Background

In recent years, improvements in marine genomic sequencing tools and technologies have allowed taxonomists to better investigate the evolutionary relationships of less explored taxa. Correct identification of these tie results is also essential for improving the knowledge of some linked topics, such as evolutionary biology, ethology, anatomy, and functional morphology of these rare species, which are often characterized by fragmented and vague information [1]. Moreover, rare species constitute an active part of marine trophic webs and have trophic relationships with key species, as confirmed by some authors for Lampriformes species [2, 3]. Owing to their predatory habits, these species play essential roles in regulating ecosystem function and species diversity [4]. Moreover, these organisms interact with pathogens, becoming vectors for the dissemination of sometimes unknown metazoan species [5].

A powerful tool for monitoring aquatic ecosystems is the environmental DNA (eDNA) method, which allows the detection of living organisms’ presence with reasonable accuracy in a noninvasive way [6]. This is only possible if the species to be identified is annotated in the reference sequence databases commonly used, such as GenBank [7]. These aspects are even more critical when considering rare or elusive species, which are sometimes difficult to detect with only one method; for this reason, most of their features are still unknown. Moreover, assessing the genetic structure of many organisms could lead to a better comprehension of their evolution and adaptation during their history of environmental variability and influence in the following years.

Phylogeny and teleost classification have been rapidly changing with increasing use of molecular phylogenetic approaches because of increasingly taxonomically valid datasets. The development of the first molecular markers as new evaluation criteria, first based on sequences of mitochondrial DNA (mtDNA) genes or even complete mitogenomes, provides new essential information across broader taxonomic scales within teleost [8, 9]. Sequencing technologies, combined with a massive reduction in time, effort, and cost of methodologies, have allowed more straightforward and more widespread use of these techniques, resulting in a considerable increase in the effectiveness and quality of data [10]. The diffusion of this approach has recently made it possible to study rare and less commercially important but biologically relevant species [11].

The deep marine environment is characterized by very harsh and selective conditions, which lead living organisms to develop morphological and genetic adaptations [12]. Although bony fishes represent the most abundant component of deep-sea megafauna [13], very little is known about their specific genetic adaptations [14] due to the difficulty of sampling often rare species and the scarcity of related research projects [15]. Recently, owing to their ecological importance, there has been increased interest in mesopelagic and bathypelagic species of the Mediterranean Sea [16, 17]. Nevertheless, information about the distribution of Lampriformes is very scarce and fragmented [18]. This fish order comprises very interesting species that are often characterized by a particular functional morphology [19, 20]. Currently, the species reported in the Mediterranean basin, with the documented capture of very few samples, are Lampris guttatus (Brünnich, 1788), Lophotus lacepede (Giorna, 1809), Trachipterus arcticus (Brünnich, 1788), Trachipterus trachypterus (Gmelin, 1789), Regalecus glesne (Ascanius, 1772) and Zu cristatus (Bonelli, 1819) [21, 22]. Owing to the scarcity of data, the evolutionary relationships within Trachipteridae are poorly resolved [9, 22, 23]. Moreover, some confusion about nomenclature and classification of trachipterid genera and species has been recorded over time [24]. Owing to the known allometric growth that characterizes this family, drastic morphological changes occur during ontogeny, with significant morphological and functional differences between the juvenile and adult stages within the same species [25]. Indeed, within this order some species prefer shallow water during larval/juvenile stages while moving to the deep-sea environment during adult life. These transition from shallow, nearshore habitats to the deep open ocean is one of the two significant events hypothesized to have characterized the evolution of the order Lampriformes [26]. This variability, combined with their rarity, has led to the description of different life stages as separate species rather than as part of the ontogenetic continuum of a single species, thus influencing the apparent diversity at the species level [9, 27]. However, the study related to adaptations of Lampriformes species to deep-sea life are very scarce and, in considerations of their morphological features, needs further in-depth investigations, especially from a functional and genetic points of view [28, 29]. Hypothesizing that these adaptations could reflect a genetic evolution within teleost, deepening the knowledge of this taxon since new molecular information will be essential for assessing the phylogenetic relationships within the order and investigating their adaptative responses during evolution.

Z. cristatus (Bonelli, 1820), commonly referred to as scalloped ribbonfish, is a mesopelagic species with a wide distribution. Indeed, its presence was reported in the Mediterranean Sea, the Azores and Madeira in the Atlantic, and in the Pacific and Indian Oceans [30]. Despite its distribution, information on the biological and ecological features of this elusive species is scarce and fragmented. Features of the species are connected to the differences in juvenile to adult life habits. Indeed, from video-recorded images of live juvenile species, it is prominent how these long filaments are the main locomotory engine in this life cycle stage. This species is also known for its locomotory features, particularly its head-up swimming style [31]. Even more than adults with a more developed muscle mass, the ability to swim in the typical vertical position is possible in the juvenile stage thanks to the dorsal/caudal fins undulations, favorited by the scalloped shape of the body with the tapering of the caudal portion.

The metabolism of animal generally decreases with the increasing of depth. This reduction represses the selective pressure on locomotory capacity, causing a reduction in the metabolic rate [32]. Hence, studing species which are involved in bathymetrical migration linked to their ontogenetic development may shed light on interesting insights. Several authors have reported that this species is the only member of the genus Zu present within the Mediterranean Sea [20, 30, 33] considering the species as rare in the basin. Differences in the life history characteristics of each teleost taxon are evident [34], hence each contribution on less studied species could result essential. The present manuscript describes the first complete assembly of mitochondrial DNA and its annotation of Z. cristatus from the Ionian Sea, providing further information on its features and the phylogenetic position of the species.

Methods

Sample collection

An adult sample of Z. cristatus was captured in May 2020 at a depth of approximately 720 m by longline swordfish fisheries off the coast of Noto, Syracuse, Italy (36°50’05” N 15°16’49” E) (Fig. 1).

