Otomorphs (= otocephalans or ostarioclupeomorphs) revisited

The largest neopterygian clade, the Infraclass Teleostei, has undergone major changes in the last 20 years mainly due to the development of molecular techniques and studies that changed the content, position and phylogenetic interpretations of numerous clades that were historically based on morphological characters (Arratia, 2015); changes in the Teleostei tree include the recognition of new taxa and the deletion of some nominal orders and families. Such new approaches, plus the discoveries of new fossil taxa and re-evaluation of others, have given a new understanding in age, content and evolutionary changes of the Infraclass Teleostei of Betancur-R. et al. (2017) or the Division Teleosteomorpha of Arratia (2001, 2013, 2017) and Nelson et al. (2016) or the Subdivision Teleostei of Nelson et al. (2016), despite the assignment of this group to various high-level ranks. An excellent example of major changes revealed by molecular studies is the present content and classification of internal groups comprising the large supercohort Clupeocephala versus previous classifications in Nelson’s Fishes of the World (1994, 2006). Independent of the major developments and changes from molecular studies and the discovery of new fossils, morphological studies of extant and fossil taxa still play a major, unique role in providing the characters that support the monophyly of taxa and taxonomic diagnoses (Arratia, 2015). Elopomorpha is currently interpreted as the most primitive taxon and the sister of Osteoglossomorpha plus Clupeocephala in the crown group Teleostei (or Teleocephala of de Pinna, 1996 or Osteoglossocephala of Arratia 1999 or Osteoglossocephalai of Betancur-R et al., 2017). This phylogenetic hypothesis is supported by morphologiOtomorphs (= otocephalans or ostarioclupeomorphs) revisited


Introduction
The largest neopterygian clade, the Infraclass Teleostei, has undergone major changes in the last 20 years mainly due to the development of molecular techniques and studies that changed the content, position and phylogenetic interpretations of numerous clades that were historically based on morphological characters (Arratia, 2015); changes in the Teleostei tree include the recognition of new taxa and the deletion of some nominal orders and families. Such new approaches, plus the discoveries of new fossil taxa and re-evaluation of others, have given a new understanding in age, content and evolutionary changes of the Infraclass Teleostei of Betancur-R. et al. (2017) or the Division Teleosteomorpha of Arratia (2001Arratia ( , 2013Arratia ( , 2017 and Nelson et al. (2016) or the Subdivision Teleostei of Nelson et al. (2016), despite the assignment of this group to various high-level ranks. An excellent example of major changes revealed by molecular studies is the present content and classification of internal groups comprising the large supercohort Clupeocephala versus previous classifications in Nelson's Fishes of the World (1994, 2006. Independent of the major developments and changes from molecular studies and the discovery of new fossils, morphological studies of extant and fossil taxa still play a major, unique role in providing the characters that support the monophyly of taxa and taxonomic diagnoses (Arratia, 2015). Elopomorpha is currently interpreted as the most primitive taxon and the sister of Osteoglossomorpha plus Clupeocephala in the crown group Teleostei (or Teleocephala of de Pinna, 1996 or Osteoglossocephala of Arratia 1999 or Osteoglossocephalai of Betancur-R et al., 2017). This phylogenetic hypothesis is supported by morphologi-2 e180079 [2] cal and molecular studies (Arratia, 1997(Arratia, , 1999(Arratia, , 2013(Arratia, , 2017Near et al., 2012;Betancur-R et al., 2013Betancur-R et al., , 2017. Currently, the supercohort Clupeocephala sensu Arratia (2010) includes the cohorts Otomorpha (sensu Wiley, Johnson, 2010 andBetancur-R et al., 2017 or Otocephala of Johnson, Patterson, 1996 or Ostarioclupeomorpha of Arratia 1996Arratia , 1997Arratia , 1999 and Euteleosteomorpha (sensu Betancur-R et al., 2017 or Euteleostei sensu Johnson, Patterson, 1996). The taxon Clupeocephala was erected by Patterson, Rosen (1977) to contain the Clupeomorpha (e.g., engraulids, clupeids, and pristigasterids) and the Euteleostei (e.g., cyprinids, catfishes, salmonids, atherinids, and cyprinodontids), in other words, the majority of extant teleosts. This interpretation of Clupeocephala represented the approach of the time based on morphological evidence, and it was accepted until 1996, when our understanding of primitive clupeocephalans was changed by both molecular and morphological evidence.
Until the 1970s, ostariophysans and gonorynchiforms were interpreted as disparate groups within the Euteleostei, following mainly the phylogenetic hypothesis of Greenwood et al. (1966: their Division III). Gonorynchiforms were accepted as part of the ostariophysans after the work of Rosen, Greenwood (1970), and consequently, under this new understanding, the Ostariophysi continued being interpreted as Euteleostei, an interpretation that was also followed by Fink, Fink (1981), who assumed that ostariophysans were primitive euteleosts sensu Patterson, Rosen (1977). The interpretation of ostariophysans as euteleosts was supported by numerous morphological studies published between the 1970s and 1990s where ostariophysans appeared nested with esocoids, argentinoids, osmerids and/or salmonids (see Fig. 1a-h).
