Phylogenetic Position of the Morphologically Ambiguous Genus Leiochrides (Annelida: Capitellidae) Revealed by Its First Complete Mitogenome

Abstract

The family Capitellidae performs critical roles in bioturbation and sediment remediation within global marine benthic ecosystems. However, they are a taxonomically challenging group due to their simple morphology and a ‘morphological mosaic’, where traditional classificatory traits, such as thoracic chaetiger counts, appear convergently across genera. Previous multi-locus studies (using 18S, 28S, H3, and COI) first highlighted this conflict, revealing the polyphyly of major genera like Notomastus and even Leiochrides itself (based on unidentified specimens). More recently, mitogenomic studies uncovered massive gene order rearrangements and a conflicting topology but did not include Leiochrides. Critically, with no complete mitogenome reported for a formally identified Leiochrides species, its true phylogenetic position and the validity of its polyphyly remain unresolved. To address this critical gap, we sequenced and characterized the first complete mitochondrial genome from a formally identified species, Leiochrides yokjidoensis, recently described from Korean waters. The complete mitogenome was 17,933 bp in length and included the typical 13 protein-coding genes (PCGs), 2 ribosomal RNAs (rRNAs), and 22 transfer RNAs (tRNAs). Gene order (GO) analysis revealed the occurrence of gene rearrangements in Capitellidae and in its sister clade, Opheliidae. A phylogenomic analysis using the amino acid sequences of 13 PCGs from 30 species established the first robust systematic position for the genus Leiochrides (based on this formally identified species). Phylogenetic results recovered Leiochrides as a sister group to the clade comprising Mediomastus, Barantolla, Heteromastus, and Notomastus hemipodus (BS 99%). This distinct placement confirms that Leiochrides represents an independent evolutionary lineage, phylogenetically separate from the polyphyletic Notomastus complex, despite their morphological similarities. Furthermore, our analysis confirmed the polyphyly of Notomastus, with N. hemipodus clustering distinctly from other Notomastus species. Additionally, signatures of positive selection were detected in ND4, and ND5 genes, suggesting potential adaptive evolution to the subtidal environment. This placement provides a critical, high-confidence anchor point for the genus Leiochrides. It provides a reliable reference to investigate the unresolved polyphyly suggested by previous multi-locus studies and provides compelling evidence for the hypothesis that thoracic chaetiger counts are of limited value for inferring phylogenetic relationships. This study provides the foundational genomic cornerstone for Leiochrides, representing an essential first step toward resolving the systematics of this taxonomically challenging family.

1. Introduction

Annelida, a diverse phylum of segmented worms, constitutes a critical component of global marine biodiversity and ecological functioning, particularly within soft-bottom benthic ecosystems [1]. The family Capitellidae Grube, 1862 [2], has a complex systematic history; it was once grouped with Clitellata due to similarities in external and internal (pharyngeal) morphology, followed by a long association with other sedentary polychaetes. The family currently comprises 42 genera and 224 recognized species [3] and is ecologically renowned. As key subsurface deposit-feeders, they ingest buried sediment particles, playing a foundational role in nutrient cycling [4]. Their continuous burrowing and ventilation activities (bioirrigation) are fundamental to benthic biogeochemical cycling, as they transport oxygen into anoxic sediment layers, stimulate microbial decomposition, and thus facilitate the remediation of organically enriched or polluted sediments [5,6]. Consequently, many capitellid species are recognized as opportunistic indicators, often dominating marine sediments following major disturbance events [7].

Despite this ecological ubiquity, Capitellidae remains one of the most taxonomically challenging groups in Polychaeta [7,8]. The family’s classification is notoriously fragmented; the fact that nearly half of its 42 described genera (approx. 24) are considered monospecific (containing only a single species) highlights the historical difficulty in establishing robust generic definitions [3,9]. This chaos stems from their “notoriously” simple body plan, which lacks complex appendages (like palps or elaborate parapodia) and exhibits a uniform, thread-like appearance [9,10,11]. Traditionally, generic-level classification has relied heavily on a limited suite of morphological characters: primarily the number of thoracic chaetigers and the distribution patterns of chaetae (i.e., capillaries vs. hooded hooks) [7,9,10,11].

This reliance on a limited set of characters has created a significant ‘morphological mosaic’, leading to profound taxonomic ambiguity. The systematic placement of the genus Leiochrides Augener, 1914 [12], exemplifies this challenge. Leiochrides is defined by possessing 12 thoracic chaetigers [9,13]. This count numerically separates it from numerous key genera such as Notomastus (11 chaetigers), Heteromastus (11 chaetigers), and Decamastus (10 chaetigers), as well as from Dasybranchus (13 chaetigers) and Leiocapitella (14 chaetigers) [9,13,14,15,16].

