Comparative analysis of complete chloroplast genomes reveals the phylogenetic relationships of Lygodium Sw. (Lygodiaceae) species in China

Abstract

Background

The family Lygodiaceae occupies a key position in leptosporangiate fern phylogeny. However, its intrafamilial relationships and chloroplast (cp.) genome evolution remain insufficiently explored. The lack of comprehensive cp. genomic data from Chinese species, which represent significant diversity in the family, hinders the development of robust molecular markers and precise phylogenetic resolution.

Results

The complete cp. genomes of seven Chinese Lygodium species were assembled and characterized. Comparative analyses across these and one published genome (L. merrillii) showed highly conserved structure and gene order, but total length varied (157,001-164,439 bp). This size variation was primarily driven by large-fragment insertions in intergenic spacers of L. microphyllum and L. longifolium. All genomes shared a common preference for A/U-ending codons and contained 31–65 simple sequence repeats (SSRs) and 31–91 tandem/dispersed repeats. Critically, nucleotide diversity analysis identified nine highly variable regions, including four coding (ycf1, ndhF, matK, and rpoC2) and five non-coding hotspots (e.g., rpl32-trnP-GGG, rpoB-trnC-GCA, trnI-CAU-trnR-ACG), which are proposed as promising novel molecular markers for species delimitation. Phylogenomic reconstruction based on cp. genomes strongly supported the monophyly of Lygodiaceae and clarified interspecific relationships, thereby consolidating previous molecular findings from a genomic perspective.

Conclusions

This study provides a comprehensive cp. genomic resource for major Chinese Lygodiaceae species. The identified hypervariable regions offer powerful tools for future phylogenetic and population studies. Furthermore, the resolved phylogeny enhances understanding of the family’s evolutionary history and provides a solid framework for its global systematics.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-026-12561-w.

Introduction

Chloroplasts are plant organelles in which photosynthesis takes place, harbour a double-stranded genome and their own genetic system that operates independently of the nuclear genome [1, 2]. In recent years, the chloroplast (cp.) genome has become a highly valuable system for comparative genomic studies, owing to its typically uniparental inheritance, compact structure, lack of recombination, and moderate evolutionary rate compared to nuclear and mitochondrial genomes [3, 4]. Additionally, because the cp. genome is relatively easy to sequence, assemble, and annotate, with high efficiency and low cost, it is now widely used in plant phylogenetics, evolutionary studies, species identification, and taxonomy [5, 6]. For example, comparative cp. genomic analyses have been applied to investigate evolutionary questions in East Asian lobelias (Lobelioideae), including the identification of closely related species, gene structural variations and adaptive characteristics [7, 8]. Similarly, cp. genome data have contributed to resolving phylogenetic relationships, revealing cryptic species in China, and enabling DNA barcoding-based identification of Cibotium barometz (L.) J. Sm [7, 8]. These findings thus provide important foundations for taxonomic classification and resource management of these genera [7, 8].

As the second largest group of vascular plants, ferns play significant ecological roles and are crucial for phylogenetic studies [9]. Research on fern cp. genomes has significantly advanced our understanding of plastomic diversity and evolution in land plants. Fern cp. genomes are highly conserved in size and structure, typically forming circular DNA molecules ranging from 120 to 170 kb and consisting of four regions: a pair of inverted repeats (IRs), a large single-copy (LSC) region, and a small single-copy (SSC) region [10]. However, certain structural variations exist, including uneven expansion and contraction of IR regions in core leptosporangiate ferns (Salviniales, Cyatheales, and Polypodiales) and Schizaeales [11–13]. To date, studies on fern cp. genomes have primarily focused on the phylogenetics of Polypodiaceae, Cyatheaceae and Pteridaceae [14–16]. In contrast, research at the family or genus level remains notably limited.

Lygodiaceae M. Roem. is a monogenetic family comprising 26 species worldwide, which are naturally distributed mainly in tropical regions and extend into subtropical and temperate zones [17]. Within China, nine species have been identified, including two endemics [17]. Members of this family are characterized by hairy and non-scaly rhizomes, climbing fronds by means of a twining rachis, often free veins, rarely sparsely reticulate, fertile pinnules bearing dual parallel rows of sporangia forming a fimbriated spike that exerts beyond the pinnule margin, sporangia located on the apex of small veins, and tetrahedral spores [17]. The family’s evolutionary history traces back to the Jurassic period, as evidenced by fossils of Lygodium japonicum (Thunb.) Sw [18]. Beyond their paleontological significance, many species within the genus hold substantial medicinal value in traditional Chinese medicine. Four species (L. japonicum, L. microphyllum (Cav.) R. Br., L. flexuosum (L.) Sw., and L. circinnatum (Burm. f.) Sw.) have been used in herbal remedies [17], with contemporary pharmacological studies confirming their broad-spectrum bioactivities, including antibacterial, antiviral, choleretic, diuretic, and hepatoprotective properties [19]. These species are particularly valued for treating urinary tract calculi and renal disorders. However, accurate species identification remains challenging due to pronounced morphological similarities within the genus Lygodium Sw., leading to frequent misidentification that affects the quality of medicinal materials and clinical safety. This taxonomic ambiguity is reflected in the nomenclature, where, for instance, only 31 of over 120 names listed in World Flora Online are scientifically accepted, underscoring the need for systematic verification. Furthermore, understanding of the genetic polymorphisms within this family remains insufficient, and a robust molecular phylogenetic framework based on genome-scale data has not yet been established, limiting our understanding of their genetic diversity and evolutionary history. Given these challenges, investigating the plastomic features of Lygodiaceae is crucial. Such research would not only establish scientific criteria for authenticating medicinal resources but also provide novel phylogenetic insights from polymorphism analysis of chloroplast markers, thereby resolving longstanding taxonomic ambiguities and supporting both conservation and pharmaceutical applications.

To achieve these objectives, we employed high-throughput sequencing to assemble the cp. genomes of Lygodiaceae species and conducted a comparative analysis that included the published cp. genome of Lygodium merrillii Copel. This study aimed to: (1) characterize the genome structure and gene content of Lygodiaceae cp. genomes; (2) examine fine-scale structural variations; (3) identify highly variable loci as potential molecular markers; and (4) clarify the phylogenetic position of Lygodiaceae within ferns.