Fig. 1
figure 1

Map of the catch area of the Z. cristatus with the highlighted point in the Ionian Sea (red star) (36°50’05” N 15°16’49” E). In the red box the Z. cristatus adult specimens analyzed in this study. Images were taken at the Messina University laboratory before the necroscopy of the fish

Full size image

Historically, the fisheries of the southern Ionian Sea focused on capturing commercially important species such as Lepidopus caudatus and Xiphias gladius. Hence, Lampriformes, when it occurs, are considered bycatch and are often discarded as fishing waste [35]. The sample used in this study was kept dead from fishing waste upon request to fisherman, partially identified on board, and stored at -20° for further analysis.

Morphological identification of the specimen and data collection

The necropsy and identification procedures were performed at the Department of Chemical, Biological, Pharmacological and Environmental Sciences of the University of Messina, Italy. The sample was transported frozen and immediately processed after arriving at the laboratory. All the measurements for identification were taken following identification keys proposed by Olney in 1999 for the Trachipteridae family [36]. Epaxial muscle samples were taken for further molecular identification and stored at -80 °C. Length data were collected via a standard icthyometer of 100 cm length and precision of 1 mm, plus other precision measuring sticks (0.1 mm in accuracy) for the fins and detailed measures, whereas a precision scale was used for the total weight (UW8200S, Shimadzu Corporation, Kyoto, Japan). The biometric and meristic features of this sample were reported in detail by Albano et al. [19].

DNA extraction

Total genomic DNA was extracted from 0.25 mg of epaxial muscle via a mechanical glass-bead disruption method following the standard protocol suggested by Qiagen for the Blood and Tissue Kit. Once isolated, the DNA was stored at − 20 °C. DNA concentration and purity were assessed on a NanoPhotometer N60 (Implen, Munich, Germany), and integrity on 1.2% agarose gels [37].

mtDNA sequencing, assembly and annotation

Total genomic DNA was sent to Galseq (www.galseq.com) for the preparation of two libraries containing different insert sizes (300 and 550 bp). The libraries were sequenced in paired-end mode (2 × 150 bp) on an Illumina HiSeq 4000 platform. After sequencing, Illumina raw reads were first inspected via FASTQC [38] and then cleaned with Trimmomatic v.0.39 [39] with the following options: HEADCROP:10, ILLUMINACLIP:~/Trimmomatic-0.39/adapters/All_adapters.fa:2:25:10, LEADING:25, TRAILING:25, SLIDINGWINDOW:4:25, MINLEN:35. The cleaned reads (Phred score ≥ 25) were then assembled by MitoFinder v.1.4.1 [40] using the Zu cristatus mitogenome (GenBank Accession: AP002926.1) as a reference. To determine the genome coverage, the clean reads were mapped back against the assembled mitochondrial genome via the bwa algorithm (v. 0.7.17.r1188) [41], and then, Qualimap 2 software [42] was used to evaluate the alignment data and determine the mapping statistics and other genome metrics. The mitochondrial genome was annotated via the MitoFish web server [43]. Finally, the sequence was deposited in GenBank as BioSample SAMN28862047 with accession number PRJNA845808. The nucleotide content values of each sequence were obtained via MEGA X [44]. The composition bias based on the asymmetry values of all the sequences was estimated via the following formulas: AT skew = (A% - T%)/(A% + T%), and GC skew = (G% - Ci%)/(G% + C%) [45]. The relative synonymous codon usage (RSCU) was also estimated via MEGA X [44].

Selection analysis of protein-coding genes (PCGs)

Phylogenetic analysis was conducted on 13 concatenated protein-coding genes (ATP6, ATP8, COXI, COXII, COXIII, CYTB, ND1, ND2, ND3, ND4, ND4L, ND5, and ND6) for 43 species of fishes, as listed in Supplementary Table 1. Before running the analysis, each protein-coding gene sequence was manually cleaned to remove stop codons and obtain the number of nucleotides as a multiple of 3. All genes were concatenated and aligned via MAFFT [46] (v7.475) with the following options: --globalpair --reorder. The output generated was then used as input in iqtree2 [47] (version 2.2.0; flags -m TESTONLY --mset raxml) to find the best evolutionary model to use during the phylogenetic tree reconstruction step performed with RAxML-NG [48] (v. 1.1.0; options: --model GTR + F + I + G4 -bs-trees 1000 --all). We used a gene-level analysis using omega (ω), which is based on the nonsynonymous (dN)-to-synonymous (dS) ratio (ω = dN/dS), to identify positively selected sites.

Positive selection analysis was conducted for 13 genes and was limited to the Lampriformes group (composed of the following: (1) Lampris guttatus; (2) Regalecus glesne; (3) Trachipterus trachypterus; (4) Zu cristatus) compared with the remaining mitogenomes in the first row. Subsequently, we repeated the analysis using Z. cristatus as the foreground branch, and finally, we restricted the analysis to Z. cristatus, excluding all other Lampriformes, to discover sites positively selected in this species only. The analysis was performed via EasycodeML [49] with the branch-site model preset (nested model). To assess the goodness of fit of the two models (modelA vs. nullmodelA), a likelihood ratio test (LRT) was calculated with the built-in utility (Run LRTs) of the EasycodeML algorithm. Protein-coding genes with p values ≤ 0.05 and posterior probabilities > 0.95 were considered under positive selection.

We used PCG sequences of Z. cristatus and 42 other species to evaluate the pairwise distances among these species. Species were selected in relation to their life habits, habitat features and taxonomical distances (Supplementary Table 1). The GenBank accession IDs used to obtain the single PCGs are listed in Supplementary Table 2. Each of the thirteen Z. cristatus PCG sequences (ATP6, ATP8, COI, COII, COIII, CYTB, ND1, ND2, ND3, ND4, ND4L, ND5, and ND6) was first aligned with the corresponding 42 other species. Analyses were conducted via the maximum composite likelihood model [50]. This analysis involved 43 nucleotide sequences, for a total of 559 PCGs. The codon positions included were 1st + 2nd + 3rd + Noncoding. All ambiguous positions were removed for each sequence pair (pairwise deletion option). Evolutionary analyses were conducted in MEGA11 [51]. DNA conservation sequence analyses were carried out via DnaSP (v.6) [52].