However, the phylogenetic value of this segment count is highly questionable, as the chaetal arrangement patterns appear to be homoplastic across these numerical boundaries. The generic diagnosis of Leiochrides itself has a convoluted history, reflecting this ambiguity. While historically defined by having all capillary chaetae [9,13], this definition was challenged by Hartman [17,18], who argued for the inclusion of species with “transitional” segments (e.g., presence of notopodial capillaries and neuropodial hooded hooks in the same segment) on the posterior-most chaetigers. This expanded definition [19,20] gained wide acceptance, particularly after the synonymization of Pseudomastus Capaccioni-Azzati & Martin, 1992 [21] (which was defined by this transitional trait) under Leiochrides by Jeong et al. [13]. This taxonomic decision confirmed that this trait is likely an intraspecific or intrageneric variation, not a genus-level diagnostic feature [13]. This exact ‘transitional’ pattern also appears convergently across multiple genera with different thoracic counts: for instance, Not omastus (11 chaetigers) includes species such as N. sunae and N. angelicae, which are characterized by possessing neuropodial hooded hooks on the last thoracic segment (TC11) [22,23]. Decamastus (10 chaetigers), such as in the type species D. gracilis, possesses hooks in the neuropodia of the last one or two thoracic segments (TC9–10) [20,24]; and Leiocapitella (12–17 chaetigers) is likewise diagnosed by a transitional final thoracic segment [9,25,26]. Furthermore, the genus Scyphoproctus—defined by its unique anal funnel, not its thorax [20]—exhibits extreme ambiguity. Its thoracic segment count is highly variable (reported as 9–14 chaetigers), and it contains species such as S. oculatus (mirroring the ‘all capillary’ Leiochrides pattern) as well as species like S. lumenalis (possessing the ‘transitional’ pattern) [19,27]. This morphological mosaic, where key chaetal patterns (e.g., ‘transitional segments’) converge across genera with 9–17 chaetigers, makes it impossible to validate these genera or infer their relationships based on morphology alone.

Previous molecular-phylogenetic studies initiated the effort to resolve this conflict between morphological definitions and evolutionary lineages. A foundational study by Tomioka et al. [8], focusing on Japanese taxa including key problematic genera like Notomastus and Heteromastus, provided the first major test of the traditional classification using concatenated data of four nuclear and mitochondrial gene fragments (18S rDNA, 28S rDNA, H3, and COI). While this study successfully confirmed the monophyly of Capitellidae (in contrast to [1]), its findings provided the first direct molecular evidence of the failure of the classical system. It demonstrated that traditional diagnostic characters—specifically, “head type; number of thoracic segments; number of segments with capillary chaetae”—were “not informative” for delineating natural groups. This was exemplified by their key finding that the genera Notomastus and Barantolla were not monophyletic, instead forming a mixed, well-supported clade. This result strongly suggests that the true phylogenetic relationships are more complex than the traditional generic definitions allow. Although this study was pivotal in establishing this problem awareness, most other nodes in their analysis remained weakly supported, highlighting that the limited number of genetic loci (4 partial genes) provided insufficient phylogenetic signal to fully resolve the family’s backbone.

More recently, the first mitogenomic study of the family by Su et al. [28] revealed even deeper complexities. This study, using 13 PCGs (and a larger concatenated dataset including nuclear 18S, 28S, and H3 genes) from 8 species, uncovered massive gene order (GO) rearrangements within the family, challenging the long-held view of conserved annelid mitogenomes. Their analysis proposed a new framework, splitting the family (excluding Capitella) into a “Conserved Group” (Mediomastus, Barantolla) and a “Rearranged Group” (Notomastus, Notodasus), with the latter even possessing a rare Group II intron. Critically, this new mitogenome-dominated phylogenomic topology revealed a significant conflict with the nuclear-dominated multi-locus topology of Tomioka et al. [8]. Specifically, the former (driven by 13 mitochondrial PCGs + 3 partial nuclear genes) recovered the Mediomastus group (Clade 1) as the basal lineage, whereas the latter topology (driven by 3 partial nuclear genes + 1 partial mitochondrial gene) had recovered Capitella (Clade B) as basal. This discordance, resulting from the different genomic data types and their respective weightings, suggests that the family’s evolution is fraught with complex genomic histories and potential analytical artifacts (like Long Branch Attraction).

This leads to the critical gap addressed by our study. While these pivotal molecular studies [8,28] did include genetic data from Leiochrides, the data itself was the source of ambiguity. Both datasets relied on three unidentified Leiochrides specimens (sp. 1 (Hiroshima, Japan), sp. 2 (Kanagawa, Japan), sp. 3 (Okinawa, Japan)), which were shown to be polyphyletic. Consequently, although the polyphyletic placement was confirmed, the systematic implications remained unresolved. Furthermore, publicly available mitochondrial DNA information for this genus remains limited to a single partial 529 bp COI sequence from another unverified Leiochrides sp1. To date, no complete mitochondrial genome has been reported for Leiochrides based on a formally identified species.

This significant data gap prevents a comprehensive understanding of genome evolution within the family. Therefore, this study presents the first complete mitochondrial genome for a formally identified species of the genus Leiochrides, using the recently described Korean species Leiochrides yokjidoensis Jeong, Wi & Suh, 2017 [13]. The primary aim of this study is to (1) characterize its complete genomic structure, gene content, and organization, and (2) utilize this new genomic data to conduct a robust phylogenomic analysis across 7 families to establish the systematic position of Leiochrides within the recently proposed mitogenomic framework; and (3) examine signatures of positive selection to elucidate the adaptive evolution of this species. This placement provides a critical, high-confidence anchor point for the genus, allowing for a re-evaluation of the polyphyly hinted at by previous studies. Although a definitive test of generic monophyly requires sequencing additional Leiochrides species, this study provides the foundational genomic cornerstone for the genus. It represents a critical first step, providing a reliable reference point to which future phylogenomic investigations into this taxonomically challenging family can be compared.