Although the inclusion of alepocephaliforms within the cohort Otomorpha has been proposed by different molecular studies, there is no morphological synapomorphy supporting the monophyly of Otomorpha comprising the subcohorts Clupei (= Clupeomorpha of Greenwood et al., 1966), Alepocephali, and Ostariophysi (but see below the section on Analysis of characters). Consequently, the main goal of this contribution is to search for and discuss potential morphological synapomorphies of this clade. A brief introduction on the three taxa is presented below.

Content of Otomorpha
The name Otomorpha was introduced by Wiley, Johnson (2010) as synonym of Otocephala of Johnson, Patterson (1996) and Ostarioclupeomorpha of Arratia (1997Arratia ( , 1999 so that in this context Otomorpha contains only Clupeomorpha and Ostariophysi. Betancur-R et al. (2017) retained the name Otomorpha, but included three main groups: clupeiforms, ostariophysans, and alepocephaliforms, whereas Nelson et al. (2016) retained the name Otocephala but added alepocephaloids. Thus, Otomorpha varies in content depending on time and authorship. Otomorpha is defined here as the clade containing clupeiforms, ostariophysans and alepocephaliforms ( Fig. 4a-g). The original taxonomic content of Otocephala or Ostarioclupeomorpha is retained here, i.e., the clade including only clupeomorphs and ostariophysans. Otomorpha is the most primitive clade in Clupeocephala with about 11,000 extant species, 1,500 extant genera, and 90 extant families, making it one of the largest clades among the crown-group Teleostei.
southern Mexico to the southernmost region of temperate South America, with an extraordinary diversification of siluriforms and characiforms, including gymnotiforms, which are unique to the region. While the cypriniforms are present in the southern part of North America and Central America, they are not native members of South America [A piece of a fossil bone was interpreted by Gayet (1982)  that has not been corroborated (Arratia, Cione, 1996;Fink, Fink, 1996)]. In contrast to most otophysans, clupeomorphs and alepocephaliforms are mainly or exclusively marine forms, respectively.
The oldest-known otomorph, †Tischlingerichthys (Fig.  5a), was recovered in marine sediments from the Upper Jurassic, Upper Tithonian of the Mörnsheim Formation, Bavaria, Germany (Arratia, 1997;Arratia, Schultze, 2015). Other comparatively younger otomorphs (Early Cretaceous) have been recovered in marine, estuarine and/or freshwater, including the oldest marine gonorynchid ( †Sapperichthys from Cenomanian strata of Mexico; Amaral et al., 2013) and a clupeomorph that has been interpreted as the most primitive within the group ( Fig. 5b; †Ranulfoichthys from marine Albian strata of Mexico; Alvarado-Ortega, 2014). In com-parison, Denticipitoidei is a younger group, with the oldest representatives- †Paleodenticeps tanganikae Greenwood, 1960 and †Paleodenticeps sp. (Fig. 6)-known from deposits in Mahenge, Tanzania, which are assigned an Eocene age (about 46 Ma; Harrison et al., 2001;Kaiser et al., 2006). Although gonorynchiforms were well represented in the Lower Cretaceous of the Neotropical region, being very abundant in Brazilian basins (for a review see Brito, Amaral, 2008), modern forms do not inhabit the Neotropical Region, except, perhaps, for Chanos, which has an Indo-Pacific distribution that extends from California to Ecuador (although this distribution may show strong periodic changes associated with the El Niño/La Niña currents).   Numerous important fossils representing some stem clupeomorphs (e.g., †Santanaclupea silvasantosi; Fig. 5c) and otophysans are known from Lower Cretaceous strata of the Neotropical Region, especially Brazil, and others with peculiar cranial morphologies -and still unknown phylogenetic relationships within siluriforms-are known from the Cretaceous-Paleocene of Bolivia (Arratia, Gayet, 1985).
Clupei. Clupei (sensu Betancur-R et al., 2017) or Clupeiformes are characterized by a series of characters, with the most conspicuous being an otophysic connection between the swim bladder and inner ear involving a pair of anterior extensions of the swim bladder that enters the skull through the exoccipital to connect with the utriculus of the inner ear, forming ossified bullae in the prootic, and usually also in the pterotic too; base of hypural 2 fused to ural centrum 2 (= of polyural terminology sensu Schultze, Arratia, 2013;Wiley et al., 2015) from early ontogenetic stages and an autogenous hypural 1 without connection with the vertebral column from early ontogeny; one or more abdominal scutes (including pelvic scutes), each of a single element that crosses the ventral midline; and development of dorsal scutes with a median keel. For other characters see Grande (1985), Di Dario (2004), and Di Dario, de Pinna (2006) and the compilation of synapomorphies by Wiley, Johnson (2010).
Ostariophysi. Ostariophysans are characterized by numerous features, with only a few listed here: basisphenoid absent; sacculi and lagena with a posterior position and nearer midline; dermal portion of the palatine (= dermopalatine) absent and represented only by the autopalatine; gas bladder divided into smaller anterior and larger posterior chambers; presence of a unique alarm substance in epidermis (absent in gymnotiforms, with electrical signaling); and presence of nuptial tubercles with well-developed keratinous caps. For other characters and comments see Fink, Fink (1981, 1996, , Lecointre, Nelson (1996), and Wiley, Johnson (2010).
