Taxonomic revision of Galeocharax ( Characiformes : Characidae : Characinae )

The taxonomy of Galeocharax, a genus of freshwater fishes distributed in most South American cis-Andean river systems, except the rio Negro, rio São Francisco and rio Xingu basins and the eastern drainages of Brazil, is herein revised. A total of 1229 specimens were examined from which 680 had meristic and morphometric data taken. Osteological and morphological features were also examined through radiographs, scanning electron microscopy and in cleared and stained specimens. Three of the four species previously considered as valid are herein recognized: Galeocharax humeralis from rio Paraguay and lower rio Paraná basins; Galeocharax goeldii, from rio Madeira basin, with records of geographical distribution expanded to the río Madre de Dios, río Beni, rio Mamoré and rio Guaporé basins and Galeocharax gulo, which is widespread throughout rio Amazonas, río Orinoco, rio Tocantins, and upper rio Paraná basins. Galeocharax knerii is herein considered a junior synonym of Galeocharax gulo. A key to species of Galeocharax is presented.


Introduction
is a genus of freshwater fishes that reach up to 25 cm in length.They are popularly known as "saicanga", "peixe-cigarra", "peixe-cadela", and "dentudo" in the Brazilian territory (Graça, Pavanelli, 2007;Langeani, Rêgo, 2014) and as "dentón" and "dientón alargado" in Peru and Venezuela (Taphorn, 1992;Ortega et al., 2012).Species of the genus are found in most of the main river systems of the cis-Andean South America with the exception of the rio São Francisco, rio Negro, rio Xingu basins, and the eastern drainages of Brazil.
They are carnivorous, feeding on small fishes and invertebrates (Ribeiro Neto et al., 1998;Gandini et al., 2012).They reproduce throughout the year and there is no record of parental care or migratory behavior (Magalhães et al., 2004).The presence of the parasitic isopoda (Braga cigarra, Cymothoidae), popularly known as "cigarrinha", in the oral cavity of specimens of Galeocharax in the Paraná basin is fairly common (Brandão et al., 2013), justifying the popular name peixe cigarra.
The genus Galeocharax is included in the family Characidae, which is the most species rich family in the order Characiformes with approximately 165 genera and more than 1300 species (Nelson et al., 2016), from which more than 600 occur in the Brazilian territory (Buckup et al., 2007).Within the Characidae, Galeocharax has been traditionally included in the subfamily Characinae (Lucena, Menezes, 2003;Mirande, 2010;Mattox, Toledo-Piza, 2012), and more closely related to Acestrocephalus Eigenmann and Cynopotamus Valenciennes, all three genera forming a monophyletic taxon that currently includes 24 species, diagnosed, among other features, by the presence of spinoid scales (Menezes, 1976).

e160040[2]
The latest taxonomic revision of the genus was carried out by Menezes (1976) who recognized three valid species: Galeocharax humeralis (Valenciennes, 1834) from the rio Paraguai and the lower rio Paraná basins; Galeocharax gulo (Cope, 1870) from the rio Amazonas and the rio Tocantins basins and Galeocharax knerii (Steindachner, 1879) from the upper rio Paraná basin.At that time, Menezes considered Galeocharax goeldii (Fowler, 1913) from the rio Madeira basin, as a synonym of G. gulo, but more recently, after the examination of additional specimens, the species was recognized as valid (Menezes, 2007a).
The number of specimens of Galeocharax available in ichthyological collections increased considerably in the past decades, showing that the genus is more widely distributed than previously recorded.Therefore, the aim of the present study was to undertake a taxonomic study of the genus including the highest number of specimens as possible in order to determine how many and which species can be recognized as valid, to reevaluate their diagnostic features and to estimate their geographic distribution.