2. Materials and Methods

2.1. Sample Collection and Processing

Leiochrides yokjidoensis specimens were obtained from subtidal zones (26–60 m depth) across the southern coast of Korea, including Geomundo Island, Busan, and the outer main channel of Gwangyang Bay, from March 2024 to May 2025 (Type Locality: Yokjido Island) (Figure 1). Samples were obtained using a Smith–McIntyre grab sampler. The collected sediments were sieved through a 0.5 mm mesh, and specimens were sorted live on board. Specimens were relaxed using a 7.5% MgCl₂ solution to prevent contraction. For morphological examination, samples were fixed in 10% neutral buffered formalin and subsequently transferred to 80% ethanol. Specimens were examined and photographed with a Stemi 305 stereomicroscope (ZEISS, Oberkochen, Germany) and an Eclipse Ci-L compound microscope (Nikon, Tokyo, Japan), both equipped with digital cameras. Two staining methods were employed to visualize different morphological features. To observe the species-specific patterns of glandular tissues, specimens were stained with a methyl green solution. Specimens were briefly immersed in the staining solution (approx. 60 s) until a distinct color pattern emerged, then rinsed in 80% ethanol for differentiation. Separately, Shirlastain A solution (SDLATLAS, Rock Hill, SC, USA) was used to enhance the overall contrast of morphological structures and facilitate observation. Specimens were immersed for a short period (approx. 30 s) and subsequently rinsed. Staining results from both methods were documented through photography.

2.2. Species Identification Criteria

The specimen used for genomic sequencing was unequivocally identified as Leiochrides yokjidoensis Jeong, Wi & Suh, 2017 [13], based on the combination of the following key diagnostic characters, which distinguish it from other congeners (Figure 2):

• Chaetal Formula: Thoracic chaetigers 1–11 bear only capillary chaetae in both noto- and neuropodia. The final thoracic segment, TC12, is a transitional segment bearing capillary chaetae in the notopodium and hooded hooks in the neuropodium.
• Chaetiger 1: The first thoracic chaetiger (TC1) is uniramous, possessing only notopodial capillary chaetae (neuropodia are absent).
• Hook Dentition: Abdominal hooded hooks, examined under high magnification (1000×) with the compound microscope, showed a main fang surmounted by 8–10 small apical teeth.
• Branchiae: Branchiae present on posterior abdominal segments as 2–4 digitate filaments near notopodia. There are approximately ten preanal chaetigers without branchiae.
• MGSP: The staining pattern was consistent with the type description of L. yokjidoensis, showing strong staining on TC6–9 and distinct patterns on posterior segments.

A posterior portion of the specimen was subsampled for molecular analysis, while the anterior portion, including the thorax, was retained as a voucher specimen. All examined specimens, including newly collected material, are deposited at Chonnam National University Specimen Collection, South Korea (JUMA_20251211_001, JUMA_20250901_007, JUMA_20260106_001).

2.3. Mitogenome Sequencing and Reconstruction

Total genomic DNA was extracted using the QIAGEN Blood and Cell Culture DNA Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. A paired-end library (151 bp) was constructed using the TruSeq DNA Nano 550 Kit (Illumina, Inc., San Diego, CA, USA), and sequencing was performed on an Illumina Novaseq X Plus platform by JS Link Inc. (Seoul, Republic of Korea) (

Table 1. Summary of statistics for next-generation sequencing data and the mitochondrial genome assembly workflow.

Leiochrides yokjidoensis
Sequencing	Sequencing System	Illumina Novaseq X Plus
	Library Construction	TruSeq DNA Nano
	Sequencing layout (bp)	151 × 2
	Fragment size (bp)	550
Raw Data	Raw reads	115,488,548
	Raw bases (bp)	17,438,770,748
	Q30 (%)	94.90
Filtered Data	Clean reads	115,488,548
	Clean bases (bp)	16,697,248,234
	Total length (bp)	17,933
Mitochondrial genome assembly	GC content (%)	45.08
	Number of protein-coding genes	13

).

Before assembly, raw reads were processed with Trim_Galore (ver. 0.6.10) [29] to eliminate adapters and low-quality bases (Phred score < 30). The de novo assembly of the Leiochrides yokjidoensis mitogenome was carried out using MitoZ (ver. 3.6) [30] based on the clean reads. Subsequent gene annotation was performed using MITOS2 (ver. 2.1.9) [31] and ARWEN (ver. 1.2.3) [32]. Finally, the circular map of the assembled genome was generated using Circos (ver. 0.69-8) [33,34,35].

2.4. Phylogenomic Reconstruction

To investigate phylogenetic relationships within the subclass Sedentaria, we performed a phylogenomic analysis based on the amino acid sequences of 13 mitochondrial protein-coding genes (PCGs). The dataset comprised 30 species, including Leiochrides yokjidoensis, with two species from the subclass Errantia used as outgroups. Publicly available mitogenome sequences were retrieved from the NCBI database; for entries lacking annotation, the 13 PCGs were identified using MitoZ (ver. 3.6) [30] and MITOS2 (ver. 2.1.9) [31]. Sequence alignment for each PCG was conducted using MAFFT (ver. 7.525) [36] and PAL2NAL (ver. 14.1) [37].

