Cyanocharax, Malabarba & Weitzman, 2003

Malabarba, Luiz R. & Weitzman, Stanley H., 2003, Description Of A New Genus With Six New Species From Southern Brazil, Uruguay And Argentina, With A Discussion Of A Putative Characid Clade (Teleostei: Characiformes: Characidae), Comun. Mus. Ciênc. Tecnol. PUCRS 16 (1), pp. 67-151 : 73-92

Cyanocharax Relationships

The phylogenetic analysis presented below is not an in-depth phyloge ­ ny of characid species. Instead, it is an up-dated presentation of previously published hypotheses of relationships among characids plus their relationships with Cyanocharax . Only characters found to be consistent in hypothe - sizing relationships and diagnosing monophyletic groups among characids are discussed. This excludes several ambiguous characters presented in the hypotheses of Lucena (1993) and Buckup (1998: 125- 134). The simplified cladogram presented in Fig. 2 summarizes current knowledge about characid phylogeny, with the addition of a few characters we found informative.

1) The presence of hooks on fin rays. The presence ofhooks on the anal-and pelvic-fin rays of males, often on the caudal fin and rarely on the dorsal and/or pectoral fins has been described in Gasteropelecidae ( Weitzman, 1954; Collette, 1977), Serrasalminae ( Jégu et al., 1989:56, Fig. 8 ) and in several repre ­ sentatives of the Characidae such as Cheirodontinae ( Malabarba, 1998; Malabarba & Weitzman, 1999, 2000), Glandulocaudinae ( Weitzman & Fink, 1985; Menezes & Weitzman, 1990; Weitzman et al., 1988; Géry, 1977; Nelson, 1964, Weitzman & Thomerson, 1970; Weitzman & Menezes, 1998), Stethaprioninae ( Reis, 1989), Charax and related genera ( Charax, Lucena, 1987 ; Phenacogaster, Malabarba & Lucena, 1995 ; Priocharax, Weitzman & Vari, 1987 ; Roeboides, Lucena 2000 a, 2000b ), Aphyocharacinae ( Aphyocharax, Wiley & Collette, 1970 ), Iguanodectinae ( Böhlke, 1954), Rhoadsiinae (pers. obs.), and several Tetragonopterinae and incertae sedis characid genera (the literature cited is not exhaustive): Astyanax, Garutti (1990) ; Attonitus, Vari & Ortega (2000) ; Aulixidens, Böhlke (1952) ; Bario, Wiley & Collette (1970) ; Boehlkea, Géry (1966) ; Bramocharax (pers. obs.); Bryconexodon, Jégu et al. (1991:780 , Fig. 7 ); Brittanichthys, Malabarba & Weitzman (1999) ; Brycon (Flávio Lima, pers. comm, and unpublished M.Sc. thesis); Bryconamericus, Malabarba & Kindel (1995) ; Bryconops, Wiley & Collette (1970) ; Caiapobrycon, Malabarba & Vari (2000) ; Creagrutus, Vari & Harold (2001) ; Eretmobrycon (pers. obs.); Genycharax (pers. obs.); Gynmocorymbus (pers. obs.); Hemibrycon, Wiley & Collette (1970) ; Hemigrammus, Wiley & Collette (1970) ; Hollandichthys (pers. obs.); Hyphessobrycon, Wiley & Collette (1970) ; Jupiaba, Zanata (1997) ; Knodus (pers. obs.); Lignobrycon, M. Malabarba (1998) ; Microschemobrycon, Böhlke (1953) , Géry (1973); Moenkhausia, Wiley & Collette (1970) ; Nematocharax, Weitzman et al. (1986) ; Odontostoechus (pers. obs.); Oligosarcus, Wiley & Collette (1970 ; as Acestrorhamphus bolivianus ), Paracheirodon, Weitzman & Fink (1983) ; Petitella (pers. obs.); Piabarchus (pers. obs.); Piabina, Van & Harold (2001) ; Prionobrama (pers. obs.); Rachoviscus, Weitzman & Cruz (1981) ; Rhinobrycon (pers. obs.); Rhinopetitia (pers. obs.); Salminus, Morais-Filho & Schubart (1955 , figs. 2-3), Tyttobrycon, Géry (1973) .

