Phylogeography and conservation genetics of the Amazonian freshwater stingray Paratrygon aiereba Müller & Henle , 1841 ( Chondrichthyes : Potamotrygonidae )

The family Potamotrygonidae is monophyletic comprising three genera: Paratrygon Duméril, Potamotrygon Garman and Plesiotrygon Rosa, Castello & Thorson. The distribution of most species in this family is restricted to a single basin or fluvial system. Only Potamotrygon motoro, Potamotrygon orbignyi and Paratrygon aiereba are found in more than one river basin. In this study we investigate genetic structuring of Paratrygon aiereba, from five rivers of the Amazon region: Negro, SolimõesAmazon-Estuary system, Tapajós, Xingu and Araguaia. Sixty-three individuals were sequenced for ATPase 6, and a representative subsample of 27 individuals was sequenced for COI. The COI dataset analysis indicated that Paratrygon is sister to all other potamotrygonid genera and species. Population parameters inferred from the analysis of ATPase 6 sequences revealed that the populations of this species are structured within each river, with no or nearly non-existent gene flow occurring between rivers and a positive correlation between geographic and genetic distances. Paratrygon aiereba is comprised of three geographically restricted clades with K2P interclade distances of at least 2%. Intraspecific divergence within P. aiereba is similar to the interspecific divergence observed in Potamotrygon spp. sampled throughout the same geographic area. Using the premises of COI barcoding and the allopatric distribution of the three P. aiereba clades, the taxon P. aiereba most likely comprises three distinct biological species. Since freshwater stingrays of the family Potamotrygonidae are highly exploited for the aquarium trade, management and conservation strategies need to be implemented at the level of each river basin, rather than at the level of the Amazon basin.


