Background

Libanotis Haller ex Zinn, nom. cons., belonging to the tribe Selineae of the family Apiaceae, includes approximately 30 species distributed throughout Eurasia, with 19 species found in China [1,2,3,4,5,6,7,8,9,10,11,12]. Libanotis as an independent genus was supported by de Candolle [13], Schischkin [14], Korovin [15], Rechinger [16], Fu [17, 18], Shan, Watson and Sheh [1, 2, 11, 19]. They thought conspicuous calyx teeth, separated bracteoles, and hairy mericarps easily distinguished Libanotis from Seseli. But the genus then has been suggested to merge into Seseli L. to establish broad sense Seseli genus by Drude [20], Ball [21], Kljuykov and Pimenov [1, 22,23,24,25,26], because they think the above diagnostic features are not sufficient to distinguish them. The views of the above taxonomists are all based on morphology, and in the Chinese taxa the taxonomists are equally sharply divided between these two schools of thought, and some taxonomists all agree that Libanotis should be retained rather than merged [1,2,3, 12, 17,18,19, 27]. By 2015, new Libanotis taxa (L. laoshanensis W.Zhou & Q.X.Liu) were still being published [10]. Pimenov set aside the retention of the taxonomic status of Libanotis for this species untreated in the 2017 treatment of the Chinese Apiaceae taxa [24]. The above indicates that a thorough phylogenetic analysis of Libanotis is necessary. Regrettably, there has been no prior phylogenetic investigation conducted concerning this contentious genus Libanotis. Furthermore, all phylogenetic analyses have consistently demonstrated that the Seseli genus, in its broader sense, is polyphyletic, owing to the complex and perplexing variations in mericarps and vegetative body morphology. [26, 28,29,30,31,32,33,34,35,36,37]. Recently, Seseli s.s. was established and several phylogenetic studies using molecular fragments (nrITS and nrETS) robustly supported that L. sibirica C. A. Mey. (type species of Libanotis, L. montana Crantz ≡ Seseli libanotis W.D.J.Koch = L. sibirica) did not cluster with Seseli tortuosum L. (type species of Seseli) into a monophyletic branch [32, 33]. Hence, we believe that the taxonomic status of Libanotis needs to be discussed again, especially in China. Nevertheless, the delimitation of Libanotis genus still faced severely challenge. All previous phylogenetic studies showed that Libanotis was not a monophyletic group and members of this genus scattered in the Selineae tribe [6, 7, 35, 37, 38]. Due to limited sample and molecular fragments contained few informative loci, these studies all generated the phylogenetic framework with weak support and low resolution, which was insufficient to aid to the taxonomic revision of Libanotis members. Hence, it is imperative to establish a more comprehensive phylogenetic framework for Libanotis to address the controversy surrounding its evolutionary relationships and taxonomic status.

Due to the large number of species under the genus Libanotis, many taxonomists have established sections under the genus. Among the opinions in favor of the independence of Libanotis, de Candolle [13] was the first to group them, arguing that Libanotis could be divided into two sections, Sect. Eriotis and Sect. Eulibanotis: in which Sect. Eriotis are “Petals covered with short fascicular hairs on the outside leaves coriaceous, thickish, shiny.”; and Sect. Eulibanotis are “Petals dorsally glabrous or with sparse simple short hairs, leaves not coriaceous, not shiny.” After a series of species transfers and the publication of new species, Schischkin [14] added two sections: Sect. Pseudolibanotis and Sect. Schultziopsis, Sect. Pseudolibanotis were described by “Main stem not developed, the root neck bearing slightly leafy, sometimes nearly leafless shoots which spread along ground or ascend.” And the trait of Sect. Schultziopsis are special, their subcapitate umbel surrounded by the rounded sheaths of terminal leaves. The four-sections system was widely accepted by taxonomists that supported the independence of Libanotis [5, 15]. L. monstrosa (Willd.) DC., the only species in Sect. Schultziopsis, has been used as the type species for the establishment of the new monotypic genus Sajanella Soják [39] and is therefore excluded from this study, while the other all three sections are included for phylogenetic analysis (Fig. 1, Table 1). Based on these section characteristics, the newly published species L. jinanensis [3] and the newly transferred species L. grubovii [19] could be included in Sect. Eriotis according to their descriptions (Table 1, Table S10).

