DNA barcoding reveals hidden diversity in the Neotropical freshwater fish Piabina argentea (Characiformes: Characidae) from the Upper Paraná Basin of Brazil

Abstract

Background and aims. We analyzed a small and wide geographically distributed Neotropical freshwater fish, the Piabina argentea from the Upper Paraná Basin, to check the hypothesis that this species is composed of more than one biological unit, since it has a limited dispersion, through the DNA barcode technique. Materials and methods. Partial mitochondrial COI and CytB gene sequences were obtained for 58 specimens drawn from 13 localities. Results. Phylogenetic analysis revealed six major clusters of P. argentea. Kimura-two-parameter (K2P) genetic divergences among these six P. argentea clusters ranged from 2 to 5.6% and from 2.3 to 5.4% for COI and CytB genes, respectively, and these values were on average approximately nine times greater than intra-cluster K2P divergences. The fixation index (F_ST) among clusters showed very high values and the haplotype network analysis displayed seven unconnected units. Conclusion. These results reinforce the hypothesis that the widely distributed P. argentea species concept as currently conceived actually represents more than one species (possibly six). These results demonstrate the efficacy of DNA barcoding for the discovery of hidden diversity in Neotropical freshwater fishes, and we conclude that barcoding is a useful tool for alpha taxonomy.

Keywords::

• DNA barcode
• Neotropical region
• freshwater fishes
• COI
• CytB
• mitochondrial DNA

Introduction

The Neotropical freshwater fish fauna is one of the richest in the world (Schaefer 1998), with about 6000 species recognized in this region, out of which 4475 are actually considered valid and 1550 are recognized but not yet described (Reis et al. 2003). In Brazil, there are about 2587 valid species and many others to be described (Buckup et al. 2007), but, even so, the sampling of species is insufficient and many regions remain almost unexplored (Langeani et al. 2006; Junk 2007). Schaefer (1998) estimates that there may be as many as 8000 species in the Neotropical region. For example, in a recent study of the fish fauna from the Upper Paraná Basin, the best studied region in Neotropical area, Langeani et al. (2006) made an inventory, which revealed that about 15% (∼50) represent new species. Many other works pointed that the number of fish species tends to increase mainly among those fish belonging to small-sized groups and that inhabit headwaters streams (Schaefer 1998; Vari and Malabarba 1998; Castro et al. 2003, 2004, 2005; Langeani et al. 2006). Additionally, the geographic distribution pattern of the Neotropical fish species is very complex, with some species having a very restricted distribution (e.g. Trichomycterus maracaya, Characidium xanthopterum) occurring mostly in headwaters, and others having a wide distribution (e.g. Hoplias malabaricus, Astyanax paranae) sometimes occurring in more than one hydrographic basin (Reis et al. 2003; Junk 2007).

The genus Piabina, composed by small fishes (∼50 mm), belongs to the family Characidae but its relationship with the remaining characids is uncertain (Lima et al. 2003). Two species are assigned to Piabina: Piabina argentea Reinhardt, 1867 and Piabina anhembi da Silva and Kaefer, 2003. P. argentea has a wide geographic distribution occurring in the Upper Paraná Basin (the same region of this work); in the São Francisco Basin (type-locality); and in the Itapirucu, Paraíba do Sul, and Itapemirim rivers (eastern Brazilian basins) (Vari and Harold 2001). P. anhembi is restricted to its type-locality (Upper Tietê River, Salesópolis, São Paulo, Brazil) (da Silva and Kaefer 2003). These two species differ from each other by the teeth position, head size, and mouth proportions (da Silva and Kaefer 2003). Piabina differs from its putative sister group, Creagrutus, only by two subtle characters: the fourth infraorbital bone morphology and the teeth position (Vari and Harold 2001). Creagrutus and Piabina were allopatric (Vari and Harold 2001) until the discovery of a new Creagrutus species in the Upper Paraná River Basin (Ribeiro et al. 2004). The Piabina species populations have a limited dispersion, usually living in a restricted hydrographic region (Lowe-Mcconnell 1999). Castro (1999) suggests a limited dispersion to small fishes, which restricts their geographical distribution and may facilitate the population geographical subdivision enabling the possible creation of new species by geographic isolation (allopatry).

