Taxonomic Review and Phylogenetic Analysis of Enchodontoidei (teleostei: Aulopiformes)

Enchodontoidei are extinct marine teleost fishes with a long temporal range and a wide geographic distribution. As there has been no comprehensive phylogenetic study of this taxon, we performed a parsimony analysis using a data matrix with 87 characters, 31 terminal taxa for ingroup, and three taxa for outgroup. The analysis produced 93 equally parsimonious trees (L = 437 steps; CI = 0.24; RI = 0.49). The topology of the majority rule consensus tree was: (Dercetoides + Hastichthys)))))). The group Enchodontoidei is not monophyletic. Dercetidae form a clade supported by the presence of very reduced neural spines and possess a new composition. Enchodontidae are monophyletic by the presence of middorsal scutes, and Rharbichthys was excluded. Halecidae possess a new composition, with the exclusion of Hemisaurida. This taxon and Nardorex are Aulopiformes incertae sedis.


INTRODUCTION
Enchodontoidei are extinct marine teleosts generally with an elongate body and long and narrow rod-like maxilla included in the mouth gape (Nelson 1994).They possess a long temporal range, extending from the Early Cretaceous to the Early Eocene, and a wide geographic distribution in sedimentary deposits of South America (e.g., Bolivia and Brazil), Africa (e.g., Democratic Republic of Congo, Egypt, and Morocco), Asia (e.g., Arabian Peninsula, India, Israel, Japan, and Lebanon), Europe (e.g., Belgium, England, Germany, Holland, Italy, and Sweden), and North America (Canada, Mexico, and United States) (e.g., Goody 1969, Chalifa 1996, Fielitz 2004, Figueiredo and Gallo 2006, Gallo et al. 2006).
The taxon was erected by Berg (1937) as a suborder, which included only the family Enchodontidae.According to this author, enchodontid fishes were similar to the members of the suborder Stomiatoidei also created by him, but the enchodontids bear a median row of dorsal scutes and their vertebrae do not possess parapophyses.Goody (1969) accomplished a comprehensive review of certain Late Cretaceous teleosteans, considering Enchodontoidei as part of the order Salmoniformes, together with three other suborders, Ichthyotringoidei, Cimolichthyoidei, and Halecoidei. Rosen (1973) erected the order Aulopiformes comprising 15 living families, and the suborder Alepisau-roidei with 15 fossil genera, without dividing them systematically (i.e., Ichthyotringa, Apateodus, Apateopholis, Cimolichthys, Dercetis, Rhynchodercetis, Pelargorhynchus, Prionolepis, Enchodus, Palaeolycus, Eurypholis, Saurorhamphus, Halec, Phylactocephalus, and Hemisaurida), as well as the fossil genera incertae sedis of the superfamily Synodontoidea, Sardinius and Volcichthys.Nelson (1994) recognized the order Aulopiformes by Rosen (1973), as well as the suborders proposed by Goody (1969) as superfamilies, putting them in a single suborder, Enchodontoidei, composed of four superfamilies: Enchodontoidea (Enchodus, Parenchodus, Palaeolycus, Eurypholis, and Saurorhamphus), Cimolichthyoidea (Cimolichthys, Prionolepis, Benthesikyme, Cyranichthys, Dercetis, Dercetoides, Pelargorhynchus, Rhynchodercetis, and Stratodus), Halecoidea (Halec, Hemisaurida, and Phylactocephalus) and Ichthyotringoidea (Ichthyotringa and Apateodus).However, this classification was not developed in a phylogenetic framework.Baldwin and Johnson (1996) accomplished a cladistic analysis of Aulopiformes, including only extant taxa.The authors maintained the monophyly of the taxon and added synapomorphies to those proposed by Rosen (1973), which are mainly related to the morphology of the dorsal portion of the gill arches.Their new synapomorphies are from the intermuscular system, internal soft anatomy, pigmentation pattern of larvae, and morphology of the pelvic girdle.Most of these features are very difficult to assess in fossil specimens.Sato and Nakabo (2002) accomplished a phylogenetic analysis of living Aulopiformes based on morphological and molecular data.They divided it into the suborders Synodontoidei, Chlorophthalmoidei, Alepisauroidei, and Giganturoidei.Moreover, the authors proposed a new family of Aulopiformes (i.e., Paraulopidae).Fielitz (2004) and Gallo et al. (2005) proposed hypotheses of the phylogenetic relationships of some fossil Aulopiformes (Enchodontoidea and Dercetidae, respectively).Nelson (2006) placed the extinct aulopiforms in three suborders: Ichthyotringoidei, comprising the families Ichthyotringidae, Dercetidae, and Prionolepidae; Halecoidei, with a single family, Halecidae; and Alepisauroidei, with two families, Cimolichthyidae and Enchodontidae.The extant aulopiforms were classified in Synodontoidei (with four families), Chlorophthalmoidei (with six families), and Giganturoidei (with two families).Additionally, four living families were placed in the suborder Alepisauroidei.
