Phylogenetic position of Ligusticopsis (Apiaceae, Apioideae): evidence from molecular data and carpological characters

Abstract Ligusticopsis (Apiaceae, Apioideae) is now considered to have an East-Asia and Sino–Himalaya distribution. The genus was not recognized as a natural and separate genus and was treated as a synonym of Ligusticum both in Flora Reipublicae Popularis Sinicae and Flora of China since first established, though Pimenov et al. have made many taxonomic revisions to Ligusticopsis, phylogenetic relationships between Ligusticopsis and Ligusticum have been in dispute. Thirty-four plastomes and 35 nrITS from Apioideae were analysed by RAxML and MrBayes to reconstruct the phylogenetic relationships, along with carpology of 10 species and comparative analyses of 17 plastomes to investigate the evidence supporting the independence of Ligusticopsis. As a result, nine species suggested to be Ligusticopsis formed a highly supported monophyletic branch (Subclade A) inside Selineae both in maximum likelihood and Bayesian inference; the results of the comparative analyses further supported the monophyly of Subclade A, mainly in the location of genes at the IRa/LSC boundary, the sequence diversity exhibited by various genes (e.g. trnH-GUG–psbA and ycf2) and same codon biases in terminator TAA (relative synonymous codon usage = 1.75). Species in Subclade A also had shared characters in mericarps, combined with other characters of the plant, ‘base clothed in fibrous remnant sheaths, pinnate bracts, pinnate bracteoles longer than rays of umbellule, mericarps strongly compressed dorsally, median and lateral ribs filiform or keeled, marginal ribs winged, and numerous vittae in commissure and each furrow’ should be the most important and diagnostic characters of Ligusticopsis. Our phylogenetic trees and other analyses supported the previous taxonomic treatments of Pimenov et al. that Ligusticopsis should be a natural and separate genus rather than a synonym of Ligusticum.


Introduction
Ligusticopsis was established by Leute in 1969. By researching the carpology of several Ligusticum species from high elevations in China, Leute found that most of these species had mericarp types that clearly corresponded to the 'Ligusticeen-Typ' but differed from the known types. The new mericarp type was 'mericarps strongly compressed dorsally, without mechanical cells arranged in groups or rings', and these differences in mericarps could be used to distinguish these species from Selinum by one vitta per furrow and from Cortia by groups of mechanical cells in the rib bases. Leute considered that there was sufficient reason to establish a new genus, and then, the genus Ligusticopsis, containing 14 species from China, was established with Ligusticopsis rechingeriana (Fig.  1) being selected as the type species. An evolutionary hypothesis for the genus was proposed alongside its classification and concluded that Ligusticopsis is a relatively ancient genus (Leute 1969). The main features of the genus Ligusticopsis were described by Leute as 'strongly dorsally flattened mericarps, vallecular vittae numerous, calyx teeth presence'. In 1970, Leute compared Ligusticum with the new genus and found that the most important difference between these two was that the genus Ligusticopsis has conspicuous calyx teeth (Leute 1970).
The taxonomy of Ligusticopsis (Apiaceae subfamily Apioideae) has been controversial as many botanists did not recognize Ligusticopsis as a separate genus because of no clear morphological delimitations between Ligusticopsis and the nearby genera (Zhou et al. 2008a(Zhou et al. , 2009Sun and He 2010). Additionally, several species included in Ligusticopsis by Leute do not have conspicuous calyx teeth, which blurred the boundaries of the genus (Leute 1970). Studies, including Flora Reipublicae Popularis Sinicae and Flora of China, have treated Ligusticopsis as the synonym of Ligusticum, and Flora of China treated 'Ligusticum in the broad sense' (including accepted species in Flora of China named by Ligusticum) as an artificial assemblage (Hz 1979;Pu 1991;Pu and Watson 2005;Sun et al. 2008).
