Molecular phylogeny of Planaltina Böhlke (Characidae: Stevardiinae) and comments on the definition and geographic distribution of the genus, with description of a new species

A molecular phylogeny of Planaltina, including the three previously described species and an undescribed species, is presented. The monophyly of the genus, included in Diapomini, is strongly supported. Its sister group, the remaining Diapomini, includes only species without modified caudal-fin squamation in the present analysis (species of Diapoma with caudal organs were not sampled). Creagrutus is sister to Planaltina plus remaining Diapomini instead of Planaltina being sister to Creagrutus plus Diapomini, as a previous analysis had suggested. Species of Planaltina form two clades: P. britskii plus the new species, with low support (< 50%); and P. myersi plus P. glandipedis, with higher support. Planaltina is rediagnosed from all Characidae based on the morphology of the caudal organ, the absence of a humeral spot and the presence of a complete lateral line. Comments on the caudal-fin squamation of Diapoma and Lepidocharax burnsi, on the type-series of L. burnsi and on the geographic distribution of Planaltina and Lepidocharax species are provided. Finally, a formal description of the aforementioned new species and a novel identification key to Planaltina are presented.


Introduction
The past ten years brought several advancements to the understanding of the characid phylogeny. Much work was directed to the Stevardiinae (as defined by Thomaz et al. [1]), with highly diverging results. Data sources included the traditional external morphology and osteology [2,3], but also histology [4,5] and DNA [1, 6,7]. Available phenotypic studies [5,8] seem to have overestimated resemblance between caudal organs in the former members of the Glandulocaudinae (sensu Weitzman & Menezes [8]), masking their now-evident polyphyly [1,7] shown by late molecular studies, which also deeply rearranged the classification of the group, sinking a few "specialized" genera and giving new combinations for species previously placed in large, "catch-all" genera [1]. The apparently homoplastic evolution of caudal organs, specialized teeth etc. in stevardiine fishes sheds doubt on the monophyletic origin of some poorly defined genera. Böhlke [9] described Planaltina and its type-species P. myersi Böhlke based on a single male specimen, and diagnosed it from all other Characidae by an extensive combination of characters, mostly meristic. Weitzman & Menezes [8] observed that in Planaltina males and females have externally indistinguishable caudal organs, a character shared only with Acrobrycon Eigenmann & Pearson and Diapoma Cope within Stevardiinae, interpreting it as a synapomorphy for those genera and placing them in a redefined tribe Diapomini (posteriorly found to be polyphyletic [1,7]). Subsequently, Menezes et al. [10] redefined Planaltina simply by diagnosing it from Acrobrycon and Diapoma, based mainly in the presence of 1-2 scales in the dorsal border of the caudal pouch opening (vs. 4-7 and 3-6, respectively; posteriorly, Menezes & Weitzman [11] and Arcila et al. [12] recorded 4-8 scales in both genera) and the nearly spherical sperm nuclei. Nevertheless, it is possible to observe in Menezes et al. ([10], Figs. 28, 33) that P. glandipedis Menezes, Weitzman & Burns occasionally exhibits more than two scales in the dorsal border of the caudal pouch opening and P. britskii Menezes, Weitzman & Burns is sexually dimorphic for the shape of the caudal-fin scales, which means some changes in the definition of Planaltina are needed.
The known geographic distribution of Planaltina presently includes only those localities recorded by Menezes et al. [10], Graça & Pavanelli [13] and Araújo & Tejerina-Garro [14]: P. britskii has been reported from the Grande, São José dos Dourados and Tietê river basins, as well as from other, smaller tributaries of the rio Paraná in the State of São Paulo and from the upper rio Paraná floodplain; P. glandipedis, from the portion of the rio Tietê basin draining the cuestas of the State of São Paulo; and P. myersi, from the Corumbá and Ouvidor river basins, both emptying in the rio Paranaíba. A recently discovered species (Planaltina sp. of Frota et al. [15]) from the Ivaí and Piquiri river basins, south to the known geographic range of Planaltina, agrees only partially with the diagnosis of the genus by Menezes et al. [10]; as in P. glandipedis, some individuals of the new species present as much as four scales in the dorsal border of the caudal pouch.
This paper aims to investigate the relationships of Planaltina based on a molecular phylogenetic analysis, including the new species, in order to test its monophyly, hence permitting the proposition of a revised diagnosis of the genus (with comments on the caudal-fin squamation in Diapoma and Lepidocharax). It also expands the known geographic range of all previously described Planaltina species. Additionally, we provide a formal description of the aforementioned new species, a novel identification key to the species of Planaltina and comments on the type-series of L. burnsi.

