Introduction

The tribe Arethuseae Lindl. belonging to the subfamily Epidendroideae can be divided into two monophyletic lineages, corresponding to the subtribes Coelogyninae Benth. and Arethusinae Benth. [1,2,3,4,5]. As one of the most diverse and economically important group within the tribe, Coelogyninae in its well-known delimitation comprises 22 genera with approximately 740 species [2, 4, 5] and is widespread in tropical Asia and the Pacific. Many species of Coelogyninae are horticultural plants with great ornamental value, while others have long been used in traditional medicine practices due to their nutritious or medical efficacy [6]. Although it is generally accepted as a monophyletic group, the circumscriptions of some Coelogyninae genera are not well resolved. Previous phylogenetic studies on the subtribe Coelogyninae were based on either traditional morphological characters or only a few loci, which may indicate some limitations. The phylogenetic relationships of the major clades in the Coelogyninae group, particularly those of Coelogyne and its related genera remain unclear. Gravendeel et al. [7] delimited Coelogyne in the strict sense by including Neogyna and Pholidota species and removing those species with hairy ovaries and ovate-oblong petals. In contrast, Chase et al. [8] expanded the genus by transferring the former 14 genera into Coelogyne s.l., which comprises about 550 species. It is worth noting that Coelogyninae has been circumscribed to consist of only eight genera after this merging.

Compared with relatively short DNA fragments, the chloroplast (cp.) genomes in vascular plants contain more abundant informative sites. Chloroplast genomes have been increasingly used for phylogenetic reconstruction and genome-wide patterns of nucleotide polymorphism in the family Orchidaceae. Insights gained from whole cp. genome data have greatly improved our understanding of the evolution and diversification of Orchidaceae [9, 10]. Over the past few years, phylogenomic analyses and comparative chloroplast genomics have been conducted in some genera of subtribe Coelogyninae, e.g., Bletilla, Thuniopsis and Pholidota [11,12,13]. These phylogenomic studies have proven to be effective in improving phylogenetic resolution. For example, Li et al. [13] initially probed the structural patterns of Pholidota plastomes and provided new insight into the phylogenetic relationships among Pholidota and its related genera using the whole-genome data. So far, however, the deep phylogenetic relationships and genome-wide patterns within the subtribe Coelogyninae have not been thoroughly investigated. The intergeneric and infrageneric classification and relationships within the subtribe remain controversy.

This study represents the most comprehensive taxonomic sampling of Coelogyninae thus far using complete plastomes. We newly sequenced 15 cp. genomes of Coelogyninae species and retrieved nine previously published cp. genomes for comparative and phylogenetic analyses. Here, in order to better show the systematic positions of Coelogyne and its related genera, the traditional concept of each genus was adopted.

The major objectives of this study were: (1) to investigate the characteristics and evolutionary patterns of cp. genomes among species in Coelogyninae; (2) to detect variation and mutation hotspots among these cp. genomes; (3) to explore the phylogenetic relationships among the major clades of Coelogyninae.

Results

Plastome features of Coelogyninae species

The complete chloroplast genomes newly obtained from the 15 Coelogyninae species, plus previously sequenced nine species were very similar in the overall structure, gene content, order and GC content (Figs. 1 and 2; Additional file 1: Fig. S1). The lengths of the 24 Coelogyninae plastomes ranged between 158,394 bp (Pleione maculata) and 160,280 bp (Chelonistele sulphurea). All of the cp genomes possessed the typical four conjoined structures, including a pair of IRs (26,489–26,796 bp) separated by one LSC (86,603–88,206 bp) and one SSC (18,499–18,885 bp) region. The GC content in the IR regions (43.18–43.34%) was higher than those in the LSC (34.83–35.27%) and SSC (29.9–30.58%) regions (Additional file 2: Table S1). Each of the cp genome comprised 135 genes, including 89 protein-coding genes (PCGs), 38 tRNAs and eight rRNAs. There were 80 genes (59 PCGs and 21 tRNA genes) in the LSC region, 11 genes (10 PCGs and one tRNA gene) in the SSC region, 20 genes included two copies because of their location in the IR regions, including eight protein-coding genes (rpl2, rpl23, rps7, rps12, rps19, ycf2, ycf15, and ndhB), four rRNA genes (rrn16, rrn23, rrn4.5 and rrn5), and eight tRNA genes (trnA-UGC, trnH-GUG, trnI-CAU, trnI-GAU, trnL-CAA, trnN-GUU, trnR-ACG, and trnV-GAC). Among these genes, 19 genes contained two exons, including 11 coding genes (atpF, ndhA, two ndhB, petB, petD, rpl2, rpl16, rpoC1, rps12 and rps16) and eight tRNA genes (two trnA-UGC, trnG-UCC, two trnI-GAU, trnK-UUU, trnL-UAA and trnV-UAC). Four coding genes (two rps12, clpP1 and paf1) contained three exons. In all these plastomes, the gene ycf1 crossed the SSC and IRA junction. The SSC and IRB boundaries were positioned in the genes ycf1b and ndhF. The rps12 gene was recognized as a trans-spliced gene, with 5’-end exon located in the LSC region and two 3’-end exons located in IR regions (Fig. 1).