Phylogeny

The phylogenetic analysis of Z. cristatus mt-Co1 revealed all the currently available sequences in the NCBI database of 21 species of the order annotated in the GenBank database (Supplementary Table 3). The coding sequences of mt-co1 from 21 Lampriformes species were aligned with Muscle and then trimmed with Gblocks. A total of 542 positions were selected for Bayesian phylogenetic reconstruction via the GTR + I nucleotide substitution model. The consensus tree was built after burning 25% of the trees from 500,000 generations. Bayesian posterior probabilities are represented as percentages. This analysis was performed with the pipeline NGPhylogeny.fr [53]. For phylogenetic inference evaluation, our complete Z. cristatus mitogenome was compared with a subset of 44 fish mitochondrial genomes downloaded from the MitoFish database (Supplementary Table 2). Our dataset comprises all major ray-finned fish lineages and major taxa, such as Teleostea, Clupeocephala, Euteleostea, Neoteleostea, Acanthomorpha, and Percomorpha. Because of this variety, we provided our analysis to expand the phylogenetic relationships among Z. cristatus, the other Lampriformes species with annotated mtDNA genomes, and the remaining related groups with similar biological and ecological habits in most cases, plus one freshwater outgroup, Danio rerio, chosen based on an ecological relatedness. Moreover, as confirmed by other authors, this species, as a cyprinid, could represent a useful outgroup in saltwater species phylogenetic reconstructions based on mt-Co1 [54], complete mitogenome [55], and other genes [56]. This analysis was performed via the JolyTree utility v.1.1b.191021ac [57] by applying the “-r 1000” flag (number of steps when performing the ratchet-based BME tree search).

Results

Species identification

Morphological identification revealed our sample as Zu cristatus, a species belonging to the Trachipteridae family [36, 58] (Fig. 1). The specimen was an adult female 121 cm in length and 4005 g in weight.

Next-generation sequencing (NGS), mtDNA assembly and annotation

A total of 169,273,456 and 219,941,430 raw reads were obtained from NGS sequencing of the 300 bp and 550 bp insert-size libraries, respectively. After quality filtering and trimming, over 93% (149,168,979 and 176,325,972 reads from the 300 bp and 550 bp libraries, respectively) of the total reads were used for mitogenome assembly, resulting in a single, circular DNA molecule of 17,450 bp, with a mean coverage of 1371 X (standard deviation: ± 238 X). The complete mitochondrial DNA sequence is available in GenBank under the accession number ON695781.

Mitogenome features

A total of 37 genes (22 tRNAs, 13 protein-coding genes (PCGs) and 2 rRNAs), including the D-loop region, never annotated for this species, were identified in the circular double-stranded mtDNA molecule of 17,450 bp. This reference represents the first complete mitochondrial DNA for Z. cristatus. The gene positions in the H-strand (heavy strand) and L-strand (light strand) strands are reported in Table 1.

Table 1 Organization of the Z. cristatus mitochondrial genome
Full size table

A graphical reconstruction of the mitochondrial genome is shown in Fig. 2, and the coverage plot representing per-base sequencing depth of combined libraries was showed in Supplementary Fig. 3. An in-depth analysis of the Z. cristatus mtDNA isolated in this study was conducted to elucidate its features.

Fig. 2
figure 2

Structure of the mitochondrial genome of Z. cristatus isolated in this study. A total of 37 genes (22 tRNAs, 13 protein-coding genes and 2 rRNAs), including the D-loop region, were identified in the genome assembly. The internal circle represents the G + C content per 5 bp (darker lines represent higher G + C contents). The external circle represents the two mtDNA strands (H eternally, L internally). The red, black and avana blocks indicate tRNAs, PCGs, and rRNAs, respectively. The D-loop region is colored brown. The graphical representation of the genome was created via Mitofish, and the Z. cristatus illustration was created via Dr. Serena Savoca

Full size image

An analysis of the asymmetry of the Z. cristatus whole mitochondrial genome revealed negative AT skews (− 0.0814) and negative GC skews (− 0.1898) (Table 2), indicating that A and C (adenine and cytosine) were present in greater proportions in the heavy chain than were T and G (thymine and guanine).

Table 2 Total length, composition and skewness of the sequenced Z. cristatus whole mitogenome and concatenated pcgs, tRNAs, rRNAs and D-loops (control region)
Full size table

The total length of the PCGs was 11,415 bp, with an average AT content of 56.4% (Supplementary Table 4). The asymmetry analysis of these concatenated regions revealed negative AT skews (− 0.1850) and negative GC skews (− 0.2070) (Supplementary Table 4). The proportions of T and G to A and C were almost equal in this region. Individually evaluated, PCGs presented average AT contents ranging from 49.2% (ND6) to 60.07% (ATPase8) (Supplementary Table 4). Additionally, all PCGs had negative AT skews and, in most cases, negative GC skews, except for ND6.

The mitogenome of Z. cristatus contained a set of 22 tRNAs, with a length of 1460 bp, an average AT content of 55.0%, and positive AT/GC skews (Table 2). More specifically, the AT content ranged from 40.8% (tRNA-Thr) to 65.2% (tRNA-Arg, tRNA-Gly) (Supplementary Table 4), whereas the gene lengths ranged from 58 bp (tRNA-Phe) to 74 bp (tRNA-Leu and tRNA-Lys), as shown in Supplementary Table 5. The secondary structure of the interestingly short tRNA-Phe was reported in Supplementary Figure S1.

The 16 S rRNA and 12 S rRNA genes presented a total length of 2,626 bp and an average AT content of 54.1%, with positive AT skews and negative GC skews. In particular, the 12 S rRNA gene was 949 bp long with an AT content of 51.8%, whereas 16 S rRNA had an AT content of 55.4%, with a gene length of 1,677 bp (Supplementary Table 6).