Molecular data collection
Total DNA was extracted from ethanol-preserved muscle samples using the DNeasy Tissue Extraction Kit (Qiagen) following the manufacturer's instructions. Partial sequences of the mitochondrial genes 16SrRNA and Cytochrome b (CytB) and the nuclear genes recombination activating gene 1 (Rag1), recombination activating gene 2 (Rag2) and myosin heavy chain 6 cardiac muscle alpha (Myh6) were amplified by polymerase chain reaction (PCR) with the primers listed in S1 File. Amplifications were performed in a total volume of 25 μl consisting of 2.5 μl 10X buffer (10 mM Tris-HCL, 15 mM MgCl 2 ), 0.5 μl MgCl 2 (50 mM), 0.5 μl each primer (5 μM); 0.4 μl of dNTPs (200 nM of each), 0.2 μl Taq Platinum polymerase (Invitrogen; 5 U/μl), 1 μl template DNA (10-50 ng) and 19.4 μl ddH 2 O. The thermocycler profile used for the fragments 16SrRNA and CytB consisted of 35 cycles of 30 s at 95˚C, 45-120 s at 50-55˚C, and 90 s at 72˚C. Nested PCR was used to amplify the nuclear genes Rag1, Rag2 and Myh6. Amplification conditions for these genes in both rounds of PCR consisted of 15 cycles of 30 s at 95˚C, 45 s at 56˚C (according to primer), and 30 s at 72˚C followed by 15 cycles of 30 s at 95˚C, 45 s at 54˚C (according to primer), and 90 s at 72˚C. PCR products were purified using ExoSap-IT1 (USB Corporation), sequenced using the Big DyeTM Terminator v 3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems), purified again by ethanol precipitation and loaded into an automatic sequencer 3130 Genetic Analyzer (Applied Biosystems) at Instituto de Biociências, Universidade Estadual Paulista, Botucatu, São Paulo, Brazil. Contigs were assembled and edited in Geneious Pro 8.1.8 [16]. In cases of unclear nucleotide identity, IUPAC ambiguity codes were applied. All obtained sequences were deposited in GenBank.

Alignment and phylogenetic analysis
Sequences of each gene were aligned in Geneious Pro 8.1.8 [16] using the MUSCLE algorithm under default parameters, and the alignments were inspected by eye for any obvious misalignments that were subsequently corrected. Genetic distances among sequences were calculated in Mega 6 [17]. We estimated the index of substitution saturation (Iss) in DAMBE 5.2.31 [18] as described in Xia et al. [19,20].
A set of six partitioning schemes ranging from 1 to 13 partitions was tested following the procedures outlined by Li et al. [21] using the AICc (Akaike Information Criterion, corrected for finite sample sizes). The best-fit model of nucleotide substitution was searched in Mega 6 [17] under default parameters using the AICc (see Posada & Buckley [22] for justification).
RAxML [23] running in the web servers RAxML-HPC2 on TG [24; 25] was used for all maximum likelihood analyses with a mixed partition model. Random starting trees were run for each independent ML tree search, and all other parameters were set to default values. All ML analyses were conducted following the 13 partitions scheme as suggested by the AICc. Topological robustness was investigated using 1,000 non-parametric bootstrap replicates.

Morphological data
Counts and measurements are the same as in Menezes et al. [10]. Principal caudal-fin rays are numbered from the longest dorsal unbranched ray (1) to the longest ventral unbranched ray (19). Count values were taken mostly from the left side of the specimens, but information about right-side counts is given when pertinent. Meristic data marked by an asterisk are the values for holotype. Numbers in parentheses after a count value represent the frequency of occurrence in the sample. Lots with measured specimens are marked with an asterisk. Diagnostic characters were examined in all specimens. The numbers of caudal-fin procurrent rays and posterior dentary teeth were annotated only for cleared and stained specimens and for entire alcohol-preserved specimens in which a precise count was possible. Sex was determined by the presence or absence of hooks on anal-fin rays. Specimens smaller than the minimumsized hook-bearing specimens of each species were considered as young; above that size, specimens not bearing hooks were treated as females. Lots excluded from the type-series are poorly preserved and have important structures damaged. Complete geographic references are provided for all specimens of Planaltina and Lepidocharax, but not for remaining comparative specimens. Elevations of collection sites were estimated in the Google1 Earth1 version 7.1.5.1557 software.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org: pub:573B0705-D1F7-4C26-8E56-BD2F2D18E7B6. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Relationships of Planaltina
Planaltina was recovered within Diapomini, as sister to all other taxa included in the tribe (Fig  1). Of these, all, except some non-analyzed species of Diapoma, lack modified caudal-fin squamation. This contrasts with the results by Oliveira et al. [7], in which Planaltina is sister to Creagrutus plus Diapomini.
