DNA barcodes analyses provide insights into species delineation and possible cryptic species in Amentotaxus (Taxaceae)

Abstract

Background

Accurate species identification and species delineation is crucial for biodiversity conservation and biological research. DNA barcoding is an efficient tool for species identification and delineation, especially when using next-generation DNA barcodes. Amentotaxus is a small genus of coniferous plants, with all six recognized species currently assessed as threatened. However, species delineation within Amentotaxus has not been thoroughly evaluated using DNA barcodes.

Results

In this study, we investigated the species discrimination capabilities and delineation using multiple samples per species, employing standard, super-, and next-generation DNA barcodes and their combinations alongside Skmer analysis with genome skimming sequencing data. Our results revealed that the nrDNA cistron exhibited the highest species discriminatory power, successfully resolving five out of the six species. This was followed by ITS, the next-generation barcode (plastome + nrDNA), super-barcode (plastome), and Skmer analysis, each identifying four species. The combinations of the three plastid standard barcodes (rbcL + matK + trnH-psbA) and all four standard DNA barcodes (rbcL + matK + trnH-psbA + ITS) successfully identified only two and three species, respectively. We propose plastid gene spacer clpP-rps12 as an Amentotaxus-specific DNA barcode since it alone delineated four species. Additionally, our results suggested that A. poilanei has recently diverged from A. hekouensis as a budding speciation event. The discordance in the placement of sample AM61 in the nuclear and plastid phylogenetic trees suggests a hybrid origin for this sample, which resulted in the failure to discriminate A. assamica using plastome data. Additionally, the results also imply the potential existence of an undescribed cryptic Amentotaxus species in north Myanmar.

Conclusion

Our findings enhance our current understanding of species diversity within Amentotaxus and provide critical insights for conservation efforts aimed at preventing the extinction of the undescribed potentially cryptic species in the wild.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08719-z.

Introduction

Accurate species identification and species delineation is crucial for biodiversity conservation, ecosystem management, and biological research [1–3]. Traditional morphological-based species delineation and identification methods often encounter difficulties, particularly when distinguishing morphologically similar but genetically distinct species [4]. In response to these challenges, DNA barcoding has emerged as a powerful alternative identification tool [5, 6]. The combination of two plastid DNA regions (rbcL and matK) and the internal transcribed spacer (ITS) of nuclear ribosomal DNA (nrDNA) have been recommended as core barcodes for land plants [4, 7], with the plastid region trnH-psbA as supplementary barcode [8]. However, the standard DNA barcodes have demonstrated limited efficacy in distinguishing closely related species due to a lack of genetic variation [9–12].

In contrast to standard DNA barcodes, super-barcodes (the complete plastome) and next-generation DNA barcodes, which include the complete plastome and the nuclear ribosomal DNA (nrDNA) cistron (including18S-ITS1-5.8S-ITS2-26S), can be rapidly and accurately assembled from genome skimming data generated by cost-effective next-generation sequencing (NGS) technologies [13–15]. This approach overcomes the inherent limitations of standard barcodes by providing more variable and informative sites, thereby showing great promise in reliably distinguishing closely related species [2, 15–17]. The plastome is typically inherited uniparentally in most plants [18, 19], with conifers such as Pinaceae, Cephalotaxaceae, Taxaceae being notable exceptions due to their predominant paternal inheritance [20]. While the plastome is widely recognized for its utility in phylogeny and species delimitation, it captures only one parental lineage.

Despite the advantages of genome skimming, current applications often require plastid genome assembly, which discards a significant proportion of nuclear genomic information and consequently reduces species discriminatory power [21–23]. Moreover, this process frequently necessitates time-consuming manual curation to minimize assembly errors [24]. To address these issues, an assembly-free and alignment-free approach known as “Skmer” has been proposed to calculate genomic distances using genome skims [25]. Although recent studies on Skmer for species identification remain limited, it has already demonstrated significant potential as a promising, reliable, and convenient approach [3, 26, 27].

Amentotaxus Pilg. is a small genus within the family Taxaceae, mainly distributed across southern China and parts of Indo-China, extending westward into the eastern Himalayas [28, 29]. The genus currently comprises six recognized species and one natural hybrid, A. × hybridia L.M. Gao & G.L. Zhang [30]. The species of Amentotaxus are evergreen shrubs or small trees, including A. assamica D.K. Ferguson, A. argotaenia (Hance) Pilg., A. formosana H.L. Li, A. hekouensis L.M. Gao, A. poilanei D.K. Ferguson, and A. yunnanensis H.L. Li. This genus has a complex and controversial taxonomic history, primarily due to significant morphological similarities and overlaps among species and substantial intraspecific variation. These factors give rise to taxonomic challenges, particularly the frequent occurrence of intermediate morphotypes in key diagnostic traits such as stomatal band width and leaf shape. Traditional identification of Amentotaxus species has largely depended on geographic origin and a limited set of morphological traits, including leaf shape and stomatal characteristics [28, 31–33]. However, geographic indicators have limitations due to overlapping species distributions and morphological variability, which further complicates species identification [34–37]. Moreover, recent studies have revealed inconsistencies in diagnostic features, such as stomatal band color, which can be influenced by leaf age and drying methods, calling into question their taxonomic reliability [38]. Accurate species identification is particularly important in Amentotaxus, as all recognized species are currently assessed as threatened at either national or global level due to habitat loss and deforestation [29, 39]. In order to devise effective conservation and sustainable management of the extant Amentotaxus species, accurate species delineation is fundamental.

