The samples of A. gracilis (Voucher specimen, No. DY159) and A. henryi (Voucher specimen, No. DY133) utilized were collected from Guangwu Mountain, Bazhong City, Sichuan Province, China. The voucher specimens were then deposited at the herbarium of Sichuan Normal University (SCNU), Chengdu City, Sichuan Province, China (contact: Dr. Prof. Zhixi Fu, fuzx2017@sicnu.edu.cn). The extraction of total genomic DNA from fresh leaves was performed following the CTAB DNA extraction protocol (Allen et al., 2006). The quantification and evaluation of the total genomic DNA integrity were assessed using the NanoDrop 2000 Spectrophotometer and Qubit 4 Fluorometer (Thermo Fisher Scientific, Wilmington, DE, USA). DNA libraries were constructed using the Illumina Paired-End DNA Library Kit (Illumina Inc., San Diego, CA, USA) and subsequently sequenced on the NovaSeq 6000 platform with 150 bp paired-end reads (NovoGene Inc., Beijing, China). Eventually, the Illumina Genome Analyzer (Hiseq 2000, Illumina, San Diego, CA, USA) was employed to obtain the raw sequence data. The raw data were then subjected to primary and secondary quality control to yield clean data.

Two sets of clean data, comprising the sequence reads of A. gracilis and A. henryi, were mapped to the reference sequence of Ainsliaea latifolia (D. Don) Sch.-Bip. using Bowtie2 v.2.4.5-Linux (Langmead et al., 2019). Following this, SAMtools v.1.15-Linux (Danecek et al., 2021) was employed to selectively retain only the reads mapped to the reference sequence for subsequent assembly. The cp genomes of two Ainsliaea species were assembled by the SPAdes v.3.15.1-Linux with default parameters (Prjibelski et al., 2020). The assembly results were visualized, untangled, inspected, and exported to two complete cp genome sequences using Bandage v.0.9.0-Linux (Wick et al., 2015). Collinearity analysis of the two sequences was performed using MUMmer v4.0.0-Linux (Marcais et al., 2018), with A. latifolia as the reference. Subsequently, the obtained sequences were annotated utilizing the Plastid Genome Annotator (PGA) (Qu et al., 2019), referencing the sequences of Amborella trichopoda Baill. and A. latifolia. The annotation results were examined utilizing Geneious R11 (https://www.geneious.com) and adjusted manually as needed. Eventually, the circular map of cp genomes was generated through OrganellarGenomeDRAW (OGDRAW) (Greiner et al., 2019).

The simple repeat sequence (SSR) was detected with MISA (Beier et al., 2017). The minimum number of repeats for each nucleotide type was set as follows: 10 repeats for mononucleotide (mono-), 5 repeats for dinucleotide (din-), 4 repeats for trinucleotide (tri-), and 3 repeats for tetranucleotide (tetra-), pentanucleotide (penta-), and hexanucleotide (hexan-). The long repeats were analyzed by REPuter (Kurtz et al., 2001). The analysis encompassed forward (F), reverse (R), complement (C), and palindromic (P) repeats. The maximal repeat size was set at 50 bp, the minimal repeat size was set at 30 bp, and the hamming distance was set at 8.

The sequences of A. gracilis (No. OQ723680), A. henryi (No. PP175243), A. latifolia (No. MW316662), Myripnois dioica Bunge (No. MK784068), Pertya multiflora Cai F. Zhang & T. G. Gao (No. MW148616), and Pertya phylicoides J. F. Jeffrey. (No. MN935435) were retrieved from GenBank in the NCBI database for comparative cp genome analysis. The contraction and extension of the IR borders within the four major regions (LSC/IRb/SSC/IRa) of the six cp genome sequences were visualized using IRscope (Amiryousefi et al., 2018). The online software mVISTA, with the Shuffle-LAGAN mode (Brudno et al., 2003; Frazer et al., 2004), was employed to compare the plastomes using the sequence of A. gracilis as a reference.

Python scripts were used to extract the coding regions (CDS) and non-coding regions (IGS) from six sequences for subsequent alignment. These sequences included two from Ainsliaea and four from Pertyeae. The sequences were aligned together using the “auto” strategy of MAFFT v.7.475 (Katoh et al., 2019). Nucleotide diversity (Pi) was subsequently calculated using the sliding window of DnaSP v.6.12.03 with window length of 600 bp and step size of 200 bp (Rozas et al., 2017). The Pi values were visualized using an R script.

