Fresh leaves of the eight Aster species were gathered from the wild ( Table 1 ). The formal identification of the plant material was undertaken by Dr. Zhixi Fu. The voucher specimens were then preserved in the herbarium at Sichuan Normal University in China (SCNU) (contact person: Dr. Zhixi Fu, fuzx2017@sicnu.edu.cn). As these species are not included in List of National Key Protected Wild Plants in China, there was no need to obtain a permit for their collection. Following the CTAB DNA extraction protocol (Allen et al., 2006), genomic DNA was extracted using the Plant Genomic DNA Kit (Tiangen, Beijing, China). The construction of the DNA library was carried out using the Illumina Paired-End DNA Library Kit (Illumina Inc., San Diego, CA, USA), and, subsequently, sequencing was performed on the Illumina Genome Analyzer (Hiseq 2000, Illumina, San Diego, CA, USA). The resulting raw data for each of the eight species consisted of approximately 150-bp paired-end read lengths. The 27 complete chloroplast genomic datasets are available for download on NCBI ( Table 2 ). In the final supermatrix, names of species were checked based on Flora of China (Chen et al., 2011).

For the assembly of the chloroplast genome, the software SPAdes 3.15.1-Linux was employed, utilizing the default parameters (Prjibelski et al., 2020). To evaluate the assembly quality, the circular maps were identified using Bandage software (Wick et al., 2015). Subsequently, the resulting assembly was annotated using PGA (Qu et al., 2019), referencing the chloroplast genome sequence of Eschenbachia blinii (H.Lév.) (NC 037605.1). The annotation results were then checked using Geneious R11 (Kearse et al., 2012). The chloroplast genome map was visualized using OGDRAW (https://chlorobox.mpimp-golm.mpg.de/OGDraw.html). Additionally, the tRNA sequences were validated using tRNAscan-SE v2.0 (Chan et al., 2021), available on the Geseq platform (https://chlorobox.mpimp-golm.mpg.de/geseq.html). The annotated chloroplast genomes have been submitted to GenBank ( Table 2 ). Analysis of plastid information was conducted using Geneious R11.

To identify potential IR expansion or contraction in eight Aster species, the reference species was used from A. ageratoides. This analysis was conducted using the perl script of Irscope (Amiryousefi et al., 2018). With the A. albescens as reference, the homology of these sequences was visualized using the mVISTA program (Frazer et al., 2004, https://genome.lbl.gov/vista/mvista/submit.shtml) with the LAGAN mode (Brudno et al., 2003).

In this study, the identification of direct (forward), inverted (palindromic), complement, and reverse repeats elements was identified by REPuter (Kurtz et al., 2001), with maximum computed repeats equal to 50 bp, hamming distance of 3, and minimal repeat size of 30 bp. Furthermore, the detection of simple sequence repeats (SSRs) within the complete chloroplast genomes was performed using Microsatellite (MISA) (Beier et al., 2017). The thresholds for SSR detection were set to 10, 5, 4, 3, 3, and 3, for mono-, di-, tri-, tetra-, penta-, and hexa-nucleotides, respectively.

The alignment of all sequences from the eight Aster species was performed using the “–auto” strategy of Multiple Alignment using Fast Fourier Transform (MAFFT). Nucleotide diversity was then calculated using a sliding window approach in DnaSP v.6.12.03 (Rozas et al., 2017) with a window length of 600 bp and a step size of 200 bp.

MEGA v 7.0 was used to analyze the synonymous codon usage and the relative synonymous codon usage (RSCU) of the Aster cp genomes. RSCU values >1 represent frequently used codons than expected, whereas values <1 signify the opposite. Codons having no preference value are set to 1.00.

The phylogenetic analysis of the complete chloroplast genomic dataset, consisting of 27 species of Astereae, was performed using the maximum likelihood (ML) method implemented in RAxML. The species of Nannoglottis ravida and Llerasia caucana from basal group of Astereae were selected as outgroups (Stamatakis et al., 2008). The analysis was performed on the CIPRES platform (Miller et al., 2010) (https://www.phylo.org/portal2/). ModelTest (Posada, 2006) was employed to determine the most suitable model for the dataset. The molecular model GTRCAT was applied for the analysis. For bootstrap support assessment, the fast bootstrap option with 1,000 replicates was utilized in RAxML from CIPRES platform. The morphological identification characteristics of the genus Aster and its related genera have been described more clearly by Chen et al. (2011). Therefore, we define the key to the Aster and related species with reference to the criteria proposed by Chen et al. (2011) in combination with classification of previous studies (Li et al., 2012; Jafari et al., 2015; Korolyuk et al., 2015).