To reduce the potential effects of long-branch attraction, only the first and second codon positions of the 13 PCGs were retained for subsequent phylogenetic analyses. Optimal evolutionary models were selected based on the Bayesian Information Criterion (BIC) using PartitionFinder2 (ver. 2.1.1) [38]. The dataset was partitioned into three subsets, and the GTR + I + G model (General Time-Reversible model with Invariant sites and Gamma-distributed rate variation) was selected as the best-fit model for all subsets. Phylogenetic reconstruction was performed using Maximum Likelihood (ML) and Bayesian Inference (BI) approaches. The ML analysis was executed in RAxML-NG (ver. 1.2.2) [39] with 1000 bootstrap replicates to evaluate nodal support. BI analysis was conducted using MrBayes (ver. 3.2.7) [40]. We ran two independent Markov Chain Monte Carlo (MCMC) chains for 1 × 10⁷ generations, sampling every 1000 generations with a chain temperature of 0.02. Convergence was confirmed when the average standard deviation of split frequencies dropped below 0.01. After discarding the first 25% of trees as burn-in, a majority-rule consensus tree was computed. The final tree topologies were visualized using FigTree (ver. 1.4.4) [41].

2.5. Tests for Positive Selection

To detect genomic signatures of adaptive evolution in response to environmental pressures, we identified positively selected genes (PSGs) among the mitochondrial PCGs using the CodeML program in PAML [42]. The input topology was the unrooted ML tree derived from the 13 PCGs with branch lengths removed. Nucleotide sequences were codon-aligned using PAL2NAL (ver. 14.1) [37]. We applied both the site model and the branch-site model for selection analysis. We analyzed site-specific selection using site models (model = 0, NSsites = 0 1 2 3 7 8, fix_omega = 0). We then performed the branch-site test by comparing a null model (model = 2, NSsites = 2, fix_omega = 1) against an alternative model (model = 2, NSsites = 2, fix_omega = 0). The site model allows the nonsynonymous-to-synonymous substitution ratio (ω = dN/dS) to vary across codon sites, whereas the branch-site model allows ω to diverge in the foreground branch relative to the background branches. Statistical significance was assessed using Likelihood Ratio Tests (LRTs) approximating a chi-square (${\chi }^2$) distribution. Sites were considered to be under positive selection if the LRT was significant (p < 0.05) and the calculated ω exceeded 1.

3. Results and Discussion

3.1. General Features of the Mitochondrial Genome

We successfully assembled the complete circular mitogenome of Leiochrides yokjidoensis (17,933 bp; GenBank Accession No. PX614612) using high-quality sequencing data. The assembly was based on 115,488,548 paired-end reads generated via the Illumina platform, from which 16.7 Gb of clean bases were retained (Q30 ≥ 94.90%) after rigorous filtering (Table 1).

The genome architecture comprises the typical metazoan set of 37 genes—13 protein-coding genes (PCGs), 2 ribosomal RNAs (12S and 16S), and 22 transfer RNAs—all of which are encoded on the heavy strand (H-strand) (

Table 2. Gene organization and annotated features of the Leiochrides yokjidoensis mitochondrial genome.

Gene	Position	Length (bp)	Initiation Codon	Stop Codon	Anti- Codon	Strand
tRNA-Trp (trnW)	617–680	65			TCA	+
NADH dehydrogenase subunit 3 (ND3)	685–1032	349	ATG	TAG		+
NADH dehydrogenase subunit 4L (ND4L)	1041–1335	295	ATG	TAG		+
tRNA-Tyr (trnY)	1336–1400	66			GTA	+
ATP synthase F0 subunit 6 (ATP6)	1401–2045	645	ATG	TAA		+
tRNA-Ser (trnS)	2105–2171	68			TGA	+
tRNA-Val (trnV)	2171–2235	66			TAC	+
tRNA-Thr (trnT)	2236–2298	64			TGT	+
NADH dehydrogenase subunit 2 (ND2)	2303–3265	963	ATG	TAA		+
Cytochrome c oxidase subunit III (COX3)	3302–4093	792	GTG	TAG		+
tRNA-Leu (trnL)	4208–4269	63			TAG	+
NADH dehydrogenase subunit 1 (ND1)	4270–5208	940	GTG	TAA		+
NADH dehydrogenase subunit 6 (ND6)	5209–5691	484	ATG	TAA		+
Cytochrome b (COB)	5765–6874	1111	ATG	TAA		+
NADH dehydrogenase subunit 5 (ND5)	8547–10,277	1732	GTG	TAA		+
NADH dehydrogenase subunit 4 (ND4)	10280–11,572	1294	ATG	TAG		+
tRNA-Asn (trnN)	11,574–11,636	64			GTT	+
tRNA-Met (trnM)	11,637–11,700	65			CAT	+
tRNA-His (trnH)	11,733–11,794	63			GTG	+
tRNA-Leu (trnL)	11,794–11,856	64			TAA	+
tRNA-Ile (trnI)	11,858–11,921	65			GAT	+
tRNA-Arg (trnR)	11,923–11,986	65			TCG	+
tRNA-Phe (trnF)	11,987–12,051	66			GAA	+
tRNA-Glu (trnE)	12,284–12,348	66			TCC	+
12S ribosomal RNA (s-rRNA)	12,349–13,186	839				+
tRNA-Gln (trnQ)	13,184–13,251	69			TGG	+
16S ribosomal RNA (l-rRNA)	13,250–14,494	1246				+
tRNA-Ala (trnA)	14,470–14,533	65			TGV	+
tRNA-Asp (trnD)	14,535–14,598	65			GTC	+
ATP synthase F0 subunit 8 (ATP8)	14,599–14,760	163	ATG	TAA		+
tRNA-Pro (trnP)	15,017–15,081	66			TGG	+
tRNA-Lys (trnK)	15,927–15,994	69			TTT	+
tRNA-Ser (trnS)	16,013–16,077	65			TCT
tRNA-Cys (trnC)	16,104–16,168	66			GCA	+
tRNA-Gly (trnG)	16,171–16,235	66			TCC	+
Cytochrome c oxidase subunit II (COX2)	16,236–16,946	712	ATG	TAG		+
Cytochrome c oxidase subunit I (COX1)	17,004-618	1548	ATA	TAG		+

