Integrative COI Barcoding and Species Delimitation in Echinodermata from Vietnam

Abstract

Echinoderms are marine invertebrates that play important roles in structuring marine benthic ecosystems. DNA barcoding has become a valuable tool for species identification; however, reference DNA barcode libraries for echinoderms remain incomplete. This study aims to: (i) develop a COI-5′ reference dataset for echinoderms from Vietnam by integrating DNA barcodes with morphological data; (ii) evaluate species resolution and barcode gaps using multiple analytical approaches; (iii) assess the consistency of species assignments from BOLD and GenBank for echinoderms collected in Vietnam; (iv) make barcode data publicly available to support global reference database development. Thirty-two echinoderm specimens representing 16 species were analyzed for COI-5′ sequences, and BLAST assignments were highly concordant with those from GenBank and BOLD. Integrative validation confirmed that all taxa were monophyletic in the Neighbor Joining Tree, formed single OTUs in Cluster Sequences, and exhibited clear barcode gaps greater than 3% to the nearest-neighbor species. These results provided species-level resolution for 75% and genus-level resolution for 90% of the records. The dataset, spanning four classes, eight orders, and eleven families, enhances barcode coverage and contributes records (ProcessIDs. BINs; GenBank accessions) to public repositories. This study delivers the first curated COI-5′ reference library, supporting regional baselines for taxonomy, conservation, and biodiversity assessment.

1. Introduction

Echinoderms are a phylum of marine invertebrates widely distributed throughout the world’s oceans. They consist of more than 7000 living species divided into five distinct taxonomic classes: Asteroidea (incl. starfishes); Echinoidea (incl. sea urchins, sand dollars, and sea biscuits); Crinoidea (incl. sea lilies and feather stars); Ophiuroidea (incl. basket stars and brittle stars); and Holothuroidea (incl. sea cucumbers) [1]. Currently, the known number of species is believed to be far lower than the actual amount, as new species are being discovered and named every year [2,3,4,5,6]. They are often key, long-living species that shape and maintain the status of many marine ecosystems, holding significant economic and medical value for humans [7,8,9].

Although they have very important roles, echinoderms are considered one of the most morphologically enigmatic animal groups with complex and confusing anatomical organization [10,11,12,13]; this makes their classification challenging and subject to ongoing revisions across all classes [14,15,16,17]. Such problems have led to the development of molecular genetic identification techniques such as DNA barcoding and advanced their application [18,19,20,21].

DNA barcoding was proposed by Hebert et al. [18] based on a standardized fragment of the mitochondrial cytochrome c oxidase subunit I (COI) gene at the 5 end (COI-5′). This method has become a vital tool for cataloging life, enabling scientists to differentiate species, identify cryptic lineages, and monitor biodiversity on a large scale for various organisms [22,23,24,25]. Universal primers [26] and the development of the Barcode of Life Data System (BOLD) [27], along with the Barcode Index Number (BIN) system [28], have further established DNA barcoding as a reproducible and scalable framework for linking sequences with voucher specimens and taxonomy. This technique has unlocked new perspectives for the accurate identification and inventory of echinoderms. In an early global survey, low intraspecific COI (~0.6% on average) and high interspecific divergences (~15%) were reported, enabling the accurate resolution of ~98% of the species analyzed [20]. Regional libraries, such as those developed for Canadian waters [29], have confirmed the presence of clear barcode gaps, further demonstrating the utility of COI for echinoderm systematics and biodiversity research. However, DNA barcode coverage of echinoderms remains geographically uneven and taxonomically incomplete, particularly in tropical regions. For example, while more than 34,000 echinoderm COI records are publicly available in BOLD, coverage for some regions, such as southern Africa and Asia, remains below 500 sequences each [30,31]. Similarly, a recent survey of shallow-water echinoderms from Central America documented 324 species, yet only 118 were represented in GenBank and 110 in BOLD, resulting in identification success rates of ~25% and misidentification rates of ~37% [32]. In Southeast Asia, national efforts are expanding: the Philippines published COI barcodes for 19 commercially important holothuroids with well-characterized K2P distances and BIN partitions [33], and COI data have been generated for Stichopus spp. from Karimunjawa National Park, Indonesia [34]. These research gaps highlight persistent challenges, including uneven sampling, misidentified reference records, and variable primer performance across taxa, all of which limit the reliability of current databases.