Ostariophysans ( Fig. 4c-g) comprise ca. 10,400 species contained in about 1,350 genera, 80 families and five orders (Nelson et al., 2016) included in two sections, Anatophysa (sensu Betancur-R et al., 2017or Anatophysi of Fink, Fink, 1981, which contains Gonorynchiformes and Otophysa (sensu Betancur-R et al., 2017or Otophysi of Fink, Fink, 1981, which in turn, includes Cypriniformes, Characiformes, Siluriformes, and Gymnotiformes. Although there are differences between morphological and most molecular phylogenetic hypotheses of the orders, there is now consensus concerning the monophyly of the group (Fink, Fink, 1981, 1996Betancur-R et al., 2017;Arcila et al., 2017) and also of the otophysans. Major differences include: (1) a few molecular studies have questioned the monophyly and relationships of the Characiformes (Nakatani et al., 2011;Chen et al., 2013;Chakrabarty et al., 2017), but monophyly of the group and its position as the sister to Gymnotiformes + Siluriformes was recently supported by other molecular studies (Arcila et al., 2017;Betancur-R et al., 2017); (2) the family Diplomystidae appears as the most primitive catfish group in morphological studies of Siluriformes (e.g., Arratia, 1987;Grande, 1987), whereas Nematogenyidae plus other loricarioids is the sister to Diplomystidae plus all other catfishes in molecular studies (e.g., Sullivan et al., 2006;Arcila et al., 2017;Betancur-R et al., 2017). For other references on the subject see Arratia, Quezada-Romegialli (2017). Most extant ostariophysans are primarily freshwater fishes with a global distribution (except Antarctica, Greenland, and New Zealand), but there are about 120 marine species restricted to the Siluriformes (Nelson et al., 2016). Numerous fossils interpreted as possible stem taxa have been recovered in marine or estuarine strata, a fact that opens interesting questions concerning the evolution of the environment of Ostariophysi or of its particular orders.
Contrary to the current state of clupeiform research, ostariophysans -fossil and extant species-receive an enormous amount of attention, especially those living in the Neotropical Region; this is reflected in hundreds of publications produced by special projects, such as the "All Catfish Species Inventory" and "Cypriniformes Tree of Life" and other special publications (Malabarba et al., 1998;Reis et al., 2003;Arratia et al., 2003;Albert, Reis, 2011).
Alepocephali. Alepocephaliforms (Fig. 4b) are characterized by the presence of parietal bones [of traditional terminology] separated by the supraoccipital; branchiostegal cartilages; one postcleithrum; epipleural bones extending forward to about abdominal vertebra 3; opercle reduced dorsally and by several absent structures such as posttemporal fossa; gas bladder; adipose fin; and urodermal bone (Johnson, Patterson, 1996). Although general descriptions of some alepocephaliform structures, such as the skull roof bones, crumeral organs, and caudal skeletons, were published by Greenwood, Rosen (1971), new information on the posterior part of the skull roof, including extrascapulars 7 e180079 [7] and the ontogenetic transformation of vertebral structures and caudal skeleton, are provided below.
In comparison to clupeiforms and ostariophysans, the Alepocephali comprise a small group of 137 species, 32 genera, three families and one order, all living in marine waters (Nelson et al., 2016). Most species of the assemblage inhabit meso-to bathypelagic environments (Nelson et al., 2016).
The inclusion of alepocephalids within the argentinoids is based on the presence of a crumenal organ, the descended position of the distal part of the two to four epineurals, and caudal median cartilages supporting the lowermost ray of the upper caudal lobe (Johnson, Patterson, 1996). Although those features are present in alepocephalids and argentinoids, under the new taxonomic interpretations given by molecular studies, they need further revisions. For instance: 1. A crumenal organ is a specialization of alepocephaliforms and argentinoids. It is a posterior branchial structure that was referred to as an epibranchial organ until Greenwood, Rosen (1971) named it "crumenal organ". The main distinction between the crumenal organ and the epibranchial organ is the presence of a distinct accessory cartilage that may have arisen by segmentation from the posterior articular surface of ceratobranchial 5 in alepocephalids and argentinoids (Nelson, 1967;Greenwood, Rosen, 1971) and in Denticeps (de Pinna, Di Dario, 2010). Thus, the finding of this accessory cartilage in Denticeps opened the necessity for extensive comparative studies in various teleosts.
2. The third feature proposed as shared by argentinoids and alepocephalids, the caudal median cartilages supporting the lowermost ray of the upper caudal lobe (Johnson, Patterson, 1996) is controversial due to its variability. The caudal medial cartilages (mc) are commonly associated with the middle principal caudal rays (PR) in argentinoids (mc versus PR10 and 11) and not with the lowermost ray of the upper caudal lobe. The condition as shown in Fig. 7 for Argentina is also found in alepocephalids (see below, the section on Analysis of characters), salmonids (Fujita, 1990;Arratia, Schultze, 1992), and other euteleosts (Fujita, 1990).