Fig. 1
figure 1

Chloroplast genome organization and characteristics of Coelogyninae. Genes drawn inside and outside of the circle are transcribed in clockwise and counterclockwise directions separately. Darker and lighter gray area in the inner circle corresponds to the GC and AT content, respectively. Genes belonging to different functional groups are color-coded in the outmost circle

Full size image

The graphical map of circular genomes was generated by GView to assess sequence variations among the 24 chloroplast genomes in Coelogyninae (Fig. 2). In all the examined plastomes, sequences in the LSC and SSC regions varied greatly across different genera and species. The genome comparison revealed that the sequences in the two IR regions were less divergent than those of the LSC and SSC regions. The intergenic regions exhibited higher divergence than the coding regions.

Fig. 2
figure 2

Graphical map of circular genomes retrieved by Gview, providing the overall visualization of the 24 Coelogyninae plastomes. From the inside to the outside, the positions of CDS (blue), rRNA (orange) and tRNA (green) genes on both the forward and reverse strand, followed by GC skew (yellow, red lines), GC content (cerulean, black lines) are shown in sequence. The outer circle indicates the genome size in kbp. The remaining circles display BLAST comparisons of plastome sequences. The innermost circle represents the reference genome of Thuniopsis cleistogama. The similar and divergent locations in the plastomes are shown in continuous and interrupted track lines, respectively. The lightly screened area stretching from the inner toward the outer circle indicates different regions with large sequence divergences

Full size image

Identification of SSRs and long repetitive sequences

Simple sequence repeats (SSRs) or microsatellites, consisting of short DNA motifs (1–6 bp in length) are specific, highly polymorphic and informative genetic markers and are widely distributed in the genomes. In this study, we examined the SSRs among the 24 Coelogyninae cp. genomes and analyzed their distribution and frequency in different genomic regions. The SSR screening resulted in the identification of 1,014 SSRs, ranged from 31 (Otochilus porrectus) to 53 (Bletilla striata and Panisea tricallosa) (Fig. 3A). The detailed frequency of SSRs with different repeat motif and number is shown in Additional file 3: Table S2. Of the total 1,013 SSRs, 972 were simple repeat motifs (95.95%) and 41 were present in compound formation (4.05%). Among the SSRs, the highest proportion was represented by mononucleotide (p1) repeats, with a proportion of 84.11% and dinucleotide (p2) repeats accounted only for 9.77%. The trinucleotide (p3) and tetranucleotide (p4) repeats were rather rare, accounting for 1.87% and 0.2%, respectively. We also detected the frequency of motif composition in these genomes. The most abundant mononucleotide SSRs were composed primarily of A and T repeat units (96.83%), with only 3.17% composed of C/G. Dinucleotide repeats always consisted of AT and TA (100%), with an obvious A/T bias. For trinucleotide (p3) repeats, TTC repeats were most abundant in many species (89.48%), with ATT repeat motif (5.26%) only detected in Pholidota imbricata cp. genome, while TCT repeat motif (5.26%) only occurred in Chelonistele sulphurea cp. genome. In addition, tetranucleotide (p4) repeats were not found in most of these genomes, except for one tetranucleotide repeat motif (ATAG) detected in the cp. genome of Pholidota ventricosa, and Thunia alba, respectively. These results were consistent with previous investigations [12, 13].

The SSRs ranged in size from 10 to 60 bp, with the majority in 10–15 bp length, accounting for 87.71%, followed by 15–20 bp (7.26%), 20–30 bp (1.5%), 30–60 bp (1.39%) and 60 bp (2.14%). The most abundant SSRs in 10–15 bp as well as a wide range of all sizes from 15 to 60 bp were detected in the cp. genomes of Bl. striata and P. tricallosa, respectively, while in the T. cleistogama cp. genome where SSRs preferred lengths were generally longer (30 to 60 bp) in comparison with SSRs in most other cp. genomes (Fig. 3B; Additional file 4: Table S3).

Fig. 3
figure 3

Repeats in the 24 Coelogyninae plastomes. (A) Number of different SSRs types (p1–p4 indicate mono-, di-, tri-, and tetranucleotide, respectively; c indicates compound SSRs); (B) Number of repeats by length; (C) Frequency of SSRs in the LSC, IR, SSC region; (D) Frequency of SSRs in the intergenic regions (IGS), protein-coding (CDS) genes and introns

Full size image

Furthermore, the SSRs in the 24 Coelogyninae cp. genomes were more frequently located in the LSC region (73.54%) than in the SSC region (18.36%), and only a minority (8.1%) was dispersed within the IR regions (Fig. 3C; Additional file 5: Table S4). Likewise, SSRs (70.48%) in these cp. genomes primarily occurred in the intergenic spacer (IGS) regions, with a small portion (19.55%) distributed in introns, while only a few (9.97%) of SSRs was found in CDS regions (Fig. 3D; Table S4).

