Monophyly of myctophiformes and the two families
This study recovered a monophyletic Myctophiformes and monophyletic families with Neoscopelidae as the sister family of the highly diversified Myctophidae (Figures 4 and 5; [3, 5, 81, 82] although see Rosen ). We found Scopelengys tristis was a sister taxon to the remaining neoscopelids, including the monotypic endemic Sulu Sea taxon Solivomer arenidens, in congruence with morphology (Figure 2). Stiassny  presented characters in support for Scopelengys as sister taxa to the remaining neoscopelids, although she did note a possibility of apparently derived characters in Neoscopelus and Solivomer being secondarily reduced in Scopelengys. Analyses from the present study are congruent on this matter although support values are alarmingly low concerning the phylogenetic position of Solivomer arenidens. We found Notolychnus valdiviae to be the sister taxon to the remaining myctophids; however, morphological character states of Notolychnus such as dorsally located photophores, small larvae without pigmentation and a reduced otolith, have been extremely difficult to interpret concerning early myctophid divergence. Photophores are present in all myctophids except Taaningichthys paurolychnus and present only in one Neoscopelidae genus Neoscopelus. Our results show both of these genera nested within their respective families and the evolution of photophores within Myctophiformes is difficult to assess, i.e. whether photophores have evolved once or twice within ancestral myctophiforms. This question is pertinent from morphological considerations on photophore structure in Myctophidae and Neoscopelidae, as it is quite different in the two groups [85, 86]. The phylogenetic positions of Scopelengys/Solivomer and Notolychnus in this study revives an interesting observation by Paxton  concerning parallel evolutionary trends for Solivomer/Lampanyctinae and for Neoscopelus/Myctophinae, regarding changes in jaw length and characters possibly associated with these changes. These characters have been difficult to interpret in terms of polarity and this study supports long jaw length as the plesiomorphic state in Neoscopelidae and Myctophidae from the basal branchings of Scopelengys/Solivomer and Notolychnus. Also, Stiassny  noted that important myctophiform character states like elongated jaw length and raised photophores show ambiguity concerning early myctophiform divergence.
A monophyletic Myctophidae was strongly supported by mitogenome sequences, a WANYC gene order and an INC-spacer between tRNA-LeuUUR and ND1 (Figures 3 and 4). Rearrangement of the canonical vertebrate WANCY gene order was shared by all myctophids with the tRNA-Cys relocated downstream resulting in a synapomorphic WANYC gene order (Figure 5). A prolonged sequence (putative OL + INC-sequence) in the canonical position of the putative origin of L-strand replication was found within all myctophids (see Additional file 4), although fishes within the tribe Electronini were difficult to determine due to additionally rearranged gene orders in this region (Figure 3). An INC-region between the tRNA-LeuUUR and ND1 also represents synapomorphic evidence for Myctophidae (Figure 3). The size of this spacer ranged from 45–79 base pairs and we found no obvious tendency regarding size change for this INC-region (see Additional file 4). Neither the OL + INC-region nor the LeuUUR/ND1 spacers showed sequence similarity in any species to any other mitochondrial genes. Neither could they be folded into clover shape secondary structures, so their origins are elusive. On the other hand, assembly of the INC-spacers showed clear sequence similarity within genera indicating the INC-regions to be inherited synapomorphic characters within the family. Smith et al.  reported a length expansion for a similar small INC-sequence located between tRNA-Val and 12S rRNA in a clade of serranid fairy basslet fishes (Serranidae). They showed the INC-sequence to be homologous for all genera from simultaneous analysis of nucleotides and gene order – a very difficult task for myctophids considering multiple gene order rearrangements and putative associated INC-sequences. The third myctophid synapomorphy from this study is an INC-region between ATP6 and CO3 ranging from 288 base pairs in Notolychnus to 9–43 base pairs in the remaining myctophids included (see Additional file 4). No conserved sequence motifs were found in this region. This is a peculiar spacer in the sense that the canonical vertebrate gene order for this region is ATP6/CO3 with no tRNA gene acting as punctuation mark for RNA processing.