3. The caudal skeleton is a complex structure in need of further investigation in alepocephaliforms (see below, Analysis of characters) and in argentinoids. Unlike most other euteleosts (and also in alepocephaliforms), argentinoids have one vertebral centrum bearing the parhypural and hypurals 1 and 2, a centrum that was interpreted as formed by preural centrum 1 plus ural centrum 1 [of the diural terminology] by Patterson (1970:figs. 38, 40] in Argentina sialis and preural centrum 1 plus ural centrum 1 plus uroneural 1 in Bathylagus antarcticus. However, it is unknown how many vertebral centra form this centrum that I interpret, in a preliminary way, as compound (CC in Fig. 7), but noting that its origin and composition is still unknown (currently under study by GA based on its ontogenetic series). A similar structure has been interpreted as a compound centrum in adult engraulids, e.g., Engraulis and Coilia, where the centrum includes preural centrum 1 plus ural centra 2 and 3+4 [of the polyural terminology] (Schultze, Arratia, 2013:figs. 18, 19A-D), and in ostariophysans (e.g., Monod, 1968;Lundberg, Baskin, 1969;Fink, Fink, 1981;Schultze, Arratia, 1989, 2013Fujita, 1990). However, it is unknown whether the structure interpreted as a compound centrum in ostariophysans forms the same way in different ostariophysan subgroups (see Schultze, Arratia, 2013;Wiley et al., 2015). The fossil ostariophysan †Tischlingerichthys also has a long vertebral centrum bearing the parhypural and hypurals 1 and 2 (Arratia, 1997:fig. 67). Consequently, argentinoids differ from many other euteleost clades, and also from alepocephaliforms, in the presence of a possible compound vertebral centrum that articulates with the parhypural and hypurals 1 and 2, ventrally.
4. The most anterior uroneural or modified uroneural or stegural with an antero-dorsal membranous outgrowth is interpreted as a euteleostean synapomorphy, but its absence in argentinoids is considered to be secondary by parsimony optimization (Wiley, Johnson, 2010). The element interpreted as the most anterior uroneural in argentinoids and alepocephaliforms is different in both groups. The most anterior uroneural has a complex structure in argentinoids. It was identified as uroneural 1 by Patterson (1970: fig. 37), Greenwood, Rosen (1971:figs. 12, 14), and Fujita (1990:figs. 56-59). According to my studies, the first and enlarged uroneural bearing an expanded membranous outgrowth is an autogenous element lying on the dorsal surface of the compound centrum in juvenile and sub-adult specimens of Argentina sialis and has an overall resemblance to the stegural of salmonids (see Arratia, Schultze, 1992;Grünbaum, Cloutier, 2010) and other euteleosts. However, the antero-lateral base of this element fuses to the dorso-lateral surface of the centrum in larger specimens, resembling a pleurostyle (Fig. 7a, b). Consequently, this element differs from the euteleostean stegural (that is always an autogenous element; Arratia, Schultze, 1992;Grünbaum, Cloutier, 2010), but it also differs from the pleurostyle present in ostariophysans and clupeiforms, except for Denticeps (for information on the pleurostyle see below, section on Analysis of characters). Thus, I term this element a "pseudopleurostyle" to note its different formation than that of the pleurostyle and stegural. Summarizing, there are disagreements concerning the inter-  interpretative drawing. Note that the posterior hemi-lepidotrichia of principal ray 10 is displaced. The color orange is used to denote a modified uroneural named pseudopleurostyle herein. CC = compound vertebral centrum; E = epural; H1,3,6 = hypurals 1,3,6; hsPU2 = haemal spine of preural centrum 2; naPU1 = neural arch of preural centrum 1; PR1 = 10,11,19 = principal caudal ray 1,10,11,19; PU3 = preural centrum 3; 'UD' = tendon-bone urodermal; UN = uroneural. pretations of alepocephaliforms as argentinoids and also of alepocephaliforms as salmoniforms based on morphological characters.

Analysis of characters.
The search for potential synapomorphies of Otomorpha was focused on skeletal structures beginning with a revision of characters previously proposed as synapomorphies of clupeiforms plus ostariophysans. The presentation of characters follows the sequence as presented in Lecointre, Nelson (1996).
Ontogenetic studies of clupeiforms and ostariophysans have shown that the so-called processes of the preural centrum 1, or pleurostyle of Monod (1968), may have different origins and topographic relationships. In members of the Clupeoidei, the pleurostyle forms as independent, paired, membranous ossifications of the neural arches of an ural centrum (ural centrum 2 of the polyural terminology) that fuses to the dorsoposterior corner of preural centrum 1 ( Fig.  9a; Arratia, 2010:fig. 13c, d;Schultze, Arratia, 2013:figs. 18, 19, tb. 3). In contrast, in ostariophysans the paired, membranous ossification or pleurostyle (modified neural arch of ural centrum 1 or 2 or 3 of the polyural terminology) may fuse to the dorsoposterior corner of the ural centrum 1+2 or ural centrum 1+2+3 (polyural terminology), forming part of the so-called compound centrum (Fig. 9d, e; Arratia, 2010:fig. 13a, b;Schultze, Arratia, 2013:figs. 4, 7b, c, 20, tb. 3). Consequently, the origin of the pleurostyle differs in groups of otomorphs. For instance, the pleurostyle may originate as a transformation of ural neural arch 2 or 3 in Chanos chanos or as a transformation of ural neural arch 1 or 2 in catostomids (Schultze, Arratia, 2013:fig. 7b) and the cyprinid Danio rerio (Schultze, Arratia, 2013:fig. 20). It is still unknown how the pleurostyle of most ostariophysans originates, because it is a common practice in ontogenetic studies of teleosts not to mention the origin and development of vertebral centra (e.g., chordacentra or autocentra) nor the development of the pleurostyle or other uroneurals. In summation, according to available information, the homology of the structure named "pleurostyle" in clupeoids and ostariophysans is unclear.
