New species of Moenkhausia Eigenmann , 1903 ( Characiformes : Characidae ) with comments on the Moenkhausia oligolepis species complex

A new species of Moenkhausia is described from tributaries of the rio Paraguay, Brazil. The new species is diagnosed from congeners by characters related to body coloration, the number of lateral line scales, the degree of poring of the lateral line, and number of scales rows above and below the lateral line. Molecular analyses using partial sequences of the mitochondrial gene Cytochrome Oxidase I from specimens of the new species and specimens belonging to morphologically similar species demonstrated that the new species is easily differentiated by their high genetic distance and by their position in the phylogenetic hypothesis obtained through the Maximum Parsimony methodology. The analyses of three samples of M. oligolepis also revealed that they have high genetic distances and belong to different monophyletic groups suggesting that this species corresponds to a species complex rather than a single species.


Introduction
The distributed characid genus Moenkhausia was proposed by Eigenmann (1903) based mainly on the combination of the presence of premaxillary teeth in two rows, an inner premaxillary tooth row with five or more teeth, a completely pored lateral line without a downward curve anteriorly, and a sheath of scales covering the base of caudalfin lobes, characters still used to substantiate the generic allocation of new species to Moenkhausia (see Benine et al., 2007;Lima et al., 2007;Lucinda et al., 2007), considering the nonexistence of a phylogenetic definition.Currently allocated as incertae sedis in Characidae, Moenkhausia comprises, at present, 65 valid species widely distributed in the Neotropical Cis-Andean river basins, except for those in Patagonia, with its greatest diversity occurring in the basins of the Amazon and Guianas (Lima et al., 2003).
Albeit the genus Moenkhausia cannot be defined as monophyletic at this time, a group of species within Moenkhausia shares an identical or, at least, very similar color pattern consisting of an overall reticulate pattern formed by dark borders on the scales, a vertically-elongate humeral blotch, a conspicuous dark blotch on the caudal peduncle preceded by a lighter area, and frequently a deep red eye in life.This group consists of M. oligolepis Günther, M. sanctaefilomenae Steindachner, M. cotinho Eigenmann, and M. pyrophthalma Costa, as suggested by Costa (1994).Lima & Toledo-Piza (2001) discussed a putatively close relationship of M. dyktiota Lima & Toledo-Piza with these cited species also on the basis of its overall similar colour pattern, albeit without the characteristic round, dark, blotch on the caudal peduncle.
Despite the similar color pattern, these species are easily differentiated from each other on the basis of number of lateral line scales and number of series of scales above and below lateral line.Nonetheless, misidentifications between Moenkhausia oligolepis and M. sanctaefilomenae are common.According to Lima et al. (2003), M. oligolepis occurs in the Guianas and in the Amazon River basin, whereas M. sanctaefilomenae is distributed in the Parnaíba, São Francisco, upper Paraná, Paraguay and Uruguay River systems.Detailed examination revealed that not does only Moenkhausia oligolepis occur in the rio Paraguay basin but also that new similar species occur in sympatry revealing a species complex.In the present paper, we describe a new species of the M. oligolepis species complex and provide an analysis of its phylogenetic position based on molecular data.Morphometric and meristic data were taken following Fink & Weitzman (1974), except for head depth (HD), which was measured in the vertical through the middle of the orbit.Counts are followed by their frequency in parentheses.Asterisks indicate values for the holotype.Vertebrae of the Weberian apparatus were counted as four elements and the fused PU1+U1 as a single element.Vertebra and supraneural counts were taken from two cleared and stained (c&s) specimens prepared following the method of Taylor & Van Dyke (1985).

