Evidence from Phylogenomics and Morphology Provide Insights into the Phylogeny, Plastome Evolution, and Taxonomy of Kitagawia

Kitagawia Pimenov is one of the segregate genera of Peucedanum sensu lato within the Apiaceae. The phylogenetic position and morphological delimitation of Kitagawia have been controversial. In this study, we used plastid genome (plastome) and nuclear ribosomal DNA (nrDNA) sequences to reconstruct the phylogeny of Kitagawia, along with comparative plastome and morphological analyses between Kitagawia and related taxa. The phylogenetic results identified that all examined Kitagawia species were divided into Subclade I and Subclade II within the tribe Selineae, and they were all distant from the representative members of Peucedanum sensu stricto. The plastomes of Kitagawia and related taxa showed visible differences in the LSC/IRa junction (JLA) and several hypervariable regions, which separated Subclade I and Subclade II from other taxa. Fruit anatomical and micromorphological characteristics, as well as general morphological characteristics, distinguished the four Kitagawia species within Subclade I from Subclade II and other related genera. This study supported the separation of Kitagawia from Peucedanum sensu lato, confirmed that Kitagawia belongs to Selineae, and two species (K. praeruptora and K. formosana) within Subclade II should be placed in a new genus. We believe that the “core” Kitagawia should be limited to Subclade I, and this genus can be distinguished by the association of a series of morphological characteristics. Overall, our study provides new insights into the phylogeny, plastome evolution, and taxonomy of Kitagawia.


Introduction
Peucedanum sensu lato, with 100-120 species distributed in Eurasia and Africa, is taxonomically one of the most complex groups in the Apiaceae [1]. Peucedanum sensu lato has long been regarded as extremely heterogeneous and contains a great diversity of life-forms, leaf and fruit structures, and chemical constituents [2,3]. Based on morphological and molecular studies, the genus is now reduced to only a few species allied to the type species Peucedanum officinale L., called Peucedanum sensu stricto, and several segregates are recognized as distinct genera [4][5][6][7][8][9].
Kitagawia Pimenov is one of the segregate genera of Peucedanum sensu lato. This genus was first described by the Russian botanist M. G. Pimenov in 1986 [10]. By investigating the carpological, morphological, and biochemical characteristics of species of Peucedanum sensu lato from the Far East and Siberia, Pimenov identified five species and one subspecies in the new genus Kitagawia and designated Kitagawia terebinthacea (Fisch. ex Trevir.) Pimenov as the nomenclatural type [10]. According to the original description of the genus by Pimenov [10], Kitagawia possesses distinguishing characteristics, such as partial lignification of mesocarp parenchyma and the absence of several flavonoids common to Peucedanum sensu lato.
analyses of phylogenetic relationships [38][39][40]. Plastome sequences can effectively improve the support and resolution of phylogenies at the generic level and beyond [41][42][43][44][45][46]. In particular, the plastome-based phylogeny performed by Liu et al. [47] constructed a robust phylogenetic framework for Peucedanum sensu lato and provided a valuable reference to our investigation of Kitagawia. In addition, comparative analyses of plastomes can also provide useful information for eliciting evolutionary and interspecific relationships [48][49][50], which should further improve our understanding of the taxonomic classification of Kitagawia.
The combination of molecular data and morphological characteristics has proven to yield solid evidence for the phylogeny and taxonomy of many Apiaceae [51][52][53][54][55]. In the present study, we performed comprehensive phylogenetic analyses based on plastomes and nrDNA sequences, complemented by detailed comparative plastome and morphological analyses of Kitagawia and related taxa. The objectives of this study were to: (1) reconstruct the phylogeny of Kitagawia; (2) investigate the plastome features of Kitagawia and related taxa; (3) verify the taxonomic value of fruit micromorphology for Kitagawia; (4) examine previous taxonomic treatments and proposals for Kitagawia.