The pleurostyle is absent in alepocephaliforms, but they have a free structure, unfused with preural centrum 1 or a ural centrum that externally resembles a stegural (Figs. 10, 11), which is a modified uroneural characteristic of euteleosts. This is also the condition present in fossil alepocephaliforms (Jerzmanska, 1979: fig. 6). Although the daily ontogenetic formation of the stegural is well documented in salmonids (Arratia, Schultze, 1992;Grünbaum, Cloutier, 2010), a comparison with the stegural-like morphology in alepocephaliforms cannot be provided at this time due to a lack of smaller specimens than those studied here (12.4 mm SL; see Material studied). Salmonid specimens of the same size already present an elongate uroneural reaching anteriorly to the neural arch of preural centrum 2. In contrast, in alepocephaliforms of this size, the vertebral centra are not yet formed; see Fig. 10a, b). According to current ontogenetic information concerning the different origins and relationships of the pleurostyle in clupeoids and ostariophysans and the absence of a pleurostyle in Denticeps and alepocephaliforms, this character does not stand as a synapomorphy of Otomorpha. (Lecointre, Nelson, 1996: character 6). The base of hypural 2 fused to ural centrum 1 (= diural terminology) or ural centrum 2 e180079 [11] (of the polyural terminology) was proposed by Lecointre, Nelson (1996) as a possible synapomorphy of the clade formed by extant clupeomorphs and ostariophysans. The authors remarked that this fusion is lacking in gonorynchiforms, such as †Rubiesichthys, †Tharrhias, and the extant Chanos. It is also lacking in other extant gonorynchiforms (Grande, Arratia, 2010: fig. 2.12). Hypural 2 is also unfused in the Late Jurassic ostariophysan incertae sedis †Tischlingerichthys (Arratia, 1997: fig. 67).

Fusion of hypural 2 with first ural centrum
Studies of the caudal skeleton based on ontogenetic series of clupeomorphs and ostariophysans have shown that early in ontogeny the base of hypural 2 fuses to ural centrum 2 (polyural terminology) in clupeiforms, such as Denticeps and clupeoids (which are missing ural centrum 1 in front of the base of hypural 1 early in ontogeny; Arratia, 2010:fig. 13c, d;Schultze, Arratia, 2013:figs. 7c, 18, 19), as well as in ostariophysans (Arratia, 2010: fig. 13b; Schultze, Arratia, 2013:figs. 4, 18-20), but not in the gonorynchiform Chanos chanos (Fig. 9b), which has a free hypural 2 from early ontogeny. In siluriforms, at least the bases of hypurals 1 and 2 are fused to the so-called compound centrum present in most primitive forms, such as in Diplomystes and Nematogenys, but further fusions including other hypurals are observed in different catfish lineages (Lundberg, Baskin, 1969;Arratia, 1982), whereas in gymmnotiforms, the caudal skeleton is highly modified and reduced (de Santana et al., 2013).
Hypural 2 and the remaining hypurals are autogenous (Figs. 10, 11) in alepocephaliforms from early in ontogeny. Alepocephaliforms, contrary to clupeiforms and ostariophysans, present a continuous haemal cartilage (Fig. 10b-d) that extends from the haemal arch of preural centrum 5 or 6 posteriorly, reaching the base of hypural 3 in the smallest specimens studied. The haemal cartilage separates the haemal arch from the ventro-lateral wall of the notochord or the centra, depending on the growth stage. Considering the distribution of this character in a phylogenetic hypothesis either   . 11. Caudal skeleton of Alepocephalus tenebrosus (UW 042217 #4). a. photograph of specimen of 363 mm standard length; photograph courtesy of C. Quezada-Romegialli; b. interpretative drawing. Note the articulation between the proximal region of hypural 3 and the haemal cartilage (arrow) uniting hypurals 1 and 2. E = epural; H1, 3, 6 = hypural 1, 3, 6; hd = hypural diastema; hsPU2 = haemal spine of preural centrum 2; mc = median caudal cartilages; naPU1 = neural arch of preural centrum 1; no = notochord; PH = parhypural; UN = uroneurals; the small third one is not shown. 14 e180079 [14] morphological (Figs. 2,3) or molecular (Fig. 2), the lack of fusion between hypural 2 and ural centrum 2 (polyural terminology) in alepocephaliforms and most gonorynchiforms represents the plesiomorphic condition in teleosts (that is interpreted as secondary loss at this phylogenetic level); whereas their fusion in clupeiforms, a few gonorynchiforms, and otophysans is a unique feature among teleosts that could be interpreted as an otomorph synapomorphy (see Fig. 12, black star). However, ACCTRAN and DELTRAN analyses of this character show this to be ambiguous as a synapomorphy of Otomorpha, thus no conclusion can be reached as to its absence in Alepocephaliformes and most gonorynchiforms relative to other otomorphs. (Lecointre, Nelson, 1996: character 7;Arratia, 1999: character 32). The otic canal passes near the lateral margin of the (lateral) extrascapula in teleosts where the bone is present. The canal gives off a branch, the supratemporal canal or supratemporal commissure or extrascapular commissure, that extends medially, ending in the extrascapula(e) or on the parietals and supra-occipital, uniting both left and right extrascapular series. It has been assumed that when the supratemporal canal passes through parietal bones, a medial extrascapula has fused with the parietal lying below, but this assumption does not explain the canal passing above or in a groove of the supraoccipital.