Material
Total DNA was extracted from ethanol-preserved muscle tissues with the Wizard Genomic DNA Purification Kit (Promega).Partial sequences of COI gene were amplified with the primers Fish F1 and Fish R1 (Ward et al., 2005).Primer final concentrations were 5 mM.Amplifications were performed by the polymerase chain reaction (PCR) in a total volume of 25 ml for 35 cycles (30 s at 95 o C, 60 s at 50-60 o C, and 120 s at 72 o C).The PCR products were identified in a 1% agarose gel.The amplified segments were extracted from the gel with the kit GFX TM PCR DNA and Gel Purification (Amersham Pharmacy Biotech Inc).The sequencing reactions were done with the kit DYEnamic ET Terminator Cycle Sequencing (Amersham Pharmacy Biotech Inc.) and analyzed in an automated sequencer ABI Prism TM 377 DNA Sequencer (Perking-Elmer).All sequences were read at least twice (forward and reverse).Individual sequences of each species were initially analyzed using BioEdit 5.0.9 (Hall, 1999) and a consensus sequence was obtained for each DNA segment for each species.All sequences were then aligned with DAMBE (Xia & Xie, 2001).Genetic distances (Kimura, 1980) were examined using MEGA 4.0 (Kumar et al., 2004).
Phylogenetic analyses based on the maximum-parsimony method (Swofford & Berlocher, 1987) were performed using PAUP* beta version 4.0b10 (Swofford, 2002) with heuristic searches using random addition of sequences and the tree bisection and reconnection (TBR) algorithm.In all analyses the character-state optimization method employed was the accelerated transformation (ACCTRAN).Parsimony trees were generated using Ti/Tv ratio of 1:1 and considering terminal gaps as missing data.Bootstrap resampling (Felsenstein, 1985) was applied to assess support for individual nodes using 1000 replicates with random additions and TBR branch swapping.

Color in alcohol.
Overall ground coloration dark silver or yellowish tan (depending on degree of retention of guanine).Dark chromatophores concentrated on distal margin of scales resulting in conspicuous reticulated pattern.Mid-dorsal region darker than flanks.Humeral region with dark inverse triangle-shaped blotch located on second to third lateral line scales, extending 3-4 scales vertically, including lateral line.Few dark chromatophores scattered on infraorbitals and opercle.Dark thin stripe extending along horizontal septum on posterior half of body.Caudal fin with conspicuous dark blotch extending from its anterior portion onto proximal onethird of caudal-fin rays.Caudal peduncle with very widely scattered dark pigment resulting in conspicuous lighter area preceding caudal-fin blotch.Dorsal fin with scattered dark pigmentation, more concentrated on anterior half.Anal fin with scattered dark pigmentation.Paired fins hyaline with scattered dark pigmentation, more so on unbranched rays.
Sexual dimorphism.One paratype (30.9 mm SL, MZUSP 90270) has small hooks on the segments of each anal-fin ray (up to four hooks per segment in each side of the anal fin).Additional non-type material from the upper rio Paraná system (LBP 5225) collected during the reproductive season included males with the same pattern of hooks.No further dimorphic characters were observed.
Distribution.Moenkhausia forestii is known from tributaries from rio Paraguay drainage (Mato Grosso and Mato Grosso do Sul States), from rio Baía (upper rio Paraná system at Batayporã, Mato Grosso do Sul State) and from rio Paraná (at Porto Rico, Paraná State) (Fig. 3).
Etymology.The specific epithet, forestii, is in honour of Fausto Foresti for his contributions to our knowledge of fish genetics.The sequences obtained in this study have been deposited in GenBank (Table 2).Genetic distances (Kimura, 1980) range from zero among specimens of Moenkhausia forestii from Paraná basin to 0.200 ± 0.021 between M. oligolepis from the rio Araguaia Basin and M. sanctaefilomenae from the upper rio Paraná system (Table 2).The combined sequence data of the 35 specimens resulted in a matrix with 611 base pairs (bp), from which 419 were conserved sites and 181 were phylogenetically informative.The MP consensus tree obtained from the analysis of 1000 bootstrap replicates is presented in Fig. 4.

Molecular
The MP analyses showed that there are six monophyletic groups supported by values of equal or higher than 99% (Fig. 4).Three monophyletic groups corresponding to the species Moenkhausia sanctaefilomenae, M. oligolepis and M. forestii and three monophyletic groups correspond to the different samples of M. oligolepis.The relationship between  Table 2. Genetic distance (mean ± standard error) among fishes of each locality (diagonal) and among samples from different localities (below diagonal).Numbers after species names are Genbank accession numbers.
M. oligolepis from the Amazon and Paraguay river basins was weakly supported but they are apparently different from the sample from the rio Araguaia basin.