Material and Methods
A total of 1229 specimens deposited in several ichthyological collections and covering the entire area of distribution of Galeocharax were examined, from which 680 specimens had meristic and morphometric data taken.Morphometric data were taken point to point with a digital caliper with 0.1 mm precision, on the left side of the specimen, whenever possible.Nineteen different measurements were made on each specimen and they are presented in the tables and text as percentages of standard length (SL), with the exception of measurements of subunits of the head, which were expressed as percentages of head length (HL).
Data of specimens from which sex could be determined based on the examination of gonads were initially analyzed separately to check for possible sexual dimorphism, but are presented together in the tables and text since no differences were detected.Variation of the number of teeth on the posterior row of the dentary is only partially associated with growth, and was represented by scatter plots.The same association was observed for number of maxillary teeth.
Measurements follow Menezes (1969), with the exception of: head length, which was measured from the tip of the snout to the posterior bony margin of the opercle at the ventral tip of the opercle concavity.The following measurements were included: upper jaw length: distance between the tip of the snout and posterior margin of the maxilla.Dorsal-fin length: distance between the base of the first ray and the distal tip of the longest unbranched ray.Dorsal-fin base length: distance between the bases of the anterior-and posteriormost dorsal-fin rays.Anal-fin length: distance between the base of the first visible unbranched ray and the distal tip of the longest unbranched ray.Analfin base length: distance between the bases of the anteriorand posteriormost anal-fin rays.Prepectoral distance: distance between the tip of the snout and the base of the first pectoral-fin ray.Caudal-peduncle length: distance between the base of the posteriormost anal-fin ray and the posterior margin of the hypural plate.
Meristic data were taken from the left side of specimens whenever possible.Scales, fin-rays and teeth were counted with the aid of a stereomicroscope.The number of unbranched rays is expressed with roman numerals and the number of branched rays with arabic numerals.For a better visualization of the teeth, a solution of methylene blue in 70% ethanol was applied to the specimens.The stain is temporary and fades away a few minutes after returning the specimen to 70% ethanol.Counts of vertebrae and supraneurals were made on radiographs taken from a limited number of specimens and examined in the RadiAnt DICOM viewer (32-bit) software.
Counts follow Menezes (1969) with the exception of number of scales below the lateral line that was counted as the number of horizontal rows of non-perforated scales between the lateral line and the base of the first pelvicfin ray.This way of counting showed to be more precise than that made at the vertical through the first anal fin ray, since the scales in this latter region tended to be slightly disorganized.The following counts were added: number of scales around the caudal peduncle; number of scales in the anal-fin sheath: number of horizontal rows on the sheath of scales that covers the base of the anal-fin rays; number of pelvic-fin rays; number of teeth on the outer row of the dentary; number of teeth on the inner row of the dentary; number of teeth on the outer row of the premaxilla; number of teeth on the inner row of the premaxilla; total number of vertebrae; number of abdominal vertebrae: including the four anteriormost vertebrae of the Webberian apparatus until the posteriormost vertebra without a haemal spine and number of caudal vertebrae (from the first vertebra with a haemal spine to the compound ural centrum, which was counted as one).Counts of unbranched anal-fin rays, principal and procurrent caudal-fin rays were made only on cleared and stained specimens.In the descriptions, meristic values are followed by their respective frequency in parentheses, except those presented on tables.
Meristic and morphometric data were summarized using descriptive statistics with the help of MYSTAT 12 software.
Morphological details of spinoid scales, sensu Roberts (1993), and bony hooks on the anal-and pelvic-fin rays were examined through scanning electronic microscope 3 e160040 [3] images made in a Zeiss DSM940 electronic microscope.Since no diagnostic differences were observed, details of those structures are shown for a single species.
Specimens were cleared and stained following the procedure of Taylor, Van Dyke (1985).
Specimens from the same locality were initially compared and if differences on the examined characteristics were not detected they were grouped.This approach was carried out successively for localities near to each other, then to localities on the same river basins until all the specimens were analyzed and preliminary groups were defined.
In the discussion of geographic distribution, the Amazon basin was defined excluding the rio Tocantins (see Goulding et al., 2003a, for a discussion).Maps of geographic distribution were made using Google Earth and Quantum GIS Wroclaw softwares.

Galeocharax Fowler, 1910
The name Galeocharax was proposed in order to resolve a nomenclatural confusion originated by publications of Fowler (1901Fowler ( , 1904Fowler ( , 1910) ) and Eigenmann (1903Eigenmann ( , 1907Eigenmann ( , 1910) ) that was later detailed by Géry, Vu-Tân-Tuê (1963a).It should be noted that Fowler's proposition was not followed by a discussion about the species included neither presented a diagnosis of Galeocharax.Most of the publications citing Galeocharax, subsequent to the proposition of the genus name consisted of species lists that did not include any discussion or characterization of the genus or the species included.Menezes (1976) recognized a natural group comprising Cynopotamus, Acestrocephalus and Galeocharax, the Cynopotaminae on the basis of morphological characters and differentiated Galeocharax from Acestrocephalus and Cynopotamus based on the three characters mentioned above in the Diagnosis.