Fig. 1
figure 1

Some flowering Libanotis species with diverse morphology. A, B- Sect. Pseudolibanotis (A) L. depressa R.H.Shan & M.L.Sheh (B) L. acaulis R.H.Shan & M.L.Sheh C, F, G, H, I- Sect. Eriotis (C) L. iliensis (Lipsky) Korovin (F) L. buchtormensis DC. (G) L. grubovii (V.M.Vinogr. & Sanchir) M.L.Sheh & M.F.Watson (H) L. lanzhouensis K.T.Fu ex R.H.Shan & M.L.Sheh (I) L. spodotrichoma K.T.Fu. D, E- Sect. Libanotis (D) L. seseloides (Fisch. & C.A. Mey. ex Ledeb.) Turcz. (E) L. sibirica C.A.Mey. Photograph by Liu Li-Jia

Full size image
Table 1 The taxonomic treatment of sections and part of important history of nomenclature of 13 Libanotis taxa in this study. Species scientific names in this study are based on the FOC treatment (Sheh and Watson, 2005), and "—" indicates that the species were not included in that treatment
Full size table

Additionally, many plants of Libanotis have important medicinal value and are used as traditional Chinese medicinal materials. For example, six Libanotis taxa (L. buchtormensis (Fisch.) DC., L. lancifolia K.T.Fu., L. spodotrichoma K.T.Fu., L. wannienchun K.T.Fu., L. lanzhouensis K.T.Fu ex R.H.Shan & M.L.Sheh, and L. sibirica) are all known as the "Changchun Seven" in the Qinling Seven medicines, which is used to treat common cold, toothache, headache, traumatic injury, inflammation, swelling, rheumatism, respiratory diseases, as well as symptomatic coughs and dyspnea [40,41,42]. However, due to morphological feature exhibiting highly similar in inter-species and significant divergence in intra-species, the accurate identification of Libanotis species was extremely difficult [40]. Due to their morphological similarity, instances of homonym or synonym in common names exist in various regions and markets, making it challenging to distinguish them during collection, acquisition, and clinical usage. They are often mistakenly interchanged. For example, the above 'Changchun Seven' consists of six different species and is divided into sixteen varieties in herbal medicine, causing confusion in the herbal market [40]. Therefore, the selection of reliable molecular markers for ensuring the accurate identification of medicinal Libanotis species is of utmost importance.

The plastome was highly conserved in flowering plant and harbored sufficient variable loci [43, 44]. Hence, plastome data have been widely used in phylogenetic analyses and development of special DNA barcode in Apiaceae, Poaceae, Lamiaceae, Rosaceae, Liliaceae, Allium, Artemisia, and other plant taxa [44,45,46,47,48,49,50,51,52]. Regrettably, despite the presence of two Libanotis plastomes in GenBank, there has been a lack of plastid phylogenomic analysis conducted for this genus. In this study, we filled this gap by sequencing the plastid genomes of twelve taxa of Libanotis. Together with two plastomes previously reported, we conducted comprehensive analyses to (1) reveal the plastid characteristics and evolution of Libanotis; (2) identify suitable mutation hotspots from plastomes to use as candidate barcodes for species identification of Libanotis; (3) investigate the genus boundary of Libanotis and provide new sights into the phylogenetic position of this genus taxa distributed in China.

Results

Plastome features of Libanotis and repeat sequence analyses

The complete plastomes newly sequenced of 12 Libanotis species have been fully characterized, with sizes ranging from 146,836 bp (L. sibirica) to 148,100 bp (L. depressa) (Table 2, Fig. 2). Compared to other Libanotis taxa, L. depressa was particularly unique, with a significantly expanded IR region of length 19,437 and the GC content of only 43.7%. The analysis of the twelve Libanotis plastomes revealed a collective inventory of 129 genes, including 84 PCGs, 36 tRNA genes, 8 rRNA genes, and one pseudogene (Table 2, Table S3, and Fig. 2). Of particular interest, the inversion of the trnY-trnD-trnE gene, previously observed in certain species of Angelica L. and Peucedanum L., was also detected in L. incana. [44, 53]. Additionally, these thirteen Libanotis plastomes exhibited no gene rearrangements or losses (Fig. 3). Additionally, a total of 1049 simple sequence repeats (SSRs) and 549 repeats belonging to four different types were identified (Fig. S1, Table S6, S7). Compared with other related taxa of Selineae, such as Seseli and Peucedanum, Libanotis was not much difference in the analysis of repeated sequences.