DNA barcoding has demonstrated considerable promise and unique advantages for species delimitation of Amentotaxus [29, 34, 38]. For instance, A. hatuyenensis T.H. Nguyen was initially described based on specimen from a single locality in Hagiang Province, Vietnam [32], but was later treated as a synonym of A. yunnanensis based on morphological and molecular evidence [34]. Additionally, A. × hybridia was confirmed as a natural hybrid between sympatric A. yunnanensis and A. hekouensis based on morphological and DNA barcoding data [30]. However, previous studies did not include samples of A. poilanei and A. assamica. To date, no research has comprehensively evaluated the potential of multiple barcoding methods, including standard and next-generation DNA barcodes, as well as Skmer across the entire genus, nor has the development of taxon-specific DNA barcodes for this genus been fully explored.

In this study, we utilized genome skimming data to assemble the next-generation barcodes, complete plastid genome and the nrDNA cistron, and extracted standard DNA barcodes for all six recognized Amentotaxus species to assess their species discriminatory power and species delineation based on tree- and distance-based approaches. Additionally, we conducted Skmer analyseis with genome skimming data to further evaluate its species delineation power. The specific objectives of this study were: (1) to investigate species identification rates within Amentotaxus using next-generation barcodes in comparison to standard DNA barcodes; (2) to screen promising specific DNA barcode candidates for distinguishing Amentotaxus species, while testing Skmer using genome skimming data for Amentotaxus species identification; and (3) to assess the validity of the current species delimitation and discrimination within Amentotaxus. The findings will provide new and important baseline data for the development of conservation plans and management strategies for the threatened species of Amentotaxus.

Materials and methods

Taxon sampling, DNA extraction and NGS sequencing

To ensure comprehensive representation of diversity within Amentotaxus species, 3 to 5 individuals per species were collected from different populations covering the core distribution range of the genus (Table 1 and Fig. S1). A total of 23 individuals representing all six recognized Amentotaxus species were sampled. Torreya jackii and T. nucifera were selected as outgroups for phylogenetic analyses. All plant materials were collected with appropriate permissions and in accordance with institutional, local, and national regulations governing the collection of plant samples. Fresh leaves were collected and dried immediately in silica-gel, and voucher specimens were collected and deposited at our research group, the Herbarium of Kunming Institute of Botany, Chinese Academy of Sciences (KUN) and/or the Herbarium of the Royal Botanic Garden Edinburgh (E). Formal taxonomic identification of the samples was performed by Dr. Lian-Ming Gao.

Table 1.

Samples of Amentotaxus and Torreya (as outgroup) included in the present study with information on voucher, sample ID, locality, sequencing data, plastome size, GC content, mean coverage, number of genes, CDS, rRNA, tRNA, and GenBank (Sequence Read Archive, SRA) accession numbers for plastome and nrDNA cistron assemblies, and data sources