Key to the species of
Although there is overlap in the number of scales above and below the lateral line between G. goeldii and G. gulo, the former species shows a tendency towards higher values of scales above (17-20, median 18 vs.14-19, median 16, in G. gulo, Tab. 2) and below (15-17, median 16 vs.11-16, median 13 in G. gulo, Tab. 3) the lateral line, respectively.Dorsal-fin origin situated anterior to half of SL, base of last dorsal-fin ray posterior to vertical through analfin origin.Proximal-medial radial of first dorsal-fin pterygiophore inserted posterior to neural spine of ninth vertebra.Dorsal-fin rays ii,9.Profile of distal margin of dorsal fin straight in most specimens.Anal-fin origin situated at approximately half of SL.Proximal-medial radial of first anal-fin pterygiophore situated posterior to haemal spine of 17 th vertebra (1).Anal-fin rays iv (1), 40 (8), 41 (12), 42 (16), 43 (23), or 44(8).Anterior lobe of anal fin poorly developed, profile of distal margin of anal fin concave at anterior two thirds; straight to slightly convex 9 e160040 [9] at posterior third.Pectoral-fin rays i, 13 (1), 14 (28), 15 (31), or 16 (10); longest ray reaching vertical through pelvic-fin origin when fin is depressed.Profile of posterior margin of pectoral fin almost straight, with four or five proximal rays shorter than others.Specimens of at least 33.7 mm SL still retain larval rayless pectoral fin.Pelvicfin origin situated anterior to vertical through dorsal-fin origin; pelvic-fin rays i,7; longest ray reaching vertical through anterior border of urogenital opening when fin is depressed, but failing to reach anal-fin origin.Profile of posterior margin of pelvic fin slightly convex, fan-like aspect when extended.Caudal fin forked, principal rays i,9,8,i (1) dorsal procurrent rays 16 (1), ventral procurrent rays 13(1); lower lobe slightly longer and deeper than upper lobe.Adipose fin present, its base anterior to vertical through base of last anal-fin ray.
Spinoid scales (Fig. 8) distributed along entire body, except on region of axilla of paired fins and on region posterior to last ray of dorsal and anal fins where cycloid scales are present.Axillary scale present in pelvic fin.Anal-fin scale sheath with three horizontal scale rows over base of anteriormost anal-fin rays, covering one-fifth of length of anterior fin rays; number of horizontal scale rows gradually decreasing posteriorly to one scale row on posteriormost rays.Base of caudal-fin rays covered by scales.
Mouth terminal, obliquely oriented relative to horizontal body axis.Upper jaw slightly more anteriorly projected than lower jaw.Posterior margin of maxilla extending beyond vertical through posterior margin of orbit.Infraorbitals 1 to 6 present, supraorbital absent, posterior margin of infraorbital 3 not reaching preopercle, posterior margin of infraorbital 4 reaching preopercle in some specimens.Teeth conical.Premaxilla with two teeth rows, inner row with two teeth oriented to interior of mouth, outer row with 8 (4), 9 ( 28  Coloration.Overall body coloration yellowish, dorsal region darker than ventral region due to presence of chromatophores distributed on posterior field of dorsal scales.Dorsal median dark band from supraoccipital spine to dorsal-fin origin, not distinguishable posterior to dorsal fin base.
Dorsal region of head darker until approximately horizontal line through dorsal margin of orbit.Few specimens with scattered chromatophores on infraorbitals, mainly on 4, 5 and 6.Skin bordering mouth more pigmented on premaxilla, maxilla and dentary, mainly on base of canine teeth of lower jaw.Lateral surface of maxilla with sparse chromatophores anteriorly to approximately half of bone length.
Oval shaped blotch on humeral region, largest diameter along vertical axis.Few specimens with crescent shaped humeral blotch with tips of crescent pointed anteriorly (Fig. 6b).
Silvery midlateral longitudinal band on body, dorsal to lateral line, ventral margin of band reaching lateral line in most specimens.Silvery midlateral longitudinal band extending from posterior margin of supracleithrum to posterior margin of hypural plate, narrower at anterior-and posteriormost regions in most specimens, broader at region of vertical through dorsal-fin base.Silvery midlateral longitudinal band widening from caudal peduncle to posterior margin of hypural plate in some specimens, and with ventral margin extending beyond lateral line on region of caudal peduncle.
Dark midlateral longitudinal band dorsal to lateral line extending from humeral region, to approximately half length of median caudal-fin rays, narrower at anterior-and posteriormost regions, broader at region of vertical through dorsal-fin.Some specimens with dark midlateral longitudinal band evident from posterior margin of supracleithrum to median caudal-fin rays.Dark midlateral longitudinal band forming a diamond shaped blotch over caudal peduncle and base of median caudal-fin rays.Dark midlateral longitudinal band not as wide as silvery midlateral longitudinal band except on caudal peduncle region.Dorsal margin of two longitudinal bands usually overlap.
Specimens smaller than 48.0 mm SL with lines of chromatophores ventral to lateral line, following myosepta on lateral surface of body, region dorsal of anal-fin base and anal-fin rays.
Pectoral and pelvic fins hyaline.Dorsal fin with lines of chromatophores along rays and scattered chromatophores on interradial membrane.Adipose fin pale, weakly pigmented except for line of chromatophores on anterodorsal margin.
Chromatophores forming arched vertical line through base of caudal fin rays; chromatophores more deeply concentrated underneath scales, arched vertical line interrupted in some specimens, with only dorsal and ventral limits evident (Fig. 6b).
Sexual dimorphism.Males of Galeocharax goeldii have bony hooks on pelvic-and anal-fin rays.Bony hooks on anal-fin ray located on lateral surface of posterior branch of branched rays; bony hooks present on last unbranched ray in few specimens.One or two, rarely three hooks per ray segment, hooks dorsoposteriorly oriented (Figs.10a-c).Bony hooks present on 10 to 23 anteriormost rays.Bony hooks on pelvic fin located on ventral surface of lepidotrichia slightly posteriorly displaced (one specimen with hooks on dorsal surface of fin ray).One or two, rarely three hooks per segment pointing towards base of ray (Fig. 11).Hooks present on four or five lateralmost branched rays.