The monophyly of Planaltina is strongly supported, although its internal relationships were resolved with lower confidence. The new species was found as sister to P. britskii; that is the clade with the lowest support (< 50%). The low support (50-69%) of the clade including the two specimens of the new species (which were collected together) is probably due to the incompleteness of some sequences from those individuals and to the conservativeness of the molecular markers employed in the study, which present a relatively low divergence among Planaltina species. The other two species, P. glandipedis and P. myersi, clustered together, forming a well-supported clade (90-100%).

Definition of Planaltina
Planaltina is diagnosed from all other Characidae by the following combination of external morphological characters: caudal pouch a single cavity completely restricted to the ventral caudal-fin lobe (vs. caudal pouch absent in most Characidae; occupying the ventral portion of the dorsal lobe in Glandulocaudini), its opening in a posteroventral position, never coriaceous or tubular (vs. opening posteriorly directed, pouch enclosed by skin ventrally in Pseudocorynopoma; tubular in Chrysobrycon, Hysteronotus); dorsal portion of caudal pouch formed by elongate, enlarged scales, which are attached to the fin by their long dorsal borders, and are never folded around themselves or ornamented with processes (vs. pouch scale deeper than long, with concave posterior margin in Corynopoma and Gephyrocharax; pouch scales not much elongate in Phenacobrycon; pouch scale ornamented with processes in Hysteronotus, Pseudocorynopoma and some Xenurobryconini); caudal pouch not including pored scales, never formed by a single enlarged scale (vs. including pored scales in Argopleura and Landonia; formed by single enlarged scale in all Xenurobryconini; see Fig 2, painted scales); humeral spot absent (vs. present in Acrobrycon and in species of Diapoma with modified caudal-fin squamation); lateral line complete (vs. incomplete in species of Diapoma with modified caudal-fin squamation);.
The following characters are also useful to distinguish Planaltina from other Stevardiinae: origin of dorsal fin slightly anterior to origin of anal fin (vs. origin of dorsal fin below base of last dorsal-fin ray in most stevardiines; anterior to origin of anal fin in several other genera, such as Hysteronotus and Pterobrycon); ventral region narrow but not keeled (vs. keeled in, e.g., Pseudocorynopoma); contact organs (hooks) never present posteriorly to the first ten branched anal-fin rays (vs. present in the first ten and in the last ten rays in Phenacobrycon); absence of modified caudal-fin rays (vs. modified rays present in, e.g., Glandulocaudini and Xenurobryconini); pelvic-fin rays usually i,5,i (rarely i,6; vs. commonly i,6 or i,7 in, e.g., Bryconamericus and Knodus). Male P. britskii can be distinguished from all other Characidae by the dorsal border of the caudal pouch opening formed by two elongate, enlarged scales; and by the presence, in the lower caudal-fin lobe, of an adnate scale (partially covered by the border of the pouch), whose posteroventral margin is bent laterad. Species of Planaltina, except P. britskii, are distinguished from all other Characidae, except Acrobrycon and Diapoma, by the equal development of the caudal pouch in both sexes (at least regarding external morphology). All species of Planaltina can be distinguished from Acrobrycon and species of Diapoma bearing modified caudal-fin scales by the absence of humeral spot; from Acrobrycon, by having few teeth in the maxilla (vs. teeth present along most of anterior margin of maxilla); and from Diapoma bearing modified caudal-fin scales, additionally by having complete lateral line (vs. incomplete).  [10]. Pink-coloured scales are those forming the dorsal rim of the caudal-pouch opening; yellow-coloured scales are those forming the anterior border of the caudal pouch opening; the blue-coloured one is the adnate scale located in the inner side of the caudal pouch of male P. britskii. Shaded areas indicate the opening of the caudal pouch. Notice, in (c), that there are two scales forming the dorsal rim of the caudal-pouch opening, which is a rare condition in female P. britskii (only one scale usually present). The shape of these scales is, however, typical (that is, not particularly elongate as in male P. britskii).
