Background

The genus Magnolia is one of the early diverged angiosperm lineages consisting of approximately 300 species across three subgenera: Gynopodium, Magnolia, and Yulania according to Figlar’s taxonomic system [1, 2]. The extensive changes in chromosome number and rare androdioecious flowers of subgenus Gynopodium make them important materials for studying the evolution and breeding of flowering plants, as they are thought to represent a key transition from bisexual flowers to unisexual flowers [3,4,5]. Furthermore, members of the subgenus Gynopodium are known for their beautiful flowers, leafy branches, and aesthetically appealing shapes having high ornamental, economic, and ecological value [6, 7]. However, over-harvesting coupled with weak regenerative capacity makes the wild populations of subgenus Gynopodium species decreased rapidly [8,9,10]. Six of the eight subgenus Gynopodium species are of conservation concern, including three critically endangered species, two endangered species, and one vulnerable species according to the IUCN Red List [11]. Despite lots of studies on phytocoenological characteristics and breeding of subgenus Gynopodium [8, 9, 12], investigations of the genomic characteristics of this subgenus remain lacking.

Compared with the nuclear genome, the chloroplast genome (CPG) has a small size, low nucleotide substitution rate, single-parental inheritance, and haploid nature, which make it a good option for the analyses of nucleotide diversity and reconstructing phylogenies of closely related species, especially among polyploid taxa [13,14,15]. Although the structure of the CPG is generally conserved consisting of a large single-copy (LSC) region, a small single-copy (SSC) region, and two inverted repeat regions (IR) [16], some structural rearrangements have been discovered, including the loss of genes or introns, as well as IR expansions and contractions [17, 18]. The comparative and phylogenetic analyses of CPGs have proved an ideal tool for species identification [19], detecting structural variation [20], assessing nucleotide diversity [21], resolving phylogenetic relationships [22], and reconstructing the evolutionary history [23]. Due to the similarity in the morphology of subgenus Gynopodium species and the complexity of their nuclear genomes associated with polyploidy [24], the CPG is suitable for exploring phylogenetic relationships, discriminating species and providing useful information for developing conservation strategies for this subgenus [25].

Here, we used the four newly sequenced CPGs of Magnolia omeiensis, Magnolia nitida, Magnolia sinica, and Magnolia kachirachirai, in addition to two previously published CPGs of Magnolia lotungensis and Magnolia yunnanensis, to (i) characterize the structural features and variations of the CPGs for the six sugenus Gynopodium species, (ii) assessing nucleotide diversity and identify hypervariable regions to developing DNA markers for species discrimination and conservation genetics studies, and (iii) resolve the evolutionary relationships of subgenus Gynopodium species.

Results

Characteristics of the CPGs

In this study, the coverage depth of each organelle genome reached over 100 × (Magnolia omeiensis: 168 × , M. sinica: 102 × , M. nitida: 132 × , M. kachirachirai: 103 ×). The six CPGs within the subgenus Gynopodium ranged in size, from 160,027 bp (M. kachirachirai) to 160,114 bp (M. lotungensis) (Table 1). All CPGs were a typical quadripartite circular structure (Fig. 1) that included a LSC region and a SSC region divided by a pair of IR regions (Fig. 1 and Table 1). The length of the LSC region ranged from 88,130 bp (M. kachirachirai) to 88,170 bp (M. yunnanensis), and the length of the SSC and IR regions ranged from 18,725 bp (M. kachirachirai) to 18,767 bp (M. lotungensis), and from 26,571 bp (M. sinica) to 26,586 bp (M. kachirachirai), respectively (Table 1). The GC-content was similar in all six CPGs. The GC content of the whole plasmid sequence was 39.3%; the GC content of the IR regions was 43.2%, which was higher than that of in LSC and SSC regions (38% and 34.3%) (Table S1). In addition, 131 genes were annotated in all six CPGs, including 37 transfer RNA (tRNA) genes, 8 ribosomal RNA (rRNA) genes, and 86 protein-coding genes (Fig. 1 and Table 1). There were two copies for seven of the protein-coding genes, seven of the tRNA genes, and four of the rRNA genes; the other 95 genes were all represented by single copies. Eleven genes possessed introns: rps16, rps12, ropC1, rpl2, rpl16, petB, petD, ndhB, ndhA, clpP1, and atpF (Table 2).