For divergence time estimation, we used the complete chloroplast sequence dataset. The BEAST v.1.8 (Drummond et al., 2012) was applied to estimate the divergence times with Bayesian uncorrelated lognormal relaxed clock model. The node of Astereae was set at 31.42 Mya according to Zhang C. F. (Zhang C. F. et al. 2021). The tree Yule model was selected. The Markov chain Monte Carlo (MCMC) was run for 10,000,000 generations and sampled every 1,000 generations. TreeAnnotator v. 1.6 (BEAST package) was used to summarize and annotate the tree, with the initial 10% of trees discarded as burn-in. Finally, the tree was visualized in the program Figtree v.1.4.4 (http://tree.bio.ed.ac.uk/) with 95% highest posterior density being shown.

In this study, the complete chloroplast genomes of the eight species of Aster were sequenced and assembled. The results revealed a high degree of conservation in the structures of these genomes ( Figure 1 ). These chloroplast genomes exhibited the standard quadripartite structure, consisting of a LSC region, a SSC region, and a pair of IR regions (IRa and IRb). The size of these genomes varied from 152,045 bp (A. polius) to 152,729 bp (A. albescens) ( Table 2 ). The GC content ranged from 37.27% (A. procerus) to 37.35% (A. polius). Overall, all chloroplast genomes have 133 genes except A. procerus and A. pycnophyllus having 132 genes, including 87 protein-coding genes, 37/38 tRNA genes, and eight rRNA genes. Additionally, 115 of these genes were unique and 18 genes were duplicated in the IR regions ( Table 3 ). The arrangement of these 133 genes in all chloroplast genomes was found to be completely collinear. There were two introns of four genes (rps12, rps12, ycf3, and clpP) and single intron of 16 genes (ndhA, ndhB, petB, petD, atpF, rbcL, rpl16, rpl2, rps16, rpoC1, trnA-UGC, trnG, trnI-GAU, trnK-UUU, trnL-UAA, and trnV-UAC). The gene rps12 was trans-spliced, and the genes ndhD and psbL experienced RNA editing.

In general, the IR/Single Copy (SC) expansion and contraction might cause the IR/SC junction position change. The IR/SC borders of the eight newly sequenced Aster chloroplast genomes were compared to analyze the expansion and contraction variation in junction regions ( Figure 2 ). Although overall genomic structure including gene order and gene number was well conserved, these genomes exhibited slight differences at four junctions (JLB, JSB, JSA, and JLA). The rps19 gene of all Aster species located the JLBs, with the IRa region including 60 bp to 62 bp, except for A. pycnophyllus (36 bp). Likewise, the JLAs of all Aster species were located between rps12 and trnH. The ndhF gene, related to photosynthesis, was entirely located in the SSC region and the distance to the junction ranged from five to 54 bp. In our newly sequenced genomes, the ycf1 pseudogene was identified in all newly sequenced genomes. The main part of ycf1 gene was in the SSC region, with other 564 bp to 567 bp in the IRa region. The same fragment was also found in the IRb region of the ycf1 pseudogene and extended to SSC region with extension region with 9 bp to 147 bp.

In the SSR analysis of the six species of Astereae, 75 (A. yunnanensis) to 99 (A. pycnophyllus) SSRs were found, showing a similar number of SSRs in Astereae ( Figure 3A ). In addition, these detected SSRs can be divided into six types, including mononucleotides (38%), dinucleotides (18.4%), trinucleotides (19.7%), tetranucleotides (17.9%), pentanucleotides (5.7%), and hexanucleotides (0.3%) ( Figure 3B ). The hexanucleotide repeats were only found in the chloroplast genomes of A. falcifolius and A. prorerus. The four dominant motif types of these SSRs were A/T (28–38), AT/AT (15–17), AAT/ATT (11–18), and AAAT/ATTT (7–10) ( Figure 3C ).