The amount of scales forming the dorsal border of the caudal pouch opening in Planaltina is 1-4, slightly overlapping the range of Acrobrycon and Diapoma, 4-8 [11,12]. However, when more than one scale is present, they are imbricate and form a continuous wall that covers the pouch cavity. In contrast, in specimens of D. speculiferum Cope (type-species) and D. terofali Géry examined herein the largest, anteriormost "pouch scale" forms a cavity by itself and the other scales are partly included in this "pouch" (Fig 3). Each of the posterior scales has its posteroventral margin somewhat bent laterad, forming a smaller cavity underneath it, which could be considered as a smaller "pouch". In Planaltina, only the adnate scale present on the inner side of the pouch of male P. britskii has its posteroventral margin bent laterad. However, it is located in a different position (Fig 2B), and is absent in females ( Fig 2C). No sexual dimorphism was observed in the caudal-fin squamation of D. speculiferum and D. terofali.
Despite the caudal-fin squamation, the genus that most closely resembles Planaltina is Lepidocharax, especially L. burnsi Ferreira, Menezes & Quagio-Grassioto, whose body and mouth shape, dentition, meristics, morphometry and coloration completely overlap those of Planaltina species. The only external difference between those taxa is the presence of exclusively non-modified scales on the caudal fin in L. burnsi, hence the common misidentifications of this species in collections. Although Ferreira et al. [5] described the caudal-fin scales in both species of Lepidocharax as adnate, the scales in L. burnsi are not completely attached to the caudal-fin rays or to the thicker skin that covers the muscles inserted on their bases. Instead, in all specimens of L. burnsi examined herein (including the holotype and 127 paratypes) the caudal-fin scales are attached to the skin only by their anterior margins, being otherwise free from it. Thus, L. burnsi differs from all Planaltina species by lacking, in the ventral caudal-fin lobe, scales attached to the skin by their dorsal margins; however, truly adnate scales are not present (in comparison, they are found, for instance, in some Knodus moenkhausii (Eigenmann & Kennedy) specimens (NUP 6522), located more distally on caudal fin, attaching directly to the portion of the rays that is not covered by thick skin).

Geographic distribution of Planaltina
Species of Planaltina are found in all major sub-basins of the upper rio Paraná biogeographic region, namely the Paranaíba, Grande, Tietê, Paranapanema, Ivaí and Piquiri river basins, as well as in the rio Paraná proper and some minor tributaries in both right and left banks ( Fig  4). Additionally, P. myersi is found in the headwaters of the rio São Francisco basin close to the headwaters of the rio Corumbá.
In contrast to previous distributional data on Planaltina species [10], our data clearly show three cases of discontinuous geographic ranges: P. glandipedis, P. myersi and the new species. Of these, P. myersi occurs in the upper Paraná and São Francisco river basins. Isolated populations of P. myersi are very close geographically and morphologically indistinguishable. In that region, the headwaters of the Corumbá (upper rio Paraná basin) and Preto (rio São Francisco basin) river basins lay only a few hundred meters from each other, and the possibility of contact between them in the present should not be discarded. Planaltina glandipedis and the new species also occur each in two separate basins. Populations of P. glandipedis from the Tietê and Paranapanema river basins are situated near to one another, occurring in waterbodies draining the cuestas of the State of São Paulo. In contrast to Planaltina glandipedis, the new species is known from rather distant localities in the Ivaí and Piquiri river basins. Although  Fig 1) have densely sampled both drainages, no Planaltina specimen was captured in their lower stretches, suggesting some sort of unknown ecological barrier. Planaltina britskii has a much wider geographic range, being notoriously allopatric to the new species and P. myersi. In fact, the only case of co-occurrence between two Planaltina species is that of P. britskii and P. glandipedis in the rio Tietê basin.
Carl Ternetz collected the holotype of P. myersi in the "Sarandi brook, Planaltina, Goyaz, Brazil" on September 21, 1923, but failed to determine whether this stream was a tributary to the upper Paraná or upper Tocantins river basin ( [9], p. 267). Menezes et al. ([10], p. 564) ruled out the latter based on the fact that all specimens available to them originated from the upper rio Paraná basin, but provided no further information on the type-locality itself. Based on maps of the region [29,30,31], both earlier and posterior to the construction of the Brazilian Federal District (Brasília), we were able to delimit the type-locality of Planaltina myersi as "córrego Sarandi (tributary to the ribeirão Mestre-d'Armas, tributary to the rio São Bartolomeu, tributary to the rio Corumbá, tributary to the rio Paranaíba), upper rio Paraná basin, Brasília, Distrito Federal, approximately between 15˚35'S 47˚41'W and 15˚35'S 47˚44'W". This locality is situated in Planaltina, an administrative region of Brasília, not in the northern portion of the former Planaltina municipality, which remains part of the State of Goiás and is almost completely drained by the rio Maranhão and its tributaries (rio Tocantins basin).