Discussion
According to Eigenmann (1907), the degree of development of the lateral-line (number and position of perforated scales) distinguishes Moenkhausia (complete lateral line) from Hemigrammus (incomplete lateral line) and Psellogrammus (interrupted lateral line).This purported difference was recently discussed by Costa (1994), Lima & Toledo-Piza (2001), who highlighted the weakness of the phylogenetic information in this feature.These authors described their new species in Moenkhausia, even though those forms demonstrated incompletely pored lateral-lines and, thus failed to conform to the traditional definition of the genus.They based their generic assignments on other evidence, specifically overall similarity with other species of Moenkhausia that they felt justified this generic assignment.Such procedure was equally adopted herein.Eigenmann (1917: 84) included specimens with incomplete, interrupted, and complete lateral line in a single species, Moenkhausia sanctaefilomenae.In fact, this author raised the potential of the existence of more than one species in those samples, noting that the forms with incomplete lateral line would correspond to his M. australe, and those with both interrupted and complete lateral line would correspond to M. sanctaefilomenae.The basis for Eigenmann's actions was unclear, but he presumably included samples with different degrees of development of the lateral line in a single species in the absence of broad samples at his disposal and consequently placed M. australe in the synonymy of M. sanctaefilomenae.Moenkhausia australe was, however, described by Eigenmann (1908) from two specimens collected at two different localities in Paraguay, Arroyo Trementina and Arroyo Chagalalina.Those specimens were examined as part of this study and, contrary to Eigenmann (1917), these specimens do not have an incompletely pored lateral line and present a higher upper jaw length in comparison with Moenkhausia forestii (48.9-52.6% vs. 39.9-43.7% in HL, respectively), thus, M. australe is not equivalent to the new species described herein.Additional material from rio Uruguay (MCP 25198 and MCP 27464) with a complete lateral line has a comparable upper jaw length (47.2-52.4% in HL) with the syntypes of M. australe.
The observed values of genetic distances (Kimura, 1980) among species were 0.182 ± 0.019 between M. sanctaefilomenae and M. forestii, 0.183 ± 0.017 between M. oligolepis and M. forestii, and 0.191 ± 0.017 between M. sanctaefilomenae and M. oligolepis.These values are higher than those obtained in the similar analyses of within-genus COI sequences (0.0993) of marine fishes from Australia (Ward et al., 2005) and of freshwater fishes from Canada (Hubert et al., 2008) and from Mexico and Guatemala (Valdez-Moreno et al., 2009).The observed values of genetic distances withinspecies were 0.003 ± 0.001 for M. sanctaefilomenae, 0.004 ± 0.001 for M. forestii, and 0.092 ± 0.009 for M. oligolepis.This high value observed for M. oligolepis leads us to hypothesize that M. oligolepis still represents a species complex rather than a single species which will be explored in a further study.
Searches through the collections at MZUSP revealed that Moenkhausia forestii has been only recently collected in the upper portions of the rio Paraná system, most probably as a consequence of movement of the species through the "Piracema Channel" (a fish ladder constructed around Itaipu Hydroelectric).This fish ladder linked the middle and upper portions of rio Paraná, that were originally separated by the natural barrier formed by the Sete Quedas falls.Several other examples of similar establishment of species in the upper Paraná as a consequence of movement via this fish ladder are cited by Graça & Pavanelli (2007).Due to this secondary establishment, the specimens from rio Paraná drainage were included as non-type material.Non-types also includes dissected, alcohol-fixed, and poorly preserved specimens.
examined in this study is deposited in the California Academy of Sciences (CAS), San Francisco; Laboratório de Biologia e Genética de Peixes do Departamento de Morfologia da Universidade Estadual Paulista (LBP), Botucatu, SP; Museu de Ciências e Tecnologia, Pontifícia Universidade Católica do Rio Grande do Sul (MCP), Porto Alegre, RS; and Museu de Zoologia da Universidade de São Paulo (MZUSP), São Paulo, SP.Vouchers of the material used in the molecular analysis are deposited at LBP.

Fig. 3 .
Fig. 3. Map of the rio La Plata basin showing the localities of collection of Moenkhausia forestii, new species.Star represents the type locality.

Fig. 4 .
Fig. 4. Maximum Parsimony consensus tree for the mitochondrial gene Cytochrome Oxidase I. Numbers represent values of 1000 bootstrap replicates.