Monophyly of the two subfamilies and five tribes
The monophyly of the two subfamilies Lampanyctinae and Myctophinae was recovered from all analyses, although the low support values associated with this split are incomprehensible from morphological characters unambigously delimiting these two groups. As with the Myctophidae, the two subfamilies are fully supported by adult and larval morphology and no discussion is needed – character states are presented by Moser & Ahlstrom , Paxton  and Stiassny . A well supported result from this study, differing from morphology, is the phylogenetic position of the tribe Diaphini within Lampanyctinae, resolved as sister tribe to the other lampanyctines. Adult and larval morphology within Lampanyctinae show incongruences and both hypotheses are different compared to our molecular results (Figures 2, 4 and 5).This shows problematic morphological characters, lampanyctine larval characters are noted as few and tentative in previous larval studies [30, 88] and of little use above genus level, as well as molecular characters, no unique gene orders connect tribes and the mitogenomic phylogenetic sequence information is insufficient, as witnessed from the three different resolved Lampanyctinae topologies from this study (Figures 4 and 5).
Monophyly of the tribes Notolychnini (clade A in Figures 4 and 5), Diaphini (B), Gymnoscopelini (C), Lampanyctini (D+E+F), Electronini (G) and two subclades of Myctophini (H and I+J) are strongly supported from all mitogenomic DNA sequence analyses and various mt gene orders discussed below. Only the node delimiting the tribe Diaphini (B) show relatively low support values. The tribe Gonichthyini is tentatively placed from a single taxon as nested within a Myctophini subclade consisting of ctenoid-scaled species. Mitogenomic DNA sequence data from this study supports recognition of 10 mutually exclusive clades within Myctophidae (Figures 4 and 5).
Phylogeny of tribes and genera
Students of lanternfish phylogenetics will note the position of Notolychnus (clade A) resolved as sister taxon to the remaining family Myctophidae from mitogenome sequences. Paxton  tentatively placed Notolychnus as sister taxon to the subfamily Lampanyctinae (Figure 2A). Subsequent parsimony analysis of the same data by Paxton et al. , including additional larval characters [27–29], placed Notolychnus in an unresolved Myctophinae-Notolychnini-Lampanyctinae trifurcation (Figure 2D). Subsequently, Stiassny  added four additional characters and reanalyzed the aforementioned data with a similar result, showing Notolychnus as the sister taxon to the remaining lampanyctines (Figure 2E). Most recently, Yamaguchi  scrutinized previously published characters (Figure 2A-E) once again and reanalyzed the data, resulting in a largely unresolved myctophid tree (Figure 2F). However, the position of Notolychnus was congruent with Paxton  and Stiassny . The novel phylogenetic position of Notolychnus from this study, as the basal branching in Myctophidae, highlight the past issues concerning subfamilial affiliation and partly explains the lack of phylogenetically informative morphological characters in relation to the two subfamilies.
A monophyletic Diaphini (clade B) was well supported from both sequences and a unique IMQ-gene order (Figure 3), all in complete agreement with morphology, showing Lobianchia (2 species) as sister group of the very speciose Diaphus (78 species) [89–91]. Four species-groups of Diaphus have been suggested from the presence of Dn-Vn, Ant, Suo and So photophores (for photophore terminology, see Figure 1) ; however, our five species of Diaphus included were insufficient to validate anything on this matter. One very rare monotypic taxon Idiolychnus urolampus, missing from our study, was transferred from Diaphus to Lobianchia by Bolin  and most lately erected into its own genus Idiolychnus by the position of VO3 and possession of two SAOs .
Gymnoscopelini (clade C) sensu Paxton  was found to be monophyletic although not fully congruent with Ahlstrom et al. . Unfortunately, the monotypic taxon Lampanyctodes hectoris is not included in this study, a myctophid showing ambiguous morphological characters concerning tribal placement. Paxton  noted Lampanyctodes as a possible early divergence based on several characters shared with the most plesiomorphic diaphinid genus Lobianchia. Conversely, Ahlstrom et al.  noted Lampanyctodes as being specialized within Gymnoscopelini from a similar larval form and development to that found within Lampadena, arguing that characters reflect habitat instead of true phylogeny.