In fact, the assemblage of extinct aulopiforms defined by Nelson (2006) corresponds to Enchodontoidei sensu Nelson (1994).However, Nelson (2006) did not discuss his reasons for disregarding the name Enchodontoidei and put its members in Alepisauroidei, Ichthyotringoidei, and Halecoidei.Moreover, Enchodontoidei were considered in the cladistic analysis of Dercetidae (Gallo et al. 2005), as well as in a preliminary approach by Silva and Gallo (2007).As there has not been a recent comprehensive phylogenetic study of Enchodontoidei, we review their classification history and provide a new cladistic analysis.

SYSTEMATIC HISTORY OF ENCHODONTOIDEI
In this paper we use the general classification of Nelson (1994), except for the family Enchodontidae (sensu Fielitz 2004).

SUPERFAMILY ICHTHYOTRINGOIDEA
According to Goody (1969), Ichthyotringoidea comprises two closely related families, Ichthyotringidae and Apateopholidae.The author considered mainly primitive features of the body and caudal skeleton, as well as a derived feature related to the rostral region.He stated that, despite the similarities shared by the taxa, Apateopholidae should be the more advanced taxon.
Family Ichthyotringidae.The family Ichthyotringidae (Table I) was created by Jordan (1905) to contain a single genus (i.e., Ichthyotringa).Goody (1969) positioned the family in the suborder Ichthyotringoidei.Later, Nelson (1994) included the taxon in the suborder Enchodontoidei together with other fossil aulopiforms.The generic epithet Ichthyotringa was created by Cope (1878) to replace the genus Rhinellus of Agassiz (1844), which was pre-occupied.The genus Ichthyotringa includes the following species: Ichthyotringa furcata (Agassiz, 1844), I. tenuirostris Cope, 1878, I. damoni (Davis, 1887), I. ferox (Davis, 1887), I. delicata (Hay, 1903), and I. africana (Arambourg, 1954).Forey et al. (2003), in the general list of fossil fishes from Lebanon, placed Apateopholis in the family Ichthyotringidae, but remarked that at least one species of Apateopholis is often misinterpreted as a species of the closely related Ichthyotringa.Only more recently a new ichthyotringoidei was reported to the El Doctor Formation in the Albian-Cenomanian of Mexico, I. mexicana Fielitz andGonzález Rodríguez, 2008. Goody (1969) ranked Apateodus as an addendum (incertae sedis) to the Ichthyotringidae with a single species (A.striatus Woodward, 1901).Nelson (1994Nelson ( , 2006) ) placed Apateodus in Apateopholidae, but he did not report the taxonomic status of Apateopholis.In Frickhinger (1995), the latter was considered an ichthyotringid and the former was not mentioned.Forey et al. (2003) positioned Apateopholis in the family Ichthyotringidae.Taverne (2004) maintained Apateodus and two other genera of Cretaceous alepisauroids (Yabrudichthys and Rharbichthys) as family incertae sedis.However, the same author (Taverne 2006c) suggested to exclude Apateodus from Ichthyotringidae.Fielitz and González Rodríguez (2008) accomplished a cladistic analysis of Ichthyotringoidea and placed Apateodus tentatively among the species of Ichthyotringa.More recently, Fielitz and Shimada (2009) described a new species of Apateodus (A. busseni) suggesting that the genus needs revision, but ranking it in Ichthyotringidae.