Table 1 Summary of the CPG features of subgenus Gynopodium
Full size table
Fig. 1
figure 1

Gene map of the CPGs of six subgenus Gynopodium species. The genes inside and outside of the circle are transcribed in the clockwise and counterclockwise directions, respectively. Genes belonging to different functional groups are shown in different colors. The darker gray area in the inner circle indicates the GC content and the lighter gray indicates the AT content of the genome. The thick lines indicate the extent of the inverted repeats (IRa and IRb) that separate the genomes into the small single-copy (SSC) and large single-copy (LSC) regions

Full size image
Table 2 Gene list in the CPGs of six subgenus Gynopodium species
Full size table

Comparative analysis of CPGs

The alignments indicated high sequence similarity among the CPGs of the six subgenus Gynopodium species. However, sequence divergence in non-coding regions was greater than that in coding regions, such as trnH-psbA, rps2-rpoC2, ycf4-cemA, petA-psbJ, and ccsA-ndhD (Fig. 2). The greatest variation among coding regions was observed in ycf1. No major genomic rearrangements or insertions were detected among the six CPGs relative to that of M. omeiensis (Fig. S1).

Fig. 2
figure 2

Sequence alignment of the CPGs of six subgenus Gynopodium species. The alignment was performed using the mVISTA program and the M. omeiensis chloroplast genome was used as a reference. The y-axis indicated the degree of identity ranging from 50 to 100%. Coding and non-coding regions were marked in blue and red, respectively. Black arrows indicated the position and direction of each gene. CNS: conserved non-coding sequences

Full size image

Expansions and contractions in the CPGs of six subgenus Gynopodium species were visualized using IRscope (Fig. 3). The gene rps19 and trnH were located in the LSC region 1 bp from the LSC/IRb border and 11 bp from the IRa/LSC border. The genes rpl2 and ndhF were located in the IRb and SSC regions, respectively, and differed slightly in their proximity to the border between the IRb and SSC regions. The gene ycf1 was located between 4,256 and 4,274 bp in the SSC region, and between 1,270 and 1,279 bp in the IRa region. In all CPGs, significant length variations were detected in the LSC and SSC regions; sequences length was more conserved in the IR regions than those in the LSC and SSC regions (Fig. 3 and Table 1).

Fig. 3
figure 3

Comparisons of the borders of the large single-copy (LSC), small single-copy (SSC), and inverted repeat (IR) regions among the CPGs of six subgenus Gynopodium species. Gaps between the ends of boundaries and adjacent genes were indicated in base pairs (bps) above the main line

Full size image

Large repeat sequences and Simple sequence repeats (SSRs) analyses

Large repeat sequences were identified using REPuter software [26]. A total of 300 repeats were identified. Palindromic repeats were the most common repeat sequences, and no complement repeat was found in the CPGs of six subgenus Gynopodium species (Fig. 4). Variation was observed in the number of palindromic repeats and reverse repeats among the six CPGs. The lowest number of palindromic repeats (19) was observed in M. sinica, followed by M. omeiensis (20), M. lotungensis (21). M. nitida (22), M. kachirachirai (22), and M. yunnanensis (22). The number of reverse repeats was less in M. nitida, M. kachirachirai, and M. yunnanensis (9) than in M. lotungensis (10), M. omeiensis (12), and M. sinica (13). Among these repeats, nine were over 30 bp and 24 were 20–29 bp; the longest repeat was 39 bp. Over half of the repeats (60%) were located in non-coding regions, and some of the repeats were located in the coding regions of genes, such as psaA, psaB, ndhC, ycf1, ycf2, rpoB, and rpoC2 (Table S2).