Long repetitive sequences were detected with a length ≥ 30 bp and similarity > 90% between two copies. In total, 1,253 long repetitive sequences were detected in the 24 Coelogyninae cp. genomes, including 10–59 forward (F) repeats, 0–11 reverse (R) repeats, 0–12 complementary (C) repeats, and 20–40 palindromic (P) repeats (Fig. 4, Additional file 6: Table S5). Of the four types of long repeats, most of those were palindromic (P) and forward (F) repeats, with percentages of 59.14% and 36.07%, respectively, with complementary (C) and reverse (R) repeats accounted for only 1.68% and 3.11%, respectively.

Fig. 4
figure 4

Frequency of four long repeat types: Forward (F), Reverse (R), Complement (C), Palindromic (P) in the 24 Coelogyninae plastomes

Full size image

Among the 24 Coelogyninae cp. genomes, Dendrochilum apoense contained the highest number (89) of long repeats, while Bulleyia yunnanensis contained the lowest (31). Only five species (Ch. sulphurea, C. rochussenii, D. cootesii, Ph. imbricata and T. cleistogama) had all four types of repeats (Fig. 4, Additional file 6: Table S5). Of these species, T. cleistogama had 11 reverse (R) repeats and 12 complementary (C) repeats, while most other species had only 1–4 reverse repeats and 1–2 complementary repeats. For three types of long repeats detected in 13 cp. genomes, most of which had 1–5 reverse (R) repeats and lacked complementary (C) repeats. Interesting, there was no reverse (R) repeat in the plastomes of four species (C. viscosa, O. porrectus, P. tricallosa and Ph. protracta), instead, there was only one complementary (C) repeat detected. The result suggests that long repeats are not only typical and but also representative in these species.

Comparison of IR/SC boundaries

The IR/LSC and IR/SSC boundary positions were visualized among the 18 plastomes from the representatives belonging to 13 genera of Coelogyninae. In all these plastomes, gene rps19 was located near the LSC/IR border, while both genes ycf1 and ndhF were located at the IR/SSC junctions. In addition, overlaps between the ycf1 and ndhF were detected at the IRb/SSC boundaries. Although these plastomes were highly similar regarding their gene content and order, the IR/SC junctions showed substantial differences among the genera and species within the subtribe (Fig. 5).

The SSC/IRa junctions were embedded in the ycf1 gene, resulting in an incomplete duplication in the IRb region, with the portion located in IRb region varied from 1,002 bp (Bulleyia yunnanensis) to 1,044 bp (Dendrochilum apoense). The ndhF gene spanned over the IRb/SSC junctions, partially overlapping with the duplicated ycf1, with the overlapping lengths at the IRB/SSC border varied from 55 to 72 bp. The length of ndhF exhibited a high degree of uniformity. The majority of these plastomes had the same length of 2,259 bp, with three exceptions (2,226 bp in D. apoense, 2,253 bp in Chelonistele sulphurea, and 2,265 bp in Thunia alba).

For the 14 Coelogyninae species representing Coelogyne and closely related genera, the rpl22 gene was completely positioned in the LSC region, with the distances from the LSC/IRb border varying from 16 bp (Coelogyne flaccida) to 93 bp (C. rochussenii). Interestingly, for the rest four Coelogyninae species, Bletilla striata, Pleione maculata, Thunia alba, and Thuniopsis cleistogama, the rpl22 gene spanned the LSC/IRb junction, with the portion located in the IRb region ranged only from 35 to 37 bp.

Fig. 5
figure 5

Comparison of the junctions between the LSC/SSC and IR regions among the 18 Coelogyninae chloroplast genomes

Full size image

Furthermore, the IRa/LSC region was located in the intergenic regions of two genes, rps19 and psbA. The gene rps19 adjacent to the IR/LSC junctions was duplicated in the IR regions, with the distances apart from the LSC/IRb border varying from 93 bp (Neogyna gardneriana) to 162 bp (D. apoense) for the 14 Coelogyninae species. However, in the four species (Bl. striata, Pl. maculata, T. alba, and T. cleistogama), the distances from the border varied little, only from 228 bp (T. cleistogama) to 235 bp (Pl. maculata). Similarly, the distance from the gene psbA to the IRa /LSC boundary ranged from 88 bp (B. yunnanensis) to 115 bp (D. apoense) for the 14 species, whereas, the distance was the same 126 bp in the four other species.

Genome comparison and sequence divergence analyses

Taking the annotated Thuniopsis cleistogama (OL809660) genome sequence as a reference, mVISTA was carried out to ascertain the divergent regions in the multiple alignments of 18 Coelogyninae cp. genomes representing 13 genera (Fig. 6). Higher degree of variations primarily occurred in the IGS regions, such as rps16-trnG-UCC, rpoB-psbD, psbK-psbI, rps12-trnV-GAC, ndhF-rp132, rp132-trnL-UAG, atpB-rbcL, atpF-atpH, atpH-atpI, trnE-UUC-trnT-GGU, psaA-pafI and psaC-ndhE. More variations were also identified in the intron-containing genes, such as rpoC1, ycf2, ccsA, ndhF and ycf1. Most of the genes in CDS region were relatively well conserved, except for the sequence variations in some genes, e.g., rps16, atpF, pafI, accD, clpP1, petD, rpl16 and ndhA. In contrast, the rRNA genes of these species were highly conserved.