Results

Chloroplast genome features

The cp. genomes of the eight Lygodiaceae species were visualized using OGDraw (Fig. 1). Their complete sequence lengths ranged from 157,001 bp to 164,439 bp, displaying the typical quadripartite circular structure common in land plants, which comprises a large single-copy (LSC) region (85,409 bp to 87,949 bp), a small single-copy (SSC) region (21,592 bp to 21,680 bp), and two inverted repeat (IR) regions (24,981 bp to 28,159 bp) (Table 1). Notably, Lygodium longifolium (Willd.) Sw. presented the largest genome, which was significantly larger than those of the other seven species, with an LSC region of 87,949 bp. L. microphyllum possessed the second largest genome and the longest IR region (28,159 bp). The overall GC content of the cp. genomes ranged from 40.6% to 41.0%. The IR regions exhibited the highest GC content (45.1% to 45.7%), followed by the LSC (from 38.8% to 39.3%) and SSC regions (36.4% to 36.7%).

Fig. 1.

Chloroplast genome map of eight species of Lygodiaceae. The different colours represent genes of different functional groups. The genes drawn inside are transcribed clockwise, whereas the outside genes are transcribed counterclockwise

Table 1.

Chloroplast genomic characteristics of Lygodiaceae plants

Species	Genome size (bp)				GC content (%)				Genes
	Total	LSC	SSC	IR	Total	LSC	SSC	IR	Total	CDS	tRNA	rRNA
Lygodium japonicum	157,001	85,409	21,630	24,981	40.6	39	36.4	45.2	117	86	27	4
L. microphyllum	163,399	85,474	21,607	28,159	40.9	38.8	36.4	45.7	117	86	27	4
L. flexuosum	162,110	85,462	21,680	27,484	40.8	39	36.5	45.3	117	86	27	4
L. circinnatum	158,643	85,859	21,620	25,582	40.7	39.1	36.5	45.1	117	86	27	4
L. longifolium	164,439	87,949	21,592	27,449	41.0	39.3	36.5	45.4	117	86	27	4
L. salicifolium	162,126	85,494	21,614	27,509	40.8	39	36.5	45.2	117	86	27	4
L. polystachyum	160,902	85,570	21,598	26,867	41.0	39.2	36.7	45.5	117	86	27	4
L. merrillii	161,997	85,520	21,571	27,453	40.9	39.2	36.5	45.3	117	86	27	4

A total of 131 genes were annotated in the eight cp genomes, comprising 86 protein-coding genes (PCGs), four rRNA genes, and 27 tRNA genes. These genes were systematically classified into three functional categories: self-replication, photosynthesis, and other genes. Among these, eight PCGs (atpF, ndhA, petB, petD, rpl16, rpl2, rps16, and ycf66) and five tRNA genes (trnA-UGC, trnG-UCC, trnI-GAU, trnL-CAA, and trnV-UAC) contained one intron, while three genes (rps12, ycf3, and clpP) contained two introns. Additionally, the rps12 gene underwent trans-splicing, with its 5’ end located in the LSC region and the 3’ end in the IR region (Table 2). It is also notable that the rpoC1 gene entirely lacked introns in all eight species, corresponding to the absence of approximately 650 bp of intronic sequence.

Table 2.

The chloroplast genome composition of Lygodiaceae plants

Category	Groups of gene	Name of genes
Self-replication	Large subunit of ribosome	rpl2*, rpl14, rpl16*, rpl20, rpl21, rpl22, rpl23, rpl32, rpl33, rpl36
	Small subunit of ribosome	rps2, rps3, rps4, rps7a×2, rps8, rps11, rps12a, b, **×2, rps14, rps15, rps16*, rps18, rps19a
	RNA polymerase subunits	rpoA, rpoB, rpoC1, rpoC2
	Ribosome RNA	rrn4.5 a ×2, rrn5 a ×2, rrn16 a ×2, rrn23 a ×2
	Transfer RNA	trnA-UGCa, *×2, trnC-GCA, trnD-GUC, trnE-UUC, trnF-GAA, trnfM-CAU, trnG-UCC*, trnG-UCC*, trnH-GUGa×2, trnI-CAU, trnI-GAUa, *×2, trnL-CAA*, trnL-UAG, trnM-CAU, trnN-GUUa×2, trnP-GGG, trnP-UGG, trnQ-UUG, trnR-ACGa×2, trnR-UCU, trnS-GCU, trnS-GGA, trnS-UGA, trnT-GGU, trnV-UAC*, trnW-CCA, trnY-GUA
Photosynthesis	Subunits of photosystem Ⅰ	psaA, psaB, psaC, psaI, psaJ, psaM
	Subunits of photosystem Ⅱ	psbA a ×2, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN, psbT, psbZ
	Subunits of NADH-dehydrogenase	ndhA*, ndhBa,*×2, ndhC, ndhD, ndhE, ndhFa,  ndhG, ndhH, ndhI, ndhJ, ndhK
	Subunits of cytochrome b/f complex	petA, petB*, petD*, petG, petL, petN
	Subunits of ATP synthase	atpA, atpB, atpE, atpF*, atpH, atpI
	Subunit of rubisco	rbcL
Other genes	Translational initiation factor	infA
	Maturase	matK
	Protease	clpP**
	Envelope membrane protein	cemA
	Subunit of acetyl-CoA- carboxylase	accD
	C-type of cytochrome synthesis gene	ccsA
	Light-independent protochlorophyllide reductase subunit	chlB, chlL, chlN
Functionally unknown genes	Conserved open reading frames	ycf1, ycf2 a ×2, ycf3**, ycf4, ycf12, ycf66

^aGene in the inverted repeat region, ^bTrans-splinting gene, * genes containing one intron, ** genes containing two introns, ×2 genes with duplicates