Remarks. Galeocharax goeldii was originally described as
Charax goeldii by Fowler (1913) based on a single specimen, 56.0 mm SL, collected in a tributary of the rio Madeira, near Porto Velho, RO, Brazil. Later, Menezes (1976) considered the species as a synonym of Galeocharax gulo, based on data presented by Géry (1972), but after examining the holotype of Charax goeldii and additional material from the rio Madeira basin, Menezes (2007a) recognized the species as valid in Galeocharax.
As part of the present study, specimens examined from the río Madre de Dios, río Beni, rio Mamoré, and rio Guaporé drainages showed that G. goeldii is more widely distributed within the rio Madeira basin.
Although the holotype of Galeocharax goeldii was not examined during this study, meristic data of the specimen presented by Fowler (1913); Géry (1972) and Menezes (2007a) are summarized on Tab. 5.There are some small differences in the numbers provided by the various authors that could be due to differences in the way of couting, or the condition of the specimen could render the counting difficult.Menezes (2007a) provided the most recent and complete information on the specimen.With the exception of the number of lateral-line sacles (87 vs. 88-97 in the present study) and the number of teeth on outer row of premaxilla (12 vs. 8-10 in the present study), data provided by that author are included within the variation observed herein for G. goeldii.No variation in meristic and morphometric data associated with any geographic pattern was observed in specimens throughout the distribution range of the species, and the slight discrepancies in those counts of the holotype provided by Menezes (2007a) seem most probably related to the relative poor condition of the specimen associated to its small size.Diagnosis.Galeocharax gulo differs from its congeners in having fewer perforated lateral line scales (77-87 vs. 90-105 in G. humeralis and 88-97 in G. goeldii, Tab. 1).Galeocharax gulo further differs from G. humeralis in the absence of small conical teeth between the third and fourth dentary canines (Fig. 7b) (vs.1-3 small conical teeth, Fig. 7c).Although there is overlap in the number of scales above and below the lateral line between G. gulo and its congeners, G. gulo shows a tendency towards lower values of number of scales above (14-19, median 16 vs.17-20, median 18 in G. humeralis and G. goeldii, Tab. 2) and below (11-16, median 13 vs. 14-16, median 15 in G. humeralis and 15-17, median 16 in G. goeldii, Tab. 3) the lateral line, respectively.
Description.Morphometric data of Galeocharax gulo presented on Tab. 4. Largest specimen examined 245.2 mm SL.Body elongated, laterally compressed and moderately deep, largest body depth at vertical through dorsal-fin origin.Dorsal profile of head slightly concave from tip of snout to posterior tip of supraoccipital spine.Dorsal profile of body slightly convex from posterior tip of supraoccipital spine to base of first dorsal-fin ray, posteroventraly slanted to slightly convex along dorsalfin base.Dorsal profile from base of last dorsal-fin ray to caudal peduncle straight, slightly concave along caudal peduncle.Ventral profile of head and body convex from tip of lower jaw to anal-fin origin, posterodorsaly slanted along anal-fin base, straight to slightly concave along caudal peduncle.Profile of posterior bony margin of opercle usually concave, rarely straight or convex.
Spinoid scales (Fig. 8) distributed along entire body, except on region of axilla of paired fins and region posterior to last ray of dorsal and anal fins where cycloid scales are present.Axillary scale present in pelvic fin.Anal-fin scale sheath with three, generaly two horizontal scale rows over base of anteriormost anal-fin rays, covering onefifth of length of anterior fin rays; number of horizontal scale rows gradually decreasing posteriorly to one or no scale row over posteriormost rays.Base of caudal-fin rays covered by scales.

e160040[16]
Pseudotympanum present, bordered anteroventraly by obliquus superioris muscle, posteriorly by obliquus superioris muscle and by rib associated with sixth vertebra, posterodorsaly by lateralis superficialis and by lateral line nerve.Rib associated with fifth vertebra visible anteriorly, but not bordering the muscle hiatus anteriorly (Fig. 9).
Coloration.Overall body coloration yellowish, dorsal region darker than ventral region due to presence of chromatophores distributed on posterior field of dorsal scales.Dorsal median dark band from supraoccipital spine to dorsal-fin origin on most specimens, not distinguishable posterior to dorsal-fin base.
Dorsal region of head darker until approximately horizontal line through dorsal margin of orbit.Few specimens with scattered chromatophores on infraorbitals, mainly on 4, 5 and 6.Skin bordering mouth more pigmented on premaxilla, maxilla, and dentary, mainly on base of canine teeth of lower jaw.Lateral surface of maxilla with sparse chromatophores anteriorly to, approximately half of bone length.
Oval shaped blotch on humeral region, largest diameter along vertical axis.A few specimens with crescent shaped humeral blotch with tips of crescent pointed anteriorly (Fig. 13c).
Silvery midlateral longitudinal band on body, dorsal to lateral line, ventral margin of band reaching lateral line in most specimens.Silvery midlateral longitudinal band extending from posterior margin of supracleithrum to posterior margin of hypural plate, narrower at anterior-and posteriormost regions in most specimens, broader at region of vertical through dorsal-fin base.Silvery midlateral longitudinal band widening from caudal peduncle to posterior margin of hypural plate in some specimens, and with ventral margin extending beyond lateral line on region of caudal peduncle.
Dark midlateral longitudinal band dorsal to lateral line extending from posterior margin of supracleithrum to approximately half length of median caudal-fin rays, narrower at anterior-and posteriormost regions, broader at region of vertical through dorsal-fin base.Dark midlateral longitudinal band forming a diamond shaped blotch over caudal peduncle and base of median caudal-fin rays.Dark midlateral longitudinal band not as wide as silvery midlateral longitudinal band except on caudal peduncle region.Dorsal margin of two longitudinal bands usually overlap.
Specimens smaller than 104.0 mm SL with lines of chromatophores ventral to lateral line, following myosepta on lateral surface of body, region dorsal of anal-fin base and anal-fin rays.
Pectoral and pelvic fins hyaline.Dorsal fin with lines of chromatophores along rays and scattered chromatophores on interradial membrane.Adipose fin pale.
Chromatophores forming arched vertical line through base of caudal-fin rays; chromatophores more deeply concentrated underneath scales; arched vertical line interrupted in some specimens, with only dorsal and ventral limits evident (Fig. 13c).
Sexual dimorphism.Males of Galeocharax gulo have bony hooks on pelvic-and anal-fin rays.Bony hooks on anal-fin rays located on lateral surface of posterior branch of branched rays; bony hooks present on last unbranched rays in a few specimens.One or two, rarely three hooks per ray segment, hooks dorsoposteriorly oriented (Figs.10a-c).Bony hooks present on 8 to 28 anteriormost rays.Bony hooks on pelvic fin located on ventral surface of lepidotrichia slightly posteriorly displaced.One or two, rarely three hooks per segment pointing towards base of ray (Fig. 11).Hooks present on one to five lateralmost rays.
Eight relatively large sized females (e.g., MZUSP 10479, 179,0 mm SL) also have bony hooks on anal fin, however less developed, fewer in number and distributed on fewer rays compared to males (Figs.10d-f).

