Systematic assessment of the Leporinus desmotes species complex , with a description of two new species

Members of the Leporinus desmotes species complex can be distinguished from other barred or banded congeners by the combination of nine distinct black bars across the head and trunk and long, pointed, laterally compressed and upward curving symphyseal dentary teeth. A taxonomic reassessment of this complex revealed two new species, one from the Orinoco and Negro rivers of Venezuela and Brazil, and the other from the Xingu and Tapajós rivers of Brazil. Both species are similar to L. desmotes and L. jatuncochi, but differ significantly in body shape morphology, coloration, and/or circumpeduncular scale counts. Genetic evidence also contributes to the recognition of both new species. This contribution also maps the geographic distribution of the four known species, and highlights the presence of an unusual meristic polymorphism within Leporinus desmotes sensu stricto that may suggest the presence of even more unrecognized diversity.

Members of Leporinus exhibit substantial morphological and ecological diversity, and vary in body size, trophic morphology, diet, habitat and color pattern.Mouth position varies from terminal to subinferior, and tooth shape ranges from rounded to multicuspidate to spatulate (Sidlauskas, Vari, 2008).Their diet is generally omnivorous, but the proportion of plant or animal material consumed can differ greatly between species (Knöeppel, 1972;Goulding et al., 1988;de Mérona et al., 2001).Most species exhibit one of three distinct color patterns: one to four midlateral blotches, one to six dark longitudinal stripes, or seven to 14 dark vertical dark bars (Sidlauskas, Vari, 2012;Sidlauskas, Birindelli, in press), and variation within these basic forms often helps to diagnose species.
Fifty-seven years after Fowler (1914) described Leporinus desmotes from specimens collected in the Rupununi River of Guyana, Ovchynnyk (1971) named Leporinus jatuncochi, a similarly barred and tusked species, from a single specimen collected in Lake Jatun Cocha within the Napo province of Ecuador.Perplexingly, Ovchynnyk (1971) never compared the newly described L. jatuncochi from the upper Amazon river basin, with the morphologically similar L. desmotes from the Essequibo river drainage.Previous work has hypothesized that L. jatuncochi may be a synonym of L. desmotes (Garavello, 1979), while others have listed both species as provisionally valid (Garavello, Britski, 2003) but no formal morphological or molecular analysis has ever tested the validity of these two species.Neither has any study investigated the taxonomic status of populations of similar fishes that have been discovered in the Orinoco river basin in Venezuela as well as in several tributaries of the Amazon River in Brazil during the century that has elapsed since the original description.Taken together, these barred and tusked Leporinus with nine branched pelvic-fin rays form a complex of populations morphologically similar to L. desmotes and L. jatuncochi, but spanning most of the cis-Andean Neotropics and potentially including unrecognized species-level diversity.We refer to these fishes here in as the Leporinus desmotes species complex.
This study reexamined the Leporinus desmotes species complex by comparing morphometrics, meristics and coloration of specimens from at or near the type localities of the two nominal species, L. desmotes and L. jatuncochi, to samples collected widely in the Amazon, Orinoco and Essequibo basins.The study also compared the genetics of a more limited sample of specimens in a molecular phylogenetic context.In so doing, we aimed to formally test the validity of the upper Amazonian L. jatuncochi as a distinct species, to evaluate the possible presence of undescribed species within the complex, and describe and diagnose such unrecognized species.

Material and Methods
Sampling design.Overall, we examined the meristics, coloration and general morphology of hundreds of specimens (see materials examined), and collected detailed morphometric data from a smaller subset of 68 specimens.Genetic data for six individuals within the Leporinus desmotes complex and several examples of other banded Leporinus were extracted from an in-progress multilocus phylogenetic study of relationships within Anostomidea.Only specimens with meristic and morphometric measurements were included in the type series for each new species discovered.Due to the small genetic sample sizes, we did not require genetic data for specimens to be assigned to a type series.Sidlauskas et al. (2011).Meristic analyses and tabular data reflect a synthesis of observations of all authors.Morphometric analysis employed thirty-seven linear measurements following Burns et al. (2014), however we used the pelvicfin and pectoral-fin origins rather than insertions for all morphometrics involving a landmark on those fins to reduce measurement variation as the insertion can be difficult to find on smaller specimens.Only specimens present at OS were used for morphometric analyses, thereby allowing a single observer (MC) to perform all measurements.Allometric differences among specimens were removed from the 37 log-transformed measurements by an allometric Burnaby technique implemented in PAST version 3.12 (Hammer, 2001).The size-standardized data were used in two different Principal Component Analyses (PCA).The first of these contained all specimens, while the second contained only the specimens from the Essequibo and upper Amazon drainages, from which the two nominal species were described.This second PCA was performed because the specimens from the Orinoco, Tapajós and Xingu populations proved to be the most morphologically distinct in the dataset, and their inclusion may have obscured more subtle differentiation between specimens from near the type localities of the described species.This more restricted PCA also helped to investigate possible morphological differentiation between specimens possessing alternative states of an intriguing scale polymorphism (see results).Because that second PCA did not reveal any greater differentiation between specimens from the Upper Amazon and Essequibo than did the global analysis, the first three eigenvectors from the global PCA, totaling 47% of the non-allometric shape variation, were employed in all further statistical analyses.The first three eigenvectors were found to account for most of the variability in the data through a scree plot investigation.