Similarity analysis of Chloroplast genomes

The sequence conservation among the eight cp. genomes was assessed using mVISTA, with L. japonicum as the reference (Fig. 2). Multiple alignments showed that all eight cp. genomes were highly conserved and syntenic. Sequence conservation was higher in the single-copy regions than in the inverted repeat (IR) regions, and in protein-coding regions compared to the non-coding regions. However, several non-coding regions were highly variable, including trnfM-CAU-rps14, trnI-CAU-trnR-ACG, rrn16-rps12, and trnR-ACG-ndhB. Additionally, a large insertion fragment of 6,191 bp was identified in the rrn16-rps12 intergenic spacer (IGS) within the IR region of L. microphyllum, which was significantly longer than the corresponding spacers in other Lygodiaceae species (2026 ~ 2038 bp) (Fig. 3). To confirm this finding, we aligned our assembly with the L. microphyllum cp. genome sequence retrieved from the NCBI database under both accessions: the standard RefSeq record (NC_039378) and its source submission (MG761729). The alignment revealed a 4,508 bp discrepancy, mainly due to a large deletion in the IGS between rrn16 and rps12 in the IRb region (Fig. S1). Specifically, these NCBI sequences showed an asymmetric IR structure, with a 6,164 bp spacer in IRa but only 2,038 bp in IRb. In contrast, our assembly contained consistent IRs, each with a 6,191 bp spacer. Therefore, our assembly was used for subsequent cross-species comparative analyses. Similarly, a large fragment of 2,513 bp was identified between the trnfM-CAU and rps14 genes in the LSC region of L. longifolium, which was also considerably longer than the corresponding regions in other species (162 ~ 305 bp) (Fig. 3).

Fig. 2.

Alignment of the SC/IR boundaries of the chloroplast genomes of eight species of Lygodiaceae. Gray arrows and thick black lines above the alignment denote gene orientation. Exons, tRNAs/rRNAs, and non-coding sequences are coloured dark blue, light blue, and red, respectively. The Y-axis indicates sequence similarity (50–100%)

Fig. 3.

Variability in the rrn16-rps12 (a) and trnfM-CAU-rps14 (b) intergenic spacers of the plastid genomes from Lygodiaceae species. The sizes are indicated in parentheses following the scientific names. Insertions containing long non-coding sequences are shown in black, and all other plastid DNA sequences in gray. Sequence deletions are shown as black lines

To investigate structural variations, including potential gene deletions, duplications, rearrangements, or translocations, we performed a structural comparison of the eight cp. genomes using Mauve software. The results revealed a single locally collinear block (LCB) across all eight species, indicating highly conserved genome structures with no rearrangements or recombinations detected (Fig. 4). This demonstrates complete collinearity among the eight cp. genomes, further confirming their high degree of structural conservation.

Fig. 4.

Mauve genome alignment of the whole chloroplast genomes. Boxes of the same colour connected with lines indicate syntenic regions. The histogram within each block shows the sequence identity profile

To identify mutational hotspots within the cp. genomes of Lygodiaceae, a sliding window analysis was performed on the aligned complete genome sequences using DnaSP v6.0. The resulting nucleotide diversity value (Pi) across the eight cp. genomes ranged from 0.00286 to 0.11762 (average: 0.03164), with variable regions distributed relatively evenly (Fig. 5a). The five most polymorphic segments—rrn16-rps12 in the IR region; trnfM-CAU-rps14, rpoB-trnC-GCA, and trnI-CAU-trnR-ACG in the LSC region; and rpl32-trnP-GGG in the SSC region—are indicated in Fig. 5a. Overall, the SSC region exhibited the highest sequence variation (average Pi = 0.09329), followed by the LSC (Pi = 0.03827) and IR (Pi = 0.01240) regions.

Fig. 5.

Nucleotide diversity (Pi) and hypervariable region analyses of the chloroplast genomes in eight Lygodiaceae species. a Nucleotide diversity (Pi) analysis of the Lygodiaceae cp. genomes. b Nucleotide diversity (Pi) analysis of the coding sequence. The X-axis indicates the nucleotide position, and the Y-axis shows the Pi value (window length: 600 bp; step size: 200 bp). The five regions with the highest Pi values are labelled

Nucleotide polymorphism analysis of the protein-coding sequences alone revealed lower Pi values, ranging from 0.00286 to 0.07802 (average: 0.02545) (Fig. 5b). A total of five highly divergent loci were identified, including rrn16-rps12 in the IR region and four protein-coding genes (matK and rpoC2 in the LSC region; ndhF and ycf1 in the SSC region). Among the protein-coding genes, ycf1 was the most differentiated. These hypervariable regions exhibited strong potential as DNA barcodes for species identification within Lygodiaceae.

Inverted repeat/single-copy boundary analysis

To visualize the four junction sites of the cp. genome, we used CPJSdraw, which provides an intuitive representation of size variations at the boundaries between the IRs, LSC, and SSC regions. The boundaries in Lygodiaceae were found to be relatively conserved, with the IR, LSC, SSC, and junction regions overall exhibiting high consistency and conservation, only minor interspecific differences.

In all eight species, the genes near the IR/LSC and IR/SSC boundaries included trnR-ACG, ndhB, trnI-CAU, trnN-GUU, ndhF, and chlL (Fig. 6). Specifically, the trnN-GUU gene was located within the IR region adjacent to either the ndhF or the chlL at the SSC/IR border. The ndhF gene crossed the SSC/IRb boundary, extending 33–34 bp into the IRb region, while the chlL gene was located near the this boundary. At the other end, the IRa/LSC border was located within the ndhB coding region, resulting in a 762 bp pseudogene at the corresponding LSC/IRb border in all eight genomes. Moreover, although the trnR-ACG gene was located in the IRa region close to the LSC/IRa boundary in all eight genomes, its distance to the boundary was markedly shorter in L. japonicum (3684 bp), L. microphyllum (2723 bp), and L. circinnatum (4261 bp) than in the other five species (5560 ~ 6196 bp). Thus, despite strong overall conservation, the SC/IR boundaries are not identical, suggesting that the IRs have undergone independent and repeated expansions and contractions throughout evolution.

Fig. 6.