Although the fin-ray hooks, almost always secondary sexual characters of males, may greatly differ in shape, position and possibly function among the species of these groups, they are always observed to be a bony spiny process developed on the surface of individual segments of lepidotrichia. We hypothesize that these hooks represent an expression of a gene or genes absent in other American or African characiforms, namely the Distichodontidae , Citharinidae , Parodontidae , Curimatidae , Prochilodontidae , Anostomidae , Chilodontidae , Crenuchidae , Hemiodontidae , Alestidae , Acestrorhynchidae , Erythrinidae , Lebiasinidae , Ctenoluciidae and Hepsetidae . Fin-ray hooks are also absent in siluriforms and cypriniforms, suggesting that the presence of fin hooks is a derived character synapomorphic for certain subgroups of characiforms ( Fig. 2 ), consisting of the Gasteropelecidae and a large part of the Characidae . Outside of these groups, hooks were found only in Gilbertolus (pers. obs.) and Roestes (Lucena, pers. com.) of the subfamily Roestinae of the Cynodontidae , but are absent in Cynodon , Rhaphiodon and Hydrolycus (Toledo-Piza, pers. com.) of the subfamily Cynodontinae . Based on the uniqueness of this character among ostariophysans and the apparent homology be - tween the fin hooks in Roestinae and the groups listed above, we suggest that a re-analysis of Roestinae and Cynodontinae relationships is needed.

Among the remaining so-called characid fishes, the presence of hooks on the fins is to date unknown in the Agoniatinae, and the following genera: Aphyocharacidium , Aphyodite , Astyanacinus , Atopomesus , Axelrodia , Bryconacidnus , Bryconella , Ceratobranchia , Chalceus , Clupeacharax , Coptobrycon , Deuterodon , Engraulisoma , Exodon , Grundulus , Gymnocharacinus , Gymnotichthys , Hasemania , Leptagoniates , Leptobrycon, Lobodeuterodon , Markiana , Microgenys , Mixobrycon , Monotocheirodon , Nematobrycon , Oligobrycon , Othonocheirodus , Oxybrycon , Paragoniates , Parapristella , Parecbasis , Phenagoniates , Pristella , Probolodus , Psalidodon , Psellogrammus , Pseudochalceus , Roeboexodon , Schultzites , Scissor , Serrabrycon , Stichonodon , Stygichthys , Tetragonopterus , Thayeria , Thrissobrycon , Tucanoichthys , and Xenagoniates .

Since hooks are usually sexually dimorphic features and in most species are observed only in mature males, we reserve judgment about the lack of hooks in the genera listed above, most of which are represented by a few lots in collections and in which positive identification of the presence of sexually mature males is lacking. However, hooks were consistently reported as absent in Inpaichthys kerri (by Géry & Junk, 1977), Hypobrycon maromba and H. leptorhynchus (by Silva & Malabarba, 1996), and Triportheus (M. Malabarba, 1998) . We have examined mature males of Carlastyanax with enlarged fin rays that lack hooks.

Among these genera, Inpaichthys , Carlastyanax and Hypobrycon also lack a supraorbital bone (see character 2), and Carlastyanax and Hypobrycon have dorsal-fin rays reduced to two unbranched plus eight branched elements (see character 3) suggesting that the lack of fin hooks on these genera is probably a reversal. Triportheus has been hypothesized as closely related to Lignobrycon and Brycon (M. Malabarba, 1998) , and since both of these genera have bony hooks, it is suggested that the absence ofbony hooks in Triportheus is also a reversal.

Cyanocharax species have pelvic- and anal-fin hooks ( Figs. 3 -8 ), suggesting they belong to a clade of South American characiforms with fin hooks.

2) Absence of a supraorbital. Among Characiformes , the absence of a supraorbital is found only in the clade formed by Erythrinidae and Lebiasinidae ( Weitzman, 1964:141) , and in the Characidae members of the Aphyocharacinae, Characinae, Cheirodontinae, Glandulocaudinae, Paragoniatinae, Rhoadsiinae, Stethaprioninae , and Tetragonopterinae, as defined by Géry (1977). Cyanocharax also lacks a supraorbital bone, a reductive character that may possibly group it with the subfamilies listed above ( Fig. 2 ) in a monophyletic Characidae ( Weitzman & Malabarba, 1998; note that we failed to cite the presence of a supraorbital in Agoniatinae that paper). The lack of supraorbital in the clade Erythrinidae and Lebiasinidae is clearly homoplastic with respect to the absence of that bone in the Cyanocharax and the other groups listed for the Characidae , according to the hypotheses of relationships relating the Erythrinidae and Lebiasinidae to the Ctenoluciidae and Hepsetidae presented by Vari (1995) and Buckup (1998).

,

3) Dorsal-fin rays reduced to two unbranched plus eight branched elements. The number of elements in the dorsal-fin skeleton of characid fishes rarely varies. The usual characid fin-ray count is two anterior unbranched rays plus nine branched rays. This putative primitive number (ii,9) is found among all the species of the so-called Characidae that bear fin-ray hooks and retain a supraorbital bone (the species of Brycon , Bryconops , Lignobrycon , Salminus ), with the exception of the Serrasalminae which has higher counts and the Gasteropelecidae in which the number of rays may vary from ii,6-8 or iii,5-7 to ii,10- 15 or iii,9- 14 among the three genera ( Weitzman, 1960). Among species that bear no hooks on fins, the Agoniatinae and Triportheus also have ii,9 rays; Chalceus ii, 10; and the genera Clupeacharax and Engraulisoma ii,8.