With the progress of molecular phylogeny, a series of studies have shown that neither the genera Ligusticum nor Ligusticopsis is monophyletic (Downie et al. 2010). The latest phylogenetic research using nrDNA sequences (nrITS) has shown that species from genus Ligusticum were located in six different clades, while several species that previously belonged to Ligusticopsis clustered with other species from nearby genera (e.g. Cortiella, Ligusticum, Pachypleurum and Selinum) and formed the 'Chinese Ligusticum Clade' inside Selineae, but the support between genera and some species was not strong enough to tell from each other. (Valiejo-Roman et al. 2006;Zhou et al. 2008aZhou et al. , 2009Zhou et al. , 2020. The carpology and mericarp anatomical characters of the taxa from the Apiaceae are of considerable taxonomic and evolutionary importance, which have been increasingly used in conjunction with molecular phylogenetic analyses to indicate stable taxonomy and evolutionary history between species, genera and even within the entire family (Liu et al. 2009;Wen et al. 2020;Xiao et al. 2021). Pimenov et al. supported the establishment of Ligusticopsis and made several taxonomic revisions of species in Ligusticopsis mainly based on carpoanatomical results. Selinum wallichianum was placed into Ligusticopsis by Pimenov et al. (2001), and three species (Ligusticopsis acuminata, Ligusticopsis tenuisecta and Ligusticopsis pteridophylla) previously belonging to Ligusticopsis were transferred to Conioselinum by Pimenov et al. (2003) (Pimenov et al. 2001). Recently Pimenov (2017) summarized the nomenclatural combinations of Ligusticopsis in China based on reviews of the type specimens and morphological evidence, 18 species (four were newly proposed, and 13 were endemic to China) were recognized, all of which were distributed in the Hengduan Mountains and Qinghai-Tibet Plateau regions of China, while one species published in 2001 was outside China (L. coniifolia, West Himalaya, India and Pakistan). And this checklist also recognized that Ligusticopsis angelicifolia should be transferred to Angelica, while Ligusticum hispidum and Ligusticum involucratum should be placed into Ligusticopsis (Pimenov 2017). Overall, the distribution pattern of these 19 species is from the Hengduan Mountains of China towards the Himalayas (including parts of Qinghai-Tibet Plateau) and up to the western and southern sides (Bhutan, India, Myanmar, Nepal and Pakistan) of the Himalayas.
Several species (e.g. Ligusticum sinense in SW China, Ligusticopsis brachyloba in Yunnan Province) from 'Ligusticum in the broad sense' are used as herbs in traditional Chinese medicine, but many are labelled with the same Chinese drug name 'Gao-Ben' (Li et al. 2001;Pu 1991). These types of herbs are believed to have medicinal value, particularly anthelmintic properties. The extraction of anthelmintic active substances and coumarins has been reported in species transferred into genus Ligusticopsis (Kondo et al. 2008;Qi et al. 2020). Therefore, unclear taxonomy can make scientific investigation of any medical properties difficult. Similarly, the regulated use of Chinese traditional herbs and a search for alternatives can be problematic without clear taxonomy and phylogeny, with molecular authentication of Ligusticum species based on nuclear DNA (nrITS2) being reported as a method for determining medically valuable species from other 'Gao-Ben' labelled specimens (Liu et al. 2019).
In general, clarification of phylogenetic relationships between genera Ligusticum and Ligusticopsis is necessary. The fact that both Ligusticopsis and Ligusticum are Morphological characters of transverse sections and mericarp, terminologies followed Kljuykov et al. (2004), co = commissure; cv = commissural vittae; lr = lateral rib; mar = marginal ribs; mer = median rib; s = seed; vv = vallecular vittae. polyphyletic, and the conflicting views of different botanists on the validity or otherwise of Ligusticopsis, implies that the subsumption of Ligusticopsis into Ligusticum should be approached with caution. Low-resolution phylogenetic relationships based on nuclear gene regions required additional DNA data sets to clarify. Plastomes (i.e. plastid genomes) are one of the good choices, as there have been many reports on the use of plastomes and comparative analyses to construct and resolve the phylogenetic relationships of genera under Apiaceae subfamily Apioideae (Gou et al. 2020;Guo et al. 2020), while latest study has used plastomes to indicate backbone phylogeny and evolution of subfamily Apioideae with good contributions (Wen et al. 2021). What is more, it is reasonable to use the mapping of carpological to phylogenetic results in finding diagnostic characters (Yu et al. 2011).
Plastids are important organelles in plants, whose genome DNA (plastome) is usually more conserved than the mitochondrial and nuclear genomes (Marechal and Brisson 2010). Most angiosperm plastomes have a quadripartite circular DNA organization with two copies of inverted repeat (IR) regions, a large single copy (LSC) region and a small single region (SSC) (Wicke et al. 2011). Plastomes are maternally inherited, have a high conservation of gene content and genome structure and have more variable sites, making plastomes a significant resource for phylogenetic reconstruction and complicated evolutionary relationships resolution (Ingvarsson et al. 2003;Kang et al. 2019;Xu et al. 2021), and have been widely used for phylogenetic analyses in angiosperms (Jose et al. 2015;Wu et al. 2020).
In this study, with the addition of seven new plastomes and 11 nrITS sequences, 34 plastomes and 35 nrITS sequences from Apiaceae subfamily Apioideae were analysed, combining phylogenetic results with carpology and comparative analyses of plastomes. We aimed to (i) reconstruct the phylogenetic relationships of Ligusticopsis and investigate the rationale for separating Ligusticopsis from Ligusticum; (ii) indicate evidence from plastomes and carpology in support of phylogenetic relationships and find the diagnostic characters; (iii) investigate the previous taxonomic treatments.