The advance of molecular techniques has proven a useful tool in biodiversity studies, mainly in those cases where the traditional tools are insufficient or unable to identify species. The use of genetic techniques has revealed that some species are actually species complexes (Agostinho et al. 2007). Bickford et al. (2006) showed that there has been an increased recognition of cryptic species from different groups of animals and plants in the past two decades due to the use of molecular methods. Hebert et al. (2003) proposed the DNA barcoding technique as a useful molecular tool for the identification of species, and many published works have shown the efficacy of this methodology for the identification of several organisms (Hebert et al. 2004a; Ward et al. 2005; Clare et al. 2007; Kelly et al. 2007; Hubert et al. 2008; Valdez-Moreno et al. 2009). Hebert et al. (2004b) proposed a threshold to delimit species that are 10 × larger than the intraspecific average values. New species have been proposed with DNA barcoding data and some of these species have been formally described later (Smith et al. 2005; Witt et al. 2006; Ward 2007; Nguyen and Seifert 2008; Ward et al. 2008; Yassin et al. 2008).

Considering the wide distribution and limited dispersion of small fish P. argentea and the promising use of DNA barcodes for flagging new species, the present work assessed samples of P. argentea from the Upper Paraná and São Francisco basins to check the hypotheses that this species could represent more than one biological unit.

Materials and methods

Specimen collection

Fifty-three P. argentea specimens from 12 sites located in the Upper Paraná Basin and one in the São Francisco Basin and five P. anhembi specimens from the Upper Paraná River Basin were collected (Table I and Figure 1). The Velhas River in the São Francisco Basin was sampled because this is the type locality of P. argentea. Additionally, two Creagrutus specimens (Creagrutus meridionalis and Creagrutus paraguayensis) from the Paraguay River Basin were used as outgroup (Table I). All specimens had a fresh fragment tissue removed and preserved in absolute ethanol at − 20°C. Voucher specimens were deposited in the collection of Laboratório de Biologia e Genética de Peixes, Departamento de Morfologia, Instituto de Biociências, UNESP, Botucatu, São Paulo, Brazil. The specimens' provenance data were deposited in BOLD Project EFUPR (Ratnasingham and Hebert 2007).

Table I.  Specimen data.