Among characid fishes lacking the supraorbital, the apparently primitive dorsal-fin ray number is equal to ii,9 in all the examined species of the Cheirodontinae (genera Cheirodon , Nanocheirodon , Heterocheirodon , Spintherobolus , Serrapinnus, Megacheirodon , Compsura , Saccoderma , Macropsobrycon , Acinocheirodon , Kolpotocheirodon , Odontostilbe , Prodontocharax , Cheirodontops , Aphyocheirodon , and Pseudocheirodon ), the Stethaprioninae (ii,9-ii, 10), the Iguanodectinae (ii,9 in Iguanodectes and Piabucus ; ii,8-9 in some Piabucus species), the “Characinae” ( Charax Clade of Lucena, 1998), Aphyocharacinae ( Aphyocharax ), and the Rhoadsiinae ( Carlana , Rhoadsia and Parastremma ).

We found two states of this character among genera of the so-called Tetragonopterinae, as well as among the Glandulocaudinae: the putative primitive state (ii,9), and a hypothesized derived condition of two anterior unbranched rays plus eight branched rays (ii,8).

The primitive state (ii,9) is found among the examined species of the tetragonopterine or incertae sedis characid genera Aphyocharacidium , Aphyodite , Astyanacinus , Astyanax , Atopomesus , Aulixidens , Axelrodia , Bario , Bramocharax , Brittanichthys , Bryconella , Bryconexodon , Coptobrycon , Ctenobrycon , Deuterodon , Eretmobrycon , Exodon , Genycharax , Grundulus , Gymnocharacinus , Gymnocorymbus , Gymnotichthys , Hasemania, Hemigramnus , Hollandichthys , Hyphessobrycon , Inpaichthys , Jupiaba , Leptobrycon , Markiana , Microschemobrycon , Mixobrycon , Moenkhausia , Nematobrycon , Nematocharax , Oligobrycon , Oligosarcus , Oxybrycon , Paracheirodon (ii,8 or ii,9), Parapristella , Parecbasis , Petitella , Pristella , Probolodus , Psalidodon , Psellogrammus , Pseudochalceus , Roeboexodon , Schultzites , Scissor , Serrabrycon (ii, 8 or ii,9), Stichonodon , Stygichthys , Tetragonopterus , Thayeria , Thrissobrycon , Tucanoichthys , Tyttobrycon (ii,8 or ii,9).

The derived condition (ii,8) is found among all the examined species of the genera of Géry (1966) ’ s Hemibryconini ( Boehlkea , Bryconacidnus , Bryconamericus , Ceratobranchia , Hemibrycon , Knodus , Microgenys , and Rhinobrycon ), and Géry (1966) ’ s Creagrutini ( Creagrutus , Creagrudite (= Creagrutus ), Piabarchus , and Piabina ). A number ofother genera were referred by Géry (1966) as possibly belonging to his Hemibryconini, or to the group called “ Hemibrycon and allied genera” ( Géry, 1977). Among these genera, only Carlastyanax and Rhinopetitia have ii.8 dorsal-fin rays. A number of ii,8 dorsal-fin rays is also present in Attontus, Caiapobrycon , Hypobrycon , Monotocheirodon , Odontostoechus , Othonocheirodus , " Astyanax ” alburnus and the species of the new genus Cyanocharax ( Fig. 9 ).

Among characid fishes lacking a supraorbital, only glandulocaudines share the derived presence of ii.8 dorsal-fin rays with the taxa listed in the preceding paragraph. Glandulocaudines. however, also show the largest variability in dorsal-fin ray number when compared to other chaiacid subgroups, as such cheirodontines, characines, aphyocharacines, iguanodectines, rhoadsiines and stethaprionines. The it,8 dorsal-fin ray formula is present in all genera of the Diapomini ( Diapoma , Planaltina and Acrobrycon ) and Corynopomini ( Gephyrocharax , Pterobrycon and Corynopoma ). Among the Hysteronotini, both species of Hysteronotus and Pseudocorynopoma have ii,8 or ii,9, (e.g. ii, 8 in Pseudocorynopoma heterandria and ii,9 in P. doriae ), In the Glandulocaudini, the basal genus Glandulocauda (see Menezes & Weitzman, 1990) and most species of Mimagoniates (e. g. M. microlepis and M. sylvicola ) have ii,8 dorsal-fin rays, but one of the most derived species of the tribe, M. rheocharis , has an increased number of dorsal-fin rays and an intraspecific range from ii, 8 to ii, 12. Xenurobryconins have two unbranched and a highly variable number of branched dorsal-fin rays, ranging from 6-7, rarely 8 in the small sized species of Xenurobrycon ; 7 or 8 in Tyttocharax ; 8 in the relatively basal genera lotabrycon and Argopleura ; 9 in Ptychocharax , and 9 or 10 in Chrysobrycon that includes the largest species of the tribe.