Morphology. Meristic counts follow the methodology of
Specimens for morphometric analysis were assigned a priori to four geographically circumscribed groups, each representing a potentially distinct species or population due their separation into distinct drainage basins.These included specimens from the Essequibo drainage (including the Rupununi River, type locality of Leporinus desmotes), the upper Amazon (from which L. jatuncochi was described), the Xingu, and the Orinoco basins.A MANOVA implemented in PAST tested the overall statistical distinctiveness of each group using the first three principal components from the global PCA.Pairwise Tukey's post-hoc comparisons tested the significance of differences between the group means (p < 0.05) on each morphospace axis.The untransformed variables that explained the highest proportion of variance in the PCA were regressed against standard length (SL) for each species to determine if the ontogenetic trajectories for these variables differed among species.This standardized-majoraxis (Type II) regression, as well as tests of equality of slope and intercept, were done in the smart package in R (Warton et al., 2012).

Molecular data.
Taxon sampling in the ingroup included seven individuals from the Leporinus desmotes species group, including three from the upper Amazon, one from the Orinoco, one from the Essequibo, and two from the Xingu drainages.These samples were subsetted from a more encompassing multilocus study of Anostomoidea, and represent all tissues available at the time of analysis.The voucher specimen for the tissue sample representing the Essequibo drainage has 15 circumpeduncular scales (see discussion for the relevance of this detail).The outgroup includes four other species of the banded species group (L.affinis, L. altipinnis, L. fasciatus and L. pearsoni), as well as the more distantly related L. brunneus and L. ecuadorensis.
Sequences of each gene for each individual were aligned using default parameters in the Muscle (Edgar 2004) plugin for Geneious version 9.1 (Kearse et al., 2012) A maximum likelihood (ML) topology was inferred with a partitioned RAxML analysis run through the CIPRES web server (Miller et al., 2010) with Leporinus brunneus and L. ecuadorensis set as the outgroups.A random starting tree was used for the ML search with all other parameters set to default values.All ML analyses were performed under GTR + G since RAxML only applies this model (Stamatakis et al., 2008).The robustness of the ML topology was tested with 1000 bootstrap replicates.Partitions were determined by Partitionfinder (Lanfear et al., 2012) and parameters for each partition estimated using JModelTest2 (Darriba et al., 2012).See Tab. 2 for partitions and individual models of evolution.A Bayesian topology was inferred using MrBayes 3.12 (Ronquist, Huelsenbeck, 2003) run through the CIPRES web server (Miller et al., 2010).We performed 10 million generations, sampling a tree every 1000 generations.The distribution of log likelihood scores was examined in Tracer 1.5 (Rambaut, Drummond, 2007).We determined whether runs had reached convergence by checking to ensure that ESS scores exceeded 200 and that log likelihood scores had achieved stationarity.The first 10% of the posterior was disregarded as burnin after visualizing the posterior distribution in Tracer 1.5 (Rambaut, Drummond, 2007), with the remaining trees used to construct a maximum clade credibility tree.
Tab. 1. Voucher and GenBank numbers for tissues used in phylogenetic analysis.