DNA extraction, sequencing, assembly and annotation ITS sequencing was required for the above 11 species. Total genomic DNA was extracted from basal leaf materials using the modified CTAB procedure (Doyle 1987). Amplifications were conducted using 2 μL extracted total genomic DNA, 10 μL ddH 2 O, 1.5 μL of 10 pmol μL −1 forward primers, 1.5 μL of 10 pmol μL −1 reverse primers and 15 μL Taq MasterMix (CWBio, Beijing, China). The nrITS sequences were amplified with primers ITS-4 (5ʹ-TCC TCC GCT TAT TGA TAT GC-3ʹ) and ITS-5 (5ʹ-GGA AGT AAA AGT CGT AAC AAG G-3ʹ) (White et al. 1990). Then, PCR cycling profile included a denaturing step at 94 °C for 4 min, followed by 30 cycles of 45 s at 94 °C, annealing at 54 °C for 45 s and extension at 72 °C for 1 min, with a final extension for 10 min at 72 °C. All PCR products were separated using a 1.5 % (w v −1 ) agarose TAE gel and delegated Sangon (Shanghai, China) for sequencing. DNAstar-SeqMan (Burland 2000) was used to edit the newly sequenced DNA and obtain consensus sequences.
Plastomes of seven newly collected species need to be sequenced, which were L. acuminata, L. angelicifolia, L. brachyloba, L. modesta, L. rechingeriana, L. tenuisecta and S. wallichianum. We provided 20 µL of total genomic DNA per species for the sequencing process, and total genomic DNA was sequenced using an Illumina Novaseq 6000 platform (Illumina, San Diego, CA, USA) by Personalbio (Shanghai, China). Libraries were constructed with an average length of 350 bp, and the average length of generated reads was 150 bp. FastQ v0.19.7 (Chen et al. 2018) was used for quality control of the raw reads, and at least 5 GB clean reads per species were yielded. Then, the remaining clean data were assembled using NOVOPlasty v4.2.1 (Nicolas et al. 2020) with K-mer 39, where rbcL gene of L. involucratum (GenBank accession no. NC049054) was used as the seed input. The assembled plastomes were annotated via Plastid Genome Annotator (Qu et al. 2019) with the sequences and annotations of L. involucratum and Melanosciadium pimpinelloideum (GenBank accession no. MN810920) as the reference. The annotation results were then revised manually according to the other plastomes released on GenBank (L. integrifolia, NC049055; L. pteridophylla, NC049056; L. hispidum, NC049053) via Generous v9.0.2 (Drummond 2012). The physical maps of seven new plastomes were generated using OGDraw v1.3.1 (Marc et al. 2019).
These seven new annotated plastomes and 11 nrITS sequences data have been submitted to the GenBank under accession numbers MZ491174-MZ491177, and OL547614-OL547616 for plastomes sequences while MZ497218-MZ497222, MZ505394 and OL600820-OL600824 for nrITS sequences. Of 34 plastome sequences and 35 nrITS sequences used in this study, 27 plastome sequences and 31 nrITS sequences were from our lab, and all the sequences are available in GenBank [see Supporting Information- Table S1].
The name of the main tribes refers to contributions of Downie et al. (2010) and Gou et al. (2021).
The gene spacer regions and single-copy coding sequences (CDS) from 34 complete plastomes (10 Ligusticopsis plastomes and 24 plastomes from subfamily Apioideae) were extracted and connected, respectively, using Generous v9.0.2 and PhyloSuite v1.2.2. The data sets of plastome CDS, gene spacer regions and 35 nrITS from subfamily Apioideae were used to conduct the phylogenetic analysis, respectively. Data sets were aligned using MAFFT v7.402 (Standley 2013). RAxML v7.2.8 (Stamatakis 2014) with GTR+G was used for maximum likelihood (ML) analyses, and the best-fit model was selected by ModelFinder (Kalyaanamoorthy et al. 2017) with 1000 bootstrap replicates. And Bayesian inference (BI) was performed in MrBayes v.3.2.7a (Huelsenbeck 2012) under the GTR+G model. The Markov chain Monte Carlo algorithm was performed for 10 000 000 generations and trees were sampled every 1000 generations for each data partition. The first 20 % of trees were discarded as burn-in and the remaining trees were used to build a 50 % majority-rule consensus tree.
IRscope (Amiryousefi et al. 2018) was used to indicate the boundaries between the IR and SC regions of 17 plastomes, and the results were used as references to visualize the final view after manual revision. Sequence divergence of 17 whole plastomes was performed using the mVISTA (Frazer et al. 2004) software in Shuffle-LAGAN mode, with L. rechingeriana as the reference, and L. thomsonii was chosen as the X-axis to indicate sequence divergence globally.