Comparison of chloroplast genome boundary regions of eight Lygodiaceae species. Genes transcribed on the positive strand are represented from right to left above the path line, whereas genes on the negative strand are represented from left to right below it. The numbers on the arrows indicate the distance from the gene to the SC/IR boundary

SSR and repeat analysis

In this study, mono-, di-, tri-, tetra-, and hexanucleotide SSRs were identified in the eight species’ cp. genomes. A total of 31 to 65 SSRs were detected across the eight Lygodiaceae species (Fig. 7a). Monomeric SSRs were predominant, accounting for 50% to 73.85% of the total, followed by tetranucleotide SSRs (7.69% to 15%). Notably, hexanucleotide SSRs were unique to L. circinnatum, and no pentanucleotide SSRs were found in any species. Among mononucleotide SSRs, the A/T motif was the most abundant, comprising between 32.50% (L. japonicum) and 63.08% (L. merrillii) of all SSRs (Fig. 7b).

Fig. 7.

Comparative analysis of repeated sequences in the eight cp. genomes. a Number of SSRs. b Frequency of identified SSR motifs in different repeat class types. c Number of dispersed repeats. d Number of tandem repeats

Sequence repeats in the cp. genomes include both tandem and dispersed repeats. Dispersed repeats, analyzed using REPuter, were classified as palindromic (P), forward (F), reverse (R), or complementary (C). The F (9–26) and P (12–27) types were the most common. A small number of R repeats were detected only in L. circinnatum (3), L. flexuosum (1), and L. longifolium (1). Complementary repeats were absent in five species and present only in L. longifolium (5), L. circinnatum (3), and L. polystachyum Wall. (1) (Fig. 7c). Additionally, 8 to 41 tandem repeats were identified (Fig. 7d), located primarily in the LSC and IR regions. L. microphyllum contained the most tandem repeats, followed by L. circinnatum, while L. japonicum had the fewest. Specifically, the tandem repeats in L. microphyllum were concentrated in the IR region, whereas those in L. circinnatum were mainly in the LSC region.

Codon usage analysis

Analysis of codon usage across the eight Lygodiaceae species identified a total of 61 synonymous codons. Among these, 30 were high-frequency codons, defined by relative synonymous codon usage (RSCU) values greater than 1.0. Notably, all eight cp. genomes exhibited similar codon preferences. For example, leucine (Leu) showed the strongest preference, with UUA having the highest average RSCU value (1.74) across all chloroplast coding sequences (CDS). This was followed by alanine (Ala), which displayed a pronounced preference for GCU (average RSCU = 1.65). Similarly, arginine (Arg), glycine (Gly), and aspartic acid (Asp) strongly favoured the codons AGA, GGA, and GAU, respectively, each with an average RSCU exceeding 1.50 (Fig. 8a). Additionally, most codons with RSCU values above 1.0 ended in A or U, with UUG—a leucine codon ending in G—being the sole exception. In contrast, the majority of codons with RSCU values below 1.0 ended in C or G. This nucleotide-ending bias is consistent with previous studies on cp. genomes [20, 21].

Fig. 8.

RSCU and optimal codon heatmap of the cp. genomes of eight Lygodiaceae species. a RSCU heatmap. Darker colours denote higher RSCU values for codons. Codons with RSCU > 1 are highlighted in red. b Optimal codon heatmap. Optimal codons are highlighted in red

Furthermore, optimal codons in the eight cp. genomes were determined based on the RSCU values. The number of optimal codons per genome ranged from 13 to 17 (Fig. 8b). A greater abundance of optimal codons suggests a stronger codon usage bias (CUB) in these genomes. Overall, the sets of optimal codons were highly consistent across the eight species, with only minor variations. GAU, GGU, UUA, AUU, and GAA were identified as optimal codons common to all eight Lygodiaceae species, indicating a shared preference for these codons. Interestingly, GUU was shared as an optimal codon by seven species, with L. longifolium being the exception. In contrast, two optimal codons (UUU and GCA) were unique to L. longifolium, while two others (GUA and CAC) were specific to L. microphyllum and L. circinnatum, respectively.

Phylogenetic analysis

The phylogenetic relationships among the eight Lygodiaceae species and 20 other published fern species (Table S2) were investigated using maximum likelihood (ML) and Bayesian inference (BI) methods. The taxon sampling covered the major clades at the order level according to the Pteridophyte Phylogeny Group I (PPGI) system. The analysis was based on six datasets: the complete cp. genome, 69 coding sequences (CDS), and four highly variable regions (ndhF, rpoC2, ycf1, and matK) (Fig. 9, Figs. S2-S10). The resulting phylogenetic topologies were largely congruent across all six datasets, though bootstrap support values of some branches varied. Among these, the tree based on the complete cp. genomes (32 genomes representing 27 species) received the highest support, with most branches having 100% bootstrap support (BS) and a posterior probability (pp) of 1, the lowest node support was 98%. This indicates a stable topology in which all fern lineages formed monophyletic groups.

Fig. 9.

Molecular phylogenetic analysis of the chloroplast genomes of Lygodiaceae. a Maximum likelihood (ML) phylogeny inferred from the complete cp. genomes of all 27 ferns. Numbers above branches are ML bootstrap score (BS), those below are Bayesian posterior probabilities (pp). Newly sequenced species are indicated by asterisks. b Enlargement of the region marked by the red box in (a). Haploid chromosome numbers are indicated at the nodes, based on the Lygodiaceae species tree and hypothesized reticulation events from Pelosi [22]

The monophyly of Lygodiaceae was strongly supported in all analyses (BS = 100%; pp = 1). Lygodiaceae showed a relatively close genetic relationship with Schizaeaceae Kaulf. and Anemiaceae Link., together forming the order Schizaeales. This order occupies an intermediate phylogenetic position between Gleicheniales and the core leptosporangiate ferns (Cyatheales and Polypodiales). This suggests that Schizaeales represents a transitional evolutionary stage from the “basal” to the “core” fern lineages. Within Lygodiaceae, Lygodium microphyllum was the first to diverge, followed by a clade comprising L. circinnatum, L. japonicum, and L. polystachyum, while L. flexuosum and L. salicifolium C. Presl formed another clade, suggesting a closer genetic relationship between them—a finding consistent with their morphological similarities. Consistent with the findings of Pelosi et al. [22], we also recovered L. flexuosum and L. salicifolium as a sister group to L. longifolium, with this combined clade being sister to L. merrillii.