The control region of Z. cristatus had a length of 1,727 bp and an AT content of 59.4%, with a positive AT skew and a negative GC skew (Table 2). Our data are the first to report the presence of a D-loop region in this species.

The mitochondrial DNA sequence of Z. cristatus also contained 15 intergenic nucleotide (IGN) sites (199 nucleotides in total) ranging in size from 1 to 130 nucleotides (Table 1), with an average length of 13.27 nucleotides. Interestingly, the Ala IGN region was unexpectedly large (130 nucleotides), affecting the total length and average size of all the IGN sites. Data on the AT content of the 37 mitochondrial genes and AT and GC skews are graphically shown in Fig. 3C.

Fig. 3
figure 3

(A) Information on the AT content (%) of Z. cristatus mitogenome genes. This figure was generated via GraphPad Prism 8.0.1 [121]. (B) Relative synonymous codon usage (RSCU) analysis of the Z. cristatus mitogenome. The RSCU values are represented on the y-axis, and families of synonymous codons and their respective amino acids are indicated on the x-axis. This figure was generated via GraphPad Prism 8.0.1 [121]. C) AT- and GC-skew values of the Z. cristatus whole mitogenome; PCGs concatenated and singularly concatenated; tRNAs, rRNAs and D-loops. This figure was generated via GraphPad Prism 8.0.1 [121]

Full size image

Characteristics of protein-coding genes (PCGs)

The mitogenome of Z. cristatus contains twelve PCGs (ND1, ND2, COI, COII, ATP8, ATP6, COIII, ND3, ND4L, ND4, ND5, and CYT B) arranged on the H-strand and just the ND6 gene on the L-strand. A total of 3,710 codons, excluding the stop codons, are used in the Z. cristatus mitogenome. When the PCGs start codons were analyzed, most genes presented ATG as the standard start codon, except for COI, which uses GTG (Table 1). Our data also revealed several incomplete stop codons: five T (ND2, COII, ND3, ND4, and CYT B) and two TA (ATP6 and COIII) stop codons on PCGs.

The analysis of relative synonymous codon usage (RSCU, Supplementary Table 7) revealed that almost all codons are present in our sequenced mitogenomes, except for AGA, which codes for arginine. Thus, the most frequently used codons were, in decreasing order, CGA (Arg) (RSCU value 2.34), CUU (Leu) (RSCU value 2.01), and CGC (Arg) (RSCU value 1.79), whereas AGG (Arg) (RSCU value 0.08), CCG (Pro) (RSCU value 0.19), ACG (Thr) and GCG (Ala) (RSCU value 0.24) were rarely used. The frequencies of the RSCU are graphically represented in Fig. 3B.

Selection analysis of protein-coding genes (PCGs)

In this study, which included 42 fish species, the predicted best evolutionary model explained by multiple sequence alignment resulted in a general time reversible (GTR) model with an unequal aminoacidic frequency (+ F), a discrete gamma model with 4 categories (+ G4) and a proportion of invariable sites (+ I). This model was applied for the reconstruction of the maximum likelihood tree generated by RAxML and is shown in Figs. 4 and S2.

Fig. 4
figure 4

Maximum likelihood tree of mitogenomes belonging to 42 fish species generated via RAxML software. The members of the family Trachipteridae were highlighted in green; the members of the family Lampridae were highlighted in yellow; the members of the family Regalecidae were highlighted in red

Full size image

EasycodeML analysis of the 13 mitochondrial protein-coding genes of the Lampriformes group revealed that only 4 of these genes are under positive selection: cytb; nd4; nd5; and nd6 (p values, posterior probabilities and positively selected sites are shown in Table 3).

Table 3 Mitochondrial genes under positive selection within the Lampriformes group, with amino acid substitutions relative to the Z. cristatus sequences. # refers to Z. cristatus amino acid mutations, excluding other Lampriformes sequences
Full size table

Pairwise distance

The number of base substitutions per site among the gene sequences is shown in Supplementary Tables 8 and in Fig. 5. The analysis of the p-genetic distance among all PCGs revealed that the NAD group genes are more divergent than the COI and CYTB genes are (Fig. 5).

Fig. 5
figure 5

Overall mean p-genetic distance ±SD in the PCG comparison of Z. cristatus and the other 42 species selected. The number of base substitutions per site from the average over all sequence pairs is shown. The p-distances are on the y-axis, whereas single PCGs are on the x-axis. Circles and asterisks identify the outliers

Full size image

The ND genes presented a relatively high degree of sequence variability among the studied species. Nevertheless, only ND3 and ND4 presented relatively high sequence conservation rates, 0.401 and 0.460 (p-distance), respectively.

Phylogeny of the mitochondrial genome of Zu cristatus

Owing to the uncertainty in the historical taxonomy of Lampriformes species, molecular analysis is essential to better evaluate the taxonomic status of these taxa, which will likely be subject to revisions in the future.

Phylogenetic relationships of Lampriformes based on mt-co1 sequences

Here, we report the results of the phylogenetic reconstruction of Lampriformes based on the currently available cytochrome oxidase subunit 1 (mt-co1) gene sequences (Supplementary Table 3). The data revealed, with high posterior probability support, that the topology of the tree only partially matches the currently accepted taxonomic relationships among the families. A graphical reconstruction obtained with iTOL is shown in Fig. 6. The Supplementary Table 9 reports the evolutionary divergence among the 21 mt-co1 sequences analyzed as p-distance.