Meristic differentiation.
With the important exception of circumpeduncular scale counts, the four geographically defined groups do not differ meristically (Tabs.3-4).However, circumpeduncular scale counts vary from population to population, with some fixed at 14 or 16 scales, and others polymorphic.All specimens from the Orinoco, the Casiquiare canal, the upper Negro drainage, the middle and upper Xingu and upper Tapajós drainages (Fig. 2) have 16 circumpeduncular scales, while all specimens from the upper, middle and lower Amazon, including samples from the lower Xingu and the Jamanxim River (the latter of which is part of the Tapajós drainage) have 14.The specimens of Leporinus desmotes from the Essequibo drainage (including the paratype series at ANSP) are polymorphic for this character, with most specimens possessing either 14 (four specimens) or 16 scales (four specimens) around the caudal peduncle, and one specimen possessing 15.The holotype of 5 e160166 [5] L. desmotes (ANSP 39324) possesses 16 circumpeduncular scales, and the holotype of L. jatuncochi (MSUM 5868) possesses 14.The same polymorphism occurs in specimens from the Araguaia and Tocantins rivers and tributaries, with 17 specimens possessing 14 and six possessing 16 scale rows around the caudal peduncle.Interestingly, sympatric specimens from the Essequibo and from the Araguaia-Tocantins have both distinct counts (i.e., specimens from the same lot exhibit 14 and 16 scale rows around caudal peduncle), as in the paratype series of L. desmotes and in several lots from the Araguaia-Tocantins (e.g., UNT 9095, INPA 1555).In this case, variation is not related to sex or size, as specimens of the same sex and size exhibit distinct scale counts.
Tab. 3. Meristic counts of Leporinus enyae and L. villasboasorum summarized as ranges followed by the mean ± 1 standard deviation in parentheses.Single values in a cell indicate invariant counts among the examined specimens for a given species.In this morphospace, specimens from the Essequibo and upper Amazon drainages overlap substantially in body shape (Fig. 3), while those from the Xingu and Orinoco basins have obviously different centroids and overlap little with each other or with specimens from the Essequibo River (Fig. 3).The MANOVA for the first three Principal Components revealed statistically significant differences overall (Wilks' λ = 0.9, p < 0.0001).Statistically significant differences on at least one morphospace axis exist among all pairwise comparisons of groups, (Tab.5) with the specimens from the Xingu basin particularly divergent from the other three groups along PC1, and those from the Orinoco basin and Essequibo basin diverging in opposite directions on PC2.The most similar groups are those from the upper Amazon and Essequibo, which are indistinguishable on PC1, and only slightly separated (p = 0.025) on PC2.A morphospace constructed from just the upper Amazon and Essequibo specimens fails to reveal any major morphological differences between them, Wilks' λ =.76, p = 0.08 (Fig. 4).In this morphospace, the Essequibo specimens with 14 (two specimens) versus 16 (four specimens) circumpeduncular scales separate slightly along PC1, with the specimens with 14 circumpeduncular scales having a larger distance between the dorsal-fin origin and dorsal-fin insertion to the pelvic-fin origin than the specimens with 16 circumpeduncular scales.However, the small sample sizes make it unclear whether this separation reflects biological reality, or a sampling artifact (Fig. 4).99.6% sequence similarity), with that clade sister to the specimen from the Orinoco (74 bootstrap, .99posterior probability).The specimens from the Xingu separate from all others at the basal split within the complex.The relatively low support values for the placement of the Orinoco specimen suggest that additional sampling could easily alter that result, and that this phylogeny should not be considered canonical with respect to interspecific relationships within the complex.
That uncertainty notwithstanding, the sequenced individuals separate into three groups, each of which is genetically distinct and subtended by a long branch.The average pairwise distances between the specimens from the Xingu basin, that from the Orinoco basin, and the clade containing the specimens from the Essequibo and upper Amazon drainage is 5.3% for the mtDNA and 0.85% for the nDNA.Though the sample sizes are too low to be conclusive on their own, this level of differentiation suggests the presence of at least three distinct species under a phylogenetic species concept.Similar levels of differentiation have been found among other closely related species in Leporinus (Avelino et al., 2015), as well as many other Neotropical fish groups (Gomes et al., 2015;Pugedo et al., 2016).
Overall, specimens from the Xingu and Orinoco drainages each proved statistically separable from all other populations on morphometric grounds, and easily diagnosable from all specimens from the upper Amazon River and most specimens from the Essequibo, Araguaia and Tocantins rivers via circumpeduncular counts.Furthermore, the Xingu and Orinoco samples sequenced proved genetically divergent from each other and from the available samples from the upper Amazon and Essequibo.Taken together, these data indicate that that Xingu and Orinoco populations represent novel species, which we describe formally below.Leporinus desmotes - Camargo et al., 2004:132 (in     Diagnosis.Leporinus villasboasorum can be separated from all other members of Leporinus except L. desmotes, L. jatuncochi, and L. enyae by possession of laterally compressed, elongate (more than twice as long as the second dentary tooth) tusk-like symphyseal dentary teeth (Fig. 1b), versus the more truncate conical or incisiform teeth (less than twice as long as the second dentary tooth) typical of other members of Leporinus (Fig. 1c), and from all other members of Leporinus except Leporinus bleheri, L. desmotes, L. enyae, L. jatuncochi, and L. yophorus by a color pattern on the head and trunk consisting of nine dark bars, some of which fork dorsally or ventrally, the fifth of which continues onto the dorsal and pelvic fins, versus various other color patterns in the approximately 90 other described species of Leporinus.It separates most easily from Leporinus jatuncochi and some specimens of L. desmotes by the possession of 16, versus 14 circumpeduncular scales.It separates from Leporinus enyae by having the dark bar anterior to the dorsal fin well-divided dorsally, versus having that bar undivided or barely divided dorsally.It is most similar in meristics and coloration to specimens of Leporinus desmotes with 16 circumpeduncular scales, but can be separated from those specimens morphometrically.Of the numerous morphometrics that differ in mean values between Leporinus villaboasorum and Leporinus desmotes, the most diagnostic are the shallower body depth in the former (23.0-24.9%SL versus 25.1-28.9%SL) and the shorter distance from the pelvic-fin origin to adipose-fin origin (36.3-38.6%SL versus 38.5-43.4% SL) with the regression lines differing greatly between the species (Fig. 8).Description.Holotype and paratype meristic values in Tab. 3, and morphometric values in Tab. 6. Body fusiform, elongate, slender, slightly compressed, deepest and widest at dorsal origin.Slight keel anterior of dorsal insertion and posterior to supraoccipital crest, all other margins of body gently rounded.Caudal peduncle compressed and equally long as deep.Head conical, compressed slightly dorsally and more so ventrally.Dorsal surface of head straight and inclined in lateral profile.Anterior profile of head forms a rounded point.Head widest at back of opercle.Eye rounded and laterally placed.Anterior nostril tubular, flared at distal end, positioned just anterior of second black transverse bar on head and overhanging the upper lip.Posterior nostril in form of sharply upturned slit positioned slightly posterior to anterior nostril at angle paralleling angle of lateral profile of head.Mouth small and subterminal.Lips fleshy, thick with fleshy, deeply-cleft lamellae.Upper lip with two series of lamellae, outer series longer than inner.Bottom lip with three or more series of lamellae much shorter than those of top lip and irregular in pattern, with outer series terminating on either side with a deep-cleft narrow fleshy lobe.Upper teeth three on each premaxilla, broader than the lower teeth and slightly spoon shaped, vaguely tricuspid and curving faintly posteriorly, with lateral teeth on smaller than those more medial.Lower teeth three on each dentary in number with lateral teeth much smaller than symphyseal.Symphyseal teeth long, pointed, laterally compressed and upward curving.Gill openings restricted.Opercle and preopercle smooth, with opercular flap pronounced and extending posteriorly from operculum.Four branchiostegal rays on each side.
Scales almost uniformly large with only slightly smaller scales in predorsal and ventral regions.Lateral line well defined, complete and straight with simple well-defined tubes.Pointed sheath scale twice as long as fin base on pelvic, starting at pelvic origin and pointing posteriorly.
Dorsal-fin origin located about halfway between snout tip and adipose origin.Pelvic-fin insertion located at vertical through origin of seventh branched dorsal-fin ray.Adipose-fin origin located at vertical through origin of third branched anal fin ray.Distal end of posterior branched anal-fin ray located at vertical through shallowest point of caudal peduncle.Dorsal-and pectoral-fins sharply falcate, all other fins come to slightly rounded points.First unbranched fin ray about half as long as first branched fin ray, with first branched fin ray being the longest, on both dorsal-and anal-fins.Caudal-fin deeply forked.Vent located anterior of anal origin about onethird the distance between anal origin and pelvic insertion.