SUPERFAMILY CIMOLICHTHYOIDEA
This superfamily includes three families (i.e., Cimolichthyidae, Dercetidae, and Prionolepididae).According to Goody (1969), Cimolichthyidae and Dercetidae show a great similarity regarding the structures of the skull and body, especially in the rostral region.Moreover, general body squamation is lacking and two or three rows of isolated scutes are present on the flanks.Regarding the family Prionolepididae, the author pointed out some problems concerning its taxonomic placement.Unlike Woodward (1901), who assigned the genus Prionolepis to the Enchodontidae, Goody (1969) considered it closely related to dercetids and cimolichthyids.
Family Cimolichthyidae.The family (Table II) was erected by Goody (1969) to include the single genus Cimolichthys, which was designated by Leidy (1857) with the species C. levesiensis.Cope (1872), studying specimens from Niobrara (USA), recognized five species of Cimolichthys: C. nepaholica, C. sulcatus, C. semianceps, C. contracta, and C. merrillii. Later, Hay (1903) recognized only the species C. nepaholica, as other species were based mainly on isolated teeth and fragments of the jaws.
Family Dercetidae.Traditionally the creation of this family is attributed to Pictet (1850).However, he did not use the name Dercetidae, defining only a Dercetis group with D. tenuis, D. triqueter, and D. linguifer.As far as we know, the name Dercetidae was used for the first time by Woodward (1901).
The genus Benthesikyme was created by White and Moy-Thomas (1940), including new species of Leptotrachelus described by several authors at the end of the nineteenth century and during the twentieth century.Taverne (2005b) furnished a comprehensive review of the genera Dercetis, Leptotrachelus, and Benthesikyme, in which L. sagittatus is probably a synonym of D. elongatus, L. virgulatus of D. triqueter, and L. longipinnis of Benthesikyme gracilis.The taxonomic status of D. reussi, D. latiscutatus, D. maximus, and L. serpentinus was also discussed but not in a conclusive way.Yet, Taverne (2005a, b) erected three new monotypic genera within the family Dercetidae, Ophidercetis (O.italiensis), Nardodercetis (N.vandewallei), and Scandiadercetis (S. limhamnensis).The latter had been proposed originally by Davis (1890) as Dercetis limhamnensis.
Blanco-Piñon and Alvarado-Ortega (2005) briefly described Robertichthys riograndensis, which is the sec-An Acad Bras Cienc (2011) 83  (2) ond record of Dercetidae in the Turonian of Mexico.Blanco et al. (2008) provided a redescription of this taxon, as well as a discussion on its relationships.Figueiredo and Gallo (2006) described Brazilodercetis longirostris, which is the first record of the family in South America.
Family Prionolepididae.The genus Prionolepis was created by Egerton (in Dixon 1850) with only one species, P. angustus.Later, Pictet and Humbert (1866) included one more species in the genus, P. cataphractus.Goody (1969) reviewed P. cataphractus and proposed the family Prionolepididae (Table II).
SUPERFAMILY ENCHODONTOIDEA Family Enchodontidae.Previously to Woodward (1901), the genera assigned to Enchodontoidea were allocated in different families (Agassiz 1835, Pictet 1850, Cope 1872, 1874).The first attempt to classify the enchodontoids in a separate group was proposed by Woodward (1901), in which the author erected the family Enchodontidae and put it in the Isospondyli.He divided the family into two main groups based on the presence or the absence of a single tooth in palatine.In the first group, the author included the genera Enchodus, Palaeolycus, Eurypholis, and Saurorhamphus; the second consisted of the genera Halec, Cimolichthys, Prionolepis, Leptecodon, and Pantopholis.However, Woodward (1901) stated that the living families more closely related to the Enchodontidae were Alepisauridae and Odontostomidae, which possess the border of the upper jaw formed exclusively by the premaxilla, the maxilla being untoothed and excluded from the mouth gape.