Fig. 4
figure 4

Comparison of the numbers of repeats among the CPGs of six subgenus Gynopodium species: Magnolia omeiensis, Magnolia sinica, Magnolia nitida, Magnolia kachirachirai, Magnolia lotungensis, and Magnolia yunnanensis. (F: Forward, P: Palindromic, and R: Reverse repeats)

Full size image

A total of 847 SSRs were identified in the CPGs of six subgenus Gynopodium species, ranging from 140 to 142 in each species, among which 117–119 were mononucleotides, 9 were dinucleotides, 3–4 were trinucleotides, 9 were tetranucleotides, and 2 were pentanucleotides (Fig. 5a and Table 3). There was no marked variation in the number of SSRs among the six species; however, slight differences were observed in the number of mononucleotides and trinucleotides. Over 80% of SSRs were mononucleotide repeats consisting of 112 A/T repeats and five C/G repeats. All the dinucleotides consisted of multiple copies of AT/TA repeats and AG/CT repeats (Fig. 5b). SSRs were mostly located in intergenic spacer regions (IGS) (69.29%), followed by coding regions (17.86%) and introns (12.86%) (Fig. S2, Table 3). The SSRs in the coding regions were located in 12 protein-coding genes (rpoC1, rpoC2, rpoB, psbC, cemA, rps3, rps19, ndhF, ndhD, ycf1, ycf2, and ycf4) (Table S3). Few SSRs were located in the IR regions (10–12 SSRs); most were located in the LSC region (104–106 SSRs), followed by the SSC region (24–25 SSRs; Table S3).

Fig. 5
figure 5

The number of microsatellite loci with different types of repeats (A) and repeat units (B) detected in the CPGs of six subgenus Gynopodium species

Full size image
Table 3 The number and location of simple sequence repeats (SSRs) in the subgenus Gynopodium species
Full size table

Identification of highly variable regions

The nucleotide diversity within a 600-bp window was calculated for all six CPGs, which ranged from 0 to 0.008 (Fig. 6). There were five highly variable regions with Pi values greater than 0.004, including the ycf1 gene and four intergenic regions (psbA-trnH-GUG, petA-psbJ, rpl32-trnL-UAG and ccsA-ndhD). Pi was greatest (0.007) for the intergenic region between ccsA and ndhD. Highly variable regions were located in the LSC region (2) and SSC region (3); no highly variable region was detected in the IR region (Fig. 6), which reflects similar patterns with structure variability of CPGs. In addition, we evaluated the potential utility of the five highly variable regions. The rpl32-trnL-UAG marker (π = 0.007) with the highest discriminatory power can discriminate six haplotypes from the six subgenus Gynopodium species (Table 4). The psbA-trnH-GUG marker (π = 0.006) with high haplotype diversity can discriminate five haplotypes. Similarly, the marker petA-psbJ (π = 0.005), ccsA-ndhD (π = 0.007), and ycf1 (π = 0.004) can discriminate three haplotypes from the six subgenus Gynopodium species (Table 4).

Fig. 6
figure 6

Sliding window test of nucleotide diversity (Pi) in the multiple alignments of six subgenus Gynopodium species (window length: 600 bp; step size: 200 bp). The X-axis indicates the position of the midpoint of the window; the Y-axis indicates the nucleotide diversity of each window

Full size image
Table 4 Nucleotide diversity and discriminatory power of the subgenus Gynopodium chloroplast markers
Full size table

Phylogenetic relationships

Phylogenetic relationships were reconstructed using both ML and BI approaches, based on the whole CPGs of 22 species covering all known sections within Magnoliaceae. Topologies of the ML and BI trees were concordant and confirmed that Magnoliaceae comprised two subfamilies (Liriodendroideae and Magnolioideae), each with one genus (Liriodendron and Magnolia). Within Magnolia, subgenus Gynopodium was sister to the subgenus Yulania (BS = 100%, PP = 1.00) (Fig. 7). However, due to the non‐monophyly of subgenus Magnolia, three previously established subgenera in Magnolia were not supported (Fig. 7). Subgenus Gynopodium should be treated as a section of genus Magnolia following Wang et al. (2021) [27]. Within Subgenus Gynopodium, M. sinica diverged first (PP = 1, BS = 100), followed by M. nitida, M. kachirachirai, and M. lotungensis (albeit with relatively low support values), and M. omeiensis was sister to M. yunnanensis (PP = 0.97, BS = 50) (Fig. 7, Fig. S3).

