Our teams sampled Trichiurus specimens from the China Seas, including the Yellow Sea, East China Sea, and South China Sea in October 2017 and August 2019 by longline, gill net, and trawl net with fishermen (Fig. 1A, Suppl. material 1: Table S1). In total, 1.311 specimens were collected. Traditional caliper measurements were performed, which identified 14 landmarks (a–n, Fig. 2).

Figure 1. 

A Eighteen sampling localities of the genus the Trichiurus along the Chinese coast and the species composition after our surveys. Refer to Suppl. material 1: Table S1 for the abbreviations of localities. B The maximum-likelihood (ML) tree of these four Trichiurus species along the coast based on the COI gene. The numbers at the nodes are bootstrap values of the ML and NJ (neighbor-joining) analyses. The sampling size (n) indicated in parentheses C The photographs of four Trichiurus species used in the mitogenomes analyses.

Figure 2. 

Positions of 14 (a–n) landmarks used to contrast the morphological differences between Trichiurus species.

A portion of the muscle tissues from 1.311 specimens was stored in 100% ethanol. Total genomic DNA was extracted from muscle tissue using a Genomic DNA Purification Kit (Gentra Systems, Valencia, CA). The COI gene was amplified by polymerase chain reaction (PCR) using the primers Fish-F2 (5´-ACCTCTGTGTGTGGGGCTAC-3´) and Fish-R2 (5´-GTGATGCATTGGCTTGAAA-3´) (Gu et al. 2021). Each 50-µl PCR mixture contained 5 ng of template DNA, 5 µl of 10× reaction buffer, 4 µl of dNTP mix (10 mM), 5 pmol of each primer and 2 U of Taq polymerase (TaKaRa, Taq polymerase). PCR was conducted on an MJ Thermal Cycler using the following cycling parameters: one cycle of denaturation at 94 °C for 3 min, 40 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for 1 min and 30 s, followed by a 72 °C extension for 10 min and storage at 4 °C. The purified PCR products were sequenced using an ABI 377 automated sequencer (Applied Biosystems, Foster City, CA, U.S.A.). The resulting chromatograms were assessed using CHROMAS software (Technelysium), and the sequences were manually edited using BIOEDIT 6.0.7 (Hall 1999). In totally, 1.311 sequences were obtained, and the haplotypes were deposited in GenBank under accessions MZ959870 - MZ959999, MZ960057-MZ960127, OK053821 - OK054341 and OL539388-OL539398. The nucleotide sequences were aligned in Clustal X 1.81 (Thompson et al. 1997). Selection of the best-fit nucleotide substitution models was performed using the Bayesian information criterion (BIC) in jModelTest 2.0 (Darriba et al. 2012). The most appropriate nucleotide substitution model was GTR+I+G for COI. Maximum likelihood (ML) and neighbor-joining (NJ) phylogenetic analysis were performed with MEGA-X (Kumar et al. 2018). Bootstrapping was implemented with 1000 replications. In addition, Shen et al. (2016) proposed that the use of the K2P (Kimura’s two-parameter) distance in barcode analyses has been challenged and the p-distance has been proposed to be a better model. Thus, the p-distances between Trichiurus species were estimated in MEGA-X.

Next-generation sequencing (NGS) was performed to obtain complete mitogenome sequences. Complete mitogenomes were obtained from high-throughput sequencing with a HiSeqX Ten platform (Illumina, San Diego, CA) with a paired-end, 150-bp approach. All the reads were mapped to the full mitogenome reference sequences of other Trichiurus species (Table 1) using SOAPdenovo v.2.04 (https://github.com/aquaskline/SOAPdenovo2). The remaining high-quality reads were assembled using SPAdes v3.10 (https://github.com/ablab/spades). Compared with the corresponding complete mitogenome sequences of the genus Trichiurus (Liu and Cui 2009; Liu et al. 2013; Xu et al. 2019; Zheng et al. 2019; Mukundan et al. 2020; Table 1), protein-coding genes (PCGs), tRNA-coding genes and ribosome-coding genes were identified by BLAST. Codon usage, nucleotide substitution and base composition were determined using MEGA-X and DnaSP version 5.10 (Librado and Rozas 2009), and the rules for the vertebrate mitochondrial genetic code was used. AT skewing and GC skewing of the nucleotide composition were measured according to the following formulae: AT skew = (A – T)/(A + T) and GC skew = (G – C)/(G + C) (Perna and Kocher 1995).