Discussion

Structural characteristics of the Chloroplast genomes of Lygodiaceae plants

In this study, the complete cp. genomes of seven Lygodiaceae species were assembled and annotated, and a comparative genomic analysis across eight species was performed. The results revealed that the eight cp. genomes of Lygodiaceae were highly conserved in structure, gene order, and content, showing completely collinear without any rearrangement or recombination.

We further analyzed the cp. genomes of 27 fern species. Notably, the IR regions in Schizaeaceae are significantly expanded due to the acquisition of additional genes from the LSC and SSC regions. In two Schizaeaceae species, Schizaea elegans (Vahl) Sw. and Schizaea pectinate (L.) Sw., the SSC region is reduced to only three genes (chlN, chlI, and trnN-GUU) [23]. In contrast, across the rest of Schizaeales (i.e., Anemiaceae and Lygodiaceae), the gene order and content of the IR regions are highly similar to those in core leptosporangiate ferns.

Additionally, the genomic locations of the ndhB gene across these 27 fern species were found to be variable, which fell into three distinct structural types (including one exon 1 plus one exon 2, one exon 1 plus two exon 2, and two exon 1 plus two exon 2), consistent with a previous report [24]. Specifically, in Lygodiaceae, the ndhB gene has been displaced from the IR region to a position near the LSC/IR boundary. The relocation of other genes, such as ycf2 from the LSC into the IRs, has contributed to a near-doubling of IR length and gene content compared to those in basal ferns (e.g., Equisetum hyemale L., Ophiopterales, Osmundales, and Gleicheniales), a characteristic feature of core leptosporangiate ferns.

Moreover, an intron loss in the rpoC1 gene was detected in all eight Lygodiaceae species. Such gene or intron losses are common in fern evolution. Some of these losses represent synapomorphies for high-level taxonomic groups (at or above the family or order level). For example, the rps16 gene and the rps12 intron are absent in Equisetales, Ophioglossales, and Psilotales [25, 26]; trnS-CGA, trnT-UGU, and trnK-UUU are lost in leptosporangiate ferns except for Osmundales; trnV-GAC is lost in core leptosporangiate ferns; and the psaM gene has been lost in Polypodiales [26]. Whereas other losses appear more randomly distributed or convergent, likely representing autapomorphies of terminal lineages (below family or genus level), such as the loss of chlB, chlL, and chlN genes in Psilotaceae [25, 27].

Sequence variation analysis

Our analysis showed that the four boundary regions were highly conserved across the eight Lygodiaceae cp. genomes. However, the total length of these genomes varied considerably, ranging from 157,001 to 164,439 bp. In land plants, cp. genome size variation is typically attributed to expansion or contraction of the IR boundary regions [28]. However, the minor IR boundary shifts observed here cannot explain the substantial length differences in Lygodiaceae.

Instead, a fine-scale comparison of cp. genomes revealed that their size variation is primarily associated with large noncoding insertions in the IR or LSC regions. For example, the IGS between rrn16 and rps12 in L. microphyllum (6,191 bp) and the IGS between trnfM-CAU and rps14 in L. longifolium (2,513 bp) are substantially longer than those in other Lygodiaceae species. Thus, in these two species, the large species-specific fragment insertions—rather than fluctuations in the SC/IR boundaries—are responsible for the cp. genome size expansion. Critically, these insertions also lead to relatively high local nucleotide diversity, with a maximum Pi value of 0.11762. Although this value may appear unusually high, similar or even higher Pi values have been reported in Polypodiaceae species (e.g., averaging 0.10103, with maxima exceeding 0.2 in the SSC region) and are often attributed to large-fragment insertions [14]. Our data support a similar mechanistic explanation for the elevated Pi observed in Lygodiaceae.

A similar phenomenon has been reported in Adiantum malesianum Ghatak, in which the rpoB-trnD-GUC IGS (6876 bp) is considerably longer than that in other Adiantum species (669 ~ 945 bp), resulting in an enlarged cp. genome [29]. However, a key difference is that this elongation in A. malesianum also led to sequential relocation of LSC genes, causing trnI-CAU to be “squeezed” into the IRb region, effectively replacing part of trnT-UGU without expanding the IR length. Such large IGSs appear common in fern cp. genomes. Robison et al. [30] systematically studied the plastid genome evolution in ferns and identified a novel group of mobile insertion sequences, termed MORFFO (Mobile Open Reading Frames in Fern Organelles). These movable insertion sequences (morffo1/2/3) are primarily distributed in the intergenic spacer regions of rrn16-rps12 and rpoB-trnD, and exhibit considerable diversity in their insertion positions, lengths, arrangements, and orientations. In particular, the rrnl6-rps12 spacer, located in the IR region, is among the most frequent regions for long noncoding sequence insertions [30]. Therefore, these mobile sequences are believed to play a key role in the structural variation and evolution of fern cp. genomes [30].

BLAST searches suggest that some of these long insertions may be of mitochondrial origin, possibly transferred to the cp. genome through horizontal gene transfer [31]. For example, certain unusually long IGSs (e.g., rps12-rrn16) align with sequences from the mitochondrial genome of Haplopteris ensiformis (Sw.) E. H. Crane (Pteridaceae), indicating possible interorganellar gene transfer. However, due to the limited availability of mitochondrial and nuclear genomic data for ferns, it remains challenging to identify highly homologous sequences for these insertions. Therefore, their origins, distribution patterns, insertion mechanisms, and evolutionary implications require further investigation [30].

Subsequently, nine divergence hotspots between Lygodiaceae species were identified, including five noncoding regions (rrn16-rps12, trnfM-CAU-rps14, rpoB-trnC-GCA, trnI-CAU-trnR-ACG, and rpl32-trnP-GGG) and four protein-coding genes (matK and rpoC2 in the LSC region; ndhF and ycf1 in the SSC region). Although the rrn16-rps12 region showed a relatively high Pi value (> 0.1), the presence of a large species-specific insertion in the L. microphyllum cp. genome means it may not be a reliable, universally amplifiable hotspot for Lygodiaceae. This underscores the importance of meticulous data exploration when selecting mutational hotspots to avoid bias from structural variations. Among coding regions, matK is highly variable in ferns and is widely used for phylogenetic reconstruction [32]. The ycf1 gene, a large open reading frame often regarded as an indispensable pseudogene with no specific function [33], contains a total of 723 variant sites and is the most frequently mutated gene in this study, supporting its recent proposal as a promising molecular marker [34]. These hypervariable regions, rich in phylogenetic information, can serve as effective molecular markers for species identification, population genetics, and phylogeny within Lygodiaceae. Furthermore, when combined with nuclear gene markers and phylogenetic analyses of parental lineages, they hold promise for investigating hybrid origins and polyploid species formation in the family.