Geographical distribution.
Galeocharax gulo is widely distributed in the Amazon basin including most of its tributaries except the rio Negro and rio Xingu.It also occurs in the río Orinoco, rio Oyapok, rio Araguaia, rio Tocantins, and the upper rio Paraná basins (Fig. 12).

Geographic variation.
Although there is overlap in the range of variation of all meristic and morphometric characters (Tabs.6-7) throughout the geographic distribution of the species, specimens from the upper rio Paraná and the rio Tocantins basins are more similar to each other, than they are to specimens from the rio Amazonas (excluding specimens from the rio Tapajós basin) and the río Orinoco basins, regarding mainly the number of scales above the lateral line (Tab.8), number of scales below the lateral line (Tab.9) and number of maxillary teeth (Tab.10).
Specimens from the upper rio Paraná basin show lower values of interorbital distance compared to specimens from the rio Amazonas basin (Fig. 14).However, when specimens from other drainages are included in the comparison (rio Tocantins and rio Orinoco basis) such difference becomes less evident (Fig. 15).Similarly, specimens from the rio Tocantins basin show higher values of orbital diameter compared to specimens from the rio Amazonas basin (Fig. 16).However, when specimens from the upper rio Paraná and río Orinoco basins are included in the comparison, their values are intermediate between those from specimens from the rio Tocantins and rio Amazonas basins (Fig. 17).
Specimens from the rio Tocantins basin show a tendency toward lower values in the number of branched anal-fin rays (Tab.11) and total number of vertebrae (Tab.12) when compared to specimens from the remainder of the geographic distribution of the species.About 60% of the examined specimens from the upper rio Paraná basin have 17 abdominal vertebrae (n = 33), while all specimens from the rest of the area of distribution of the species have 16 abdominal vertebrae (n = 91).e160040 [17] Tab. 6. Morphometric data of specimens of Galeocharax gulo from rio Amazonas, upper rio Paraná and rio Tocantins basins.SD = standard deviation.Standard length in mm.Very few specimens from the rio Tapajós are present in collections, only nine specimens of three lots (MNRJ 23654, MZUSP 62839, and MZUSP 62841) were examined in the present study.Meristic and morphometric characters of those specimens are summarized in Tabs.7 and 13.All those specimens show the diagnostic characters of G. gulo, however there are some subtle differences between the specimens from the rio Tapajós and those from the rest of the distribution range of G. gulo concerning the range of variation and frequency distribution of the number of scales between the lateral line and the pelvic-fin origin 10 (1), 11 (4), or 12 (2) (vs.11-16), number of maxillary teeth 47 (1), 48 (1), 52 (2), 53 (1), 54 (1), 56 (2), or 57 (1) (vs.35-55), and number of branched anal-fin rays 34 (2), 35 (2), 36 (4), or 37 (1) (vs. 33-46).The frequencies of distribution of theses features for the specimens from the rio Tapajós are more similar to the frequencies of distribution observed for specimens from the upper rio Paraná and the rio Tocantins basins (Tabs.9-11, respectively).Specimens from the rio Tapajós have 40 (7) total vertebrae which is also more similar to the counts of vertebrae of specimens from the rio Tocantins (Tab.12).The range of variation of the number of teeth on the external row of the premaxilla 10 (2), 11 (3), or 13 (4) of the specimens from the rio Tapajós also differs from that of the rest of the distribution range of G. gulo, however overlapping also occurs.Specimens from the rio Tapajós have 5 (7) supraneurals, which is relatively uncommon when compared with the rest of the specimens of G. gulo 4 (80) or 5 (7).In view of the lack of any features that unambiguously characterize, the specimens from rio Tapajós as a distinct species, they are tentatively identified as G. gulo.Tab.12. Frequency distribution of the total number of vertebrae of specimens of  Planquette et al. (1996) reported the occurrence of specimens of Galeocharax in the rio Oyapok, French Guiana, and identified them as Galeocharax aff.gulo, based on an apparently lower number of lateral line scales than the reported by them for G. gulo (76-84 vs. 81-86).
In the present study two specimens from the rio Oyapok drainage were examined and have 78 and 84 lateral-line scales.The large number of specimens examined in the present study showed that the range of variation in the number of lateral-line scales is larger than previously recognized for the species.In addition, all other meristic and morphometric characters of the specimens from the Oyapok river fall into the variation ranges herein observed for G. gulo.
No geographic variation in body proportions other than those discussed above was observed (Tabs.6-7).Although it was possible to identify a few patterns of geographic variation in G. gulo, it was not possible to unequivocally diagnose different species based on the characteristics examined in the present study.