The relative synonymous codon usage (RSCU), nonsynonymous codon usage (Ka) and synonymous codon usage (Ks) of all PCGs were analyzed using DnaSP. Comparison of the rates of Ka/Ks provides insight into changes in selective pressure: Ka/Ks values > 1 indicate positive selection; Ka/Ks = 1 indicates neutral selection; and Ka/Ks < 1 indicates negative or purifying selection. Some mitogenomes of the family Trichiuridae were downloaded from GenBank (NCBI database, Table 1). The most appropriate nucleotide substitution model was GTR+I+G for the mitogenome. The ML and NJ phylogenetic analysis were performed with MEGA-X. Bootstrapping was implemented with 1000 replications. The p-distances between Trichiurus species (interspecific) and between genera within Trichiuridae (intergeneric) were estimated in MEGA-X.

Measurements were referred to the truss network (Humphries et al. 1981) and some additional landmarks, forming 19 distances from 14 landmarks (Fig. 2). The morphometric characteristics were measured to nearest 0.1 and 0.01 cm using traditional calipers. In total, 225 specimens from South China Sea were measured. Values of the distances between landmarks were measured, and their means and standard deviations (S.D.) were calculated.

A total of 1.311 specimens were collected. Species were first identified by morphology. Two species groups were recognized, the T. lepturus complex, which has the anal opening positioned vertically at the 38^th–41^st dorsal fin rays, and the T. russelli complex, which has the anal opening positioned vertically at the 34^th and 35^th dorsal fin rays (Burhanuddin et al. 2002). Within the T. lepturus complex, T. japonicus has a longer tail, and T. lepturus has a whitish dorsal fin when fresh; by contrast, T. nanhaiensis has a yellowish green dorsal fin (Hsu et al. 2009). Besides, our study found that from the front view of the heads preserved specimens, the frontal bone of T. nanhaiensis is very smooth (Suppl. material 1: Fig. S1A), the frontal bone of T. japonicus is slightly inverted (Suppl. material 1: Fig. S1B), and the frontal bone of T. lepturus is obviously inverted and bulges in the upper part of the orbit and is accompanied by an indentation (Suppl. material 1: Fig. S1C). Four species belonging to the two species complexes were collected. We used COI sequences to identify species (Hebert et al. 2003). Our study sequenced complete COI gene (1551 bp) in all specimens. The phylogenetic trees reconstructed within ML and NJ were identical. In the ML tree (Fig. 1B), all specimens were grouped into four lineages with strong bootstrap support. After BLAST, we ensured these four lineages corresponded to four Trichiurus species: T. japonicus (n = 958), T. lepturus (n = 27), T. nanhaiensis (n = 120) and T. brevis (n = 206).

Trichiurus japonicas is distributed in the China Sea; T. lepturus, T. nanhaiensis, and T. brevis are distributed in the South China Sea. The results from the morphological and molecular data were the same. However, our study revealed that T. lepturus is very rare in the South China Sea (Fig. 1, Suppl. material 1: Table S1). Additionally, the results showed that T. lepturus complex was not a monophyletic group because T. brevis, belonging to T. russelli complex, was nested with T. nanhaiensis. Our study considers that this is because information is lacking.