					GPS			GenBank accession number
Cluster	Species name	Voucher number	Collection code	Locality state	Latitude	Longitude	BOLD process ID	COI	CytB
A	P. argentea	36233	LBP 7329	Paraná	− 25.092	− 52.495	FUPR120-09	HM144073	GU908189
A	P. argentea	34659	LBP 7111	Paraná	− 23.938	− 50.729	FUPR181-09	HM144065	GU908182
A	P. argentea	36235	LBP 7329	Paraná	− 25.092	− 52.495	FUPR122-09	HM144071	GU908187
A	P. argentea	36234	LBP 7329	Paraná	− 25.092	− 52.495	FUPR121-09	HM144072	GU908188
A	P. argentea	17229	LBP 2594	São Paulo	− 21.013	− 49.690	FUPR089-09	HM144104	GU908219
A	P. argentea	25091	LBP 6745	São Paulo	− 22.341	− 48.935	FUPR102-09	HM144091	GU908207
A	P. argentea	14154	LBP 1996	São Paulo	− 22.917	− 48.500	FUPR062-09	HM144107	GU908222
A	P. argentea	21305	LBP 3509	São Paulo	− 22.941	− 48.584	FUPR091-09	HM144102	GU908217
A	P. argentea	25050	LBP 6741	São Paulo	− 23.231	− 48.533	FUPR101-09	HM144092	GU908208
A	P. argentea	23459	LBP 6741	São Paulo	− 23.024	− 48.826	FUPR092-09	HM144101	GU908216
A	P. argentea	20764	LBP 6741	São Paulo	− 23.024	− 48.826	FUPR090-09	HM144103	GU908218
A	P. argentea	25410	LBP 6745	São Paulo	− 22.341	− 48.935	FUPR108-09	HM144085	GU908201
A	P. argentea	17228	LBP 2594	São Paulo	− 21.013	− 49.690	FUPR088-09	HM144105	GU908220
A	P. argentea	17227	LBP 2594	São Paulo	− 21.013	− 49.690	FUPR087-09	HM144106	GU908221
A	P. argentea	29280	LBP 6268	Minas Gerais	− 21.285	− 46.493	FUPR201-09	HM144056	GU908174
A	P. argentea	29282	LBP 6268	Minas Gerais	− 21.285	− 46.493	FUPR203-09	HM144054	GU908172
B	P. argentea	25214	LBP 6743	São Paulo	− 22.786	− 48.481	FUPR238-09	HM144052	GU908170
B	P. argentea	23222	LBP 6743	São Paulo	− 22.786	− 48.481	FUPR237-09	HM144053	GU908171
B	P. argentea	25219	LBP 6743	São Paulo	− 22.786	− 48.481	FUPR107-09	HM144086	GU908202
B	P. argentea	22884	LBP 4032	Distrito Federal	− 15.115	− 47.046	FUPR189-09	HM144063	GU908181
B	P. argentea	23220	LBP 6743	São Paulo	− 22.786	− 48.481	FUPR109-09	HM144084	GU908200
B	P. argentea	35870	LBP 7280	Goiás	− 17.120	− 48.740	FUPR192-09	HM144060	GU908178
B	P. argentea	35871	LBP 7280	Goiás	− 17.120	− 48.740	FUPR193-09	HM144059	GU908177
B	P. argentea	35849	LBP 7680	Goiás	− 17.099	− 48.762	FUPR191-09	HM144061	GU908179
B	P. argentea	35904	LBP 7292	Goiás	− 17.801	− 48.372	FUPR115-09	HM144078	GU908194
B	P. argentea	35906	LBP 7292	Goiás	− 17.801	− 48.372	FUPR117-09	HM144076	GU908192
B	P. argentea	35905	LBP 7292	Goiás	− 17.801	− 48.372	FUPR116-09	HM144077	GU908193
B	P. argentea	35848	LBP 7680	Goiás	− 17.099	− 48.762	FUPR190-09	HM144062	GU908180
C	P. argentea	31608	LBP 6513	Minas Gerais	− 19.385	− 43.659	FUPR113-09	HM144080	GU908196