Coloration in alcohol.
Ground color white to brownish tan, weakly countershaded.Nine well-defined dark black transverse bars encircling head and trunk in specimens larger than 30 mm SL, eight such bars in small juveniles.Head with one dark bar on top of snout, one dark stripe running from tip of lower jaw, through end of maxillary bone, to anterior margin of orbit, conspicuously present in specimens of 30 mm SL or less, and slightly less conspicuous and distinctly arranged in larger specimens (with faint connections to the second dark bar).Head with one dark bar extending from dorsalmost portion of head (between frontals and parietals) to dorsal margin of eye; bar extended below ventral margin of eye only in specimens of 50 mm SL or more.Bars on main body slope slightly toward posterior from dorsal to ventral.Trunk with six dark bars in specimens of 30 mm SL or less and with seven in larger specimens.Variation in bar number stems from division of bar dorsal to pelvic fin over ontogeny.First dark bar of trunk encompasses opercle and much of the isthmus with a slight ventral division in line with gill opening and black resuming in space between pectoral-fins ventrally.Second dark bar of body between head and dorsal-fin; bar wide and undivided in specimens of approximately 20 mm SL, bar wide and only slightly divided into two parallel bars in specimens of around 25 mm SL, bar completely dorsally and ventrally shaped as an "X" in specimens of 60 mm SL or larger.Third and fourth dark bar united into a single bar in specimens of 20 mm SL, divided dorsally and ventrally in specimens of 25 mm SL, and completely divided into two bars in specimens larger than 60 mm SL.Third dark bar extended on base on dorsal-fin rays, forming a dark triangle, with anteriormost rays darker than posteriormost; and extended on pelvic-fin.Pelvic-fin completely dark, except anteriormost tip, in specimens of 25 mm SL or less; pelvicfin dark except for anteriormost tip, first two unbranched rays and posterior margin in larger specimens.Fifth dark bar in front of adipose fin and extended to base of four to six anteriormost rays of anal-fin.Sixth dark bar running from posterior portion of adipose to posterior portion of anal-fin.Adipose-fin dark except for middle portion in specimens of 25 mm SL or less, and dark except for insertion and distal margin in larger specimens.Seventh dark bar covering posterior third of caudal peduncle.Specimens with 25 mm SL or less with a dark semicircular bar extended on base of caudal-fin rays, and a conspicuous dark spot on base of middle caudal-fin rays.Caudal-fin generally dusk with a clear semicircular bar anteriorly in specimens of 25 mm SL or less; and hyaline in larger specimens.

