Description of a new species of Moenkhausia (Characiformes: Characidae) from the upper Paraguay basin, Central Brazil, with comments on its phylogenetic relationships

A new species of Moenkhausia is described from tributaries of the upper rio Sepotuba, Paraguay basin, Brazil. The new species is distinguished from its congeners by a combination of characters, including an inconspicuous oval-shaped vertically elongated humeral blotch, extending horizontally from third through five lateral-line scales, and vertically from third row above lateral line to first row below it, followed by a diffuse field of dark chromatophores in the flank, combined with a welldefined dark line at the base of the anal fin. Furthermore, the phylogenetic position of the new species is presented based on molecular data, showing a close relationship among species of Moenkhausia and Hemigrammus that have a well-defined dark line at the base of the anal fin. Until this moment, this species is only known from in the upper rio Sepotuba basin.

Despite of these recent efforts in trying to identify the sister species of Moenkhausia xinguensis and Hemigrammus unilineatus, and propose a phylogenetic definition for both genera, the low number of taxa included in those analyses, along with the lack of understanding of closely related, species-rich genera such as Astyanax and Hyphessobrycon, has precluded a more consistent and useful definition for these and other characid genera.
Collecting efforts in headwater streams of the upper rio Sepotuba, rio Paraguay basin, revealed an undescribed species of small characin that, despite of its similarity to species of Hemigrammus, such as H. lunatus, would better fit to the artificial and traditional definition of Moenkhausia, since it has a completely pored lateral line.Herein, we present a phylogenetic analysis based on molecular data to infer and discuss its phylogenetic relationships within Characidae and formally describe this new species.

