From light into shadow: comparative plastomes in Petrocosmea and implications for low light adaptation

Abstract

Background

Plastids originated from an ancient endosymbiotic event and evolved into the photosynthetic organelles in plant cells. They absorb light energy and carbon dioxide, converting them into chemical energy and oxygen, which are crucial for plant development and adaptation. However, little is known about the plastid genome to light adaptation. Petrocosmea, a member of the Gesneriaceae family, comprises approximately 70 species with diverse light environment, serve as an ideal subject for studying plastomes adapt to light.

Results

In this study, we selected ten representative species of Petrocosmea from diverse light environments, assembled their plastid genomes, and conducted a comparative genomic analysis. We found that the plastid genome of Petrocosmea is highly conserved in both structure and gene content. The phylogenetic relationships reconstructed based on the plastid genes were divided into five clades, which is consistent with the results of previous studies. The vast majority of plastid protein-coding genes were under purifying selection, with only the rps8 and rps16 genes identified under positive selection in different light environments. Notably, significant differences of evolutionary rate were observed in NADH dehydrogenase, ATPase ribosome, and RNA polymerase between Clade A and the other clades. Additionally, we identified ycf1 and several intergenic regions (trnH-psbA, trnK-rps16, rpoB-trnC, petA-psbJ, ccsA-trnL, rps16-trnQ, and trnS-trnG) as candidate barcodes for this emerging ornamental horticulture.

Conclusion

We newly assembled ten plastid genomes of Petrocosmea and identified several hypervariable regions, providing genetic resources and candidate markers for this promising emerging ornamental horticulture. Furthermore, our study suggested that rps8 and rps16 were under positive selection and that the evolutionary patterns of NADH dehydrogenase, ATPase ribosome, and RNA polymerase were related to the diversity light environment in Petrocosmea. This revealed an evolutionary scenario for light adaptation of the plastid genome in plants.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-024-05669-2.

Background

Plastids serve as the place where green plants perform photosynthesis, capturing energy from sunlight and storing it in ATP and NADPH, while releasing oxygen from water, and then assimilating carbon dioxide to produce organic molecules through the Calvin cycle [1, 2]. The core enzymes crucial for this process were co-encoded by the plastome and the nuclear genome [3–5]. Thus, the plastid acts as the receptor of light energy and the executor of photosynthesis. Plastids originated from the endosymbiosis of cyanobacteria billions of years ago [6]. Their genomes have undergone significant simplification, containing only several dozen genes related to photosynthesis, as well as their genome replication and transcription [7, 8]. Therefore, the evolution of the plastome was closely related to adapting to the varied light environments [9, 10]. Typical angiosperm plastomes are relatively conserved, usually around 150 kb in size [11]. They are inherited uniparentally, so recombination rarely occurs [12]. Additionally, the plastome exhibits a very high copy number in cells [13]. In summary, plastid DNA is relatively easy to enrich, and plastomes are easy to assemble. Therefore, plastomes are widely used in molecular phylogenetic studies at various taxonomic levels and in the development of DNA barcoding [14–16].

The light intensity, temperature, water stress or atmospheric CO₂ concentration can impact photosynthesis efficiency, influencing the evolutionary pattern of plastomes [9, 17–22]. A large amount studies have corroborated this notion. For instance, a comparison of Ferula plastomes revealed that twelve plastid genes encoding PS II (Photosystem II), ATPase and NADH complexes underwent positive selection, potentially aiding this genus in adapting to extreme habitats [9]. The psbA gene encoding the PS II complex, underwent positive selection in shade-tolerant and sun-loving rice species [21]. Whereas the cemA experienced positive selection in Swertia, possibly related to the decrease of carbon dioxide concentration during the late Miocene [22]. To advance the study of plastomes and their adaptation to varied light conditions, it is imperative to investigate a monophyletic group with diverse light environments.