Coloration in life.
Body yellow to brownish.Patterns of dark bars otherwise as described for preserved specimens.Dorsal margin of maxilla dark, with coloration extending posteriorly to vertical through posterior margin of outer fleshy lamellae of lower lip.Coloration faintly connecting dorsally to the dark bar through nares with only slight lightening in color of space between.Lips generally yellow with darkening of dorsal edge of upper lip and dark black on both sides of upper lip.

Geographical distribution.
Leporinus villasboasorum occurs throughout the middle and upper Xingu drainage, from Volta Grande rapids near Belo Monte, downstream of the city of Altamira in Pará State to the headwaters of the Xingu River in Mato Grosso State.It also occurs in the Teles Pires and Juruena rivers, which are tributaries of the upper Tapajós River, in Mato Grosso State, Brazil (Fig. 9).Conservation status.Leporinus villasboasorum is relatively frequent and abundant in the Xingu (520,300 km 2 ) and Tapajós (489, 628 km 2 ) river basins.It is widespread throughout the both basins and the species can be categorized as Least Concern (LC) according to the International Union for Conservation of Nature (IUCN) categories and criteria (IUCN Standards and Petitions Subcommittee, 2016), as the species occurs in several areas of preservation priority, such as indigenous parks.
Remarks.Leporinus villasboasorum occurs in the middle and upper Xingu and in the upper Tapajós basins.Since the Xingu and Tapajós basins are nowadays not connected it is possible that future more detailed studies will recognize the populations of those two basins as not conspecific as considered here.To prevent a polytypic type series in the advent of such a split, only specimens from the Xingu basin were included as paratypes of L. villasboasorum.Scales almost uniformly large with only slightly smaller scales in predorsal and ventral regions.Lateral line well defined, complete and straight with simple well-defined tubes.Pointed sheath scale twice as long as fin base on pelvic, starting at pelvic-fin origin and pointing posteriorly.Dorsal-fin origin located about halfway between snout tip and adipose-fin origin.Pelvic-fin insertion located at vertical through origin of third branched dorsal-fin ray.Adipose-fin origin located at vertical through origin of third branched anal-fin ray.Distal end of posterior branched anal-fin ray located at vertical through caudal peduncle.All fins sharply falcate.First unbranched fin ray about half as long as first branched fin ray, with first branched fin ray being the longest, on both dorsal-and anal-fins.Caudalfin deeply forked.Vent located anteriorly of anal-fin origin about one-third the distance between anal-fin origin and pelvic-fin insertion.

