Petulanos, Sidlauskas & Vari, 2008

Sidlauskas, Brian L. & Vari, Richard P., 2008, Phylogenetic relationships within the South American fish family Anostomidae (Teleostei, Ostariophysi, Characiformes), Zoological Journal of the Linnean Society 154 (1), pp. 70-210 : 167-172

GENUS PETULANOS GEN. NOV.

Type species: Anostomus plicatus Eigenmann, 1912: 296 , designated herein

Diagnosis: Petulanos may be distinguished from the other members of the Anostomidae with strongly upturned mouths ( Anostomus , Gnathodolus , Pseudanos , Sartor and Synaptolaemus ) by the following combination of external characters: a series of dark spots greater than one scale wide along the lateral line, a series of dark bars across the dorsal surface of the body, four branchiostegal rays, a lack of dermal papillae on the lower lip, two cusps occurring on the symphyseal tooth of the premaxilla, the possession in adults of only a single cusp on the symphyseal and second tooth of the dentary, and the presence of three cusps on the third tooth of the dentary. Based on skeletal anatomy, Petulanos may be distinguished from all other members of the Anostomidae by the possession of a symplectic with a distinctive triangular ventral lamina. Petulanos further differs from all examined members of Anostomus , Gnathodolus , Pseudanos , Sartor and Synaptolaemus in possession of two intermediate pores along the ossified portion of the sensory canal of the second infraorbital (versus one), and from all members of the genera cited above except Pseudanos gracilis by possession of only one intermediate pore along the ossified portion of the sensory canal of the first infraorbital (versus two). It can be further distinguished from Anostomus and Pseudanos by a very wide ventral portion of the maxilla, a high urohyal with the angle between the dorsal and ventral margins greater than 45°, and the anterior termination of the lateral shelf of the quadrate at a point posterior of the articular condyle on that bone. Petulanos can be further distinguished from Gnathodolus , Sartor and Synaptolaemus by the posterior expansion of the sixth infraorbital and its contact or fusion with the suprapreopercle, the rotation of the anterolateral flange of the maxilla and the ventral portion of that bone relative to each other, and the lack of a triangular process on the ascending arm of the preopercle.

Etymology: Derived from the Latin petulans (impudent or petulant) and the first four letters of the generic name Anostomus from which Petulanos was separated; both in allusion to the characteristic upturned mouth that gives these fishes the appearance of a perpetual pout.

SUBFAMILIAL CLASSIFICATION IN ANOSTOMIDAE

The Anostomidae of the present study has in the past also been considered a subfamily Anostominae (e.g. Géry, 1961a). The Anostominae has also been utilized as a subfamily within the Anostomidae in several senses. Some authors (e.g. Géry, 1977; López, Menni & Miquelarena, 1987) recognized an Anostominae that included all of the Anostomidae with the exception of Leporellus that they placed in its own subfamily, the Leporellinae. Winterbottom (1980) alternatively recognized a more restrictive Anostominae consisting of the species in Anostomus (then including Petulanos ), Gnathodolus , Pseudanos , Sartor and Synaptolaemus . Interestingly, each of these subfamilial concepts delimits a monophyletic assemblage under the results of this study. The recognition of a Leporellinae and Anostominae sensu Géry (1977) and López et al. (1987) would highlight the basal dichotomy in the family, but would provide no sense of structure within the broadly encompassing Anostominae of those authors. Winterbottom’s (1980) more restricted Anostominae highlights the clade of species with superior mouths, but within the context of our phylogenetic results would require the proposal of several other subfamilies in order to satisfy the requirement that all subfamilies in the Anostomidae be monophyletic (even leaving aside the issues presented by the lack of resolution within Leporinus ). Such a course of action is premature in light of the issues we discuss above, and we consequently do not recognize any subfamilies within the Anostomidae at this time.

MAJOR SHIFTS IN JAW MORPHOLOGY

Nine major shifts in jaw morphology occurred during the evolution of the Anostomidae . Most of these represent stages of a transition from a plesiomorphically subterminal mouth to a terminal and then increasingly upturned mouth. One transition represents a return to a subterminal mouth. The nine transitions are scattered more or less evenly throughout the phylogeny of the family, suggesting that jaw diversification occurred throughout the history of the Anostomidae . In this light, the very highly modified, superior jaws possessed by some genera ( Gnathodolus , Sartor ) can be understood as the product of several progressive shifts away from a more plesiomorphic subterminal jaw morphology (e.g. Leporellus ).