Repeat sequence analysis

Repetitive sequences are key factors influencing genetic diversity and can promote the rearrangement of cp. genes, thereby playing an important role in species diversification. SSRs and other repetitive sequences are widespread in cp. genomes. They are highly polymorphic, rich in genetic information, easily amplified, and suitable for applications such as genetic diversity assessment, kinship analysis, and molecular-assisted breeding [35]. This study identified numerous repeat regions, including forward, reverse, palindromic, and tandem repeats, which may represent important hotspots for genomic reconfiguration [36]. Notably, L. microphyllum exhibited a greater number of dispersed and tandem repeats compared to other species, likely due to the unusually long IGSs in its cp. genome [16].

Moreover, the abundance of repetitive sequences in cp. genomes can enhance population-level genetic diversity and is associated with multiple phylogenetic signals [37]. These repeats offer a wealth of informative sites that can be used to develop molecular markers for phylogenetic and population genetic studies [38]. Such markers are crucial for species identification within Lygodiaceae, population polymorphism screening, and phylogeny reconstruction. However, the functional significance of repetitive sequences in the cp. genomes of Lygodiaceae remains to be fully elucidated.

Codon bias analysis

Codon usage bias is ubiquitous across organisms. Closely related species, or those inhabiting similar environments, often share similar codon usage strategies. Previous studies have indicated that variations in codon usage patterns can provide important insights into species classification and molecular evolution, making the analysis of codon bias an essential component of cp. genomic research [39]. Codon usage patterns have been analyzed in numerous plant groups, and it is established that natural selection significantly influences these patterns. Furthermore, phylogenetic trees based on relative synonymous codon usage (RSCU) often yield topologies consistent with molecular phylogenetic trees [40, 41].

Our analysis indicated that all eight examined species share similar codon preferences in their cp. genomes, which aligns with findings from other cp. genomic studies [20, 21]. Additionally, 27 genes across the eight Lygodiaceae sequences were identified to contain alternative start codons, stop codons, or both. Specifically, 12 genes had an alternative start codon, 16 genes contained an alternative stop codon, and the ycf2 gene had both an alternative start codon and an internal stop codon. This pattern is a common feature in the cp. genomes of leptosporangiate ferns [42].

Phylogenetic analysis

Phylogenetic analysis confirmed that Lygodiaceae is a strongly supported monophyletic group, consistent with the findings of previous studies [22, 43, 44]. The family shows a very close genetic relationship with Schizaeaceae and Anemiaceae, and together they form a highly supported clade, corresponding to the order Schizaeales. Within Lygodiaceae, the phylogenetic topology is consistent with the molecular trees constructed from short sequences such as rbcL alone or rbcL plus trnL-F [43, 44], as well as with the tree reconstructed by Pelosi et al.. based on multilocus plastid and nuclear gene data obtained by targeted capture sequencing [22].

However, notable discrepancies exist between molecular-based phylogenies and the morphological phylogeny proposed by Hanks, which relied on plant morphological traits and chemical characteristics, and the subgenera or sections defined in her classification are not well supported in the molecular trees [45]. For example, species with palmate-lobed pinnae (e.g., L. longifolium and L. circinnatum) did not form a clade, nor did those with pinnate-lobed pinnae (including L. japonicum and L. flexuosum). Similarly, L. merrillii, characterized by reticulate venation, did not cluster separately from species with bifurcated venation (Fig. 10). Given that reticulate evolutionary events are very common in ferns [46], these inconsistencies between molecular and morphological classifications may be explained by polyploidy and hybridization events, as revealed by Pelosi et al. [22]. The molecular phylogeny showed stronger congruence with chromosome numbers (Fig. 9b). For example, L. microphyllum—located at the base of the phylogenetic tree—has a chromosome number of x = 30; species positioned toward the top of the tree, including L. flexuosum, L. salicifolium, and L. longifolium, exhibit x = 28; while the remaining four species have x = 29 [22].

Fig. 10.

Diversity of the morphological characteristics of Lygodium plants. A Fertile pinna of Lygodium japonicum. B Fertile pinnae of L. microphyllum. C Fertile pinna of L. flexuosum. D Sterile pinnae of L. salicifolium. E-F Fertile pinnae of L. polystachyum. G Fertile pinna of L. circinnatum. H Sterile pinnae of L. longifolium. I L. merrillii. J Bifurcated venation of L. circinnatum. K Bifurcated venation of L. longifolium. L Reticulate venation of L. merrillii. (A-C), (G), (J-L), by J.T.; (D), (H-I), by Z.F.; (E-F), by J.L