Coloration in alcohol.
Pale white to brownish with nine well-defined dark black transverse bars which, on bars posterior to the dorsal-fin origin, meet dorsally and ventrally to form bands encircling head and trunk.First bar at snout, second at eye, third at opercle, and remaining six on trunk.Bars on main body sloping slightly toward posterior from dorsal to ventral.Head with one dark bar on e160166 [16] snout, starting just posterior of origin of anterior tubular nostril and extending to just anterior of eye, with faint connections to the second dark bar extending at a slight posterior slope above and below posterior nostril.Head with second dark bar on dorsal and lateral surfaces of head, located on interorbital and infraorbital region, about twothirds width of eye and centered on eye, with no connections to third dark bar.Trunk with seven dark bars.First dark bar on trunk encompasses operculum but does not extend onto branchiostegals or isthmus, with black only occurring in space between the pectoral-fins ventrally.Second dark bar is located anterior of dorsal-fin origin and is often distinctly forked below the midway point between lateral line and ventral surface of body.Occasionally second dark bar slightly forked above lateral line at midway point between lateral line and dorsal surface of the body.Third dark bar includes entire dorsal base, anterior dorsal-fin rays and entire pelvic-fin except first unbranched fin ray.Fourth dark bar begins just posterior to dorsal-fin insertion and lies midway between pelvic insertion and anal origin and does not include vent.Fifth dark bar lies anterior of adipose-fin and extends ventrally to include first three rays of anal-fin.Sixth dark band begins at adipose insertion and extends to include caudal peduncle and posterior half of anal-fin as well as the distal two-thirds of adipose.Seventh dark bar includes caudal base with no intrusion of color onto caudal-fin rays.
Lips generally whitish with faint darkening of dorsal margin of upper lip and dark black on both sides of upper lip.Fleshy lobes on both sides of lower lip dark black.Dark pigmentation on snout tip and dorsal margin of maxilla, extending posteriorly to vertical through posterior margin of outer fleshy lamellae of lower lip.Pigmentation on upper lip connects faintly to dark bar through nares with only slight lightening in color of space between.Iris greyblack.Fins whitish except where noted.Ventral portion of head whitish.

Coloration in life.
Pale yellow to brownish dorsally.Coloration and pattern of bars otherwise similar to that described in preservative.Lips dark black with ventral portion of lower lip appearing lighter in color.Fins generally light in color with dark pigmentation occurring near bars on the body.

Geographical distribution.
Leporinus enyae occurs throughout the southern tributaries of the Orinoco river drainage in Venezuela, including the Ventuari, Parcucito, Atabapo, and Caura rivers, and in the upper Negro River in Brazil and Venezuela, and in the Casiquiare canal, which connects the Negro and Orinoco rivers (Fig. 9).
Etymology.Named in honor of the singer Enya, whose beautiful song "Orinoco Flow" celebrates the flow of the mighty Orinoco River, which the new species inhabits.A noun in the genitive case.
Conservation status.Leporinus enyae is relatively frequent and abundant in the Orinoco River basin.It is widespread throughout the 880,000 km 2 of the Orinoco River basin and the species can be categorized as Least Concern (LC) according to the International Union for Conservation of Nature (IUCN) categories and criteria (IUCN Standards and Petitions Subcommittee, 2016).

