Comparative chloroplast genome analysis of Ardisia (Myrsinoideae, Primulaceae) in China and implications for phylogenetic relationships and adaptive evolution

Abstract

Background

Numerous species of Ardisia are widely used for their medicinal and ornamental values in China. However, accurately identifying Ardisia species at the molecular level remains a challenge due to the morphological similarities among different species, the complexity of interspecific variation, and the limited availability of genetic markers. In this study, we reported 20 chloroplast genomes of Ardisia species from China and combined them with 8 previously published chloroplast genomes to conduct a comprehensive analysis for phylogenetic relationships and adaptive evolution.

Results

For the 28 Ardisia species analyzed in this study, the size of the chloroplast genomes ranged from 155,088 bp to 156,999 bp, and all exhibited a typical tetrad structure with conserved gene content and number. Each genome contained 85–88 protein-coding genes, 36–37 tRNA genes, and 8 rRNA genes. Comparative analysis showed that the genomic structures and gene order were relatively conserved with slight variations in the inverted repeat regions (IRs). Simple sequence repeats (SSRs) were predominantly single nucleotide repeats, while repeat sequences were mainly composed of palindromic and forward repeats. Twelve highly variable regions were identified as potential DNA barcodes for species identification and phylogenetic analysis of Ardisia. The phylogenetic tree supported the division of the subgenus Bladhia s.l. into two subgenera: Bladhia s.str. and Odontophylla (Yang) Huang. Further investigation revealed that two protein-coding genes (rbcL and rpoC2) were under positive selection and might be associated with the adaptation of Ardisia species to shaded environments.

Conclusion

Our study analyzed the chloroplast genomes of 20 Ardisia species from China to explore their phylogenetic relationships and adaptive evolution. By combining these results with data from eight previously published chloroplast genomes, the essential characteristics of Ardisia chloroplast genomes were clarified. The research establishes a theoretical basis for the classification, identification, and comprehension of the adaptive evolution of Ardisia species.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-024-05892-x.

Background

Ardisia, the largest genus of Myrsinoideae (Primulaceae), is primarily found in tropical and subtropical regions and consists of over 700 accepted species [1]. The classification of Ardisia species in China follows the taxonomic system described in the Flora of China, including six sections: Sect. Pimelandra, Sect. Acrardisia, Sect. Tinus, Sect. Akosmos, Sect. Crispardisia, and Sect. Bladhia [2]. These species are primarily found in the southern regions of the Yangtze River, with approximately 65 recorded species and 12 varieties [3]. Ardisia species often used as ornamental plants in China are typically small trees or shrubs with brightly colored red fruits. In addition to their ornamental uses, several Ardisia plants, including A. japonica, A. crenata, A. gigantifolia, and A. crispa, have been used as traditional folk medicine herbs in southern China since ancient times, with various medicinal purposes, such as alleviating cough and phlegm, promoting blood circulation, reducing fatigue, and reducing swelling [4, 5]. Aidicha (the whole dried plant of A. japonica) and Zhushagen (the dried root of A. crenata) are included in the Pharmacopoeia of the People’s Republic of China (2020) (https://db.ouryao.com/yd2020/). Currently, Ardisia plants are used to produce several products, including compounded Aidicha tablets, Aidicha capsules, and Zhushagen dispensing granules, which are widely utilized in clinical applications in China [6]. Modern pharmacological studies have identified various chemical compounds in Ardisia plants, such as bergenin, ardisicrenoside, benzoquinone, triterpenes, and flavonoids [7, 8].

The morphological and clinical similarities between different Ardisia species often result in confusion and errors in taxonomic records, making accurate species identification challenging [9]. Ensuring the safety and quality of raw materials of Ardisia plants holds paramount importance in preserving the authenticity and efficacy of herbal products within the pharmaceutical supply chain. Four DNA barcodes (ITS, psbA-trnH, rbcL, and matK) were evaluated for Chinese Ardisia species, using a sample of 121 individuals from 33 species, and the results showed that the ITS fragment had a higher identification rate compared to the other three barcodes [10]. However, the accuracy of species identification using the ITS fragment was below 85%, indicating the need for further improvement in DNA barcoding techniques for Ardisia species. Additionally, the phylogenetic relationships between Asian Ardisia species and its relatives in the Myrsinoideae (Primulaceae) were analyzed using ITS, psbA-trnH and rpl32-trnL sequences [1]. The relationships among the subgenera remained poorly understood due to the low support rate of the inferred phylogenetic tree, highlighting the requirement for more powerful molecular markers.

In recent years, advances in sequencing assembly and annotation technology have significantly reduced the cost of chloroplast genome sequencing. Consequently, more and more chloroplast genome data has been successfully applied to plant phylogeny and evolutionary studies, including the reconstruction of phylogenetic relationships of plants [11]. The complete chloroplast genome along with its derived barcode has emerged as an ideal tool for species identification of economic plants. However, the number of published chloroplast genome sequences of Ardisia species is limited currently [12–14]. Comparative analysis of chloroplast genomes using a larger number of Ardisia species would greatly contribute to understand chloroplast genome evolution and reconstruct the phylogenetic relationships of Ardisia. The highly variable region of the chloroplast genome could provide potential genetic markers for species identification and phylogenetic analysis of Ardisia.