The first shifts in jaw morphology (transition 1) within the Anostomidae are synapomorphies of the whole clade. All anostomids have the anguloarticular–quadrate joint in a more anterior position relative to the ventral wing of the lateral ethmoid and the orbit than is typical of other characiforms. The repositioning of the joint of the lower jaw was represented in the data matrix by character 83, the elongation of the interopercular–mandibular ligament. This modification also reflects the elongation of the anterior portion of the quadrate (not explicitly coded). Other transitions associated with jaw architecture at the base of the Anostomidae are the positioning of the retroarticular in a deeply recessed pocket of the dentary (character 65), the evolution of spade- or chisel-like teeth on the dentary (character 37), the evolution of a well-developed triangular ascending process of the premaxilla (character 45), the thickening of the lateral ethmoid–ectopterygoid ligament and the relocation of the insertion of that ligament to the ventral surface of the lateral ethmoid (character 24). This ligament, which binds the dorsal portion of the ectopterygoid to the neurocranium, appears to reduce or eliminate motion of the suspensorium relative to the neurocranium and may create the fixed component in a four-bar linkage (sensu Westneat, 2003, 2004, 2006). The single species in the Anostomidae to have lost this ligament ( Gnathodolus bidens , character 23) possesses a highly modified sliding relationship between the suspensorium and anteroventral portion of the neurocranium that is much different from that found elsewhere in the family, or indeed anywhere in the Characiformes .

The stabilization of the suspensorium provided by the lateral ethmoid–ectopterygoid ligament may be important to the function of the anteriorly shifted jaws in the Anostomidae exclusive of Gnathodolus , and the linkages between cranial elements in the family should be examined and modelled and the functional implications analysed. The diversification in jaw orientation within the Anostomidae may also represent diversification of linkage geometry, mechanical advantage and feeding ecology, but this conjecture remains to be tested across the family (for studies that explicitly link jaw shape to function and ecology, see Westneat, 1995; Hulsey & Wainwright, 2002; Wainwright et al., 2004).

The most basal members of the Anostomidae ( Leporellus , Fig. 33 ) have slightly downturned mouths similar to that in Caenotropus in the closest outgroup, the Chilodontidae . The members of the next most derived genus in the Anostomidae (Hypomasticus) also have subterminal mouths, and the downturned mouth position appears to be the plesiomorphic condition for the Anostomidae . Hypomasticus and basal members of its sister taxon ( Leporinus within clade 7) have, however, modified the plesiomorphic jaw morphology in different ways.

The species in Hypomasticus (formerly a subgenus of Leporinus ) evolved a more exaggerated subterminal mouth position (transition 2; Fig. 34 ) characterized by a vertically orientated premaxilla (character 46) and a widened, trough-like morphology of the anguloarticular (characters 55 and 56). The apparent function of this trough was discussed in the character descriptions, but briefly, it appears to function in the manner of a pulley, altering the direction of the force vector applied by the adductor mandibulae to the maxilla.

In the sister group to Hypomasticus (clade 7), the mouth became fully terminal (transition 3) and the basal components of this clade have assumed the condition typical of Leporinus ( Fig. 35 ). The anterior portion of the mesethmoid which attaches to the premaxilla shifted from ventrally hooked ( Figs 16 , 17 ) to straight (character 15; Fig. 18 ) and the process of the mesethmoid connecting to the vomer became posteroventrally aligned (character 17; Fig. 18 ). The maxilla acquired a vertical or posterodorsal orientation (character 48; Fig. 35 ) and the ectopterygoid assumed a vertical orientation (character 72; Fig. 41 ) as the joint of the lower jaw with the quadrate became more anteriorly positioned than was the case in Leporellus , Hypomasticus and the proximate outgroup, the Chilodontidae .