After Woodward (1901), there was a long debate about the relationships of enchodontids and living fish families.Jordan (1905), Gregory (1933), and Arambourg (1954) agreed with the hypothesis of Woodward (1901) regarding the relationships of enchodontids and alepisaurids, but they positioned Enchodontidae in the suborder Iniomi.On the other hand, Regan (1911) and Romer (1945) rejected the hypothesis of Woodward (1901) and allocated the family into the Stomiatoidei, belonging to the suborder Isospondyli.Berg (1940) also rejected the enchodontids plus alepisaurids hypothesis and put enchodontids in the suborder Enchodontoidei into the Clupeiformes, as synonym of Isospondyli.The generic composition of Enchodontidae remained stable for some time, except for Halec, which was moved to the family Halecidae by Goody (1969), and for the inclusion of Rharbichthys by Arambourg (1954).Goody (1969) accomplished a comprehensive review of the Enchodontidae, including the genera Enchodus and Palaeolycus.Also, he created the family Eurypholidae into Enchodontoidei to comprise the genera Eurypholis and Saurorhamphus.Sorbini (1976) proposed a relationship between Rharbichthys and Cimolichthys.Taverne (1985) studied Rharbichthys and stated that it probably possessed close affinity with the halecids regarding the general proportions and the head shape.
Although the monophyly of Enchodontidae seems to be widely accepted, the previous diagnoses (e.g., Goody 1969, Rosen 1973, Chalifa 1989b) were not deduced from cladistic analyses.
Recently, Fielitz (2004) tested the monophyly of the family Enchodontidae, including living and extinct aulopiforms.The clade is supported by three synapomorphies: single dermopalatine tooth, dermopalatine bone with same length or shorter than the tooth, and interopercle absent.Alepisauridae appear in the analysis as the sister group of the clade formed by the extinct Aulopiformes.Enchodontidae were divided into four subfamilies: Rharbichthinae (with Rharbichthys), Palaeolycinae (with Palaeolycus), Eurypholinae (with Eurypholis and Saurorhamphus) and Enchodontinae (with Enchodus).The genus Parenchodus was put in synonymy with Enchodus.
The subfamily Palaeolycinae (Table III) is monotypic, comprising only Palaeolycus dreginensis described by von der Marck (1863).Later, it was reviewed by Siegfried (1954), who pointed out morphological similarities and putative relationships with the An Acad Bras Cienc (2011) 83 (2) 490 HILDA M.A. SILVA and VALÉRIA GALLO   Chalifa (1987) extant genus Odontostomus.This genus is in synonymy with Evermanella, which is in the family Evermannellidae of the suborder Alepisauroidei.
The subfamily Eurypholinae (Table III) was originally proposed as a family (Eurypholidae) by Goody (1969) to encompass the genus Eurypholis by Pictet (1850).This genus comprises only the type-species (E.boissieri Pictet, 1850) and another one initially proposed as Enchodus pulchellus by Woodward (1901), but later redefined by Goody (1969) as Eurypholis pulchellus.Nelson (1994) placed the family Eurypholidae in the superfamily Enchodontoidea in the suborder Enchodontoidei.Fielitz (2004) suggested the arrangement of the genera Eurypholis and Saurorhamphus in the subfamily Eurypholinae.
Parenchodus longipterygius was described by Raab and Chalifa (1987) as belonging to the family Enchodontidae.The authors suggested a relationship with the genus Enchodus, due to similarities in some structures of the head, the absence of scales, and the fusion of the elements of the caudal fin.Fielitz (2004) put the genus in synonymy with Enchodus.
SUPERFAMILY HALECOIDEA Family Halecidae.The family Halecidae (Table IV) was originally proposed by Agassiz (1834) including forms similar to the clupeoids and salmonoids.The grouping and its name were used only by Pictet (1850) and Davis (1887), being disused later.The family was re-erected by Goody (1969) into the suborder Halecoidei containing three genera: Halec, Phylactocephalus, and Hemisaurida.Nelson (1994) put Halecidae in the superfamily Halecoidea in the suborder Enchodontoidei.Later, Nelson (2006) opted to use the suborder Halecoidei by Goody (1969).