and Figure 3). Coding regions account for 81.10% of the entire genome, with PCGs alone constituting 61.50%.

In terms of nucleotide composition, the genome exhibits a distinct A + T bias (54.92% overall), with base proportions of 19.15% A, 35.77% T, 27.89% G, and 17.19% C. Analysis of the 13 PCGs revealed that while the canonical ATG start codon is used in nine genes, COX3, ND1, and ND5 utilize GTG, and COX1 initiates with ATA. Termination codons consist of either TAA (eight genes) or TAG (five genes) (Table 2).

3.2. Phylogeny and Synteny

To overcome the limitations of morphological identification—specifically the scarcity of type specimens and ambiguous original descriptions—we employed a phylogenomic approach to resolve the systematic position of Leiochrides yokjidoensis. While morphology alone often yields uncertain taxonomic results, genomic data provide a robust framework for addressing these complexities. We constructed a Maximum Likelihood (ML) tree based on the amino acid sequences of 13 mitochondrial protein-coding genes (PCGs) from 26 representative species of the subclass Sedentaria (

Table 3. List of the 26 mitochondrial genomes utilized in this phylogenomic analysis, including taxonomic details and GenBank accession numbers.

Family	Genus	Species	Accession Number	Type Locality	Sampling Locality	Reference
Capitellidae	Leiochrides	Leiochrides yokjidoensis	PX614612	Yokjido, Korea	Geomundo, Korea	This study
	Notomastus	Notomastus sp.	LC661358.1	-	Wakayama, Japan	Kobayashi et al. [45]
		Notomastus sp. A	PP133661.1	-	Hong Kong, China	Su et al. [28]
		Notomastus sp. B	PQ010758.1	-	Hainan, China	Su et al. [28]
		Notomastus hemipodus	PV742383.1	North Carolina, USA	Vancouver, Canada	Acharya-Patel et al. [46]
	Notodasus	Notodasus sp. A	PP133663.1	-	Hong Kong, China	Su et al. [28]
		Notodasus sp. B	PQ010757.1	-	Hainan, China	Su et al. [28]
		Notodasus sp. C	PP133662.1	-	Fujian, China	Su et al. [28]
	Capitella	Capitella teleta	PP133665.1	Massachusetts, USA	Shandong, China	Su et al. [28]
		Capitella capitata	PV742352.1	Greenland	Vancouver, Canada	Acharya-Patel et al. [46]
	Heteromastus	Heteromastus filobranchus	PV742385.1	Vancouver, Canada	Canada	Acharya-Patel et al. [46]
	Mediomastus	Mediomastus sp.	PP133664.1	-	Guangdong, China	Su et al. [28]
	Barantolla	Barantolla sp.	PQ010756.1	-	Hainan, China	Su et al. [28]
Arenicolidae	Abarenicola	Abarenicola claparedi	LC707921.1	Mediterranean Sea	Hokkaido, Japan	Kobayashi et al. [47]
Maldanidae	Asychis	Asychis amphiglyptus	NC_069297.1	South Georgia, UK	-	-
	Clymenella	Clymenella koellikeri	PQ593498.1	Fiji, EEZ	-	-
		Clymenella torquata	NC_006321.1	New Jersey, USA	Massachusetts, USA	Jennings and Halanych [48]
	Euclymene	Euclymene annandalei	OM273170.1	India	-	-
	Lumbriclymenella	Lumbriclymenella robusta	OP537514.1	South Georgia, UK	-	-
	Maldane	Maldane sarsi	OP694172.1	Sweden	-	-
	Metasychis	Metasychis gotoi	OP605751.1	Japan, EEZ	-	-
	Nicomache	Nicomache sp.	PQ593499.1	-	-	-
	Petaloproctus	Petaloproctus sp.	PQ593501.1	-	-	-
	Praxillella	Praxillella affinis	PQ593500.1	Norway, EEZ	-	-
	Sabaco	Sabaco sinicus	PQ741859.1	China	-	-
Cossuridae	Cossura	Cossura aciculata	PV151471.1	China	China	-
Opheliidae	Armandia	Armandia sp.	LC661359.1	-	Wakayama, Japan	Kobayashi et al. [45]
Orbiniidae	Orbinia	Orbinia latreillii	NC_007933.1	La Rochelle, France	Bretagne, France	Bleidorn et al. [49]
Phyllodocidae	Phyllodoce	Phyllodoce medipapillata	NC_087881.1	USA	California, USA	Huč et al. [50]
		Phyllodoce koreana	PQ510072.1	Korea	Korea	Kim et al. [51]

). This multi-locus dataset offers significantly higher resolution for phylogenetic inference than single-gene markers [43,44].