Morphology observations of mericarps
Mericarps of 10 species were photographed using a stereomicroscope (SMZ25, Nikon Corp., Tokyo, Japan). The structures and details for observation were dorsal side views, commissural side views, transverse section views, mericarp shape, calyx teeth, rib shape and vittae. And 10 randomly selected mature mericarps per species for morphological measurements using KaryoType (Altinordu et al. 2016). Mericarp terminology followed Kljuykov et al. (2004).

Results
The plastomes of seven new species The seven new complete plastome sequences ranged from 148 005 bp (L. acuminata) to 163 810 bp (L. angelicifolia) in length ( Under the uniform annotation rules, except for L. angelicifolia, the remaining six new plastomes contained 128 genes with 84 protein-coding genes (PCGs), 36 transfer RNA genes (tRNAs), eight ribosomal RNA genes (rRNAs) and two genes identified as pseudogenes (ψycf1 and ψycf2). And L. angelicifolia plastome contains 143 genes with 98 PCGs, 37 tRNAs, eight rRNAs and one gene identified as a pseudogene (ψycf1). The length, GC content and gene components of the seven new plastomes were summarized in Table 1 [see  Supporting Information-Table S2].

Phylogenetic analyses
Single-copy CDS sequences of 34 plastomes were used to execute the phylogenetic analyses. There were seven major tribes, of which the Ligusticopsis species treated by Leute (1969) Fig. 2A) with strong support (BS = 100 %, PP = 1.00), and all species located in this branch were treated as 'true Ligusticopsis' by Pimenov. Inside Subclade A, L. brachyloba allied with S. wallichianum forming a branch (BS = 100 %, PP = 1.00) that was sister to seven other species. Ligusticopsis rechingerana was sister to L. involucratum and allied with L. hispidum forming a branch with strong support (BS = 100 %, PP = 1.00), and this branch was sister to the other four species with moderate support in ML (BS = 72 %) but in BI the support was rather weak (PP = 0.55). Ligusticopsis capillacea was sister to L. scapiformis and L. integrifolia was also sister to L. modesta, these two sister branches had strong support (BS = 100 %, PP = 1.00), respectively. We also found that L. acuminata, L. tenuisecta and L. pteridophylla were in Hymenidium Clade. Ligusticopsis acuminata and L. tenuisecta were sister species and clustered with three Ligusticum species formed a clade (Subclade B) with strong support (BS = 100 %, PP = 1.00). Inside Subclade B, L. sinense and L. jeholense were sister species (BS = 100 %, PP = 1.00) and this branch became a sister branch (BS = 100 %, PP = 1.00) to L. acuminata and L. tenuisecta, while L. tenuissimum was sister (BS = 100 %, PP = 1.00) to the four species above. Ligusticopsis pteridophylla belonged to Hymenidium Clade in its own branch and it was identified as a sister (BS = 100 %, PP = 1.00) to other species in Hymenidium Clade and Selineae. Ligusticum delavayi was a sister species to M. yunnanensis and located in Acronema Clade with strong support (BS = 100 %, PP = 1.00), which was one of the outgroups for Hymenidium Clade and Selineae. A similar topology occurred in the phylogenetic tree based on 35 nrITS from Apiaceae subfamily Apioideae but some species had a different phylogenetic position, and some branches had weak support (Fig. 2B). Ligusticopsis pteridophylla was no longer sister to the Hymenidium Clade and Selineae but formed a sister branch (BS = 94 %, PP = 1.00) to L. sinense, and L. jeholense inside the Hymenidium Clade (Subclade B, Fig. 2B). Ligusticum thomsonii did not cluster with Angelica species but was a sister to Subclade A with strong support (BS = 92 %, PP = 1.00). Inside Subclade A, L. modesta was still clustered with L. integrifolia but with very weak support (BS = 35 %), L. hispidum here was a sister to L. modesta and L. integrifolia but also had weak support (BS = 33 %), and L. involucratum was sister to these three species with strong support (BS = 96 %, PP = 0.91). Weak support (BS = 57 %) occurred on the branches between L. capillacea and the other four species (L. involucratum, L. hispidum, L. modesta and L. integrifolia), and then L. rechingerana was sister to these five species with strong support (BS = 87 %, PP = 0.93). Ligusticopsis scapiformis was sister to six species above with strong support (BS = 93 %, PP = 1.00). As for the phylogenetic tree of plastome gene spacer regions, the topological differences were mainly in the positions of L. thomsonii and L. pteridophylla as well as species within Subclade A [see Supporting Information- Fig. S3]. Ligusticum thomsonii was the sister species to 16 other species from Selineae with strong support (BS = 100 %, PP = 1.00). Ligusticopsis pteridophylla was no longer a sister species to Selineae and Hymenidium Clade but allied and formed sister branches (BS = 78 %, PP = 0.85) with L. acuminata, L. tenuisecta and three other Ligusticum species inside Subclade B. Within Subclade A, L. integrifolia allied with L. scapiformis and formed a sister branch to L. modesta with weak support. The branch including L. brachyloba and S. wallichianum here was more related to L. integrifolia, L. scapiformis, L. modesta and L. capillacea with strong support (BS = 88 %, PP = 0.91).