The next major shift in jaw morphology (transition 4) occurred in clades 13 and 14, all members of which possess upturned mouths at some point in ontogeny. Rhytiodus and Schizodon have supraterminal mouths only as small juveniles and larvae and have terminal or subterminal mouths as adults. Anostomoides has only a slightly supraterminal mouth in the adult condition (juveniles of the genus not examined), while Laemolyta , Pseudanos , Anostomus , Petulanos , Synaptolaemus , Sartor and Gnathodolus have supraterminal or fully superior mouths apparently throughout ontogeny (small juveniles of Synaptolaemus , Sartor and Gnathodolus were not examined). Transition 4 involved a ventral shift in the position of the retroarticular in clade 13 (character 65; compare Figs 35 , 36 ). In the slightly more derived clade 14, the functional edge of the lower teeth shifted from the posterior lamina to the distal margin of the tooth (compare Fig. 32A–C with Fig. 32E, F ), and the orientation of the teeth in the dentary relative to that bone also changed (compare Fig. 35 with Figs 36 , 38 ; see discussion of character 58). The shift in orientation may have changed the angle of force applied by the teeth. A biomechanical model [e.g. that of Westneat (2003), or a modification thereof] should be applied to test this conjecture and fully characterize the transition in jaw function in clades 13 and 14.

The jaw of Schizodon nasutus ( Fig. 39 ) within clade 13 displays the most remarkable instance of convergence in the Anostomidae (transition 5). This species evolved a strongly subterminal mouth similar to, but independent of, the condition in Hypomasticus ( Fig. 34 ). Schizodon nasutus also independently acquired three of the four skeletal morphologies that diagnose Hypomasticus : the transversely widened anguloarticular (character 55, also shared with all species of Schizodon ), the trough-like ascending process of the anguloarticular (character 56) and the vertically aligned premaxilla (character 46). The convergence between Hypomasticus and Schizodon nasutus appears to be an excellent example of two lineages evolving the same solution to a comparable mechanical problem, effective functioning of ventrally rotated oral jaws. It has not, however, been demonstrated that the subterminal jaws of Schizodon nasutus and Hypomasticus are convergent functionally as well as morphologically.

The next major shift in jaw evolution in the Anostomidae (transition 6) occurred in clade 21, all the members of which have fully upturned jaws throughout ontogeny. Transition 6 is characterized by a fully horizontal orientation of the premaxilla (character 46; Figs 36 , 37 ) and the reduction or elimination of the posterior lamina of the dentary teeth (compare Fig. 32E with 31E or 32F). The ontogenetic shift in mouth position in Rhytiodus and Schizodon (see discussion in Sidlauskas et al., 2007), which together form the sister to clade 21, suggests that the strongly upturned mouths of the taxa in clade 21 may have resulted from heterochrony (paedomorphic fixation of the juvenile state in adults). The conjecture of heterochrony should be investigated further by comparison of ontogenetic series of skeletal preparations.

The six genera in clade 25 have the most strongly upturned jaws in the Anostomidae , and indeed proximate outgroup families, with the mouth opening completely on the dorsal surface of the head. Several morphological synapomorphies characterize this shift (transition 7). The ventral process of the mesethmoid runs vertically or nearly so (character 17), the ventral surface of the vomer developed a pentagonal raised area that contacts the restructured palatine (characters 21 and 67), and the ectopterygoid became almost totally posterodorsally inclined (character 66; Fig. 54) as the joint of the lower jaw with the quadrate moved ever more anteriorly. Somewhere near the base of this clade, a notch also developed in the lateral ethmoid to accommodate the now nearly horizontal alignment of the ectopterygoid–lateral ethmoid ligament (character 25; Fig. 26 ). The level at which this notch of the lateral ethmoid is a synapomorphy is uncertain due to the lack of resolution concerning Pseudanos .

The penultimate shift in jaw orientation in the Anostomidae occurred in the lineage leading to Sartor and Gnathodolus (transition 8). These two genera have the mouth so strongly upturned (backwards facing in Sartor ) that the quadrate–anguloarticular joint is relocated to anterior of the opening of the mouth ( Fig. 37 ). The maxilla also became plate-like with a distinctive medial process (character 54), the dentary teeth became elongate and recurved (character 37; Figs 31H, I , 37 ), and the premaxillary teeth became bowed (character 35; Fig. 29G, H ).