After identifying species by morphology and DNA barcoding, the complete mitochondrial genomes of four Trichiurus species were sequenced (Fig. 1C). These four specimens were fixed in 10% formalin, transferred to 70% ethanol, and deposited in the Guangdong Ocean University, Zhanjiang, China as voucher specimens (GOU101614, GOU101617, GOU102735, and TLYJ041). The lengths of the complete mitogenomes of T. japonicus (MW719077), T. lepturus (MW719078), T. nanhaiensis (MW719076), and T. brevis (MW694877) were 16.685 bp, 16.568 bp, 16.801 bp, and 16.733 bp, respectively. To confirm the taxonomy of Trichiurus species, the phylogeny of Trichiuridae was analyzed using mitogenome sequences (Fig. 3, Table 1). The phylogenetic trees reconstructed within ML and NJ were identical. In ML tree (Fig. 3), the sequences of the genus Trichiurus were grouped into five lineages (I–V). Trichiurus haumela (MH846121 in Zheng et al. 2019) was included within T. japonicus (lineage I), and T. lepturus (MW719078) in our study and “T. lepturus” (MK333401 in Mukundan et al. 2020) were not considered monophyletic (lineages II and III). Thus, our study used COI sequences to examine the taxonomic status of Trichiurus species. All COI sequences of Trichiurus species in GenBank (NCBI database) were downloaded. After alignment, 477 bp were analyzed. The phylogenetic trees reconstructed within ML and NJ were identical, with only small differences in bootstrap values. In the COI phylogenetic analyses (ML tree, Fig. 4), all sequences were grouped into six lineages (A–F). Lineage F included T. brevis within the T. russelli complex. Trichiurus haumela (MH846121 in Zheng et al. 2019) was also included within T. japonicus in lineage A. The specimen from the Indian Ocean (MK333401 in Mukundan et al. 2020) might be not T. lepturus, as it was grouped with other specimens of T. gangeticus in lineage E. The genetic distance within the six lineages ranged from 0.0013 (lineage F, T. brevis) to 0.0333 (lineage C, T. lepturus), and the genetic distance between lineages ranged from 0.0435 (between T. japonicus and T. auriga) to 0.1600 (between T. japonicus and T. brevis) (Table 2). Based on the mitogenomes, the genetic distances between these five species ranged from 0.0507 (T. nanhaiensis and T. gangeticus) to 0.1331 (T. gangeticus and T. brevis), including the d-loop region, and from 0.0476 (T. nanhaiensis and T. gangeticus) to 0.1288 (T. lepturus and T. brevis), excluding the d-loop region (Table 2). Moreover, the mitogenome p-distances between T. japonicus and T. haumela, including and excluding the d-loop region, were 0.0067 and 0.0047, respectively.

Figure 3. 

The maximum-likelihood (ML) tree of the Trichiuridae based on the sequences of mitogenome (excluding d-loop). The numbers at the nodes are bootstrap values of the ML and NJ (neighbor-joining) analyses.

Figure 4. 

The maximum-likelihood (ML) tree of six Trichiurus species in the world based on the COI gene. The numbers at the nodes are bootstrap values of the ML and NJ (neighbor-joining) analyses.

After identifying species by morphological characters and phylogenetic analysis, traditional caliper measurements were performed, which produced 14 landmark sites (a–n, Fig. 2). Trichiurus brevis could not be identified by these standard morphological characteristics (Table 3), but it is easy to distinguish from T. lepturus complex in the anal opening positioned vertically at the 34^th and 35^th dorsal fin rays. Thus, our morphological comparison is mainly concentrated in the T. lepturus complex (Table 3, Fig. 5). Our study compared many numerical values based on the external morphology of various body ratios (more than 20 counts, Suppl. material 1: Fig. S2). Our study found that the caudal length is longer in T. japonicus [D(i,n/i,m) = 3.08 ± 0.32, 2.61 ± 0.09 and 2.74 ± 0.18 in T. japonicus, T. lepturus, and T. nanhaiensis; Table 3, Fig. 5A]; the body depth at the anus is wider in T. nanhaiensis [D(m,n/e,f) = 11.60 ± 1.79, 10.29 ± 0.87 and 7.77 ± 1.38 in T. japonicus, T. lepturus, and T. nanhaiensis; Table 3, Fig. 5B]; the orbital length is larger in T. lepturus [D(d,o/j,k) = 3.00 ± 0.42, 2.20 ± 0.32 and 3.57 ± 0.44 in T. japonicus, T. lepturus, and T. nanhaiensis; Table 3, Fig. 5C]; and the head is slenderer in T. lepturus [D(i,l/d,o) = 2.22 ± 0.14, 2.73 ± 0.13 and 1.92 ± 0.18 in T. japonicus, T. lepturus and T. nanhaiensis; Table 3, Fig. 5D].

Figure 5. 

The simple regression and the boxplot analysis in T. japonicus (blue), T. lepturus (orange) and T. nanhaiensis (grey) A Total length [D(i,n)] and Preanal length [D(i,m)] B Caudal length [D(m,n)] and Body depth at anus [D(e,f)] C Head depth [D(d,o)] and Orbital length [D(j,k)] and D Head length [D(i,l)] and Head depth [D(d,o)]. The landmarks are illustrated in Fig. 2.