Vietnam’s extensive coastline and diverse marine ecosystems support a rich echinoderm fauna of approximately 350 species [35] belonging to five classes, with high concentrations in families such as Holothuriidae, Astropectinidae, Cucumariidae, and Amphiuridae. Regional inventories underscore this rich biodiversity with specific spatial structures: Surveys of the northeastern islands documented 41 species (29 genera, 18 families) distributed across Ha Long–Cat Ba (33 species), Bai Tu Long (25 species), and Co To–Thanh Lan (24 species). In south-central waters, Xuan Dai Bay (Phu Yen province) recorded 19 echinoderm species in three classes (Asteroidea, Holothuroidea, and Echinoidea) among 93 macro-invertebrate taxa, with several species listed in the Vietnam Red Data Book [36]. Fishery records from Phu Quoc (the Gulf of Thailand) noted a ~10-fold decline in total sea cucumber landings over a decade (1993: ~3 t day⁻¹ to 2003: ~0.3 t day⁻¹), with the subsequent displacement of fishing efforts into neighboring Cambodian waters as local stocks dwindled [37].

Although many of these species have broad Indo-Pacific distributions and have been barcoded in other regions, COI barcode data explicitly linked to specimens collected in Vietnam remain extremely limited. A search of the BOLD database currently retrieves only 23 COI records originating from Vietnam, all mined from GenBank and generated by international research groups. Notably, 20 of these records correspond to a single species (Holothuria fuscogilva), while the remaining three are assigned to Holothuria cf. fuscogilva. To date, COI barcode data for most Vietnamese echinoderm taxa, including Asteroidea, Echinoidea, and Crinoidea, are largely absent. This substantial gap between documented species diversity and available molecular barcode data highlights the need for DNA barcoding studies based on vouchered specimens collected within Vietnam. The present study aims to (i) establish an initial, COI-5′ reference dataset for echinoderms from Vietnam by linking DNA sequences with morphological identifications; (ii) assess species resolution and barcode gaps using multiple analytical approaches; (iii) evaluate the consistency of species assignments generated by BOLD and GenBank for locally collected echinoderm samples; and (iv) provide publicly accessible barcode records to strengthen global reference databases. By addressing a regional gap in barcode coverage, this work contributes to enchinoderm taxonomy, conservation, and biodiversity monitoring in tropical coastal ecosystems.

2. Materials and Methods

2.1. Sampling

In the present study, 32 echinoderm specimens were collected from Ha Long Bay (July 2023) in the north of Vietnam, Lang Co Bay (September 2024) and Van Phong Bay (May 2025) in the central region of the country, and Con Dao (June 2023) and Phu Quoc (February 2025) in the south. Detailed information about the sampling locations, geographical coordinates, and sampling dates is given in Figure 1 and Table S1. The collection process and field trials were performed under permits issued by provincial authorities. No protected areas or species requiring special authorization were sampled. Sampling was conducted via SCUBA diving in water depths between 3 and 12 m during either the morning or afternoon. Individuals were photographed in situ and ex situ; relaxed, when necessary, in seawater; and then preserved in 95% ethanol (Duc Giang Chemicals, Hanoi, Vietnam) or formalin (Duc Giang Chemicals, Hanoi, Vietnam) for morphological examination. All specimens were identified to species level based on morphological characteristics. Whole-organism vouchers were preserved in 95% ethanol and deposited at the Institute of Marine Fisheries Research and the Institute of Science and Technology for Energy and Environment (Hai Phong, Vietnam). Tissue samples for molecular analyses were kept on ice in the field, transferred to the Institute of Chemistry, VAST, Hanoi, Vietnam, and stored at −20 °C. The specimens examined in this study are listed in Supplementary Table S1.

2.2. DNA Extraction, COI Amplification, and Sequencing

Depending on the taxonomic group, different body parts were dissected and used for DNA extraction: tube feet or arm tip tissues of asteroids and ophiuroids, body wall tissues of holothurians, tube feet or spine base tissues of echinoids, and pinnule tissues of crinoids. Genomic DNA was extracted using the Chelex 100 method [38]. The integrity and quantity of DNA were verified using 0.8% agarose gel electrophoresis and spectrophotometry (NanoDrop 2000, Thermo Scientific, Waltham, MA, USA).