The increased dorsal-fin ray number is apparently secondary in glandulocaudines, as exemplified by Mimagoniates rheocharis and Chrysobrycon myersii . In both cases, these species correspond to derived taxa of their tribes (Glandulocaudini and Xenurobryconini), while basal species have ii,8 dorsal fin rays. The increased dorsal-fin ray number in Glandulocaudinae may be related to the elaborate courtship behavior of these fishes ( Nelson, 1964). A dorsal-fin with 9 branched rays also occurs in the only species of the Landonini , Landonia latidens and the Phenacobryconini, Phenacobrycon henni .

Hypotheses of affinities between the Glandulocaudinae and some members of the Tetragonopterinae were mentioned by Böhlke (1954) and Géry (1966). However, those hypotheses were weakly defended and mostly based on superficial similarities among the species of those groups. Further evidence of possible relationships between glandulocaudines and the genera listed above are related to the presence of insemination in Attonitus , Monotocheirodon , and some species of the genera Creagrutus , Knodus and at least one species currently assigned to Bryconamericus , a derived feature among Characidae .

4) Presence of four teeth in the inner series of the premaxilla. The presence of four teeth in the inner series of the premaxilla ( Fig. 10 ), versus the presence of five or more teeth, has long been used since Eigenmann (1917) to diagnose some genera among tetragonopterine characids. No evidence has been presented before, however, to identify any of these states as derived and thus informative for the diagnosis of monophyletic groups within the Characidae . We found, however, the presence of four teeth in the inner premaxillary tooth row in some tetragonopterine genera and basal glandulocaudines consistently associated with the derived presence of a reduced number of dorsal-fin rays (ii,8). The hypothesis of relationships supported by the preceding characters 1, 2 and 3 and presented in the cladogram of Fig. 2 , also parsimoniously supports the recognition of the presence of four teeth in the inner series of the premaxilla as derived, and as an additional synapomorphy for Clade A characids, versus the presence of five or more teeth in the inner series of the premaxilla in all other taxa bearing at least two tooth rows. Cyanocharax species have four teeth in the inner series of the premaxilla (four or five in four of the species of Cyanocharax ), which supports its inclusion among Clade A characids.

The presence of four teeth in the inner series of the premaxilla, along with the presence of a large infraorbital 3 (designated the “second suborbital") in contact with the preopercle below, has been used by Eigenmann (1917: 51) to diagnose a group of tetragonopterines in his key. Based on the same characters. Géry (1966) proposed the recognition of a subtribe named Hemibryconini as “... a group of tetragonopterine characids which seem (remotely) [sic] monophyletic and curiously well delimited, despite the apparently insignificant common characters of the forms, i. e. the presence of only four inner premaxillary teeth [italicized by Géry], frequently associated with a great development of the third suborbital [infraorbital 3]. and. quite often, the irregular implantation of the outer premaxillary row of teeth.” Géry's Hemibryconini consisted of Hemibrycon . Knodus . Bryconacidnus , Bryconamericus , Boehlkea , Microgenys , Ceratobranchia and Rhinobrycon ; along with the genera then classified in the Stevardiidi (= Glandulocaudina e), Glandulocauda (“at least the type species melanogenys "), Phenacobrycon , Argopleura , Planaltina , Acrobrycon , Pterobrycon , and Stevardia (= Corynopoma ); and the genera of a “ Tetragonopteridi ” subtribe named the Creagrutini that included Piabarchus , Creagrudite , Creagrutus . and Piabina . Further in the text, he mentioned some additional genera or species for the Hemibryconini, including: Nematobrycon , Coptobrycon , Rhinopetitia , Hyphessobrycon melanopleurus (= Glandulocauda melanogenys, Weitzman & Menezes , pers. comm.). Landonia and Gephyrocharax (the last two genera now referred to Glandulocaudinae, Weitzman & Menezes, 1998). It is noteworthy the similarities in genera composition of the group proposed by Géry and the hypothesized monophyletic Clade A proposed herein (Fig. 11).

Chernoff & Machado-Allison (1990: 266), while discussing the relationships of Ceratobranchia , pointed out that Géry's hypothesis of relationships among those genera were difficult to accept, since it was based on characters not unique for the group in question (the large infraorbital 3 is present in other tetragonopterine genera), or difficult to polarize (the number of inner premaxillary teeth). They additionally disagree with Géry's statement that the outer premaxillary teeth in Ceratobranchia , Knodus . Bryconaniericus or Hemibrycon are irregularly placed, and different from those non-Hemibryconini genera.