C	P. argentea	31605	LBP 6513	Minas Gerais	− 19.385	− 43.659	FUPR110-09	HM144083	GU908199
C	P. argentea	31609	LBP 6513	Minas Gerais	− 19.385	− 43.659	FUPR114-09	HM144079	GU908195
C	P. argentea	31606	LBP 6513	Minas Gerais	− 19.385	− 43.659	FUPR111-09	HM144082	GU908198
C	P. argentea	31607	LBP 6513	Minas Gerais	− 19.385	− 43.659	FUPR112-09	HM144081	GU908197
D	P. argentea	25162	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR103-09	HM144090	GU908206
D	P. argentea	25156	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR097-09	HM144096	GU908211
D	P. argentea	25030	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR093-09	HM144100	GU908215
D	P. argentea	25031	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR094-09	HM144099	GU908214
D	P. argentea	25167	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR106-09	HM144087	GU908203
D	P. argentea	25165	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR104-09	HM144089	GU908205
D	P. argentea	25166	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR105-09	HM144088	GU908204
D	P. argentea	25161	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR099-09	HM144094	GU908209
D	P. argentea	25157	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR098-09	HM144095	GU908210
D	P. argentea	25155	LBP 6744	São Paulo	− 23.231	− 48.533	FUPR096-09	HM144097	GU908212
E	P. argentea	29240	LBP 6226	Minas Gerais	− 21.321	− 46.511	FUPR200-09	HM144057	GU908175
E	P. argentea	29239	LBP 6226	Minas Gerais	− 21.321	− 46.511	FUPR199-09	HM144058	GU908176
E	P. argentea	29281	LBP 6268	Minas Gerais	− 21.285	− 46.493	FUPR202-09	HM144055	GU908173
F	P. argentea	35934	LBP 7301	Goiás	− 18.110	− 48.504	FUPR119-09	HM144074	GU908190
F	P. argentea	35933	LBP 7301	Goiás	− 18.110	− 48.504	FUPR118-09	HM144075	GU908191
F	P. argentea	28396	LBP 5994	Minas Gerais	− 21.732	− 46.423	FUPR129-09	HM144067	GU908184
F	P. argentea	28394	LBP 5994	Minas Gerais	− 21.732	− 46.423	FUPR127-09	HM144069	GU908185
F	P. argentea	28393	LBP 5994	Minas Gerais	− 21.732	− 46.423	FUPR126-09	HM144070	GU908186
–	P. anhembi	23373	LBP 6742	São Paulo	− 23.524	− 45.890	FUPR236-09	HM144047	GU908165
–	P. anhembi	23360	LBP 6742	São Paulo	− 23.524	− 45.890	FUPR235-09	HM144048	GU908166
–	P. anhembi	20821	LBP 6742	São Paulo	− 23.524	− 45.890	FUPR125-09	HM144049	GU908167
–	P. anhembi	23372	LBP 6742	São Paulo	− 23.524	− 45.890	FUPR124-09	HM144050	GU908168
–	C. meridionalis	35631	LBP 7557	Mato Grosso do Sul	− 20.343	− 55.726	FUPR188-09	HM144042	GU908160
–	C. paraguayensis	35628	LBP 7557	Mato Grosso do Sul	− 20.343	− 55.726	FUPR185-09	HM144045	GU908163