Adaptive evolution is considered to enhance the fitness of species to constantly changing environmental conditions [15]. Protein-coding genes of chloroplast genome often undergo adaptive evolution, for example, genes such as rbcL, ycf1, and accD, have been positively selected and are significantly correlated with environmental adaptations, including temperature, light, humidity, and atmospheric conditions [16]. In analysising six chloroplast genomes of Chrysosplenium (Saxifragaceae), 19 genes under positive selection were screened out, and most of them were involved in photosynthesis, which may be the adaptive response to shady and moist habitat [17]. Ardisia plants are also typically found in shaded habitats, such as the understory or near valleys and streams. These plants serve as an ideal model group for studying plant adaptations to low light conditions. Comparative chloroplast genome analysis of Ardisia species might provide insight into the effects of low light for angiosperms and enhance our understanding of the evolution of Ardisia species.

In this study, we sequenced the complete chloroplast genomes of 20 Ardisia species and conducted a comprehensive analysis along with 8 additional chloroplast genome data from GenBank. The aims of this study were: (a) to perform a comparative analysis of the structural characteristics of chloroplast genomes; (b) to identify highly variable regions for species identification and phylogenetic studies; (c) to conduct a phylogenetic analysis of Ardisia species in China; and (d) to examine the adaptive evolution of protein-coding genes. The complete genomes of the Ardisia species will serve as a theoretical foundation for the classification, identification, and understanding of the adaptive evolution of Ardisia species.

Results

General features of chloroplast genomes

In each Ardisia species, the chloroplast genome consisted of a typical tetrad structure, which included one small single-copy (SSC), one large single-copy (LSC), and two inverted-repeat (IR) regions (Fig. 1). Among the 28 Ardisia species, A. bullata had the lowest GC content (36.00%), while A. gigantifolia had the highest GC content (37.30%). The lengths of the chloroplast genomes of the 28 Ardisia species ranged from 155,088 bp to 156,999 bp, including the SSC region (18,093 − 18,479 bp), LSC region (84,709 − 86,989 bp), and two IR regions (25,411 − 26,236 bp each) (Table 1). Each chloroplast genome of Ardisia species contained 85–88 protein-coding genes, 36–37 tRNA genes, and 8 rRNA genes (Table 1). Among these genes, 12 protein-coding genes and 6 tRNA genes contained introns, three of these genes (clpP, rps12, and ycf3) contained two introns, while the others contained only one intron (Table 2).

Fig. 1.

Gene map of the A. argenticaulis chloroplast genome. Genes located outside the circle are transcribed counterclockwise, while genes inside the circle are transcribed clockwise. In the inner circle, the dark grey area represents GC content of the cp. genome, and the light grey area represents the AT content. Different color blocks represent genes that belong to different functional groups

Table 1.

Summary of the chloroplast genomesof 28 Ardisia species

Species	Genome Length (bp)	LSC Length (bp)	SSC Length (bp)	IR Length (bp)	GC (%)	Total Genes	CDS	t RNA	rRNA
A. argenticaulis	156,940	86,068	18,400	26,236	37.03%	131	87	36	8
A. balansana	155,307	85,673	18,342	25,646	37.21%	132	88	36	8
A. bullata	155,088	84,709	18,367	26,006	36.00%	132	88	36	8
A. carnosicaulis	156,548	86,102	18,346	26,050	37.05%	131	87	36	8
A. crenata	156,540	86,093	18,347	26,050	37.10%	132	87	36	8
A. crispa	156,709	86,300	18,411	25,999	37.06%	131	87	36	8
A. faberi	156,104	86,989	18,293	25,411	36.96%	130	85	37	8
A. fordii	156,999	86,131	18,404	26,232	37.02%	132	88	36	8
A. gigantifolia	156,216	85,725	18,397	26,047	37.30%	131	87	36	8
A. japonica	155,787	86,715	18,222	25,457	37.05%	130	85	37	8
A. lindleyana	156,741	86,359	18,380	26,001	37.06%	131	87	36	8
A. maclurei	155,751	86,725	18,104	25,467	37.02%	129	85	36	8
A. mamillata	156,757	86,325	18,434	25,999	37.10%	131	87	36	8
A. merrillii	156,746	86,351	18,387	26,004	37.08%	131	87	36	8
A. nutantiflora	156,542	86,225	18,379	25,969	37.22%	132	88	36	8
A. obtusa	156,626	86,168	18,346	26,056	37.07%	131	87	36	8
A. omissa	156,761	86,335	18,428	25,999	37.05%	132	87	37	8
A. pedalis	156,722	86,301	18,405	26,008	37.05%	132	87	37	8
A. perreticulata	156,953	86,084	18,413	26,228	37.03%	131	87	36	8
A. polysticta	156,506	86,078	18,328	26,050	37.07%	131	87	36	8
A. primulifolia	156,728	86,318	18,402	26,004	37.05%	131	87	36	8
A. pseudocrispa	156,744	86,354	18,308	25,999	37.04%	131	87	36	8
A. pusilla	155,749	86,677	18,222	25,425	37.05%	129	85	36	8
A. quinquegona	156,766	85,900	18,408	26,229	37.05%	131	87	36	8
A. replicata	156,278	86,012	18,358	26,196	37.20%	132	88	36	8
A. sieboldii	156,923	85,982	18,479	25,954	37.04%	132	88	36	8
A. solanacea	156,518	86,033	18,093	26,231	37.14%	132	88	36	8
A. villosa	156,720	86,353	18,339	26,014	37.07%	131	87	36	8

LSC large single-copy, SSC small single-copy, IR inverted repeat

Table 2.