Petrocosmea, a member of the Gesneriaceae, comprises approximately 70 species and is predominantly distributed in the south of the Qinling Mountains [23, 24]. Renowned for its vibrant green leaves, bright flowers, unique shapes, and prolonged flowering period, this genus has emerged as a prominent choice for indoor cultivation [25]. Most Petrocosmea species inhabit relatively shaded light environments; however, detailed investigations reveal variations within these light conditions. The phylogeny of this genus can be divided into five main clades (Clade A-E) based on six plastid and two nuclear fragments [26, 27]. Species in Clade A are primarily distributed on open limestone cliffs, exposed to a brief period of direct sunlight, whereas species in Clade B inhabit mountain crevices on the shady side. In Clade C, species are found in both brighter habitats at cave entrances and darker habitats inside caves. The species of Clade D were distributed on large rocks or at the entrance of caves, with occasional direct sunlight. The species of Clade E were mainly distributed on the mountains with short shrub covered or not, where they could be exposed to direct sunlight for short periods. Hence, species from this genus, with diverse light intensity habitats, serve as ideal subjects for studying how plastomes adapt to varied light conditions. Furthermore, given its significant ornamental value and the continuous discovery of new species in this genus, there is an urgent need to develop promising DNA barcodes.

In this study, we sampled ten species from each of the five main clades of Petrocosmea with diverse light conditions to sequence, assemble and annotate their plastomes. Subsequently, we conducted a comprehensive comparative analysis among Petrocosmea plastomes, encompassing gene content, repeats, boundary sliding, and phylogeny. We then accessed the nucleotide substitution rates and selective pressure of both single gene matrixes, functional categories, and the concatenated matrix to infer the light adaptation of the plastome. Furthermore, we calculated the nucleotide diversity to mine reliable candidate DNA barcodes. This study provides valuable insights into the evolutionary patterns of plastomes response to diverse light conditions and offers crucial information for the identification and conservation of this promising candidate for ornamental horticulture.

Materials and methods

Taxon sampling and sequencing

According to prior phylogenetic studies [26, 27], we collected 10 representative species from all five different clades of Petrocosmea in the wild, and planted them in the greenhouse of Qilu Normal University (Table S1). The morphological characteristics and habitats of the selected species were validated through extensive field surveys. The light intensity measurement and evaluation process for microhabitats of Petrocosmea species is as follows: (1) From September to October of 2019 through 2021, we gradually measured the light intensity in the environments of 12 populations of 10 Petrocosmea species involved in this study. (2) Measurements were conducted on sunny days between 10 AM and 4 PM, with readings taken every hour, avoiding short-term cloud cover interference (Kurzanleitung Testo 540 light meter). (3) The empirical light intensity data were preemptively categorized into five levels. (4) An ANOVA was then conducted to analyze the differences in light intensity across these levels, followed by further validation using an LSD (Least Significant Difference) test.

Fresh leaf materials were sampled, and total genomic DNA was extracted using the modified CTAB method [28, 29]. The quality control was carried out by 1% agarose gel electrophoresis and Qubit 3.0. Subsequently, DNA libraries were constructed according to the manufacturer’s instructions of the MGIEasy DNA library preparation kit (Beijing Genomics Institution, Shenzhen, CHN) and sequenced on the DNBseq platform (insert size = 300 bp, PE150). The sequencing data for each species exceeded 10 Gb (Table S1). Lysionotus pauciflorus (NC_034660.1) and Petrocodon jingxiensis (NC_044477.1) were selected as outgroups according to the molecular phylogeny of Gesneriaceae [30].

Plastid genome assembly and annotation

The raw sequencing data were assessed and quality-controlled using FastQC v0.11.8 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and Trimmomatic v 0.39 [31]. Specifically, low-quality bases and adapter sequences were removed (ILLUMINACLIP: TruSeq3-PE.fa:2:30:10 HEADCROP:10 LEADING:20 TRAILING:20 SLIDINGWINDOW:4:15 MINLEN:36 AVGQUAL:28 TOPHRED64) to obtain clean data. Clean data were de novo assembled using SPAdes v3.15.2 [32]. Subsequently, assembly graphs were simplified and manually processed by Bandage v0.8.1, according to sequencing depth and connections [33]. Finally, complete plastomes of our samples were obtained. The Plastid Genome Annotator (PGA) was used to annotate the plastomes assembled in this study, and the GenBank-formatted file of the Amborella trichopoda plastome (AJ506156) was selected as the reference [34, 35]. Then, the annotation files were manually checked using Sequin v16.0 and submitted to GenBank (Table 1). Additionally, we extracted single nucleotide polymorphism (SNP) and insertion and deletion (indel) among Petrocosmea plastomes by SNP-sites v2.5.1 [36]. Finally, Circos v0.69-9 and OGDRAW v1.3.1 were used to draw the plastome and variants circle map [37, 38].

Table 1.