LBP, Laboratório de Biologia e Genética de Peixes.

Figure 1.  Map showing the distribution of samples of Piabina. Letters correspond to P. argentea clusters. Black square represents P. anhembi species.

Extraction, PCR, and sequencing

Total genomic DNA was isolated from fin or muscle tissue of each specimen using the DNeasy Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The partial mitochondrial cytochrome c oxidase subunit I gene (COI, 648 bp) was amplified by the PCR using two sets of primers: FishF1, 5′-TCAACCAACCACAAAGACATTGGCAC-3′; FishF2, 5′-TCGACTAATCATAAAGATATCGGCAC-3′; FishR1, 5′-TAGACTTCTGGGTGGCCAAAGAATCA-3′; and FishR2, 5′-ACTTCAGGGTGACCGAAGAATCAGAA-3′ (Ward et al. 2005). The whole cytochrome b (CytB, 1118 bp) mitochondrial gene was amplified by PCR using the CytB-F, 5′-GACTTGAAAAACCAYCGTTGT-3′, and CytB-R, 5′-GCTTTGGGAGTTAGDGGTGGGAGTTAGAATC-3′ (C. Oliveira, pers. comm.). PCR was carried out on a thermocycler (Veriti^® 96-Well Thermal Cycler; Applied Biosystems, Foster City, California, USA) with a final volume of 12.5 μl containing 0.3 μl dNTP (2 mM), 1.25 μl 10 × Taq buffer (50 mM KCl, 10 mM Tris–HCl, 0.1% Triton X-100, and 1.5 mM MgCl₂), 0.3 μl each primer (10 μM), 0.7 μl MgCl₂ (50 mM), 0.05 μl Taq-Pht DNA polymerase (5 U), 1 μl template DNA (10–20 ng), and ultrapure water. The thermocycler conditions to amplify the COI were initial denaturation at 95°C for 5 min followed by 30 cycles of denaturation at 95°C for 45 s, annealing at 55°C for 30 s, and extension at 72°C for 60 s. A final extension at 72°C for 10 min was performed. The thermocycler conditions to amplify the CytB were initial denaturation at 95°C for 5 min followed by two cycles of denaturation at 95°C for 30 s, annealing at 55°C for 45 s, and extension at 72°C for 60 s; two cycles of denaturation at 95°C for 30 s, annealing at 50°C for 45 s, and extension at 72°C for 60 s; two cycles of denaturation at 95°C for 30 s, annealing at 48°C for 45 s, and extension at 72°C for 60 s; 25 cycles of denaturation at 95°C for 30 s, annealing at 50°C for 45 s, and extension at 72°C for 60 s; and a final extension at 72°C for 5 min. Amplified products were checked on 1% agarose gels stained with Blue Green Loading Dye I (LGC Biotecnologia, Cotia, Sa˜o Paulo, Brazil). The PCR products were purified with ExoSAP-IT^® (USB Corporation, Cleveland, OH, USA) following the manufacturer's protocol. The purified PCR product was used as template to sequence both DNA strands. The cycle sequencing reaction was carried out using a BigDye™ Terminator v3.1 Cycle Sequencing Ready Reaction kit (Applied Biosystems) in a final volume of 7 μl containing 1.4 μl template, 0.35 μl primer (10 μM), 1.05 μl buffer 5 × , 0.7 μl BigDye mix, and water. The cycle sequencing conditions were initial denaturation at 96°C for 2 min followed by 30 cycles of denaturation at 96°C for 45 s, annealing at 50°C for 60 s, and extension at 60°C for 4 min. The PCR sequencing products were purified with ethylenediamine tetraacetic acid/sodium acetate/alcohol following the protocol suggested in the BigDye™ Terminator v3.1 Cycle Sequencing kit's manual (Applied Biosystems). All samples were sequenced on an ABI 3130 Genetic Analyzer (Applied Biosystems) following the manufacturer's instructions. All sequences were deposited in the GenBank and in the Barcode of Life Data Systems (Project EFUPR) (Table I).

Data analysis

All sequences were analyzed using SeqScape^® software v2.6 (Applied Biosystems) to obtain the consensus sequences and check the occurrence of deletions, insertions, and stop codons. The sequences were aligned using the online version of MUSCLE (Edgar 2004). The genetic distance among and within observed clusters was calculated using the Kimura-two-parameter (K2P) distance model (Kimura 1980) for both genes separately. A neighbor-joining (NJ) tree of K2P distances using the combined COI and CytB sequences was created to provide a graphic representation of the relationships among specimens and clusters with the software MEGA 4.0 (Tamura et al. 2007). Bootstrap resampling (Felsenstein 1985) was applied to assess the support for individual nodes using 1000 pseudo-replicates.

Phylogenetic analyses using maximum parsimony were performed using PAUP* version 4.0b10 (Swofford 2002) with heuristic searches, random addition of sequences, and tree bisection and reconnection algorithms. The ACCTRAN optimization method was utilized. The parsimony trees were constructed using a 1:1 transition–transversion ratio. Cluster robustness was assessed using 1000 bootstrap pseudo-replicates (Felsenstein 1985) with the same parameters cited above.

The seven major clusters obtained were considered as different units for the fixation index (F_ST) calculation using Arlequin 3.11 (Excoffier et al. 2005). A statistical parsimony network was constructed using TCS 1.21 (Clement et al. 2000), which employs the method of Templeton et al. (1992) with a statistical confidence interval of 90%. The analyses were carried out in TCS using the “fix connection limit” option to obtain the mutational steps necessary to connect the seven observed haplotype networks. The ancestral haplotype was also identified using TCS according to the method of Castelloe and Templeton (1994).