Table 2 Features of the twelve Libanotis plastomes newly sequenced
Full size table
Fig. 2
figure 2

The gene map displays twelve newly sequenced Libanotis plastomes (L. sibirica was used as a representative). The inner circle's dark gray area indicates the GC content. Gene numbers and plastomes length are tagged inside. Genes outside the outer circle are transcribed clockwise, while inside are transcribed counterclockwise. Different gene functional groups are color-coded

Full size image
Fig. 3
figure 3

Mauve alignment of thirteen Libanotis plastomes, where blocks of the same color connected by lines indicate local collinear blocks within each alignment. The red boxes pick out are the inversion of the trnY-trnD-trnE gene

Full size image

Nucleotide diversity analyses and potential DNA barcodes

For these thirteen Libanotis, the nucleotide diversity (Pi) values for the protein-coding regions ranged from 3.58 × 10–4 (rps7 gene) to 0.01459 (ccsA gene), and the average value was 3.23 × 10–3 (Fig. 4, Table S8). The range of Pi values in non-coding regions and introns exhibits a considerable variation compared to coding regions. Among the protein-coding genes analyzed, only ccsA displayed a relatively high Pi value (> 0.01), whereas four other genes, namely matK, ycf2, ndhE, and ycf1, exhibited moderate levels of nucleotide diversity (0.007 < Pi < 0.01), making them viable alternatives for further investigation (Fig. 4A, Table S8). Furthermore, three non-coding regions and introns with high nucleotide diversity (Pi > 0 0.015) were identified: trnH-GUG-psbA, petA-psbJ, and ccsA-ndhD (Fig. 4B, Table S8). These eight highly variable regions (ccsA, matK, ycf2, ndhE, ycf1, trnH-GUG-psbA, petA-psbJ, and ccsA-ndhD) were selected as potential DNA barcodes.

Fig. 4
figure 4

Assessing nucleotide diversity (Pi) across the thirteen Libanotis plastomes through comparative analysis: (A) protein-coding genes, (B) non-coding and intron regions

Full size image

Phylogenetic analyses

Seventy-nine single-copy plastome CDS from 57 plastomes were used to reconstruct the phylogeny of Libanotis (Fig. 5, Table S4). Our analyses robustly supported that the Libanotis taxa fell into one tribe (Selineae), and they were not clustered as a monophyletic group or divided into three sections but fell into seven groups (Subclades) (Fig. 5): (I) L. sibirica and L. schrenkiana clustered with Seseli glabratum Willd. ex Schult. (PP = 1.00, BS = 100); (II) L. buchtormensis and L. seseloides was sister to Saposhnikovia divaricata (Turcz.) Schischk. (PP = 0.99, BS = 84); (III) L. incana did not clustered with other Libanotis. However, within the phylogenetic analysis, L. incana and subclades I and II formed a robust clade with high support values (PP = 1.00, BS = 100). This clade indicated that L. incana diverged first from the rest of the taxa (PP = 1.00, BS = 99); (IV) L. iliensis, L. grubovii and L. acaulis formed a clade (PP = 1.00, BS = 100), clustered with I, II, III, and some Peucedanum species. (V) this clade contained L. jinanensis, L. lanzhouensis, L. spodotrichoma, and Seseli intramongolicum Y. C. Ma. (PP = 0.99, BS = 76); (VI) L. condensata was sister to Pachypleurum alpinum (PP = 1.00, BS = 95), and P. alpinum is type of Pachypleurum; (VII) L. depressa, along with other Ligusticopsis species, established a strong and clearly separated clade, displaying high support values (PP = 1.00, BS = 100), distinguishing it from the rest of the genus. In the nrDNA-based tree (Fig. 6, Table S5), that species in the subclades were clearly divergent except for Subclade V, which was better clustered into a single branch (PP = 1.00, BS = 100). L. sibirica and S. libanotis were clearly not clustered with Seseli s.s., S. tortuosum and some Seseli species were clustered with Kitagawia, Peucedanum, L. incana, and L. lancifolia, while L. sibirica, S. libanotis and some Libanotis were clustered with Stenocoelium popovii, and several Seseli species.