Genes in the chloroplast genome of 28 Ardisia species

Category	Gene group	Gene name
Protein synthesis and DNA-replication	Ribosomal RNA genes	rrn4.5, rrn5, rrn16, rrn23
	Transfer RNA genes	trnA-UGC*, trnC-GCA, trnD-GUC, trnE-UUC, trnF-GAA, trnG-GCC, trnG-UCC*, trnH-GUG(1, 2, 3, 4), trnI-CAU, trnI-GAU*, trnK-UUU*, trnL-CAA, trnL-UAA*, trnL-UAG, trnM-CAU, trnfM-CAU, trnN-GUU, trnP-UGG, trnQ-UUG, trnR-UCU, trnR-ACG, trnS-GCU, trnS-GGA, trnS-UGA, trnT-GGU, trnT-UGU, trnV-GAC, trnV-UAC*, trnW-CCA, trnY-GUA
	Ribosomal protein genes (larger subunit)	rpl2*, rpl14, rpl16*, rpl20, rpl22, rpl23, rpl32, rpl33, rpl36
	Ribosomal protein genes (smaller subunit)	rps2, rps3, rps4, rps7, rps8, rps11, rps12**, rps14, rps15, rps16*, rps18, rps19
	RNA polymerase	rpoA, rpoB, rpoC1*, rpoC2
Photosynthesis	Photosystem I	psaA, psaB, psaC, psaI, psaJ
	Photosystem II	psbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN, psbT
	Cytochrome b/f complex	petA, petB*, petD*, petG, petL, petN
	ATP synthase	atpA, atpB, atpE, atpF*, atpH, atpI
	Rubisco large subunit	rbcL
	NADH dehydrogenase	ndhA*, ndhB*, ndhC, ndhD, ndhE, ndhF, ndhG, ndhH, ndhI, ndhJ, ndhK
Miscellaneous group	ATP-dependent protease	clpP**
	Maturase	matK
	Acetyl-CoA carboxylase	accD
	Cytochrome c biogenesis	ccsA
	Inner membrane protein	cemA
	Translation initiation factor	infA
Pseudogene unknown function	Hypothetical chloroplast reading frames (ycf)	ycf1, ycf2, ycf3**, ycf4, ycf15[2, 5, 6, 7]

*—Gene containing a single intron; **—Gene containing two introns; ()—Gene exists in some species; []—Gene do not exist in some species. Species—1, A. fordii; 2, A. japonica; 3, A. omissa; 4, A. pedalis; 5, A. faberi; 6, A. maclurei; 7, A. pusilla

Boundary regions analysis

There are four boundaries between two inverted repeats (IR), large single copy (LSC), and small single copy (SSC) regions in the chloroplast genome, known as the IRb-LSC boundary (JLB line), IRb-SSC boundary (JSB line), IRa-SSC boundary (JSA line), and IRa-LSC boundary (JLA line). The chloroplast genome structures were conserved in 28 Ardisia species (Fig. 2). At the IRb-LSC boundary, the JLB line was located in the rps19 gene region of 28 Ardisia species. At the IRb-SSC boundary, the JSB line was located in the ndhF gene region of 23 Ardisia species, while the JSB line in 5 other species was 27–75 bp away from the ndhF gene. At the IRa-SSC boundary, the JSA line was located in the ycf1 gene region of 27 Ardisia species. In A. pseudocrispa, the JSA line was 170 bp away from the ycf1 gene, and the ycf1 gene was only 4,430 bp and located in the SSC region. At the IRa-LSC boundary, the JLA line was located between the rpl2 and psbA genes in 28 Ardisia species. Notably, it was observed that one of the two copies of rps19 gene was pseudogene in A. fordii, A. sieboldii, A. solanacea, A. replicata, A. nutantiflora, A. balansana, and A. bullata, which also presented at the LSC-IRb boundary. In A. faberi, A. japonica, and A. pedalis, trnH gene was located in the LSC region, 2–10 bp away from the JLA line.

Fig. 2.

Comparison of the borders of the LSC, SSC, and IR regions among chloroplast genomes of 28 Ardisia species. JLB, JSB, JSA, and JLA denote the junction sites of LSC/IRb, IRb/SSC, SSC/IRa, and IRa/LSC respectively

Repeat sequence analysis

In this study, we employed Reputer software to analyze the repeat sequences in the chloroplast genomes of 28 Ardisia species. Each species displayed 33–66 repeat sequences, with palindromic repeats being the most prevalent (17–35 per species, accounting for 51.29% of total repeats), followed by forward repeats (15–30 per species, making up 45.65% of total repeats), reverse repeats (1–3 per species, constituting 2.72% of total repeats), and complementary repeats (0.33% of total repeats) being the least common (Fig. 3A, Table S2). Complementary repeat sequences were only found in five species (A. japonica, A. maclurei, A. mamillata, A. omissa, and A. pusilla), each with one such sequence. A. balansana had the highest number of palindromic repeats (35 sequences), followed by A. argenticaulis and A. omissa (30 sequences each), and A. replicata with the lowest count of 17 sequences. In terms of forward repeats, A. balansana, A. japonica, A. pusilla, and A. faberi showed the highest numbers (30 sequences each), while A. maclurei, A. mamillata, A. omissa, and A. pusilla displayed the fewest. A. bullata and A. replicata had the lowest number of forward repeats (15 sequences each), whereas A. carnosicaulis, A. crenata, and A. polysticta had the highest count of reverse repetitive sequences, each with three sequences (Fig. 3B, Table S2).