General information of plastomes in Petrocosmea and outgroups

Species	Code	GC content (%)				Length (bp)				Gene content				GenBank ID
		LSC	SSC	IR	Total	LSC	SSC	IRs	Total	Total	PCGs	tRNA	rRNA
Petrocodon jingxiensis		35.40	31.10	43.20	37.50	84,154	18,058	25,420	153,056	130 (113)	85 (79)	37 (30)	8 (4)	NC_044477
Lysionotus pauciflorus		35.40	31.20	43.20	37.50	85,079	17,839	25,469	153,856	132 (114)	87 (80)	37 (30)	8 (4)	NC_034660
P. menglianensis	Q1	35.60	31.30	43.20	37.60	84,586	18,205	25,456	153,703	130 (113)	85 (79)	37 (30)	8 (4)	PP997971
P. bicolor	Q2	35.60	31.30	43.20	37.60	84,679	18,123	25,454	153,710	130 (113)	85 (79)	37 (30)	8 (4)	PP997972
P. wui	Q4	35.50	31.10	43.20	37.50	84,401	18,212	25,385	153,383	130 (113)	85 (79)	37 (30)	8 (4)	PP997973
P. duclouxii	Q5	35.60	31.20	43.20	37.60	84,322	18,312	25,343	153,320	130 (113)	85 (79)	37 (30)	8 (4)	PP997974
P. shilinensis var. changhuensis	Q6	35.60	31.60	43.20	37.70	84,764	17,956	25,378	153,476	130 (113)	85 (79)	37 (30)	8 (4)	PP997975
P. serica	Q7	35.70	31.50	43.20	37.70	84,335	17,806	25,384	152,909	130 (113)	85 (79)	37 (30)	8 (4)	PP997976
P. barbata	Q8	35.50	31.10	43.20	37.50	84,328	18,342	25,440	153,550	130 (113)	85 (79)	37 (30)	8 (4)	PP997977
P. cavaleriei	Q9	35.50	31.00	43.10	37.50	84,590	18,354	25,435	153,814	130 (113)	85 (79)	37 (30)	8 (4)	PP997978
P. sinensis	Q10	35.50	31.10	43.20	37.50	84,890	18,365	25,406	154,067	130 (113)	85 (79)	37 (30)	8 (4)	PP997979
P. nervosa	Q12	35.60	31.40	43.20	37.60	84,741	18,302	25,479	154,001	130 (113)	85 (79)	37 (30)	8 (4)	PP997980

LSC, large single copy region; SSC, small single copy region; IR, inverted repeat; PCGs, protein-coding genes; tRNA, transfer RNA; rRNA, ribosomal RNA. The numbers in parenthesis represented the number of unique genes in plastome

Phylogenetic reconstruction

We reconstructed the maximum-likelihood tree of Petrocosmea based on all plastid protein-coding genes using RAxML v 8.2.12 [39], with Lysionotus pauciflorus (NC_034660.1) and Petrocodon jingxiensis (NC_044477.1) as outgroups. The specific steps were as follows: (a) The selectSites.pl script was used to extract both the nucleotide and protein sequences of protein-coding genes (PCG) from each sampled species (http://raven.wrrb.uaf.edu/~ntakebay/teaching/programming/perl-scripts/perl-scripts.html), (b) MAFFT v7.490 was used to align the protein matrix of each single gene, and pal2nal.pl v13 was used to align the DNA matrix of each single gene based on the aligned protein matrix [40, 41], (c) ZORRO was used to remove unreliable alignment regions [42], (d) FASconCAT-G v1.04 was applied to concatenate single gene DNA matrixes [43], (e) RAxML v8.2.12 was applied to reconstruct the Maximum Likelihood tree under the GTRGAMMA model, with 1,000 bootstrap replicates [39]. The phylogenetic tree was visualized and refined using iTOL v6.8.1 [44].

Comparative plastome analysis

The similarity and conservation of plastomes among Petrocosmea and outgroups were analyzed by mVISTA under Shuffle-LAGAN mode, which could also detect rearrangements at whole genome level [45, 46]. Due to the large variation in the Petrocodon jingxiensis plastome, we removed it from this analysis. Additionally, we selected outgroups as references to detect the contraction and expansion of the IR regions among Petrocosmea species, along with the sliding genes around the SC-IR (single copy region and inverted repeat region) boundary. The global pattern of structural variation was displayed according to the phylogenetic tree of Petrocosmea.