The resulting phylogeny strongly supported the monophyly of the family Capitellidae (BS = 100%, PP = 1.00). Within this clade, L. yokjidoensis was recovered as the sister group to the clade comprising Mediomastus, Barantolla, Heteromastus, and Notomastus hemipodus (BS = 99%), a placement that diverges from traditional morphological classifications. Additionally, Capitellidae formed a robust sister-group relationship with Opheliidae (BS = 99%), suggesting a shared evolutionary history. This affinity is further reinforced by the distinct patterns of mitochondrial gene rearrangement observed in both families (Figure 4).

To complement these phylogenetic findings, we performed a comparative analysis of mitochondrial gene synteny across the 30 polychaete species (Figure 4, right). Our analysis revealed a striking correlation between phylogenetic branch lengths (evolutionary rate) and genomic structural stability. While the outgroup and most other sedentarian families exhibited a conserved ancestral gene arrangement, Capitellidae showed three distinct evolutionary modes correlated with their substitution rates:

• Conserved Group (Short Branches): The clade comprising Mediomastus, Barantolla, Heteromastus, and Notomastus hemipodus exhibited relatively short branch lengths, indicating a slower rate of molecular evolution. Correspondingly, these species shared an identical and conserved gene order, suggesting high genomic stability in this lineage.
• Hyper-variable Group (Intermediate/Long Branches): The “Rearranged Group” (Notomastus sp., Notomastus sp. A, Notomastus sp. B, Notodasus sp. A, Notodasus sp. B and Notodasus sp. C) displayed longer branch lengths. Remarkably, all four examined Notomastus species (N. hemipodus, Notomastus sp., Notomastus sp. A, and Notomastus sp. B) exhibited different gene arrangements from each other. This extreme variability within a single nominal genus, coupled with their polyphyletic placement, indicates a “hotspot” of genomic instability and further supports the need for taxonomic revision of Notomastus.
• Divergent Group (Longest Branches): The distinct lineages of Leiochrides yokjidoensis and the genus Capitella exhibited the longest branch lengths in the tree, indicative of rapid accelerated evolution. Consistent with this high substitution rate, both genera exhibit unique and heavily rearranged mitochondrial gene orders relative to other groups; the two Capitella species share the same mitochondrial gene order, while Leiochrides differs markedly from Capitella, consistent with their distant phylogenetic positions.

Although limited by the current taxon sampling, these observations suggest a potential link within Capitellidae where lineages undergoing rapid sequence evolution (long branches) are also prone to extensive structural rearrangements. The unique gene order and rapid evolutionary rate of Leiochrides further highlight its distinct evolutionary trajectory, distinguishing it from the polyphyletic Notomastus complex and other related genera, despite their morphological resemblances.

3.3. Positive Selection Analysis

Although mitochondrial protein-coding genes (PCGs) are primarily constrained by purifying selection to maintain the functional integrity of oxidative phosphorylation, they can also exhibit signatures of adaptive evolution. To investigate these potential environmental adaptations, we evaluated selective pressures by calculating the ratio of non-synonymous to synonymous substitution rates (ω = dN/dS) across all 13 PCGs (

Table 4. Statistical outcomes of the positive selection analysis based on site models. Abbreviations: np, number of parameters; LRT, likelihood ratio test statistic.

Gene	M0 (37)	M1a (38)	M2a (40)	M3 (41)	M7 (38)	M8 (40)	Model Comparison (np)	LRTs
ATP6	−18,216.17475	−18,086.11357	−18,086.11357	−17,551.54513	−17,530.66051	−17,530.66246	M0 vs. M3 (4)	1329.2592
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.003888
ATP8	−4915.891856	−4815.027421	−4814.851827	−4713.267295	−4709.704374	−4709.70452	M0 vs. M3 (4)	405.24912
							M1a vs. M2a (2)	0.351188
							M7 vs. M8 (2)	0.000292
COX1	−26,830.58115	−26,700.24829	−26,700.24829	−25,753.3935	−25,737.72334	−25,736.49474	M0 vs. M3 (4)	2154.3753
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	2.457196
COX2	−14,386.37689	−14,344.30227	−14,344.30227	−13,834.47394	−13,833.0083	−13,833.00895	M0 vs. M3 (4)	1103.8059
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.001298
COX3	−14,686.20445	−14,564.92926	−14,564.92931	−14,027.43752	−14,010.6313	−14,010.63354	M0 vs. M3 (4)	1317.5339
							M1a vs. M2a (2)	0.000104
							M7 vs. M8 (2)	0.004478
COB	−23,974.19042	−23,798.6266	−23,798.6266	−22,897.66727	−22,864.7537	−22,864.75552	M0 vs. M3 (4)	2153.0463
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.003644
ND1	−21,684.92572	−21,371.14525	−21,371.14525	−20,589.39337	−20,552.08262	−20,552.02358	M0 vs. M3 (4)	2191.0647
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.11807
ND2	−28,666.00705	−28,409.10097	−28,409.10097	−27,794.59458	−27,760.01575	−27,760.0163	M0 vs. M3 (4)	1742.8249
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.001096
ND3	−8783.448956	−8603.506367	−8603.506367	−8284.0882	−8287.373649	−8287.374643	M0 vs. M3 (4)	998.72151
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.001988
ND4	−34,592.0451	−34,232.74772	−34,232.74772	−33,072.81786	−33,028.81993	−33,028.79363	M0 vs. M3 (4)	3038.4545
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.05259
ND4L	−8178.441719	−8140.187983	−8140.187983	−7922.090965	−7912.282848	−7908.560915	M0 vs. M3 (4)	512.70151
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	7.443866
ND5	−44,566.23379	−44,023.18734	−44,023.18734	−42,438.75504	−42,334.82236	−42,332.05566	M0 vs. M3 (4)	4254.9575
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	5.533394
ND6	−13,442.30588	−13,262.91993	−13,262.91993	−12,929.87655	−12,907.36782	−12,907.07133	M0 vs. M3 (4)	1024.8587
							M1a vs. M2a (2)	0
							M7 vs. M8 (2)	0.592974