Although there is a discrepancy between morphological and molecular findings, the current molecular phylogenetic tree supports the following inferences: (1) The basal position of Lygodium microphyllum among Chinese taxa is strongly supported by both our findings and that of Pelosi et al., confirming its status as an early-diverging lineage within the genus. (2) Within the genus Lygodium, the patterns of bifurcated venation appear to hold greater taxonomic and evolutionary significance than frond-lobing patterns, particularly among Chinese species with palmate pinnules. Morphologically, L. circinnatum, L. longifolium, and L. merrillii might be expected to form a monophyletic group based on their shared feature of “palmate-lobed pinnules” (Fig. 10). However, molecular results from both our study and that of Pelosi et al. [22] strongly suggest a closer relationship among L. longifolium, L. merrillii, L. flexuosum, and L. salicifolium, while L. circinnatum and L. japonicum are more distantly related. Venation analysis reveals that L. longifolium, L. merrillii, L. flexuosum, and L. salicifolium share a common pattern of “short primary vein bifurcation with multiple secondary vein bifurcations”: the primary veins bifurcate near their origin, and the secondary veins undergo repeated successive bifurcations. In contrast, L. circinnatum exhibits “variable primary bifurcation distances and infrequent unilateral secondary bifurcations”, indicating markedly different vein development compared to L. longifolium and L. merrillii, despite their superficial similarity in frond lobbing (Fig. 10). Although L. japonicum and L. polystachyum display terminal pinnule venation similar to that of L. longifolium and L. merrillii, their pinnately dissected (rather than simple) fronds result in widely spaced primary veins along the rachis and non-anastomosing secondary veins—characteristics that distinguish them from both the aforementioned species and L. circinnatum. Thus, the close genetic relationship among L. longifolium, L. merrillii, L. flexuosum, and L. salicifolium is supported by venation morphology, while the divergent venation patterns of L. japonicum, L. polystachyum and L. circinnatum support their exclusion from this clade. Further morphological and developmental studies are warranted to elucidate the overlooked systematic importance of bifurcated venation in the genus Lygodium. (3) Our results identify L. salicifolium and L. flexuosum as sister species, which conflicts with Pelosi et al.‘s topology in which L. salicifolium is nested within L. flexuosum. This discrepancy may arise from insufficient sampling of these two species or from potential polyploidization events in L. flexuosum, which could lead to ploidy-level variation among populations and affect phylogenetic reconstruction. Despite their high morphological similarity and genetic proximity (or nesting), resolving the species boundaries between L. salicifolium and L. flexuosum will require expanded sampling, cytological analyses, and population genetic studies.

This study, based on the chloroplast genomes of eight Chinese Lygodiaceae species along with previous research, provides preliminary insights into the phylogenetic relationships within the family and explores the genetic affinities among Chinese species. However, due to limited sampling, the prevalence of reticulate evolution (resulting from hybridization, polyploidization, and horizontal gene transfer) in ferns, and a lack of robust single-copy nuclear gene data, this work has inherent limitations in constructing a robust genus-wide phylogeny, clarifying species boundaries, and resolving interspecific relationships within Lygodiaceae. Future studies should integrate chromosomal karyotype analysis, single-copy nuclear gene datasets, and more comprehensive morphological and taxonomic assessments, and should combine the trait evolution with phylogenetic relationships to establish a reliable phylogenetic framework and infer well-supported interspecific affinities. For morphologically similar taxa that exhibit strong cytonuclear discordance in phylogenetic trees, population-genetic and phylogeographic approaches should be incorporated. Such an integrative taxonomic framework will help to accurately delineate species boundaries within the complex groups of the genus Lygodium.

Conclusions

The complete cp. genomes of eight Lygodiaceae species native to China were analyzed, including the first assembly and annotation of Lygodium salicifolium, L. longifolium, and L. polystachyum. A detailed comparative genomic analysis of these eight species revealed both conserved structural features and divergent characteristics within Lygodium, highlighting the evolutionary information inherent in their cp. genomes. Notably, large inserted fragments were identified in the rrn16-rps12 spacer of the IR regions in L. microphyllum and in the trnfM-CAU-rps14 spacer of the LSC region in L. longifolium, which were found to be the primary cause of cp. genome expansion in these species. These IGS regions showed a positive correlation with repetitive sequences, especially dispersed repeats—a pattern also reported in Pteridaceae. Moreover, the intron of the rpoC1 gene (approximately 650 bp in length) was completely absent in all eight Lygodiaceae species and may represent an autapomorphy for the family. Four highly variable regions (matK, rpoC2, ndhF, and ycf1) with high mutation rates and suitable length were identified as potential DNA barcodes for species identification within Lygodiaceae. Finally, phylogenetic analysis strongly supported the monophyly of Lygodiaceae and further clarified the molecular taxonomic status of related lineages, underscoring the evolutionary significance harbored within the cp. genomes of this genus. Based on these findings, future studies on the classification and identification of global Lygodiaceae species should combine genomic and integrative taxonomic approaches to clarify the taxonomic positions of other complex lineages. Additionally, integrating chloroplast and nuclear genomic data could help identify homologous sequences of these large insertions and elucidate their origin, distribution, insertion mechanisms, and evolutionary history.

Materials and methods

Sampling, DNA extraction, and genome sequencing

In this study, eight species of Lygodiaceae were collected from various locations across southern China (Supplemental Table 1). We were unable to obtain a sample of Lygodium yunnanense Ching, a species morphologically similar to L. flexuosum (L.) Sw. and endemic to China, where its distribution is largely restricted to limited areas in southwestern Yunnan Province. Fresh fronds were surface-cleaned to remove impurities and contaminants. After excess water was removed, the samples were completely desiccated in silica gel for long-term of preservation [47]. The corresponding voucher specimens for all collected samples were deposited in the Herbarium of Guangzhou University of Chinese Medicine.

The cp. genomes of the eight species were extracted using a modified CTAB method [48]. DNA quality was evaluated by electrophoresis on 1% agarose gels, and DNA concentration was measured with a microplate reader. The qualified genomic DNA was sheared into fragments with an average size of 200–300 bp using a Covaris ultrasonicator. Paired-end (PE) libraries were then constructed through terminal repair, A-tailing, adapter ligation, purification, and PCR amplification. Library size distribution and concentration were assessed with an Agilent 2100 Bioanalyzer (Agilent DNA 1000 Reagents) for preliminary quantification. Second-generation sequencing was performed on the DNBSEQ platform (BGI Technology Co., Ltd., Shenzhen, China), generating approximately 3 GB of clean data per species.

Assembly and annotation of the Chloroplast genome

De novo assembly of the cp. genomes from the clean data was performed using GetOrganelle (version 1.7.3.5) [49]. Assembly accuracy was assessed with Bandage software by examining coverage and sequencing depth [50]. Initial annotation of the cp. genome was carried out with PGA software [51], using the sequence of L. japonicum (NC_022136) as a reference. Subsequent manual corrections of problematic annotations were made in Geneious (version 9.0.2) [52]. The annotated sequences were submitted to the National Center for Biotechnology Information (NCBI) and assigned the following GenBank accession numbers: PQ580665, PQ591737, PQ591738, PQ591739, PQ591740, PQ591741, and PQ591742. Finally, the cp. genome map was visualized with OGDraw [53].