Fragments of the 5′ end of the mitochondrial cytochrome c oxidase subunit I (COI-5′) gene were amplified using primers designed for this study. Primer design was conducted using DNAMAN XL (Lynnon Biosoft, San Ramon, CA, USA) based on alignments of reference COI sequences retrieved from GenBank and BOLD. These primers were optimized in silico and validated through gradient PCR trials. All primer sequences have been submitted to the BOLD Primer Database and are accessible under primer names (Supplementary Table S2). The PCRs were performed in a final volume of 25 µL, containing 1× DreamTaq buffer (Thermo Scientific™, Vilnius, Lithuania), 2.0 mM MgCl₂ (Thermo Scientific™, Vilnius, Lithuania), 0.2 mM each dNTP (Thermo Scientific™, Vilnius, Lithuania), 0.4 µM each primer, 1U Taq DNA polymerase (Thermo Scientific™, Vilnius, Lithuania), and 10–20 ng template DNA. The thermal conditions comprised an initial step at 94 °C for 2 min followed by 35 cycles of 94 °C for 30 s, 48–53 °C for 30 s, and 72 °C for 45–60 s, followed by a final extension of 72 °C for 5–10 min. PCR conditions were optimized for each primer pair, and amplification products were verified on 1% agarose gel with Redsafe (Intron, Seongnam, Gyeonggi, Republic of Korea) in 1× TAE (Thermo Scientific, Vilnius, Lithuania). Successful amplicons were purified using the GeneJet Gel Extraction Kit (Thermo Scientific, Vilnius, Lithuania) and used for direct sequencing or cloning into the pJET1.2 vector using the CloneJet PCR Cloning Kit (Thermo Scientific, Vilnius, Lithuania) following the manufacturer’s instructions. Bidirectional Sanger sequencing was performed using First Base Pte Ltd. (Singapore).

Sequences were assembled, edited, and checked for correct reading frames based on invertebrate mitochondrial code using DNAMAN XL (Lynnon Biosoft, San Ramon, CA, USA). Sequences, specimen metadata, and trace files were uploaded to BOLD v5 under project “HMDG-DNA barcoding analysis_Echinodermata_Vietnam”. The sequences were submitted to GenBank, and their accession numbers in NCBI and ProcessID in BOLD are presented in Supplementary Table S3.

2.3. Molecular Data Analysis

The effectiveness of the DNA barcode was validated for discriminating species using different methods. For similarity-based identification, curated COI-5′ sequences were queried against GenBank (BLASTn) and the BOLD Identification Engine (BOLD Systems v5, “Barcode ID”). For each query, the top hit, percent identity, coverage, and taxon were recorded. Provisional species-level assignments required ≥97–99% identity with ≥95% coverage and concordant names across databases: 95–97% identity was treated as the genus-level threshold.

The species delimitation step entails confirming whether COI-5′ sequences are valid DNA barcodes of their identified species using the REfined Single Linkage (RESL) method. RESL is used by the BOLD Systems to group similar DNA sequences into Barcode Index Numbers (BINs) and later into operational taxonomic units (OTUs) [28]. Cluster membership was examined to determine whether all sequences attributed to a species were contained within a single OTU. The ‘BIN Discordance Report’ analysis tool available within BOLD was used to assess the concordance between barcode sequence clusters and nominal species.

The Assemble Species by Automatic Partitioning (ASAP) method [39] was further applied to find species boundaries in the DNA barcode data, ranking potential species groups by ASAP-score (lower is better) using the Kimura 2-parameter (K2P) distance model [40] on the online platform (https://spartexplorer.mnhn.fr/delimitation; assessed on 8 November 2025) [41]. The two automated genetic groups organized by RESL and ASAP were compared.