Plastome Features of Kitagawia and Related Taxa
The complete plastome sequences range in size from 146,718 bp (K. formosana) to 148,327 bp (K. terebinthacea) for the 14 species of Kitagawia and related taxa (Table 1, Figure S1). All these plastomes comprised a pair of inverted repeat (IR) regions (17,,043 bp) separated by the large single copy (LSC, 91,829-93,700 bp) and small single copy (SSC) regions (17,631 bp), exhibiting a typical quadripartite structure [43,47] (Table 1, Figure S1). The IR regions of three species (17,987 bp for P. morisonii, 18,058 bp for Angelica sylvestris L., and 18,095 bp for C. sinensis) were the shortest in length (  Figure S1). All 14 plastomes encoded 113-114 unique genes, with 80 protein-coding genes (PCGs), 29-30 transfer RNA (tRNA) genes, and 4 ribosomal RNA (rRNA) genes (Table 1 and Table S1). In comparison to other samples in this study, the plastomes of K. praeruptora, K. formosana, and Peucedanum harry-smithii var. grande lacked the trnT-GGU gene ( Figure S1, Table S1). Fifteen genes were duplicated in the IR regions, including five PCGs, six tRNA genes, and four rRNA genes (Table S1). Incomplete copy of gene ycf1 in the IRb region was regarded as a pseudogene (ψycf1) (Table S1). There were 16 intron-containing genes found, 13 of which were PCGs, and the remainder were tRNA genes (Table S1). Three genes (ycf3, clpP, and rps12) possessed two introns, whereas the others each contained only one (Table S1).

IR Border Analysis
The IR border regions and adjacent genes of the complete plastomes from 14 selected species of Kitagawia and related taxa were compared to analyze expansion and contraction in junction regions. Overall, Subclade I and Subclade II were similar at the four junctions, while both markedly differed from P. morisonii, C. sinensis, and A. sylvestris. The LSC/IRa junction (JLA) of the 14 plastomes occurred between genes trnL and trnH. There were 14-111 bp of non-coding sequence between JLA and the 3 end of gene trnH in Subclade I and Subclade II, which were considerably less than 928 bp in P. morisonii, 1068 bp in C. sinensis, and 978 bp in A. sylvestris ( Figure 3). Conversely, there were 1255-1848 bp of non-coding sequence in Subclade I and Subclade II between JLA and the 3 end of gene trnL, which were much more than 746 bp in P. morisonii, 873 bp in C. sinensis, and 882 bp in A. sylvestris ( Figure 3). The four junctions in Subclade II were relatively consistent, whereas, within Subclade I, they were more diverse. The LSC/IRb junction (JLB) was mostly located within gene ycf2, with 37-576 bp of gene ycf2 duplicated in the IRb region, except that the distance from JLB to the 3 end of gene ycf2 was 35-53 bp in K. baicalensis, K. stepposa, and P. hakuunense in Subclade I (Figure 3). Similarly, the SSC/IRb junction (JSB) were mostly located between pseudogene ψycf1 and gene ndhF, and were 3-156 bp away from the 3 end of gene ndhF, except for K. baicalensis, K. stepposa, and K. komarovii, where gene ndhF spanned JSB, and 33 bp of gene ndhF were duplicated in the IRb region ( Figure 3). In the 14 plastomes, gene ycf1 spanned the SSC/IRa junction (JSA) (Figure 3). For K. baicalensis, K. stepposa, and K. komarovii in Subclade I, 1884 bp of gene ycf1 duplicated in the IRa region, while for the remaining plastomes, the duplication portions of gene ycf1 in the IRa region were longer (1953-2269 bp) ( Figure 3).

DNA Rearrangement and Sequence Divergence Analyses
The DNA arrangement of the 14 plastomes examined was relatively conserved. No gene rearrangements were detected among them ( Figure S2). With reference to K. terebinthacea, the overall sequence identity and divergent regions across the 14 selected plastomes were analyzed. Clearly, the LSC and SSC regions were more divergent than the two IR regions, and coding regions showed more sequence conservation than non-coding regions ( Figure 4). In several highly variable regions (e.g., atpF-atpH, petN-psbM, accD-psaI, ycf2-trnL, rpl32-trnL, ycf1, and ycf2), Subclade I showed a high degree of similarity, which differed from Subclade II and P. morisonii ( Figure 4).

Fruit Anatomical and Micromorphological Examination
Fruit anatomical and micromorphological characteristics of the five Kitagawia species studied are shown in Table 2. Note: marw/fw: the ratio of the total widths of both marginal ribs to the fruit width; cow/fw: the ratio of the commissure width to the fruit width.