MISA-web [47], Tandem Repeats Finder v4.09 [48] and REPuter [49] were applied to identify simple sequence repeats (SSRs), tandem repeat sequences, and dispersed repeat sequences in Petrocosmea and outgroups plastomes, respectively. Minimum repeat units were set to 10, 6, 5, 5, 5 and 5 for mononucleotide, dinucleotide, trinucleotide, tetranucleotide, pentanucleotide and hexanucleotide to identify SSRs, respectively. Tandem Repeats Finder v4.09 [48] was configured with default parameters to identify tandem repeats. The minimal repeat size was set to 30 bp in REPuter [49] to identify dispersed repeats.

Evolutionary analysis

We evaluated the nucleotide substitution rate (d_S, synonymous substitution rate and d_N, nonsynonymous substitution rate) and selective pressure (w) from the following four aspects: (a) the nucleotide diversity (pi) in Petrocosmea plastomes at the whole genome level, (b) the d_N, d_S and w (d_N/d_S) in pairwise comparisons of the concatenated matrix of all PCGs, (c) the d_N, d_S and w of the each branch of single gene matrix, (d) identification of sites in a particular branch under positive selection.

The Petrocosmea plastomes were aligned by Bioedit v7.2.5 [50] manually, and nucleotide diversity (pi) was calculated by DnaSP v6 (word size = 600, step size = 200) [51].

EasyCodeML v1.0 [52] was employed to convert the concatenated matrix to the PAML format. Subsequently, the codeml of PAML v4.9j [53] in runmode = -2 calculated the pairwise d_N, d_S and w.

The program codeml of PAML v4.9j [53] was applied to calculate the d_N, d_S and w of each branch according to the phylogenetic tree generated in the previous step (runmode = 0, CodonFreq = F3 × 4, model = 2). Statistical analysis of the d_N, d_S and w of each branch was performed by R v4.3.1.

EasyCodeML v1.0 [52] was utilized to identify positive selective sites and branch under the branch-site model. Clade A, Clade B, Clade C, Clade D, Clade E, and P. serica were selected as foreground branch, respectively. The likelihood ratio test was conducted by EasyCodeML v1.0 [52].

Results

General features of plastomes in Petrocosmea

All ten newly assembled plastomes of Petrocosmea were assembled into a single circular molecule (Fig. 1) with sizes ranging from 152,909 bp to 154,067 bp (mean = 153,593 bp, SD = 344.4468) (Table 1). They exhibited a typical quadripartite structure,

Fig. 1.

Annotation and variants distribution map of the newly assembled Petrocosmea plastomes. From the outside to the inside, the first circle exhibited the annotation information of these plastid genomes, where the gene name located outside indicated clockwise transcription, the gene name located inside indicated counterclockwise transcription. The colors corresponded to functional categories at the bottom. The second circle exhibited the distribution of variants among these plastomes, with the blue line indicating single nucleotide polymorphism (SNP) and the red line indicating insertion and deletion (indel). The third circle exhibited the partitions of the plastomes, with the dark green indicating the large single-copy regions (LSC), the light green indicating the small single-copy regions (SSC), and the orange indicating inverted repeat regions (IR)

comprising a large single-copy region (LSC), a small single-copy region (SSC) and a pair of inverted repeat regions (IR). The sizes of LSC, SSC and IR ranged from 84,322 to 84,890 bp, 18,257 to 18,365 bp, and 25,343 to 25,479 bp, respectively. The GC content of the newly assembled plastomes ranged from 37.50 to 37.60% (mean = 37.58%, SD = 0.0008), with the GC content of LSC, SSC and IR ranging from 35.50 to 35.70%, 31.00–31.60%, and 43.10–43.20%, respectively. These ten newly assembled plastomes all contained 113 unique genes, including 79 unique protein-coding genes (PCGs), 30 unique tRNA genes and four unique rRNA genes. Six PCGs (ndhB, rpl2, rpl23, rps7, rps12 and ycf2) had two copies because they are located in the IR regions. Additionally, eight PCGs (atpF, ndhA, ndhB, petD, rpl2, rpl16, rpoC1 and rps16) contained one intron, while three PCGs (clpP, rps12 and ycf3) contained two introns. Only rps12 contained one trans-spliced intron (Fig. 1, Table S2). There was a total of 4,139 variants in the ten newly assembled plastomes, including 3,856 SNPs and 283 indels (Table S3). The density of variants in the IR regions was notably lower than that in the single-copy regions (LSC and SSC) (Fig. 1). The Petrocosmea plastomes were relatively conserved, exhibiting a very small variation in size, GC content and gene content.