The forward, palindromic, complement, and reverse repeats were detected in the eight newly sequenced chloroplast genomes ( Figure 4A ). A. procerus, A. pycnophyllus, A. yunnanensis, and A. falcifolius had all four type repeats. A. polius, A. albescens, A. argyropholis, and A. lavandulifolius had forward, palindromic, and reverse repeats. On average, 46–49 repeat sequences were identified in these genomes, with 17–23 forward repeats, 19–25 palindromic repeats, and 1–8 reverse repeats. However, complement repeats were only detected in A. procerus, A. pycnophyllus, A. yunnanensis, and A. falcifolius, with number of 1 to 3. Moreover, the repeats with 30 bp to 39 bp in length were the most common type in these genomes ( Figure 4B ), and none of the repeats with 50 bp to 59 bp in length. These interspersed repeat sequences were mainly present in the intergenic spacers, and several were observed within the coding regions and introns. The ycf15, rps12, ycf2, rrn5, rrn4.5, psbN, trnG, trnT-GGU, ycf4, cemA, trnS-GCU, trnS-UGA, trnS-GGA, psaB, psaA, accD, psal, psbE, petL, ndhD, and psaC genes contained LDRs. The interspersed repeat sequences were also more commonly detected in LSC and IR than SSC regions. The overall distribution of interspersed repeat sequences was similar in both IR regions.

The mVISTA analysis of these eight Aster species indicated the complete chloroplast genome shared high levels of sequence similarity. Genetic variability was more prevalent in the non-coding regions than in the coding regions. The five genes with the highest variation were matK, atpA, rps19, ycf2, and ycf1 ( Figure 5 ). DnaSP analysis revealed nucleotide diversity in single copy genes and intergenic regions with nucleotide diversity (Pi) ranged from 0.00068 to 0.04577. Six mutation hotspots showed significantly high Pi values (π > 0.014) ( Figure 6 ), much higher than the average pi value (π = 0.0038). Among the gene coding regions, the highest Pi values were found in ndhF, followed by ndhC, trnV (UAC), and trnM (CAU). Among intergenic regions, the highest Pi values were detected in the rpl12-ndhF region, followed by ndhC–trnV (UAC), trnV (UAC)–trnM (CAU), and trnM (CAU)–atpE ( Figure 6 ). We analyzed the nucleotide diversity of the complete chloroplast genomes and the LSC, SSC, and IR regions. The nucleotide diversity in the complete chloroplast genome was 0.0038, and higher nucleotide diversity was found in the LSC and SSC regions than the IR region, showing that the IR regions were more conserved than the single-copy regions. We found only two regions with π > 0.02, i.e., trnT (GGU)–psbD and trnL (UAA)–trnF (GAA), and three regions with π>0.015 and <0.02, i.e., trnU (UAC)–trnM (CAU), accD, and ycf4/ycf4-cemA ( Figure 6 ).

The preferences for codon are extremely similar among species. The analyses showed that 87 protein-coding genes were encoded by 64 codons (including three are stop codons: UAA, UGA, and UAG; Figure 7 ). The most prevalent amino acid was leucine. Leucine was encoded by CUA, CUC, CUG, CUU, UUA, and UUG with 2,420 codons (A. falcifolius) to 2,440 codons (A. yunnanensis). However, the rarest one was cysteine. Cysteine was encoded by UGC and UGU with 251 codons (A. falcifolius) to 252 codons (A. polius). In the complete chloroplast genome of these Aster, only codons tryptophan (encoded by UGG) exhibited no bias with RSCU = 1.00. The common start codon for the protein coding genes was AUG (M), except for the psbL, rps19, and ndhD genes, which have start codons of ACG, GUG, and GUG in all species, respectively.