Fig. 7
figure 7

Phylogenetic relationship of Magnoliaceae based on the CPGs of 20 Magnolia species and two Liriondendron species. The phylogeny was inferred by Bayesian inference. Numbers above the lines indicate the posterior probabilities from the Bayesian inference

Full size image

Discussion

Characteristics of the CPGs

The CPGs of most angiosperms varied in size from 120 to 160 kb [16]. Our results indicated that the CPGs of six subgenus Gynopodium species are similar in size (ca. 160 kb) and structure (quadripartite circular structure) to other Magnolia species [28,29,30] as well as other higher plants [31]. The total number, order, and composition of genes in the CPGs were highly conserved within subgenus Gynopodium, which is also consistent with most Magnolia species [32, 33], suggesting a very conserved structure of CPGs of subgenus Gynopodium.

The overall GC content has been reported to be associated with the phylogenetic position; specifically, the GC content tends to be higher in early diverged lineages, such as magnoliids [34]. Our results are consistent with these previous findings. Of the six subgenus Gynopodium species, the overall GC content of CPGs was approximately 39.3%, which is similar to that of other Magnolia species, such as M. shiluensis [32], M. grandiflora [35], and M. zenii [36] but higher than the average GC content (35%) of most angiosperms [37]. The GC content also varies among different regions of the CPG [34, 38]. IR region (43.2%) contains significantly higher GC content than that of the LSC (38%) and SSC regions (34.3%) (Table S1), which can attribute to the high GC content in the ribosomal RNA (rRNA) genes in IR region (Fig. 1). Identical findings have been reported in other species, such as Magnolia polytepala [39], Magnolia delavayi [40] and Datura stramonium [41].

Conservatisms of the CPGs

We compared the CPGs of six species within the subgenus Gynopodium. The results indicated that the SSC and LSC regions were more divergent than IR regions, and sequences in non-coding regions were more divergent than that in coding regions, which were consistent with previous findings in Magnolia species [29] and other flowering plants [42, 43] In this study, we identified six regions presenting significant variations in the CPGs of subgenus Gynopodium species, such as five intergenic regions: trnH-psbA, rps2-rpoC2, ycf4-cemA, petA-psbJ, and ccsA-ndhD, and one gene ycf1 (Fig. 2). No major genomic rearrangements or insertions were detected among the six CPGs, which further corroborated the results of recently published studies about Magnoliaceae [27]. Previous studies also found that variation in the size of angiosperms CPGs might be largely driven by length variation in IR regions, intergenic regions, and the number of gene copies [44,45,46]. The structure of the six CPGs within subgenus Gynopodium species was highly conserved; no major expansions or contractions were observed in the IR regions. However, variations in sequence length have been observed in both the LSC and SSC regions, which may drive variations in the size of CPGs within the subgenus Gynopodium species, as reported in other species [29, 47, 48].

Large repeats and simple sequence repeats

Knowledge of genetic diversity within subgenus Gynopodium is necessary to develop sustainable conservation management that ensures long-term maintenance of the genetic diversity within these species [3, 49]. Repeat sequences, which are dispersed in CPGs, are an important source of structural variation and play a significant role in genomic evolution [16, 50]. In our study, 300 repeats were identified, of which palindromic repeats were the most common, while complement repeats were missing in CPGs of the subgenus Gynopodium. The different number of forward repeats, palindromic repeats and reverse repeats generated the variations of CPGs [41]. Therefore, genetic variation in large repeats can provide useful information for phylogenetic research and population genetics. Previous studies have indicated that repeat sequences are mostly located in the intergenic spacer regions, followed by the coding regions [14, 32]. Our findings are consistent with this general pattern; 61.22-65.31% of the repeats were located in IGS regions, followed by coding regions and introns (34.69-38.38%) (Table S2).