Fig. 6
figure 6

Sequence alignment of the complete plastome sequences of 18 Coelogyninae species with Thuniopsis cleistogama cp. genome as a reference. Red arrows up the alignments indicate the direction of the gene. Blue and white correspond to coding regions and non-coding regions, respectively

Full size image

Additionally, the nucleotide variability (Pi) values in the LSC, SSC, and IR regions were calculated separately among the 24 Coelogyninae chloroplast genomes. The SSC regions showed the highest nucleotide diversity (Pi = 0.01747), followed by the LSC region (Pi = 0.01055), while IR regions showed the lowest nucleotide diversity (Pi = 0.00256). The results showed that the LSC and SSC regions were more divergent than the IR regions. As expected, the average Pi value in the non-coding regions (0.01113) exhibited comparably higher divergence levels when compared to that in the coding regions (0.00641).

The sliding window analysis revealed ten highly divergent hotspots among the cp. genomes with Pi values ranging from 0.02676 to 0.04849 (Fig. 7). Among the ten regions, five regions: trnS-trnG (0.03086), atpB-rbcL (0.02999), matK-rps16 (0.02788), rps16-trnQ (0.02676) and rpoB-trnC (0.02701) were located in the LSC region, and four regions: ndhF-rp132 (0.04239), psaC-ndhE (0.03781), rp132-trnL (0.03169) and ndhE (0.02739) were located in the SSC region, and only ycf1a (0.04849), crossed the SSC-IRa boundary in the coding region.

The region of ycf1a showed the highest nucleotide variability (0.04849). Three of the variable intergenic spacer (IGS): ndhF-rp132 (0.04239), psaC-ndhE (0.03781) and rp132-trnL (0.03169) had markedly higher divergence values (> 0.031). These hypervariable regions or genes were identified as potential molecular markers for further development such as DNA barcoding, molecular phylogenetics and breeding.

Fig. 7
figure 7

Sliding-window analysis based on the whole cp. genomes of 24 Coelogyninae species. X-axis shows the genomic regions; Y-axis denotes the Pi values

Full size image

Positive selection sites

We investigated the ratio of non-synonymous to synonymous substitutions (dN/dS) to estimate the selective pressure on 81 common non-redundant protein-coding genes among 20 Coelogyninae species. Positively selected sites under positive selection were detected using codon substitution models. A total of 19 genes with positive selection sites were identified (Additional file 7: Table S6), which were listed as follows: one subunit of the Acetyl-Co A-carboxylase gene (accD), one subunit of the ATP-dependent CLP protease gene (clpP1), one gene encoding the maturase K (matK), one subunit of the rubisco gene (rbcL), four genes for a component of the trans locus of an envelope protein (ycf1a, ycf2, ycfb and ycf15), one gene for photosystem I subunit (psaB), one subunit of ATP synthase gene (atpB), six genes for subunits of NADH-dehydrogenase (ndhA, ndhD, ndhE, ndhF, ndhG and ndhK), and three DNA-dependent RNA polymerase genes (rpoA, rpoB and rpoC2).

Among the 19 protein coding genes containing amino acid positive sites, it was found that the ycf1a gene located in the IR region harbored the highest number of positive selection sites (113), including 34 significant positive selection sites and 79 extremely significant positive selection sites, followed by rpoC2 (100) and ycf2 (96).

Phylogenetic relationships

Phylogenetic analyses were conducted with the ingroups using a total of 55 accessions, representing 42 species of 13 genera in Coelogyninae. Generally, the topologies of the ML and BI trees constructed with complete cp. genomes (Fig. 8, Additional file 8: Fig. S2) and CDSs (Additional file 9: Fig. S3 and Additional file 10: Fig. S4) were almost identical with only a few exceptions at the shallow nodes. Slightly different tree topologies may be mainly due to the discrepancy in the number of variable sites. Most nodes were maximally supported in the ML tree of Coelogyninae inferred from full plastome sequences. On consideration, we present here the topology resulting from plastome-based ML analysis, with posterior probability (PP) and maximum likelihood bootstrap values (BS) labeled on the tree branches. The positions of representative genera and species in each clade were labelled (Fig. 8).