Fig. 3.

Analyses of repeat sequences. A Box plot showing the distribution of four different types of repeats among 28 Ardisia species. B Histogram showing the number of repeats in the chloroplast genomes of 28 Ardisia species

Microsatellites, also known as simple sequence repeats (SSRs), are continuous 1–6 bp nucleotide repeat units found in chloroplast genomes. A statistical analysis of 28 Ardisia species revealed a total of 1,669 SSRs, with each species containing between 54 and 72 SSRs (Table S3). Mononucleotide repeats accounted for 76.21% of all SSRs, with repeat numbers ranging from 39 to 56 among species. In addition, there were 157 dinucleotide repeats, 15 trinucleotide repeats, 208 tetranucleotide repeats, 14 pentanucleotide repeats, and 3 hexanucleotide repeats, representing 9.41%, 0.90%, 12.46%, 0.84%, and 0.18% of all SSRs, respectively. The repeat numbers for each type ranged from 5 to 7, 0 to 1, 6 to 9, 0 to 1, and 0 to 1, respectively (Table S3). Among the species, A. maclurei exhibited the highest number of mononucleotide repeats (56), while A. bullata and A. quinquegona had the most dinucleotide repeats (7 each), and A. faberi showed the highest count of tetranucleotide repeats (9). Trinucleotide and pentanucleotide repeats were only observed in 15 and 14 Ardisia species, respectively, with a single repeat identified in each species. Furthermore, only one hexanucleotide repeat was detected in the chloroplast genomes of A. bullata, A. gigantifolia, and A. nutantiflora (Fig. 4, Table S4).

Fig. 4.

Distribution maps of simple sequence repeats (SSRs) in the chloroplast genomes of 28 Ardisia species. A Box plot showing distribution of six SSR types among 28 Ardisia species. B Classification of SSRs in 28 Ardisia species by repeat type: mono-, mononucleotides; di-, dinucleotides; tri-, trinucleotides; tetra-, tetranucleotides; penta-, pentanucleotides; and hexa-, hexanucleotides

Comparative genomic analysis

Using the Mauve Multiple Genome Comparison method, this study investigated rearrangements and colinearities in the chloroplast genomes of 28 Ardisia species (Fig. S2). The analysis identified two Locally Collinear Blocks (LCBs), indicating a high degree of similarity among the species. Nucleotide diversity (π) was also assessed using sliding window analysis with DnaSP to identify hotspot regions in the chloroplast genomes of Ardisia species (Fig. 5). The results revealed that intergenic regions displayed significantly higher levels of polymorphism compared to protein-coding regions. Specifically, the SSC region showed the highest nucleotide variation (average π = 0.00699), followed by the LSC region (average π = 0.00551) and the IR region (average π = 0.00271). In terms of protein-coding genes, infA, rpl22, and ycf1 exhibited higher π values (> 0.008) compared to other protein-coding genes. Nine intergenic regions (rps16-trnQ, trnG-trnR, trnT-psbD, ycf3-trnS, trnT-trnL, atpB-rbcL, petG-trnW, trnL-ndhB, and rpl32-trnL) exhibited high π values (π > 0.008).

Fig. 5.

Nucleotide diversity (π) of shared various regions in chloroplast genomes of 28 Ardisia species

Phylogenetic analysis

To assess the phylogenetic relationships among Ardisia species in China, 30 complete chloroplast genomes including the 28 Ardisia species and 2 outgroup species (Tapeinosperma multiflorum and T. netor), were used for phylogenetic analysis (Fig. 6, Table S1). Phylogenetic trees were constructed utilizing both maximum likelihood (ML) and Bayesian inference (BI) methods, considering five different data sets: complete chloroplast genome sequences, large single-copy (LSC) regions, inverted repeat (IR) regions, small single-copy (SSC) regions, and CDS datasets (Fig. 6, Fig. S3, Fig. S4, Fig. S5, Fig. S6). Analyses of the phylogenetic data from various datasets revealed subtle differences in topology, with phylogenetic trees based on complete chloroplast genomes showing the highest support. The results based on the complete chloroplast genome indicated that the 28 Ardisia species formed a monophyletic clade (BS = 100, PP = 1) which further divided into five smaller clades (Fig. 6). Among them, the subgenus Crispardisia was sister to the subgenus Akosmos, with both being sister to a group that included the species A. solanacea from the subgenus Tinus. Subgenus Odontophylla was sister to a monophyletic clade containing all remaining subgenera. Additionally, the subgenus Bladhia was identified as basal and sister to all other subgenera.

Fig. 6.