Taxon	Voucher	Sample ID	Locality information	Sequencing yield (Gb)	Plastome size (bp)	GC content	Mean coverage (×)	No. of genes	CDS	rRNA	tRNA	GenBank (SRA) Accession number of plastome	GenBank (SRA) Accession number of nrDNA cistron	Source
A. argotaenia	83	AM04	China, Sichuan, Dujiangyan	4.218	137,564	35.8%	579.7	122	84	4	34	PX703144	PX651885	This study
A. argotaenia	166	AM06	China, Chongqing, Nanchuan	2.332	137,267	35.8%	136.8	122	84	4	34	PX703146	PX651886	This study
A. argotaenia	182	AM05	China, Chongqing, Nanchuan	3.109	137,571	35.8%	230.9	122	84	4	34	PX703145	PX651904	This study
A. argotaenia	zhdq-019	AM03	China, Guizhou, Chishui	4.852	137,379	35.8%	640.3	122	84	4	34	PX703143	PX651884	This study
A. assamica	4773	AM08	China, Xizang, Motuo	6.753	137,678	35.8%	115.4	122	84	4	34	PX703148	PX651889	This study
A. assamica	2595	AM07	China, Xizang, Motuo	3.190	137,678	35.8%	101.2	122	84	4	34	PX703147	PX651887	This study
A. assamica	M1502	AM61	Myanmar, Kachin, Putao	4.954	137,500	35.8%	354.2	122	84	4	34	PX703163	PX651888	This study
A. formosana	MMO-081376	AM09	China, Taiwan, PingTung	1.810	137,347	35.8%	317.7	122	84	4	34	PX703149	PX651893	This study
A. formosana	MMO-081377	AM10	China, Taiwan, PingTung	3.662	136,361	35.8%	500	121	83	4	34	OK138557	SRR25021671	Wang et al. 2022 [3]
A. formosana	MMO-081378	AM11	China, Taiwan, PingTung	1.706	137,345	35.8%	111	122	84	4	34	PX703150	PX651894	This study
A. hekouensis	CKF250	AM24	Vietnam, Lao Cai, Muong Khuong	2.401	137,743	35.8%	237	122	84	4	34	PX703152	PX651898	This study
A. hekouensis	GLM-123943	AM25	China, Yunnan, Hekou	2.088	137,939	35.8%	193.6	122	84	4	34	PX703153	PX651897	This study
A. hekouensis	GLM-164271	AM50	China, Yunnan, Hekou	3.336	138,005	35.8%	176.5	122	84	4	34	PX703157	PX651896	This study
A. hekouensis	GX-5–4	AM15	Laos, Xiang Khouang, Phak Leung Mt	1.893	138,156	35.8%	156.4	122	84	4	34	PX703151	PX651895	This study
A. poilanei	KKK4006	AM53	Vietnam, Gia Lai, Mang Yang	2.560	137,841	35.8%	340.4	122	84	4	34	PX703158	PX651899	This study
A. poilanei	KKK4016A	AM54	Vietnam, Gia Lai, Mang Yang	2.580	137,809	35.8%	163.3	122	84	4	34	PX703159	PX651900	This study
A. poilanei	KKK4022	AM56	Vietnam, Gia Lai, Mang Yang	2.800	137,809	35.8%	281.7	122	84	4	34	PX703160	PX651902	This study
A. poilanei	KKK4030	AM59	Vietnam, Gia Lai, Mang Yang	2.795	137,841	35.8%	244.5	122	84	4	34	PX703161	PX651903	This study
A. poilanei	KKK4031	AM60	Vietnam, Gia Lai, Mang Yang	3.034	137,842	35.8%	370.7	122	84	4	34	PX703162	PX651901	This study
A. yunnanensis	CKF 043	AM49	Vietnam, Ha Giang, Quan Ba Commmune	4.540	137,574	35.9%	372.9	122	84	4	34	PX703156	PX651892	This study
A. yunnanensis	GBOWS079	AM21	China, Yunnan, Malipo	1.928	137,609	35.8%	146.2	122	84	4	34	OR197346	OR195109	Mo et al. 2023 [1]
A. yunnanensis	GLM-123952	AM26	China, Yunnan, Xichou	3.084	137,631	35.8%	273.8	122	84	4	34	PX703154	PX651890	This study
A. yunnanensis	P 11052	AM29	Vietnam, Thanh Hoa, Quan Hoa	3.739	137,554	35.8%	117.5	122	84	4	34	PX703155	PX651891	This study
Torreya jackii	GLM-07317	To07	China: Yunnan, Kunming	2.767	136,798	35.5%	433.6	121	83	4	34	OR197350	OR195113	Mo et al. 2023 [1]
Torreya nucifera	19,940,647	To36	USA: Atlanta, Atlanta Botanical Garden	5.911	136,944	35.5%	1144.5	121	83	4	34	OR197358	OR195121	Mo et al. 2023 [1]

Total genomic DNA was extracted from silica-gel dried leaves using a modified CTAB method [40]. The DNA was sheared by ultrasonication to construct an Illumina library for each individual with a mean insert size of 500 bp using a TruSeq DNA Sample Prep Kit following the manufacturer’s instructions (NEBNext® Ultra IITMDNA Library Prep Kit for Illumina®, USA). Each library was sequenced to generate approximately 2 Gb of sequencing data per sample on an Illumina HiSeq X Ten platform (Illumina, San Diego, CA, USA) with 150 bp paired-end reads at BGI Wuhan, China.

Plastome and nrDNA assembly and annotation

Adapters and low-quality reads were filtered from the raw sequencing data using Fastp v0.21.0 [41] with default parameters. The plastomes and nrDNA cistrons were assembled from the clean data using the GetOrganelle toolkit [42] with the plastome (GenBank accession: OR197346) and nrDNA cistron (OR195109) of Amentotaxus yunnanensis as references, respectively. The assembly graph was visualize using Bandage v0.8.1 [43], and the nrDNA cistrons and complete plastome sequences which were collinear with the reference were extracted for subsequent analyses. Plastid genes were annotated using the PGA tool [44] and were manually adjusted in Geneious Prime v2024.0.5 [45]. Transfer RNAs (tRNAs) were confirmed using tRNAscan-SE v2.0.3 [46]. The nrDNA cistrons were also annotated and manually adjusted in Geneious. The final circle plastome map was drawn using OGDRAW [47].