We do not challenge the Chernoff & Machado-Allison (1990) arguments regarding the variability in size ofthe infraorbital 3 and outer premaxillary tooth arrangement in tetragonopterine fishes for using these as characters defining a clade among characid genera. We hypothesize, however, the presence of four teeth in the inner premaxillary tooth row in some tetragonopterine genera and basal glandulocaudines as derived and a synapomorphy for Clade A characids, through the most parsimonious analysis of distribution of characters 1, 2, 3 and 4.

Comparisons of the number of teeth in the inner series of the premaxilla of the Clade A characids must be made with species in the basal glandulocaudine tribes and genera. The species of the genera Planaltina (N. A. Menezes, pers. commun.) and Diapoma ( Malabarba, 1983) of the basal glandulocaudine tribe Diapomini and Landonia latidens of the basal tribe Landonini have premaxillary teeth with similar arrangement and number (usually four teeth in the inner series) to those observed for most Bryconamericus or Knodus species. Among xenurobryconins, one of its basal genera, Argopleura ( Weitzman & Fink, 1985:42, Fig. 66) also has similar tooth arrangement, but other genera of this tribe and also the putatively basal genus Ptychocharax ( Weitzman et al., 1994:59, Fig. 12 ) has 5 teeth in the inner series. The genera lotabrycon, Scopaecharax, Tyttocharax and Xenurobrycon , all with derived and paedomorphic species, have completely different and specialized tooth number and arrangements ( Weitzman & Fink, 1985:40-41, Figs. 59-64). Therefore, the species of these genera are not actually pertinent to this particular problem.

Cyanocharax definition

The cladogram of Fig. 11 depicts the hypothesized relationships among the genera included in the putative monophyletic Clade A of Fig. 2 , that is characterized by the possession of ii,8 dorsal-fin rays and four teeth in the inner series of the premaxilla. Considering the presented hypothesis of monophyly of Clade A characids, Cyanocharax differs from Hemibrycon and Boehlkea by the presence of an incompletely toothed maxilla, with 2 to 8 teeth ( Figs. 23 , 28 , 32 , 38 , 44 , 49 , 51 ) whereas Boehlkea has an almost to completely toothed maxilla with 11 -21 teeth, and Hemibrycon 6-20 maxillary teeth ( Fig. 10 ).

Figure 11 also illustrates hypotheses of the monophyly and/or relationships previously proposed by Weitzman & Menezes (1998) for the Glandulocaudinae; Malabarba & Malabarba (1994) and Silva & Malabarba (1996) for Hypobrycon ; Vari & Harold (1998, 2001) for Piabina and Creagrutus ; Malabarba & Vari (2000) for Caiapobrycon , Piabina , and Creagrutus ; and Vari & Ortega (2000) for Attonitus . Cyanocharax species do not share any of the apomorphies described for those clades or genera and are therefore excluded from all of them.

Clade A encompasses most of the known inseminating species of the Characidae , including all species of Glandulocaudinae, all species of the genera Attonitus and Monotocheirodon and a few species currently placed in Creagrutus , Bryconamericus , and Knodus (J. R. Bums, pers. commun.). Cyanocharax species are not inseminating, thus lacking any further evidence to group them with the inseminating species listed above.

It is notable that the genera grouped in Clade A ( Figs. 2 and 11) include most of the tetragonopterine species with ventrally-located mouths ( Attonitus , Bryconacidnus , Caiapobrycon , Ceratobranchia , Creagrutus , Hypobrycon , Microgenys , Piabarchus , Piabina , Rhinobrycon , Rhinopetitia , and part of Bryconamericus and Knodus ) suggesting a possible close relationship among them or at least part of them. Three unlisted characid genera with ventrally-located mouths ( Monotocheirodon , Odontostoechus and Othonocheirodus) were previously included in Cheirodontinae ( Géry, 1977) and moved to incertae sedis in Characidae by Malabarba (1998). Species of these genera have a single series of teeth in the premaxilla and thus cannot be characterized by the presence of “four teeth in an inner tooth series”. Species of those genera, however, have ii,8 dorsal-fin rays suggesting their inclusion among Clade A characids. In Odontostoechus there is a single series, but Malabarba (1998) reported an undescribed species, closely related to Odontostoechus lethostigmus , in which young individuals have two series of teeth in the premaxilla with four teeth in the inner row, and “an arrangement similar to that of Bryconamericus ” corroborating their inclusion in Clade A. In Monotocheirodon there is a single series of four teeth, suggesting that the absence of an anterior series is a secondary loss, and that the remaining tooth series is homologous to the inner tooth series of the other Clade A genera, but such a hypothesis needs further support.