The phylogenetic tree is based on 30 complete chloroplast genome sequences using Bayesian inference (BI) and Maximum likelihood (ML) analyses. The number on the branches were Bayesian inference posterior probability/maximum likelihood bootstrap support values. The species marked with ▲ were newly collected in this study

Analysis of adaptive evolution

The nonsynonymous and synonymous substitution ratios (Ka/Ks) were calculated for the chloroplast genomes of 28 Ardisia species based on 79 common protein-coding genes, with Tapeinosperma netor used as a reference. Results showed that 22 genes had ratios that could not be calculated due to the absence of synonymous or nonsynonymous changes (Ka or Ks = 0), while the remaining 57 genes had average Ka/Ks ratios ranging from 0.009 (psbB) to 1.076 (rbcL). Notably, the average Ka/Ks ratio for the rbcL gene was above 1, and Ka/Ks ratios in 14 comparison groups were all greater than 1, indicating positive selection at specific chloroplast coding sites. Specifically, the Ka/Ks values for the rpoC1 gene in the comparison group between T. netor and A. crenata, A. japonica, and A. polysticta were 1.221, 1.221, and 1.003 respectively. For the rpoC2 gene in the comparison group between T. netor and A. argenticaulis, A. omissa, and A. pusilla, the Ka/Ks values were 1.202, 1.001, and 1.001 respectively. For the the rpl22 gene, the Ka/Ks values in the comparison groups of T. netor and A. balansan, A. gigantifolia, A. pedalis, and A. sieboldii were 1.422, 1.404, 1.233, and 1.123 respectively. Similarly, the Ka/Ks values of the rpl33 gene in T. netor and A. argenticaulis, A. bullata, A. crispa, A. lindleyana, A. maclureivs, A. omissa, and A. pusilla were all 1.147 (Table S5). It was observed that the Ka/Ks values of other genes were mostly less than or close to 1, suggesting that most protein-coding genes in the chloroplast genome of Ardisia species underwent purifying selection or experienced no selection pressure during the evolutionary process.

In addition, the protein-coding genes of the chloroplast genomes of 28 Ardisia species were analyzed using EasyCodeML v1.21 [18], and seven genes (cemA, ndhF, psbL, rbcL, rpoC2, ycf1, and ycf2) were identified under positive selection by a high posterior probability (> 95%) through the BEB test (Table S6). Therefore, only two genes (rbcL and rpoC2) were under positive selection using two selective pressure estimation strategies. Among the positively selected amino acid sites in the rbcL protein, two (140th and 262th) were located in the random coil, while the remaining six (142th, 225th, 251th, 279th, 407th and 461th) were located in the α-helix (Figs. 7A and 8A). In the RNA polymerase β subunit coding gene (rpoC2), one amino acid site (666th) was determined to be under positive selection (Fig. 7B), and spatial analysis showed that the site was situated in the α-helix (Fig. 8B).

Fig. 7.

Comparison of partial sites under positive selection of different genes

Fig. 8.

Spatial location of the positively selected sites in proteins of A. argenticaulis

Discussion

Sequence variation of chloroplast genomes

Within Angiosperms, it is commonly observed that the chloroplast genome adheres to a quadripartite structural arrangement, with typical sequence lengths falling within the range of 120–160 kb [19–21]. In this study, the length of complete chloroplast genomes of 28 Ardisia species ranged from 155,088 bp (A. bullata) to 156,999 bp (A. fordii) with an average length of 156,456 bp (Fig. 1; Table 1). The chloroplast genome across these species collectively comprised 129–132 genes, including 8 rRNA genes, 36–37 tRNA genes, and 85–88 protein-coding genes. The similar types and quantities of coding genes observed among these species suggest a degree of genetic preservation and stability within the chloroplast genomes of closely related plant species [22].

The inverted repeat (IR) region often undergoes contraction or expansion, which is a significant factor responsible for the variation in chloroplast genome length among angiosperm groups during evolution [23, 24]. By comparing analysis of the IR/SC boundary regions of the chloroplast genomes of the 28 Ardisia species, we observed dynamic changes in the IR regions, with some species (A. sieboldii, A. faberi, A. japonica, A. maclurei, and A. pusilla) showing noticeable expansion or contraction (Fig. 2). This phenomenon could be attributed to these species evolving at a faster rate or differentiating earlier, leading to a transformation of their genome structure during evolution, as observed in most other terrestrial plants [25].

Repeat sequences in chloroplast genome

Repeat sequences are commonly found in the chloroplast genome of plants, and their presence can result in duplication, deletion and rearrangement of segments, ultimately influencing species evolution and intraspecific genetic variation [26, 27]. In this study, we examined the chloroplast genomes of 28 Ardisia species and observed significant differences in the number and types of repeat sequences present in these genomes (Fig. 3, Table S2). Forward and palindromic repeats were the most prevalent, followed by reverse repeats, while complementary repeats were the least common. Not all species contained all four types of repeat sequences, and only six species had complementary repeats. Previous studies suggested that the number of repeat sequences could affect the stability of chloroplast genomes [28]. The variability in the number, type, and length of repeat sequences among species provides a potential basis for the development of new molecular genetic markers [29].