The final major shift in jaw morphology in the family (transition 9) occurred in the derived genus Gnathodolus , in which the dentary was reduced to a slender cylinder bearing only a single tooth (characters 33 and 57; Fig. 31I ), the ligamentous connection of the lateral ethmoid to the ectopterygoid that characterizes all other anostomids was lost (character 23), the palatine assumed the shape of an elongate hourglass (character 70) and the vomer developed a deep groove on each side into which the dorsal portions of the mesopterygoid and metapterygoid fit (character 20, Fig. 22 ; see also Winterbottom, 1980: fig. 58). Functionality of the jaws in Gnathodolus appears to require that the dorsal portion of the suspensorium slide along grooves in the ventral portion of the neurocranium. That would be impossible if the lateral ethmoid–ectopterygoid ligament were retained. If so, the linkages within the skull of Gnathodolus may differ considerably from those found in other anostomids and may require construction of a new biomechanical model.

PHYLOGENETIC BIOGEOGRAPHY

All 14 genera in the Anostomidae comprising approximately 140 species are present in the vast cis-Andean mid-elevation and lowland portions of tropical and temperate South America. Anostomids inhabit myriad river systems from the Caribbean Sea versant drainages of northern Venezuela ( Lasso et al., 2004a) through the Río Orinoco ( Lasso et al., 2004b) and Rio Amazonas ( Garavello & Britski, 2003) basins south to the Río de La Plata system ( Britski, Silimon & Lopes, 1999; López, Miquelarena & Menni, 2003; Menni, 2004). Anostomids are also diverse in the Atlantic versant rivers of the continent from the Guianas ( Lowe-McConnell, 1964; Planquette et al., 1996) south through the coastal rivers of eastern Brazil ( Britski, Sato & Rosa, 1984; Malabarba, 1989). Many fewer anostomid genera and species inhabit the less extensive but topographically complex trans-Andean region of South America. Only six species in four genera are now recognized in the arch from the Lago Maracaibo basin of northwestern Venezuela ( Schizodon, Vari & Raredon, 1991 ) through the Caribbean and Pacific slope rivers of northern and western Colombia ( Abramites , Leporellus , Leporinus, Dahl, 1971 ; Mojica-C, 1999) to the Río Guayas basin of southeastern Ecuador ( Leporinus, Eigenmann & Henn, 1916 ; Barriga, 1991).

Incomplete understanding of anostomid species diversity and geographical distribution at the species level preclude a fine-grained analysis of their phylogenetic biogeography. This problematic situation is exemplified by the continuing publications describing new species within the family and/or refining concepts of species limits and geographical ranges for previously described forms (e.g. Sidlauskas & Santos, 2005; Sidlauskas et al., 2007). Although the phylogenetic reconstruction arrived at herein included only a subset of the species and yielded incomplete resolution, particularly within Leporinus ( Fig. 5 ), it nevertheless allows estimates of minimum ages for several major intrafamilial cladogenic events.

An interesting pattern is apparent when the distribution of genera of the Anostomidae to the two sides of the Andean Cordilleras is evaluated within the context of the arrived at phylogeny ( Fig. 5 ). This analysis finds that those genera in the family with representatives on both sides of that mountain chain are all restricted to the somewhat more basal components of the phylogeny. These are Abramites with one species ( A. hypselonotus ) broadly distributed in cis- Andean regions and its congener ( A. eques ) endemic to the Río Magdalena system ( Vari & Williams, 1987; Mojica, 1999), Leporellus with several nominal species occurring east of the Andes and one form identified as L. vittatus reported from the Río Magdalena basin ( Dahl, 1971; Maldonaldo-Ocampo et al., 2005), Schizodon with representatives broadly distributed east of the Andes ( Sidlauskas et al., 2007, and references therein) and one species, S. corti , endemic to the Lago Maracaibo basin of northern Venezuela ( Vari & Raredon, 1991; Lasso et al., 2004a) and finally Leporinus , a genus with over 90 recognized species ( Garavello & Britski, 2003), only three of which are reported to occur in river systems to the west of the Cordilleras. These are L. ecuadorensis from the Río Guayas basin of Ecuador ( Eigenmann & Henn, 1916; Barriga, 1991), L. muyscorum from the Río Magdalena, Río Atrato, Río Ranchería and Río Sinú basins of northern Colombia (Mojica-C, 1999; Mojica et al., 2006), and L. striatus , reported from river systems of northern and western Colombia ( Dahl, 1971; Mojica-C, 1999).