The taxon trees were constructed by using 1000 bootstrap replicates and rooted Neighbor-Joining (NJ) with MEGA software version 11 [42] with the pairwise K2P model and a dataset containing species in the same genus. Monophyletic analysis was predicated on the taxon NJ tree Species were considered successfully identified when all individuals of a species formed a monophyletic group. In addition, another NJ tree was run based on all obtained barcodes using MEGA to determine whether a distinct and well-supported cluster corresponding to one OTU was formed for each species. All trees were constructed with the aim of obtaining support values for taxa in which genetic clusters may represent existing, new or cryptic species; it was not our objective to infer phylogenetic relationships among the analyzed species.

To examine genetic distance, a comparison between the maximum distance within species and the minimum distance between species was performed using BOLD’s Barcode Gap Analysis tool. A barcode gap was determined when the maximum intraspecific divergence was less than the nearest-neighbor distance; any overlaps indicated a potential taxonomic conflict. Due to its established robustness reported for different general species [18], fish [20], and marine mollusks [43,44,45,46], the common distance cutoff of 3% was utilized for species separations of marine echinoderms in this study. A scatter plot was created to show the maximum intra-specific distances against the NN distances, aiming to show the existence and magnitude of the barcode gap for all barcodes. Barcode gap histogram (genetic distance vs. frequency) based on COI-5′ K2P distance was also generated by the ASAP software, SPART format [41]

3. Results

3.1. Species Identification via Morphology and BLAST (GenBank and BOLD)

PCR amplification was successful for 32 echinoderm specimens, yielding COI-5′ sequences that were 454–1173 bp in length. A total of 32 new barcode sequences were uploaded to GenBank and BOLD.

Table 1. A comparison of the identification of 32 specimens based on morphology and blasting their COI-5′ sequences with GenBank and BOLD databases.

No	Sample ID	COI-5′ seq. Size (bp)	Processed ID (BOLD)	Morphological Identification	Species Names on GenBank and BOLD	Blast
						GenBank Similarity (%)	BOLD Similarity (%)
1	DG6DTh-76	758	HMDG022-25	Culcita novaeguineae	Culcita novaeguineae	99.36	98.98
2	DGVP60	747	HMDG033-25	Culcita novaeguineae	Culcita novaeguineae	99.78	99.26
3	DGVP93	747	HMDG034-25	Culcita novaeguineae	Culcita novaeguineae	99.78	99.56
4	DG3C-22	454	HMDG010-25	Echinaster sp.	Echinaster luzonicus	99.12	99.78
5	DG1THD-65	745	HMDG014-25	Echinaster sp.	Echinaster luzonicus	98.47	99.59
6	DG1Ta-48	495	HMDG015-25	Echinaster sp.	Echinaster luzonicus	99.37	98.73
7	DG2HT-108	752	HMDG032-25	Echinaster sp.	Echinaster luzonicus	98.39	99.60
8	DG6HT-11	508	HMDG017-25	Fromia milleporella	Fromia milleporella	95.15	97.24
9	DG2TT-157	509	HMDG018-25	Fromia milleporella	Fromia milleporella	95.35	97.25
10	DG5C-7	508	HMDG019-25	Fromia milleporella	Fromia milleporella	94.69	96.85
11	DG5THD-19	509	HMDG020-25	Fromia milleporella	Fromia milleporella	95.15	97.25
12	DGVP40	626	HMDG036-25	Linckia laevigata	Linckia laevigata	100	100
13	DG52-106	517	HMDG003-25	Echinotrix diadema	Diadema setosum	99.02	99.02
14	DG3TD-2	1127	HMDG011-25	Diadema savignyi	Diadema setosum	99.79	99.71
15	DG2DR-293	1127	HMDG012-25	Diadema setosum	Diadema setosum	99.73	99.81
16	DG5DTh-23	482	HMDG013-25	Diadema setosum	Diadema setosum	99.79	99.79
17	DG7THD-135	758	HMDG023-25	Toxopneustes pileolus	Toxopneustes pileolus	100	99.00
18	DG2DTh-266	614	HMDG030-25	Toxopneustes pileolus	Toxopneustes pileolus	99.67	98.84
19	DG4TD-314	614	HMDG031-25	Toxopneustes pileolus	Toxopneustes pileolus	99.51	99.01
20	DG5TT-12	1003	HMDG025-25	Stephanometra indica	Stephanometra indica	99.20	99.35
21	DG3DD-86	1004	HMDG026-25	Stephanometra indica	Stephanometra indica	99.60	99.60
22	DG4C-244	740	HMDG024-25	Clarkcomanthus sp.	Clarkcomanthus sp.	98.52	98.52
23	DG76-63	583	HMDG004-25	Colochirus quadrangularis	Colochirus quadrangularis	97.70	97.70
24	DG101-47	563	HMDG005-25	Stichopus chloronotus	Colochirus quadrangularis	99.82	99.82
25	DG3DTh-18	563	HMDG027-25	Holothuria atra	Holothuria atra	100	100
26	DG3Ta-234	562	HMDG029-25	Holothuria atra	Holothuria atra	99.82	99.82
27	DG3HT-112	533	HMDG028-25	Holothuria hilla	Holothuria hilla	100	100
28	DG4DD-28	1173	HMDG009-25	Holothuria leucospilota	Holothuria leucospilota	99.91	100
29	DGVP55-262	747	HMDG037-25	Stichopus herrmanni	Holothuria impatiens	99.85	99.85
30	DGDLB-41	616	HMDG007-25	Holothuria cinerascens	Holothuria cinerascens	99.32	98.78
31	DG1DTh-223	1049	HMDG021-25	Stichopus chloronotus	Stichopus chloronotus	99.43	99.43
32	DG8TT-20	623	HMDG038-25	Protoreaster nodosus	Protoreaster nodosus	100	100