493
The genus Phylactocephalus was erected by Davis (1887) and put in synonymy with Halec by Woodward (1901).Goody (1969) verified marked differences between the genera and separated them.He re-erected Phylactocephalus with a single species, P. microlepis Davis, 1887. Kner (1867) created the genus Hemisaurida containing a single species, H. neocomiensis.Woodward (1901) and Romer (1966) suggested that this genus could belong to the family Myctophidae.Goody (1969) rejected this hypothesis based mainly on two features present in halecoids: maxilla partially excluded from the mouth gape and premaxilla without ascending and articular processes.Yet, Goody (1969) created the species H. hakelensis.

TAXA INCERTAE SEDIS
Family Nardorexidae.The monotypic family Nardorexidae (Table V) was proposed by Taverne (2004) with the species Nardorex zorzini.He placed the family in the suborder Alepisauroidei based on putative relationships with Enchodontoidei.
Family Serrilepidae.The taxon was proposed by Chalifa (1989c) with the single species Serrilepis longidens.Forey et al. (2003) added two new species to the genus, S. prymnostrigos and S. minor.According to these authors, among the aulopiforms, Serrilepis is more closely related to Halec, Hemisaurida, and Phylactocephalus and, therefore, it should be classified into the Halecidae.This relationship is based on two synapomorphies: fusion of dorsal hypohyal and anterior ceratohyal; fusion of first and second hypurals and third and fourth hypurals.However, this latter synapomorphy is present in Atolvorator longipectoralis An Acad Bras Cienc (2011) 83 (2) (Gallo and Coelho 2008), and we opted to use Chalifa's classification with Serrilepis in the family Serrilepidae (Table V).

MATERIAL
The specimens of Enchodontoidei herein studied belong to several paleontological collections (see Appendix I).Extant aulopiform, stomiiform, and myctophiform fishes were used as comparative specimens represented by dry skeletons, alcohol-preserved, and cleared and stained specimens.They belong to the AO.UERJ, O.UERJ, and MZUSP (see Appendix I).Moreover, for the taxa of difficult access, as for instance, those deposited in Hebrew University of Jerusalem and Museo Civico di Storia Naturale di Verona, we selected information from available literature (e.g., Chalifa 1985, 1989a, b, c, 1996, Raab and Chalifa 1987, Taverne 2005a, b, 2006a, b).A data matrix was built with 87 characters, unordered and unweighted, 31 terminal taxa for ingroup, and three taxa for outgroup.The parsimony analysis was carried out using the computer program PAUP* version 4.0b10 (Swofford 2001), with the heuristic algorithm HSearch.We tried to apply the exact branch-and-bound algorithm, but due to the length and complexity of the data matrix, we failed to obtain a result even running the analysis for a few days.
Appendix II includes the coded data matrix, which was built based on the list of characters presented in the Results.Only character states that resulted in apomorphies were illustrated.Although the strict consensus is the real consensus that shows all possible topologies, we opted to present the majority rule consensus (MRC) because, in general, it possesses a better resolution.The MRC is a form of consensus that pre-An Acad Bras Cienc (2011) 83 (2) serves all clades present in the majority (i.e., in more than 50%) of the obtained set of equally parsimonious cladograms (Margush and McMorris 1981).The 50% rule ensures that all included clades are compatible (Sharkey and Leathers 2001).In spite of some criticism (e.g., Sharkey and Leathers 2001), several authors are using MRC as a method of weighting clades to solve ambiguous strict consensus trees (e.g., Swofford 1991, Candall and Fritzpatrick 1996, Titus and Larson 1996, Lutzoni 1997).

RESULTS
Eighty-seven characters were analyzed in this study (see Appendix III).The cladistic analysis produced 93 equally parsimonious trees, with a tree length of 437 steps, consistency index (CI) of 0.24, and retention index (RI) of 0.49.The majority rule consensus tree is shown in Figure 1

DISCUSSION
Before discussing the cladistic analysis per se, we will furnish a brief comment on certain characters.
According to Chalifa (1989a) and Taverne (1991), the presence of a low head is a synapomorphy of Dercetidae.In this study, it is shared with two other genera outside the group, showing a homoplastic distribution.