Fusion of extrascapular and parietal bones
A supratemporal canal or commissure primitively passing through parietals and the supraoccipital is considered a synapomorphy of clupeomorphs (e.g., Grande, 1985) including Denticeps (e.g., Di Dario, de Pinna, 2006). The presence of a medial extrascapula that may fuse or ankylose late in ontogeny with the parietal bone lying below is observed in the ontogenetic series of Chanos chanos (Fig. 13a, b) and other gonorynchiforms, with the exception of the highly modified kneerids, which lack parietal bones, even in their fossil representatives (Davis et al., 2013). A similar pattern of fusion is observed in other extant ostariophysans (Fig. 13c, d; Arratia, Huaquín, 1995: fig. 3). Lecointre, Nelson (1996:p. 202) first proposed that a fusion of a medial extrascapula and the parietal bone is a synapomorphy of ostariophysans and clupeomorphs. Later, Arratia (1999:p. 325) described the character for her Os- Fig. 12. Abbreviated phylogenetic hypothesis of teleosts (modified from Betancur-R. et al., 2017) with distribution of certain character states discussed in the text. Yellow circle = fusion or ankylosis between medial extrascapular and parietal bones; red triangle = vertebrae anterior to preural centrum 2 with haemal arches and centra fused in juvenile-adult individuals; green square = silvery areas associated with gas bladder; black star = hypural 2 fused with ural centrum 2 [of the polyural terminology; ural centrum 1 of the diural terminology]. N = non applicable.  tarioclupeomorpha (= Otocephala) as "clupeocephalans in which primitively there is an ankylosis or fusion between the medial extrascapula and parietal alone or parietal and supraoccipital".
Alepocephaliforms do not present this character. An ossicle-like lateral extrascapula is present in some genera (e.g., Leptoderma sp. and Searsioides; Fig. 13e) and a medial ossicle-like extrascapula is absent so that a supratemporal canal is absent or very short. Other alepocephaliforms have an irregular series of two or more small ossicle-like extrascapular bones, but a supratemporal canal passing on the parietals or parietals plus supraoccipital is not present (e.g., Paraholtbyrnia, Pellisolus, Mirorictus, and Searsia; Fig. 13f; see also Greenwood, Rosen, 1971).
Primitively, both the clupeiforms and ostariophysans are unique among teleosts in possessing a fusion or ankylosis of the medial extrascapula with the parietal bone so that the supratemporal canal passes through both bones. This condition is lacking in alepocephaliforms, which have one or multiple ossicle-like free extrascapulars and no medial extension of the supratemporal canal. Absence in alepocephaliforms and most advanced ostariophysans (Fig. 12, yellow circle) is secondary by parsimony optimization. Thus, following this interpretation, this character can be interpreted as a synapomorphy of otomorphs.
Fusion of hemal arches and centra anterior to preural centrum 2 (Lecointre, Nelson, 1996: character 8). The original character description indicated a fusion between haemal spines and centra. However, the centra are the ones fused to their respective arches during ontogeny.
This character was suggested first as an ostariophysan synapomorphy by Fink, Fink (1981, 1996, who also mentioned that the fusion is present in the caudal region of all clupeiforms, suggesting a possible relationship among the groups. However, clupeiforms lack an adipose fin and nuptial tubercles, which are features that are present in ostariophysans that link them to euteleosts, a view favored by Fink, Fink (1981). Lecointre, Nelson (1996:p. 202) pointed out that this feature is consistently present in all extant clupeomorphs and ostariophysans (see also Monod, 1968;Fujita, 1990) and in their fossil representatives (Arratia, pers. observations). However, and following Fujita's (1990) caudal skeleton survey, Lecointre, Nelson (1996) stated that this feature might be also present in some euteleostean subgroups of various phylogenetic positions, but not in all members of each subgroup, e.g., in Umbra among esociforms and in Bathylagus among osmeriforms. Some species with all haemal arches anterior to preural centrum 2 fused to their centra are found in some Stomiatiformes, Myctophiformes, and a few more advanced neoteleosts (e.g., Aulopiformes). Those occurrences outside the otocephalans or ostarioclupeomorphs make this feature homoplastic.