Comparative plastome analyses
Seventeen plastomes were compared to observe the gene distribution at their IR boundaries (Fig. 3), while the pseudogenes were visualized with reference to previous studies (Downie and Jansen 2015;Guo et al. 2020). The major differences between Subclade A and Subclade B, as well as the other three plastomes, were the gene distribution at IRa/LSC borders (JLA line) and LSC/IRb borders (JLB line). The trnH gene region was located near JLA line, and the distance from trnH gene to JLA line was the same (6 bp) in Subclade A, while this distance was longer and varied in Subclade B and three other plastomes. Except for L. angelicifolia and L. tenuissimum, a similar structure was identified across the remaining 15 plastomes at JLB line and JLA line, with some variations. JLB line extended 576-585 bp into the ycf2 gene in Subclade A, which was shorter than the corresponding region of Subclade B (659-701 bp), L. pteridophylla (701 bp) and L. thomsonii (694 bp). The distance from trnL gene to JLA line in Subclade A was longer (1809-2177 bp) than the corresponding region of Subclade B (1331-1345 bp), L. pteridophylla (1045 bp) and L. thomsonii (1033 bp). Ligusticopsis angelicifolia has the longest IR region such that the IRb region expanded towards the LSC region to the location of the petB gene; thus, JLB line extended 1084 bp into petB gene. The IRb region of L. tenuissimum also showed expansion towards the LSC region, but the JLB line only expands to the location of the rpl22 gene, and rpl22 gene spanned JLB line with 2 bp of extensions in IRb region.
We extracted 53 CDSs from each of the 17 plastomes for the codon usage analysis. These CDSs encoded 21 585-21 749 codons and leucine (Leu) had the highest number of codons while cysteine (Cys) had the least (Fig. 4). The codons ATG of methionine (Met) and the codons TGG of tryptophan (Trp) had an RSCU value of 1.0 across all plastomes, with 1.0 of RSCU value representing no bias of codon usage. Thirty codons had RSCU values > 1.0 and all had A/T at the third position except TTG of Leu. Among three types of terminator codons, Subclade A had the same codon biases in TAA (RSCU = 1.75), which was higher than Subclade B and the other three plastomes. It was the major difference between Subclade A and other plastomes, while the other two types of terminator codons (TAG and TGA) had RSCU < 1.0 and varied among plastomes. The other 61 types of codons had similar RSCU values in different species (Fig. 4) [see Supporting Information- Table S3].
The mVISTA software was used to compare the complete plastomes of 17 species. More sequence conservation occurred in coding regions than non-coding regions. Within Subclade A, there was a high degree of similarity in some highly divergent regions (e.g. trnH-psbA, ycf1, ycf2, rpoC2, rpl32, psaB, ndhF), but the corresponding regions were significantly different when compared to Subclade B and other three plastomes [see Supporting Information- Fig. S4].

Discussion
In this study, we reconstructed the phylogeny of Ligusticopsis and related taxa based on nrITS and plastomes; nine species previously suggested to be 'true Ligusticopsis' by Pimenov formed a strongly supported monophyletic branch (Subclade A, type species L. rechingerana included). The commonalities in carpological characters and the structural consistency in plastomes distinguished from species outside Subclade A and provided further support for the monophyly of the branch. These findings further supported Ligusticopsis should be a separate genus in Selineae, and supported the previous taxonomic treatments that L. hispidum, L. involucratum and S. wallichianum should be placed in Ligusticopsis, while L. acuminata, L. tenuisecta, L. pteridophylla and L. angelicifolia should be transferred out of Ligusticopsis.

Phylogenetic position of Ligusticopsis and morphological delimitations between related genera
The results confirmed that nine species (including the type species, i.e. L. rechingerana) treated as Ligusticopsis by Pimenov (2017) formed a monophyletic branch (i.e. Subclade A) within Selineae. A natural and independent genus should be monophyletic and stable (Funk 1985;Linder and Humphreys 2009;Gomes-Da-Silva and Souza-Chies 2018). Here we agree that treating Ligusticopsis as a natural and independent genus within Selineae is justified because (i) the phylogenetic position of the type species (L. rechingeriana) is within Selineae, differing from the phylogenetic position (in Acronema Clade) of the type species (Ligusticum scoticum, i.e. true 'Ligusticum') of Ligusticum (Downie et al. 2010;Zhou et al. 2020), and (ii) morphological delimitations between Ligusticopsis and 'Ligusticum in the broad sense' objectively exist.