Introduction
Phylogeography deals with principles and processes governing the geographic distributions of genealogical lineages (Avise, 2000).These analyses are aimed at investigating patterns of geographic distribution of taxa and understanding processes that have resulted in these patterns (Bermingham & Moritz, 1998).The Neotropical region has been the object of a large number of biogeographic studies on fishes.The complex geomorphological history of this region is reflected in its ichthyofauna, which provides a rich source of material for the study of the Neotropical biogeography (Vari & Malabarba, 1998).Many studies have been carried out with the aim of discovering these biogeographic patterns and interpreting these patterns in light of geomorphological processes such as the formation of hydrographic basins.Examples include studies of the genera Brachyhypopomus, Pimelodella, and Roeboides in Central America (Bermingham & Martin, 1998), fishes of the Neotropical family Rivulidae (Hrbek & Larson, 1999), Potamorraphis in the river basins of the Amazon and Orinoco (Lovejoy & de Araújo, 2000), Prochilodus in the Amazon, Orinoco, and Paraná River basins (Sivasundar et al., 2001), Leporinus in the Paraná River Basin (Martins et al., 2003), Hypostomus in the main river basins of the Neotropics (Montoya-Burgos, 2003), Brachyplatystoma in the Solimões-Amazon-Estuary system (Batista et al., 2004;Batista & Alves-Gomes, 2006), Arapaima gigas in the Amazon River basin (Hrbek et al., 2005), Cichla in the Orinoco and Amazon River basins (Willis et al., 2007), Symphysodon in the Amazon River basin (Farias & Hrbek, 2008) and fishes of the order Characiformes in the main river basins of South America (Hubert & Renno, 2006).
The biotic diversification of aquatic systems is directly related to the geomorphological history of rivers basins.Inter-basin connections, alterations of drainage systems, and capture of rivers and basins promoted vicariance, and allopatric divergence.Thus, the geomorphological history of river and river basin formation provides a good model for studies of diversification of aquatic fauna (Bermingham & Moritz, 1998;Lundberg et al., 1998;Montoya-Burgos, 2003).
According to Lundberg et al. (1998), a significant portion of the diversification of Neotropical freshwater fishes and other aquatic organisms occurred due to the formation of South American rivers and river basins at the end of the Cretaceous and the Cenozoic.One of the most important events in the diversification of ichthyofauna during this period involved marine incursions onto the continent.These marine incursions led to origin of freshwater fishes derived from marine groups such as anchovies, needlefish and freshwater stingrays (Lovejoy et al., 1998).
Freshwater stingrays are endemic to the Neotropical region and belong to the family Potamotrygonidae (Carvalho et al., 2003).This group is monophyletic and comprises three genera: Paratrygon Duméril (basal genus), Potamotrygon German and Plesiotrygon Rosa, Castello & Thorson (sister genera).These groups have 18 valid species, 15 of which are found in Brazil.Among these species, only Potamotrygon motoro, Potamotrygon orbignyi, and Paratrygon aiereba have wide distributions and are found in more than one hydrological basin.The group is found in diverse habitats, including beaches, flooded forests with a rocky, clay bottom and lakes (Carvalho et al., 2003).
Although an exclusively freshwater group, these taxa share some biological characteristics with marine elasmobranches, such as low fecundity, late sexual maturation and slow growth rate (Araújo et al., 2004).Such characteristics make the group vulnerable to commercial-scale exploitation.In the State of Amazonas, the family Potamotrygonidae is exploited by ornamental fisheries and is highly prized by the aquarium trade throughout the world.In the State of Pará, this family is also used as a food source, with the targeting of species that reach larger sizes, such as Paratrygon aiereba (Araújo et al., 2001;Carvalho et al., 2003).
Paratrygon aiereba is found in all types of aquatic environments of the Amazon basin (Carvalho et al., 2003).This distribution is probably due the ability to adjust to the diverse ionic characteristics of Amazon River and its tributaries (Duncan et al., 2009).The wide geographic range allied with low fecundity makes this species vulnerable to different environmental risks such as persecution, direct and indirect fisheries and habitat degradation (Araújo et al., 2004;Martin, 2005).Because of their wide distribution and potentially vulnerable status, understanding how populations are structured has a high value for conservation.