Juvenile and adult alepocephaliforms present all haemal arches anterior to preural centrum 2 fused to their centra (Fig. 11). In early ontogenetic stages (Fig. 10), the haemal arches anterior to preural centrum 2 are unfused with their respective centra represented by chordacentra. Autocentra of the caudal region begin to form in the juvenile so that the fusion between dorsal and ventral arcocentra to their respective autocentrum is comparatively later than in other otomorphs. Thus, clupeiforms, alepocephaliforms, and ostariophysan share the presence of fused arches and centra anterior to preural centrum 2 in juvenile-adult growth stages. According to the distribution of this character in the phylogenetic hypothesis shown in Fig. 12 (red triangle), this feature stands as a synapomorphy of Otomorpha, including the alepocephaliforms. Its erratic occurrence in some Stomiatiformes and Myctophiformes can be interpreted as independently acquired (This character has not been represented in Fig. 12 where it occurs only in one genus, e.g., Umbra within Esociformes). (Arratia 1999: character 135). Arratia's (1999; Fig. 3 herein) phylogenetic hypothesis included fossil and extant teleosts, and because of this and the fact that in primitive fossil clupeiforms (e.g., †Santanaclupea) and fossil ostariophysans (e.g., †Tischlingerichthys) the bases of hypurals 1 and 2 are not joined by cartilage, Arratia (1999:p. 325) specifically wrote that this feature stands as a synapomorphy at the primitive level of the Ostarioclupeomorpha (= Otocephala). Most extant clupeiforms and ostariophysans may show cartilage at the base of one hypural, or at the base of hypural 1 and arch of the parhypural, or both are fused to the autocentrum from early stages of development. In adult clupeiforms and ostariophysans, the bases of hypurals 1 and 2 are not connected by cartilage, a unique feature among teleosts (see also Monod, 1968;Fujita, 1990).

Bases of hypurals 1 and 2 not joined by cartilage at any growth stage
The poorly known caudal skeleton of the fossil alepocephaliform †Carphathichthys polonicus (Jerzmanska, 1979:fig. 6) shows spaces between the bases of hypurals 1 and 2 and the ventral margins of the centra, that may indicate the presence of cartilage during life as the condition present in extant alepocephaliforms (see Figs. 10, 11). However, I have not been able to examine the fossil material, only had access to photographs. Extant alepocephaliforms are unique in the presence of a long tube-like haemal cartilage surrounding the caudal blood vessels that extends from the haemal arches of preural centrum 4 or 5 to the base of hypural 3 in juveniles and is restricted to the bases of hypurals 1-3 in adults. The posterior part of the haemal cartilage produces an articular surface for the base of hypural 3 (see Figs. 10, 11), another unique character of this group among otomorphs. A similar articulation seems to be present as a convergence in some stomiatiforms and myctophiforms (see Fujita, 1990:figs. 76, 111).
Considering the new information on alepocephaliforms and the fact that I have not been able to examine their fossil representatives, I prefer to postpone further suggestions or conclusions about this character. Silvery area associated with gas bladder. The anterior chamber of the gas bladder partly to completely covered by a silvery peritoneal tunic was proposed as a possible synapomorphy of Clupeomorpha plus Ostariophysi by Wiley, Johnson (2010), following Fink, Fink (1996), who listed it as an ostariophysan synapomorphy, but indicated that a similar condition was also present in some clupeomorphs and therefore might be synapomorphic for the clade (that was given a name in another chapter of the same volume: Johnson, Patterson, 1996) containing the two groups. A gas bladder is characteristically absent in alepocephaliforms among teleosts (Berg, 1937;Johnson, Patterson, 1996;Wiley, Johnson, 2010); this absence makes the character ambiguous, because it is impossible to know if the structure primitively had a shiny peritoneal tunic in the group or not. Another hypothesis is that this absence in alepocephaliforms (Fig. 12, green square) represents a secondary loss among otomorphs, and consequently, this character stands as a synapomorphy of otomorphs, a hypothesis that is supported by parsimony optimization.
Additional cartilage attached to ceratobranchial 5. The finding of an additional cartilage attached to ceratobranchial 5 (Nelson, 1967) was important for separating fishes with a branchial pouch. The additional cartilage is characteristic of the crumenal organ present in alepocephalids and argentinoids (Greenwood, Rosen, 1971;Johnson, Patterson, 1996), whereas this cartilage is absent in the epibranchial organ found in gonorynchiforms with different cartilages (Fig.   14b, c). Thus, the finding of a vestigial additional cartilage attached to ceratobranchial 5 (Fig. 14a) in the clupeomorph Denticeps clupeoides (de Pinna, Di Dario, 2010) is important because it "provides circumstantial support for a possible relationship between Otocephala and argentinoid and may indicate a vestigial remnant of an epibranchial organ in Denticeps" (de Pinna, Di Dario, 2010:p. 251). It may be argued that this new information may provide a circumstantial support of a vestigial crumenal organ in Denticeps more than an epibranchial organ, because the cartilages in gonorynchiforms, as illustrated in Figure 14b, c are not the same as the vestigial cartilage associated with ceratobranchial 5 in Denticeps. This character is also present outside the clade Otomorpha in Argentiniformes.