Simple Sequence Repeats (SSRs) in the chloroplast genome are highly variable in their internal specificity, making them valuable genetic markers in population genetic and evolutionary studies [30]. Our analysis revealed that the chloroplast genomes of 28 Ardisia species contained 54 to 72 SSR loci, most of which were single nucleotide repeat sequences composed of A/T repeat sequences (Fig. 4, Table S3, Table S4). The result is consistent with the sequence composition of SSRs in other angiosperm chloroplast genomes, further supporting the fact that SSRs primarily consist of short poly-A and poly-T repeats [31]. The variation in the distribution and number of SSRs among different Ardisia species may be attributed to mutations and deletions of gene sequences during plant evolution [32]. SSR markers with rich polymorphism could be utilized for variety purity detection, genetic diversity analysis, species identification and gene mapping of Ardisia plants in the future.

Identification of highly variable region

Due to the high conservation of protein-coding regions in the chloroplast genome, they are often not effective for species identification. Genes such as ycf1 and ycf2, which have high mutation rates and lengths over 5000 bp, present challenges for PCR amplification in practical applications. Therefore, identifying genic spacer regions with high variation rates and moderate lengths as potential DNA barcodes can assist in the identification of specific plant taxa [33–35]. In this study, nucleotide diversity (π) was calculated using the complete chloroplast genome sequences and identified 12 highly variable regions, including three protein-coding genes (infA, rpl22, and ycf1) and nine intergenic regions (rps16-trnQ, trnG-trnR, trnT-psbD, ycf3-trnS, trnT-trnL, atpB-rbcL, petG-trnW, trnL-ndhB, and rpl32-trnL) (Fig. 5). However, this method has limitations as it only considers differences after aligning all sequences, and these highly variable regions may not be suitable for identifying all species. Therefore, conducting pairwise comparisons of all species is necessary to ensure that the highly variable regions have a sequence length range that allows for conventional PCR product generation and contain enough variant sites to select the most appropriate sequences as DNA barcodes for accurate species identification.

Phylogenetic relationships

Ardisia includes a diverse group of species and varieties, making accurate identification challenging. The most comprehensive revision of Ardisia conducted by Mez identified 14 subgenera based on their habitats, leaf morphology, inflorescence position, and flower morphology in 1902 [36]. Since then, most taxonomic work on Ardisia has been limited to regional revisions [37–41]. Walker divided Ardisia into five sections based on various characters such as sepals, inflorescences, glandular dots, and leaf margins in his revision of the family Myrsinaceae of East Asia [41]. Chen et al. substantially followed Walker’s classification system, but included A. aberrans in section Pimelandra, and consequently, Ardisia species in China were ultimately divided into six Sect. [2]. Based on the phylogenetic relationship reconstructed using nuclear ITS and two chloroplast intergenic spaces (psbA-trnH and rpl32-trnL), the latest revised species classification indicates that Asian Ardisia was not a monophyletic group, and the relationships between most subgenera remained unresolved due to limited molecular markers [1].

Chloroplast genomes have proven successful in resolving phylogenetic relationships in various plant groups [42–45]. In this study, phylogenetic studies were conducted on the Ardisia plants in China based on chloroplast genome data of 28 Ardisia species and two outgroups (Tapeinosperma multiflorum and T. netor). Phylogenetic analysis based on different datasets, including LSC region, SSC region, IR region, coding sequences, and complete chloroplast genome, revealed that the 28 Ardisia species formed a monophyletic group, further divided into smaller clades: subgenus Crispardisia, subgenus Akosmos, subgenus Tinus, subgenus Odontophylla, and subgenus Bladhia (Fig. 6, Fig. S3, Fig. S4, Fig. S5, Fig. S6). The result supported the recent classification revision that the traditional subgenus Bladhia s.l. should be split into two clades: subgenus Bladhia s.str. and subgenus Odontophylla (Yang) Huang which could be distinguished based on distribution range and morphology [1, 46]. It is worth noting that the phylogenetic trees generated from different datasets reveal a polytomy at the root of the clade containing all subgenera except subgenus Bladhia. As a result, the relationships between certain clades remain unclear, including: (a) the clade consisting of the subgenera Crispardisia and Akosmos; (b) the subgenus Tinus; and (c) the subgenus Odontophylla.

In this study, the subgenus Crispardisia clade comprised 12 species (Fig. 7, Fig. S3). According to Mez’s taxonomic system, A. mamillata and A. primulifolia were initially placed in the subgenus Bladhia [36]. However, Pitard argued that the presence of marginal glandular dots on the leaves were the most distinctive feature of the subgenus Crispardisia, leading to the reclassification of these two species [47]. The molecular phylogenetic analysis in this study supported Pitard’s taxonomic treatment and suggested that assigning the two species to the subgenus Crispardisia was more appropriate. As a controversial species, A. argenticaulis was placed in the subgenus Bladhia by Walker [37]. However, Wang et al. argued that A. argenticaulis should be classified into the subgenus Akosmos [48]. The phylogenetic tree in this study supported that A. argenticaulis belongs to the subgenus Akosmos (Fig. 6, Fig. S3, Fig. S4, Fig. S5, Fig. S6). Overall, the phylogenetic analysis conducted in this study provided molecular evidence to support previous findings, and enhanced our understanding of the relationships between subgenus and species within Ardisia in China.

Adaptive selection on Ardisia chloroplast genes

Ardisia species are extremely shade-tolerant and primarily found in the understory of forests or in moist areas near valleys and streams. The influence of low light as a natural selective pressure may have resulted in adaptive changes in the chloroplast genes responsible for environmental adaptation [49–51]. In Chrysosplenium (Saxifragaceae), two genes related to photosynthesis (matK and ycf2) showed positive selective pressure, which might be associated with adaptation to low light conditions [17]. Similarly, we identified two genes that were under positively selection among the 28 Ardisia species, including rbcL and rpoC2 (Figs. 7 and 8, Table S5, Table S6).