In fossil aulopiform fishes, particularly in most of Cimolichthyoidei and Enchodontoidei (sensu Goody 1969), the snout length is equivalent to the diameter of the orbit (e.g., Cimolichthys and Eurypholis, respectively).All dercetids possess an elongate snout and the extreme condition is verified in Hastichthys, in which the snout length reaches more than 12 times the diameter of the orbit.Fielitz (2004) considered the presence of vomerine teeth as a synapomorphy of the group formed by Cimolichthys and members of the family Enchodontidae.In the present study, this state of character has a homoplastic distribution, due to it is present in Nardorex and Prionolepis.Gallo et al. (2005) interpreted the presence of a mesethmoid with a bifid anterior extremity as an autapomorphy of Dercetis, but herein this condition was also verified in Nardorex and Sardinioides.Yet, Gallo et al. (2005) pointed out a mesethmoid with a bifid posterior extremity in Dercetis.However, Taverne (2005b) described this bone with an acute posterior extremity, and we confirmed this feature in Dercetis elongatus (NHM P. 49793) and D. triqueter (MNHN SHA 523).Yet, Taverne (2006c) described Ichthyotringa africana as possessing a mesethmoid with acute posterior extremity, mainly based on specimens MNHN DTS 225-228.However, observing the specimen MNHN DTS 228, we verify a mesethmoid with bifid posterior extremity.
In Gallo et al. (2005), the autosphenotic spine posteriorly curved is a synapomorphy of the clade (Rhynchodercetis, Hastichthys).However, in this study, the character shows a slightly wide distribution, being also present in Atolvorator and Trachinocephalus.Taverne (2006b) stated that the autosphenotic of Caudadercetis is hidden by the frontals.For this reason, we opted to code it as missing data.
Accordig to Taverne (1991), Pelargorhynchus is characterized by four derived features, among them the loss of the supraoccipital crest.Gallo et al. (2005) considered the absence of the supraoccipital crest as an autapomorphy of Pelargorhynchus.In the present study, the character was differently interpreted, because its absence and/or loss were verified in other two taxa (Benthesikyme and Rharbichthys).Gallo et al. (2005) pointed out the presence of a pterotic not projecting beyond the occiput in Dercetis based on available descriptions.However, we pointed out another condition (projecting beyond the occiput), following the redescription of the genus furnished by Taverne (2005b), as well as by direct observation of specimens of Dercetis triqueter and D. elongatus.
The presence of an unroofed post-temporal fossa in Palaeolycus followed Goody (1969), in contrast with Fielitz (2004) who pointed out a roofed condition.We verify a roofed condition in Eurypholis and Saurorhamphus in disagreement with Fielitz (2004).Gallo et al. (2005) pointed out a partially roofed posttemporal fossa in the genus Dercetis.However, following the redescription of the genus furnished by Taverne (2005b), as well as the direct observation of specimens of Dercetis triqueter and D. elongatus, we indicate herein a roofed post-temporal fossa to Dercetis.Rosen (1973) pointed out the absence of orbitosphenoid in enchodontids.Additionally, Taverne (1991) stated that its absence would be a primitive condition of dercetids.Gallo et al. (2005) proposed the presence of orbitosphenoid as an autapomorphy of Ichthyotringa.Generally, the common absence of an orbitosphenoid in the specimens observed directly or indirectly is probably due to its fragility, which impedes a good preservation.Gallo et al. (2005) suggested the presence of the basisphenoid as an autapomorphy of Ichthyotringa, but other taxa (i.e., Apuliadercetis tyleri, Atolvorator longipectoralis and Robertichthys riograndensis) bearing this bone were described after this study.Taverne (1985) pointed out three conditions for the presence of teeth on the ectopterygoid: bone toothless or bearing some small conic teeth in a small portion of it as in Rharbichthys ferox, from the Cenomanian of Morocco; and well-developed teeth on the ectopterygoid similar to those found in the dentary of Rharbichthys cf.ferox from the Cenomanian-Turonian of Italy (see also Sorbini 1976).For this reason, the character was regarded as polymorphic.