Remarks.
Although the holotype of G. gulo (ANSP 8053, 73.9 mm SL, Fig. 18) was not examined during this study, data of that specimen is available in the literature (Tab.5).In addition, examination of specimens originating from the region near the type locality, showed that meristic and morphometric data fall within the variation observed in the present study for G. gulo.Only data presented by Géry (1972) regarding number of teeth on the inner row of the premaxilla and number of maxillary teeth do not agree with the variation ranges observed during this study (3 vs. 2 and 31-36 vs. 35-55, respectively).Discrepancies in the number of premaxillary teeth are probably due to different ways of counting them (i.e.considering the canine teeth as part of the inner row or not).Regarding the maxillary teeth, it is common to find specimens with several fallen teeth, a fact that might have influenced the count reported by Géry (1972).
Tab. 11.Frequency distribution of the number of branched anal-fin rays of specimens of Galeocharax gulo from various drainages.Scale rows between dorsal-fin origin and lateral line 15-16 7 Scale rows between lateral line and pelvic-fin origin 10-12 7 Scale rows between lateral line and anal-fin origin 14-16 9 Scale rows around caudal peduncle 20-24 6 Branched anal-fin rays 34-37 9 Teeth on maxilla 47-57 9 Teeth on dentary inner row 7-9 6 Teeth on pre-maxilla outer row 10-13 9 Abdominal vertebrae 16 7 Caudal vertebrae 24 7 Total number of vertebrae 40 7 Galeocharax gulo was diagnosed from G. knerii on the basis of only two features: a deeper body and by the presence of fewer teeth on the posterior row of the dentary (Menezes, 1976, figs. 51 and 55, respectively).However, the examination of a larger number of specimens from the rio Amazonas, upper rio Paraná and rio Tocantins resulted in increase of the overlap of those characteristics considered as diagnostic for G. gulo and G. knerii (figs. 19;Tab. 14), and all the remaining meristic and morphometric characters (tabs. 6, 8-12, 14-16).Menezes (1976: 41) reported the existence of ontogenetic variation in the number of teeth on the posterior row of the dentary, with larger specimens having a higher number of teeth.Larger specimens examined in the present study have more teeth than some of the small specimens, but there is individual variation in this feature (Fig. 20), so that the ontogenetic variation is not clearly reflected by the data.Mattox, Toledo-Piza (2012) reported some osteological differences between specimens of Galeocharax from the rio Amazonas basin and the upper rio Paraná basin, that could help to diagnose two distinct forms.According to their observations specimens of Galeocharax from the rio Amazonas basin have the anterior margin of the palatine divided into two distinct articulation regions, a medial region associated with the neurocranium and a lateral region associated with the medial process of the maxilla.Those two regions are separated by a circular opening into which the tip of the largest dentary canine fits.Alternatively, specimens from the upper rio Paraná basin have the anterior margin of the palatine rounded, with a single articulation surface, a generalized condition in the Characiformes (Mattox, Toledo-Piza, 2012: 851).The examination of the material used by the authors in their study as well as the examination of additional cleared and stained material showed that all Galeocharax species have the same condition, which is intermediate between the two conditions described by Mattox, Toledo-Piza (2012).The anterior margin of the palatine in the specimens examined have two distinct articulation areas, one medial associated with the neurocranium and a lateral, associated with the medial process of the maxilla (Fig. 21).However, those two articular surfaces are separated by a subtle concavity on the anterior margin of the autopalatine that do not form a circular opening (the latter condition occurs in species of Cynopotamus).
Mattox, Toledo-Piza (2012: 853) also described that specimens of Galeocharax from the upper rio Paraná basin have the entire lateral margin of the ectopterygoid straight and a ligament connects the ventral surface of the maxilla to the lateral margin of the ectopterygoid.Alternatively, in specimens from the rio Amazonas basin, the lateral margin of the ectopterygoid has a bony lateral projection dorsoventraly flattened, where the cited ligament is attached (Fig. 21b).Examination of the material used by those authors confirmed the presence of the cited projection in G. gulo.However, an additional cleared and stained specimen of Galeocharax from the upper Paraná basin, of similar size to the specimen of Galeocharax from the rio Amazonas basin examined by the authors, also presents such lateral projection (Fig. 21f).This may indicate an ontogenetic variation of the character in the species, the gradual ossification of the base of the ligament connected to the lateral margin of the ectopterygoid.A third difference reported by Mattox, Toledo-Piza (2012: 854) regards the presence in specimens of Galeocharax from the upper rio Paraná basin of an antero dorsal projection of the metapterygoid medially oriented and associated with the endopterygoid (vs.projection absent in specimens from rio Amazonas basin).However, the examination of additional material lead to a reinterpretation of the character and to the conclusion that the conditions reported by the cited authors are not distinct between specimens of Galeocharax from the rio Amazonas basin and upper rio Paraná basin (Fig. 3).Therefore, it could not be used as a diagnostic character between the herein considered synonym species.
Mattox, Toledo-Piza (2012) also reported that in specimens of Galeocharax from the upper rio Paraná basin, the posterior margin of the adductor operculi and levator operculi muscles are at the same vertical, while in Galeocharax from the rio Amazonas basin the posterior margin of the levator operculi is posterior to the posterior margin of the adductor operculi.This myological character was not evaluated in the present study.
Despite differences in the range of variation of characters such as perforated lateral line scales (Tab.15), number of branched anal-fin rays (Tab.11) and total number of vertebrae (Tab.12) between specimens from the Tocantins river basin and specimens from the remaining distribution range of Galeocharax, the large overlap in the values complicates the recognition of specimens from the Tocantins as a distinct species.
None of the characteristics studied in the present study allow us to unequivocally recognize forms from the upper rio Paraná, rio Amazonas, and rio Tocantins as different species.Therefore G. knerii is herein considered a junior synonym of G. gulo, following the Principle of Priority (ICZN, 1999: art. 23).Information about the syntypes of G. knerii (Fig. 22) available in the literature are summarized on Tab. 5. Taphorn (1992: 185) reported the occurrence of Galeocharax in the río Apure, río Orinoco basin, Venezuela, but based on the range of variation of the characters of the specimens he examined, he was not able to identify the species as either G. gulo or G. knerii.The author hence suggested that the specimens from the río Apure could probably represent an undescribed species.In the present study, 42 specimens from the río Orinoco basin were examined, including 14 specimens from the río Apure drainage.The number of perforated lateral-line scales of the specimens from the río Apure drainage varies from 80-84 (n = 11), a range included in the range of variation of 79-84 reported by Taphorn (1992).The remaining meristic and morphometric data fall within the range of variation of specimens from the Amazon basin (15)(16).The specimens from the río Orinoco basin are herein identified as Galeocharax gulo.