shows species assignments, GenBank/BOLD identity percentages, and morphological corrections for all 32 specimens. BLAST comparisons against GenBank (NCBI) and the BOLD Identification Engine produced concordant assignments between the two repositories. Across the dataset, 32 sequences were mapped to 16 species, with high pairwise identity (97–100%) to named conspecific references in both databases. The exception was DG6HT-11, DG2TT-157, DG5C-7 and DG5THD-19 (originally identified as Fromia milleporella), which showed the similarity lower than 97% threshold in one or both databases. The low identity indicates insufficient confident species-level assignment, suggesting either the possibility of cryptic diversity, database under-representation, or a misidentified reference.

Overall concordance between morphological identification and COI-based BLAST results was high but not complete. Of the 32 specimens examined, 27 were morphologically identified to the species level, with 23 showing species-level agreement with COI data and four exhibiting discordant assignments. The remaining five specimens were identified only to the genus level morphologically. Notably, four specimens that could only be determined morphologically at the genus level as Echinaster sp. were identified as Echinaster luzonicus at the species level using COI-5′, with more than 98% similarity. These results illustrate that barcode sequence analysis provided additional diagnostic power for species identification in most cases whenever reference coverage was adequate. The COI-5′ sequences generated here allowed for successful initial species assignment: the BLAST results were concordant with those from GenBank and BOLD, with high identity and coverage of the matches, thus confirming the identity of the proposed taxa.

To confirm whether these sequences can be used as authoritative DNA barcodes for the studied species, three validation methods were performed: (i) monophyly on taxon Tree (NJ, K2P); (ii) BIN/OTU and ASAP partition cohesion using BOLD’s Cluster Sequences and ASAP method; (iii) barcode gap analysis to verify whether the maximal intraspecific divergence within species remained less than 3% and the distance to the nearest-neighbor species was more than 3%. Agreement across (i)–(iii) substantiated species-level validity, and discrepancies were flagged for targeted re-evaluation.

3.2. Monophyletic and Cluster Sequence Analysis

In our study, the BIN system assigned 16 echinoderm taxa to 16 BINs. Neighbor joining (NJ) analysis for each genus showed that 16 BIN-assigned species formed a monophyletic group. All barcodes provide 100% species monophyly and cluster as a single species corresponding (Supplementary File S1) to a single BIN. An example for Toxopneustes pileolus is shown in Figure 2, where three samples were grouped into a monophyletic cluster with other sequences of this species.