The aims of the present study were to employ phylogeographic and population genetic analyses to obtain information on the distribution of genetic diversity of Paratrygon aiereba; to determine whether there is genetic structuring and whether such structuring is explained by river basins; and to discuss how these results can be used for the conservation of this organism.
DNA extraction and sequencing.Muscle tissue was collected and conserved in absolute alcohol and frozen at -20°C until DNA extraction.Tissue was dissolved using a SDS/proteinase K solution, and total genomic DNA was isolated using a standard phenol-chloroform extraction and ethanol precipitation (Sambrook et al., 1989).
Two gene fragments of mitochondrial DNA were selected for the present study: cytochrome oxidase I (COI) and ATPase 6. Partial sequences from COI gene were used to estimate genetic divergences, and to identify putative lineages or species within the P. aiereba, according to the DNA barcode guidelines.Other species of Potamotrygonidae and two outgroup species (Himantura pacifica, Himantura schmardae) available in the GenBank (data from Toffoli et al., 2008) were also included in analyses.
The partial sequence of ATPase 6 was used for population genetic analyses, as well as to estimate intra-specific divergences and reconstruct phylogenetic hypothesis for the genus Paratrygon.Sequences of ATPase 6 of Potamotrygon and Plesiotrygon were used as outgroup.
The primers COIf.1 (5´-CTTAACACAACWTTCTTTGACCC-3´) and COIr.3 (5´-ACGTTTTGATGCRAAKGCYTCTC-3´) were kindly provided by Tomas Hrbek and used for the amplification of mitochondrial COI, using the same PCR conditions as for ATPase, but changing the annealing temperature to 53ºC.These primers amplify the COI region spanned by the primers COIf and COIa reported in Palumbi (1996); this region corresponds to the 3' half of the COI gene, while the standard COI Barcoding region corresponds to the 5' half of the COI gene.
The BigDye Cycle Sequencing kit (Applied Biosystems) was used for the COI sequences and the DYEnamic ET Terminator kit (GE-Healthcare) was used for the ATPase 6 sequences, following the manufacturers' instructions.Sequencing was preformed on the ABI 3130 x l automatic sequencer (Applied Biosystems) and MEGAbase 1000 (GE-Healthcare).
Data analysis.The sequences were manually edited and aligned with the aid of the BioEdit program (Hall, 1999).The most appropriate model of molecular evolution for each gene was selected using ModelTest (Posada & Crandall, 1998).
The haplotype tree phylogeny for COI sequences were constructed in Treefinder (Jobb et al., 2004) under the Maximum Likelihood (ML) model of molecular evolution with parameters selected in the program ModelTest 3.7 (Posada, 2004), and neighbor-joining (NJ) method assuming the Kimura-twoparameter model of molecular evolution (barcoding standard - Hebert et al., 2003).The NJ tree topology and statistical robustness (using 2000 bootstrap replicates) were estimated in PAUP* 4.0b10 (Swofford, 2002).Sequence data of other potamotrygonid species available in the GenBank (data from Toffoli et al., 2008), were included in the analyses to calculate genetic divergence between genera and for the identification of putative lineages or species within the P. aiereba, according to the DNA barcode guidelines.The same two shark species used om Toffoli et al. (2008) were used as outgroups.
A haplotype tree of ATPase 6 was constructed in the program Treefinder (Jobb et al., 2004) under the Maximum Likelihood (ML).The most appropriate molecular model was HKY+G as inferred in the program ModelTest 3.7 (Posada, 2004).
Based on the results obtained from the NJ and ML analyzes and delimiting clades by river systems, we subdivided the data into the following groups for further population-level analyses: the Solimões-Amazonas-Estuary system, Negro, Araguaia, Tocantins, and Xingu rivers.Due to the small number of specimens from the Tapajós River (N = 2), these individuals were excluded from the population level analyses.Molecular population level analyses such as nucleotide diversity (S), haplotype diversity (H), and number of polymorphic sites were performed using the program Arlequin 3.11 (Excoffier et al., 2005).Tajima's D (Tajima, 1989) and Fu's Fs (Fu, 1997) neutrality tests were used to determine whether the samples from the different locations were at mutation-migration-drift equilibrium.
Population subdivision and structure were examined using analysis of molecular variance (AMOVA) (Excoffier et al., 1992) and the pair-wise ) ST , which is analogous to F ST (Weir & Cockerham, 1984), both implemented in Arlequin 3.11 (Excoffier et al., 2005).Significance of the correlation between the matrix of genetic distance and geographic distance between P. aiereba localities was tested by the Mantel test (Mantel, 1967) implemented in Arlequin 3.11 (Excoffier et al., 2005).All tests that required multiple comparisons were corrected using the Bonferroni procedure (Rice, 1989).