A survey of other osteological structures, cephalic sensory canals, and pit lines did not provide further support for the monophyly of otomorphs.

Conclusions
Considering that without exception recently published molecular studies have indicated that the alepocephaliforms, clupeiforms, and ostariophysans form a monophyletic assemblage, I have revisited this phylogenetic hypothesis from a morphological point of view. This was based on the fact that all synapomorphies previously cited for otocephalans or ostarioclupeomorphs are mainly osteological (Lecointre, Nelson, 1996;Arratia, 1997Arratia, , 1999, except one (Fink, Fink, 18 e180079[18] 1996). Thus, a study of alepocephaliform osteology, sensory canals and pit lines was executed for comparisons among the three groups. Among otomorphs, alepocephaliforms are characterized by the loss of several structures and at least four of them concern characters previously proposed for otocephalans. The results show that the potential characters supporting the monophyly of Otomorpha are homoplastic: 1. The fusion of hemal arches and centra anterior to preural centrum 2 in juvenile and adult individuals is a synapomorphy of Otomorpha (Fig. 12, red triangle), although this fusion also occurs in a few unrelated euteleosts (e.g., stomiatiforms and myctophiforms). Under current knowledge, this is an unambiguous synapomorphy supporting the clade.
2. A fusion of ural centrum 2 [polyural terminology] with the base of hypural 2 is present in clupeomorphs, cypriniforms, characiforms, and siluriforms. The highly modified caudal skeleton of gymnotiforms may include reduction and fusion of all its endoskeletal structures. Most gonorynchiforms have unfused elements, as do alepocephaliforms. The fusion between hypural 2 and ural centrum 2 is unique to these fishes among the crown Teleostei (Fig. 12, black stars) and its absence among primitive otomorphs (alepocephaliforms and most gonorynchiforms) is ambiguous.
3. The (medial) extrascapula and parietal bone are fused so that the supratemporal canal extends into the medial extrascapula or in both the medial extrascapula and supraoccipital. This character is not present in alepocephaliforms due to the absence of well-defined extrascapular bones and a supratemporal canal (Fig. 13f, g). According to the distribution of this character among teleosts, this loss in alepocephaliforms (Fig. 12, yellow circle) is best interpreted as a secondary loss, and consequently, the character may still be interpreted as an otomorph synapomorphy. 4. A silvery area associated with the gas bladder is present in clupeomorphs and ostariophysans (Fink, Fink, 1996;Wiley, Johnson, 2010) among teleosts. In contrast, alepocephaliforms do not share this character, because they do not have a gas bladder, which is considered a synapomorphy of the group. According to the occurrence of this character in otomorphs (Fig. 12, green square) among teleosts, this loss in alepocephaliforms is interpreted as a secondary loss and the character still can be considered an otomorph synapomorphy.
Consequently, all characters proposed as synapomorphies of Otomorpha are homoplastic, which is a common condition of many characters supporting different teleostean nodes among the crown Teleostei. This is especially the case in phylogenetic studies including large numbers of taxa where losses or reversals at different levels change interpretations of characters, e.g., from those previously interpreted as uniquely derived to homoplastic.
A survey of the literature reveals that most (or all?) phylogenetic higher-level hypotheses of teleostean intrarelationships are heavily based on osteological characters, including sensory canals. Despite the great efforts spent by teleostean researchers to understand osteological characters, numerous cases remain unclear, especially those concerning interpretations of homologous characters of the caudal fin. See above for information concerning the so-called uroneural 1 in argentinoids, alepocephaliforms, and ostariophysans (Figs. 7,9,10,11) or the so-called compound centrum in some argentinoids, clupeomorphs, ostariophysans (Figs. 7,9). Although knowledge on branchial arches has increased during recent years, more effort should be spent on investigations concerning branchial arches, and their cartilages and muscles, especially within the large diversity of otomorphs and closely related clades. In the particular case of otomorphs, it is suggested that future researchers explore other anatomical structures outside osteology such as muscles and ligaments, and soft anatomical complexes, such as the nervous, circulatory, or digestive systems.
Finally, I would like to call the attention to the fact that among otomorphs there are three main groups characterized by numerous autapomorphies: Denticipitidae, Alepocephaliformes, and Gymnotiformes. The three taxa share being the youngest clades in Otomorpha, which has a long fossil history of at least 145 Ma. Among otomorphs, the youngest ones seem to be the gymnotiforms, with their oldest record being from the upper Miocene (about 23 Ma; Gayet, Meunier, 1991) and the alepocephaliforms, with the oldest fossil record of about 30-23 Ma. Has the late origin of these taxa some bearing on their uniqueness that poses difficulties in the interpretation of characters? Material Studied. In addition to the otomorph species, the material listed below also includes other teleosts used in comparisons. The specimens are listed following a taxonomic arrangement and ordered alphabetically within each taxon. Extinct taxa are identified with the sign ' †'; cl&st = cleared and double stained specimens for cartilage and bone; skl = dry skeleton.