The rbcL gene for the Rubisco large subunit was identified under positive selection (Figs. 7 and 8). Being the selection target of multiple environmental factors related to light, temperature, and carbon dioxide concentration, the rbcL gene is often under positive selection [52]. The positive selection of rbcL gene related with photosynthesis, suggested their role in the adaptation of Ardisia species to low-light habitats. Plastid-encoded RNA polymerase (PEP) is composed of four subunits (ɑ, β, β’, β’’) which are encoded by the rpoA, rpoB, rpoC1, and rpoC2 genes respectively [53–55]. Previous studies indicate that PEP is a crucial enzyme responsible for the transcription of photosynthesis genes in chloroplasts [56]. We identified an amino acid site under positive selection in the rpoC2 gene of Ardisia species in this study (Figs. 7 and 8), which suggested that PEP might play a significant role in expression of photosynthesis genes of adapting to the environment. In summary, two genes (rbcL and rpoC2) were under positive selection, which might have contributed to the adaptation of Ardisia species to various environmental conditions, particularly those characterized by low light levels.

Conclusions

The chloroplast genomes of 28 Ardisia species exhibited a typical tetrad structure, ranging in length from 155,088 bp to 156,999 bp. These genomes consisted of a large single-copy region (LSC) spanning 84,709 bp to 86,989 bp, a small single-copy region (SSC) ranging from 18,093 bp to 18,479 bp, and a pair of inverted repeat regions (IR) spanning 25,411 bp to 26,236 bp. The GC content ranged from 36.0 to 37.3%. The gene composition remained relatively conserved, comprising 129–132 coding genes including 85–88 protein-coding genes, 36–37 tRNA genes, and 8 rRNA genes, with 18 genes containing introns. Palindromic repeats were the most common among the repeat sequences, and the chloroplast genome sequences contained numerous SSR sites, with A/T being the predominant component. Comparative analysis of the chloroplast genome identified 12 highly variable regions, including 3 protein-coding genes (infA, rpl22, and ycf1) and 9 intergenic regions (rps16-trnQ, trnG-trnR, trnT-psbD, ycf3-trnS, trnT-trnL, atpB-rbcL, petG-trnW, trnL-ndhB, and rpl32-trnL), which could serve as potential DNA barcode markers for identifying Ardisia species. The phylogenetic tree supported the division of the subgenus Bladhia s.l. into subgenus Bladhia s.str. and subgenus Odontophylla (Yang) Huang. Notably, two genes (rbcL and rpoC2) exhibited positive selection, potentially linked to the adaptation of Ardisia species to low-light environments. This comprehensive study offers valuable insights into the chloroplast genome of Ardisia species, facilitating species identification and the utilization of germplasm resources.

Materials and methods

Plant materials and DNA extraction and sequencing

Twenty Ardisia species were collected in Guangxi, China and cultivated at the Guangxi Institute of Botany (Fig. S1, Table S1). Genomic DNA extraction was carried out from fresh green leaves using a modified CTAB method [57]. The extracted DNA was subjected to a series of meticulous procedures, including mechanical fragmentation, purification, terminal repair, and other necessary treatments prior to sequencing. Fragments of 350 bp were selected through agarose gel electrophoresis and subsequently amplified through PCR to establish a sequence library. High-throughput sequencing of the library was performed using the Illumina HiSeq 2000 sequencer (Illumina Biotechnology Company, San Diego, CA, USA) to generate paired-end (PE) reads. The genome was reassembled from the filtered data using NOVOPlasty (v.2.7.2) [58]. To ensure assembly accuracy, Bowtie2 (v2.0.1) was used to map all high-quality clean reads to the assembled genome sequence [59]. Finally, complete chloroplast genome sequences of 20 Ardisia species (accession number: OK054492-OK514747) were obtained, and a combined analysis of these genomes, along with the other 8 published chloroplast genomes acquired from GenBank was conducted (Table S1).

Genome annotation and sequence characterization

The annotation results were compared with the reference genome annotation information using Geneious v8.0.2 software to confirm the annotation. The positions of the stop and start codons of some protein-coding genes were manually adjusted. Geneious v8.0.2 software was also used for chloroplast genome boundary annotation [60]. A circular chloroplast genome map was created using the web program Organellar Genome DRAW [61].

Boundary regions and genome comparative analysis

The annotation of chloroplast genomes for 28 Ardisia species, including the boundaries of LSC, SSC, two IRs regions and genes near each boundary was mapped onto a simplified chloroplast genome structure map using IRscope (https://irscope.shinyapps.io/irapp/). Mauve Alignment in Geneious Prime was used for the chloroplast genome collinear analysis among the 28 Ardisia species [62]. Nucleotide polymorphisms (π) were calculated among the 28 Ardisia species using DnaSP v6.0 software, with specific parameter settings comprising a window length of 600 and a step length of 200 [63].