Although some species of the Anostomidae live in mid-level, upland river systems (e.g. Leporellus in the Río Magdalena system, Maldonaldo-Ocampo et al., 2005), most occur in lower elevation settings and no members of the family are inhabitants of the precipitous drainage systems of the higher regions of the Andean Cordilleras. As such, the uplift of the Andean Cordilleras probably served as a vicariant event or sequential events dividing the ancestral components of Abramites , Leporellus , Leporinus and Schizodon into cis- and trans-Andean components.

Schizodon (clade 18) is the most deeply nested of the clades within the Anostomidae that includes cis- and trans-Andean components. Schizodon corti , an endemic to the trans-Andean Lago Maracaibo basin ( Vari & Raredon, 1991; Lasso et al., 2004a) and the sole member of the genus found in a trans-Andean drainage, was not incorporated into the phylogenetic analysis of this study. Schizodon corti shares, however, the externally obvious attributes of its congeners and it is reasonable to assume that S. corti is a component of clade 18. It follows directly that clades arising at progressively more inclusive nodes within the phylogeny of the Anostomidae below clade 18 (the ancestors of clades 1, 3, 7, 8, 12, 13, 14 and 15; Fig. 5 ) evolved prior to this isolation of components of that family to each side of the uplifting Cordilleras. Furthermore, the ancestors of the three genera with members to the two sides of the Andes ( Abramites , Leporellus , Leporinus ) evolved prior to the uplift and the same applies to Anostomoides and Hypomasticus , genera limited to river basins east of the Andes. Given that clade 15 (containing Schizodon and Rhytiodus ) is the sister to clade 21 ( Laemolyta , Pseudanos , Anostomus , Petulanos , Synaptolaemus , Gnathodolus , and Sartor ), stem members of clade 21 must also have evolved by the final uplift of the Cordilleras.

The specific geological event associated with final vicariance into cis- and trans-Andean components of the Anostomidae cannot be identified in so far as it is impossible to determine the degree of uplift that resulted in the ultimate division of that component of the ichthyofauna into eastern and western subunits. Nonetheless, the minimum age for such vicariance would be the closure of the north-flowing Maracaibo– Falcon outlet of the Río Orinoco. That closure occurred approximately 8 Mya during the final major uplift sequence of the northern portions of the Andean Cordilleras ( Hoorn, 1993; Hoorn et al., 1995; Lundberg et al., 1998). Evidence from the phylogeny when correlated with geological information is, however, further informative in terms of older likely vicariance events for the Anostomidae .

Schizodon corti , which is endemic to the Lago Maracaibo basin, is the sole trans-Andean species of the Anostomidae in that basin ( Lasso et al., 2004a). The other trans-Andean species of the Anostomidae (species of Abramites , Leporellus and Leporinus ) inhabit river systems that lie west of the central and western Andean Cordilleras in the portions of South America from the Río Guayas to the Río Ranchería. These mountain ranges underwent uplift prior to the closure of the Maracaibo– Falcon outlet with the resultant isolation of Lago Maracaibo from what is now the Río Orinoco basin. The most recent and thus pertinent of the major uplift events in that region that is relevant to the question of a minimum date for the division of the anostomid faunas east and west of the Central and Western Andean Cordilleras involved the Sierra de Perija, the mountain range between the Lago Maracaibo basin and the river systems that drain regions to the west of those uplands. This uplift event of that portion of the northeastern Andes took place approximately 11.8 Mya as evidenced by the westward shifts of the palaeocurrents within the present valley of the Río Magdalena system ( Hoorn et al., 1995; Lundberg et al., 1998). The uplift of the Sierra de Perija 11.8 Mya provides a minimum age for the Anostomidae and for the ancestors of Abramites , Leporellus , Hypomasticus and Leporinus (clades 1, 2, 3, 7, 8, 9; Fig. 5 ). Although Abramites and Leporinus form a largely unresolved polytomy in our final phylogeny ( Fig. 5 ), that lack of resolution does not bear on the hypothesis of the age of these genera as the phylogenetic hypothesis clusters all members of those genera at clade 8. As such the ancestor of the subunit of the Anostomidae at that level of the phylogeny also was present prior to 11.8 Mya.