8
AoB PLANTS, 2022, Vol. 14, No. 2 A (i.e. Ligusticopsis branch) hold the shared characters 'base clothed in fibrous remnant sheaths, pinnate bracts, pinnate bracteoles longer than rays of umbellule, subequal rays, obcordate petals, apex notched with incurved apical lobule, calyx teeth presence, mericarps strongly compressed dorsally, median and lateral ribs filiform or keeled, marginal ribs winged, and numerous vittae in commissure (4-8) and each furrow (1-4, not solitary)'. These shared characters can be well distinguished from 'base without fibrous remnant sheaths, linear bracts and bracteoles, calyx teeth absent, mericarps slightly compressed dorsally' possessed by Subclade B and L. pteridophylla. Shared characters can also be distinguished from 'base without fibrous remnant sheaths, linear bracteoles, calyx teeth absent, linear purple petals, mericarps slightly compressed dorsally, median and lateral ribs narrowly winged' possessed by L. angelicifolia further support for Ligusticopsis to be a separate genus. What is more, conspicuous calyx teeth were the main boundary between Ligusticopsis and Ligusticum according to Leute, but this character was not observed clearly enough on the specimens and fieldwork. Through our research, the combination of mericarp, bract and bracteole morphologies can be more clearly distinguished between Ligusticopsis and 'Ligusticum in the broad sense', which is also consistent with Leute's view that the differences under genus Ligusticopsis come from a combination of a series of characters (Leute 1969). Based on these findings, the morphological delimitations between Ligusticopsis and 'Ligusticum in the broad sense' should be 'base clothed in fibrous remnant sheaths, pinnate bracts, pinnate bracteoles longer than rays of umbellule, subequal rays, obcordate petals and apex notched with incurved apical lobule, calyx teeth presence, mericarps strongly compressed dorsally, median and lateral ribs filiform, marginal ribs winged, and numerous vittae in commissure (4-8) and each furrow (1-4, not solitary)'. Zhou et al. (2020) determined that species from Ligusticopsis were related to Cortiella, Ligusticum, Oreocome and Selinum within the 'Chinese Ligusticum Clade' in Selineae ). However, a limited number of plastomes meant that a comprehensive evaluation of the phylogenetic relationships of these genera was difficult for our research. Based on previous reports on the mericarp morphology of these two genera (Cortiella and Oreocome), we believe that the mericarp type of Ligusticopsis is not a transitional form of these two genera, which indicates the objective existence of morphological divergence between Ligusticopsis and these two genera. The mericarp of Oreocome candollei (the type species of Oreocome) was described as median and lateral ribs all broadly winged and narrow mericarp commissure. And the mericarp of Cortiella hookeri (the type species of Cortiella) was described as dorsal (median and lateral) ribs broadly winged, often convoluted and crowded when mature, and had vittae two in commissure and one in each furrow, while the most unusual external characters of C. hookeri were 'acaulescent caulescent, forming compact rosettes closely appressed to soil surface'. The morphology described above can be clearly distinguished from Ligusticopsis, and the monophyletic relationships exhibited by the nrITS sequences in previous studies, we believe, were most likely due to hybridization, as mixed growth of Ligusticopsis species (S. wallichianum) and O. candollei in the same habitat has also been reported (Norman 1937;Pimenov et al. 2001).
Ligusticum delavayi and M. yunnanensis all formed sister species with strong support (BS = 100 %, PP = 1.00) in Acronema Clade, occurring to the three different data sets-based phylogenetic trees. Recent studies have reported that L. delavayi allied to several genera (e.g. Meeboldia, Pleurospermum, Tilingia and Rupiphila) within Acronema Clade Gou et al. 2021), with complex phylogenetic relationships that requiring more extensive sampling as well as more in-depth studies. Therefore, L. delavayi was only one of the outgroups and will not be analysed temporarily in this study. More comprehensive studies on L. delavayi and Acronema Clade are in progress.