As shown in Figure 3, NJ analyses were conducted on the 32 barcode–COI sequences provided, revealing that specimens across the northern site, central mainland coast, and two southern islands were classified as 16 species spanning four classes, eight orders, and eleven families. Five belonged to Asteroidea (five clusters = five OTUs: OTU5, 6, 8, 14 and 15), two to Echinoidea (two clusters = two OTUs: OTU1 and 9), two to Crinoidea (two clusters = two OTUs: OTU10 and 11) and seven to Holothuroidea (seven clusters = seven OTUs: OTU2, 3, 4, 7, 12, 13 and 16). The ASAP delimitation method provides similar results as BIN clustering. The lowest ASAP-score (1.50) provides the best-fit scenario at the threshold distance of 2.68% with 16 species (File S2).

Therefore, the tree analysis also substantiates the classification of the obtained samples into 16 distinct species.

3.3. Barcode Gap Analysis

Figure 4 shows a clear barcode gap with no overlap between conspecific and congeneric distances computed by both the barcode gap analysis tool on BOLD and ASAP method.

A barcode gap for each taxon analyzed on BOLD was provided in

Table 2. Barcode gap analysis and cluster sequence analysis of 16 echinoderm species based on the divergence of their COI-5′ sequences.

No.	Species	BINs 1	Max. Intra Sp. 2 (%)	Min Inter Sp. 3 (%)
1	Clarkcomanthus sp.	BOLD:AHA1014 (*)	0	3.06
2	Colochirus quadrangularis	BOLD:ABA3321	2.73	10.89
3	Culcita novaeguineae	BOLD:ABA1943	3.23	75
4	Diadema setosum	BOLD:ABA3950	2.46	11.7
5	Echinaster luzonicus	BOLD:AAF8608	2.28	7.07
6	Fromia milleporella	BOLD:AHA5186	1.2	2.84
7	Holothuria atra	BOLD:AAB2668	2.96	13.29
8	Holothuria cinerascens	BOLD:ABA2112	1.62	10.59
9	Holothuria hilla	BOLD:AAF0588	1.14	12.44
10	Holothuria impatiens	BOLD:ABA1766	1.03	17.13
11	Holothuria leucospilota	BOLD:AAI8215	5.29	14.38
12	Linckia laevigata	BOLD:AAA1544	5.17	15.79
13	Protoreaster nodosus	BOLD:AAA4270	1.75	7.67
14	Stephanometra indica	BOLD:ABA5375	2.53	8.98
15	Stichopus chloronotus	BOLD:AAB6442	1.14	12.36
16	Toxopneustes pileolus	BOLD:ACB6445	1.18	2.1

¹ BINs—Barcode Index Numbers assigned in BOLD (RESL clusters approximating species). “*”: singleton = one record only. ² Max. Intra Sp. (%)—maximum intraspecific K2P distance (pairwise deletion) among sequences of the named species. ³ Min. Inter Sp. (%)—minimum interspecific (nearest neighbor, dNN) K2P distance to the closest congeneric sequence.

. In this study, we applied 3% as the cutoff threshold to separate echinoderm species; i.e., the maximum intraspecific distance within species (Max. Intra. Sp.) was determined to be <3% and the distance of a species to the nearest-neighbor species (Min Inter. Sp.) was >3%. A Max. Intra. Sp. value lower than 3% (from 0 to 2.96%) and Min Inter. Sp. value higher than 3% (3.06–75%) can be observed in 11 species. For three species, Culcita novaeguineae (3.23% vs. 75%), Holothuria leucospilota (5.29% vs. 14.38%), and Linckia laevigata (5.17% vs. 15.79%), the Max. Intra. Sp. and Min Inter. Sp. exceeded 3%. For the last two species, Fromia milleporella and Toxopneustes pileolus, both their Min Inter. Sp. and Max. Intra. Sp. were lower than the 3% cutoff threshold. The taxon Clarkcomanthus sp. could not be determined at the species level due to the fact that only one representative of this genetic lineage (in this study) could be found in BOLD; however, its genetic distance to its nearest neighbor species was 3% (Table 2). Altogether, our results indicate that COI-5′ is a suitable DNA barcode for identifying the 16 echinoderm species in this study.