Results
Cytochrome Oxidase I -COI.Considering that we were interested to check for inter and intra group divergences, we did not amplify all the individuals of P. aiereba for the COI gene.nucleotide substitutions found in the COI sequences was Hasegawa-Kishino-Yano.This model uses a variable nucleotide frequency and variable transition and transversion frequencies (Hasegawa et al., 1985), with the following parameters: TS/TV ratio = 7.7586; and nucleotide frequency of A=0.2775, T=0.2817, C=0.2753 and G=0.1655.The topologies and support values generated in the NJ (Fig. 2) and ML analyses (not shown) were very similar.The lineage formed by the genus Paratrygon was monophyletic and sister to the genera Potamotrygon and Plesiotrygon.The genetic distances between Paratrygon and the other genera and species of Potamotrygon based in Kimura-two-parameter model of evolution was very high, varying from 13% to 22% (Table 2).The inter-locality Paratrygon genetic distances varied from 0.37% (between SAE -NEG) to 4.9% (between SAE -ARA), while intra-locality genetic distances raged from 0.17% to 0.26%.
Within the genus Paratrygon, the group comprised of individuals collected from the Solimões-Amazon-Estuary system (SAE) and Negro River (NEG) formed a monophyletic group with a moderately high support value.The SAE+NEG clade was sister to individuals collected from Xingu River (XIN), with high support values.The clade formed by individuals collected from Araguaia River (ARA) was sister to the SAE+NEG+XIN clade (Fig. 2).
The genetic distances between Paratrygon aireba from the SAE system and NEG were low (Table 2).However, the divergence between XIN -SAE and XIN -NEG was 2.4% and 2.1%, respectively.The divergence between ARA -SAE and ARA -NEG was higher at 4.9% and 4.7%, respectively.Additionally, genetic distances within the genus Paratrygon were larger than those found among species of the rosette-spot group of Potamotrygon (cf.Toffoli et al., 2008) comprises the following species: Potamotrygon motoro, P. scobina, P. orbignyi, P. leopoldi, P. falkneri, and P. henlei.Within Paratrygon, intraclade divergences were less than 1%, as were the inter-specific divergences observed in the rosette-spot group of Potamotrygon (Table 2) (Toffoli et al., 2008).ATPase 6.Among the 584 bp sequenced for 63 individuals of P. aiereba considered for the analysis, 498 characters were constant and 86 were variable.The model that best explained the patterns of nucleotide substitutions found in the ATPase sequences was Hasegawa-Kishino-Yano+Gamma (HKY+G) with the following parameters: gamma=0.35,TS/TV ratio = 0.3006; and nucleotide frequency of A=0.30900, T=0.25140, C=0.34030 and G=0.09930.These parameters were used for the Maximum-Likelihood (ML) analysis (Fig. 3).
The haplotypes ML phylogenetic tree shows the same group observed in the COI tree, formed by SAE and NEG group, but with a low support value.Another group was formed by the TAP, XIN and ARA groups.A small group was formed by one individual from ARA and one from SAE showing no support values.
polymorphic sites (S) and greatest nucleotide diversity were found in the group from the Araguaia River.The greatest gene diversity values were found in the groups from the Solimões-Amazon-Estuary system, Negro, and Araguaia Rivers.Xingu River had the lowest DNA polymorphism values.Tajima's D was significant for the Solimões-Amazon system and Negro River populations, the latter of which also had significant Fu's Fs values (Table 3).
The AMOVA results revealed strong genetic structuring () ST = 0.71697, P< 0.001), with 71.69% variation between rivers and 28.30% variation within rivers.Pairwise ) ST analyses also indicated strong differentiation among localities, and lack of exchange of individuals (Table 4).