Fig. 5
figure 5

The plastome CDS-based phylogenetic tree constructed by Bayesian inference (BI) and maximum likelihood (ML) with the posterior probabilities (PP) of BI and the bootstrap values (BS) of ML above the branches. The topology of the tree is derived from the optimal tree of the maximum likelihood method, and the unaligned tree is labeled in the upper left corner. Respectively, (*) represents maximum support in both two analyses, (#) represents those nodes not occurring in the BI strict consensus tree. The red is the newly sequenced Libanotis in this study, and the orange is the Libanotis sequences downloaded from Genebank. Different subclades are colored differently. Details of the sections labeled with different symbols are shown in Table 1

Full size image
Fig. 6
figure 6

The plastome nrDNA-based (ITS + ETS) phylogenetic tree constructed by Bayesian inference (BI) and maximum likelihood (ML) with the posterior probabilities (PP) of BI and the bootstrap values (BS) of ML above the branches. The topology of the tree is derived from the optimal tree of the bayesian inference method, and the unaligned tree is labeled in the upper left corner. (*) represents maximum support in both two analyses. Bolded are the sequences newly sequenced in this study, and the type species of Seseli and Libanotis are highlighted in red in the figure

Full size image

In terms of morphological sections of CDS-based tree, only L. sibirica and L. schrenkiana (Subclade I) can be retained from the five species of the core group Sect. Libanotis, with the remaining three species each dispersed in three other branches (Subclade II, III, VI). The six species of Sect. Eriotis are also not monophyletic, with the exception of L. buchtormensis which is better concentrated in two branches (Subclade IV, V), and the two species of Sect. Pseudolibanotis are also separated, one clustered with Sect. Eriotis and one within the genus Ligusticopsis (Subclade IV, VII). On the nrDNA-based tree (Fig. 7), Sect. Libanotis except L. incana clustered together. It is noteworthy that these species of Libanotis sect. Eulibanotis included in Schischk [14] (Table 1) included in this tree (L. montana (≡ S. libanotis), L. sibirica, L. schrenkiana, L. condensata, L. seseloides, L. transcaucasica (≡ S. transcaucasicum)) clustered into a highly supported monophyletic clade (PP = 1.00, BS = 100). Sect. Eriotis apart from L. iliensis and L. lancifolia also clustered into a monophyletic clade (PP = 1.00, BS = 97), but within this clade were also included two narrowly-fielded Seseli species published in recent years. The two species of Sect. Pseudolibanotis are also separated.

Fig. 7
figure 7

Comparison of two trees constructed based on different datasets. The left one is a CDS-based phylogenetic tree, and the right one is an nrDNA-based (ITS + ETS) phylogenetic tree. The same species with different subclades in the two trees are connected to each other by a line representing the color of that subclade. Details of the sections labeled with different symbols are shown in Table 1

Full size image

Comparative plastome analyses

The boundaries of these species were not too distinctly different or regular, either on the basis of phylogenetic subclades or on the basis of former taxonomic sections (Fig. S2). There is no doubt that the plastome structure of Libanotis is relatively conserved. The Relative Synonymous Codon Usage (RSCU) values across all codons exhibited a spectrum from 0.33 to 2.02, as depicted in Figure S3 and detailed in Table S9. Notably, L. depressa (Subclade VII) exhibited lower RSCU values for UGA termination codon (RSCU = 0.57), whereas L. condensata (Subclade VI) showed lower RSCU values for UAG termination codon (RSCU = 0.62) and higher values for UGA termination codon (RSCU = 0.74) compared to other subclades. The usage of specific codons within the remaining subclades, apart from the aforementioned individual subclades, shows no significant differences. (Fig. S3; Table S9).

The divergence analysis of thirteen Libanotis plastomes revealed that the coding regions exhibited higher conservation compared to the non-coding regions (Fig. 8). Compared with other taxa, L. schrenkiana was highly similar to the reference L. sibirica. Furthermore, the plastid divergence between Subclades I and II is relatively low, as is the divergence between IV and V, while the remaining three separate subclades exhibit distinct differences (Fig. 8). Interestingly, some subclades exhibit a certain degree of conservation when compared to the rest of the subclades, while others show significant differences. For instance, L. incana (Subclades III) displays significant distinctions from the rest of the sequences in the region from trnD-GUC to trnE-UUC, likely due to gene inversion, which aligns with the above analysis. It also exhibits noticeable differences from other sequences in the region from psbL-psbF-psbE.