4. Discussion

This study discerned that, for approximately 90% of the echinoderm species examined, morphological species identification was concordant with the results of the genetic analyses, highlighting the reliability of our morphological approach for most of the species examined. However, in many cases, relying on morphology alone can be limiting for echinoderms as their diagnostic traits are often flexible, variable in the development within a species, or similar between closely related species [20,47]. In our study, four discordant cases were found in taxa where families with similar morphologies are common—e.g., Holothuroidea (samples DG101-47 and DGVP55-262) and the sea urchins Diadema setosum versus D. savignyi—and intergeneric confusion can arise when the spines of specimens are worn, as in Echinothrix diadema. These patterns underscore the need to anchor names to vouchers and to cross-validate morphology with sequence-based evidence [20,48].

The use of molecular tools provided a useful complementary approach to species identification. In this study, we sought to match genetic barcodes from 32 COI-5′ sequences with the results from morphological analyses of 90% of echinoderms across four marine bays and islands of Vietnam. Additionally, the results from our monophyletic and barcode gap analyses, based on all related gene sequences available in GenBank and BOLD, exhibited >97 % identity with our queries, providing strong support for the identification of all species using COI-5′ barcodes. However, due to the limited availability of DNA sequence data for some species and genetic markers in public repositories, identification was not possible for certain species: while the DG4C-244 specimen in this study exhibited a ~99% match to Clarkcomanthus sp. and formed a distinct OTU, it lacked a close conspecific reference in BOLD (Supplementary File S1) and the analysis remained at the genus level due to the absence of species-level COI-5′ records in both databases. These limitations indicate the need to expand curated reference libraries and, where necessary, provide more complementary markers and broader geographic sampling to ensure accurate species-level identifications [29,48]. While the aforementioned record was identified as a singleton with a unique BIN, this was because the sequence did not show sufficient identity with any existing BINs in the database, rather than merely being a new entry in BOLD. The presence of a unique BIN within Clarkcomanthus, combined with limited reference data for the genus, suggests the potential existence of undocumented diversity, highlighting the need for integrative taxonomic approaches.

Our NJ method resolves 16 species across four classes, eight orders, and eleven families, providing a first barcode-supported overview of Vietnamese echinoderms. The patterns identified herein match those found in regional studies. Philippine surveys of economically important sea cucumbers revealed the formation of monophyletic clades and paraphyly/elevated intraspecific K2P in several Holothuria and allied taxa, indicating the presence of cryptic structures that require integrative examinations [33]. Similarly, Indonesian analyses of commercial Stichopus showed that the application of COI (±ITS) was crucial for species identification, regardless of confusing vernacular morphs, underscoring the value of using barcodes for the taxonomic classification of species under fishery pressure [34]. While Vietnamese taxa that form cohesive clades support the routine use of COI identification, internally split or genus-only cases (e.g., Clarkcomanthus sp.) require a more comprehensive approach, i.e., an integrative pipeline (OTU cohesion + barcode gap + morphology), to resolve the lineage status and establish the names of species for monitoring and management strategies.

The initial 2% DNA divergence cutoff proposed by Hebert et al. [18] for species identification is a guideline; however, the most appropriate cutoff value varies according to the specific group of taxa being studied [18,29,43,46,49]. In addition, the use of fixed distance cutoffs (2 or 3%) is insufficient for accurately defining species for all organisms; instead, analyses should rely on concordance among barcode gap diagnostics, NJ/ML monophyly, OTU cohesion, and name/biogeography, rather than distances alone [30,50]. The threshold of our sequences estimated via the ASAP method revealed the value of 2.86% to be the best partition of 16 species, like other identification methods (morphological and BIN bases). Therefore, a 3% DNA barcode divergence threshold was appropriate for identifying echinoderm species in this specific study. Calculating the maximum intraspecific distance using COI barcodes also facilitated the prediction of possible divergence within a species with an intraspecific genetic distance higher than 3%, as in the cases of our echinoderm species Culcita novaeguineae, Holothuria leucospilota, and Linckia laevigata. This finding prompts further investigation to determine if they are complexes of multiple cryptic or previously overlooked species [51]. In contrast, when both the maximum intraspecific and minimum interspecific distances of Fromia milleporella and Toxopneustes pileolus were less than 3%, it indicated a potential issue with species delimitation, such as recent divergence, hybridization, or the possibility that the identified species and its nearest neighbor were closely related. The high internal and low external divergence, coupled with the presence of a barcode gap, suggests that the two species may have a pronounced population structure or are in the process of incipient speciation, therefore warranting expanded geographic sampling, voucher-based morphological reassessments, and multilocus analyses to test their lineage status [50].