Phylogenetic Position and Non-Monophyly of Kitagawia
Previous phylogenetic analyses did not specifically examine the genus Kitagawia and may have been misled by misidentifications of some Kitagawia accessions, failing to identify a clear phylogenetic position for Kitagawia [27,29,30]. Recent studies have revealed that Kitagawia is distantly related to Peucedanum sensu stricto and other segregates of Peucedanum sensu lato, but they relied on single DNA fragments and were limited by low support [5,9,31]. In this study, we combined complete plastome sequences with nrDNA sequences to produce a well-supported phylogeny for Kitagawia.
In our study, single-copy CDS plastome sequences produced a robust phylogeny for Kitagawia. In the CDS tree, all six Kitagawia species examined fell into Selineae and were divided into two branches (Subclade I and Subclade II) (Figure 1). They were all considerably distant from P. morisonii, the member of Peucedanum sensu stricto confirmed in previous studies [2,8,29] (Figure 1). Similar results were presented in the nrDNA (ITS + ETS) tree ( Figure 1). Peucedanum officinale L., the type species of Peucedanum sensu stricto, formed a sister branch to P. morisonii, which was closely related to four Angelica species and more distantly related to the six Kitagawia species examined ( Figure 1). In both phylogenetic trees, Subclade I and C. sinensis, and Subclade II and S. divaricata formed sister branches, respectively ( Figure 1). Although there were differences in the topologies of plastome CDS and nrDNA (ITS + ETS) trees, they consistently indicated considerably distant phylogenetic relationships between Kitagawia and Peucedanum sensu stricto, thus providing strong support for the separation of Kitagawia. These trees also revealed that Kitagawia, as defined by Pimenov, was not monophyletic [4,10,17,18].
We believe that the "core" Kitagawia should be limited to Subclade I, while K. praeruptora and K. formosana should be removed from the genus. Furthermore, P. hakuunense was treated as a synonym of K. komarovii by Pimenov [18]; while our results showed that P. hakuunense should be identified as a distinct species, both P. hakuunense and P. chujaense may belong to Kitagawia. Our phylogenetic analyses confirmed that Kitagawia does not belong to the Acronema Clade but to Selineae. There was no doubt that some accessions of Kitagawia used in previous studies were misidentified [27,29,30]. Our results demonstrate the rationale behind the separation of Kitagawia and the necessity for its further taxonomic revision.

Plastome Structure and Evolution and Their Phylogenetic Implications
Our comparative analyses revealed that all 14 plastomes examined displayed a typical quadripartite structure [43,47] ( Figure S1). The plastome structure, GC content, and gene order were similar to those of species in the Apiaceae tribe Selineae (Figures S1 and S2, Table 1), implying that plastome structure was highly conserved [43,47,55]. Among the 14 species examined, the plastomes of K. praeruptora, K. formosana, and P. harry-smithii var. grande lacked the trnT-GGU gene (Table S2), which may suggest that they have different evolutionary histories when compared to other plastomes [56,57].
SSRs have been widely used as molecular markers in plant population genetics and evolutionary studies [58][59][60]. In the 14 studied plastomes of Kitagawia and related taxa, mononucleotides were the most abundant SSRs, followed by dinucleotides, tetranucleotides, trinucleotides, pentanucleotides, and hexanucleotides. Such findings are widespread in Apiaceae, Liliaceae, and Allium L. [47,61,62]. All identified SSRs were mainly composed of bases A and T, causing an AT richness in the overall plastome [43,63]. IR expansion and contraction have been recognized as evolutionary markers for illustrating relationships among taxa [43,55,64]. In this study, the JLA of species within Subclade I and Subclade II visibly differed from other species examined (Figure 3). Distances between gene trnL and JLA of species within Subclade I and Subclade II were considerably longer than P. morisonii, C. sinensis, and A. sylvestris (Figure 3). In contrast, distances from gene trnH to JLA of species within Subclade I and Subclade II were shorter than for the three above-mentioned species (Figure 3). Furthermore, the lengths of the IR regions of P. morisonii, C. sinensis, and A. sylvestris were shorter than for other species examined (18,457-19,043 bp) ( Figure 3, Table 1). The similar features of JLA and the IR length of P. morisonii, C. sinensis, and A. sylvestris may imply a common but different evolutionary history that differentiated them from the others (Subclade I and II and S. divaricata). In addition, sequence divergences noted in several regions (e.g., ycf1, ycf2, and accD-psaI) separated Subclade I from Subclade II and P. morisonii (Figure 4). These differences in plastomes may reflect distinct phylogenetic affinities between the species of Subclade I and the others [55]. Overall, the results of comparative plastome analyses revealed different evolutionary relationships of Subclade I and Subclade II to other taxa examined and added some support to the distinct phylogenetic position and non-monophyly of Kitagawia.
The association of a series of morphological characteristics is regarded as a more powerful tool to distinguish taxonomically difficult taxa compared to a single morphological characteristic [55]. Through our analyses, the morphological delimitation of Kitagawia (the four "core" Kitagawia species within Subclade I) should be "stems glabrous, leaf blades 2-3-pinnate, ultimate leaf segments linear to lanceolate, bracts and bracteoles lanceolate, rays adaxial surface hispid, petals white, calyx teeth subulate or triangular, stylopodia conical, mericarps strongly compressed dorsally, elliptic and glabrous, dorsal ribs filiform and slightly prominent, marginal ribs winged, commissure broad, solitary vitta in each furrow, two vittae on commissure, partly lignified parenchyma cells with pitted walls in marginal ribs, 1-2 layers of mesocarp cells closest to the epidermis, inconspicuous cell borders of fruit surfaces, without epicuticular wax". Considering that P. hakuunense and P. chujaense probably belong to Kitagawia, the morphological delimitation of Kitagawia may need further revision.