In the cladistic analysis herein performed (Fig. 1), we demonstrate that the suborder Enchodontoidei is not a monophyletic group, as two genera of the outgroup belonging to another suborder and even to another order (Trachinocephalus and Protostomias, respectively) went to the ingroup.Although Trachinocephalus belongs to the same order of the taxa herein studied (Aulopiformes), it is allocated into the suborder Synodontoidei, together with other extant members of Aulopiformes (Baldwin andJohnson 1996, Sato andNakabo 2002).Arambourg (1954) and Taverne (1991) included Protostomias in the order Stomiifomes, based on generalized anatomical features, such as general shape of the body and a massive and tooth-bearing dentary, as well as the position of the median fins.Taverne (1992), in his comprehensive review of Protostomias, retained its placement in Stomiiformes.
Yet, the paraphyly of Enchodontoidei not allowed their taxonomic classification in the cladistic context.
The family Apateopholidae is not a monophyletic group, as Apateopholis is the sister-taxon of the clade including the family Enchodontidae (new usage) and the genera Cimolichthys, Prionolepis, Halec, Phylactocephalus, and Serrilepis.Apateodus, often placed with the Apateopholidae, is the sister-group of Ichthyotringa.
The monophyly of Dercetidae proposed by Gallo et al. (2005) and Blanco et al. (2008) was corroborated and supported by a single synapomorphy (very reduced neural spine; character 71) (Fig. 2), but the inclusion of new taxa changed the interrelationships of the family.The genus Apuliadercetis is the sister-group of Brazilodercetis and they are the basal clade of Dercetidae.Cyranichthys is the sister-taxon of Robertichthys, and Dercetis is the sister-taxon of Ophidercetis.These two clades plus Benthesikyme form a new clade, which is an intermediary group between the basal and crown groups.However, the relationships within this intermediary group are uncertain.Caudadercetis appears as the most basal taxon in the major clade of Dercetidae (Caudadercetis, (Pelargorhynchus, (Nardodercetis, (Rhynchodercetis, (Hastichthys, Dercetoides)))).This clade is sustained by the unique presence of a convoluted suture marking the contact between second and third hypurals (character 83; Fig. 3B), although this character in Pelargorhynchus was coded as missing data, as its caudal skeleton is unknown.A pipe-shaped preopercle is an autapomorphy of Brazilodercetis (character 56; Fig. 4B).Hastichthys appears as the sister-taxon of Dercetoides.Taverne (2006b) suggested Dercetis as the most basal dercetids, and Ophidercetis as the sister-taxon of Cyranichthys.However, we obtained different results: Dercetis forms a clade with Ophidercetis, and Cyranichthys is related to Robertichthys.
The family Enchodontidae is confirmed as monophyletic, as has already been proposed by Fielitz (2004), but herein it possesses a new composition.Parenchodus forms a clade with Palaeolycus, and Enchodus is the sister-group of this clade; the clade formed by (Eurypholis, Saurorhamphus) is the basal sister-group.Rharbichthys was excluded from the Enchodontidae, being recognized in this analysis as the sister-group of the clade formed by the Dercetidae plus the genus Trachinocephalus.Fielitz (2004) proposed three synapomorphies of Enchodontidae: the presence of a single dermopalatine tooth, dermopalatine bone with same length or shorter than the tooth, and the absence of an interopercle.In this analysis, these features were not corroborated as synapomorphies.For instance, a single tooth on the dermopalatine is also present in Ophidercetis, a genus of Dercetidae (Taverne 2005b).The remaining synapomorphy of this family is the presence of middorsal scutes (character 87; Fig. 5).In addition, the clade (Eurypholis, Saurorhamphus) is supported by two synapomorphies: quadrate-mandibular articulation hidden (character 52; Fig. 6) and the presence of a spine on posterior border of the opercle (character 61; Fig. 7).
Halecidae possess a new composition: Halec is the sister-group of Phylactocephalus; and Hemisaurida was excluded from the family, being considered Aulopiformes incertae sedis, like Nardorex.

Fig. 1 -
Fig. 1 -Majority rule consensus.Values represented at the right side in each branch indicate the percentage of the trees in which each clade is present.Numbers at the left side associated with a black bar indicate the characters that resulted in synapomorphies or autapomorphies.