Concerning the presence of Planaltina myersi in the rio São Francisco basin, it had already been reported by Géry ([32], p. 358, key). Nevertheless, Géry's [32] book lacks a list of examined material, thus it is possible that he has confused Planaltina myersi with Lepidocharax White signs represent the new species of Planaltina described herein; green signs, P. britskii; red signs, P. glandipedis; yellow signs, P. myersi; black signs, L. burnsi; purple signs, L. diamantina. Stars represent type-localities; circles represent localities with specimens analyzed herein; and the lozenge-shaped mark represents the locality of one lot of P. myersi that was not analyzed herein, but whose identification was confirmed. Notice that a few previously listed localities [5,10] are not shown herein.
https://doi.org/10.1371/journal.pone.0196291.g004 burnsi, which does not occur in sympatry with any Planaltina species, but is much more widely distributed in the rio São Francisco basin (Fig 4). Since no description of the caudal-fin squamation is given, it is better to disregard Géry's [32] observation. At least two other papers have suggested the presence of Planaltina in the rio São Francisco basin. Alves & Pompeu ( [33], p. 592), studying the rio das Velhas fish fauna, reported a putative new species to the genus; no formal description or diagnostic characters were given. Santos et al. [34], who presented a checklist of the rio Pandeiros ichthyofauna, recorded Planaltina sp. We were unable to access the lots listed by Alves & Pompeu [33] (MZUSP 73709 and MZUSP 73785). The specimens assigned by Santos et al. [34] to Planaltina sp. (UFRGS 10158) can be assigned to Lepidocharax cf. diamantina by the following characters: base of first dorsal-fin ray slightly anterior to base of first anal-fin ray, a few apparently adnate scales on proximal portion of caudal fin, 8 scalerows between lateral line and base of first dorsal-fin ray. They also presented 22-24 branched anal-fin rays, 4-5 teeth in the inner premaxillary series, 4-5 maxillary teeth, 17-19 circumpeduncular scale rows. Unfortunately, all five specimens are young (about 20 mm SL), which makes it difficult to give them a precise identification. If they in fact belong to L. diamantina, then the distribution of that species is wider than that reported by Ferreira et al. [5] (Fig 4).   Diagnosis. Planaltina kaingang is distinguished from all other congeners by the presence of melanophores embedded in all caudal-fin rays and interradial membranes (usually a few in each ray segment in specimens larger than about 25.0 mm SL) vs. melanophores located only on the interradial membranes, especially those between rays 8 to 15, but never embedded in the rays, regardless of SL (Fig 6). Planaltina kaingang is further distinguished from P. britskii by the absence of a modified, adnate scale located on the inner side of the caudal pouch of males, vs. presence (Fig 2A and 2B); the presence of 2-4 very elongate scales along the dorsal rim of the caudal-pouch opening in both sexes, vs. males always with 2 very elongate scales and females usually with only 1 (occasionally 2) enlarged, but not elongate scale (Fig 2A-2C); 15-17 circumpeduncular scale series, vs. 12-15. From P. glandipedis, by the presence of adipose fin, vs. absence; and the presence of tetra-and pentacuspid teeth in the inner series of the premaxilla and in the anterior portion of dentary (Fig 7), vs. all oral teeth with up to three cusps. From P. myersi, by the presence of 2-4 very elongate scales along the dorsal rim of the caudal-pouch opening in both sexes, vs. only one very enlarged scale in both sexes (Fig 2A and  2D); and by the presence of 2-3 distinct scales forming the anterior portion of caudal pouch opening, vs. anterior border of the pouch opening formed by the fusion of several scales into one big scale (Fig 2A and 2D). Description. Morphometric data in Table 1. Dorsal profile slightly convex from tip of premaxilla to insertion of first dorsal-fin ray; straight from this point to adipose fin; slightly concave on caudal peduncle. Ventral profile distinctly convex from lower jaw to anal-fin origin; slightly concave on anterior half of anal-fin; slightly convex on posterior half of anal-fin; slightly concave on caudal peduncle. Insertion of first dorsal-fin ray slightly anterior to vertical through first anal-fin ray, posterior to middle of standard length; posterior end of adipose-fin base at about vertical through insertion of last anal-fin ray. Tip of longest pectoral-fin rays reaching or slightly surpassing origin of pelvic fin; tip of longest pelvic-fin rays almost reaching or slightly surpassing anal-fin origin.