A sliding window analysis was conducted using DnaSP v6.12.03 [48] with a step size of 200 bp and a window length of 600 bp across the newly assembled 23 plastomes of all Amentotaxus species. Nucleotide diversity (Pi) was calculated to identify differences in plastome sequences among species with DnaSP. Based on Pi values, ten hypervariable loci with the highest nucleotide diversity were retained for subsequent analyses of species discriminatory rates. Besides, comparative analyses of the plastid genomes among the Amentotaxus species were performed using mVISTA in Shuffle-LAGAN mode [49] to visualize overall sequence conservation and to identify highly variable regions.

Distance-based identification analyses

The standard DNA barcodes, the nuclear ITS region (including ITS1-5.8S-ITS2) and the plastid barcodes rbcL, matK and trnH–psbA, and the ten hypervariable plastome loci identified based on nucleotide diversity using DnaSP, were extracted using Geneious, and combinations of multiple loci were concatenated using AMAS [50], resulting in three datasets (ITS; rbcL + matK + trnH–psbA; rbcL + matK + trnH–psbA + ITS). All extracted barcodes (including the ten hypervariable loci), the nrDNA cistron, and the complete plastome were aligned using MAFFT v7.407 [51] and manually adjusted in Geneious where necessary.

Summary statistics of all single or combinations of multiple barcodes, hypervariable loci, nrDNA cistron and plastome — including taxon occupancy, alignment length, and parsimony-informative sites (PI) for the Amentotaxus species — were calculated using AMAS. For each retained single barcode, multiple-barcode dataset, hypervariable loci, nrDNA cistron, and plastome, pairwise distances were estimated using the Kimura 2-parameter distance for the distance-based species discrimination in MEGA v11.0.13 [52]. The efficiency of species discrimination was evaluated by assessing the DNA barcoding gap, which involved comparing the maximum intraspecific distance with the minimum interspecific distance for each locus.

Tree-based identification analyses

Maximum likelihood (ML) phylogenetic trees were constructed for all datasets under the GTR + Γ model with 1000 rapid bootstrap replicates as implemented in RAxML v8.2.7 [53]. Following the methodology of Wang et al. [2], taxa with multiple samples forming taxon-clade with bootstrap support (BS) ≥ 50% were considered successfully discriminated.

Skmer analysis

An assembly-free method (Skmer) for estimating genomic distances was used for the species identification in Amentotaxus. Unassembled clean reads of all Amentotaxus samples were analyzed using Skmer v3.2.1 [25]. Before Skmer analysis, the clean reads were mapped to reference plastomes of A. argotaenia AM05, A. formosana AM09 and A. yunnanensis AM49, as well as mitochondrial genome of Taxus cuspidata (MN593023) using Bowtie2 v2.3.4 [54] to remove organelle-related reads. The remaining reads were then used as input for Skmer analysis. Following the recommended workflow [25], we obtained the distance matrix for the sequencing reads. Subsequently, FastME v2.1.6.4 [55] was utilized to infer the backbone tree, and RAxML was employed to generate the phylogenetic trees with support values using default parameters (https://github.com/shahab-sarmashghi/Skmer).

Results

General features of Amentotaxus plastomes

High-quality complete circular plastomes were assembled for all 23 Amentotaxus individuals with average sequencing depths ranging from 101 to 640 × (Fig. 1, Table 1). The newly generated plastomes had a length of 137,267 to 138,156 bp with similar GC contents (35.8–35.9%). Identical to the published plastomes of Amentotaxus species (A. formosana NC_024945 and OK138557, A. argotaenia NC_027581), the newly assembled plastomes lacked an inverted repeat region. The Amentotaxus plastomes consisted of 118 unique genes, comprising 83 protein-coding genes, four rRNA genes and 31 tRNA genes. Three tRNA genes, trnI-CAU, trnN-GUU and trnQ-UUG, were present in two copies across all Amentotaxus individuals, while the protein-coding gene infA had two copies in all individuals except accession A. formosana AM10. The orange graph in the inner circle indicates the nucleotide diversity among the Amentotaxus plastomes, as calculated by sliding window analysis, showing that sequence variation was generally low across the plastomes, with slightly higher diversity observed in several intergenic spacer regions (Fig. 1). Consistently, the mVISTA comparison further revealed a high level of sequence conservation among Amentotaxus species, with variation mainly occurring in non-coding regions (Fig. S2).

Fig. 1.