Taxonomic Suggestions for Six Species from Kitagawia and Peucedanum Sensu Lato
Pimenov stated that the generic boundaries of Kitagawia might need to be expanded and highlighted several Peucedanum sensu lato species that may belong to Kitagawia [10,17]. One example was discussed in the taxonomic treatment of K. formosana [18].
In this study, two Peucedanum sensu lato species considered as candidates for Kitagawia, P. ampliatum and P. harry-smithii var. grande, were included to investigate the generic boundaries of Kitagawia. In our phylogenetic trees, P. ampliatum and P. harry-smithii var. grande clustered with K. praeruptora and K. formosana, forming Subclade II (Figure 1). The phylogenetic positions of the four species within Subclade II were distant from Subclade I ("core" Kitagawia) and Peucedanum sensu stricto (Figure 1). Instead, they are closely related to S. divaricata, the type species of a monotypic genus Saposhnikovia (Figure 1). The rather distant relationship of Subclade II to Subclade I and Peucedanum sensu stricto suggest that the four species within Subclade II (including K. praeruptora and K. formosana) should neither be considered members of Kitagawia nor included in Peucedanum sensu lato, which is consistent with previous phylogenetic results [32,33,47,69].
In our morphological analyses, the four species within Subclade II, K. praeruptora, K. formosana, P. ampliatum, and P. harry-smithii var. grande, shared characteristics "stems pubescent or tomentose, bracts absent or few, linear or lanceolate, bracteoles numerous, linear or lanceolate, rays pubescent or tomentose on adaxial surface or throughout, petals white, mericarps pubescent or hispid, dorsal ribs filiform and prominent, marginal ribs narrowly winged, 3-5 vittae in each furrow, 6-8 vittae on commissure" [11,12,70], which differed from the four Kitagawia species within Subclade I, supporting the distant phylogenetic relationship of K. praeruptora and K. formosana to "core" Kitagawia ( Figure 1, Table S3). These characteristics shared by the four species within Subclade II could also distinguish them from Peucedanum sensu stricto [4] and Saposhnikovia (the latter have glabrous stems, rays, and mericarp, few bracteoles, solitary vitta in each furrow, two vittae on commissure, and one pseudovitta under each primary rib) [71] (Table S3). As far as the present study is concerned, the members of Subclade II possessed unique fruit anatomical and micromorphological characteristics, as well as general morphological characteristics, that could not be identified in other related genera (Tables S2 and S3, Figures 5 and 6). This may suggest that their current taxonomic placements are unreasonable. The phylogenetic and morphological results consistently suggested that K. praeruptora and K. formosana should be placed in a new genus and the two possible candidate taxa (P. ampliatum and P. harry-smithii var. grande) should not be included in Kitagawia. For the possible new genus, more extensive population sampling and study of additional, potentially related taxa are required to determine generic boundaries and morphological delimitations.
Two Peucedanum sensu lato species found in Korea, P. hakuunense and P. chujaense, fell into Subclade I. Of these, P. hakuunense was treated as a synonym of K. komarovii [18]. Peucedanum chujaense K. Kim, S.H. Oh, Chan S. Kim & C.W. Park was thought to be morphologically most similar to Kitagawia species [15]. Our phylogenetic analyses revealed that P. hakuunense and P. chujaense nested into Subclade I, suggesting they may need to be included in Kitagawia (Figure 1). However, specimens of P. hakuunense and P. chujaense were not available and morphological evaluations could not be undertaken. In our morphological analyses, therefore, the general morphological characteristics of P. hakuunense and P. chujaense were directly referenced from previous literature [15,72,73] to document the morphological similarity of these two species with the four Kitagawia species within Subclade I (Table S3). Phylogenetic and morphological evidence suggests that P. hakuunense and P. chujaense probably belong to Kitagawia. More evidence is needed for a complete taxonomic treatment of P. hakuunense and P. chujaense and for the clarification of interspecific relationships within Subclade I. To that end, international cooperation will be needed to produce a more detailed revision of Kitagawia.