Discussion
Barcoding.In order to facilitate a more rapid identification of species, the use of the barcoding, which promises the discrimination of species through the analysis of a small segment of the mitochondrial COI gene, is an innovative and efficient approach for the characterization of biological diversity (Hebert et al., 2003).A growing number of studies have demonstrated that interspecific genetic distances observed between the COI mtDNA fragments is generally higher than 2%, which allows the differentiation of species that are more closely related and enables identification with a high degree of confidence (http://barcoding.si.edu).
The taxonomy and identification of species from the family Potamotrygonidae is difficult and new species are being described (Carvalho et al., 2003;Carvalho & Lovejoy, 2011).The genus Paratrygon is monotypic (Rosa & Carvalho, 2007).However, due to its wide distribution, which ranges from the Orinoco River basin to the Amazon River basin, number of researchers have suggested that it a species complex (Charvett- Almeida, pers. comm.).The genetic analyses performed in this study are lending support to these conclusions.The COI fragment separated this species into three large groups: (1) the group from the Solimões-Amazon-Estuary system (SAE) and the Negro River (NEG); and (2) the group from the Xingu (XIN) River; (3) other group from Araguaia (ARA) River.The greatest genetic distance was observed between individuals from Araguaia River and the other localities with values above 4% of sequence divergence.The Xingu locality showed more than 2% sequence divergence from all other localities.These divergences are compatible with interspecific divergences under the molecular barcoding criteria (www.barcodinglife.org).
According to Moritz & Cicero (2004), a diagnosis with good accuracy depends on lower intra-specific variation in relation to inter-specific variation, the so called barcoding gap.Pairwise distances found between the SAE and NEG groups in relation to Xingu (~2%) and Araguaia (~4%) clades were roughly 10-fold greater than distances within each clade.According to Hebert et al. (2003), a threshold divergence 10fold greater between clades than within each clade is normally found between different species.However, diagnosability of potamotrygonid species through DNA barcodes is not always efficient.Species of the Potamotrygon rosette-spot group composed of Potamotrygon motoro, P. scobina, P. orbignyi, P. leopoldi, P. falkneri, and P. henlei, share haplotypes and thus are not monophyletic (Toffoli et al., 2008).Haplotype sharing in this group could be due to incomplete lineage sorting in this recently diversified group, or due to introgressive hybridization.Intraspecific divergences are as large as interspecific divergences, and also exceed the suggested 2% divergence threshold (Toffoli et al., 2008), limiting the value of DNA barcoding for the Potamotrygon rosette-spot group.
Although barcoding has not functioned well for rosettespot group of Potamotrygon (Toffoli et al., 2008), the identification system based on the COI fragment may be an initial step toward the delineation of species (Hebert et al., 2003).In P. aiereba, the sister group of Potamotrygon + Plesiotrygon, COI has proven to be a good tool for the determination of evolutionarily significant units, and even potential new species, as the geographically restricted clades have lower within-clade genetic divergence than between clades, which is one of the premises of diagnostic accuracy based on the barcoding system (Hebert et al., 2003).Biogeography.The results from the present study of the ATPase 6 reveal a separation into two large groups: one formed by the Solimões-Amazon-Estuary system and Negro River; and the other formed by the Tapajós, Xingu, and Araguaia Rivers, with the exclusion of the small group formed of one individual from SAE and one from ARA.These results match the pattern of distribution found by Hubert & Renno (2006), who analyzed 345 species of Characiformes using parsimony analysis of endemicity and also found a separation into two large groups: (1) one denominated "Lower Amazon", corresponding to the Amazon, Negro, lower Madeira, and Branco Rivers; and (2) the other denominated "Xingu-Tocantins", corresponding to the Tocantins, Araguaia, Xingu, and Tapajós Rivers.
The Xingu, Tapajós, and Araguaia Rivers have their origin on the Central Brazilian Shield.The individuals collected from the Tapajós River, near the city of Itaituba, Pará State, and in the Xingu River, above Belo Monte, Pará, occur at the limit between the sedimentary basin and the Brazilian Shield, where large waterfalls are common, which may serve as a geographic barrier for this species, thereby playing an important role in the isolation of these populations.
The geomorphological events that resulted in the establishment of current river basins may have been agents of vicariant diversification, fragmenting and isolating these populations.Thus, one of the hypotheses that could explain the genetic pattern found in the present study is allopatric fragmentation and diversification of the P. aiereba populations.A number of biogeographic studies (Lovejoy et al., 1998;Albert et al., 2006;Hubert & Renno, 2006) have shown that the events of the Miocene period are the most important for the modeling of the distribution and evolution of Neotropical ichthyofauna.
According to Wesselingh & Salo (2006), the Amazon basin began to be delineated in their current configuration as a consequence of the emergence of the Andes, together with a variation in sea level due to the glaciers of the Quaternary, which carved out the rivers into what is more or less the current definition of the valleys.A number of authors agree with this formation process of the Amazon Basin, but the period in which it occurred is not yet well defined and ranges from 8 and 2.5 mya (Lundberg et al., 1998;Campbell Jr. et al., 2006;Wesselingh & Salo, 2006).
The family Potamotrygonidae is monophyletic, originating from a marine ancestor that colonized South America (Lovejoy et al., 1998;Marques, 2000;Carvalho et al., 2003;Carvalho et al., 2004).The timing of the colonization remains tentative, ranging from 10 to 50 mya (Lovejoy et al., 1998;Carvalho et al., 2004).However, it uncontroversially precedes the final formation of the Amazon basin drainage system, and thus vicariance driven diversification of the potamotrygonid stingrays is a likely hypothesis.