Circular map of Amentotaxus plastomes. The colored boxes outside and inside the circle represent the direction of transcription of genes, counterclockwise and clockwise, respectively. The orange graph in the inner circle illustrates the nucleotide diversity across the 23 plastomes calculated by a sliding window analysis (window length: 600 bp, step size: 200 bp)

Species discrimination based on standard DNA barcodes, multiple barcode combinations, next-generation DNA barcodes and specific DNA barcodes

For the standard DNA barcodes, ITS exhibited the highest percentage of PI sites at 2.3%, followed by the combination of all four barcodes (rbcL + matK + trnH-psbA + ITS) at 0.9%, and the combination of the three plastid barcodes (rbcL + matK + trnH-psbA) at 0.4% (Table 2). Interestingly, the species identification rates corresponded well to the percentage of PI sites observed in the tree-based analyses. ITS alone successfully identified four out of the six Amentotaxus species (Amentotaxus assamica and A. hekouensis failed to be distinguished), resulting in the highest species identification rate (66.7%), while the combinations of rbcL + matK + trnH-psbA + ITS and rbcL + matK + trnH-psbA only resolved three and two species, respectively (Table 3, Figs. S3-S5). In the distance-based analyses, rbcL + matK + trnH-psbA + ITS displayed the highest species identification ability, successfully identifying four species, followed by ITS with three species, and rbcL + matK + trnH-psbA identifying two species (Table S1).

Table 2.

Sequence characteristics of DNA barcodes, plastome, nrDNA, and their combinations as well as ten selected hypervariable plastome loci used in Amentotaxus

DNA barcode	No. of taxa	Alignment length (bp)	No. of variable sites	No. of parsimony informative sites	GC content
ITS	23	1,050	33(3.1%)	24(2.3%)	63.6%
rbcL + matK + trnH-psbA	23	2,785	12(0.4%)	12(0.4%)	37.1%
rbcL + matK + trnH-psbA + ITS	23	3,835	45(1.2%)	36(0.9%)	44.4%
nrDNA cistron	23	5,984	41(0.7%)	29(0.5%)	55.5%
plastome	23	139,017	754(0.5%)	581(0.4%)	35.8%
plastome + nrDNA cistron	23	145,001	795(0.5%)	610(0.4%)	36.6%
rrn16-trnV	23	1,520	49(3.2%)	45(3.0%)	28.4%
trnL-F	23	932	16(1.7%)	14(1.5%)	32.7%
trnR-ndhF	23	1,189	15(1.3%)	15(1.3%)	38.6%
clpP-rps12	23	614	8(1.3%)	8(1.3%)	19.2%
accD	23	2,997	43(1.4%)	35(1.2%)	32.7%
trnL-UAA	23	562	8(1.4%)	7(1.2%)	37.3%
ycf1-chlN	23	987	14(1.4%)	12(1.2%)	29.4%
accD-rbcL	23	1,372	22(1.6%)	14(1.0%)	32.9%
psbM-trnD	23	1,344	14(1.0%)	11(0.8%)	34.9%
ndhC-trnV	23	1,977	32(1.6%)	13(0.7%)	34.0%

Table 3.

Species discrimination of standard DNA barcodes, plastome, nrDNA, and their combinations as well as ten selected hypervariable plastome loci based on tree-based analyses across Amentotaxus. Numbers in each species column indicate bootstrap support (BS) values (%) supporting the monophyly of conspecific accessions with BS ≥ 50%; “N” indicates that the corresponding species was not resolved as monophyletic or had BS < 50% in the phylogenetic tree

DNA barcode	A. yunnanensis	A. argotaenia	A. poilanei	A. assamica	A. formosana	A. hekouensis	discrimination rate
ITS	99	88	N	93	99	N	4 (66.7%)
rbcL + matK + trnH-psbA	65	N	N	N	100	N	2 (33.3%)
rbcL + matK + trnH-psbA + ITS	100	92	N	N	100	N	3 (50%)
nrDNA cistron	99	93	52	95	100	N	5 (83.3%)
plastome	100	100	99	N	100	N	4 (66.7%)
plastome + nrDNA cistron	100	100	99	N	100	N	4 (66.7%)
rrn16-trnV	88	91	N	N	100	N	3 (50%)
trnL-F	55	81	N	N	100	N	3 (50%)
trnR-ndhF	N	N	N	N	100	N	1 (16.7%)
clpP-rps12	96	53	76	N	98	N	4 (66.7%)
accD	N	96	66	N	98	N	3 (50%)
trnL-UAA	N	N	N	N	100	N	1 (16.7%)
ycf1-chlN	N	N	N	N	100	N	1 (16.7%)
accD-rbcL	87	56	N	N	100	N	3 (50%)
psbM-trnD	N	56	N	N	100	N	2 (33.3%)
ndhC-trnV	N	N	N	N	100	N	1 (16.7%)

The complete plastome contained 581 PI sites (0.4%), while the nrDNA cistron included 29 PI sites (0.5%), and their combination (plastome + nrDNA) contained 610 PI sites (0.4%) (Table 2). Based on the tree-based analyses, all Amentotaxus species except A. hekouensis could be identified by nrDNA (83.3%), the plastome and its combination with nrDNA were unable to distinguish A. assamica and A. hekouensis, although other species were successfully identified (66.7%) (Table 3, Fig. S6). Furthermore, the plastome + nrDNA combination provided significantly higher support values for the identification of A. poilanei compared to nrDNA alone. Based on genetic distance analyses (Table S1), nrDNA successfully identified A. yunnanensis, A. assamica, and A. formosana (50%). However, both the plastome and plastome + nrDNA successfully identified four species (except for A. assamica and A. hekouensis), achieving a success rate of 66.7%.