Hypothesis of relationships among genera with ventrally-located mouths have already been discussed by Vari & Harold (1998, 2001) for Piabina and Creagrutus and by Malabarba & Vari (2000) for Piabina , Creagrutus , and Caiapobrycon . Malabarba & Vari (2000) have further discussed the possible relationships among tetragonopterines with ventrally-located mouths that share a reduced anal-fin ray number, as found in Caiapobrycon , Attonitus , Ceratobranchia , Rhinobrycon , Bryconacidnus and Creagrutus . Cyanocharax species do not share any of the specialized jaw bones and teeth such as observed in Attonitus , Bryconacidnus , Caiapobrycon , Ceratobranchia , Creagrutus , Hypobrycon , Microgenys , Piabarchus , Piabina , Rhinobrycon , Rhinopetitia, Bryconameric us, Knodus , Monotocheirodon , Odontostoechus and Othonocheirodus or the reduced anal-fin ray number that would support the inclusion of the species described herein in any of those genera.

Although we cannot phylogenetically diagnose Cyanocharax based on exclusively derived features, the synapomorphies diagnosing the internal clades within that genus and which are discussed below further support the close relationships among included species. Characters that allow the recognition of the Cyanocharax species are listed under “Distinguishing characters.”

Relationships among Cyanocharax species (Note: characters used in a cladistic analysis of the species included in the discussion below are numbered and presented in the cladogram of Fig. 12 )

Possible derived features diagnosing Cyanocharax among Clade A characids are: (1) A maxilla with an incomplete dentition consisting of 2-8 teeth that occupy less than half the length of the maxilla or the anteriormost portion of this bone ( Figs. 23 , 28 , 32 , 38 , 44 , 49 , 51 ). Hemibrycon and Boehlkea (with 11 -21 maxillary teeth) are apparently the most basal genera in Clade A, lacking all specializations related to insemination, development of caudal and/or anal glands or the jaw and teeth modifications related to a ventral mouth, as observed in the remaining genera and species of Clade A. Compared to Cyanocharax , the genera Hemibrycon and Boehlkea basically differ in the longer toothed portion of the maxilla ( Fig. 10 ) that extends beyond its half-length. Although we can hypothesize the condition found in Cyanocharax as derived compared to the long toothed portion of the maxilla as found in the outgroup characid genus Brycon , such a character is also found in several representatives of Clade A characids, and cannot be assumed as a synapomorphy for Cyanocharax until the relationships among the members of Clade A are better hypothesized. However, the facts above do allow a diagnosis of Cyanocharax that distinguishes it from Hemibrycon and Boehlkea .

(2) The presence of an intense blue color in sexually mature specimens somewhat similar to that observed in Mimagoniates is also hypothesized to be derived ( Figs. 20A , 47 ). This color has been observed in C. itaimbe , C. dicropotamicus , C. lepiclastus , C. alegretensis , and C. macropinna . No mature specimens of C. tipiaia were available for evaluation of this charac - ter and such coloration is absent in C. alburnus . The latter species’ systematic position, formerly placed in Astyanax , has long been problematic due to the variable presence of four or five teeth in the inner series of the premax - illa. It has consequently been described in the literature as Bryconamericus alburnus and Astyanax hasemani , and lately referred as A. alburnus . Its relationships were briefly discussed by Malabarba (1983), who pointed out a possible relationship to Bryconamericus or to the glandulocaudine species Diapoma speculiferum and Glandulocauda terofali (now = Diapoma terofali ), instead of Astyanax .

We hypothesize that “ Astyanax alburnus " belongs to Clade A characids on the basis of its possession of ii, 8 dorsal-fin rays and the presence of a variable number of four or five teeth in the inner series of the premaxilla. It furthermore lacks the derived features diagnosing most of the other genera of Clade A and has a short section of the maxilla toothed, a feature that excludes it from Hemibrycon and Boehlkea , but is congruent with a hypothesis of a relationship to Cyanocharax . This species, now referred to as C. alburnus , a new combination, would seem to be basal among the species of Cyanocharax . It lacks the intense blue color of sexually mature specimens, a putative derived feature grouping the remaining Cyanocharax spe - cies, and, in addition, the derived features discussed below that define internal Cyanocharax subclades. Cyanocharax alburnus also has the largest geographical distribution among Cyanocharax species, occurring in all river systems where the remaining species of the genus are found, although not necessarily syntopic.

(3) Among the remaining Cyanocharax species, C. itaimbe and C. dicropotamicus are hypothesized to constitute a clade defined by the presence of a black pigmented adipose fin, that is absent in all other Cyanocharax species and outgroup taxa.