Fig. 8
figure 8

Phylogenetic tree of 42 Coelogyninae species using the maximum likelihood (ML) analysis based on whole chloroplast genomes with Arundina graminifolia as outgroup. Clades discussed in the text are signified by Roman numerals. The phylogenetic positions of different clades are highlighted in different colors. Numbers at nodes indicate ML bootstrap values (BS; before the slash) and Bayesian posterior probabilities (PP; after the slash). Dash (–) indicates nodes with PP < 0.5 which is incongruent with ML analysis

Full size image

As seen in Fig. 8, Both ML and Bayesian analyses of the whole plastomes recovered four major monophyletic lineages of Coelogyninae species: clade A consisting species of Coelogyne and related genera was strongly supported (BS: 100, PP: 1.0) and further split into several subclades; clade B, Pleione clade (BS: 100, PP: 1.0); clade C, Thuniopsis clade (BS: 100, PP: 1.0); and clade D, Bletilla + Thunia (BS: 99, PP: 1.0). Arundina graminifolia occurred in a basal position, as an unsupported sister to Coelogyninae.

Clade A was further divided into five highly-supported subclades (A1–A5): A1, composed of a mixed subclade (Coelogyne s.s. + Pholidota s.s. + Neogyna and Otochilus s.s.); A2, corresponding to the C. fimbriata and its allies clade; A3, Panisea + Ischnogyne clade; A4, Bulleyia clade, and A5, the Dendrochilum + Chelonistele clade always appeared among the early branching clade.

The monophyletic clade A1 (BS: 100, PP: 1.0), representing the core members of Coelogyne, Neogyna, Otochilus and Pholidota respectively, was further resolved to three subclades and each received high supports. The small clade A1-I, including the generic type of Otochilus: O. porrectus, and several sampled species of Pholidota (P. articulata and P. chinensis) was recovered as a monophyletic lineage (BS: 100, PP: 1.0); the small clade A1-II comprising the generic type of Pholidota: P. imbricata, its sister P. pallida and monotypic Neogyna was retrieved as monophyletic (BS: 100, PP: 1.0). The type species of Coelogyne, C. cristata was well supported (BS: 100) as a sister to three samples for sect. Ocellatae (C. corymbosa and C. punctulata), and together formed a moderately supported sister clade A1-III (BS: 76) to three samples for sect. Flaccidae (C. flaccida and C. viscosa). However, in the BI tree (Additional file 8: Fig. S2), C. cristata was a close sister (PP: 1.0) to the clade consisting of Pholidota s.s. and Neogyna. The second clade A2 (BS: 100, PP: 1.0) including four samples for Coelogyne sect. Fuliginosae (C. fimbriata and C. ovalis) and sect. Elatae (C. barbata) formed a well-supported monophyletic lineage. Clade A3 containing two species of Panisea, monotypic Ischnogyne and three species of Pholidota (P. longipes, P. niana and P. protractas) was strongly recovered as monophyletic (BS: 100, PP: 1.0) and together referred to as the Panisea clade. Clade A4 including monotypic Bulleyia and six accessions of five Pholidota species (P. cantonensis, P. leveilleana, P. missionariuorum, P. wenshanica and P. yunnanensis) were recovered with full support (BS: 100, PP: 1.0), referred to here as the Bulleyia clade. Two sampled species of Dendrochilum, type species of Chelonistele: Ch. sulphurea and its sister: Pholidota ventricosa, together with Coelogyne rochussenii, clustered into the monophyletic clade A5 (BS: 100, PP: 1.0) and referred to as the Dendrochilum + Chelonistele clade. The five clades A1–A5 were well supported as successive sister groups to each other and relationships among major lineages have received consistent support.

Discussion

As displayed in Figs. 1 and 2, the 24 Coelogyninae cp. genomes are extremely well conserved with regard to structure, gene composition and order, similar to those previously reported from other species in the subtribe Coelogyninae [11,12,13]. General characteristics of all 24 Coelogyninae plastomes are shown in Additional file 2: Table S1. Among these cp. genomes, the cp. genome of Pleione maculata showed the smallest size (158,394 bp) compared with those of the other Coelogyninae species. Similarly, it also had the smallest LSC and SSC regions (86,603 and 18,499 bp, respectively). The overall GC content detected in these plastomes had less fluctuation, ranging from 37.05 to 37.41%, but varied greatly among different genomes and different genomic regions within a genome. Generally, the IR regions comprised the highest GC content (43.18–43.34%), whereas the SSC region contained the lowest GC content (29.9–30.58%). The GC content values were regular as expected based on AT-rich plastome organization, which were similar to those of most other sequenced angiosperm genomes so far. The IR regions contained consistently higher GC content than that of the SC regions, probably because of four copies of GC-rich rRNA genes (rrn4.5, rrn5, rrn16 and rrn23) located in the IR regions [14].

The expansion and contraction of IR/SC boundary regions are considered to be common evolutionary events and may cause size variations [15,16,17,18]. As seen in Fig. 5, comparison of the IR/SC junction positons among the Coelogyninae plastomes revealed obvious variations. The rpl22 gene was consistently situated in the LSC region and separated from the LSC/IRb border across the 14 Coelogyninae cp. genomes. However, this gene crossed the LSC/IRb junctions for the cp. genomes of four species, Bl. striata, Pl. maculata, T. alba and T. cleistogama. Overall, the IR/SC boundary positions of these four species were more similar than those of the other 14 Coelogyninae species. In most cases, more closely related species are often assumed to have similar responses to environmental changes, so these IR boundary shifts tend to be relatively minor, involving only a small number of genes [19, 20]. The patterns of variation observed in these cp. genomes were generally consistent with the main clades (nodes A, B, C and D, Fig. 8) of Coelogyninae species recovered in the phylogenetic analyses.