In this study, the complete chloroplast genomes of 27 Astereae species were used to perform phylogenetic reconstruction, with Nannoglottis ravida (C.Winkl.) Y.L.Chen and Llerasia caucana (S.F.Blake) Cuatrec used as outgroup. Phylogenetic analyses of the supermatrix of 25 taxa (not include outgroups) using the ML methods yielded a topology ( Figure 8 ) with in-group fell into five clades: clade A, clade B, clade C, clade D, and North American clade ( Figure 8 ). Clade A was the largest clade and was strongly supported (Bootstrap value (BP) = 97/100), containing 12 Aster species and two Heteroplexis species. The newly sequenced species, namely, A. falcifolius, A. pycnophyllus, and A. procerus, were nested in clade A. Other four newly sequenced species, namely, A. argyropholis, A. albescens, A. lavandulifolius, and A. polius, together with Aster hypoleucus Hand.-Mazz. formed the strongly supported clade B. A. hersileoides lonely formed clade C, as the sister group of clade B with a moderate support (BS = 68). Symphyotrichum subulatum (Michx.) G.L. Nesom and Erigeron canadensis L. formed the North American clade, and it was a sister group of clades A, B, and C with a high support (BS = 100). The newly sequenced A. yunnanensis were placed together with A. flaccidus and A. batangensis in clade D with strong support (BS = 100).

On the basis of the newly reconstructed phylogeny, the origin and divergence times of lineages within the genus Aster were estimated ( Figure 9 ). Divergent time estimate showed that the divergent time of clade A was dated back to 23.15 Mya. clades B, C, and D were divergent from 15.13 Mya, 24.29 Mya, and 21.66 Mya, respectively.

The structure, gene position, size, orientation, and gene content of the plastid genomes of the eight Aster species were highly conserved (Doorduin et al., 2011; Curci et al., 2015; Cheon et al., 2017; Chen et al., 2018). These genomes of Aster have a standard quadripartite structure, including a LSC, a SSC, and a pair of IRs (IRa and IRb), which was the same as that reported for most other Aster (Shen et al., 2018; Wang et al., 2019; Zhang X. et al., 2021). The sizes of plastomes of the eight Aster species are between 152,045 bp and 152,729 bp ( Table 2 ). In addition, these plastomes did not have any loss of introns. Additionally, the GC contents, which are a crucial factor in genome organization and stability, of the chloroplast genomes were low (37.3%) and were similar to that of other Asteraceae species, such as Aster spathulifolius Maxim. (37.28%) and A. hypoleucus (37.3%) (Ravi et al., 2008; Tyagi et al., 2019; Wang et al., 2019). We detected losses of the trnT (GGU) gene in A. procerus and A. pycnophyllus. In previous study, the loss of the tRNA was detected in some Asteraceae species (Lee et al., 2017).

The IR regions are known to be highly conserved in the genome of chloroplasts. During evolution, the expansion and contraction of the IR, LSC, and SSC regions are common, which leads to variability in genome length (Kim and Lee, 2004). In this study, the examination of chloroplast genome variation ( Figure 2 ) showed that great expansion or contraction of the chloroplast IR region was not detected. However, ycf1 and ndhF genes located at the SSC/IR border had the slight variation in position and length in the eight Aster chloroplast genomes, suggesting boundary contraction and expansion between the SSC/IR regions in Aster (Liu et al., 2018).

The SSRs are effective molecular markers, and they are often used for species identification and population genetic analyses (Thiel et al., 2003). In the eight Aster species analyzed here, A/T repeats, AT/AT repeats, AAT/ATT repeats, and AAAT/ATTT repeats were commonly detected ( Figure 3C ). This phenomenon may be related to that the AT preference pattern is widely reported in many plant plastids (Somaratne et al., 2019). In the rearrangement of the complete chloroplast genomes and sequence divergence, larger and more complex repeat sequences may play an important role (Weng et al., 2014). The interspersed repeat sequences were more prevalent in the non-coding regions than the coding regions (Kim et al., 2015). In our study, the ycf2 gene includes rich repeats, which contained many repeats: forward and palindromic. This result was consistent with the previous analysis that showed the gene has already been shown to be associated with many evolutionary events (Huang et al., 2010).

DNA barcoding technology has been widely used in the species identification, phylogeny, and evolution (Doorduin et al., 2011; Palazzesi et al., 2022). In mVISTA analysis, the matK, atpA, rps19, ycf2, and ycf1 genes had large differences and were putative markers for population genetic and barcoding analyses ( Figure 5 ). Among these genes, the matK and ycf1 genes have been used in previous plant phylogenetic and DNA barcoding analyses for land plants (Dong et al., 2015). Some regions of the plastomes of the eight Aster species showed high sequence divergence and might be used for phylogenetic reconstruction. However, these regions are different from the phylogenetic markers previously reported for Asteraceae (Do et al., 2019). Therefore, the complete chloroplast genome sequences and molecular markers might provide fundamental data for further studies on genus of Aster and related species in tribe Astereae.