Discussion
Taxonomic implications of circumpeduncular scale counts.Circumpeduncular scale counts are among the most frequently used diagnostic characteristics separating anostomid species.That ubiquity stems from the largescale size of most anostomids, which in turn fixes the circumpeduncular scale counts in most species.Almost all known species of Leporinus possess a fixed count of either 12 or 16 circumpeduncular scale rows, excluding aberrant or deformed individuals.The only exceptions are L. jatuncochi, with 14 as discussed herein, and some specimens of L. taeniatus that also have 14 (Birindelli et al., 2016).Therefore, the circumpeduncular polymorphism within the Essequibo, Tocantins and Araguaia populations of Leporinus desmotes (in which counts of 14 and 16 circumpeduncular scales are both frequent), is highly unusual for a Leporinus species.
Given the rarity of such polymorphism, one must consider whether the presence of 14-and 16-scaled individuals in the Essequibo and lower Amazon tributaries indicates that these regions harbor not one, but two distinct species in the Leporinus desmotes group, with the paratype series of L. desmotes including individuals of both.If two species do exist in the Essequibo, the name Leporinus desmotes would apply to the 16-scaled species, because the holotype possesses that count.The 14-scaled species would then either receive a new name, or fall into the synonymy of Leporinus jatuncochi from the upper Amazon, specimens of which universally possess 14 circumpeduncular scales.The morphometric separation between 14 and 16 scaled individuals (Fig. 4) also hints at the presence of two species in the Essequibo, though increased sample sizes may blur that distinction.The Tocantins specimens are remarkably similar to the Essequibo specimens in body coloration (Fig. 11), and possess the same circumpeduncular polymorphism.Thus, it may also be that only a single species in this complex (Leporinus desmotes sensu stricto) occupies the Essequibo and Tocantins.If so, the circumpenducular polymorphism would be an emergent characteristic helping to characterize the distinctiveness of that species, as no other species in the barred Leporinus group possesses such a polymorphism.With only one genetic sample of L. desmotes currently sequenced (and with that specimen, somewhat frustratingly, being the only known specimen with 15 circumpeduncular scales) it is impossible to test the conspecificity of the two meristic morphotypes, nor can we assess the distinctiveness of the population in the Essequibo versus that in the lower Amazon tributaries.On the status of Leporinus jatuncochi.Despite recognizing two new species in the Leporinus desmotes species complex our analysis was not fully able to resolve the taxonomic status of the upper Amazonian specimens assignable to L. jatuncochi.The mean body shape of these specimens differs only slightly from the Essequibo specimens assignable to L. desmotes Tab. 5), and the genetic separation between the three sequenced upper Amazonian specimens and the single sequenced Essequibo specimens is much less pronounced than the difference between either of these groups and the sequenced specimens of L. enyae or L. villasboasorum.Furthermore, specimens of L. jatuncochi are meristically identical to the portion of L. desmotes with 14 circumpeduncular scales.Ultimately, the taxonomic status of L. jatuncochi depends upon the resolution of the circumpeduncular conundrum described above.If all Essequibo and lower Amazonian specimens of L. desmotes are conspecific, then L. jatuncochi might be best considered a junior synonym (and geographically distinct population) of L. desmotes.However, if the circumpeduncular polymorphism with L. desmotes indicates the presence of two species in that drainage, L. jatuncochi might be a valid name, and potentially (but not necessarily) applicable to the 14-scaled species that occurs in the Essequibo.Given the overall low genetic sample sizes currently available, we prefer to place the question of the validity of L. jatuncochi into abeyance until increased genetic sampling allows a rigorous test of the possible presence of two species within the present concept of L. desmotes in the Essequibo, Araguaia and Tocantins, and until the levels of genetic divergence between the upper and lower Amazonian populations can be more fully characterized.
Geographic limits to species distributions.As herein delimited, Leporinus desmotes possesses two distinct populations, one from the Essequibo and another from the Tocantins, apparently separated by the main channel of the Amazon River and its left-margin's tributaries.The distribution of L. villasboasorum and L. jatuncochi provides another likely example of a species boundary formed by the Volta Grande rapids, as also shown by Doras higuchii (Sabaj Pérez, Birindelli, 2008).In the Xingu, L. villasboasorum is apparently restricted to the stretch above the great rapids of Volta Grande, near the city of Altamira, where Belo Monte dam was recently constructed.Specimens collected below the Volta Grande rapids belong to L. jatuncochi.That said, no obvious biogeographic barrier separates the distribution of these species in the Tapajós basin.L. villasboasorum occurs in the upper Tapajós River basin, including the Teles Pires and Juruena rivers, but is absent in the middle and lower Tapajós basin, including the Jamaxim River, where L. jatuncochi occurs.
New species of Leporinus continue to be described yearly (Birindelli, Britski, 2013;Birindelli et al., 2016), but as our analysis has shown, many more undescribed species still exist across South America.The uncertain delimitation of even well-known species, such as L. desmotes and L. jatuncochi, stresses the importance of combining morphological and genetic approaches in the description and revision of species.An in-depth phylogenetic and biogeographic analysis of barred Leporinus species is still needed to fully understand the species diversity in this highly diverse genus.(MHNG), Paulo Buckup and Marcelo Britto (MNRJ); Wolmar Wosiacki, Alberto Akama and André Netto-Ferreira (MPEG), Mary Burridge, Erling Holm and Hernán López-Fernández (ROM) and Larry Page (UF).Mark Sabaj and Barbara Lundrigen prepared photographs of the holotypes of Leporinus desmotes and L. jatuncochi, respectively.Mark Sabaj Pérez and Leandro Sousa provided live photographs of live specimens.Tyler Lyman assisted with the early stages of morphological specimen examination.NSF grant DEB -1257898 provided support to BLS and MDB.Fundação Araucária (research grant 641/2014) and CNPq (research projects 478900/2013-9 and 420255/2016-8) provided support to JLOB.Specimens were collected during expeditions funded by the South American Characiformes Inventory (FAPESP 2011/50282-7, http://www.usp.br/peixes), the All Catfish Species Inventory (NSF DEB-0315963), the iXingu Project (NSF DEB-1257813), the National Geographic Committee for Research and Exploration, and Aquatic Critter Inc.A visit to the MCP fish collection by JLOB was funded by "Projeto qualificação taxonômica da base de dados da coleção de peixes do MCT-PUCRS para uso on-line" (MCT/CNPq no 45/2012, 504177/2012-5).