The mitogenomes of all four Trichiurus species contain 36 mitochondrial genes (13 PCGs, 21 tRNA-coding genes and 2 rRNA-coding genes) and two noncoding regions (OL and d-loop, control region) (Table 4). One of the 13 PCGs (ND6), seven tRNA-coding genes (Gln, Ala, Asn, Cys, Tyr, Ser, Glu), and one noncoding region (OL) are encoded on the L-strand, and the other 28 genes (12 PCGs, 14 tRNA-coding genes, and 2 rRNA-coding genes) and d-loop are encoded on the H-strand. The composition and arrangement of the mitochondrial genes in these four species were the same as those in T. gangeticus in the Indian Ocean (MK333401 in Mukundan et al. 2020). To characterize variation among the Trichiurus mitogenomes, we analyzed the base composition of T. gangeticus in the Indian Ocean (Mukundan et al. 2020). We found the mean AT nucleotide content of the five complete mitogenomes to be similar (55.0% in T. japonicus, 55.1% in T. lepturus, 54.4% in T. nanhaiensis, 54.3% in T. gangeticus and 54.6% in T. brevis; Table 5). All mitogenomes had high A + T content: 54.3%–55.1% (53.3%–54.1% for PCGs, 54.5%–56.9% for light tRNA genes, 53.4%–54.5% for heavy tRNA genes, 52.3%–52.6% for rRNA genes, and 63.5%–67.1% for d-loop). The overall AT skews in the five entire mitogenomes were 0.06006, 0.04465, 0.05775, 0.04891 and 0.06789, and the overall GC skews were -0.17695, -0.17258, -0.18480, -0.18396 and -0.19633 (Table 5).

The total lengths of PCGs in the five Trichiurus species ranged from 11.530 to 11538 bp, accounting for 68.47%–69.59% of the entire mitogenome. The mitogenomes could be translated into 3.809–3.810 amino acid-coding codons, excluding stop codons. ND5 and ATP8 were the largest and smallest genes, respectively. The majority of PCGs start with an NTN (ATG/GTG/ATT) start codon and are terminated with the stop codons TAA, TAG, and AGA (Table 4). Most of the AT skew and GC skew values of the PCGs in the five species were negative, indicating that the bases T and C were more plentiful than A and G (Table 5). Moreover, the A + T content and AT skew differed among PCGs (Suppl. material 1: Table S2, Fig. 6). The AT skew values of five genes (ND2, COII, ATP8, ND4 and ND5) were positive, and those of other genes were negative. The GC skew value was positive only for ND6.

Figure 6. 

A AT-skew in 13genes. B GC-skew in in 13 genes.

To better understand the role of selection in the evolution of the PCGs, the Ka/Ks value of each PCG was calculated (Fig. 7A). All the PCGs, excluding ND6, showed signatures of purifying selection (Ka/Ks < 1). The ND6 and ATP8 genes had the highest Ka/Ks values (1.18 and 0.13), and the COI and cyt b genes had the lowest Ka/Ks values (0.04). A lower Ka/Ks value indicates less variation in amino acids. For the ND6 gene, the highest Ka/Ks value was observed between T. nanhaiensis and T. gangeticus (Fig. 7B). For the ATP8 gene, the highest Ka/Ks value was observed between T. brevis and other Trichiurus species (Fig. 7C). Summaries of the relative synonymous codon usage and number of amino acids in the annotated PCGs are presented in Figs 8, 9 and Suppl. material 1: Table S3. Overall codon usage among the sequenced Trichiurus mitogenomes was similar; Leu, Ala, Thr, Ile, and Ser were the five most common amino acids.

Figure 7. 

A Mean evolutionary rates for each protein coding gene in mitogenomes of five Trichiurus species B Evolutionary rates of ND6 gene of five Trichiurus species. C Evolutionary rates of Ka/Ks in ATP8 gene of five Trichiurus species. Indicated the rates of non-synonymous substitutions to the rate of synonymous substitutions (ka/ks). T. japonicus (TJ), T. lepturus (TL), T. nanhaiensis (TN), T. gangeticus (TG) and T. brevis (TB).

Figure 8. 

Relative synonymous codon usage (RSCU) of the mitogenomes of the five Trichiurus species; the stop codon is not included. T. japonicus (TJ), T. lepturus (TL), T. nanhaiensis (TN), T. gangeticus (TG) and T. brevis (TB).

Figure 9. 