Results

Sequence data

Sequence data for a 648 bp fragment of COI and 1118 bp of CytB were obtained for a total of 58 Piabina specimens (53 P. argentea and 5 P. anhembi). We also obtained the COI and CytB sequences from two specimens of Creagrutus (C. meridionalis and C. paraguayensis) used as an outgroup. No sequences showed insertions, deletions, or stop-codons, and the global transition–transversion ratio was 4.4. A total of 233 nucleotides (72 in COI and 161 in CytB) were variable in the data set of Piabina specimens (∼13%—outgroup not considered) and 209 of them were informative in the parsimony analyses. These variations defined a total of 42 haplotypes (COI and CytB displayed 28 and 39 haplotypes, respectively). The two methods of tree construction (NJ and maximum parsimony) resulted in the same topology (except for some internal taxa in the subclusters; data not shown), which showed seven major clusters with high support values (Figure 2). P. anhembi samples formed one cluster and P. argentea samples were divided into six clusters, one corresponding to the sample from the São Francisco River Basin (Cluster C) and five clusters representing P. argentea samples from the Upper Paraná Basin (Figure 2). These seven clusters are divided into two major groups, one containing Cluster A and a second group with the other clusters (Figure 2). We use two different methodologies of tree construction to check the robustness of the data.

Figure 2.  NJ tree of COI/CytB showing the seven major clusters obtained among Piabina specimens (A–F represent P. argentea). Node values represent statistic support: upper values, NJ bootstrap (1000 pseudo-replicates); lower values, maximum parsimony bootstrap (1000 pseudo-replicates). Numbers on fishes represent voucher number and size of photographed specimens (left and right, respectively).

The inter-cluster K2P genetic distance values ranged from 2% (Clusters D × E) to 5.6% (Clusters A × C) and from 2.3% (Clusters B × C) to 5.4% (Clusters A × E) for COI and CytB, respectively (Table II). The average intra-cluster K2P distance ranged from 0 to 0.9% (average = 0.36%) for COI and from 0.1 to 1% (average = 0.5%) for CytB (Table II).

Table II.  K2P genetic distance obtained among the seven major Piabina clusters.

	A	B	C	D	E	F	P. anhembi
A	0.009/0.010	0.047	0.047	0.042	0.054	0.052	0.048
B	0.042	0.007/0.010	0.023	0.030	0.047	0.050	0.027
C	0.056	0.025	0.003/0.004	0.031	0.053	0.053	0.030
D	0.035	0.021	0.029	0.002/0.002	0.041	0.049	0.025
E	0.048	0.025	0.031	0.020	0/0.005	0.049	0.049
F	0.047	0.039	0.045	0.032	0.036	0.003/0.003	0.051
P. anhembi	0.040	0.025	0.034	0.022	0.028	0.034	0.001/0.001

Note: COI below diagonal and CytB above diagonal. The average values of intra-K2P distance represented in bold on the diagonal (COI/CytB).

Cluster comparisons

The pairwise F_ST index among the seven clusters identified showed values from 0.77 to 0.98 for COI and from 0.66 to 0.96 for CytB, all highly significant (p < 0.001) (Table III).

Table III.  Pairwise F_ST index obtained among the seven major Piabina clusters.

	A	B	C	D	E	F	P. anhembi
A	–	0.79254	0.84386	0.85336	0.85659	0.86067	0.86193
B	0.80038	–	0.65594	0.77328	0.78754	0.82325	0.70487
C	0.85826	0.76602	–	0.92187	0.91004	0.92957	0.91288
D	0.82645	0.78760	0.92876	–	0.94973	0.96113	0.95335
E	0.83305	0.77467	0.93190	0.93346	–	0.93750	0.95991
F	0.83610	0.84631	0.92541	0.92958	0.93352	–	0.96486
P. anhembi	0.81920	0.79240	0.93925	0.93510	0.97734	0.93127	–

Note: COI gene below diagonal, CytB gene above diagonal.