Comparative analysis of the chloroplast genome

Owing to the thick and protruding leaf veins of L. merrillii, we were unable to extract the DNA of sufficient quality for sequencing; therefore, the chloroplast genome sequence (NC_084182) downloaded from the NCBI GenBank database was used for subsequent comparative analysis. To characterize the cp. genome of the eight Lygodiaceae species, we employed Geneious software [52]. Multiple genome alignments of the complete cp. genomes were conducted using MAFFT to assess possible structural rearrangements [54]. Whole-genome alignments were also performed using the ProgressiveMauve algorithm in Mauve to visualize the synteny, gene rearrangements, and inversions [55]. Additionally, the online tool mVISTA (http://genome.lbl.gov/vista/index.shtml) was used in Shuffle-LAGAN mode to compare and visualize structural differences among the cp. genomes [56]. The boundaries of the IR, SSC, and LSC regions were determined and plotted using the CPJSdraw boundary-map tool [57].

To estimate nucleotide diversity (Pi) and identify mutational hotspots across the Lygodiaceae species, the eight cp. genomes were aligned pairwise with MAFFT [54]. Nucleotide polymorphism across the aligned sequences was assessed by sliding-window analysis using DnaSP v6.0 [58], with a window length of 600 bp and a step size of 200 bp. Potential hypervariable regions were evaluated based on their length, number of variable sites (S), and nucleotide diversity (Pi). Several fragments with relatively high Pi values were identified as hypervariable regions within the cp. genomes of Lygodiaceae, and their genomic positions were mapped according to the gene annotations.

Codon usage analysis

Codons, which serve as the link between DNA and proteins, play an essential role in transmitting genetic information [59]. Codon usage bias (CUB) refers to the nonrandom use of synonymous codons during amino acid encoding in protein synthesis [60, 61]. Relative synonymous codon usage (RSCU) quantifies the deviation of a codon’s observed frequency from its expected frequency under random synonymous codon usage [62]. The RSCU value is interpreted by comparison with 1: an RSCU > 1 indicates that the codon is used more frequently than other synonymous codons [21]. The effective number of codons (ENC) reflects the degree of codon bias: higher ENC values indicate weaker bias, whereas lower values indicate stronger bias. Highly expressed genes typically exhibit lower ENC values. Only coding sequences (CDS) longer than or equal to 300 bp were included in the subsequent analysis, and they were required to start with ATG as the initiator codon and end with TAA, TAG, or TGA as the terminator codon. CodonW software was used to calculate the RSCU and ENC values for the 56 protein-coding genes in each species. The filtered CDS were first ranked by their ENC values [63]. Subsequently, the top 10% (low ENC) and bottom 10% (high ENC) of genes were selected as high‑expression and low‑expression groups, respectively. Their corresponding RSCU values were calculated using an online tool (http://cloud.genepioneer.com:9929/#/tool/alltool/detail/291). Codons meeting the criteria of RSCU > 1 and ∆RSCU > 0.08 were defined as optimal codons [63].

Statistical analysis of repeated sequences

Simple sequence repeats (SSRs) were identified using MISA [64]. The minimum repeat thresholds were set as follows: 10 repeats for mononucleotide motifs, 5 for dinucleotides, 4 for trinucleotides, 3 for tetranucleotides, 3 for pentanucleotides, and 3 for hexanucleotides. Tandem repeats in the eight cp. genomes were detected with Tandem Repeats Finder (TRF) [65], with parameters set to match = 2, mismatch = 7, and indel = 7. The minimum alignment score was set to 90 and the maximum period size to 500. Dispersed repeats (including forward, reverse, complementary, and palindromic repeats) were identified using the REPuter online tool [66] (https://bibiserv.cebitec.uni-bielefeld.de/reputer), with the analysis parameters were set to a minimum repeat size of ≥ 30 bp and a Hamming distance of 3.

Phylogenetic analysis

To investigate the phylogenetic relationships among the eight studied species of Lygodiaceae, phylogenetic trees were constructed using maximum likelihood (ML) and Bayesian inference (BI) methods. The genome sequences were initially aligned using MAFFT (v7.490) under default settings [67]. Then, 69 protein-coding sequences (CDS) were extracted from the aligned genomes using Phylosuite [68]. Subsequently, the CDS alignments were concatenated into a supermatrix using the pxcat program from the Phyx package [69]. ML analysis was performed with IQ-TREE (v 1.6.0) [70] using 1000 bootstrap replicates; branches with bootstrap (BS) values ≥ 70% were considered well supported [71]. BI analysis was carried out in MrBayes (v3.2.7) [72], the best-fit substitution model was selected with ModelFinder based on the Bayesian information criterion (BIC) scores [73]. The MrBayes analysis was run for 2 million generations, sampling every 100 generations [72]. Convergence was assessed using the average standard deviation of split frequencies (ASDF), with a threshold of ASDF < 0.01. Branches with Bayesian posterior probabilities (pp) ≥ 0.95 were considered statistically supported. Both phylogenetic trees were visualized using the ChiPlot online tool (https://www.chiplot.online/normalTree.html).

Authors’ contributions

Conceptualization, J.T.; Resources, W.L., J.T. and J.L.; Investigation, Formal analysis, W.L. and J.L.; Visualization, W.L. and Z.F.; Original draft preparation, W.L.; Review & Editing, J.T. and Y.T.; Funding Acquisition, H.H. and R.Z. All the authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Traditional Chinese Medicine Standardization Project (Xiaoke Pill Standardization Construction) (ZYBZH-C-GD-07) and Agricultural Product-Oriented Innovation Team Construction Project for Guangdong Modern Agricultural Industry Technology System (Nan-Yao Industry Technology System) (2024CXTD24).

Data availability

The complete cp. genome was deposited in the NCBI GenBank under the accession numbers PQ580665, PQ591737, PQ591738, PQ591739, PQ591740, PQ591741 and PQ591742.

Declarations

Ethics approval and consent to participate

The collection of the plant materials complied with all relevant institutional, national, and international guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.