Topologic Incongruence between Phylogenies based on Plastome and NrDNA Sequences
Previous molecular phylogenetic studies have identified a common phylogenetic incongruence between plastid and nuclear gene trees [43,74]. Several evolutionary processes, such as hybridization, introgression, and incomplete lineage sorting (ILS), are regarded as plausible explanations for the discordance of plastid and nuclear DNA phylogenies [75,76].
In our phylogenetic analyses, the phylogenetic positions of three main clades (i.e., Tordyliinae, Coriandreae, and Hymenidium Clade) differed between the two phylogenetic trees (Figure 2). These differences were also reported by Wen et al. [46]. We agree with Wen et al. [46] that the conflicting positions of the three main clades (Tordyliinae, Coriandreae, and Hymenidium Clade) between the plastome CDS and nrDNA (ITS + ETS) trees were mainly caused by chloroplast capture. In Selineae, some branches (e.g., the branch Subclade I + C. sinensis and the branch Subclade II + S. divaricata) and even some species within the same branch differed in their phylogenetic placements in the phylogenies based on plastome CDS and nrDNA (ITS + ETS) sequences (Figure 1). Similar to the three main clades (Tordyliinae, Coriandreae, and Hymenidium Clade), the incongruent positions of the four branches (i.e., "Peucedanum sensu stricto group", "Angelica group", Subclade I + Carlesia, and Subclade II + Saposhnikovia) between the plastome CDS and nrDNA (ITS + ETS) trees may also have been generated by early chloroplast capture events due to hybridization. Our phylogenetic results supported a chloroplast capture hypothesis that the early lineage of the branch Subclade I + Carlesia may have successively captured the chloroplast genome of ancestors of the "Angelica group". The strongly supported polytomy of Subclade II in the nrDNA (ITS + ETS) tree suggested the insufficiency of phylogenetic informative characters in the nrDNA dataset, which is also thought to contribute to the incongruent topologies between plastid and nuclear gene trees [77,78]. The lack of informative characters is usually derived from rapid radiation and hybridization [79,80]. In our study, hybridization is not a reasonable explanation for the polytomy of Subclade II because the four species within Subclade II included two stenochoric species (K. formosana and P. ampliatum) with non-overlapping distributions [12,70]. Rapid radiation may be the most likely mechanism for the polytomy of Subclade II in the nrDNA (ITS + ETS) tree. The hairy stems and fruits and numerous mericarp vittae may be synapomorphies of the four species within Subclade II (Table S4), and they may have undergone rapid morphological differentiation by a founder effect [81], which can coincide with rapid radiation.

Taxon Sampling
Fresh green leaves from adult plants of seven taxa, including five Kitagawia species (K. baicalensis, K. komarovii, K. praeruptora, K. stepposa, and K. terebinthacea), C. sinensis, and S. divaricata were collected from the wild, and then immediately dried with silica gel to preserve them for DNA extraction. Special permission was not required to collect these materials because they are not key-protected plants. In addition, we requested and obtained the genomic DNA of K. formosana from the Herbarium of the Institute of Botany (PE), Chinese Academy of Sciences. The formal identification of all samples was undertaken by Professor Xingjin He (Sichuan University). The voucher information of the eight samples is summarized in Table S4.