When we consider the degree of morphological diversity present between the Anostomidae and its close relatives, the Chilodontidae , Curimatidae and Prochilodontidae , the latter two of which also have cis- and trans-Andean components ( Vari, 1988; Castro & Vari, 2004) and the fact that a series of cladogenic events within the Anostomidae (the basal splits within clades 1, 3, 7 and 8) preceded the vicariant events of 11.8 Mya, it seems very likely that the origin of the family and its basal clades most likely significantly pre-dated that time. Evidence from the fossil record of the Anostomidae and its close relatives is informative on this question.

Fossils of the Anostomidae have been rarely cited in the literature. It is uncertain whether this situation represents the rarity of such material in deposits within South America that include remains of freshwater taxa or the fact that fossils of the family remain unrecognized among samples of fossil fishes from such strata. Two reports of fossil anostomids are that of Roberts (1975) for remains from the Cuenca basin of southern Ecuador at approximately 19 Mya ( Roberts, 1975: 261), and the more recent citation by Lundberg (1997: 73) based on material from the La Venta formation in Colombia from between 13.5 and 11.5 million years old ( Guerrero, 1997: 41). Both of these samples are limited to teeth identified as having originated with a species of Leporinus (jaw and pharyngeal teeth in the Ecuadorian samples and a single premaxillary tooth in the case of the La Venta material). The age of these deposits minimally confirms the existence of species of Leporinus at the time of the uplift of the Sierra de Perija 11.8 Mya, a minimum age for the genus independently postulated above on the basis of distributional data evaluated in a geological context. The oldest possible age for the fossils (19 Myr for the Cuenca material) would indicate that the family and clades including Leporinus significantly pre-date that time period.

Fossils of close relatives of the Anostomidae within the Characiformes are also very limited and only one is informative as to the minimum age of the more inclusive clade including that family plus its close relatives, the Chilodontidae , Curimatidae , and Prochilodontidae . Malabarba (1996) demonstrated that a fossil fish described by Travassos & Santos (1955) as Curimata mosesi from Oligocene deposits of the Tremembé basin of eastern Brazil was assignable to the curimatid genus Cyphocharax , a component of a terminal polytomy deeply nested within the phylogeny of the Curimatidae ( Vari, 1989a) . The deposits from which the specimens of Cyphocharax mosesi originated are at least 22.5 million years old ( Malabarba, 1998; Reis, 1998) and the ancestors of the clade formed by the Curimatidae along with the Prochilodontidae and of that clade plus the Anostomidae and Chilodontidae therefore pre-date that time. Although not directly informative as to the age of the Anostomidae , this older date of 22.5 Myr is, nonetheless, congruent with the likely diversification of the Anostomidae well in advance of the uplift episodes of the Andean Cordilleras cited above as pertinent to likely vicariance events within the family. That date also distinctly pre-dates the minimum age of 11.5–13.5 Myr for the Anostomidae indicated by the limited fossil evidence for that family. The numerous cladogenic events between the divergence of the anostomid–chilodontid clade and the ancestor of Cyphocharax in conjunction with the pronounced degree of morphological divergences in each of those lineages are presumably indicative of a lengthy intervening time period, but it is impossible to determine whether the approximately 10 Myr between those events would accommodate that level of divergence.

Although not directly informative on that issue, evidence does indicate that the Characiformes evolved long in advance of even the 22.5-Myr date. The Characiformes as a whole long pre-dates the earliest fossils assignable to the Anostomidae as evidenced by both Old World fossil characiforms of the genera Alestes , Brycinus and Bryconaethiops in the family Alestidae ( Zanata & Vari, 2005) that extend back approximately 49–54.8 Myr. It is also noteworthy that several instances of trans-Atlantic relationships within the Characiformes (see summary in Zanata & Vari, 2005: 120) would extend the African– South American drift vicariance at several phylogenetic levels within the Characiformes back approximately 90–112 Mya. There was thus a lengthy time period to accommodate the accumulation of evolutionary novelties between the origin of the Characiformes and the earliest fossils known at the level of the clade formed by the Anostomidae , Chilodontidae , Curimatidae and Prochilodontidae or of Leporinus in the Anostomidae .

Figures

Figure 5. Strict consensus of 1312 most parsimonious trees generated in PAUP* with branches of zero maximum length collapsed and all characters unordered. Repeated analysis with branches of zero minimum branch length collapsed and/or multistate characters ordered yielded the same consensus topology. Each of the most parsimonious trees has length = 529, CI = 0.425 and = 0.805. Numbers above nodes indicate Bremer support values, numbers below nodes indicate bootstrap percentages. The letter C indicates where the Citharinidae and Distichodontidae were constrained to cluster via their joint designation as a monophyletic root.