Fig. 8
figure 8

mVISTA-based sequence identity plots for the thirteen plastomes with L. sibirica as the reference. The different colors and Roman numerals correspond to the different subclades separated by the plastome CDS-based phylogenetic tree in Fig. 5

Full size image

Discussion

Comparison of the Libanotis plastomes and Potential DNA barcodes

In this study, we sequenced and assembled twelve plastomes of Libanotis and performed comprehensive comparative analyses of these plastomes with one other published plastomes of this genus obtained from GeneBank. All Libanotis plastomes exhibited the typical quadripartite structure with various features displaying similarity. And the genome length (146,836 BP- 148,100 bp), IR/ SC borders and gene numbers and arrangements (129) of each Subclades formed by Libanotis species were not significantly different. These results suggested that Libanotis plastomes were highly conserved between different subclades, while the coding regions were more conserved than the non-coding regions, the IR regions were more conserved than the single copy regions. Nevertheless, we identified eight mutation hotspot regions, each spanning over 200 bp, with elevated Pi values. These regions, including the matK gene, ycf2 gene, ccsA gene, ndhE gene, ycf1 gene, trnH-GUG-psbA, petA-psbJ, and ccsA-ndhD, were selected as potential DNA barcodes for the purposes of phylogenetic analysis and species identification within the Libanotis genus.

Phylogeny analyses and taxonomic inference

We have reconstructed the phylogenetic relationships using 13 Libanotis species plastomes sample (Fig. 5). This work provides a solid and high-resolution phylogenetic tree of Libanotis, revealing inconsistencies between molecular systematics and traditional taxonomic studies. According to the current research results, the genus Libanotis is obviously polyphyletic, and L. sibirica (type species) and L. schrenkiana (Subclade I) formed a monophyletic clade with strong supports. Meanwhile, the clade could be recognized by leaf segments ovate-rhombic or lanceolate, surfaces glaucous on the back of the leaves and sparsely puberulent; bracts absent or few, subulate to linear, small, easy to loss; bracteoles several, linear; petals abaxially glabrous; calyx teeth conspicuous, triangular-lanceolate; fruit ovoid-ellipsoid, dorsally compressed, densely pubescent when young, becoming sparsely puberulent or glabrous; ribs subequal, shortly keeled; vittae 1 in each furrow, 2 on commissure [1, 2]. As Pimenov argues [22], Libanotis s.s. and Seseli s.s. do not differ in fruit morphology up to the genus level, and their main morphological differences are in less commonly used morphologies (characteristics of bracteoles, bracts and leaf segments, stem branching, stem and petiole pubescent etc.). But the results of the phylogeny suggest that we cannot simply merge them because the monophyly of the genera is not supported, and we cannot rule out the effects of homoplasy or reversals. Just like the concept of cryptic species, Libanotis is in a sense a cryptic genus. Therefore, we propose to accept this genus in narrow sense, namely Libanotis s.s., and identify only two members for the time being. According to the type specimens and literature records, other possible members of Libanotis s.s. are Seseli junatovii V. M. Vinogradova and Seseli salsugineum A.Duran & Lyskov. However, due to the limited sampling and the lack of sufficient reliable morphological information, we would not make taxonomic treatments for now. The results of the comparison between nrDNA-based tree and the CDS-based tree (Fig. 7) showed that there was nucleoplasmic conflict in Libanotis, these may be due to incomplete lineage sorting and introgression. In these conflicts, we found that species with similar leaf morphology tended to cluster more in the nrDNA-based tree: in the branch where L. sibirica located, the four Libanotis species (L. sibirica, L. schrenkiana, L. seseloides, and L. condensata) all have green, thin, papery leaf blades, and leaf abaxial surfaces sometimes gray-green; whereas the eight Libanotis species (L. abolinii, L. grubovii, L. buchtormensis, L. laticalycina, L. spodotrichoma, L. wannienchun, L. jinanensis, and L. lanzhouensis) in the branch beneath them have leathery to fleshy leaf blades, the leaf blades mostly blue-green or gray-green overall. The rest of the dispersed species have distinctive vegetative body morphology. Meanwhile, species with similar morphology of mericarps tended to cluster together (such as subclades I and V) in the CDS-based tree (Table S10).