(4) Cyanocharax lepiclastus , C. alegretensis , C. macropinna , and C. tipiaia share the presence of an interrupted lateral line or an alternating series of perforated and non-perforated lateral line scales. Although the presence of a complete or reduced lateral line is a character highly variable among Clade A species and thus difficult to polarize for the species of Cyanocharax , the hypothesis supported by character 2 that places all Cyanocharax species as a sister group to C.albumus ,allowstheuseofthe last species as a functional outgroup to polarizethischaracteramongthe remaining Cyanocharax species. Such a proceduresupportstherecognition of the reduced lateral line as derived and asa synapomorphyforC. lepiclastus , C. alegretensis , C. macropinna ,and C.tipiaia .

Further supporting this hypothesis isa series ofmodificationsofthe anal fin of males and females. The anal-fin shape is variableinCyanocharax, with the adult males ofthe putative basalspecies, C. albumusplus C. itaimbe ( Figs. 13 , 17 ) and C. dicropotamicus ( Figs.6 ,24),having(5) an anal fin with a concave distal border(5 state0), whereasC. tipiaiaandC. lepiclastus ( Fig. 33 ) have a nearly straightdistalanal-finmargin(5state1) and C. alegretensis ( Figs. 7 , 15, 39-40)and C. macropinna ( Figs.8 , 16 , 45 ) a deeply convex margin (5 state 2). By parsimony,a nearlystraight analfin distal border is derived and a synapomorphy supportinga cladeformedby Cyanocharax lepiclastus , C. alegretensis , C. macropinna , andC. tipiaia , and, in a transformation series, the deeply convexanalfindistalbordera synapomorphy for the subclade C. macropinna and C. alegretensis .

All Cyanocharax species have a(6)scalesheathcoveringthebasalportions of the anal-fin rays. The extension ofthisscalesheathoverthebasalfin rays varies according to the species. In the putativebasalspeciesC. albumus as well as in C. itaimbe ( Fig. 13 ), C. dicropotamicus ( Fig.14) andC. tipiaia we found the scales forming a sheath in each speciestovary from 7 to13 scales (6 state 0) covering the bases of the anteriorbranchedanal-finrays. Cyanocharax lepiclastus and C. alegretensis ( Fig. 15 )havea greaterseries with 12 to 20 scale s (6 state 1), while C. macropinna ( Fig.16 )havea greater series of 20 to 30 scales (6 state 2) covering thebaseofmostanal-finrays. Again, comparing to the most basal species, thelargerseriesofscalescovering anal-fin rays bases is considered apomorphicanda synapomorphywithin a clade consisting of C. lepiclastus , C. alegretensisandC. macropinna (6state 1). The highest number observed in C. macropinna isalsoapomorphicina transformation series and an autapomorphy ofthisspecies(6state2).

Finally, C. macropinna is hypothesized to havethemostderivedanal fin among all Cyanocharax species. The numberofbranchedanal-finrays (7) in C. macropinna varies from 29-35 (x̄= 31.0,n = 120),whileallother Cyanocharax species have smaller countsofbranchedanal-finrays,asfollows: C. alburnus , 20-23; C. itaimbe , x̄= 23.9,n = 88; C. dicropotamicus , x ̄ = 24.6, n = 130; C. tipiaia , x̄ = 22.3, n = 15; C. lepiclastus, x ̄= 26.2, n = 133; C. alegretensis , x̄ = 26.6, n = 185. The highercountobservedinC. macropinna is considered derived and an autapomorphyforthisspecies.

Figures

Figure 2. Cladogram summarizing present knowledge on the relationships of characids, gasteropelecids and roestines. Lettered boxes A through E represent hypotheses supported by characters presented by A - M. Malabarba (1998); B - Malabarba (1998); C - Lucena (1998); D - Reis (1989) and E - Vari (1977). Lettered boxes F and G arc characters putatively derived that represent possible synapomorphies for the corresponding groups (F - the large fourth infraorbital in all Aphyocharax species; G - a peculiar ontogenetic development of teeth and jaws that includes changes in tooth shape, number of tooth scries in the premaxilla and extension of toothed portion of the premaxilla in the Rhoadsiinae). Numbered boxes represent characters discussed herein (sec under “Cyanocharax relationships").

Figure 3. Ventral view of the left pelvic fin of a male of Cyanocharax dicropotamicus showing the distribution of the pelvic-fin hooks (MCP 19510, 41.9 mm SL).

Figure 4. Ventral view of the left pelvic fin of a male of Cyanocharax alegretensis showing the distribution of the pelvic-fin hooks (MCP 11232, 39.7 mm SL).

Figure 5. Ventral view of the left pelvic fin of a male of Cyanocharax macropinna showing the distribution of the pelvic-fin hooks (MCP 16382, 46.4 mm SL).

Figure 6. Left view of the anal fin of a male of Cyanocharax dicropotamicus showing the distribution of the anal-fin hooks (MCP 19510,41.9 mm SL).