Planaltina kaingang, new species
Head and trunk compressed, somewhat ovoid in cross-section, broader dorsally, tapering ventrally. Pre-anal region somewhat keeled, pre-dorsal region somewhat round. Greatest width just posterior to supracleithrum. Distance between insertions of contralateral first pelvic-fin rays small, half distance between insertions of contralateral first pectoral-fin rays.
Head longer than deep. Dentary very slightly projecting anterior to premaxilla. Mouth in same horizontal as center of pupil. Tip of maxilla at vertical through anterior margin of pupil, reaching about junction of infraorbitals 2 and 3. Nostril at horizontal through dorsal margin of pupil. Eye little anterior to middle of head length, slightly dorsal to center of head depth. Infraorbital 3 touching dorsal margin of horizontal arm and anterior margin of vertical arm of preopercle. Dorsal end of gill opening in same horizontal as dorsal margin of eye.
Color in alcohol. Background light yellow. Infraorbital and opercular series and isthmus silvery. Otic region of skull brownish. Top of head and snout and anterior portion of lower jaw darkened by numerous, highly concentrated melanophores, especially in specimens smaller than 35 mm SL. Melanophores on midline of dorsum forming thin stripe from tip of supraoccipital to caudal-fin base. On remainder of dorsal region of flank and caudal peduncle, melanophores concentrated on scale margins, forming reticulate pattern. Dark longitudinal stripe from just dorsal to gill opening to base of caudal-fin rays; centered on horizontal myoseptum, along which melanophores concentration is highest; about one-scale deep from about vertical through base of first dorsal-fin ray to vertical through adipose fin, broadening slightly towards end of caudal peduncle, tapering anteriorly to vertical through first dorsal-fin ray. Longitudinal stripe occasionally covered by silvery pigmentation, partially hiding melanophores. In humeral region, melanophores of longitudinal stripe scattered, never forming spot. Melanophores absent or scarcely concentrated immediately dorsal to longitudinal stripe, leaving clear area between reticulated area dorsally and longitudinal stripe. Ventral to it, melanophores almost absent, except for a few scattered over anal-fin pterygiophores. All fins with hyaline background. Dorsal fin with melanophores on interradial membranes only, mostly edging rays. Adipose fin usually with no melanophores, rarely with a few on its anterior margin. Caudal fin with melanophores scattered over interradial membranes and embedded in rays; interradial membranes of middle rays usually with higher concentration of melanophores, but not continuing longitudinal stripe of flank; melanophores embedded in caudal-fin rays usually one in each segment, occasionally none or more than one. Anal fin with melanophores only on interradial membranes, as in dorsal fin. Pelvic fin usually with no melanophore, rarely a few embedded only in first ray. Pectoral fin with no melanophores or with some embedded in first one to three rays.
Color in life. See S2 Fig. Head silvery. Trunk somewhat translucent, with silvery glow; peritoneum silvery; lateral band distinct, but completely covered by silver pigment, hiding black pigment underneath. Melanophores evident on base of median caudal-fin rays and distal portion of all caudal-fin rays.
Sexual dimorphism. Males with hooks on anal and pelvic fins and fusion of anteriormost external ceratobranchial 1 gill filaments; pelvic-fin length apparently unrelated to sex (Table  1). Anal-fin hooks from longest unbranched to 9 th or 10 th branched ray, always 1 on each ray segment, raising from posterior margin of ray or ray branches, laterally directed (but tip directed to base of ray when hook well developed). One to 7 hooks distributed along middle to distal portion of longest unbranched ray. Usually 2-4 hooks on branched rays. First branched-ray hooks usually on proximal segments of posterior branch, but in occasional specimens also present in one or two segments immediately proximal to ramification. Second to fifth branched-ray hooks usually on proximal segments of posterior branch, but some specimens with fewer poorly developed hooks on anterior branch. Sixth to ninth or tenth branched-ray hooks on segments immediately proximal to ramification. Pelvic-fin hooks lacking on first (unbranched) ray, but occasionally present in each of remaining rays; always one on each segment, raising from medial margin of ray or ray branches, medially directed (but tip directed to base of ray when hook well developed). Each ray with up to eleven hooks from proximal region to almost tip of medial branch and only a few poorly developed ones on lateral branch, if any. Gill filaments incompletely fused, attached to one another but still seen as separate units externally.

Distribution.
Planaltina kaingang is known from the rio Keller, from the rio Ubazinho basin and from the rio Barra Grande, all immediate tributaries to the rio Ivaí; and from the rio Piquiri, upper rio Paraná basin, Paraná State, Brazil (Fig 4).