When considering the outcomes of morphological sections within the phylogenetic tree, it becomes evident that the alignment is less than ideal. None of the three sections of taxa appear to be monophyletic. This situation is not unique among Apiaceae family. The Apiaceae, located on the upper echelons of angiosperms, signify a taxon along the path of divergence, and belong to the most complicated families of flowering plants, also in terms of species identification [54,55,56]. The reliability of diagnostic features between and within genera may be affected by homoplasy and reversals, and that traditional printed dichotomous keys in large “Floras” are far from satisfying. Recent times have witnessed a rapid reconfiguration of species within the tribe Selineae, marked by the revision of established genera and the independence of new ones [57,58,59]. In our assessment, we propose that all thirteen species in this study, except L. sibirica and L. schrenkiana, should be transferred and revised, but not transferred to Seseli to further confuse the polygenetic genus. It is worth noting that Peucedanum, Saposhnikovia and Kitagawia, which are close relatives of Libanotis and Seseli, also suffer from the problems mentioned above. Comprehensive sampling of Seseli and Peucedanum, two the world-wide complex genera with a mass of species, will be crucial to the taxonomic system of Libanotis and the entire tribe Selineae.

Except for them, the members of Libanotis were scattered among the branches, and the phylogenetic positions of L. condensata and L. depressa are particularly noteworthy. L. condensata (Subclade VI) and Pachypleurum alpinum Ledeb. (Type species of Pachypleurum) clustered together. Their morphology is similar in vegetative body which both have solitary stem with branched above or simple, hollow, glabrous, and striate, base densely clothed with fibrous leaf remains, and oblong leaf blade, but quite different in mericarps especially ribs all winged, subequal in Pachypleurum. Thus, L. condensata may be more closely related to Pachypleurum than Seseli s.s. or Libanotis s.s., but its transfer to Pachypleurum seems inappropriate unless the definition of Pachypleurum is reconstructed. Other molecular evidence [7] also supports the view that L. condensata does not belong to Libanotis. In nrDNA sequence (ITS) phylogenetic results [7], L. condensata is obviously separated from above genera, and is located in Pilopleura Schischk. While our nrDNA tree (ITS + ETS) showed that L. condensata was inserted into Libanotis s.s.. Unfortunately, due to the chloroplast genome and ETS sequence of Pilopleura absented, we could not confirm the relationship between Pilopleura and L. condensata. L. depressa clustered with Ligusticopsis species. However, we found L. depressa develops few bracts, lanceolate and very unequal bracteoles, and mericarps with few vittae in the furrow (1) and commissure (2), not strongly compressed and marginal ribs not winged, which are distinguishable from Ligusticopsis. Consequently, L. depressa should be treated as an independent taxon distinct from Ligusticopsis or Libanotis. Pimenov [24] argued that L. depressa should be transferred to Stenocoelium. Our results showed that Stenocoelium popovii clustered with some Seseli species and was far away from L. depressa. L. depressa also does not conform to the unique mericarp morphology of Stenocoelium that ribs are thick-obtuse, very prominent, irregularly denticulate especially along ribs and furrows are narrow. Due to conflicting and partial lack of morphological data sampling, we will detailedly discuss its taxonomic status in future research. The other species of Libanotis (Subclade II, III, IV, V) were clustered into some relatively single clades: L. incana (Subclade III) was alone; L. seseloides and L. buchtormensis were gathered in one branch (Subclade II) and then sister to Saposhnikovia, but the shape of mericarps and the numbers of vittae are quite different among them; L. acaulis and L. grubovii, L. iliensis clustered together (Subclade IV); L. lanzhouensis, L. jinanensis, and L. spodotrichoma formed a clade (Subclade V). Compared with Libanotis s.s., they belong to Sect. Eriotis or Sect. Pseudolibanotis, and petals are densely coated with soft hairs or stem not developed, which is easy to distinguish. In conclusion, our results showed that Libanotis s.s. has a need to be retained, but other eleven species that thought to be attributed to Libanotis should be transferred out Libanotis genus but their taxonomic status needs to be further studied by adding more species.