Thirty-one codons with RSCU value >1 were found, indicating that these codons are preferentially used in coding amino acids. An identical trend was discovered among the eight species ( Figure 7 ). Leucine was the most abundant amino acid, whereas the cysteine was the least abundant amino acid, which is consistent with other Asteraceae species (Salih et al., 2017; Shen et al., 2017). In addition, most of A/U-ending codons had RSCU values >1; meanwhile, most of G/C-ending codons had RSCU values <1, indicating that amino acids tended to using A/U-ending codons, similar to a previous study (Zhao et al., 2021).

In the tribe Astereae, there are numerous morphologically similar but distantly related taxa, such as some species of Aster (Noyes and Rieseberg, 1999). Whether some taxa should remain as genera or be merged into a single genus remains to be determined, such as genus Kalimeris, Heteropappus, and Heteroplexis, as wells as Aster series Albescentes, Aster Ser. Hersileoides, and Aster section Alpigenia (Nesom, 1994b; Noyes and Rieseberg, 1999; Li et al., 2012; Jafari et al., 2015; Korolyuk et al., 2015; Nesom, 2020). Nesom (1994b) proposed that Aster series Albescentes should be removed from Aster. In the Flora of China, Aster series Albescentes species A. nitidus and A. hersileoides were treated as the unplaced Aster group. The molecular phylogeny of Li et al. (2012) suggested that Aster section Alpigenia should be elevated to the new genera, series Albescentes is considered to be more closely related to section Alpigenia, and the Aster series Hersileoides is a well-supported monophyletic group. Therefore, according to the results of previous studies and the phylogenetic tree of this study, we classified the 25 species (not including outgroups) into five clades: clade A (core Aster), clade B (Aster series Albescentes), clade C (Aster Ser. Hersileoides), clade D (Alpine Aster, Aster section Alpigenia), and North American clade. Besides, the phylogenetic analysis of complete chloroplast genomes provided strong supports for these five clades (clade A, BS = 100; clade B, BS = 100; clade C, BS = 100; North American Clade, BS = 100; and clade D, BS = 100).

The divergence time estimation of Aster relied on secondary calibration because of the lack of fossil record for most Aster taxa. Most species of Aster and its related genera are distributed in East Asia (Nesom, 1994b; Brouillet et al., 2009; Chen et al., 2011). The result of Brouillet et al. (2009) indicated that Aster originated from a clade with a dispersal from Australasia into East Asia. The result of our molecular dating suggests that clades A and C began to diversify in the late Oligocene (23.15 Mya and 24.29Mya, respectively). Clades B and D originated in the Early Miocene (15.13 Mya and 21.66Mya, respectively). The rapid radiation may be related to collisions between geological plates (Audley-Charles, 1987; Liu et al., 2002). Geologic uplift events (first of which began at about 50 Ma) have taken place in the Tibetan Plateau during at least four different periods since the early Miocene, i.e., 22 Mya, 15–13 Mya, 8–7 Mya, and 3.5–1.6 Mya (Shi et al., 1998; Guo et al., 2002; Spicer et al., 2003). The origin of the four clades (Clade A-D) likely occurred independently at first two stages of the uplift and formation of the Tibetan Plateau. Geological evidence suggests that the strong uplift of the Tibetan Plateau, coupled with favorable oceanic and continental environments, produced a strong Asian monsoon dominated by the Summer Monsoon (Shi et al., 1998; Ding et al., 2020). During this uplift movement, the original Planetary Wind System in East Asia was changed and the arid zone retreated to the northwest (Shi et al., 1998; Ding et al., 2020). Eastern China was gradually covered by tropical or subtropical forests. This scenario also correlates with the current habitat preferences of the studied taxa (clade A, understorey vegetation; clades B and C, dry slopes and scree regions; and clade D, cold and dry alpine meadows). The similar rapid radiation has also been found in other groups of Asteraceae in the Tibetan Plateau, such as Saussurea (Liu et al., 2002) and the Dolomiaea-Diplazoptilon-Xanthopappus group (Wang et al., 2007).