Figure 16. Right anterior portion of neurocranium of Leporellus vittatus, INHS 56128, 78.2 mm SL; lateral view, drawing reversed to place anterior at left.

Figure 17. Left anterior portion of neurocranium of Hypomasticus mormyrops, USNM 318140, 104 mm SL; lateral view.

Figure 18. Left anterior portion of neurocranium of Leporinus fasciatus, USNM 225991, 87.8 mm SL; lateral view.

Figure 22. Anterior portion of neurocranium of Gnathodolus bidens, USNM 389623, 71 mm SL; ventral view.

Figure 26. Anterior portion of neurocranium of Petulanos plicatus, USNM 225396, 95.3 mm SL; ventral view.

Figure 29. Dentition of right premaxilla in (A) Leporellus vittatus, INHS 56128; (B) Hypomasticus mormyrops, FMNH 112963, margin of symphyseal tooth slightly damaged; (C) Hypomasticus despaxi, MCZ 56552; (D) Leporinus jatuncochi; INHS 38940, posterior portion of premaxilla damaged and missing; (E) Rhytiodus argenteofuscus, INHS 66082; (F) Schizodon fasciatus, FMNH 111351; (G) Sartor elongatus (paratype), INPA 1168; (H) Gnathodolus bidens, ANSP 159389; lateral view, scale bar = 200 Mm, images rotated to standardize orientation of premaxilla, anterior to right of image in A–C, to bottom-right in D–F and to bottom in G and H.

Figure 31. Dentition of right dentary in (A) Leporellus vittatus, INHS 56128; (B) Hypomasticus despaxi, MCZ 56552; (C) Leporinus jatuncochi, INHS 38940; (D) Leporinus cf. fasciatus, FMNH 103450; (E) Laemolyta taeniata, INHS 61513; (F) Petulanos intermedius, INPA 15184; (G) Synaptolaemus cingulatus, FMNH 103455; (H) Sartor elongatus, INPA 1168; (I) Gnathodolus bidens, ANSP 159389; lateral view, scale bar = 200 Mm, images rotated to standardize orientation of dentary, anterior to top of image in A and B, to right in C and D, to bottom right in E–G, and to bottom in H and I.

Figure 32. Dentition of right dentary in (A) Leporinus cf. moralesi, UMMZ 216435; (B) Abramites hypselonotus, FMNH 106574; (C) Anostomoides laticeps, INHS53677, posterior (fourth) tooth dislodged from socket; (D) Rhytiodus argenteofuscus, INHS 66082; (E) Schizodon fasciatus, FMNH 11135; (F) Pseudanos trimaculatus, FMNH 102122; medial view, scale bar = 200 Mm, images rotated to standardize orientation of dentary, anterior to left in A, B, D and E, to bottom left in C and to bottom in F.

Figure 33. Left upper and lower jaws of Leporellus vittatus, INHS 56128, 78.2 mm SL; lateral view.

Figure 34. Left upper and lower jaws and anterior palatine arch, Hypomasticus mormyrops, USNM 318140, 104 mm SL; lateral view, jaws spread wider in drawing of dissected specimen than typical in life.

Figure 35. Left upper and lower jaws, Leporinus fasciatus, USNM 225991, 87.8 mm SL; lateral view, jaws spread wider in drawing of dissected specimen than typical in life.

Figure 36. Left upper and lower jaws, Laemolyta taeniata, USNM 280708, 72.9 mm SL; jaws spread wider in drawing of dissected specimen than typical in life.

Figure 37. Right upper and lower jaws, Sartor elongatus, INPA 1168 (paratype), 76.1 mm SL; lateral view, jaws spread wider in drawing of dissected specimen than typical in life, drawing reversed to place anterior at left.

Figure 38. Right lower jaw, Schizodon fasciatus, USNM 280739, 103.2 mm SL; lateral view, drawing reversed to place anterior at left.

Figure 39. Left lower jaw, Schizodon nasutus, USNM 326944, 111 mm SL; lateral view.

Figure 41. Anterior portion of left suspensorium, Leporinus fasciatus, USNM 225991, 87.8 mm SL; lateral view.