Conclusion

This study marks the inaugural endeavor to conduct a comprehensive exploration of plastome characteristics and to deduce the phylogeny of the Libanotis genus, encompassing a total of thirteen Libanotis species. In the course of this investigation, we conducted the fresh sequencing, assembly, and annotation of complete plastomes for twelve Libanotis species along with two closely related taxa. These results suggested that Libanotis plastomes were conserved between different subclades, while the coding regions were more conserved than the non-coding regions, and the IR regions were more conserved than the single copy regions. Nevertheless, eight mutation hotspot regions (matK gene, ycf2 gene, ccsA gene, ndhE gene, ycf1 gene, trnH-GUG -psbA, petA-psbJ, ccsA-ndhD) longer than 200 bp with high Pi values were chosen as potential DNA barcodes for the purpose of both phylogenetic investigation and species identification used in materia medica of Libanotis. 78 common single-copy CDS from fifty-seven plastomes sequences and 144 nrDNA (72 ETS + 72 ITS) sequences were used to perform the phylogenetic analysis of Libanotis. Plastid phylogenomic analyses confirmed the efficacy of plastome data in enhancing the support and resolution of Libanotis phylogeny, firmly showing that Libanotis belong to Selineae and not a monophyletic genus, and the species within the sections in the original morphological framework are also polyphyletic. We finished the delimitation of Libanotis by establishing Libanotis s.s. and provided some taxonomic suggestions for other species in the genus, especially L. depressa and L. condensata. In short, our study can provide new insights into the plastome evolution of Libanotis and promoted the improvement of taxonomic system for Aipaceae family.

Methods

Taxon sampling and DNA sequencing

A total of 57 plastomes from 56 taxa and 144 nrDNA sequences (72 ITS + 72 ETS) from 67 taxa were used in this study, of which 48 plastomes and 96 nrDNA originated from us (Table S4, S5). We collected fresh and fully developed leaves from twelve different Libanotis species, which included the type species L. sibirica (L.) C.A.Mey. and then dried with silica gel (Table S1). These sections of Libanotis species reference FRPS [2], including all three sections (one previous section has been used to create the new monotypic genus Sajanella) to establish a more complete phylogenetic framework (Table 1, Fig. 1). Additionally, we expanded our sampling efforts for Pachypleurum alpinum Ledeb. and Stenocoelium popovii V.M.Vinogr. & Fedor., based on prior experimentation and taxonomic studies [24, 60]. In addition to the 14 plastomes, we newly measured 24 ETSs as well as 24 ITSs containing 17 species of Libanotis and two closely related species (Table S2). The formal identifications of all collected samples were identified by Liu Li-Jia and Professor He Xing-Jin from Sichuan University. Specimens vouchering the mentioned taxa were stored in the herbarium of Sichuan University (SZ) and the herbarium of the Kunming Institute of Botany (KUN), and the details of these vouchers can be found in Table S1 and S2. In order to distinguish them from other genera, all the scientific names of Libanotis species and sections in this study were based on the taxonomic treatment of FOC and IPNI [1, 61], but the scientific names in Table 1 followed the authors' original records. The newly published species L. jinanensis and the newly transferred species L. grubovii are included in Sect. Eriotis according to their morphologic descriptions (Table 1, Table S10).

We began by extracting total DNA from approximately 20 ~ 30 mg of silica gel-dried leaves using the CTAB (Cetyl trimethylammonium bromide) method [62]. We conducted Polymerase chain reactions (PCRs) to amplify ITS (Internal Transcribed Spacer) and ETS (External Transcribed Spacer) sequences using the following primers: ITS-4, ITS-5 [63], 18S-ETS [64], and Umb-ETS [65]. Each PCR reaction had a 30 µL volume with 2 µL plant DNA, 1.5 µL forward primer,1.5 µL reverse primer, 15 µL of 2 × Taq MasterMix (cwbio, Beijing, China), and 10 µL of ddH2O. We used Geneious v2023.0.4 [66] for sequence editing and assembly. The newly acquired sequences have been officially submitted in GenBank (accession numbers in Table S2). For plastomes, we fragmented the genomic DNA into 150 bp fragments to create a pair-end library, adhering to the manufacturer's instructions provided by Illumina in San Diego, CA, USA. The sequencing of these libraries took place on the Illumina NovaSeq platform at Personalbio in Shanghai, China. We applied fastP v0.15.0 [67] to filter the raw data, and these high-quality reads were then assembled for the whole plastomes using GetOrganelle v1.7.7.0 [68].