Plastid phylogeny and habitat of Petrocosmea

The phylogenetic tree of Petrocosmea species, reconstructed based on all plastid PCGs, can be divided into five main clades [26, 27]. Clade A, D and E were all supported by 100% bootstrap values (Fig. 2A). P. bicolor and P. menglianensis formed the first diverged clade (Clade A), followed by the divergence of Clade B and C. Clade D and E formed the closely related sister clades, consistent with the previous studies. The monophyly formed by Clade B, C, D and E was supported by 100% bootstrap value. However, the support value of the relationship of Clade C, D and E was only 55%, in line with the phylogenetic tree based on the CDS (1st + 2nd) of PCGs (Fig. S1), indicating the phylogenetic position of Clade C may be controversial.

Fig. 2.

Phylogenetic relationships and habitats of the sampled species in this study. (A) Maximum likelihood tree reconstructed based on all PCGs, with numbers at nodes indicated bootstrap values. (B) Light intensity of each species’ habitat assessed based on extensive field surveys. (C) The morphology and habitat characteristics of the sampled species

Through a long-term field survey, we found that the species in Clade A are generally distributed in open limestone cliffs under scattered small shrubs, experiencing direct sunlight for a short period of time, so they are ranked in terms of relatively strong light intensity (Fig. 2B and C, Table S5). Moreover, the leaves of species in Clade A differentiated into inner and outer, with the outer leaves being larger, reaching lengths to 10–30 cm. The two species of Clade B were mostly distributed in the crevices of mountains, where the light intensity was weak. There was no differentiation between inner and outer leaves, and the leaf size was around 10 cm. P. shilinenisis var. changhuensis in Clade C was also mostly distributed in crevices of mountains with weak light, while P. serica was mainly distributed in the cracks of large rocks with a darker environment. The leaves of both species show no differentiation between inner and outer and are smaller compared with Clade A species. The two species of Clade D were distributed on large rocks or at the entrance of caves, with occasional direct sunlight. The leaves also showed no inner or outer differentiation. The two species of the Clade E were mainly distributed on the mountains with short shrub covered or not, where they could be exposed to direct sunlight for short periods.

Repeat content and structural variations

A comparative analysis was conducted on tandem and dispersed repeats within the plastid genomes of Petrocosmea and outgroups. Tandem repeats were categorized into simple sequence repeats (SSR, size = 1 ~ 6 bp) and general tandem repeats (> 6 bp) according to the length of the repeat unit. Mononucleotide repeats dominated the simple sequence repeats in the Petrocosmea plastomes, followed by dinucleotide and tetranucleotide repeats. Trinucleotide repeats were rare, and no pentanucleotide repeats were observed compared to the outgroups (Fig. 3A, Table S5). The general tandem repeats predominantly ranged from 10 to 30 bp, with fewer instances greater than 30 bp. The two species from Clade B, P. duclouxii and P. wui, contained fewer tandem repeats in terms of both number and accumulated length (Fig. 3B, Table S6). According to the match direction of the dispersed repeats, they were categorized into four types: forward (F), reverse (R), complement (C), and palindromic (P) matches. Among them, forward and palindromic matches predominated in Petrocosmea plastomes (Fig. 3C, Table S7). The dispersed repeats were further categorized into four categories according to their length. In Petrocosmea plastomes, the majority of the dispersed repeats ranged from 30 to 45 bp (Fig. 3D, Table S8). Remarkably, P. wui from Clade B accumulated the largest number of dispersed repeats, whereas P. serica from Clade C accumulated the longest length, reaching 1432 bp.

Fig. 3.