Fig. 6
figure 6

Unrooted radiation tree illustrating the phylogenetic relationships of Lampriformes species based on the currently available mt-co1 sequences. The scale bar corresponds to an estimated evolutionary distance of 0.1. The tree was generated via the iTOL utility. The members of the family Trachipteridae were highlighted in green; the members of the family Lampridae were highlighted in yellow; the members of the family Regalecidae were highlighted in red

Full size image

Our analysis revealed that the members of the Lampridae and Veliferidae families clustered on two close monophyletic branches (estimated p-distance 0.196). In contrast, the three species of the Regalecidae family were grouped into two separate branches from the same node, one related to the two species of the genus Regalecus and the other consisting only of A. parkeri (branch length 0.271). The other three branches originated from the same node, two of which are related to the Trachipteridae genera Trachipterus and Zu, and the other branch concerning the species E. fiski resulted separately from the other Lophotidae (estimated p-distance 0.196). Similarly, the two species of the genus Desmodema clustered separately, but closed to the rest of the Trachipteridae genera (estimated p-distance 0.198). T. arcticus grouped separately from the congeneric species (branch length 0,335) on a shared branch with Regalecus species and originated from the same node as the rest of Trachipterus.

Phylogenetic relationships between Zu cristatus and related taxa based on mitogenome sequences

Here, we report our results from a phylogenetic inference analysis in which our complete Z. cristatus mitogenome was compared with 43 fish mitochondrial genomes of related species, plus one outgroup (Supplementary Table 2). The resulting tree was visualized with iTOL v6.5.6 [59] (Fig. 7).

Fig. 7
figure 7

Unrooted radiation tree inferred via JolyTree via 43 mitogenomes downloaded from the MitoFish database (Supplementary Table 2). The scale bar corresponds to an estimated evolutionary distance of 0.1. The tree was generated via the iTOL utility. The members of the family Trachipteridae were highlighted in green; the members of the family Lampridae were highlighted in yellow; the members of the family Regalecidae were highlighted in red; the members of the family Lophotidae were highlighted in light blue; the members of the family Veliferidae were highlighted in orange

Full size image

Our results revealed that the trachipterids Z. cristatus and T. trachypterus grouped, with high bootstrap values, on the same branch, followed by R. glesne and L. guttatus, respectively. In addition, according to the present taxonomy, our analysis confirms the current position of the Stylephoridae, Cetomimidae and Atelopodidae taxa out of the Lampriformes order. The clustering of the species D. rerio, employed as outgroup, confirmed the validity of our analysis.

Discussion

The study of mitochondrial DNA is of fundamental importance for reviewing and establishing proper phylogenetic relationships in bony and cartilaginous fishes [60]. The features of this molecule make it essential in population biology and stock assessment. Moreover, the functional role of several mitochondrial genes covers essential aspects of bony fish adaptation to bathymetrical gradients or extreme environments, such as thermal tolerance or respiratory capacity in oxygen-limited conditions [61]. Indeed, evolutionary adaptations to deep-sea life appear to have occurred independently in at least 22 fish orders [28].

In general, the mitochondrial genome (mtDNA) structure of bony fishes consists of a small circular double-stranded DNA molecule (approximately 15–20 kb in length) containing 37 functional subunits, 13 protein-coding genes (PCGs), 22 transfer RNAs (tRNAs), and two ribosomal RNAs (rRNAs) [62]. Moreover, this molecule contains a particular control region called the “D-loop”, which contains the origin of heavy-strand (H-strand) replication. It is characterized by highly repeated sequences, AT contents and high variability in teleost, with a mutation rate five times higher than that of the rest of the organelle genome [63]. However, the high conservation rate and absence of recombination processes in some mitogenome PCGs (i.e., COI genes) and rRNAs support their use as standard markers for evolutionary and taxonomic studies in teleost species [8]. Moreover, studying the polymorphisms of these genes could lead to interesting insights from an adaptative point of view when comparing organisms from different habitats. The increasing use of mtDNA gene sequences in populations and evolutionary studies led the MitoFish database [43] to collect all the annotated mitochondrial sequences of fishes in a single repository.

Within Lampriformes, annotations of whole-genome or complete mtDNA are very scarce [22]. With respect to the scalloped ribbonfish, the literature contains only one incomplete mitogenome annotation (without the D-loop region), which was made by Miya and colleagues in 2001 [8]. Therefore, to better understand the evolution and phylogenetic relationships of the various taxa within the Lampriformes order and the Trachipteridae family, enriching the current literature with complete and well-annotated mitogenome sequences of other species is fundamental. The current state of the art is affected by the limitations of professional fisheries, which focus on a restricted number of commercially important species [35, 64]. Lampriformes species play a confirmed ecological role in mesopelagic trophic webs. They arouse the scientific community’s interest concerning their adaptations to life in deep-sea environments. Indeed, the process by which several Lampriformes species pass from the shallow waters in which they live in juvenile stages to the high-depth oceans in adult life hides adaptive mechanisms that are still largely unknown [65].

The family Trachipteridae, order Lampriformes, is the most widespread in the Mediterranean Sea, with the confirmed presence of at least three species: T. arcticus, T. trachypterus, and Z. cristatus [22]. Nevertheless, their knowledge is still scarce, especially from a molecular point of view, for which our data make the first-ever contribution. This is essential, especially for species such as Z. cristatus, which undergo metamorphosis during their life cycle from the juvenile stage to the adult stage. Within the Mediterranean basin, this fish species is often caught accidentally, as a bycatch, by professional fisheries through deep-sea longlines and mid-trawling nets, but this occurs very rarely [30, 33]. Owing to its bathymetrical migrative nature, this species has been reported from shallow water in the juvenile stage to deep-sea environments as adults [33]. The scalloped ribbonfish is characterized by a striking shape and color and a particular head-up swimming style [31]. Adult scalloped ribbonfish commonly feed on small fish and cephalopods [66]. The few records of Z. cristatus in the Mediterranean basin report its widespread presence, with an apparent prevalence in the Central Mediterranean Sea [19, 30]. Because of the scarce literature, the accidental fishing of scallops seems to have occurred more frequently during the spawning season [67].