Evidence from comparative analyses to support the monophyly of Subclade A Consistency in the distribution of genes around JLA line, the same codon usage bias (RSCU = 1.75) in terminator TAA, and consistency in sequence divergence possessed by some regions (e.g. trnH-psbA, ycf1, ycf2, rpoC2, rpl32, psaB, ndhF), all the consistency possessed by Ligusticopsis branch (i.e. Subclade A) were different from Subclade B and L. angelicifolia. The trnH-psbA regions are highly variable and have been suggested to be used as universal barcodes for taxonomic identification (Kress and Erickson 2007;Dong et al. 2012). In sequence divergence analysis, the high degree of consistency shown within trnH-psbA regions supported the monophyly of Ligusticopsis branch and can be distinguished from Subclade B. This finding was also supported by 'the same distance (6 bp) from the trnH gene to the JLA line within Ligusticopsis branch' from IR boundary analyses, while in Subclade B this distance was much longer (7-944 bp). The location changes of genes trnL and trnH may be due to longer non-coding region fragments during the evolution history of Ligusticopsis; this kind of events (rearrangement for structural, frequent losses, and few gains) for genes of plastomes have been discovered in angiosperms (Millen 2001;Jansen 2011). What is more, the same codon usage bias in terminator TAA was the most significant difference between Ligusticopsis branch and Subclade B. All these findings gave support to the monophyly of Subclade A (true Ligusticopsis). In addition, the RSCU values of the other two terminator codons were different from each other between species, along with the high diversity of other regions, may provide additional information to clarify interspecific relationships (Dong et al. 2012).

Validated taxonomic treatments from previously contributions
We agree with the treatment by Pimenov et al. (2001Pimenov et al. ( , 2017 that three species (L. hispidum, L. involucratum and S. wallichianum) should be placed into Ligusticopsis (Pimenov et al. 2001). These three species clustered with Ligusticopsis branch both in plastome and nrITS phylogenetic trees and were consistent with Ligusticopsis branch (Subclade A) in the gene locations around JLA line and codon usage bias in TAA (RSCU = 1.75). Although L. hispidum held the rays extremely elongated, both it and other two species share characters (e.g. base clothed in fibrous remnant sheaths, pinnate bracteoles longer than rays of umbellule, pinnate bracts, mericarps strongly compressed dorsally, median and lateral ribs filiform or keeled and marginal ribs winged) consistent with Ligusticopsis. What is more, Pimenov et al. (2001) clearly described that the closest relative to S. wallichianum in mericarp anatomy was L. brachyloba. We confirmed this finding cause S. wallichianum was sister to L. brachyloba, and plastomes of these two species had some commonalities in the same length (5508 bp) of ycf1 gene and same distance (59 bp) from ndhF gene to IRb/SSC boundary, while the codon usage of three types of terminators was also the same (RSCU = 1.75 in TAA, RSCU = 0.68 in TAG and RSCU = 0.57 in TGA).
Three species, L. acuminata, L. tenuisecta and L. pteridophylla, were transferred into Conioselinum back in 2003 by Pimenov et al. through mericarp anatomy (Pimenov et al. 2003). This treatment was better supported by our results that these three species located outside Subclade A, while there three species have shared characters 'linear bracts and bracteoles, median and lateral ribs winged, marginal ribs more broadly winged, and calyx teeth obsolete' against macromorphological features of Ligusticopsis branch. The plastomes of these three species were similar to those of L. jeholense and L. sinense in terms of gene distribution at the IR boundaries and sequence diversity and were therefore distinguished from Ligusticopsis branch. Ligusticopsis pteridophylla was not a sister to Hymenidium Clade and Selineae anymore but allied with L. sinense, and L. jeholense both in nrITS tree and plastome gene spacer regions tree, while it has been reported to be tetraploid (Zhou et al. 2008b), its unique phylogenetic founding suggests that its evolutionary history may have been complex.

Interspecific relationships within Ligusticopsis
Topology incongruence between phylogenetic trees constructed based on three data sets (plastome CDSs, plastome gene spacer regions and nrITS) occurred mainly within the Ligusticopsis branch (Subclade A), which led to unsatisfactory interspecific relationships under genus Ligusticopsis. Inconsistencies imply that Ligusticopsis may have a complex history of evolution and differentiation.
Several plant lineages have been confirmed to hold inconsistencies in phylogenies between nrDNA and plastomes (Alejandro et al. 2016), with many contributions indicated different interspecific relationships may be due to hybridization and very different genetic pathways and mutation rates between plastomes and the nuclear DNA (Marechal and Brisson 2010;Pelser et al. 2010). And interspecific hybridization may cause chloroplast capture events during the rapid radiation of species (Rieseberg and Soltis 1991;Fehrer et al. 2007), which have also been reported in Apiaceae subfamily Apioideae (Wen et al. 2021). We believe interspecific hybridization is more reasonable as many Ligusticopsis species have very close or even the same localities, which provided geographical conditions for hybridization. Fieldwork and type specimens have also revealed large morphological differences between species within Ligusticopsis, including but not limited to the shape of basal leaves, the degree of hispid and the texture of bracteoles and bracts. For example, both L. modesta and L. integrifolia are in the same alpine meadows of Likiang Snow Range (type locality) and have very different external morphologies [see Supporting Information- Table S5], but their mericarps have considerable structural similarity. The hybridization may have caused chloroplast capture events between these two species, which led these two to be sister species in plastome tree, but the influence on nrITS was too weak to well support these two to be sister species in nrITS tree.