Spinoid scales (Fig. 8) distributed along entire body, except on region of axilla of paired fins and on posterior to last ray of dorsal and anal fins where cycloid scales are present.Axillary scale present in pelvic fin.Anal fin scale sheath with three, in most specimens four horizontal scale rows over base of anteriormost anal-fin rays, covering onefith of length of anterior fin rays; number of horizontal scale rows gradually decreasing posteriorly to one scale row over posteriormost rays.Caudal fin rays base covered by scales.
Pseudotympanum present, bordered anteroventrally by obliquus superioris muscle, posteriorly by obliquus superioris muscle and by rib associated with sixth vertebra, posterodorsally by lateralis superficialis muscle and lateralline nerve.Rib associated with fifth vertebra visible anteriorly, but not bordering muscle hiatus anteriorly (Fig. 9).
Coloration.Overall body coloration yellowish, dorsal region darker than ventral region due to presence of chromatophores distributed on posterior field of dorsal scales, mainly near scale margins.Dorsal median dark band from supraoccipital spine to dorsal-fin origin, not distinguishable posterior to dorsal-fin base.
Dorsal region of head darker until approximately horizontal line through dorsal margin of orbit.A few specimens with scattered chromatophores on infraorbitals, mainly on 4, 5 and 6.Skin bordering mouth more pigmented on premaxilla, maxilla and dentary, mainly on base of canine teeth of lower jaw.Lateral surface of maxilla with sparse chromatophores anteriorly to approximately half of bone length.
Oval shaped blotch on humeral region, largest diameter along vertical axis.A few specimens with crescent shaped humeral blotch with tips of crescent pointed anteriorly (Fig. 23b).
Silvery midlateral longitudinal band on body, dorsal to lateral line, ventral margin of band reaching lateral line in most specimens.Longitudinal band extending from posterior margin of supracleithrum to posterior margin of hypural plate, narrower at anterior-and posteriormost regions in most specimens, broader at region of vertical through dorsal-fin base.Silvery midlateral longitudinal band widening from caudal peduncle to margin of hypural plate in some specimens, and with ventral margin of silvery midlateral longitudinal band extending beyond lateral line on region of caudal peduncle.
Dark midlateral longitudinal band dorsal to lateral line extending from posterior margin of supracleithrum to approximately half length of middle caudal-fin rays, narrower at anterior-and posteriormost regions, broader at region of vertical through dorsal-fin base.Dark midlateral longitudinal band forming a diamond shaped blotch at caudal peduncle and base of median caudal-fin rays.Dark midlateral longitudinal band not as wide as silvery midlateral longitudinal band except on caudal peduncle region.Dorsal margin of two longitudinal bands usually overlap.
e160040 [27] Specimens smaller than 126.0 mm SL with lines of chromatophores ventral to lateral line, following myosepta on lateral surface of body, region dorsal of anal-fin base and anal-fin rays.
Pectoral and pelvic fins hyaline.Dorsal fin with lines of chromatophores along rays and scattered chromatophores on interradial membrane.Adipose fin pale on most specimens, adipose fin with similar pattern of chromatophore distribution of dorsal region of body in a few specimens.Arched vertical line of chromatophores on base of caudal-fin base absent.
Sexual dimorphism.Males of Galeocharax humeralis have bony hooks on pelvic-and anal-fin rays.Bony hooks on anal-fin rays located on lateral surface of posterior branch of branched rays; bony hooks present on last unbranched ray in a few specimens.One or two, rarely three hooks per ray segment, dorsoposteriorly oriented (Figs.10a-c).Bony hooks present on 12 to 27 anteriormost rays.Bony hooks on pelvic fin located on ventral surface of lepidotrichia, slightly posteriorly displaced on posterior surface of posterior branch of branched rays.One or two, rarely three hooks per segment pointing towards base of ray (Fig. 11).Hooks present on two to six lateralmost rays.
Two relatively large sized females (MZUSP 90473 210.5 mm SL and MCP 28624 157.4 mm SL) have bony hooks on anal fin, however less developed, smaller in number and distributed on fewer rays compared to males (Figs.10d-f).
Geographical distribution.Galeocharax humeralis occurs in the rio Paraguai, rio Uruguai and lower rio Paraná basins (Fig. 12).The only record of G. humeralis in the upper rio Paraná is in the Itaipu reservoir (NUP 255,.Those specimens were collected between 1986 and 1989.With the formation of the Itaipu Reservoir, in October of 1982 some fish species that occurred in the 150 km stretch between Sete Quedas (Guaíra) and the Itaipu Power Plant (Foz do Iguaçu), became established above the former basins divider, that currently is the Itaipu Power Plant (Graça, Pavanelli, 2007).More detailed studies are necessary in order to understand if the presence of the species in this region is associated with this event.Remarks.Hydrocyon humeralis (= G. humeralis) was described based on an illustration from material deposited in the collection of the MNHN by D'Orbigny (Valenciennes 1834), with no further description.Valenciennes (1847) refered to the drawing of the previous publication and presented a brief description of the species.A more complete description was presented by Valenciennes (1849), who included the species in the genus Cynopotamus.The type locality was indicated as Buenos Aires, Argentina.Valenciennes (1849: 321) pointed out that the type specimen of Cynopotamus humeralis was obtained in a market at Buenos Aires, thus it would be more adequate to infer that the specimen originated in some locality within the río de La Plata basin.
The first mention that we found in the literature to the holotype of G. humeralis was made by Bertin (1948) in a catalogue of the types deposited in the Muséum National d'Histoire Naturelle in Paris.
The holotype of G. humeralis (MNHN 9848,117.6 mm SL,Fig. 24) was not examined during this study, however data of the specimen is available in the literature (Tab.5).
There are a few discrepancies among the data of the holotype presented by Valenciennes (1849) and Géry, Vu-Tân-Tuê (1963a).The number of lateral line scales presented by Valenciennes (1849) and the number of scales below the lateral line presented by Géry, Vu-Tân-Tuê (1963a) also fall outside the range of variation presented herein.These differences could be explained by different ways of counting scales and by the fact that the scales below the lateral line are somewhat disorganized, leading to some imprecision of the counts.