Through sliding window analyses of the newly assembled Amentotaxus plastid genomes, 10 highly variable regions (designated ‘mutational hotspots’) were identified and assessed for their species discriminatory power. These included accD, trnL-UAA, trnR-ndhF, ycf1-chlN, accD-rbcL, clpP-rps12, ndhC-trnV, psbM-trnD, rrn16-trnV and trnL-F. The rrn16-trnV region exhibited the highest proportion of PI sites (3.0%), followed by trnL-F (1.5%), trnR-ndhF and clpP-rps12 (both 1.3%), accD, trnL-UAA, ycf1-chlN (all 1.2%), accD-rbcL (1.0%), psbM-trnD (0.8%) and then ndhC-trnV (0.7%) (Table 2). In the tree-based analyses, only clpP-rps12 showed high species resolution, identifying four out of six species (except A. hekouensis and A. assamica), achieving a success rate of 66.7% (Fig. S7). Four loci, accD, accD-rbcL, trnL-F, rrn16-trnV, were able to identify three species (50%), while the remaining loci identify only one or two species (Table 3). In the distance-based analyses, clpP-rps12 again distinguished the four species (except A. hekouensis and A. assamica) with a success rate of 66.7%. The loci accD, psbM-trnD, trnL-F, and ycf1-chlN identified three species (50%), while the remaining loci identify only one or two species (Table S1). The results of tree-based species identification were used for subsequent discussion.

Species discrimination of plastome, nrDNA cistron, and Skmer analysis

In the plastome phylogenetic tree (Fig. 2a), four out of the six Amentotaxus species, A. poilanei, A. formosana, A. argotaenia, and A. yunnanensis, formed species-specific clades with strong support values. Samples of A. hekouensis were not resolved as monophyletic and fell in grades, with A. poilanei nesting within the A. hekouensis clade. Additionally, samples of A. assamica also were not resolved as monophyletic due to one individual (AM61) from north Myanmar formed a distinct branch that was sister to the clade of A. hekouensis + A. poilanei, while the other two individuals (AM07, AM08) formed a maximally-supported clade.

Fig. 2.

Maximum likelihood tree across Amentotaxus. a ML tree constructed based on entire plastomes. b ML tree constructed based on nrDNA cistron sequences. Bootstrap support (BS) values are given for each taxon node, with “*” indicating BS = 100%

In line with the plastome tree, the four species, A. formosana, A. argotaenia, A. yunnanensis, and A. assamica, were successfully identified in the nrDNA cistron phylogenetic tree, each forming a species-specific clade with high bootstrap support (BS > 90%) (Fig. 2b). In addition, the samples of A. assamica were resolved as monophyletic with maximum support. Conversely, A. hekouensis was not resolved as monophyletic: three accessions formed a clade with moderate support (BS = 78%), while one accession (Am24) was sister to the A. poilanei clade with weak support (BS = 37%). The samples of A. poilanei were resolved as monophyletic (BS = 52%), and also nested within the clade of A. hekouensis (Fig. 2b).

The Skmer analysis based on nuclear genome skimming data, excluding organelle reads, showed that A. poilanei, A. formosana, A. argotaenia, and A. assamica were resolved as monophyletic with strong support. While, A. hekouensis and A. yunnanensis were not resolved as monophyletic. The sample AM50 of A. hekouensis was sister to the clade comprising the remaining samples of A. hekouensis and A. poilanei, while A. yunnanensis did not form a cohesive species-specific clade and sister to the clade of A. assamica (Fig. 3).

Fig. 3.

Skmer tree of Amentotaxus species based on unassembled sequencing data without organelle reads. Branches of each species are designated in different colors. Bootstrap support (BS) values are given for each node

Discussion

Performance of species discrimination across the standard, Super-, next-generation, specific DNA barcodes, and Skmer analysis

The decreasing costs of next-generation sequencing (NGS) have facilitated the extensive application of next generation barcodes, demonstrating high efficacy for distinguishing plant species [2, 14, 16, 56, 57]. In this study, we found that compared to the standard barcodes (rbcL + matK + trnH-psbA + ITS), the super-barcode (the complete plastome) and next-generation barcode (plastome + nrDNA cistron) showed relatively higher species discrimination rates, successfully distinguishing four out of the six recognized species of Amentotaxus with greater support values. Interestingly, ITS alone achieved an equivalent species discriminatory power to both the super-barcode and the next-generation barcode, and outperformed the combinations of the three standard plastid DNA barcodes and the combination of all four standard barcodes including ITS (Table 3). While the nrDNA cistron provided the highest species resolution, successfully identifying all but one of the six species; A. hekouensis could not be distinguished in any DNA barcodes dataset due to A. poilanei consistently nesting in its clade. Our findings indicated that the complete plastome consistently offered higher resolution for species discrimination compared to standard barcodes, as reported in previous studies (e.g., [2, 56, 58, 59]). The next-generation DNA barcode, which encompassed a greater number of variable sites, may enhance species discrimination capabilities when excluding the hybrid accessions [57, 59].