Material and Methods
Morphological data.Counts and measurements followed Fink, Weitzman (1974), except for counts of scale rows, which follow Lima et al. (2007) and with the addition of pelvic-fin origin to anal-fin origin measured at origin of pelvic-fin through the anal-fin origin.
Measurements were taken point to point with a digital caliper on the left side of specimens whenever possible (precision of 0.1 mm).All measurements are expressed as percentage of standard length (SL) or head length (HL).Values in the parentheses indicate the number of specimens with a particular count and an asterisk indicates values of the holotype.Vertebrae of the Weberian apparatus were counted as four elements and the fused PU1+U1 as a single element.Vertebrae, supraneural counts and gillrakers of first arch were taken from seven cleared and stained (cs) specimens prepared following the method of Taylor, Van Dyke (1985).
Molecular data.The present study included 30 species currently allocated in Moenkhausia, including the new species.As a framework for the selection of species to be included in our analysis we follow the results presented in Oliveira et al. (2011), selecting representatives of their node 31, which includes the Acestrorhynchidae, Bryconidae, Chalceidae, Characidae, Gasteropelecidae, Iguanodectidae, and Triportheidae.Molecular methods.Total DNA was extracted from ethanol preserved muscle samples using DNeasy Tissue Extraction Kit (Qiagen), following manufacturer's instructions.Partial sequences of the genes 16SrRNA, Cytochrome b (Cytb), 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 same primers utilized by Oliveira et al. (2011).Amplifications were performed in a total volume of 25 ml with 2.5 ml of 10X buffer (10mM Tris-HCL + 15mM MgCl2 buffer), 0.5 ml MgCl2, 0.5 ml each primer (5 mM); 0.4 ml dNTPs (200 nM of each), 0.2 ml Taq Platinum polymerase (Invitrogen), 1 ml template DNA (10-50ng) and 19.4 ml ddH2O.The thermo-cycler profile used for the fragments 16SrRNA and Cyt b was with 35 cycles and annealing temperature of 50-55 o C. Nested-PCR was used to amplify the nuclear genes rag1, rag2, and Myh6.Conditions for amplification of these genes for both rounds of PCR used 15 cycles with annealing temperature at 56 o C followed by 15 cycles with annealing temperature at 54 o C. PCR products were purified using ExoSap-IT® (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 on an automatic sequencer 3130-Genetic Analyzer (Applied Biosystems) at Instituto de Biociências, Universidade Estadual Paulista, Botucatu, São Paulo, Brazil.
Alignment and phylogenetic analyses.Contigs were assembled and edited in Geneious v 5.4 software (Kearse et al., 2012).Sequences were independently aligned using the muscle algorithm under default parameters (Edgar, 2004), and alignments were inspected for any problems.The first alignments were initially analyzed by neighbour-joining (NJ) using MEGA 7.0 software (Kumar et al., 2016) to control potential sequencing errors.The sequences were translated using Geneious v 5.4 (Kearse et al., 2012) to check for the unexpected occurrence of stop codons.PartitionFinder v1.1.1 (Lanfear et al., 2012) was used to determine codon-specific models of molecular evolution for each gene under the Bayesian information criterion (BIC).A generalized time reversible model with rate heterogeneity of the remainder being model by gamma distribution with a proportion of invariable sites (GTR+Gamma+I) was identified as the best model of molecular evolution for the first, second e170086[3] and third codon position.Phylogenetic hypotheses were inferred from the data set with maximum-likelihood (ML), maximum parsimony (MP) and Bayesian inference (B).Maximum-likelihood (ML) analysis was performed using RAxML Web-Servers Black-Box (Stamatakis et al., 2008) with a mixed partition approach, with GTR + G + I as the model.Partitioned analysis was performed, with the data set divided into three sections corresponding to the first, second and third positions of the gene.Random starting trees were used for each independent ML tree search, and all other parameters were set at default values.Topological robustness was investigated using 1000 nonparametric bootstrap pseudoreplicates (Felsenstein, 1985).Maximum parsimony analyses were conducted with PAUP* 4.0b10 (Swofford, 2002).Heuristic searches were performed with random addition replicates (minimally 1000) and TBR branch swapping.Character transformations were equally weighted and branches with a maximum length of zero were collapsed.Gaps were treated as missing data.Clade robustness was assessed using 1000 bootstrap pseudoreplicates with the same parameters as above (Felsenstein, 1985).Phylogenetic analysis using a partitioned Bayesian approach was conducted in MrBayes 3.1.2(Ronquist, Huelsenbeck, 2003) with the models and partitions determined as the best scheme according PartitionFinder (Lanfear et al., 2012).MrBayes was configured to run for 10 million generations using eight chains (nchain = 8, being two parallel runs with one cold and 7 hot chains each; temperature parameter set to default).Results were analyzed in the Tracer v. 1.6 (Rambaut et al., 2014), TreeAnnotator v. 1.8.0 (available as part of the BEAST package, Drummond et al., 2012) and FigTree v. 1.4.2(Rambaut, Drummond, 2012) software programs, after deleting burn-in trees, which were approximately 30%.Moenkhausia flava is quite similar with H. barrigonae and M. conspicua by sharing overall body shape and similar color pattern.In addition, some specimens of H. barrigonae also present complete lateral line (see Géry, 1977:503;Soares, Bührnheim, 2016: 398), similar to the new species and M. conspicua.Moenkhausia flava differs from H. barrigonae and M. conspicua by presenting inner having premaxillary and dentary teeth pentacuspid (vs.inner premaxillary and dentary teeth heptacuspid in both species).Additionally, differs from H. barrigonae by larger body depth (31.0-42.6%SL, mean = 35.8%SL vs. 30.4-34.8%SL, mean = 32% SL); and differs of M. conspicua by the dorsal-fin length (24.3-31.3% vs. 31.2-36.8%SL, respectively), anal-fin length (14.8-22.1% vs. 22.6-27.1% SL, respectively) and orbital diameter (34.8-43.4% vs. 44.9-56.0%HL, respectively).Moenkhausia flava can be easily distinguished from H. machadoi and H. lunatus by an inconspicuous oval-shaped vertically elongated humeral blotch, extending horizontally from third through five lateralline scales, and vertically from third row above lateral line to first row below it (vs.a conspicuous vertically elongated dark humeral blotch, extending horizontally from second through sixth lateral-line scales, and vertically from third row above lateral line to first row below it in H. machadoi; and a small roundish humeral blotch in H. lunatus).The new species differs from M. collettii and M. copei by having six longitudinal rows of scales above lateral line (vs.five longitudinal rows of scales).Additionally, it is distinguished from M. copei by a greater number of branched analfin rays (20-23 vs. 15-17 in M. copei) and distinguished from H. machadoi, H. lunatus and M. collettii by a threescale deep band of sparse, scattered dark chromatophores extending along midlateral body (vs. a line of concentrated chromatophores along midlateral body).
Tab. 1. Morphometrics data of holotype and paratypes of Moenkhausia flava.Standard length (SL) is expressed in mm, all other measurements are expressed as percentages of SL, except for subunits of head that are expressed as percentages of head length (HL).N=46.Sexual dimorphism.Adult males with small hooks on the last unbranched and anterior four branched anal-fin rays.Also, found on the first and second pelvic-fin rays branched.In both fins, there are four to seven hooks per fin ray, located on the distal segments.
Color in alcohol.Humeral region with an inconspicuous oval-shaped vertically elongated humeral blotch located on second to fourth lateral line scales, extending from 3 horizontal series of scales above lateral to the series of scales immediately below the lateral line.Dark chromatophores scattered on infraorbitals and opercle, longitudinal dark 6 e170086 [6] lighter stripe along on the eye (Fig. 2).A dark thin stripe extending along horizontal septum, from humeral region to caudal peduncle, more evident on posterior half of the body.A three-scale deep band of sparse, scattered dark chromatophores extending along midlateral body.Dorsal fin rays with few dispersed chromatophores, more concentrated on anterior half.Anal-fin rays with few dispersed chromatophores, more concentrated along its proximal and distal extension, resulting in a lighter medial area.Tip of anterior anal-fin rays densely pigmented by dark chromatophores resulting in a dark dash in this area.Paired fins hyaline with scattered dark pigmentation, more concentrated on unbranched rays.Caudal fin with a narrow field of dark chromatophores on its distal margin.Conspicuous dark line at anal-fin base.All fins hyaline with few dispersed chromatophores.Anterior rays of the anal fin and base of caudal-fin lobe present orange pigmentation in freshly preserved specimens.
Color in life.Body general color pattern pale yellowish.Dorsal region olive.Abdomen whitish to light yellow.Pelvic fin and adipose fin with yellow-orangish coloration.Dorsal fin and caudal fin with orange-reddish coloration.Anal fin with first rays with orange-reddish coloration and the remaining rays with hyaline coloration.A conspicuous dark line at the base of the anal fin.Pectoral fin yellowish (Fig. 4).