Implications for conservation.
As the genetic diversity of species represents the range of evolutionary and ecological adaptations in relation to a given environment, phylogeographic studies can generate useful information for the conservation of organisms (Avise, 2000).Based on the genetic divergence (see www.barcodinglife.org) of the populations from the Xingu, Araguaia, and the Solimões-Amazonas-Estuary system, the present study indicates that there is more than one species within what currently is considered P. aiereba.If these lineages represent different species, management and conservation policies will need to be modified to reflect this.
In the Brazilian State of Amazonas, freshwater rays are exploited by the ornamental fish trade for export to Europe, Japan, and the United States (Araújo et al., 2004;Moreau & Coomes, 2007).In other regions of the Amazon, this group is also exploited as a food source throughout the Solimões-Amazon system, however in a small scale compared with the commercialization for the ornamental fish trade.Freshwater stingrays have also suffered from large-scale commercial fishing, in which they are caught as bycatch in gillnets (Araújo et al., 2004).
In Brazil, at the end of the 1990s, the fishing of freshwater stingrays was banned by the Brazilian Environmental Agency (Instituto Brasileiro de Meio Ambiente e Recursos Naturais Renováveis -IBAMA (Araújo et al., 2004).Since then, this activity has been permitted again and once again banned.Currently, the activity is permitted through the Normative Instruction n o 204/2008, which allows the capture of only six species from this family, with annual quotas, in the Amazon and Araguaia-Tocantins River basins in the states of Pará and Amazonas, (IBAMA, 2008).However, this norm is currently under review by IBAMA.Paratrygon aiereba is not among the six species that can be commercialized and its capture is prohibited.In Brazil, P. aiereba has a vulnerable status and it is prohibited to be exported as ornamental fish, but has been exported by Peru and Colombia in small quantities in the last fifteen years.
In other regions of South America, this species has been exploited by direct and indirect fisheries (Araújo et al., 2004;Barbarino & Lasso, 2005).However, other factors contribute to the vulnerability of this species: a low rate of population increment (low fertility) (Barbarino & Lasso, 2005;Charvet-Almeida et al., 2005), low abundance rates (Almeida et al., 2009); anthropogenic action in the environment (damns,  being unlawfully exported due to the lack of adequate infrastructure for fiscal policies in the region. Using DNA barcoding one could identify putative illegal stingrays catches and exports.Although DNA barcoding cannot discriminate between the species of the Potamotrygon rosette-spot group, it can differentiate mines, dragging), persecution (Araújo et al., 2004;Martin, 2005), and climate changes which affect the events of reproductive cycle.Further, with the enormous area to be covered, protective policies in the Amazon are very difficult to be implemented and reinforced.A large number of protected fish species, including the stingrays, are currently unambiguously between genera, as well as many species within the genus Potamotrygon.In the case of Paratrygon aiereba it can also differentiate between the geographic origins of the specimens.Considering that DNA barcoding is cheap and fast, it can be used to effectively implement conservation programs for freshwater stingrays by verifying that what is being exported or which areas are being fished, is in reality what is permitted to be sold.

Fig 1 .
Fig 1. Map with the sample locations.In some localities more than one individual was sampled.

Fig 2 .
Fig 2. Neighbor Joining (NJ) tree for partial COI gene of Paratrygon aiereba combined with data from Toffoli et al. (2008).Numbers above the line are bootstrap support values.

Fig 3 .
Fig 3. Maximum likelihood (ML) tree based on the analysis of the ATPase 6 gene of the Paratrygon aiereba.Numbers above the line are maximum likelihood edge support values.

Table 2 .
The greatest number of

Table 1 .
Vouchers of tissue sampling in Inpa collection.
We randomly chose few individuals representative from each group.Among the 610 bp sequenced for the COI gene in 27 specimens of P. aiereba, 570 sites were monomorphic and 40 variable.The model that best explained the patterns of

Table 2 .
Genetic distance using COI sequence data based in Kimura-two-parameter model of evolution.Pair-wise comparison between P. aiereba and other species/group from data of