Although the plastome serves as a powerful super-barcode for species discrimination, its application may be limited by the relative high sequencing costs, data complexity, and analytical challenges, as plastome analyses require processing large genomic datasets and relying on more sophisticated bioinformatic pipelines compared to standard barcode approaches [2, 16]. Consequently, lineage-specific barcodes have been proposed to improve species discrimination [60]. In this study, the clpP-rps12 plastid region, one of the ten highly variable plastid loci, showed the same species discriminatory power as the complete plastome, successfully identifying four out of the six species (66.7%). Considering the appropriate length of clpP-rps12 (608–613 bp in length) for Sanger PCR and sequencing, we proposed this region as an effective special DNA barcode for discriminating Amentotaxus species.

The Skmer analysis is an assembly-free method for estimating genomic distances between a query and reference genome skims [25] and has shown promising potential for species identification [3, 57, 61]. In this study, the Skmer analysis yielded results consistent with those obtained from nrDNA, with the exception of A. yunnanensis and A. hekouensis, which could not be resolved as a distinct species (Fig. 3). It is important to note that Amentotaxus has a large genome size of c. 30G [62], while only approximately 2G (1.7–6.8 G) sequencing data were utilized here, which may be insufficient for effective species discrimination using Skmer [25]. Increasing the amount of sequencing data may enhance species resolution, although this requires further evaluation.

Based on the monophyly assumption of barcode species delineation, A. hekouensis and A. poilanei should be treated as a single species. However, there are consistent morphological differences between the two species. If one assumed their conspecificity, it would increase the species discrimination rate with the plastome, plastome + nrDNA cistron and Skmer analysis from 66.7% to 80%, and nrDNA cistron from 83.3% to 100%, respectively (but see below).

Species delimitation and possible cryptic species

The taxonomy of Amentotaxus has primarily relied on morphological characters, including stomatal band width and color, as well as leaf shape and size [28, 34]. However, these traits are often variable and can exhibit overlapping states among different species [33, 34, 38]. In this study, three species, A. argotaenia, A. yunnanensis, and A. formosana, were consistently distinguished using standard, super-, next-generation, and Amentotaxus-specific DNA barcodes, suggesting the robustness and reliability of the species delimitation for these taxa [28, 29]. Based on morphological and distributional characteristics, these species are all also considered “good” species without taxonomic controversy.

Samples of A. poilanei collected from central Vietnam were found to be monophyletic based on standard, super-barcode, nrDNA, next-generation, and Amentotaxus-specific DNA barcodes, as well as Skmer analysis. Although A. poilanei appears to be a delineated species, it consistently nested within the clade of A. hekouensis, leading to the paraphyly of this species (Figs. 2 and 3). Genetically, these two species should be merged into a single species, as they cluster within a clade characterized by relatively short branch lengths indicating their young age. Amentotaxus hekouensis, a recently described new species, is mainly distributed in the neighboring regions between China (southeast Yunnan) and north Vietnam, and Laos [29], while A. poilanei is only known from its type locality in southern Vietnam (Gia lai-Kon tum) [63]. However, the two species show consistent morphological differences: Amentotaxus hekouensis differs from the latter by its linear or linear-lanceolate leaves and larger leaf size with 8–12.5 cm × 9–14 mm (vs. linear-lanceolate leaves and 5–9 cm × 5.5–8.4 mm), long acuminate apex and flat with slightly wavy leaf margin (vs. acute apex and strongly downcurved leaf margin), and stomatal bands equal to or slightly narrower than marginal bands (vs. stomates bands c. 2 times as wide as the marginal bands) [29, 63] (Fig. S8).

We hypothesize that A. poilanei originated from a dispersal event or fragmentation from a marginal population of A. hekouensis that has adapted to a local environment in southern Central Vietnam, resulting in morphological divergence due to morphological drift or establishment of isolation barriers, serve as a budding speciation event [64, 65]. Further study with more populations is needed to fully uncover its speciation mechanism.