Frequencies of different amino acids in the mitogenomes of the five Trichiurus species; the stop codon is not included.

The lengths of 16S rRNA genes ranged from 1.725 (T. gangeticus) to 1.742 (T. lepturus), whereas those of 12S rRNAs ranged from 957 (T. gangeticus and T. nanhaiensis) to 959 (T. brevius). These rRNA genes are located between tRNA^Phe and tRNA^Leu and are separated by tRNA^Val. The AT content of the rRNA genes ranged from 52.3% to 52.6% (Table 5). The total lengths of the 21 tRNA genes ranged from 1.483 (T. japonicus and T. brevis) to 1.487 bp (T. lepturus), and individual tRNA genes typically ranged in size from 66 to 76 bp. No sequence similarity to the tRNA^Pro gene was observed elsewhere in the mitogenome. The d-loop in Trichiurus mitogenomes is located between tRNA^Phe and tRNA^Thr. The A + T content (63.5%–67.1%) of the d-loop was higher than that of the whole genome (54.3%–55.1%), rRNA-coding genes (52.3%–52.6%), and tRNA-coding genes (54.4%–55.2%) (Table 5). Furthermore, compositional analysis revealed that the mitogenome of T. nanhaiensis and T. gangeticus had a negative AT skew (-0.01349 and -0.03428) in the d-loop.

To determine molecular markers that could be used to examine the phylogeny and identify species, the overall interspecific and intergeneric p-distance was used to describe the evolutionary rate of two rRNA-coding genes, 13 PCGs and the mitogenome, excluding the d-loop region (Fig. 10, Table 6, Suppl. material 1: Table S4). The maximum interspecific p-distance (mean = 0.189, range = 0.073-0.241) was observed for the ND6 gene, and the maximum intergeneric p-distance (mean = 0.369, range = 0.257-0.470) was observed for the ATP8 gene. Among these 16 markers, 9 markers (e.g., 12S rRNA, ATP6 and ND1 genes) displayed overlapping interspecific and intergeneric p-distances (Fig. 10). Among the four genes in the oxidase family, only COII showed overlap between interspecific and intergeneric p-distances. Furthermore, the range of pairwise interspecific p-distances among five Trichiurus species based on the 16S rRNA and cyt b genes ranged from 0.015 (between T. gangeticus and T. nanhaiensis) to 0.077 (between T. lepturus and T. brevis) and from 0.072 (between T. gangeticus and T. nanhaiensis) to 0.143 (between T. nanhaiensis and T. brevis) (Table 6). In addition, our study found that the 16S rRNA genetic distances between T. brevis (short-tailed species complex) and other Trichiurus species (T. lepturus complex or large-head species complex) were not higher than those within the T. lepturus complex (Table 6). The results based on cyt b and 16S rRNA differed.

Figure 10 

. The mean partwise interspecific (gray) and intergeneric (black) p-distance in each gene.

The mitogenomes of Trichiurus species encode 36 typical mitochondrial genes (13 protein-coding, 2 ribosomal RNA-coding genes, and 21 transfer RNA-coding genes) and two typical noncoding control regions, the d-loop and origin of the light strand (OL) (Table 4). Overall, the mitogenomes of four Trichiurus species in the present study have a gene order and composition similar to other Trichiurus mitogenomes in previous studies (e.g., Liu and Cui 2009; Liu et al. 2013; Zheng et al. 2019; Mukundan et al. 2020). The gene order of the mitogenomes in Trichiuridae was similar to that in most teleosts, although different types of gene rearrangements were observed within Pleuronectiformes (Gong et al. 2015) and Stomiiformes (Arrondo et al. 2020).

In addition, our study found that the tRNA^Pro gene was absent in the Trichiurus mitogenomes. Previous studies of the complete mitogenomes of Trichiurus species have also obtained similar findings (e.g., Liu and Cui 2009; Liu et al. 2013; Zheng et al. 2019; Mukundan et al. 2020). Our data indicate that this event only occurred in the genus Trichiurus and not in other teleosts. Adams and Palmer (2003) proposed that the mitochondrial gene content is highly variable across eukaryotes. However, most previous studies have been conducted on plants (Adams et al. 2001; Adams and Palmer 2003). In addition, the loss of genes in vertebrate mitogenomes is rare. In teleosts, loss of the ND6 gene was observed only in Antarctic fish mitogenomes (Papetti et al. 2007), and no cases of tRNA gene loss were observed.