Repeat sequences analysis

Repeat sequences, encompassing forward, reverse, complement, and palindromic repeats, were quantified employing the online tool REPuter, accessible at https://bibiserv.cebitec.uni-bielefeld.de/reputer/manual.html, utilizing the following specific parameters: minimum repeat sequence length > 30 bp, repeat sequence similarity > 90%, and Hamming distance = 3 [64]. Simple sequence repeats were analyzed using the online tool MISA-web for identification and statistical analysis [65]. The parameters were set as follows: mononucleotide ≥ 10 repeat units, dinucleotide ≥ 5 repeat units, trinucleotide ≥ 4 repeat units, and tetra-, penta-, and hexanucleotides with at least 3 repeat units.

Phylogenetic Analysis

To infer phylogenetic relationships among Ardisia species, we independently analyzed the complete chloroplast genome and specific DNA fragments. The outgroup species (Tapeinosperma multiflorum and T. netor) were selected based on the research of Yan et al. [66]. The chloroplast genome sequences and specific DNA fragments were aligned using MAFFT (version 7.222) [67]. Maximum likelihood (ML) tree analysis of the aligned sequences was performed using RAxML 7.2.8 software with the best model of TVM + F + I + I + R4, determined through 1,000 guided repeat tests [68]. The Bayesian Inference (BI) tree was constructed using MrBayes v.3.2.6 software [69]. The Markov Chain Monte Carlo (MCMC) algorithm was run for 1,000,000 generations under the TPM1uf + I + G model, sampling every 1,000 generations. We ensured that the average standard deviation of the split frequencies remained below 0.01. Additionally, 25% of the samples from the burn-in phase were discarded before computing the consensus tree and determining the Bayesian posterior probabilities (PP).

Analysis of adaptive evolution

Twenty-eight Ardisia species and Tapeinosperma netor (used as a reference) were selected for the analysis of selection pressure on chloroplast genome protein-coding genes. The sequences of 79 shared chloroplast protein-coding genes were compared individually, and stop codons were removed. The ratio of paired nonsynonymous substitution rate (Ka) to synonymous substitution rate (Ks) was then calculated for all species using KaKs Calculator v2.0 [70]. To predict selection for each gene, we considered the ratios Ka/Ks. A ratio of Ka/Ks < 1 indicates purifying selection, a ratio of Ka/Ks = 1 indicates neutral selection, and a ratio of Ka/Ks > 1 indicates positive selection [71]. If Ks was 0, the Ka/Ks values were expressed as NA.

Furthermore, the site model (set to seqtype = 1, model = 0, NSsites = 0, 1, 2, 3, 7, 8) was used to perform the LRT (likelihood ratio test). The log-likelihoods of each model and the neutral model were compared using LRT to test for statistical significance. The alternative hypothesis model M8 was accepted if the p-value was less than 0.05 in the LRT results; otherwise, the null hypothesis model M7 was accepted. Genes with positive selection sites under the M8 model, where p < 0.05 and the Bayesian Empirical Bayes (BEB) posterior probability exceeded 0.95, were considered potential positively selected genes [72]. Upon detection of positive selection by LRT, the Bayesian Empirical Bayes (BEB) method was applied to estimate the posterior probability of each codon from the positive selection site category in models M2a and M8. The Bayesian Empirical Bayes method represents an improvement over the previous Naïve Empirical Bayes method, accounting for sampling errors of maximum likelihood estimates within the model [73–76]. Amino acid sequences were visualized using Jalview software to highlight positively selected sites [77]. To gain further insight into the structural characteristics of these genes, we utilized the online protein structure prediction tool SWISS-MODEL, with A. argenticaulis as an example [78].

Abbreviations

Cp

Chloroplast

LSC

Large single-copy region

IR

Inverted repeat region

SSC

Small single-copy region

CDS

Protein-coding sequences

tRNAs

Transport RNAs

rRNAs

Ribosomal RNAs

SSRs

Simple sequence repeats

π

Nucleotide diversity

Ka/Ks

The rate of non-synonymous substitutions to the rate of synonymous substitutions

RSCU

Relative synonymous codon usage

Pi

Nucleotide diversity

ML

Maximum-likelihood

BEB

Bayes empirical bayes

NCBI

National Center for Biotechnology Information

Authors’ contributions

ZJ: conceived the study, performed data analysis and drafted the manuscript; NY: collected and identifed the species of sample, designed the experiments, analyzed the data; LJ, DY and WL: assisted in conceptualizing and designing experiments; MS: collected and identifed the species of sample, data curation； ZB: funding acquisition, reviewed the manuscript critically. All authors have read and agreed with the contents of the manuscript

Funding

This work was supported by National Natural Science Foundation of China (32060090), Natural Science Foundation of Guangxi Province (2021JJA130119) and Science and technology plan project of Guangzhou Construction Group ([2022]-KJ019), ([2021]-KJ014).

Data availability

All data generated or analysed during this study are included in this manuscript. All the annotated chloroplast sequences data reported here were deposited in GenBank (https://www.ncbi.nlm.nih.gov/) with accession numbers and voucher number shown in Table S1.

Declarations

Ethics approval and consent to participate

The collection of all samples completely complies with national and local legislation permission. Plant samples used in the study were not included in the list of national key protected plants and were not collected from the national park or nature reserve when we collected them. According to national and local legislation, no specifc permission was required for collecting these plants when we collected them. Voucher specimens were prepared and deposited at the Herbarium of Guangxi Institute of Botany, Chinese Academy of Sciences (IBK).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.