Repetitive sequence analysis indicated uneven distribution of polymorphic SSRs across the 24 Coelogyninae plastid genomes with variations in number, size and type of SSR motifs (Fig. 3; Additional file 3–5: Tables S2–S4). Likewise, long repetitive sequences in these genomes also displayed distinct proportion of repeat types (Fig. 4, Additional file 6: Table S5). The accumulation of specific motifs for different SSR types might be the result of selective constraints [21]. Carmona et al. [22] suggested that changes in the quantities and distribution of repetitive DNA sequences are major driving forces of genome evolution and speciation.

To evaluate the sequence divergence level across the 24 Coelogyninae species, we compared the nucleotide diversity in the LSC, SSC and IR regions of the cp. genomes (Fig. 7). The IR regions have lower nucleotide diversity than that of the LSC and SSC regions, possibly due to copy correction between IR sequences caused by gene conversion [23, 24]. As expected, the greatest sequence divergence among the genomes was located in the intergenic areas, which is a common phenomenon in cp. genomes [25]. The dN/dS values varied among groups of different functional genes and provide measures of adaptation or functional constraint in protein-coding genes [26]. Although the majority of protein-coding genes were found to have been subjected to purifying selection, we identified 19 genes that evolved under positive selection (Additional file 7: Table S6). The genes accD, rpoC2, ycf1 and ycf2 have been reported under positive selection in orchid species [9]. Moreover, we found that genes ycf1a, rpoC2 and ycf2 possessed higher number of positive amino acid sites (113, 100, 96, respectively) within Coelogyninae species. The gene rpoC2 encodes the subunit of DNA-dependent RNA polymerase gene, and ycf1 and ycf2 genes encode unknown function proteins. These genes may play an important role in the adaptation to diverse environments for Coelogyninae species.

Overall, all phylogenetic analyses resulted in largely identical tree topologies of Coelogyninae with the incongruence mainly occurred in the interspecific relationships within clades (Fig. 8; also see Additional file 8: Fig. S2, Additional file 9: Fig. S3, and Additional file 10: Fig. S4). For the whole plastome data, Coelogyninae species fell into four major clades (A, B, C, and D), with clade A consisting of four to five subclades. The monophyly of the four major clades of Coelogyninae represented in this study was well-resolved, with each clade recovered with strong to full support, except for minor interspecific differences. The early diverging Coelogyninae species composed of Bletillia + Thunia and Thuniopsis were successive sister lineages to the clade containing species of Pleione, Coelogyne and related genera. This relationship was compatible with our former phylogenetic study [5] inferred from the nuclear ITS and two plastid regions.

Presence of storage organs such as rhizomes, succulent stems, tubers, corms or pseudobulbs is important for understanding the adaptations of orchids to survive and thrive in harsh environments. Morphologically, species of Bletilla, Thunia and Thuniopsis all lack heteroblastic pseudobulbs [8, 27]. Bletilla species are characterized by having short stems with two to four leaves and irregular subterranean tuberous rhizomes. However, the latter two possess elongated stems with many distichous leaves [4]. For Thunia species, they have no subterranean corms or tubers but possess biennial, thick, bamboo-like stems that develop from the previous season’s growth, wither and die during the second season, while Thuniopsis has prominent subterranean corms which normally become dormant under stressful conditions [4, 5, 8]. The material of Aglossorrhyncha, Dilochia and Glomera was not available. In our former study, Dilochia was moderately supported as a close sister to Thunia [5]. Despite its superficial resemblance to Thunia, Dilochia species can be distinguished by having long-lived, stout but not fleshy stems. Instead, Aglossorrhyncha and Glomera species can usually be distinguished by more or less rhizomatous, often branching and basally rooting stems [4]. These two genera may also form separate lineages inferred from morphological comparisons and previous molecular data [5, 28, 29]. In contrast to all these species, Coelogyne and Pleione species have heteroblastic pseudobulbs with one or two apical leaves [4, 8]. Unlike the former, Pleione species have only short-lived deciduous pseudobulbs and leaves, and annually renew their pseudobulbs [4], which seem to be an intermediate state of true pseudobulbs. In our phylogenetic analysis (Fig. 8), the sampled species of Pleione formed a monophyletic clade B, which was resolved as sister to the clade A containing Coelogyne and related species.

