and I. wilsonii appear to overlap, there is no indication that the two species grow together in the wild and, therefore, their natural hybridization is likely to be ruled out [11]. Moreover, I. forrestii blooms about two weeks earlier than I. wilsonii [65]. Our results suggest a phylogenetic affinity of I. wilsonii with I. clarkei. In addition, the molecular data shows that I. delavayi is the closest species to the species of the bulleyana group and appears as a sister taxon in the same cluster with them. This species is distinct in morphology (Table 3), being, however, more genetically similar to I. chrysographes than to any other of the I. subser. Chrysographes species. Thus, our analyses support the recognition of I. clarkei, I. delavayi, and I. wilsonii as distinct species.
Traditionally, I. clarkei is regarded as unique and holding an isolated position within I. ser. Sibiricae, as it has a solid flowering stem (e.g., References [10,11,13,14]). However, according to some authors [20,66] and the herbarium specimens (e.g., L3912484; see Annex 1), the flowering stem in I. clarkei is considered to be solid with a small central hollow. We assume that, generally or in some cases, the flowering stem in I. clarkei is not completely solid, with the central part of the stem filled with a broad expanse of pith. Similarly, it is generally accepted that I. wilsonii has an unbranched flowering stem with a terminal inflorescence of two flowers, as we observed when examining the herbarium specimens. Contrary to this, we found specimens of I. wilsonii heaving 1-branched flowering stems. These plants were collected in Zhaojue County, Sichuan Province (the gatherings "W. Sun 15" and "Sichuan Vegetation Team 12818"; see Annex 1).
There is a high degree of morphological similarity between specimens belonging to the bulleyana group, which probably results from their close relationship, as indicated by many researchers. As pointed out by Stapf [67] and Dykes [13,22,68], the author of the bulleyana group taxa, I. chrysographes is closely allied to I. forrestii, while differing in color of flowers and in habitats, and I. bulleyana also strongly resembles I. forrestii. According to Noltie [10,21], the evidence of the variability observed in the northwestern Yunnan Province in 1993 provides much support to the idea of treating I. chrysographes and I. bulleyana as forms of the same species differing mainly in the flower color, and our results suggest that this assumption is reasonable. In fact, it was confirmed on the basis of the morphological results that I. bulleyana, I. chrysographes, and I. forrestii have considerable morphological variation, and, even though the studied specimens exhibit continuity of morphological characters, they, however, differ in flower color (Table 3). In addition, the morphometric analysis showed one overlapping group in which the specimens of the bulleyana group were indistinctly separated from each other and formed a single aggregation ( Figure 5).
In addition, our examination of the herbarium specimens showed (see Annex 1) that the species of the bulleyana group have the same distribution in China. In the Yunnan Province, these are common in Dali Bai Autonomous Prefecture, Diqing Tibetan Autonomous Prefecture (Haba Snow Mountain, Shangri-La county-level city), the Lijiang prefecture-level city (Yulong Naxi Autonomous County, including I. bulleyana f. alba), and Nujiang Lisu Autonomous Prefecture. In the Sichuan Province, these are common in Garzê Tibetan Autonomous Prefecture (Jiulong County), and Liangshan Yi Autonomous Prefecture (Muli and Yanyuan counties); additionally, I. bulleyana and I. chrysographes occur in Ngawa Tibetan and Qiang Autonomous Prefecture (Mianning County). Moreover, as was reported in previous works [11,14] and is confirmed in the present study, I. bulleyana is sometimes found in associations with I. chrysographes and I. forrestii (e.g., the gathering "C.W. Wang 63721"; see Annex 1) or in mixed populations of I. bulleyana and I. chrysographes (e.g., the gatherings "G. Thus, I. bulleyana, I. chrysographes, and I. forrestii do not differ in characters that are significant for taxonomic species differentiation within the genus, and the molecular data and morphometric characters selected here are sufficient to confirm this assumption. In such a situation, combining the critical taxa into a single species seems to be appropriate, and many researchers have adopted this approach (e.g., Reference [69]). For this reason, further determination of the three species with their overlapping distribution ranges is obviously not justified, and, therefore, we suggest reconsidering their taxonomic status.