Fig. 5 .
Fig. 5. Phylogenetic relationships within the Leporinus desmotes species complex showing the ML tree.Node values represent bootstrap support and posterior probabilities, respectively.

Fig. 7 .
Fig. 7. Rio Culuene, tributary of Xingu river in Brazil, at the type locality of Leporinus villasboasorum.The river stretch herein illustrated is currently dried by the diversion of the river channel due to the construction of the Paranatinga 2 hydroelectric dam.

Fig. 8 .
Fig. 8. Regressions of body depth and the distance between the pelvic-fin and adipose-fin origins against SL for Leporinus desmotes (blue) and L. villasboasorum (brown) showing different slopes.
Named in honor of Orlando, Cláudio and Leonardo Villas-Bôas, in recognition of their pioneering efforts to conserve and protect the rio Xingu's marvelous biodiversity, of which Leporinus villasboasorum forms part.A noun in the genitive case.
Linear morphometrics for all examined species summarized as observed ranges, with mean ± standard deviation in parentheses.Measures 1 to 29 are percentages of SL, measures 30-37 are percentages of head length.

.
Leporinus enyae can be separated from all other members of Leporinus except Leporinus desmotes, L. jatuncochi, and L. villasboasorum by possession of long, pointed, laterally compressed and upward curving symphyseal dentary teeth, versus the more truncate conical or incisiform teeth typical of other members of Leporinus, and from Leporinus bleheri, L. desmotes, L. jatuncochi, L. villasboasorum, and L. yophorus by a color pattern consisting of nine dark bars on the head and body, some of which fork dorsally or ventrally, the fifth of which continues onto the dorsal and pelvic fins, versus various other color patterns in the approximately 90 other described species of Leporinus.Holotype and paratype meristic values in Tab. 3, morphometric values in Tab. 6. Body fusiform, elongate, slightly compressed, deepest and widest at dorsal origin.Very slight keel between dorsal insertion and supraoccipital crest, all other margins of body gently rounded.Caudal peduncle compressed and slightly longer than its depth.Head conical, compressed slightly dorsally and more so ventrally.Lateral profile of head straight and inclined.Anterior profile of head forming rounded point.Head broadest at posterior margin of opercle part of head.Eye round and laterally placed.Anterior nostril tubular, flared at distal end, positioned just anterior of first black transverse bar and overhanging upper lip slightly.Posterior nostril a sharply upturned slit positioned slightly posterior of anterior nostril at angle paralleling lateral profile of the head.Mouth small and subterminal.Lips fleshy and thick with deeply-cleft lamellae.Upper lip with two series of lamellae of which outer longer than inner.Bottom lip with two or more series of lamellae, outermost terminating on either side with deep-cleft narrow fleshy lobe.Inner series irregular in pattern and much shorter than series on upper lip or of outer series of bottom lip.Teeth three on each premaxilla, all broader than dentary teeth and slightly spoon shaped, pointed and jutting slightly anteriorly, with lateral teeth on each side smaller than medial teeth.Teeth three on each dentary with lateral teeth much smaller than symphyseal.Symphyseal dentary teeth long, pointed, laterally compressed and upward curving.Gill openings restricted.Opercle and preopercle smooth, with opercular flap pronounced and extending posteriorly from operculum.Four branchiostegal rays on each side.
It separates most easily from Leporinus jatuncochi and some specimens of L. desmotes by the possession of 16, versus 14 circumpeduncular scales.It separates from Leporinus villasboasorum and specimens of Leporinus desmotes with 16 circumpeduncular scales by having the dark bar immediately anterior of the dorsal fin undivided or barely divided dorsally, versus well-divided dorsally.Description.