Statistics on various types of repeats in the sampled plastomes. (A) The number of SSR with different lengths. “mono”, “di”, “tri”, “tetra”, “penta” indicated simple mono-, di-, tri-, tetra- and pentanucleotide microsatellites, respectively. (B) The number and accumulated length of tandem repeats with different lengths. The line chart indicated the accumulated length, while the bar chart indicated the number of repeats. (C) The number of the different type dispersed repeats. “F” indicated the forward repeats, “R” indicated the reverse repeats, “C” indicated the complement repeats, and “P” indicated the palindromic repeats. (D) The number and accumulated length of dispersed repeats with different length. The line chart indicated the accumulated length, and the bar chart indicated the number of repeats

There were no rearrangements detected in the plastomes of Petrocosmea and Lysionotus pauciflorus by mVISTA at the whole genome level (Fig. 4). The SC-IR boundary was relatively conservative in Petrocosmea plastomes (Fig. 5). The border of the LSC-IRb region was in the intergenic region, and the distances from rps19 and rpl2 to the boundary varied. Only in P. serica, the LSC-IRb boundary shifted 10 bp towards IRb, resulting in a distance of 237 bp from rps19 to the boundary and 42 bp from rpl2 to the boundary. In other species, the distance from rps19 to the boundary ranged from 245 to 247 bp, and the distance from rpl2 to the boundary ranged from 32 to 34 bp. The ycf1 gene was located at the boundary of IRb-SSC, and its length in the IR region was 802 bp in the two species from Clade A, while it ranged between 770 and 782 bp in other species. The ndhF gene was located at the boundary of SSC-IRa in P. menglianensis, P. bicolor, P. wui, P. serica, and P. barbata, while it was entirely located in the SSC region in P. duclouxii, P. shilinensis var. changhuensis, P. cavaleriei, P. sinensis, and P. nervosa. The distance of rpl2 from the IRa-LSC boundary varied from 87 to 97 bp, but the distance of psbA from the boundary ranged from 371 to 481 bp.

Fig. 4.

Sequence conservation and identity among the sampled plastomes

Fig. 5.

Schematic diagram of LSC, SSC and IR boundary sliding in Petrocosmea and outgroup. The squares indicated the border genes, and the gene names were marked at the top. The arrows and numbers indicated the distance between the border and the border gene

Nucleotide diversity and evolutionary rate

We calculated the nucleotide diversity across the complete plastome based on 600 bp as sliding windows and 200 bp as steps, and found that the nucleotide diversity of the single copy regions (LSC 0.00885 and SSC 0.01293) was higher than that of the IR regions (0.00188), where the nucleotide diversity of ycf1 and several intergenic regions (trnH-psbA, trnK-rps16, rpoB-trnC, petA-psbJ, ccsA-trnL, rps16-trnQ and trnS-trnG) were higher than 0.02 (Fig. 6A).

Fig. 6.

Nucleotide diversity and pairwise selective pressure in Petrocosmea plastomes. (A) Nucleotide diversity in Petrocosmea plastomes. The color blocks on the top indicated genes and functional categories consistent with Fig. 1. (B) Pairwise selective pressure in Petrocosmea plastomes. The lower triangle indicated the heat map of pairwise selective pressure, and the upper triangle was the corresponding values

The selective pressures of the concatenated all PCGs matrix between any two species were below 1, and the majority of values were under 0.5. Only the values between P. bicolor and P. menglianensis and between P. sinensis and P. nervosa were greater than 0.5 (Fig. 6B). These results indicated that the plastid PCGs were generally under purifying selection and were relatively conserved. Furthermore, we also calculated the selective pressure for each PCG. The d_S values of petG and psbJ were too small, causing their w values to be too large. Only the w values of rps8 and rps16 were greater than 1, indicating that they were under positive selection. Whereas the w values of the other genes were below 1, with most of them even below 0.5, these results also supported that most plastid genes were under purifying selection (Table S9).

To evaluate whether selective pressures varied among branches, we calculated the d_S, d_N and w values of each PCG in five branches. We also classified the PCGs according to their functions and calculated the selective pressures experienced by different functional categories among branches (Table S2). We found that the d_S and d_N of clade A, located in a relatively strong light density environment, differed from those of clade E in a strong light density environment, and were significantly different from those of Clade B, Clade C and Clade D which located in a relatively weak lightly environment (Fig. S2 and S3). A similar tendency was observed in the plastid genes encoding the subunits of NADH dehydrogenase, Ribosome, and RNA polymerase, but this tendency was not found in PSI, PSII, ATP synthase and cytochrome b/f complex. However, there was no difference in selective pressure in either all genes or genes in each functional category among five clades (Fig. 7, Table S10). The w values of rps8 were greater than 1 in P. wui, P. shilinensis var. changhuensis, P. serica, P. barbata and P. cavaleriei, and the w values of rps16 were also greater than 1 in P. bicolor, P. wui and P. serica, indicating that most PCGs experienced purifying selection, and only rps8 and rps16 under positive selection (Table S10).