Ceratoscopelus and Bolinichthys formed a clade (D) as either sister group to the remaining lampanyctines or included within them (Figures 4 and 5, respectively). Paxton  included these genera within Lampanyctini as closely related, whereas Ahlstrom et al.  placed them within Gymnoscopelini. Our study supports inclusion in Lampanyctini despite unresolved intrarelationships of this tribe. The clade itself is very distinct in terms of mt sequences and supported by both osteology  and larval development .
A tRNA-Glu gene could be detected in the canonical vertebrate position between ND6 and Cytb in all myctophids included in this study, except in a Lampanyctini subclade (E) comprising Stenobrachius, Triphoturus, Nannobrachium and Lampanyctus (Figures 3). In Nannobrachium ritteri and Lampanyctus crocodilus, we were able to determine a novel location of tRNA-Glu in the highly polymorphic region downstream Cytb including the tRNA-Thr and Pro genes, the CR and various INC-regions. Sequence similarities between tRNA-Glu from N. ritteri and L. crocodilus and the INC-regions between ND6 and Cytb were evident from sequence comparisons and putative secondary structures (see Additional file 3). Interestingly, this INC-region showed no sign of redundancy since all five taxa have retained a fragment equal in size to the typical length of tRNA genes (68–73 base pairs). We tentatively assign this gene order to the lampanyctine subclade E and note that this clade is congruent with Paxton  and Paxton et al. , although the latter notes that no synapomorphic character is present to actually define the clade. We expect this feature to be found within the unsampled genus Parvilux as well. A new position of tRNA-Glu and retention of an INC-sequence between ND6 and Cytb has also been found in other vertebrates, e.g. the amphisbaenian reptile family Bipedidae , and seems to be one of the most common gene order rearrangements in vertebrates.
A clade consisting of Lampadena and Taaningichthys (F) is found monophyletic and completely congruent with morphology. Taaningichthys was previously a part of the genus Lampadena although it was separated by Bolin  based on characters such as reduced number of photophores and flaccid body structure. Nafpaktitis and Paxton  noted that Lampadena chavesi, L. dea and L. speculigera were more closely related to Taaningichthys than the remaining species of Lampadena based on expanded neural arches on the anterior vertebrae as found in Taaningichthys. In fact, Paxton  noted that discriminating characters between the two genera were hard to find. Lampadena and especially Taaningichthys are among the largest and deepest living myctophids, with the latter showing reduction of the lateral line and in the number of body photophores. This apparent reductional trend has resulted in one taxon, Taaningichthys paurolychnus, having lost all body photophores although retained the supra- and infracaudal glands. Lampadena yaquinae is found nested within the Lampadena species included and corroborates Paxton  in synonymizing Dorsadena with Lampadena. Likewise, the subgenera Lampadena and Lychnophora proposed by Fraser-Brunner , supported by only an elevated PO4, are not supported from this study. Increased mitogenomic taxon sampling of these two genera is interesting concerning evolution of photophores within these deep dwelling genera showing atypical character states within the Myctophidae .
The tribe Electronini (G), erected from low-leveled PVO and PLO photophores , composes a subset of the tribe Myctophini as recognized by Paxton . Our results clearly support Electronini as a separate tribe  from both DNA sequences and long (>200 base pairs) synapomorphic INC-regions upstream the tRNA-Ile gene (Figures 3 and 5). We recovered a sister relationship of Electronini to the rest of Myctophinae, a result also presented by Paxton et al. , although cladistic analyses of morphological characters failed to recover this result (Figures 2E-F).