The phylogenetic relationships among the major clades inferred by ML and BI analyses were almost congruent. Two sampled species of Coelogyne and Pholidota clustered into the Chelonistele + Dendrochilum clade, which was resolved as a distinct and divergent lineage, well separated from the Coelogyne s.s. and Pholidota s.s. clade. The result was consistent with previous studies using nuclear DNA and partial plastid sequences [5, 7, 30]. It was also clear that three sampled species of Pholidota and a well-supported small clade composed of monotypic Ischnogyne and Panisea species formed a monophyletic clade with maximum support, which we referred to as the Panisea + Ischnogyne clade. Similarly, monotypic Bulleyia and five sampled Pholidota species were consistently placed as sister group that together formed the Bulleyia clade with full support. The close affinity between Bulleyia, Panisea and these species of Pholidota was also indicated in our former results [13].

Despite these consistencies, topological ambiguities occur in the placement of a few Coelogyne species. The plastome-based ML analysis clearly placed C. cristata (type species of Coelogyne) as a sister to species sampled for section Ocellatae (C. corymbosa and C. punctulata) with maximal statistical support (100%), which was also highly supported (0.94) by BI analysis using the CDS sequences (Additional file 9: Fig. S3). These species clustered together as a moderately supported sister (76%) to C. flaccida and C. viscosa (Fig. 8). In plastome-based BI tree, however, C. cristata was inferred as sister taxon to a clade comprising P. imbricata (type species of Pholidota), P. pallida and Neogyna gardneriana (Additional file 8: Fig. S2). In all the analyses, species sampled for sect. Ocellatae (C. corymbosa and C. punctulata) were always resolved as a monophyletic clade with a high-resolution value. The plastome-based ML analysis showed that this clade was sister to C. cristata, whereas, it was clustered within the clade composed of the sampled species for Coelogyne sect. Fuliginosae (C. fimbriata and C. ovalis) and sect. Elatae (C. barbata) in plastome-based BI tree. Furthermore, the phylogenetic positions of C. flaccida and C. viscosa, were not consistent across the different datasets.

Based on the available molecular data, the placement of some Coelogyne species endemic to southwest China and the Sino-Himalayan Mountains has been unstable, probably because the limited sampling might have influenced the analyses. For such a complex, heterogeneous and large genus which circumscribes diverse taxa with varying morphological characteristics, the depth of sampling is far insufficient. The intergeneric relationships between Coelogyne and related genera remain unclear and require further study. Some uncertainties persist and resolving the phylogenetic relationships within certain taxa remains challenging. To clarify their relationship, further investigations using additional data may depend strongly on the inclusion of Coelogyne in a wider geographical distribution range, particularly species occurring in the Sino-Himalaya and adjacent regions.

Conclusion

Herein, we determined the genomic features, sequence divergences and mutation patterns in the chloroplast genomes of 24 Coelogyninae species. Comparison of genomic sequences revealed genomic variations across genera and species and provided valuable insights into the general evolutionary dynamics of Coelogyninae. Our phylogenomic analyses yielded a robust backbone phylogeny of Coelogyninae with its deep nodes well resolved. The results provide strong support for the relationships among the major groups, but also indicate non-monophyly of several genera. Although we obtained a well-supported topology consistent with earlier studies, it remains challenging to identify associated morphological characteristics in Coelogyne and its related genera. Thus, future studies with extensive taxon sampling and morphological evidence are needed.

Materials and methods

Sampling, DNA extraction and sequencing

Fresh mature leaves were plucked from the living plants cultivated at the greenhouse of South China Botanical Garden, Chinese Academy of Science (SCBG, CAS). Leaf tissues from each accession were immediately dried with silica gel for further DNA extraction. We sampled 15 species for Coelogyninae, including six Coelogyne species (C. corymbosa, C. cristata, C. flaccida, C. punctulata, C. rochussenii and C. viscosa), two Otochilus species (O. fuscus and O. porrectus), one Panisea species (P. tricallosa), two Dendrochilus species (D. apoense and D. cootesii), one Chelonistele species (Ch. sulphurea), representatives of monotypic Bulleyia (B. yunnanensis), Ischnogyne (I. mandarinorum) and Neogyna (N. gardneriana), respectively. The following cp. genome sequences were retrieved from the NCBI database: Bletilla striata (accession No: MT193723), Coelogyne barbata (accession No: NC_050858), Pholidota chinensis (accession No: ON880553), P. imbricata (accession No: ON880554), P. protracta (accession No: ON880560), P. ventricosa (accession No: ON880561), Pleione maculata (accession No: MW699846), Thunia alba (accession No: OL809658) and Thuniopsis cleistogama (accession No: OL809660).

Total genomic DNA was extracted using the Trelief plant genomic DNA kit, manufactured by TsingKe Biological Technology, Beijing, China. The quality of each extracted DNA samples was evaluated by 1% agarose gel electrophoresis. Libraries were constructed with insert sizes of approximately 250 to 350 bp from the same DNA sample. These qualified libraries were sequenced on an Illumina HiSeq TM2500 platform in 150 bp paired-end format performed by Novogene Bioinformatics Technology Co., Ltd., Beijing, China. For each sample, approximately 3G raw data were generated. Clean reads were obtained by removing adapter sequences, reads containing poly-N and low-quality ones.