Note. The critical values are ${\chi }_{2.5\%}^2$ = 5.99 (i.e., 2 degrees of freedom at 5% significance), ${\chi }_{2.1\%}^2$ = 9.21 (i.e., 2 degrees of freedom at 1% significance), ${\chi }_{4.5\%}^2$ = 9.48 (i.e., 4 degrees of freedom at 5% significance), and ${\chi }_{4.1\%}^2$ = 13.27 (i.e., 4 degrees of freedom at 1% significance). The LRT statistic, 2Δℓ, is reported for all model comparisons.

and

Table 5. Statistical outcomes of the positive selection analysis based on the branch-site model. Abbreviations: np, number of parameters; LRT, likelihood ratio test statistic.

Gene Name	Null Model (np)	Alternative Model (np)	LRTs (p-Value)	Site Class	0	1	2a	2b
ATP6	−17,913.70553 (64)	−17,912.76472 (65)	1.881618 (0.17)	Proportion	0.51033	0.20768	0.20043	0.08156
				Background ω	0.09429	1	0.09429	1
				Foreground ω	0.09429	1	45.4215	45.4215
ATP8	−4672.201808 (64)	−4672.201805 (65)	0.000006 (0.998)	Proportion	0.21817	0.78183	0	0
				Background ω	0.086	1	0.086	1
				Foreground ω	0.086	1	1	1
COX1	−26,375.29311 (64)	−26,375.29226 (65)	0.0017 (0.967)	Proportion	0.8193	0.022	0.15455	0.00415
				Background ω	0.01875	1	0.01875	1
				Foreground ω	0.01875	1	1.01768	1.01768
COX2	−14,280.19673 (64)	−14,280.19673 (65)	0(1)	Proportion	0.60394	0.02154	0.36162	0.0129
				Background ω	0.03379	1	0.03379	1
				Foreground ω	0.03379	1	1	1
COX3	−14,342.36051 (64)	−14,342.36051 (65)	0(1)	Proportion	0.7732	0.05583	0.15946	0.01151
				Background ω	0.03566	1	0.03566	1
				Foreground ω	0.03566	1	1	1
CYTB	−23,569.97597 (64)	−23,568.82145 (65)	2.30904 (0.128)	Proportion	0.6602	0.12651	0.17899	0.0343
				Background ω	0.05486	1	0.05486	1
				Foreground ω	0.05486	1	1.99219	1.99219
ND1	−21,178.57394 (64)	−21,178.44627 (65)	0.25534 (0.613)	Proportion	0.7132	0.16247	0.10126	0.02307
				Background ω	0.06267	1	0.06267	1
				Foreground ω	0.06267	1	1.3493	1.3493
ND2	−27,990.92862 (64)	−27,990.92862 (65)	0(1)	Proportion	0.61005	0.38995	0	0
				Background ω	0.15035	1	0.15035	1
				Foreground ω	0.15035	1	1	1
ND3	−8436.363774 (64)	−8436.363774 (65)	0(1)	Proportion	0.51587	0.48413	0	0
				Background ω	0.06222	1	0.06222	1
				Foreground ω	0.06222	1	2.98105	2.98105
ND4	−33,740.52873 (64)	−33,737.52529 (65)	6.00689 (0.14)	Proportion	0.51219	0.35569	0.07797	0.05415
				Background ω	0.10263	1	0.10263	1
				Foreground ω	0.10263	1	54.54094	54.54094
ND4L	−7938.586789 (64)	−7938.586789 (65)	0(1)	Proportion	0.51716	0.12794	0.28452	0.07039
				Background ω	0.11906	1	0.11906	1
				Foreground ω	0.11906	1	1	1
ND5	−43,443.13895 (64)	−43,439.61876 (65)	7.040374 (0.007)	Proportion	0.48923	0.31608	0.11827	0.07641
				Background ω	0.09697	1	0.09697	1
				Foreground ω	0.09697	1	80.31	80.31
ND6	−13,066.49994 (64)	−13,069.9364 (65)	6.872916 (0.008)	Proportion	0.42345	0.57655	0	0
				Background ω	0.13626	1	0.13626	1
				Foreground ω	0.13626	1	18.70366	18.70366

). The Site model analysis generally indicated purifying selection across the phylogeny. Although the likelihood ratio test comparing models M7 and M8 initially detected signals in both ND4L and ND5 (Table 4), examination of parameter estimates revealed an ω value of 1 for ND4L (

Table 6. Parameter Estimates and Positively Selected Sites Identified Using the BEB Method for the Site Models (Model 8).