In this study, A. assamica was resolved as monophyletic based on nuclear DNA barcodes (ITS and nrDNA) and Skmer analysis, but exhibited non-monophyly with plastid DNA barcodes due to accession AM61 being sister to the clade of A. hekouensis + A. poilanei, which lead to failure of its species identification (Figs. 2 and 3, Fig. S3, Table 3). The differing inheritance patterns of plastid (paternal) and nuclear (biparentally) genomes in Taxaceae [20] suggest that the discordance of the placement of AM61 between nuclear and plastid phylogenetic trees indicating a possible hybrid event [2, 57]. The accession AM61 identified as A. assamica was collected from Kachin, north Myanmar, representing a new record of Amentotaxus from this area [66]. In the plastome phylogenetic tree, AM61 formed a deep diverged distinct lineage with a relatively long branch, positioned as sister to the A. hekouensis + A. poilanei clade, indicating that its parent species may represent an undescribed cryptic Amentotaxus species, geographically situated between A. assamica and A. hekouensis/A. poilanei. However, more field surveys are necessary to discover a potential cryptic species in this region.

If excluding the possible hybrid accession AM61, A. assamica can be successfully identified using all DNA barcodes tested here, improving the species discrimination power. Considering the likely recent evolutionary divergence of A. poilanei, the delineation of all currently recognized Amentotaxus species based on molecular, morphological, and geographic evidence seems justified. The integration of the entire plastid genome and nrDNA cistron data was also found to provide a more reliable and robust framework for distinguishing closely related species in other plants [2, 3, 27]. Our findings underscore the importance of integrating molecular data and morphological characteristics with comprehensive sampling across the species’ range when evaluating species boundaries, thereby enhancing our understanding of the complex evolutionary history surrounding Amentotaxus.

Implications for conservation of Amentotaxus

Accurate species delineation and identification are essential for effective species management and conservation [1, 2]. Currently, all species of Amentotaxus are classified as threatened at either the national or global level (IUCN, https://www.iucnredlist.org). Therefore, accurate identification of these species is crucial for developing effective conservation strategies for this group. In this study, all but A. hekouensis could be accurately delineated using DNA barcodes. Amentotaxus hekouenis could not be identified due to the nesting of the A. poilanei clade in its clade. Although these two species exhibit distinct morphological characteristics and differ in their distribution range, suggesting they are separate species, their conservation managements must be addressed independently due to their differing conservation statuses: A. hekouensis is classified as Endangered (EN) [29], while A. poilanei is classified as Vulnerable (VU) [67].

This study investigated the species delineation and discrimination power of Amentotaxus using standard-, super-, next-generation and Skmer method, which is significant for guiding species conservation and the sustainable use of biological resources. For instance, Gu et al. [68] utilized genomic evidence to identify a new pangolin species, highlighting the importance of molecular data in preventing the loss of unrecognized biodiversity. Our findings suggest the presence of a possible cryptic species and a natural hybrid (AM61) of Amentotaxus never undescribed before in north Myanmar, thereby expanding our current knowledge of species diversity and evolution. Nevertheless, future field investigations in this region, together with extensive sampling and population genetic studies, are needed to ascertain the origin of this lineage. Such research will have important implications for conservation efforts aimed at preventing the extinction of a potential undescribed cryptic species in the wild.

Conclusion

We generated genome skimming data for all six currently recognized Amentotaxus species using multiple accessions per species to evaluate the standard DNA barcode, super-barcode, next-generation barcode, and Skmer analysis for species delineation and discrimination capabilities. Comparing the combinations of standard DNA barcodes, the next-generation barcode, the super-barcode, and the nrDNA cistron all exhibited comparable higher species discrimination rates, with the last achieving the highest species discriminatory rate, successfully resolving five out of the six recognized species, followed by ITS, the next-generation barcode, super-barcode, and Skmer analysis, each identifying four species. As clpP-rps12 showed the highest species discriminatory power and possesses an appropriate sequence length for Sanger sequencing, we proposed this region as a promising Amentotaxus-specific DNA barcode.

Amentotaxus poilanei consistently nested within the clade of A. hekouensis, leading to the failure to delineate this species. We hypothesize that A. poilanei originated from a population of A. hekouensis that adapted to a local environment, resulting in morphological divergence and the recent budding speciation of A. poilanei. Our findings indicated a possible cryptic species undescribed and hybrid origin of accession AM61 from Kachin, north Myanmar, though more research is necessary to ascertain their origin. The findings greatly enhance the knowledge of species limits and diversity within Amentotaxus and offer important insights for conservation efforts to preventing the extinction of the undescribed cryptic species in the wild.

Authors’ contributions

L.G. and Z.M. designed the experiments. L.G. and P.T. collected the plant samples., D.L., J.W., S.C., and C.F. performed the experiments and analyzed the data. D.L., Z.M., L.G., M.M. and J.Y. wrote and revised the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by the National Key Research and Development Program of China (2023YFA0915800), the 811 program of Shandong Agricultural University, and the Key Basic & Applied Research Program of Yunnan Province, China (202101BC070003).

Data availability

The original contributions presented in the study are publicly available. Plastome and nrDNA sequences generated in this study have been deposited in GenBank under accession numbers PX703143 to PX703163 and PX651884 to PX651904, respectively (Table 1). The information of the primer sequences for clpP-rps12 and the detailed PCR conditions are provided in the Supplementary Materials (Table S2) of the manuscript.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.