Because the systematics of many species remain unresolved, many studies have employed molecular, phylogenetics and DNA barcoding approaches (e.g., Hebert et al. 2003; Hsu et al. 2009; Han et al. 2019). Among all molecular markers, the mtDNA COI and cyt b genes have been the most frequently used (Yang et al. 2016; Han et al. 2019; Hsu et al. 2020; Ju et al. 2021). In addition, some studies have used mtDNA rRNA (12S and 16S) sequences to resolve phylogenetic relationships and taxonomy (Byrne et al. 2010; Herler et al. 2013; Zheng et al. 2016). However, the results of pairwise p-distances based on 16S rRNA and cyt b genes differed in this study (Table 6). These results, coupled with the results of the Ka/Ks analyses (Fig. 7), suggested that the evolutionary rates of these genes differed. Our study indicated that overlap between the interspecific and intergeneric distances might affect phylogenetic reconstruction and molecular species identification. For example, the intergeneric 12S rRNA p-distance between Evoxymetopon and Assurger was 0.056, which is smaller than the interspecific 12S rRNA p-distance of Trichiurus (Table 6, Fig. 10). Thus, some genes that were used to resolve the phylogeny and identify species should be evaluated. Actually, this question has been intensively discussed, and has been applied to mitochondrial genes (i.e., Zardoya and Meyer 1996; Miya and Nishida 2000). However, some studies ignored this question (our observations).

Moreover, variation in the Ka/Ks values was greater for ATP8 than for other genes (Fig. 7), and the length of ATP8 (approximately 168 bp) was shorter. Thus, our study suggested that only COI, COIII, cyt b, ND5 and mitogenome (excluding d-loop) sequences could be used to identify Trichiurus species and examine the phylogeny of Trichiuridae. However, these genes may also display a limited ability to identify complex evolutionary relationships in many fishes (Mirande 2018). For example, the Ka/Ks values of the COI and cyt b genes were the lowest (0.04). Lower Ka/Ks values indicate less variation in amino acids (Brookfiel 2000; Li et al. 2020; Sun et al. 2021). Therefore, an increasing number of studies have used complete mitogenome data to resolve animal phylogenies and identify species because they provide more information (Ajene et al. 2020; González-Castellano et al. 2020; Irisarri et al. 2020); the results of our study support this hypothesis.

The taxonomy of the genus Trichiurus remains unresolved because of the high degree of morphological similarity within the genus in terms of bodily appearance and silvery coloration. Our study also showed that identifying Trichiurus species by morphological characters is very difficult (Fig. 5, Table 3). Phylogenetic analyses based on the complete mitogenome (Fig. 3) and COI gene (Fig. 4) showed that T. haumela was clustered with T. japonicus. Moreover, T. japonicus is synonymous with T. lepturus in FishBase (Froese and Pauly 2021), but the present results (Figs 3, 4; Table 2) indicated that T. haumela is synonymous with T. japonicus and that T. japonicus is a valid species (Hsu et al. 2009; Fricke et al. 2021). Moreover, the systematic position of T. brevis is still not resolved in this study because we did not analyze other species of Trichiurus russelli complex, and did not provide enough information.

In addition, the results suggested that specimens in the Indian Ocean (MK333401 in Mukundan et al. 2020) are not “T. lepturus” (Figs 3, 4; Table 2). In the phylogenetic tree based on COI (Fig. 4), MK333401 was grouped with other specimens in the Indian Ocean as lineage E. Within lineage E, most specimens were identified as “T. lepturus”, and only MK340737 in Bangladesh was identified as T. gangeticus. According to these results, members of lineage E could not be identified as T. lepturus; our data suggest that they should be recognized as T. gangeticus. Similarly, within lineage B, some specimens were identified as Trichiurus sp. (Isari et al. 2017), and some specimens were identified as T. auriga. We thus recognized lineage B as T. auriga (Fig. 4). Accordingly, our study suggests that the Trichiurus specimens in the Indian Ocean are not T. lepturus calls into question many previous studies (e.g., Chakraborty et al. 2006a; Jahromi et al. 2016; Mukundan et al. 2020).