The haplotype network based on Templeton's method (Templeton et al. 1992) with the combined data set (COI/CytB) displayed seven unconnected networks, one representing P. anhembi and the other six representing P. argentea (Figure 3). This result is consistent with the seven clusters identified through the phylogenetic analysis (Figure 2). The number of haplotypes present in each network range from 3 (Cluster E) to 11 (Cluster A) (Figure 3), and the number of mutational steps necessary to connect the independent P. argentea networks ranged from 45 to 110 (Figure 3, dashed lines). The haplotype network was constructed for each separate gene to check whether the same seven unconnected networks would be obtained. Both genes displayed the same result, with 14-34 (COI) and 25-71 (CytB) mutational steps necessary to connect the independent networks (networks not shown).

Figure 3.  Seven unconnected haplotype networks among Piabina specimens. P. anhembi is represented in gray. Numbers inside the figures represent specimens that share the same haplotype. Numbers on lines represent the mutational steps between haplotypes. Dashed lines represent the necessary steps to connect the independent networks.

Discussion

The specimens of Piabina were divided into seven clusters in the phylogenetic analysis, one cluster representing P. anhembi and the six other representing P. argentea (Figure 2). The data showed the absence of genetic flow among local samples and permit one to suggest that P. argentea represents six different biological units (meaning a minimum of five new species). These seven clusters were confirmed by haplotype network (Figure 3) and are divided into two major groups (Figure 2). The first group contains the Cluster A and is the sister group of the second group, composed by the remaining clusters, including P. anhembi (Figure 2). The average inter-cluster K2P distance values among P. argentea were about nine times greater than the average intra-cluster values found for the COI (from 5.6 to 15.6 × ) and CytB genes (from 4.6 to 10.8 × ) (Table II) and the inter-cluster values among the P. argentea units were similar to the values between P. argentea clusters and their congeners P. anhembi (average = 3.0 and 3.8% for COI and CytB, respectively), reinforcing the hypothesis of the existence of more than one biological unit for P. argentea (Table II). These results corroborate the hypothesis of limited dispersion for Piabina species (Lowe-Mcconnell 1999) and other small fishes (Castro 1999), which facilitates the population geographical subdivision enabling the possible creation of new species by geographic isolation (allopatry).

Hebert et al. (2004b) suggested a threshold to delimit species with DNA barcode data. These values should be at least 10 × the average intraspecific values. The average intra-cluster values of the six P. argentea clusters were 0.4% and 0.56% for COI and CytB, respectively, and some inter-cluster divergences within P. argentea are slightly below this limit (see Table II). However, a recent review of “barcoded” fishes (Ward 2009) noted that about 17% of the genetic divergence values among congeneric species were less than 3% divergent and that a further 3.7% of congeners are less than 1% divergent. The author suggests that if the unknown specimen is more than 2% divergent from the known specimen, it is very likely that this is a different species with a probability greater than 95%. Hence, the threshold limit proposed by Hebert et al. (2004b) as an indicator of cryptic speciation should be carefully analyzed for each group.