Ecology and habitat. Planaltina kaingang was collected in streams about 8-10 m wide, a few centimeters to little more than 1 m deep. A few specimens were collected in streams with bedrock bottom (e. g., arroio Lajeadão and rio Keller a few kilometers downstream from the stretch where the holotype was captured), more specimens were collected over fine gravel (rio Keller where the holotype was captured), but their abundance was by far the greatest in the rio Maria Flora, on sandy bottom, about 1.2 m deep. This locality also yielded the largest specimens. Planaltina kaingang was captured mostly close to the water surface, where it probably forages on allochthonous insects, which were found in some stomachs observed. Planaltina kaingang in the rio Keller was collected along with Piabarchus aff. stramineus Eigenmann (a species to which it is overwhelmingly similar, also lacking a humeral mark), apparently forming a single school in which the latter was more abundant. In the rio Maria Flora, P. kaingang was by far the most abundant species and was caught along with Oligosarcus cf. paranensis Menezes & Géry (small-sized individuals) and Bryconamericus coeruleus Jerep & Shibatta, which has a subterminal mouth and is also abundant. The maximum size reached by P. kaingang is evidently superior to that observed for other congeners, reaching up to 54.6 (female) and 53.7 (male) mm SL. The smallest males of P. kaingang observed to present fully developed hooks are about 32.1 mm SL. The maximum SL for each species was as follows: P. britskii, 37.4 (female) and 37.4 mm SL (male; fully developed hooks already at 26.0 mm SL); P. glandipedis, 30.4 (female) and 29.5 mm SL (male; fully developed hooks already at 25.6 mm SL); P. myersi from the rio Paranaíba basin, 36.6 (female) and 36.7 mm SL (male; fully developed hooks already at 29.8 mm SL); and P. myersi from the rio São Francisco basin, 42.1 (female) and 31.1 mm SL (male; fully developed hooks already at 28.4 mm SL).
Etymology. Planaltina kaingang is named after the Kaingang, an ethnic group that have been inhabiting the southern portion of Brazil for centuries, including some areas in the vicinities of the rio Ubazinho drainage, where the new species was first recognized.

Key to the species of Planaltina
1. Adipose fin absent; all premaxillary and dentary teeth with up to three cusps _______________ __________________________________________________________________ P. glandipedis 1'. Adipose fin present; at least some of the inner-series premaxillary teeth and anterior dentary teeth tetra-and pentacuspid _______________________________________________ 2 2. A single, very elongate scale forming the dorsal margin of the caudal pouch opening in both males and females; a complex of fused scales forming the anterior margin of the caudal pouch opening (Fig 2D) ________________________________________________ P. myersi 2'. More than one scale forming the dorsal margin of the caudal pouch opening (except female of P. britskii, but then scale is not elongate); no fused scales in the anteroventral margin of the pouch (Fig 2A-2C) _____________________________________________________ 3 3. Melanophores present on some caudal-fin interradial membranes, but absent from caudal-fin rays ( Fig 6B); males with two scales forming the dorsal margin of the caudal pouch opening and one adnate scale forming the medial wall of the pouch (Fig 2B), females with only one scale (rarely two) forming the dorsal margin of the caudal pouch opening, not elongate ( Fig 2C); 12-15 (rarely 15) circumpeduncular scale rows _____________________ P. britskii 3' Melanophores embedded in all caudal-fin rays and interradial membranes ( Fig 6A); 2-4 elongate scales forming the dorsal margin of the caudal pouch opening regardless of gender (Fig 2A); 15-17 circumpeduncular scale rows _____________________________ P. kaingang

Comments on Lepidocharax burnsi
The specimen of Lepidocharax burnsi depicted by Ferreira et al. [5] is not the holotype, as the caption claims, but the paratype MCP 31798, which was collected along with the holotype (MCP 45718). All of the morphometric and meristic data presented by Ferreira et al. [5] as belonging to the holotype has, however, been checked and considered to be correct. A picture of the true holotype is given in Fig 8A. Both specimens are very badly preserved regarding their color patterns, but the MCP 31798 specimen also seems to present a malformation in the snout, which allowed us to identify the mistake. The holotype and especially the remainder of the paratypes and the non-type specimens look exactly like a Planaltina species regarding color pattern and body shape (Fig 8). However, as mentioned before, they differ by the caudalfin squamation pattern.