Plastome assembly and annotation

All raw data were first filtered for trimming of any low-quality bases or adapter sequences by FASTP [31]. Then clean reads were used to perform a de novo assembly by NOVOPlasty v.4.2.1 [32] with the Bletilla striata cp. genome from NCBI (MT193723) as a reference. Reference-guided connecting was subsequently conducted using Bandage v.5.6.0 [33]. The quality of the newly assembled genomes was assessed on read level basis by aligning the trimmed raw reads to the de novo assemblies using Geneious v.9.1.8 [34]. The assembled cp. genomes were annotated using the online program DOGMA [35] and GeSeq [36], and further manually adjusted and confirmed using Geneious. Additionally, for predicting transfer RNA (tRNA) genes in genomic sequences, tRNAscan-SE v.1.31 [37] was performed with the default parameters. The circular chloroplast genome maps were drawn by OGDRAW v.1.3.1 [38]. All of the 15 newly generated Coelogyninae cp. genome sequences were deposited in the NCBI GenBank database under accession numbers OR687499–OR687513.

Repeat sequence analyses

The simple sequence repeat (SSR) loci in the 15 newly sequenced genomes, together with nine published genomes were examined via Perl script MISA v.1.01 [39]. To facilitate SSR detection, the search criteria were set as follows: 10 minimum repeat units for mononucleotide (p1) repeats, 5 for dinucleotide (p2) repeats, 4 for trinucleotide (p3), and 3 for tetranucleotide (p4), penta-, and hexa-nucleotides, respectively. The maximum distance between two adjacent SSRs in a compound SSRs (c SSRs) was less than 100 bp according to previous reports [40] and the default parameters of the MISA software. REPuter v.2.74 program [41] was utilized to analyze long repetitive sequences including forward (F), reverse (R) complement (C), and palindromic (P) repeats in the 24 Coelogyninae plastomes. The minimal repeat size was set to 30 bp long per repeat unit with Hamming distance of 3 bp.

Plastome comparison and sequence divergence analyses

Chloroplast genome similarity was assessed using BLAST Atlas on the GView Server (https://server.gview.ca/) with 100 bp connection windows [42] with Thuniopsis cleistogama OL809660 as a reference. The position changes in the IR/SC junction and their adjacent genes of these cp. genomes were assessed and compared using the IRscope online program [43]. For identifying hypervariable regions, the whole-genome alignment was visualized using Shuffle-LAGAN mode [44] included in mVISTA v.2.0 [45]. Nucleotide diversity (Pi) values were calculated by DnaSP v6.12.03 software [46] with a sliding window analysis. The window length was set to 600 bp with a step size of 200 bp.

Positive selective tests

Non-synonymous (dN) and synonymous (dS) nucleotide substitution rates, as well as their ratios (ω = dN/dS) were analyzed using CODEML program from the PAML package v.4.8a to detect sites under diversifying selection [27, 47]. The codon substitution model enables to estimate the selective pressures on protein-coding regions across divergent lineages via comparing their ratios. The ω ratio measures the mode of natural selection acting on the protein genes, with ω > 1 indicating positive (adaptive) selection, ω = 1 indicating neutral evolution, while ω < 1 signifying negative (purifying) selection. Bayes Empirical Bayes (BEB) inferences [48] were used to assess the statistical significance of potential positively selected sites. In the BEB analysis, posterior probability higher than 0.95 and 0.99 indicated sites that were under positive selection and strong positive selection, respectively (Additional file 7: Table S6).

Phylogenetic analyses

To infer the phylogenetic relationships within Coelogyninae, we performed a series of phylogenetic analyses using both maximum likelihood (ML) and Bayesian inference (BI) methods based on two datasets (complete plastome sequences and CDS sequences). A total of 56 accessions from 42 species of 13 genera (Bletilla, Bulleyia, Chelonistele, Coelogyne, Dendrochilus, Ischnogyne, Neogyna, Otochilus, Panisea, Pholidota, Pleione, Thunia and Thuniopsis) representing the main lineages of Coelogyninae were included, plus one outgroup species. We were unable to obtain the material of Aglossorrhyncha, Dilochia, and Glomera, which were not included in the analyses. All the plastome sequences were aligned using MAFFT [49] and adjusted manually by BioEdit [50]. The ML tree was generated using FastTree 2 [51] and implemented in RAxML v.8.2.11 [52] under the generalized time-reversible GTR + G model. Nodes were evaluated by Shimodaira–Hasegawa (SH) tests [53] to detect significant topology. BI analyses of phylogeny were performed in MrBayes v.3.2.7 [54]. The Markov Chain Monte Carlo (MCMC) analyses were run for 10 million generations, employing TVM + F + R3 as the optimal model, as determined by ModelTest-NG 0.1.6 [55]. These trees were sampled every 1,000 generations, with a burn-in of 25% and the remaining trees were used to generate a majority-rule consensus tree and calculate the probability of each branch of the posterior probability (PP).