SSRs are useful molecular markers that have been widely used in species discrimination, breeding and conservation, and phylogenetic studies [51,52,53,54]. In the CPGs of six subgenus Gynopodium species, the number of SSRs located in the LSC and SSC regions accounted for 92.86% of all SSRs, and only ten SSRs were located in the IR region (Table S3). Our findings were consistent with the general pattern of angiosperm that most of the repeats were located in the LSC and SSC regions of CPGs [36, 48]. The SSRs of the CPGs of six subgenus Gynopodium species identified in our study provided valuable sources for developing primers of specific SSR loci and a useful tool for species identification.

Highly variable regions

Highly variable regions provide abundant phylogenetic information and can be used as potential molecular markers to delimit closely related taxa [55]. The Pi of highly variable regions within subgenus Gynopodium species was lower (< 0.008) compared with previously published values of other species [56, 57] and some of Magnolia species [29, 30]. The low genetic diversity of subgenus Gynopodium species and other Magnolia species, e.g., Magnolia ashei may relate to their limited habitat and small populations as threatened species [54, 58, 59].

In the Magnoliaceae, several highly variable regions, such as, matk, ycf1, psbA-trnH and atpB-rbcL have been recognized as potential sites for DNA barcoding [39, 60]. In this study, we recognized five highly variable regions with Pi values greater than 0.004, including one gene (ycf1) and four intergenic regions (psbA-trnH-GUG, petA-psbJ, rpl32-trnL-UAG and ccsA-ndhD). The highly variable regions identified here have high discriminatory power to distinguish 6 (rpl32-trnL-UAG), 5 (psbA-trnH-GUG), 3 (petA-psbJ), 3 (ccsA-ndhD), and 3 (ycf1) plastid haplotypes from six subgenus Gynopodium species (Table 4). These regions could be considered as potential barcoding markers for species identification of subgenus Gynopodium.

Phylogenetic relationship

CPGs have shown substantial power in solving phylogenetic relationships among angiosperms [61]. However, it is still controversial regarding the boundaries of the genera of Magnoliaceae [1, 6]. Based on the whole CPGs of 22 species covering all known sections of Magnoliaceae, topologies of the ML and BI trees all supported that Magnoliaceae consisted of two subfamilies Magnolioideae and Liriodendroideae, each with one genus, Magnolia and Liriodendron, respectively. However, due to the non‐monophyly of subgenus Magnolia, three previously established subgenera in Magnolia were not supported. Our results supported the infrageneric circumscriptions reported by Wang et al. that classified Magnolia into 15 clades corresponding to 15 sections and subgenus Gynopodium treated as a section of Magnolia [27, 62]. And our results also supported merging section Manglietiastrum into section Gynopodium as reported previously [62, 63].

Although we recovered the phylogenetic relationship within subgenus Gynopodium, some of the nodes were poorly supported (Fig. 7). The low nucleotide diversity and nucleotide substitution rate in the CPGs of subgenus Gynopodium species and other Magnolia species might contribute to the lack of phylogenetic resolution in Magnoliaceae [62, 64, 65]. Consequently, genetic markers from the mitochondrial and nuclear genomes should be developed to reconstruct more robust phylogenies of subgenus Gynopodium species.

Conclusions

We compared the complete CPGs of six subgenus Gynopodium species (four newly sequenced and two obtained from previous studies). All CPGs exhibited the typical quadripartite structure of most angiosperms. The number, composition, and order of genes in the CPGs of subgenus Gynopodium species were similar to those of other species in the Magnoliaceae. We detected neither major expansions or contractions in the IR region, nor rearrangements or insertions. We identified large repeats, SSRs, and highly variable regions within subgenus Gynopodium, getting knowledge of the extremely low genetic diversity in these species. The six highly variable regions identified here will be useful for species delimitation within the subgenus Gynopodium. Overall, our findings and genetic resources presented here will facilitate future studies of subgenus Gynopodium and aid in species discrimination and conservation strategy development for threatened species in this subgenus.