Geographic distribution
Despite all species of the genus inhabit the upper rio Paraná basin, the geographic distribution of Planaltina species is basically allopatric: P. kaingang was found in the Ivaí and Piquiri river basins, and P. myersi in the rio Paranaíba basin; both the Ivaí and Paranaíba drainages have been considered as endemic regions by previous authors [15,35]. The rivers draining the cuestas of the State of São Paulo, where P. glandipedis is endemic, have also been suggested to have their own exclusive species [36].
The rate of endemism in the rio Ivaí is equivalent to that of the rio Jordão, in which 11% of native species are endemic [40]. In spite of being much smaller, the rio Jordão is part of the lower rio Iguaçu basin, where major adaptive radiations took place due to a series of waterfalls that isolated its tributaries from one another and from the rio Paraná basin [41,42]. Species from the rio Ivaí, on the other hand, are not geographically restricted by self-evident barriers. This is curious, because there is noticeably a number of species that are present in the upper rio Paraná floodplain and widespread in the upper rio Paraná basin, but not in the rio Ivaí basin (except, in some cases, in the very lower portion). Some examples are Aphyocheirodon hemigrammus, Cyphocharax vanderi, Steindachnerina insculpta, Leporellus cf. vittatus, Schizodon altoparanae, Microglanis garavelloi, Rhyacoglanis paranensis and Piabarchus stramineus (P. aff. stramineus, a species without humeral spot and a diffuse spot on caudal fin, is present). This suggests a double-way barrier, whose nature is unknown, preventing some species to disperse upstream, and others to disperse downstream.

Diagnostic characters
Planaltina species, other than P. kaingang, have the caudal fin roughly hyaline except for the median portion, through which the lateral band of the flank continues posteriorly almost to the tip of the rays; a close observation reveals that melanophores may be present throughout the fin, usually bordering the fin rays, never on the rays themselves. In P. kaingang, melanophores are more homogeneously distributed, even embedded in the fin rays, and, to the naked eye, the caudal fin looks dusky. This character was analyzed in a very large sample and is conservative among Planaltina species, thus permitting the ready recognition of the new species. Among other Stevardiinae available to the study, those presenting the most similar condition to P. kaingang are Diapoma speculiferum, D. terofali and Mimagoniates microlepis (Steindachner). In those species, there are usually one or two melanophores in each ray segment throughout the caudal fin, but they are also spread on the interradial membranes and not particularly concentrated along the borders of the rays. The latter condition, in which the melanophores are united along the borders of the rays, forming thin black lines, is evident in, e.g., Bryconamericus exodon Eigenmann (especially on distal portion of caudal-fin lobes), B. aff. iheringii (Boulenger) [15]) and Piabina argentea Reinhardt. Occasional melanophores embedded in caudal-fin rays appear in specimens of most aforementioned species, but rarely in all segments of all fin rays, and they are usually more concentrated along the margins of the rays. In any case, this character seems to be less conservative in those species than in Planaltina species. Lepidocharax burnsi presented a pattern similar to that of P. britskii, P. glandipedis and P. myersi.
The caudal-fin squamation pattern, which is important to diagnose Planaltina species, is very conservative as well. Both P. britskii and P. myersi exhibit conditions that are probably autapomorphic: P. britskii is the only species in Diapomini to present a sexually dimorphic caudal-fin squamation, and P. myersi is the only one to exhibit a fusion of scales surrounding the opening of the caudal pouch. Only P. glandipedis presents a similar squamation in comparison to P. kaingang, but the two species are easily distinguished by the presence or absence of the adipose fin and by the dentition. Both characters are invariable in the specimens examined herein.

Sexual dimorphism
Sexual dimorphism in Planaltina species is homogeneous regarding fin hooks (see Menezes et al. [10]). Gill glands, in contrast, were observed herein in the new species and also in all males of P. britskii and P. myersi, but not in P. glandipedis, despite the fact that several hookbearing individuals were analyzed for it. However, it is also possible that the gland is present in P. glandipedis, but much less developed than in the other species (see Terán et al. [48] for the presence of underdeveloped glands in some species of Astyanax Baird & Girard). Burns and Weitzman [49] first described a similar organ in Corynopoma riisei Gill, and posterior papers described it for several other characids [48,50]. Mirande [2] was the first to employ this character in a phylogenetic analysis of the Characidae, finding it to have been independently acquired at least two times and lost a few times throughout the evolution of the group. Herein this character is better understood as reversed in P. glandipedis. Sexual dimorphism is otherwise found in Planaltina only in the caudal-fin squamation of P. britskii, as commented above.