Chakraborty et al. (2006a) and Chakraborty and Iwatsuki (2006) found that T. lepturus in Indo-Pacific differed from that in Atlantic using 16S rRNA sequences. However, Hsu et al. (2009) identified the specimens of T. lepturus in the Indo-Pacific in Chakraborty et al. (2006a) and Chakraborty and Iwatsuki (2006) as “Trichiurus sp. 2” (synonym of T. nanhaiensis). Jahromi et al. (2016) examined the phylogenetic relationship of T. lepturus from the Persian Gulf using 16S rRNA sequences, and suggested homogeneity between Persian Gulf and the other Indo-Pacific individuals. However, Lin et al. (2021) found that the specimens in Jahromi et al. (2016) were identified as T. japonicus, T. lepturus, and T. nanhaiensis and the specimens in the Persian Gulf was nested with T. nanhaiensis using 16S rRNA sequences. Besides, Lin et al. (2021) found T. nanhaiensis could be divide as two groups, Indo-Pacific and West Indian. The results of our study indicate that T. gangeticus was more similar to T. nanhaiensis (Fig. 4, Table 2). Thus, our study considers these two groups might be T. nanhaiensis and T. gangeticus, although we did not collect the COI data of T. nanhaiensis in the east Indian Ocean. In addition, our study found some specimens from the Gulf of Oman referred to in Lin et al. (2021) were in fact T. lepturus. However, T. lepturus had the highest intraspecific diversity (Table 2). Thus, our study suggests that systematics of T. lepturus species complex and T. lepturus both need to be reviewed.

Bingpeng et al. (2018) used the COI gene to identify fish at the species level in the Taiwan Strait and proposed that the average p-distances within species, genera, families, orders, and classes were 0.0021, 0.0650, 0.2370, and 0.2560, respectively. Our study revealed that the range of COI interspecific distances in Trichiurus ranged from 0.0435 to 0.1600, and the intraspecific distance within lineage C (T. lepturus) was 0.0333 (Table 2, Fig. 4). These results suggest that there were cryptic species within lineage C. Lineage C could be divided into three sublineages C1–C3 (Fig. 4). Lineage C1 was distributed in the West Pacific Ocean; lineage C2 was distributed in the Northwest Atlantic Ocean; and lineage C3 was distributed in the East Pacific and Southwest Atlantic oceans. The range of the pairwise genetic distances ranged from 0.0308 to 0.0529. Thus, these three sublineages should be recognized as three different species. Within lineage C3, most specimens were identified as T. lepturus, but some specimens (MF957079-MF957087) were identified as T. nitens. Trichiurus nitens was described in 1899, and it is distributed in the eastern Pacific, from California south to Peru. Nakamura and Parin (1993) considered it synonymous with T. lepturus, but some researchers have suggested that it is the real T. nitens (Eschmeyer and Herald 1983; Burhanuddin and Parin 2008; Robertson et al. 2017). In addition, T. margarites is considered a valid species in FishBase (Froese and Pauly 2021) and ECoF (Fricke et al. 2021). Trichiurus margarites is distributed in the South China Sea (Li 1992; Fricke et al. 2021), but this species has not yet been detected along Chinese coastal waters. However, it is possible that the lineage C1 is T. margarites (Fig. 4). Thus, our study suggested that the systematics within lineage C require careful evaluation. In future studies, a careful morphological comparative work within lineage C is needed.

Tzeng et al. (2007) analyzed the morphometry from T. japonicus and T. lepturus, and found that it exhibited high intraspecific variations. Our study also found the same (Fig. 5, Table 3). However, although Tzeng et al. (2007) found a decisive specific gap of non-overlapping scattering using discriminant function analysis, they did not provide a reference key to identify the Trichiurus species because it is very difficult. Lee et al. (1977) proposed that T. japonicus and T. lepturus can been distinguished based on the external morphology of various body ratios. Thus, our study calculated some body ratios (Table 3, Fig. 5) and only found that the ratio between length and depth of head can been used to distinguish T. lepturus and other species (Fig. 5D). In T. lepturus, the ratio between distance of head length [D(i,l)] and distance of head depth [D(d,o)] was larger than 2.5. Our study also did not find a reference key to distinguish these three species within T. lepturus complex; we used the complex indexes to distinguish them. Trichiurus japonicus has a longer body and tail (Fig. 5A, B); T. lepturus has a shorter tail, longer head, and bigger eye (Fig. 5A, C, D); and T. nanhaiensis has a wider tail, smaller eye, and shorter head (Fig. 5B, C, D).