Fig. 7.

The selective pressure of plastid PCGs from different functional categories across five clades (Clade A, Clade B, Clade C, Clade D and Clade E) in Petrocosmea (refer to Fig. 2A). (A) The selective pressures of all plastid PCGs. (B) The selective pressures of plastid PCGs encoded ATP synthase. (C) The selective pressures of plastid PCGs encoded NADH dehydrogenase. (D) The selective pressures of plastid PCGs encoded Photosystem I (PS I). (E) The selective pressures of plastid PCGs encoded Photosystem II (PS II). (F) The selective pressures of plastid PCGs encoded Cytochrome b/f complex. (G) The selective pressures of plastid PCGs encoded Ribosome. (H) The selective pressures of plastid PCGs encoded RNA Polymerase. (I) The selective pressures of plastid PCGs encoded Maturase, Protease, Envelope membrane protein, Acetyl-CoA carboxylase, C-type cytochrome synthesis gene, Translation initiation factor and Proteins of unknown function (Other). Detailed information on the gene functional classification can be found in Table S10

In addition, to identify branches and sites under positive selection, five clades A, B, C, D, E, B + C + D + E, and P. serica were used as foreground branches, respectively. The LTR P-values of all genes were greater than 0.05, and some sites under positive selection were identified in certain genes, indicating that all genes of each clade were subject to purifying selection, but some sites were under positive selection, particularly when Clade A was used as the foreground branch (Table S11-S16).

Discussion

Relatively conserved plastomes in Petrocosmea

In this study, we newly sequenced, assembled and annotated ten plastomes of ten representative species from five main clades in Petrocosmea. The interspecific relationships reconstructed through phylogenomic analysis in Petrocosmea were consistent with previous researches [26, 27]. The ten newly sequenced plastomes were a single quadripartite molecule, composed of LSC, SSC, and a pair of IRs, which were similar to the plastomes of other genera in Gesneriaceae and photoautotrophic angiosperm [11, 54–57]. Notably, all ten newly sequenced plastomes contained the same set of genes, including 79 unique PCGs, 30 unique tRNA genes and 4 unique rRNA genes. The gene content is more conserved than that in other genera of Gesneriaceae (i.e., Haberlea, Paraboea, Primulina and Streptocarpus) [56, 58]. Those results indicated that the plastomes of Petrocosmea were conserved in gene content.

There were slight variations in size, GC content, repeat content and the sliding of the SC-IR boundary. Among the newly sequenced plastomes, the plastome of P. sinensis (154,067 bp) was the largest, while the plastome of P. serica (152,909 bp) was the smallest. However, the lengths of their IR were 25,406 bp and 25,384 bp respectively, with a difference of only 22 bp. Therefore, the expansion and contraction of the IR region were not the main factors causing the size variation of plastomes in Petrocosmea. Secondly, the tandem repeats accumulated more in the P. sinensis plastome, while the dispersed repeats accumulated more in the P. serica plastome. Therefore, the accumulated repeats were also not the main factor causing the size variation of plastomes in Petrocosmea. According to our observation, the P. serica plastome has two large deletions of intergenic regions in LSC (rpoB - trnC-GCA) and SSC (rpl32 - ndhF). In summary, the deletions of the intergenic region were the main factor causing the size variation of plastomes in Petrocosmea.

Adaptation to diversity light environment of Petrocosmea plastomes

The newly sequenced plastomes from ten representative species of Petrocosmea spanned a range of light intensities, encompassing Clade A with a relatively strong intensity habitat to Clade C, which thrives in relatively dark caves. This also covered the five main clades of this genus. The selective pressure of the concatenated matrix of PCGs between any two species was lower than 1 and all PCGs were lower than 1 except rps8 and rps16. This result was consistent with the observation in 773 angiosperm plastomes that the selective pressure was below 1, indicating the plastid PCGs were under purifying selection [59]. The above phenomenon may be related to uniparental inheritance [12], or it may be because the plastomes were extremely simplified compared with their ancestor and contained housekeeping genes related to plastid energy metabolism and gene replication and expression [6, 59]. Alternatively, there were a plenty of complexes encoded by both the plastome and nuclear genome, this may be another force constraining the evolution of plastomes [4, 60]. Additionally, there were some nuclear genes responsible for plastome replication, repair, and recombination, maintaining a low evolutionary rate and selective pressure [61].