DNA Extraction, Sequencing, Assembly, and Annotation
The total genomic DNA of the eight taxa was extracted from silica gel-dried leaves or herbarium specimens with a modified CTAB protocol [82]. The primers ITS4 (5 -TCC TCC GCT TAT TGA TAT GC-3 ) and ITS5 (5 -GGA AGT AAA AGT CGT AAC AAG G-3 ) were used in the polymerase chain reaction (PCR) amplification of the complete ITS region [83]. The ETS sequences were amplified with primers 18S-ETS (5 -ACT TAC ACA TGC ATG GCT TAA TCT-3 ) and Umb-ETS (5 -GCG CAT GAG TGG TGA WTK GTA-3 ) [84,85]. The 30-µL PCR reactions contained 2 µL extracted total genomic DNA, 10 µL ddH 2 O, 1.5 µL each of 10 pmol µL −1 forward and reverse primers, and 15 µL Taq MasterMix (CWBio, Beijing, China). The PCR program of ITS and ETS regions started with an initial denaturation at 94 • C for 4 min, followed by 30 cycles of denaturation at 94 • C for 45 s, annealing at 54 • C for 45 s, and extension at 72 • C for 1 min, with a final extension at 72 • C for 10 min. All PCR products were separated on a 1.5% (w v −1 ) agarose TAE gel and sent to Sangon (Shanghai, China) for sequencing. The newly sequenced ITS and ETS sequences of the eight taxa were examined and edited with Geneious v9.0.2 [86], and consensus sequences were obtained separately.

Phylogenetic Analyses
After preliminary analyses, 28 species from Selineae, Tordyliinae, Coriandreae, Hymenidium Clade, Apieae, Careae, Pyramidoptereae, and Acronema Clade were selected to conduct the phylogenetic reconstruction. The type species of Peucedanum sensu stricto and Angelica, P. officinale and A. sylvestris, were included due to the blurred boundaries and potential expansion of Kitagawia [10]. Similarly, P. ampliatum and P. harry-smithii var. grande were added to examine the possible candidates for Kitagawia according to Pimenov and Ostroumova [17]. Pternopetalum davidii Franch. and Pternopetalum vulgare (Dunn) Hand.-Mazz., which belong to Acronema Clade, were chosen as the outgroup to root the phylogenetic tree, based on the results of Downie et al. [30]. The names of the main clades refer to the contributions of Downie et al. [30] and Gou et al. [51].
Two datasets were constructed for the phylogenetic analyses. The CDS dataset consisted of single-copy CDS sequences from complete plastomes of 27 taxa (6 species from Kitagawia and 21 species from other genera in the above-mentioned clades). To provide greater branch support [85], the nrDNA dataset concatenated the complete ITS and ETS regions from 27 taxa of the same genera used in the CDS dataset. Peucedanum officinale L. and P. chujaense could not be added to the CDS dataset and nrDNA dataset, respectively, due to a lack of online data. All sequences used in phylogenetic analyses are available in GenBank (Table S5).
The 79 single-copy CDS sequences from 27 complete plastomes were extracted and connected using PhyloSuite v1.2.2 [91]. The nrDNA (ITS + ETS) sequences of 27 taxa were also concatenated with PhyloSuite v1.2.2 [91]. Sequences of the two datasets were aligned with MAFFT v7.221 [92] and then manually corrected with MEGA7 [93]. Maximum likelihood (ML) and Bayesian inference (BI) methods were adopted to infer phylogenetic relationships. RAxML v8.2.10 [94] was used to perform the ML analyses for the two datasets based on the GTRGAMMA model and 1000 rapid bootstrap replicates. MrBayes v3.2.7 [95] was used to perform the BI analyses with the best substitution model as determined by MrModeltest v2.4 [96]. The selected models for the CDS dataset and nrDNA dataset in BI analyses were GTR + I + G and GTR + G, respectively. Four independent Markov chains were run for 10,000,000 generations with random initial trees, sampling every 1000 generations. The first 25% of trees were discarded as burn-in, and the remaining trees were used to build a 50% majority-rule consensus tree. The phylogenetic trees were visualized and edited with FigTree v1.4.4 [97] and MEGA7 [93].
The expansion or contraction between IR border regions of the 14 plastomes of Kitagawia and related taxa were drawn by IRscope [99] and adjusted manually.