Generally, fisheries waste includes rare and elusive fish that are not commonly detected with professional fishing technologies; therefore, the results of scientific investigations are crucial for their detection. In this sense, the number of field studies involving eDNA detection is increasing, opening new frontiers in the detection of living organisms from challenging habitats. This innovative approach, however, requires a well-stocked database of highly shared genes or genomes such as mitochondrial DNA. Hence, the contributions of new mtDNA sequences to databases are essential and will be helpful for researchers in this area to move from a single-gene metabarcoding approach to multigene metagenomics, providing more accurate detection of biodiversity.

Our genome reference represents the first complete mitochondrial DNA for Z. cristatus. The patterns of similar gene positioning in the H- and L-strands are comparable to the mtDNA structure of other teleost species, both freshwater and saltwater, with complete and annotated genome sequences [68]. By analyzing the asymmetry of the molecule, our findings follow the pattern of nucleotide composition observed in other teleost species, such as Cyprinodon rubrofluviatilis and Kryptolebias marmoratus, which presented both negative AT and GC skews [69], but are not consistent with other species, such as the Sciaenidae species Johnius belangerii and Argyrosomus japonicus, which presented negative AT skews and positive GC skews and vice versa, respectively [70]. The strand asymmetry of mostly metazoan mitogenomes often results in positive AT skew values, which reveal more A than T on the strand, and negative GC skew values, which indicate more C than G [71]. Despite this, it is not rare that this strand bias is inversed, with negative AT skews and positive GC skews on the majority strand, or as in our case, both are negative. This trend is attributable to an alteration, often an inversion, of the replication origin (ROI) located in the control region (D-loop), resulting in altered strand asymmetry [72]. However, owing to the low shared contents of G, it seems to be a generalized feature of the fish mitogenome to have a greater AT percentage than that of GC and, in the same manner, that of AC and TG [73]. The asymmetry analysis of PCG concatenated regions revealed negative AT skews (− 0.1850) and negative GC skews (− 0.2070); a similar trend was recently reported for Pseudocaranx dentex by Li and colleagues [74]. The almost equal proportions of T and G to A and C in this region follow the values of asymmetry obtained in several mitogenome fish studies of various ecological and biological habits, representing a generalized trend [74, 75]. Additionally, all PCGs showed negative AT skews and, in most cases, negative GC skews, except for ND6. These data are in accordance with studies on species with similar habits to those of Z. cristatus [76] but contrast with some studies on Anguilliformes fish of the Muraenidae family [77], highlighting different skewness patterns, probably concerning their behaviors.

The mitogenome of Z. cristatus comprises a set of 22 tRNAs, with features in total length, AT contents and AT/GC skews comparable to recently reported data for freshwater species such as Gobiobotia naktogensis [78] but contrasting findings on Sinorhodeus microlepis, which reported an unusual negative AT skew on tRNAs [79]. Individually, the range of tRNA lengths was wider than that normally reported for teleost, particularly due to the abnormally short length of tRNA-Phe (58 bp), which contrasts with the common values of 68–70 bp reported for this value in several species [80]. Further studies are needed to better understand this strange feature of the Z. cristatus mitogenome, which represents the lowest length value for this tRNA in bony fishes, value confirmed in the first incomplete annotation of the Z. cristatus mitogenome [8]. The 16 S and 12 S rRNA lengths, AT contents and skews were within the normal ranges for marine teleost. Additionally, studies in the literature reported comparable values for marine teleost [74], whereas freshwater species commonly reported relatively high rRNA AT content values [81].

The control region (D-loop) of Z. cristatus annotated in this study represents the first region recorded for this species. The features of this region were interesting; indeed, the control region of the few bony fishes for which this sequence was annotated presented a relatively high AT content. Moreover, the length of the D-loop region sequenced in our study was among the longest in the literature, comparable to that reported for Dicentrarchus labrax, which was first reported at approximately 2.5 kb by Cecconi et al. in 1995 [63] and more recently reported by Tine and colleagues, with a length of 1950 bp [82]. Several other authors have reported very short control regions compared with these values, ranging on average values of 900 bp [83, 84], with rare cases of longer sequences, such as Pagrus major reported by Xia et al. [85] or Epinephelus epistictus recently reported by Vella N. and Vella A [86]., together with related species that had shorter lengths of this sequence. One factor that should not be underestimated is that P. major represents the species most similar from a biological and ecological point of view to Z. cristatus among the cited literature. Moreover, D. labrax has aspects of its life cycle that allow it to adapt to somewhat different environments, even if, in this case, mainly for salinity variations, in a similar way to our species. Hence, these two species have significant control region length values that are comparable to those of Z. cristatus. However, further investigation is needed, as very few mitogenomes in fish have complete D-loop sequences, so data are still lacking. As reported by Cecconi et al., these variations could be attributable to the high percentage of repetitive regions in this mitogenome [63]. Despite the novelty of our complete annotation, the absence of a sufficient number of annotated control regions among Lampriformes order species led us to exclude this region from our phylogenetic analyses, avoiding biases. Extending structural analyses in teleost species in this mitogenome region is required before proposing phylogenetic in-depth analyses on this particular order.

Our data concerning the 15 noncoding intergenic regions (IGNs) found in the Z. cristatus mitogenome were affected by the large size of Ala IGN (130 nucleotides), which altered the total length and size range of these regions. Indeed, IGN sizes commonly range from 1 to 20 bp, with 40–50 maximums not uncommonly reported [87]. However, owing to their assessed variability and polymorphisms, which are influenced by many factors, such as the sex of the sample, even for these features, a vast basin of annotated sequences could reveal interesting insights [88]. The identified intergenic regions were correctly identified (no variants were found after mapping of reads against the mitogenome), supporting the correctness of the data presented in the manuscript and opening to further perspectives. The features of the PCGs resulted in the same general pattern reported for several bony fish sequences [68].