Taxonomic Treatment
In the present study, we confirm that I. ser. Sibiricae is divided into two groups, the autonymic subseries with a single species I. sibirica [1] and I. subser. Chrysographes. As a consequence of the present work, we consider form rank to be the most suitable option for I. bulleyana, I. chrysographes, and I. forrestii. The earlier described names, I. bulleyana and I. forrestii, were simultaneously published by Dykes [70], both with equal priorities until now. Here, we combine these taxa and establish priority of I. bulleyana over I. forrestii, the other competing name (see Art. 11.5, Note 3 of the ICN).

The List of Taxa
A list of the taxa within I. subser. Chrysographes accepted in the present work and information on species distributions, habitats, and types is provided below.
( Distribution and habitat-This species has a more westerly distribution and is native to the Central and Eastern Himalayas, particularly to northeastern India (the states of Manipur, Sikkim, and West Bengal and the Ladakh union territory), central and eastern Nepal, the Haa and Paro valleys in western Bhutan [71], and northern Myanmar [72]. In southwestern China, it is distributed in the southeastern Xizang Province (the Nyingchi and Shigatse prefecture-level cities) and northwestern Yunnan Province (Nujiang Lisu Autonomous Prefecture). It grows commonly in shady places in marshes, wet meadows, woodland margins, and beside streams and lakes at elevations of 2300-4300 m.
( however, we could not confirm its distribution there with any reliable herbarium specimens and, therefore, it requires verification. The populations of this semi-aquatic species are severely fragmented in distribution. They generally grow in damp places along ditches and streams, in wet or marshy mountain meadows, and swampy places at elevations of 2300-3400 m.
( Distribution and habitat-This species is endemic to the Hengduan Mountains in southwestern China, where it is very common. It grows in moist areas among grasses on hillsides or forest edges, in meadows, pastures, and beside streams at elevations of 1800-4800 m. have shown that the chloroplast markers trnS-trnG, trnL-trnF, rps4-trnS GGA , and psbA-trnH provide a reliable resolution of the species and are optimal molecular markers for identifying taxonomic and phylogenetic relationships within critical taxa of the genus Iris.
The four lineages of I. subser. Chrysographes correspond to four morphologically distinct, biogeographically congruent groups: I. clarkei in the western part of the range, I. wilsonii in the eastern part, and I. delavayi and a complex of I. bulleyana, I. chrysographes, and I. forrestii in the central part of the range. In addition, our data show that the morphological characters of I. bulleyana, I. chrysographes, and I. forrestii were within the range of variation of a single species to which they are assigned. A morphometric analysis based on nine morphological characters has not revealed any separation between the three taxa. In congruence with the molecular data, I. bulleyana, I. chrysographes, and I. forrestii show weak morphological differentiation and, thus, are better to be treated as color forms of the same species. Based on our present results, we accept I. clarkei, I. delavayi, and I. wilsonii in their traditional concepts and recognize the other three taxa as a single species. Therefore, we suggest two combinations, I. bulleyana f. forrestii and I. bulleyana f. chrysographes.
Molecular evidence obtained in this study contribute to the knowledge of the taxonomy of irises and their distribution in China. Thus, one of the goals set for the future is to produce a thoroughly verified, sufficiently sampled, and robust phylogenetic tree that would provide a basis for a revised phylogeny of Iris s.l. Furthermore, a thorough re-examination of some morphological characters is also needed, using a broader set of samples across the entire distribution ranges of the species considered, since we still lack adequate understanding of the flowering stem structure in I. clarkei and the flowering stem branching in I. wilsonii.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/plants10112232/s1, Annex 1: Complete list of specimens examined in the morphological analysis, Table S1: Raw data of the morphological analysis, Table S2: The results of the variance analysis of the Iris subser. Chrysographes species, Table S3: Nucleotide divergence between groups identified by the MJ (four haplogroups) and phylogenetic analyses (three clusters) of Iris subser. Chrysographes from 25 localities, and also between I. bulleyana, I. forrestii, and I. chrysographes (haplogroup A) and I. delavayi (haplogroup B) as inferred from the cpDNA data.