The IQM-regions of species included in the genera Benthosema and Diogenichthys (clade H) were peculiar and the IQM-region and WANYC-region of both Electronini and Benthosema-Diogenichthys were highly differentiated compared to all other myctophids. Benthosema glaciale showed new positions of tRNA-Gln and -Met to the downstream region of the WANYC-region (Figure 3) representing a novel gene order in vertebrate mitogenomes. Diogenichthys atlanticus showed an IQMM gene order and the canonical vertebrate gene order was observed within Benthosema pterotum and B. fibulatum. Taxon sampling prevents us from discussing rearrangements for these taxa at this point; however, a close relationship between the two genera is consistent with morphology. Diogenichthys was separated from Benthosema by ventrally leveled Prc 1 and 2 in addition to various jaw differences and hooked dentary teeth, a character state only found within lampanyctines. Similar to Notolychnus, D. atlanticus is a diminutive species with adult maximum size < 30 millimeters and could represent a problematic branch within this clade considering character states related to miniaturization. Diminutive species are also found within the genus Diaphus. The Benthosema-Diogenichthys clade was supported from DNA sequences and an INC-region up- or downstream tRNA-Gly; all three species of Benthosema showed an INC-region flanked by an upstream CO3 gene and downstream tRNA-Gly. In addition, B. glaciale showed an INC-region after tRNA-Gly. In D. atlanticus, however, the INC-spacer was located downstream from the tRNA-Gly gene (Figure 3). Interestingly, two tRNA-Gly genes separated by a large INC-region were observed in this same region in Electrona antarctica, although none of the INC-regions could be shown to have any sequence similarity to each other or to tRNA-Gly. We have assumed an INC-sequence associated with tRNA-Gly as a character state for the Benthosema-Diogenichthys clade, although with reservations. In view of E. antarctica possesing two tRNA-Gly genes we note that an association of this region in E. antarctica, D. atlanticus and Benthosema spp. is plausible and should be elucidated with increased mitogenomic taxon sampling.
The tribe Myctophini was not supported from this study as the sister clade to Electronini, because the single representative of the tribe Gonichthyini (Centrobranchus choerocephalus) included in this study renders Myctophini paraphyletic. We note that broader taxonomic representation of the genera Gonichthys, Centrobranchus, Tarletonbeania and Loweina is necessary concerning phylogenetic position of Gonichthyini within Myctophinae. However, evidence from mt sequences and a unique gene order in part of Myctophini (Figure 5) is consistent with the results from Paxton  stating that”..A number of characters suggest that the Myctophum-Symbolophorus line gave rise to the slendertails [ed. tribe Gonichthyini]..”. Additional evidence for this result comes from the non-monophyly of Gonichthyini based on larval characters  (Figure 2C) and the analyses by Stiassny  and Yamaguchi  (Figure 2E-F) showing Symbolophorus a basal branching within Gonichthyini. This study confidently resolves the genus Myctophum into two different clades from mitogenome sequences and also from a unique gene order involving ND6, tRNA-Glu and Cytb (Figures 3, 5). Myctophum affine, M. punctatum and M. nitidulum (clade I) showed this particular gene arrangement whereas M. asperum and M. orientale showed the typical myctophid gene order also found in Centrobranchus choerocephalus and Symbolophorus californiensis, the latter four forming a well supported clade J (Figures 4 and 5). Two species-groups are currently recognized in the genus Myctophum based on cycloid versus ctenoid scales  corresponding to the two separated groups of Myctophum found in this study. Moser and Ahlstrom  discussed eye shape within Myctophum and noted this genus as difficult to distinguish phylogenetically, with stalked eyes usually absent, although present in e.g. Myctophum nitidulum and M. punctatum corresponding to the cycloid-scaled Myctophum lineage from this study. This result is, however, complicated by stalked eyes also being found in all species of Symbolophorus. Clearly, mitogenome sequences and a unique gene order in the cycloid-scaled species of Myctophum strongly suggest that the two groups are in fact different lineages within Myctophini and that the tribe Gonichthyini is nested within the ctenoid-scaled group. This indicates ambiguous relationships from adult and larval morphology or convergent evolution within Myctophini lineages. Paxton  noted that no additional characters are present within Myctophum to support a split between cycloid- and ctenoid-scaled Myctophum species. The term”near-surface” myctophids has been used for a group of myctophids represented within the genera Loweina, Tarletonbeania, Gonichthys, Centrobranchus, Symbolophorus and Myctophum with most species showing DVM behaviour and most being easily netted in surface layers at night time . Denser taxon sampling within Gonichthyini, and the ctenoid-scaled Myctophini group as recognized from this study, should clarify if DVM patterns and phylogeny are correlated.