Discussion
The taxonomic distinction between Moenkhausia and Hemigrammus is currently based solely on a single character state: completely pored lateral line vs. incompletely pored lateral line respectively, as proposed by Eigenmann (1903), Eigenmann, Myers (1917) andDurbin (in Eigenmann, 1918).Moenkhausia flava possesses a completely pored lateral line, but a body pattern similar to species of the genus Hemigrammus.
Recent phylogenetic hypothesis including species of both Moenkhausia and Hemigrammus (Mirande, 2009(Mirande, , 2010;;Oliveira et al., 2011;Mariguela et al., 2013) showed the non-monophyletic condition of these genera.In Mariguela et al. (2013), 50% (14 species) of the analyzed species of Moenkhausia (29 in total) grouped with its type species, Moenkhausia xinguensis.It is noteworthy that this group also included the incomplete lateral-lined Hemigramus marginatus Ellis, 1911; Hasemania sp., and Nematocharax venustus Weitzman, Menezes & Britski, 1986.According to our results, Moenkhausia flava is more close related to H. unilineatus (the type species of Hemigrammus) than to M. xinguensis (type species of Moenkhausia), reinforcing the weakness of the diagnostic characters between Moenkhausia (completely pored lateral line) and Hemigrammus (incompletely pored lateral line).Our new species is also close related to Moenkhausia copei, Hemigrammus barrigonae, M. collettii, M. aff.copei, and H. ulreyi.Such a close relationship may be phenotypically supported by the general color pattern of these species.However, until more comprehensive analyses encompassing more representatives of both genera are available, we traditionally and conservatively opted to describe this new species in Moenkhausia.
Moenkhausia flava has an inconspicuous oval-shaped vertically elongated humeral blotch, somewhat diffuse or slightly pigmented, and a diffuse longitudinal stripe along the flank and a dark line at the anal-fin base, quite similar to the pattern described for M. copei (sensu Géry, 1977).This author affirmed that M. copei has a humeral spot absent or slightly pigmented, and that M. collettii has a conspicuous humeral spot.We observed that M. copei has an inconspicuous oval-shaped humeral mark, and a slightly dark longitudinal stripe along the body.Moenkhausia collettii presents a conspicuous humeral spot varying from round to square-shaped crossed by a longitudinal dark stripe extending along the flank, and both species present a dark line at the base of the anal fin (Eigenmann, Myers, 1917;Géry, 1977).Hemigrammus ulreyi also has a black line on the anal fin base, as found in M. copei, M. collettii, M. flava and, Hemigrammus barrigonae (Géry, 1977).Therefore, these species present a characteristic color pattern, which basically consists of a well-marked dark line at the base of the anal fin and a longitudinal dark stripe from posterior margin of the operculum to the caudal peduncle; and a humeral spot more or less conspicuous, but always present, varying from horizontally elongated to square-shaped.Hemigrammus machadoi, H. lunatus, Moenkhausia conspicua, and M. venerei, not included in our analysis, also share this color pattern, and probably belong to the same clade.Ota et al. (2014) had already suggested that the shared color pattern might support a monophyletic group, which they called "Hemigrammus lunatus species-group".
Although the color based characters used by Mirande (2009;2010) showed mostly homoplastic, our results are evidence that, for some groups of species, the color pattern is useful for detecting potential natural groups and should be tested in congruence analysis.It is noteworthy that Costa (1994), based on color pattern, discussed the putative monophyly of M. oligolepis, M. sanctaefilomenae, and M. pyrophthalma, species with complete, interrupted and incomplete lateral lines, respectively, which was posteriorly corroborated by Mariguela et al. (2013).

Fig. 5 .
Fig. 5. Map showing the localities of Moenkhausia flava.Red star represents the type locality.Black square represents the Salto das Nuvens fall and the white square represents Salto Maciel fall.

Fig. 7 .
Fig. 7. Maximum likelihood phylogenetic tree based on Partial sequences of two mitochondrial genes and three nuclear genes.The series of three numbers (e.g.100/94/1) at each of the main nodes represents the percentage of bootstrap support obtained by Maximum Likelihood (ML), the percentage of bootstrap support obtained by the Maximum Parsimony (MP) analysis and the posterior probability for that split obtained in Bayesian analysis (B),respectively (1000 bootstrap replicates).Dashes represent values < 50% (ML, MP) or <0.5 (B).