Only rps8 and rps16 were subjected to positive selection, which was consistent with previous research results that plastid genes related to information processing (e.g., rps2, rpl2, and rpoA) were more trend to relax selection than genes related to energy production (e.g., psaA, psbA and ndhA) [59]. More importantly, rps8 and rps16 encoded 30 S ribosomal subunits, which played primary functions in plastid protein translation [62]. Therefore, the positive selection on those genes may affect the synthesis of other plastid proteins, thereby influencing the adaptation of the other genes.

Although there was no difference in the selective pressure among the five main clades in Petrocosmea, the synonymous and non-synonymous substitution rate of NADH dehydrogenase, Ribosome and RNA polymerase complexes in Clade B, Clade C, and Clade D, which were distributed to a relatively weak light intensity habitats, were significantly higher than those in Clade A, which had a relatively strong light intensity habitat. The NADH dehydrogenase, located in the thylakoid membrane, participates in cyclic electron transport [63]. And it was closely related to the maintenance of normal growth and yield of rice in low light conditions [64, 65]. Whereas the Ribosome and RNA Polymerase played essential roles in the transcription and translation of plastid PCGs, which may affect photosynthetic efficiency through protein abundance [66, 67]. Correspondingly, we also identified sites under positive selection in atpB, ndhB, rpl20, rpoA and others. Those results suggested that the mutation rate of those plastid PCGs in species from Clade B, Clade C and Clade D, distributed in low-light environments, were accelerated, providing raw materials for adaptation [68].

Promising DNA barcodes in Petrocosmea plastomes

Petrocosmea, characterized by its unique plant architecture and elegant floral pigmentation, represents a promising candidate for ornamental horticulture [25]. Nonetheless, the taxonomic intricacies arising from the high morphological similarity among closely related species, particularly in the absence of discernible floral organs, present challenges to accurate classification based solely on vegetative characteristics [23, 24, 27, 69]. The continuous influx of newly reported new taxa has expanded the taxonomic breadth of the genus, growing from 42 species in 2015 to approximately 70 at present [26, 69]. Leveraging the diverse gene segments within the plastome as robust DNA barcodes emerges as a viable strategy for precise and reliable species identification [70, 71]. Indeed, Qiu et al. [26] have used six plastid fragments (atpI-H, matK, trnH-psbA, rps16, trnL-trnF, trnT-trnL) and two nuclear fragments to reconstruct the interspecies relationships of Petrocosmea, and the relationships are consistent with those reconstructed based on the PCGs of plastome, indicating that the plastome and its highly variable regions can be used as candidate molecular markers. In this study, we identified ycf1 and several intergenic regions (trnH-psbA, trnK-rps16, rpoB-trnC, petA-psbJ, ccsA-trnL, rps16-trnQ and trnS-trnG) were higher than 0.02 in Petrocosmea plastomes. Notably, ycf1 and trnH-psbA are the most promising DNA barcodes of plants [72]. Therefore, those hypervariable regions can be used as the candidate barcodes for identification of this genus.

Conclusion

In this study, based on previous research, we selected species representing five Clades of Petrocosmea, with different typical habitats. We reconstructed the phylogenetic relationships of ten species based on plastid genes, which is consistent with previous research, dividing this genus into five main lineages. We conducted a comparative analysis of the plastomes and found that the plastomes were relatively conserved in Petrocosmea. We analyzed the plastid genes of species in different habitats and found that the rps8 and rps16 genes were under positive selection, while other genes were mainly subject to strong negative selection. When we divided genes into functional categories, we found significant differences in genes such as NADH dehydrogenase, ATPase ribosome and RNA polymerase between Clade A and the rest Clades. In addition, through branch site model analysis, although most genes undergo negative selection, there are still many sites that are subject to positive selection. We also found that ycf1 and several intergenic regions (trnH-psbA, trnK-rps16, rpoB-trnC, petA-psbJ, ccsA-trnL, rps16-trnQ and trnS-trnG) can be used as a candidate DNA barcode for this promising candidate for ornamental horticulture.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

WZ and CQL designed the research; SLK, XJS, LY, HLZ, and MQ performed the experiments; SLK, XJS, and LY contributed to data analysis; and CQL, WZ and SLK wrote the manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (grant nos. 82173936 and 32200187).

Data availability

Raw sequence data is available through the NCBI SRA under BioProject accession PRJNA1130731. The plastomes are deposited in the GenBank under accessions PP997971- PP997980.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.