Figure 7. Left view of the anal fin of a male of Cyanocharax alegretensis showing the distribution of the anal-fin hooks (MCP 11232,39.3 mm SL).

Figure 8. Left view of the anal fin of a male of Cyanocharax macropinna showing the distribution of the anal-fin hooks (MCP 16382,43.5 mm SL). K

Figure 9. Left view of the dorsal tin of male Cyanocharax alegretensis (MCP 11232, 39.7 mm SL) showing the two unbranched plus eight branched dorsal-fin rays shared with other Cyanocharax species and putatively related genera

Figure 10. Jaws and teeth of Hemibrycon dariensis (MCP 27072, 47.0 mm SL). A-B - Maxilla and detailed view of teeth; C-D - Premaxilla and detailed view of teeth; E-F - Dentary and detailed view of teeth.

Figure 12. Hypothesized relationships among Cyanocharax species based on the characters described under “Relationships among Cyanocharax species”.

Figure 13. Left view of the anal fin of male Cyanocharax itaimbe, showing the sheath of scales covering the base of the anterior anal-fin rays (USNM 323415, 45.3 mm SL).

Figure 14. Left view of the anal fin of male Cyanocharax dicropotamicus, showing the sheath of scales covering the base of the anterior anal-fin rays (USNM 337612,40.8 mm SL).

Figure 15. Left view of the anal fin of male Cyanocharax alegretensis, showing the sheath of scales covering the base ofthe anterior anal-fin rays (USNM 357245, 39.3 mm SL).

Figure 16. Left view of the anal fin of male Cyanocharax macropinna, showing the sheath of scales covering the base of the anterior anal-fin rays (USNM 319750, 43.5 mmSL).

Figure 17. Cyanocharax itaimbe new species, holotype, male, 49.6 mm SL, arroio Facão at Mãe dos Homens, tributary of rio Canoas - rio Mampituba, Praia Grande, Santa Catarina, Brazil (MCP 25972).

Figure 20A-B. Cyanocharax itaimbe, Balneário Maquiné, rio Maquiné, Maquiné, Rio Grande do Sul, Brazil (UFRGS 4998). A - Photograph taken just after fixation in formalin, showing the blue color pattern. B - Photograph of a second specimen collected with specimen above after losing the blue color pattern.

Figure 23A-F. Jaws and teeth of Cyanocharax itaimbe (USNM 323409, 48.5 mm SL). A-B - Maxilla and detailed view of teeth; C-D - Premaxilla and detailed view of teeth; E-F - Dentary and detailed view of teeth.

Figure 28. Teeth of Cyanocharax dicropotamicus (MCP 19510, 40.0 mm SL). Jaws seem decalcified and deformed. A - Maxilla, B - Premaxilla and C - Dentary teeth.

Figure 32. Jaws and teeth of Cyanocharax tipiaia (MCP 22712, 38.0 mm SL). A-B - Maxilla and detailed view of teeth; C-D - Premaxilla and detailed view of teeth; E-F - Dentary and detailed view of teeth.

Figure 33. Cyanocharax lepiclastus, male, 45.2 mm SL, arroio Lageado Uniao, Linha dos Lima, Palmitinho, Rio Grande do Sui, Brazil (MCP 14557).

Figure 38. Jaws and teeth of Cyanocharax lepiclastus (MCP 16382, 43.2 mm SL). A-B - Maxilla and detailed view of teeth; C-D - Premaxilla and detailed view of teeth; E-F - Dentary and detailed view of teeth.

Figure 44. Jaws and teeth of Cyanocharax alegretensis (MCP 1 1232, 44.5 mm SL). A-B - Maxilla and detailed view of teeth; C-D - Premaxilla and detailed view of teeth; E-F - Dentary and detailed view of teeth.

Figure 45. Cyanocharax macropinna, holotype, male, 44.2 mm SL, arroio Pai- Passo, on the road Santana do Livramcnto-Alegrete, Quaraí, Rio Grande do Sui, Brazil (MCP 25973).

Figure 47. Cyanocharax macropinna, male, rio Sarandi III, on the road Santana do Livramento to Quaraí, Rio Grande do Sul, Brazil (MCP 16382).

Figure 49. Jaws and teeth of Cyanocharax macropinna (MCP 16382, 43.1 mm SL). A-D - Maxilla and detailed view of teeth; B-E - Premaxilla and detailed view of teeth; C-F - Dentary and detailed view of teeth.

Figure 51. Jaws and teeth of Cyanocharax alburnus (MCP 9100, 55.0 mm SL). A-D - Maxilla and detailed view of teeth; B-E ‒ Premaxilla and detailed view of teeth; C-F- Dentary and detailed view ofteeth.