Discussion
The examination of a large number of specimens, including many from areas not sampled at the time of the last taxonomic study of Galeocharax (Menezes, 1976), made possible to corroborate previous hypotheses present in the literature, such as the recognition of G. humeralis and G. goeldii as two distinct species.In addition, the diagnostic characters and the information on the geographical distribution of these two species were refined.
The present study also provided evidence for alternative interpretations as to the pattern of geographic variation of the specimens from the rio Amazonas, rio Tocantins, rio Orinoco, and the upper rio Paraná basins, with implications in the number of species that may be recognized in these drainages.It was observed that recognizing the specimens
Fig. 8. Scanning electron microscope images of the spinoid scales of Galeocharax goeldii, UFRO-I 9668, 160.8 mm SL. a. dorsal view of scale, posterior margin of scale oriented towards upper left corner, scale bar 500 µm; b. and c. detail of posterior field of scale, scale bar 200µm, in b. dorsal view, posterior margin of scale to top, and c. dorsolateral view, posterior margin of scale to right.

Fig. 10 .
Fig. 10.Scanning electron microscope images of the second branched anal-fin ray of Galeocharax gulo, male, MZUSP 40901, 110.6 mm SL a.-c.and female, MZUSP 10479, 179.7 mm SL d.-f.. Proximal tip to upper left corner in a.-c., to top in d. and to upper right corner in e.-f.a. posterior view, scale bar 2 mm; b.Detail of middle portion, lateral view, left side, scale bar 200 µm; c.Detail of middle portion, posterior view, scale bar 200 µm; d. lateral view, right side, scale bar 1 mm; e. Detail of middle portion, lateral view, right side, scale bar 200 µm, and f.Detail of middle portion, lateral view, right side, scale bar 100 µm.

Fig. 11 .
Fig. 11.Scanning electron microscope images of the first branched pelvic-fin ray of Galeocharax gulo, male, MZUSP 40901, 110.6 mm SL.Proximal tip of ray to upper left corner.a. Ventral view, scale bar 1 mm; b.Detail of middle portion, ventral view, scale bar 200 µm.

Fig. 12 .
Fig. 12. Map of northern and central South America showing the geographical distribution of species of Galeocharax.G. goeldii (blue circles), the black triangle indicates its type locality, G. gulo (red circles), the black square indicates its type locality, and G. humeralis (green circles).

Fig. 20 .
Fig. 20.Scatter plot of number of teeth on the posterior row of the dentary vs. standard length of specimens of Galeocharax gulo from rio Amazonas (circles) and upper rio Paraná (squares) basins.

Table 8 .
Frequency distribution of number of scale rows between the dorsal-fin origin and lateral line of specimens of Galeocharax gulo from various drainages.
) (vs. small conical teeth between the third and fourth dentary canines absent).It further differs from G. gulo in the higher number of perforated lateral-line scales (90-105 vs. 77-87, Tab. 1).Although there is overlap in the number of scales above and below the lateral line between G. humeralis and G. gulo, G.