A total of 25 complete cp genomes from Asteraceae family were selected to construct phylogenetic trees to determine the systematic position of A. gracilis and A. henryi. They represented the major five subfamilies (Asteroideae, Cichorioideae, Gymnarrhenoideae, Pertyoideae, and Carduoideae). Anthriscus cerefolium Hoffm. (Apiaceae) and Kalopanax septemlobus (Thunb.) Koidz. (Araliaceae) were chosen as outgroups. Two data matrices (complete cp genomes and CDS) were selected for phylogenetic analysis. These 27 sequences data were aligned by MAFFT v.7.475 with default parameters (Katoh et al., 2019), followed by manual inspection and adjustment of sequence differences. The maximum likelihood (ML) and Bayesian inference (BI) trees was inferred with RaxML HPC 2 v8.2.12 (Stamatakis, 2014) and MrBayes v3.2.7a (Ronquist et al., 2012) accessible on CIPRES platform (https://www.phylo.org/), employing the GTRGAMMA substitution model (predicted by ModelTest-NG v0.1.7, Darriba et al., 2020), with 1000 bootstrap replicates (Miller et al., 2010). Finally, the ML and BI trees was visualized and optimized in the program FigTree v.1.4.4 (http://tree.bio.ed.ac.uk/software/figtree).

The graphical plastome maps of the newly sequenced Ainsliaea species (A. gracilis and A. henryi) are depicted in Figure 1. A total of 134 genes were identified and annotated in the cp genomes, consisting of 87 protein-coding genes, 37 tRNA genes, 8 rRNA, and 2 pseudogenes (Figure 1). Among these genes, 16 contained one intron (ndhA, ndhB, petB, petD, atpF, rpl2, rps12, rpl16, rps16, rpoC1, trnA-UGC, trnG-UCC, trnI-GAU, trnK-UUU, trnL-UAA, trnV-UAC) and 2 genes possessed two introns (clpP and ycf3). It was found to have 19 double-copy genes, including 5 protein-coding genes (ndhB, rpl2, rpl23, rps7, rps12), 4 rRNA genes (rrn16, rrn23, rrn4.5, rrn5), 8 tRNA genes (trnA-UGC, trnG-UCC, trnI-CAU, trnI-GAU, trnL-CAA, trnN-GUU, trnR-ACG, trnV-GAC), and 2 unknown function genes (ycf2, ycf15) (Table 2). The gene ycf1 also has two copies, but one of them was a pseudogene. In addition, the gene rps12 was a trans-splicing construct. It is noteworthy that the functions of ycf1, ycf2, ycf3, ycf4, and ycf15 are still unknown.

The gene structure and content of cp genomes of the two Ainsliaea species were highly similar to other four Pertyeae species. The sequence length of the two Ainsliaea and four Pertyeae species ranged from 152,805 bp (A. henryi) to 153,793 bp (M. dioica) (Figure 1; Table 1). These cp genomes exhibited comparable genome structures, comprising 131–135 genes (which include 85–87 protein coding genes, 37 tRNA genes, 8 rRNA genes, and 1-3 pseudogenes, Figure 2 and Table 1). The genes could be categorized into 4 types: photosynthesis-related genes (45), self-replication-related genes (74–75), other genes (6) and genes of unknown function (6–9) (Figure 2). The six genomes contained a pair of IRs (25,049–25,212 bp) separated by the LSC (84,084–85,012 bp) and SSC region (18,378–18,683 bp) (Table 1; Figure 2). The average GC content of the genomes was 37.6%. Compared to the LSC (35.63%–35.83%) and SSC (31.15%–31.31%) regions, the IR regions possessed a higher GC content (43.07%–43.12%)

In total 57, 58, 56, 48, 63, and 54 SSRs were found in A. gracilis, A. henryi, A. latifolia, M. dioica, P. multiflora, and P. phylicoides, respectively (Figure 3). Among these SSRs, mononucleotide repeats were the most abundant, while pentanucleotides repeats were only detected in A. gracilis and P. multiflora. Further analysis of the long repeats is provided in Figure 4. The results demonstrated that the number of long repeats in the tribe Pertyeae were highly similar. Palindromes (21–31) were the most prevalent, with a majority of them ranging from 40–49 bp in length. This was followed by forward repeats (17–29) and reverse repeats (1–7), while complement repeats were not observed.