Like other studies on bony freshwater [69] and saltwater [76] fish, uncommon start codons are very rare in bony fish, as the CCT was recently reported in Mobula tarapacana by Chandrasekaran et al. [89]. In contrast, stop codons are more variable in teleost, despite the complete stop codon TAA being the most common among PCGs. The expression of complete TAA stop codons is likely due to posttranscriptional polyadenylation events. However, the recording of incomplete stop codons (TA/T) is typical in bony fishes, as confirmed by our data on Z. cristatus, which shows seven incomplete stop codons on PCGs, following several other authors [69, 76]. Additionally, UAG is a common stop codon shared by several species on ND1 [73, 84] and, more rarely, on COI [90], confirming our findings in Z. cristatus. Other authors reported the UAG stop codon even for ND3, ND5, and ND6 in other species [91]. Our data show that the stop codon AGG on ND6 is rarely reported in bony fishes and is more frequent in freshwater species such as Caracidae [81] and in M. tarapacana [89]. Another rare stop codon is the CCT codon, which was recently reported for ND4 in M. tarapacana by Chandrasekaran et al. [89]. The RSCU trend of Z. cristatus was like that of Carangiformes [74], Pleuronectiformes [92], and freshwater species such as Cipriniformes [93]. UGA represents the most used stop codon in our sequenced genome. This codon, which is generally related to W synthesis, sometimes in fish species, represents a stop codon [94].

Research on fish mitogenome DNA has led to significant innovations in species identification and population identification [8]. Most of the studies carried out in this field have used MT genes as markers for species identification; specifically, the most used following species-specific DNA sequences are 16 S and 12 S rRNAs, cytochrome b (CYT B or Cyt b), and cytochrome c oxidase I (COI or mt-co1) [95, 96]. COI has been widely used because of its moderate variability in nucleotide composition and role as a “DNA barcode” for species identification [81]. Despite this, several studies have demonstrated that the use of only COIs in strictly related species may be limited [97, 98]. COI markers may slowly evolve, resulting in low nucleotide sequence distances in some taxonomic groups, thus preventing specific discrimination of closely related species [99] when the gene does not contain effective regions for barcoding applications [100]. Therefore, COI barcodes are not enough to indiscriminate species identification results, especially in some cases [101]. In these terms, identifying MT markers or conducting a multiple-marker approach is clearly important. This situation has led to the formulation of the proposal to study and analyze the complete mtDNA sequence to identify mitochondrial markers or multiple marker approaches [102] with increasing interspecific divergence.

Data on mtDNA are scarce and incomplete for the Lampriformes order, making species identification difficult. To create a baseline of suitable markers for Lampriformes ID, we performed the following analyses on the 13 PCGs of the Z. cristatus mtDNA sequenced. The pairwise distance method lies at the basis of the interspecific divergence and phylogenetic tree reconstruction. The value of the p-distance index provides information on the affinity among tree-like compared organisms. The kinship between species, specimens or populations is enhanced by genetic distance. Studying the pairwise distances among different species via single mitochondrial PCGs can help define the most suitable marker for generating phylogenetic trees. As expected, and confirmed by phylogenetic analyses, the minimum pairwise distances recorded were between Z. cristatus and T. trachypterus for all the mitochondrial PCGs (green highlighting in Supplementary Table 8). While the maximum pairwise distance site detected for each gene among the species selected resulted from high variability, greater pairwise PCGs were detected between Z. cristatus and D. rerio. Our results revealed that the CO family genes presented a relatively high conservation rate among the different analyzed species, confirming the role of these genes in the identification of relatively high numbers of taxa during barcoding. Thus, the ND family presented a greater p-distance, which could be targeted for primer design evaluation when species discrimination is needed, as it could provide species-level information on selected species. This highlights the need to explore better potential markers for species identification, especially regarding ND genes. In fact, for the Sparidae family, useful markers for precise species identification are the ND2 and ND5 genes [101]. Our results corroborate the importance of whole mtDNA analyses in creating unequivocal and compelling identification markers in teleost species.

The taxonomy and systematics of the order Lampriformes are still affected by the absence of a wide data basin, especially from a molecular point of view, which currently represents the main goal of research in this area [22]. For the rarest families, few or incomplete descriptions are associated with morphological variation, and the similarity between Lampriformes species leads to some uncertainty in phylogenetic relationships [18]. Our phylogenetic analysis, which is based on all the available mt-co1 genes of Lampriformes species, revealed that although we have chosen the most common gene used for this type of analysis in teleost, the remaining six species included in the order (L. guntheri, L. machadoi, R. elongatus, R. kessinger, T. fukuzakii, and T. ishikawae) still have no annotated mt-co1 sequences, highlighting the strong lack of data on the subject and the need to explore this topic further. No data are available for the entire family of Radiicephalidae. Hence, it was not possible to include them in our analysis. In contrast, all the species of the Lampridae, Veliferidae, and Regalecidae families presented mt-co1 sequences available in NCBI. Our results provide interesting insights into members of the genus Lophotus, which are somewhat distant from the species E. fiski. This grouping represents the most important difference between our analysis and the accepted phylogeny of the order. Interestingly, T. arcticus grouped separately from the congeneric T. arcticus, on a shared branch with Regalecus species, and originated from the same node as the rest of Trachipterus. We can obtain the most interesting insight from our analysis of the position within the Lophotidae and Trachipteridae families. Further in-depth analyses are needed to better assess these relationships, especially those between quite different organisms, such as Lophotidae, both morphologically and biologically [103].