Materials and methods

Plant material, DNA extraction and sequencing

Leaf samples of M. omeiensis were collected from mature trees from wild populations on Emei Mountain (Sichuan, China). Leaf samples of M. nitida were collected from Nanjing Botanical Garden. Leaf samples of M. kachirachirai and M. sinica were collected from South China Botanical Garden. The plant materials were identified by Dr. Lei Zhang and the voucher specimens (collection numbers: LiuJQ-2019–123, LiuJQ-2019–168, LiuJQ-2019–050, and ZC-1906–7) were deposited in the herbarium of Sichuan University. The CPGs of 20 species spanning all sections within Magnoliaceae were obtained from the National Center of Bio-technology Information (NCBI, https://www.ncbi.nlm.nih.gov/). Liriodendron chinense and Liriodendron tulipifera were used as outgroups, and the two CPGs for these species were downloaded from NCBI (Table S4).

Total genomic DNA was extracted from silica gel‐dried leaves using a modified CTAB method [66] and treated with RNase (TransGen, China). The DNA samples were indexed by tags and pooled together in a single lane of a Genome Analyzer (Illumina HiSeq 2000) for sequencing at BGI-Shenzhen. Paired‐end reads (2 × 150 bp) were sequenced, and more than 4.0 Gb of reads were obtained for each sample.

Assembly and annotation

The raw Illumina reads were first filtered by removing paired-end reads that contained (i) adapter sequences, (ii) more than 10% N bases, and (iii) more than 50% of bases with a Phred quality score less than ten. The filtered reads were then assembled using NOVOPlasty version 4.0 [67] and the complete plastome sequence of Magnolia biondii Pamp. (KY085894) as a reference. These assemblies were manually inspected using Geneious Prime version 9.1.8 [68]. The genome was automatically annotated using Plann version 1.1 [69] based on the well-annotated plastome of M. insignis Wall. (KY921716). All annotated CPGs were submitted to GenBank (accession numbers: OL631157, OL631158, OL631159, and OL631160). The chloroplast genomes map was generated by OGDRAW version 1.2 [70].

Comparative analysis of the CPGs of subgenus Gynopodium species

The results of the comparative analysis of the CPGs of the six subgenus Gynopodium species were visualized using online mVISTA software [71] with the annotated CPG of M. omeiensis as the reference in Shuffle-LAGAN mode. Detection of structural variation was conducted using Mauve software [72] with M. omeiensis as the reference. The borders of the four different regions among the six CPGs were visualized using IRscope [73].

Repeat structure and highly variable regions analysis

The online software REPuter [26] was used to identify repeat sequences (forward, reverse, complement, and palindromic) in CPGs with default parameters. Simple sequence repeats were examined using MISA-web [74] with minimal repeat numbers of 8, 5, 4, 3, 3, and 3 for mono-, di-, tri-, tetra-, penta-, and hexa-nucleotide repeats, respectively. To identify highly variable regions, polymorphic sites and nucleotide diversity (Pi) in the six MAFFT-aligned CPGs were assessed using a sliding window analysis in DNAsp v6.12.03, with a 200-bp step size and a 600-bp window length [75]. Regions in the CPGs with numbers of polymorphic sites greater than the sum of the average and double the standard deviation were considered highly variable regions [76]. Then we estimated the number of haplotypes, haplotype diversity, parsimony informative sites, and singleton sites to detect the discriminatory power of highly variable regions using DnaSP v6.12.03 [75].

Phylogenetic analysis

Phylogenies were reconstructed using maximum likelihood (ML) and Bayesian inference (BI) analyses with the complete CPGs of 20 Magnolia species and two Liriodendron species (Table S4). ML analysis was conducted in RAxML [77] using the GTRGAMMA model and 1000 bootstrap (BS) replicates. BI analysis was conducted in Mrbayes v 3.2.6 [78], with four independent Markov chain Monte Carlo analysis runs for 1,000,000 generations each. PartitionFinder was used to determine the optimal partitioning scheme [79]. Priors were set to default values, and trees were sampled every 1,000 generations, with the first 25% discarded as burn-in. The consensus tree was calculated from trees sampled after reaching likelihood convergence, and the posterior probabilities (PPs) of the tree nodes were calculated.