In addition, wind dispersal is one of the main pathways for the dispersal of Apiaceae (Wen et al. 2020). Mericarps strongly compressed dorsally, with weakened dorsal (median and lateral) ribs and winged marginal ribs are characters further adapting to wind dispersal, which may lead to a Sino-Himalaya distribution (from Hengduan Mountains, through the Qinghai-Tibet Plateau to the west of Himalaya) with high altitude of Ligusticopsis, and in the Qinghai-Tibet Plateau regions, natural hybrids have been recovered in many genera (Wu et al. 2021). All these imply during the rapid radiation of species, hybridization may mainly cause the topology differences between different data set-based phylogenetic trees but did not have a greater impact on the diversity of evolutionary.
In the future, the expansion of sampling coverage in Sino-Himalaya regions will help to clarify interspecific relationships under Ligusticopsis. Sampling and studying populations outside the type locality can also help to clarify interspecific relationships, as has been reported within Apiaceae subfamily Apioideae (Xiao et al. 2017). We suggest that attention be paid to transitional morphological characters that may occur in Ligusticopsis outside the type localities, which will help to understand the evolutionary history and trends during the rapid radiation of species in the Qinghai-Tibet Plateau.

Conclusion
In this study, the phylogenetic reconstruction of genus Ligusticopsis was undertaken with the addition of seven new plastomes and 11 new nrITS sequences. A total of 34 plastomes and 35 nrITS sequences were used for analyses; we found that the plastome structures and mericarp anatomical morphology of genus Ligusticopsis differed significantly from those of the 'Ligusticum in the broad sense', and that these differences matched well with the two kinds of phylogenetic trees (plastome and nrITS trees). Plastome structure mainly differs in gene distributions nearby IRa/LSC boundary, biases in the usage of the terminator TAA and sequence diversity within several gene regions (trnH-psbA, ycf1, ycf2, rpl32 and ndhF). The morphological delimitations between Ligusticopsis and 'Ligusticum in the broad sense' should be 'base clothed in fibrous remnant sheaths, pinnate bracts, pinnate bracteoles longer than rays of umbellule, obcordate petals and apex notched with incurved apical lobule, calyx teeth presence, mericarps strongly compressed dorsally, median and lateral ribs filiform or keeled, marginal ribs winged, and numerous vittae in commissure (4-8) and each furrow (1-4, not solitary)'. We support the taxonomic treatments by Pimenov that three species (L. acuminata, L. tenuisecta and L. pteridophylla) should be transferred from Ligusticopsis to Conioselinum and three species (L. hispidum, L. involucratum and S. wallichianum) should be placed in Ligusticopsis. The results of our molecular phylogenetic study show that genus Ligusticopsis is a completely independent and natural genus-level taxon, which should be separated from genus Ligusticum, and the boundary between genera is clear in morphological characteristics.

Supporting Information
The following additional information is available in the online version of this article- Figure S1. Plastome map of five Ligusticopsis and Selinum wallichianum. Genes shown outside and inside the black circle are transcribed in the clockwise and counterclockwise directions, respectively. Different colour boxes indicate different functional groups. LSC, large single copy; SSC, small single copy; IR, inverted repeat. Figure S2. Plastome map of Ligusticopsis angelicifolia. Genes shown outside and inside the black circle are transcribed in the clockwise and counterclockwise directions, respectively. Different colour boxes indicate different functional groups. LSC, large single copy; SSC, small single copy; IR, inverted repeat. Figure S3. Phylogenetic relationships of 34 Apioideae taxa based on gene spacer regions from plastomes. Tree constructed by Bayesian inference (BI) and maximum likelihood (ML) with the posterior probabilities of BI and the bootstrap values of ML above the branches, respectively, (*) represents maximum support in both two analyses. (#) represents those nodes not occurring in the BI strict consensus tree. Species from the treatments by Leute (1969) are marked in red, species suggested to be Ligusticopsis by Pimenov are marked in bold. Figure S4.  Table S1. Voucher information and GenBank accession numbers of DNA sequences used in this study. Newly sequenced plastomes and ITS are marked in bold. Table S2. List of genes encoded in Selinum wallichianum and six new Ligusticopsis plastomes. (a) to show duplicated genes, (b) to show duplicated genes only in Ligusticopsis angelicifolia, (ψ) shows pseudogenes. Table S3. Codon usage and relative synonymous codon usage (RSCU) values of 53 protein-coding genes of 17 plastomes. Table S4. Synopsis of the carpological data of 10 species. Table S5. Synopsis of the morphological data from 17 species in Selineae and Hymenidium Clade involved in this study. Some data from type specimens and Flora of China.