Our complete mitogenome phylogenetic analysis revealed that, within the Lampriformes order, the grouping is comparable to the previous grouping, which referred to mt-co1 sequences. Moreover, these results align with previously reported classifications for the order [104], particularly under the most recent one made by Yu and colleagues in 2019 related to the R. glesne mtDNA annotation. The difference between the two Z. cristatus sequences analyzed is attributable to the presence in our new annotation of the D-loop portion, which was absent in the annotation of Miya et al. 2001 [8]. The position of S. chordatus, which is currently a species of the order Stylephoriformes but once part of the current order Lampriformes as a separate family, is very interesting [36]. Our analysis confirmed the position of Stylephoridae, and despite its group being the nearest taxon to Lampriformes, distance assigned it to a different order [105]. Similarly, other taxa that were previously classified in the order Lampriformes, such as the existing Cetomimidae family of the Bercyformes order and the Atelopodidae family of the Atelopodiformes order, were analyzed [106]. With respect to the first, previously classified as the Mirapinnidae family within the Lampriformes order, the ten species analyzed presented strong relationships among them as well-assessed taxa within the tree. Indeed, the strict groupings of species of the genera Gyrinomimus and Cetomimus were in line with the classification of Colgan et al. [107] based on 16 S rDNA data, as well as the distance from the genus Cetostoma. Our data are also in accordance with the bGMYC analysis of Weber (2020) [28], which reported, in their maximum-made credibility tree for the family, the genera Danacetichthys, Cetostoma, and Procetichthys grouped in three separate branches in this order, far from the remaining Cetomimids. With respect to the Ateleopodidae family, our analysis clearly revealed relationships within the taxon for the genera Ateleopus and Ijimaia, which were grouped nearest the Cetomimidae species and far from the Lampriformes ones. These findings contrast with the older classifications [106], which are mainly based on morphological evidence, such as those reported by Sasaki and colleagues in 2005 [108]. They studied the structure of the ethmoid region and proposed a close relationship between Ateleopodidae and Lampriformes taxa. Following our results, several molecular pieces of evidence have recently contradicted these findings on a morphological basis [109]. Our findings concerning the order Aulopiformes are in accordance with those of Tan and colleagues [110] about recent mt-co1-based relationships of the genus Saurida; in the same manner, the results of Zhang and Xian [111] confirmed the relationships within the Saurida genus and those with Harpadon and Synodus, which, in our analysis, were grouped according to their classification. The slight distance found in our analysis between these genera and the Chlorophthalmus species supported the classification of Ota et al. [112] proposed in 2000 for the Order and based on Cyt b sequences. The last considered order of Mictophiformes, like Aulopiformes species, comprises meso- and bathypelagic fish involved in the same trophic webs of Lampriformes. Our analysis revealed a correct grouping of these species within them, supported by high bootstrap values, partially in accordance with the previous classification based on the morphological position of photophores, a species-specific feature of these fishes, proposed by Denton and Adams in 2015 [113]. Indeed, they proposed a stricter phylogenetic relationship between the genera Benthosema and Myctophum, whereas our analysis revealed a stronger relationship among the genera Myctophum and Electrona; however, the distances are very small in both cases, and these three genera are, in accordance with the two reconstructions, grouped quite far from Diaphus. The positions of the genera Triphoturus and Lampadena are in accordance with the reconstruction of Paxton [114], whereas the grouping of the Scopelengys and Neoscopelus genera requires further analysis for confirmation. With respect to the other taxa included in our analysis, supported by the grouping of the two outgroup species, the three orders Carangiformes, Scombriformes, and Spariformes are all commercially important orders in the Mediterranean basin and are sometimes captured with any rare Lampriformes. These genera have clustered accordingly with the existing classification with high bootstrap values, both within the three orders and with the rest of the groups analyzed in our tree [105, 109].

Our positive selection analysis of PCGs revealed that when we compared Z. cristatus with all other mitogenomes (including Lampriformes), we found no statistically significant site (p value > 0.05), probably because of the high genetic similarity within the Lampriformes group, which was included in the background by the analysis. As expected, when Lampriformes were removed from the background dataset, EasycodeML detected 4 statistically significant codons in the positively selected ND6 gene: G45Q, A112N, S147A, and F156G (p values, posterior probabilities and amino acid mutations are shown in Table 3), indicating that, specifically for Z. cristatus, these are the only sites that underwent positive selection. Despite fishes are poorly studied compared to terrestrial vertebrates, patterns which denote the selection of mitochondrial genes have been reported by some authors. Silva et al., 2014 attributed these alterations in European anchovy (Engraulis encrasicolus) to temperature clines [115], while the surface temperature seems related to the CYTB selection in Japanese sand lance (Ammodytes personatus) [116]. Also, fish commonly found in transitional and extreme environments, such as the mummichog (Fundulus heteroclitus) show this phenomenon, related to osmotic alterations (Brennan et al., 2016). Mukundan et al., 2022, showed positive selection in the genes ND5 and ND6 for some mesopelagic fish orders, highlighting functional and structural roles of the latter in the whole complex I organization [117], already hypothesised by other authors [118]. Changes in these subunits may affect other important respiratory-chain functions in vertebrates [119, 120]. Considering our limited data, more in-depth studies on these ND6 substitutions are required to understand their functional consequences and potential value for the Z. cristatus metabolism. On our current knowledge, the vertical migratory habits of the species could be related to this positive selection, as reported for other teleost species with an active lifestyle.

Conclusion

Given the ecological relevance of mesopelagic and bathypelagic organisms and the complexity of deep-sea environments, the phylogeny, genetic adaptations, and distribution triggers of rare mesopelagic species require broader and more complete knowledge. Our study aimed to investigate the mitochondrial genome structure and features of Z. cristatus in detail, annotating its first complete reference. Our analyses revealed a single circular mtDNA molecule of 17,450 bp in length. A total of 37 genes, including the first D-loop region for this species, were identified. The asymmetry was negative for both AT skews and GC skews, whereas the AT content was 56.4%. We also detected the presence of 15 small, noncoding, intergenic nucleotide (IGN) regions and some rare stop codons in bony fishes. Pairwise distance and phylogenetic analyses against a list of 42 other bony fishes provided some interesting insights. Indeed, new questions about the phylogeny of Lampriformes have arisen related to the genera of the families Lophotidae and Trachipteridae, particularly to the positions of E. fiski and T. arcticus, which are grouped separately from congeneric Lampriformes. Directional selection analysis revealed that four sites in ND6, among all PCGs, underwent positive selection. Overall, these findings highlight the need to elucidate the genetic features of bony fishes in relation to their deep-sea adaptation, with a focus on rare and interesting species.