Gene	Estimates of Parameters	Positively Selected Sites	Pr (w > 1)	Post Mean +− SE for w
ND4L	p0 = 0.98871 (p1 = 0.01129) p = 0.92443 q = 15.49543 w = 1.00000	-	-	-
ND5	p0 = 0.99558	190 -	0.556	2.865 ± 3.120
	(p1 = 0.00442)	469 F	0.577	2.806 ± 2.970
	p = 0.56877	487 I	0.694	3.343 ± 3.087
	q = 9.41966 w = 1.51283	491 -	0.656	3.245 ± 3.161

) [52]. Therefore, ND4L was under neutral evolution rather than positive selection. More importantly, the Branch-site model test, which focused specifically on the lineage leading to Leiochrides yokjidoensis, revealed stronger and statistically significant signatures of positive selection in three genes: ND4, ND5, and ND6 (Table 5). However, ND6 was excluded from the final candidates because the estimated proportion of sites under positive selection was zero (p₂ = 0) suggesting the absence of true adaptive signals. Specific amino acid sites under positive selection were identified via Bayes Empirical Bayes (BEB) analysis (Table 6 and

Table 7. Positively Selected Sites Identified Using the BEB Method for the Branch-site Model.

Gene	Positively Selected Site (BEB)
ND4	12 A (0.718), 76 S (0.632), 94 A (0.669), 142 A (0.794), 175 S (0.657), 179 C (0.591), 208 A (0.888), 252 T (0.686), 255 T (0.525), 275 A (0.635), 289 S (0.584), 296 T (0.555), 321 V (0.885), 324 S (0.648), 345 A (0.607), 362 G (0.683), 372 A (0.774), 386 S (0.737), 389 C (0.636), 401 V (0.586), 421 S (0.696), 442 S (0.630), 465 L (0.601)
ND5	16 S (0.506), 21 L (0.633), 23 M (0.521), 65 W (0.926), 79 G (0.649), 83 C (0.790), 87 S (0.833), 101 V (0.807), 118 L (0.917), 149 S (0.677), 151 S (0.792), 158 A (0.778), 173 A (0.804), 181 H (0.787), 189 G (0.956), 209 A (0.759), 219 A (0.517), 251 V (0.554), 266 T (0.792), 275 A (0.875), 286 A (0.675), 330 T (0.769), 338 S (0.701), 347 W (0.826), 349 G (0.674), 354 Y (0.726), 358 W (0.727), 363 A (0.848), 369 M (0.928), 382 M (0.792), 388 H (0.714), 393 V (0.887), 419 T (0.877), 430 Q (0.600), 432 G (0.842), 433 S (0.908), 435 S (0.790), 437 A (0.535), 452 M (0.793), 463 L (0.534), 467 V (0.849), 519 T (0.856), 520 Q (0.516), 526 F (0.781), 537 L (0.610), 547 C (0.737), 551 C (0.889), 561 A (0.749), 576 V (0.913), 584 F (0.841), 586 T (0.529)

). These results suggest that, while the mitochondrial genome is largely conserved, specific genes in L. yokjidoensis have undergone adaptive evolution, potentially facilitating metabolic adjustments to its subtidal benthic environment (26–60 m depth).

4. Conclusions

This study presents the first complete mitochondrial genome of the Korean endemic polychaete Leiochrides yokjidoensis, a species belonging to the taxonomically challenging family Capitellidae. Our phylogenomic analysis, utilizing 13 mitochondrial PCGs from 30 species, established the systematic position of Leiochrides for the first time. Unexpectedly, Leiochrides formed a sister group with clade comprising Mediomastus, Barantolla, Heteromastus, and Notomastus hemipodus, forming a distinct clade characterized by long branch lengths and unique gene rearrangements.

Our findings revealed a dynamic pattern of mitochondrial genome evolution within Capitellidae, linking evolutionary rates with genomic stability. We identified three distinct evolutionary modes: (1) a Conserved Group (including Heteromastus and Notomastus hemipodus) with stable gene orders; (2) a Hyper-variable Group (including Notodasus, other Notomastus species) showing extreme structural instability; and (3) a Divergent Group (Leiochrides and Capitella) exhibiting rapid evolution and unique gene orders. These results confirm the polyphyly of Notomastus and demonstrate that Leiochrides represents a distinct evolutionary lineage, phylogenetically distinct from the polyphyletic Notomastus complex, despite morphological similarities. Furthermore, the apparent similarity to Capitella, characterized by long branch lengths and positive selection signatures, suggests that Leiochrides may share a similar trajectory of rapid evolutionary adaptation to environmental pressures, akin to the opportunistic nature of Capitella. Additionally, the detection of positive selection in ND4 and ND5 genes suggests that L. yokjidoensis has likely undergone adaptive evolution to meet the metabolic demands of its specific subtidal benthic habitat.

By providing the first genomic reference for a formally identified Leiochrides species, this study serves as a foundational genomic cornerstone. It offers a high-confidence anchor point for resolving the “morphological mosaic” of Capitellidae and paves the way for future studies to clarify the complex systematics of this family using a phylogenomic approach.