Fruit Anatomical and Micromorphological Characteristics
Mature fruits of five Kitagawia species (K. baicalensis, K. komarovii, K. praeruptora, K. stepposa, and K. terebinthacea) were collected from the wild and preserved in formaldehydeacetic acid-alcohol (FAA) for morphological examination. Fruit anatomy was evaluated by paraffin section. For each of the five species, the fruits of at least three individuals from the same population were examined. All fruits were treated following the method of Feder and O'Brien [103] for embedding in glycol methacrylate (GMA). A Leica RM2016 microtome was used to prepare transverse sections through the center of the mericarp, about 3-5 mm in thickness, and they were stained with toluidine blue. Sections were observed and photographed under an Olympus BX43 microscope. Morphological terminology followed Kljuykov et al. [104].
Fruit micromorphology was examined by scanning electron microscope (SEM). After dehydration using graded ethanol, all mature fruits were directly mounted on clean aluminum stubs with conducting carbon adhesive tabs, coated, and then scanned with a JSM-7500F scanning electron microscope. The micromorphological characteristics studied and micromorphological terminology follow the contributions of Ostroumova [23,24].

General Morphological Characteristics
We examined the general morphological characteristics of the 14 species used in the above comparative plastome analyses together with P. officinale, the type species of Peucedanum sensu stricto. The characteristics examined included features of the stem, leaf, bract, bracteole, ray, flower, and fruit. Data from the five Kitagawia species (K. baicalensis, K. komarovii, K. praeruptora, K. stepposa, and K. terebinthacea) were obtained mainly from our field-based observations and anatomy research. For some taxa and characteristics, data were obtained directly from previous literature [11,12,15,17,[66][67][68][71][72][73].

Conclusions
In this study, we newly sequenced complete plastomes of eight species from Kitagawia, Carlesia, and Saposhnikovia. The phylogeny reconstruction of Kitagawia was performed based on plastome CDS and nrDNA (ITS + ETS) sequences. Our results revealed that Kitagawia is a member of the tribe Selineae but is divided into two branches (Subclade I and Subclade II). The distant phylogenetic relationships between Kitagawia and Peucedanum sensu stricto verified the rationale behind the separation of Kitagawia. The visible differences in JLA and several highly variable regions added some support to the distinct phylogenetic position and non-monophyly of Kitagawia. Similarly, fruit anatomical and micromorphological characteristics, as well as general morphological characteristics, distinguished Subclade I from Subclade II and other related taxa, further supporting the results of phylogenetic analyses. These findings demonstrated that P. ampliatum and P. harry-smithii var. grande should not be included in Kitagawia and suggested that K. praeruptora and K. formosana should be placed in a new genus and that P. hakuunense and P. chujaense probably belong to Kitagawia. We believe that the "core" Kitagawia should be limited to Subclade I, and this genus can be distinguished by the association of a series of morphological characteristics. In short, our study confirmed the distinct generic status of Kitagawia and clarified the morphological delimitations between Kitagawia and related taxa, providing a foundation for further taxonomic and evolutionary research on Kitagawia.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/plants11233275/s1, Figure S1: Gene maps of the eight plastomes of Kitagawia, Carlesia, and Saposhnikovia; Table S1: List of genes identified in the 14 plastomes of Kitagawia and related taxa; Table S2: Number of all identified SSRs in the 14 plastomes; Figure  S2: Mauve alignment of the 14 plastomes of Kitagawia and related taxa; Table S3: Synopsis of general morphological characteristics from the 15 species involved in this study; Table S4: Collection locality and voucher information are provided for the eight sequenced plastomes; Table S5: GenBank accession numbers of DNA sequences used in this study.  Data Availability Statement: The annotated plastomes, ITS, and ETS sequences of the eight taxa studied have been submitted to NCBI (https://www.ncbi.nlm.nih.gov (accessed on 7 September 2022)) with accession numbers that can be found in Table S5.