Genomic annotation and feature analyses

We utilized the Plastid Genome Annotator (PGA) [69] for the annotation of plastomes, employing L. buchtormensis (MZ707534) and L. spodotrichoma (MZ707535) as our reference sequences. Subsequently, we performed manual refinements using Geneious v2023.0.4 [66]. The newly acquired plastome sequences for the twelve Libanotis taxa, along with two additional sequences, have been officially submitted in GenBank (accession numbers in Table S1). To visualize the circular plastome maps for the twelve newly sequenced Libanotis taxa, we employed the online tool Organellar Genome DRAW (OGDRAW) [70]. Furthermore, we identified gene rearrangements among the thirteen Libanotis taxa including one previously published sequence, using Mauve Alignment [71] within Geneious v2023.0.4 [66].

Repeat sequence and nucleotide diversity analyses

We employed the online REPuter program [72] to identify repeat sequences in the plastomes of the thirteen Libanotis taxa and the parameters used for this analysis referred to Cai et al. [33]. Furthermore, we utilized the Perl script MISA [73], available at http://pgrc.ipk-gatersleben.de/misa/sleben.de/misa/, to detect simple sequence repeats (SSRs) within the plastomes of the thirteen Libanotis taxa. For the assessment of nucleotide diversity (Pi) within protein-coding genes, noncoding regions, and introns, we turned to DnaSP version 6.12.03 [74]. This analysis aimed to pinpoint regions with elevated mutation rates, potentially serving as valuable molecular markers for future research. Regions meeting or exceeding a length of 200 base pairs were singled out for this purpose, as described previously [33].

Sequences selection and alignment

In accordance with initial experiments and prior taxonomic assessments [24, 60], we carefully curated two dataset consisting of 57 complete plastomes derived from 56 taxa and 144 nrDNA (72 ITS + 72 ETS) from 67 taxa for the purpose of constructing phylogenetic trees. Notably, 13 plastomes of these sequences, which include 11 Libanotis species (including the type species L. sibirica) and two related taxa, Stenocoelium popovii and Pachypleurum alpinum, were being introduced for the first time into our analysis. And among these nrDNAs, all ETSs of 17 Libanotis and Stenocoelium popovii were sequenced for the first time. In recognition of the intricate relationship between Seseli and Libanotis, we incorporated Seseli into our study. To establish the root of our phylogenetic tree, we selected three species from the Tordylieae tribe: Heracleum moellendorffii Hance, Heracleum yungningense Hand.-Mazz., and Semenovia transiliensis Regel & Herder, as recommended by Wen et al. [75]. Our main clade designations were based on the contributions of Downie et al. [76] and Wen et al. [75]. We further assembled a dataset comprising 78 common single-copy coding sequences (CDSs) extracted from the 57 complete plastomes. This dataset was concatenated using PhyloSuite v1.2.2 [77]. To ensure accuracy, we aligned the sequences using MAFFT v7.221 [78] and performed inspection and manual refinements with the assistance of MEGA7 [79]. It's worth noting that all sequences data utilized in our phylogenetic analyses are readily accessible in GenBank (Table S4, S5).

Phylogenetic analyses

To elucidate phylogenetic relationships, we employed both maximum likelihood (ML) and Bayesian inference (BI) methods. For ML analyses, we utilized RAxML v8.2.10 [80] with the GTRGAMMA model, accompanied by 1000 rapid bootstrap replicates to assess node support. In the case of BI analyses, we first determined the best-fitting substitution model using MrModeltest v2.4 [81]. Subsequently, we conducted Bayesian inference with MrBayes v3.2.7 [82], employing the selected GTR + I + G parameters. The parameter settings for the BI analysis refer to previous research about Apiaceae [33, 83, 84]. Finally, we visualized and edited the resulting phylogenetic trees using FigTree v1.4 [85].

Comparative analyses of plastomes

We visualized the variations in size between the inverted repeat (IR) border regions in the plastomes of the thirteen Libanotis species using IRscope [86]. Any necessary manual adjustments were made to ensure accuracy. Subsequently, we conducted a sequence divergence analysis of these thirteen plastomes, using mVISTA [87] in Shuffle-LAGAN mode, with L. sibirica serving as the reference species. For codon usage analysis, we employed codonW [88]. To reduce the impact of sampling bias [63, 89], we selected 53 coding sequences (CDSs) from the thirteen plastomes, excluding CDSs shorter than 300 base pairs and repetitive sequences. These selected CDSs were then concatenated using PhyloSuite v1.2.2 [77]. To visualize the relative synonymous codon usage (RSCU) [90] values across the thirteen plastomes, we utilized TBtools [91].