One of the most significant challenges in DNA barcoding is incomplete reference databases; the extremely high interspecific divergence of Culcita novaeguineae may result from this issue. The limited availability of genetic data for the sea star genus Culcita in public databases such as BOLD has the effect of representing Culcita novaeguineae as a monotypic genus. Currently, only nine conspecific COI-5′ records with three new records from this study exist and were assigned to a single BIN, BOLD:ABA1943. This highlights a significant gap in genetic information, as the sequences for eleven other Culcita species [31] are not yet publicly available for reference. This could hinder the robustness of barcode gap analysis as both maximum intraspecific and nearest-neighbor distance estimates are unreliable with low conspecific and incomplete congeneric sampling. Consequently, the results for C. novaeguineae should be treated as provisional until conspecific and congeneric references are increased and curated.

A morphological–molecular discordance was observed for two specimens (DG101-47 and DGVP55-262), in which morphology supported assignment to Stichopus chloronotus or Stichopus herrmanni, whereas COI-based analyses suggested affinities with Colochirus quadrangularis or Holothuria impatiens. As these taxa belong to different orders, direct morphological confusion is unlikely. Re-examination of available morphological characters confirmed that the original identifications were based on the best external diagnostic features; however, because each specimen is represented by a single individual, a more detailed morphological reassessment could not be conducted at this stage.

The COI sequences were treated as independent molecular observations and included in clustering and species delimitation analyses, while species assignment primarily relied on morphological identification. The observed discordance was therefore interpreted with caution. It should be noted that public reference databases such as GenBank and BOLD, while essential for DNA barcoding, may occasionally contain misidentified records because taxonomic identities are not always independently verified [52]. As highlighted by Radulovici et al. [53], such limitations emphasize the need for cautious interpretation of molecular results. Nevertheless, DNA barcoding remains a powerful tool for biodiversity research when applied within an integrative framework that combines molecular data with expert morphological identification.

Taken together, the present results show that BIN, OTU, sequence clustering, and barcode gap diagnostics are concordant, indicating high species-level resolution with COI-5′ for most taxa. The maximum intraspecific distance of <3% was achieved for 27/32 samples, and all nearest-neighbor distances exceeded this value, leading to the formation of coherent clusters and supporting the use of validated COI barcodes; this is consistent with reports of low within-species versus high between-species divergence in echinoderms (Table 2). The one remaining sequence is currently pending BIN assignment in BOLD’s next RESL run. The few edge cases in this study highlight familiar challenges in this field: deep intraspecific structure (e.g., Colochirus quadrangularis) and genus-only matches where public references are sparse (e.g., Clarkcomanthus sp.).

Key limitations include incomplete conspecific/congeneric coverage in BOLD/GenBank (e.g., Culcita novaeguineae, Clarkcomanthus sp.), elevated intraspecific K2P distances (>3%) in three taxa suggesting cryptic diversity, and use of a single mitochondrial marker. Future work should expand geographic and conspecific sampling, incorporate multilocus data (e.g., ITS, 28S), and integrate genomic approaches to resolve ambiguous lineages and strengthen regional barcode libraries.

5. Conclusions

This study establishes the first curated COI-5′ reference library for Vietnamese echinoderms. COI-5′ proved effective in facilitating the morphological analysis and revealed 16 species, including one taxon determined only at the genus level. For the majority of the investigated species, the results for the morphological identification and COI sequence clusters were congruent. All taxa achieved clear species-level resolution—monophyly, single-OTU cohesion, and barcode gaps—demonstrating the reliability of COI for routine identification in this fauna. This method strengthens regional baselines for taxonomy, monitoring, and conservation and can be extended to additional sites and taxa. Future work will prioritize expanding conspecific and geographic sampling to obtain standard, full-length COI sequences and add complementary markers for taxa exhibiting split clusters or exceeding the 3% threshold, thereby strengthening the COI reference library for echinoderms in Vietnamese waters and combining high-quality sequences, vouchers, and open deposits (BOLD/GenBank).