We compared the IR/SC boundaries of the 6 species within Pertyeae (Figure 5). While the lengths of the IR regions were similar among the six species, variations in the extensions and contractions at the IR boundaries were observed. As shown in Figure 5, notably, A. gracilis exhibited the longest IR region length (25,212 bp) among the six species. Moreover, substantial differences were observed in the range of each region within Pertyeae. In particular, the rps19 gene showed expansions ranging from 60 to 61 bp, extending from the LSC to the IRb region in the five species. In M. dioica, however, this gene was entirely located within the LSC region, 68 bp away from the LSC/IRb boundary. The ndhF gene was found in the SSC regions, both positioned 23–42 bp away from the SSC/IRa boundary, except in M. dioica (101 bp away from the IRb/SSC boundary) and P. phylicoides (37 bp away from the IRb/SSC boundary). The ndhF gene is positioned near the IRa region in the four species and near the IRb region in the other two species. Similarly, the ycf1 gene was positioned at the IRb/SSC border, except for M. dioica and P. phylicoides, where it was located at the SSC/IRa border.

The complete cp genome sequences of six Pertyeae species were compared using mVISTA online software (Figure 6). While some variation was present, the results indicated a large extent of similarity among these cp genomes. It is worth noting that the SC regions were more variable compared to the IR regions. The conserved non-coding sequences (CNSs) were more diverse than the coding sequences.

Nucleotide diversity (Pi) analysis revealed that a higher proportion of variable sites in non-coding regions on average compared to coding regions (Figure 7). In the coding regions, Pi values ranged from 0 to 0.24545 (trnK-UUU), with an average of 0.00633. In the non-coding regions, Pi values ranged from 0 to 0.04218 (trnT-UGU-trnL-UAA_1), with an average of 0.00804. Notably, thirteen distinct regions showcased high variability, with Pi values exceeding 0.05 (trnK-UUU, trnG-UCC, and trnT-GGU) and 0.02 (trnT-UGU-trnL-UAA_1, rps19-rpl2_2, accD-psaI, rpl22-rps19, rpl2_2-trnH-GUG, ndhC-trnV-UAC_2, trnP-UGG-psaJ, psbB-psbT, ycf4-cemA, and rps18-rpl20). These highly variable regions possess significant potential as informative markers for future phylogenetic analysis at higher taxonomic levels within Pertyeae.

The ML (Figures 8, 9) and BI phylogenetic trees (Figures 10, 11) were constructed using the complete cp sequences and CDS from 27 species, representing five main clades in Asteraceae. In this analysis, A. cerefolium (Apiaceae) and K. septemlobus (Araliaceae) were used as outgroups. The phylogenetic relationships inferred from the ML and BI trees based on complete cp genomes (Figures 8, 10) and CDS (Figures 9, 11) were mostly consistent. The phylogenetic trees revealed that all sampled taxa in Asteraceae formed five significant main subfamilies (Asteroideae, Cichorioideae, Gymnarrhenoideae, Pertyoideae, and Carduoideae), encompassing ten tribes (Astereae, Gnaphalieae, Anthemideae, Senecioneae, Heliantheae, Inuleae, Cichorieae, Gymnarrheneae, Pertyeae, and Cardueae). Among these five subfamilies, Pertyoideae was closely allied with Gymnarrhenoideae and Carduoideae. Furthermore, the two newly sequenced species and four additional species from Pertyeae forme a clade (Pertyeae) with robust support. A. latifolia, A. henryi, and A. gracilis collectively formed a distinct clade, indicating their close evolutionary relationship. Moreover, the clade of Myripnois and Pertya was closely allied with Ainsliaea. Notably, Myripnois was identified as nested within the genus Pertya. Compared to Ainsliaea, Myripnois showed a closer relationship to Pertya. The results supported that the three genera belonged to the tribe Pertyeae and the subfamily Pertyoideae.

The two cp genomes exhibited lengths of 152,959 (A. gracilis) and 152,805 (A. henryi) bp. The GC content was 37.6% in these cp genomes, as in Dolomiaea denticulate (Y. Ling) C. Shih (Shen et al., 2020) and Sonchus brachyotus DC. (Wang et al., 2021). These genomes each contained 134 genes, congruent with those of Sinosenecio albonervius Y. Liu & Q. E. Yang (Abdullah et al., 2021), which comprised 87 protein-coding genes, 37 tRNA genes, 8 rRNA genes, and 2 pseudogenes (Table 2). Out of 134 genes, 16 contained one intron, while 2 contained two introns. A total of 19 double-copy genes were also detected (Table 2). The detected pseudogenes (rps19 and ycf1) were previously reported in Ligularia Cass. (Chen et al., 2018), Artemisia L. (Liu et al., 2013) and Parasenecio W. W. Sm. & J. Small (Liu et al., 2023). Notably, rps12 was recognized as a trans-splicing gene, previously reported in other species (Choi and Park, 2015). Moreover, the use of GTG, as a start codon for rps19, was reported in other angiosperm cp genomes (Liu and Xue, 2004; Wang et al., 2016).

The cp genomes of six Pertyeae species revealed significant conservation in size, gene content, structure, and other characteristics. A typical quadripartite circular structure was present in the genomes, with a distinct separation between the LSC and SSC regions by the IR regions. It was accordant with the cp genomes observed in other Pertyeae species (Lin et al., 2019; Wang et al., 2020; Liu et al., 2021). However, the cp genome size of the genus Ainsliaea (152,805–152,959 bp) is relatively shorter compared to the genera Pertya and Myripnois (153,379–153,793 bp). According to analysis of the cp genome divergence, Ainsliaea appears to possess low levels of sequence divergence and generally conserved plastomes. The intergenic spacers were identified as the most divergent regions, with non-coding regions showing greater divergence than coding regions. Previous multispecies investigations (Hu et al., 2020) had demonstrated that intergenic spacers were highly informative phylogenetic markers.

SSRs, composed of 1-6 nucleotide repeat units, are prevalent across cp genomes. Previous studies have employed them for species identification, genetic diversity, and polymorphism research (Deguilloux et al., 2004; Redwan et al., 2015). A total of 336 SSRs were identified in the cp genomes of Pertyeae. Mononucleotide repeats predominated in sequences. This phenomenon might be caused by the fact that SSRs commonly consist of polyA or polyT repeats (Zhang et al., 2019d). It was congruent with the cp genomes of most Asteraceae species (Choi and Park, 2015; Abdullah et al., 2021; Liu et al., 2023). The discovery of these newly identified SSRs will present valuable resources for future development of genetic markers for Ainsliaea species. Moreover, this study identified thirteen highly polymorphic loci (e.g., trnK-UUU, trnG-UCC, trnT-GGU, accD-psaI, and rpl22-rps19). In the tribe Pertyeae, these variable regions could serve as possible DNA barcodes for species identification and phylogenetic analysis.

The Asteraceae, recognized as one of two largest and most diverse families of blooming plants worldwide (Bremer and Anderberg, 1994; Funk et al., 2005; Anderberg et al., 2007), comprises sixteen subfamilies (Susanna et al., 2020; Zhang et al., 2024). In our study, 5 of the 16 subfamilies were sampled. Most nodes in the phylogenetic trees displayed high support values and were similar. The phylogenetic relationships of five subfamilies were consistent with earlier investigations (Funk et al., 2009; Fu et al., 2016; Panero and Crozier, 2016; Mandel et al., 2017). Previous studies have indicated the phylogenetic position and demographic histories of the tribe Pertyeae utilizing short DNA fragments (e.g., ndhF, rbcL, and matK) or the complete cp genomes data. For instance, Zhang (2024) reconstructed an updated phylogeny of Ainsliaea based on the plastid ndhF and nrDNA (ITS, ETS) sequences. Mitsui and Setoguchi (2012) addressed the demographic histories of adaptively diverged riparian and non-riparian species of Ainsliaea using 10 nuclear DNA loci (e.g., CHS, GTF). Fu et al. (2016) proposed that Pertyeae (as recognized by Panero and Funk, 2002) was sister to the tribes Cardueae and Gymnarrheneae and nested above the subfamily Carduoideae. Moreover, Pertyeae also has been suggested as a sister group to the tribes Cichorieae (Lin et al., 2019) and Cardueae (Wang et al., 2020). However, research on the cp genomes of Ainsliaea has not yet been conducted to date.

The phylogenetic analysis elucidated the taxonomy placement of the genus Ainsliaea within the tribe Pertyeae (Pertyoideae). Compared to A. gracilis, A. henryi and A. latifolia exhibited the closest phylogenetic relationship and clustered together, which is consistent with the findings of Zhang (2024). We also identified that the genus Ainsliaea was closely related to Myripnois and Pertya. This finding was accordant with previous studies (Kim et al., 2002; Fu et al., 2016). Based on previous morphological studies, Myripnois was delineated as a distinct genus (Gao et al., 2011). However, phylogenetic analysis confirmed that Myripnois was nested within Pertya in our study, aligning with the results derived from cladistic analysis (Freire, 2017).