Ward et al. (2007), working with sharks of the genus Squalus, observed the formation of two clusters in the species Squalus acanthias, which showed a genetic divergence of just 0.76% between them. Interestingly, these two groups had been considered as two species until the decade of 1960: S. acanthias from the Atlantic and South Pacific Oceans and Squalus suckkeyi from the North Pacific Ocean (see Jensen 1966). The authors suggested the revalidation of the second species. The comparison with values among congener species may be useful for the delimitation of a threshold. Ornelas-Garcia et al. (2008), working with species of the genus Astyanax from Mesoamerica, found that some specimens formed separate clusters and suggested the occurrence of a species complex in this genus, assigning provisional names to each cluster obtained. Ward et al. (2008), working with Asian sea bass Lates calcarifer specimens from different localities (Australia and Myanmar), found genetic distance values of 9.5% between two groups for COI (DNA barcode region) and 11.3% for CytB. The authors suggested the existence of two species. The average divergence value of “barcoded” congeneric fishes is about 8.4% (Ward 2009). Values smaller than this average, such as those observed in the present work and in the above-cited papers, can be explained in two ways: the rate of evolution can vary among different higher taxa and, consequently, the accumulation of substitutions can vary. In fact, it has been observed that different teleost orders have different evolutionary rates (Krieger and Fuerst 2002). Another possible explanation could relate to species ages, where evolutionarily “young” species may not have had sufficient time to accumulate many mutations in their barcodes. In fact, Montoya-Burgos (2003), working with species of Hypostomus from South America, suggested that the process of divergence and radiation in this genus dates back to sometime between 12 and 4 million years ago. Hubert et al. (2007), working with Serrasalmus and Pygocentrus from South America, encountered similar values suggesting that species separation dates back to sometime between 8 and 2 million years ago. Both authors suggested that this pattern is valid for most Neotropical freshwater fishes. In their studies of Rhamdia and Synbranchus fish species, Perdices et al. (2002, 2005) proposed similar patterns for Mesoamerica and Ornelas-Gacia et al. (2008) corroborated the same patterns for Astyanax. With increasing recognition that mitochondrial DNA is under strong selection, some authors caution against the use of mitochondrial DNA data for dating divergence events, but, this caveat notwithstanding, recognize that selective sweeps can be beneficial for barcoding (Galtier et al. 2009). Molecular clock approaches that infer age of the most recent common ancestor tend to be overestimated using mitochondrial DNA unless they correct for apparent rate differences between short and long time frames (Rand 2008).

The intercluster analysis performed confirmed the presence of seven dissimilar barcode sequence clusters among the Piabina specimens examined. The haplotype networks obtained using the combined data set (Figure 3) and those for each gene separately (data not shown) displayed seven unconnected networks with high numbers of mutational steps (ranged from 45 to 110; Figure 3) necessary to connect these independent networks. This situation is not expected when the specimens represent a single species (Hart and Sunday 2007), even when there is very strong structure among populations. Some pairs of P. argentea clusters need more mutational steps than others to connect with their congener P. anhembi species (Figure 3). Thus, these results support the hypothesis that P. argentea comprises more than one biological species. Kon et al. (2007), working with the gobioid fish Schindleria, obtained an unconnected haplotype network with seven independent clusters and suggested that Schindleria represents a species complex, as imparted here.

The F_ST index showed very high values among the seven clusters obtained (Table III), with similar values among P. argentea and P. anhembi clusters. Considering that F_ST values between 0 and 0.05 indicate a low genetic structure, values between 0.05 and 0.15 a moderate genetic structure, values between 0.15 and 0.25 a high genetic structure, values above 0.25 a strong genetic structure, and values close to 1 are usually found among different species (Wright 1978; Hartl and Clark 1997); the values presented in Table III strongly suggest that our seven clusters represent different species.

Many species have been discovered with the use of molecular data and some have been formally described later (Smith et al. 2005; Witt et al. 2006; Kon et al. 2007; Ward et al. 2007, 2008; Nguyen and Seifert 2008; Yassin et al. 2008), and the DNA barcode has also been utilized as part of the validation and formal description of new fish species such as Coryphopterus kuna (Victor 2007); Urolophus kapalensis (Yearsley and Last 2006); Brachionichthys autralis (Last et al. 2007); five new species of Chromis genus (Pyle et al. 2008), Dipturus argentinensis (Diaz de Astarloa et al. 2008), and Moenkhausia forestii (Benine et al. 2009). Our data suggest that the widely distributed P. argentea species represent more than one biological unit in the Upper Paraná Basin, and probably this hypothesis is valid all over the area of occurrence of this species. Interestingly, some clusters were found only in a single locality (Clusters C–F, and P. anhembi) while others are widely dispersed (Clusters A and B) (Figure 1). The fact that Clusters A and B are widely dispersed could be a cause of no prior recognition of these possible species, since the area of overlap between them could impede its recognition. Thus, we suggest that a detailed review of Piabina be conducted to validate these new species (sensu Padial et al. 2010). On the other hand, we believe that the analysis of many other widely distributed fish species may also disclose new species.

Conclusions

Our data demonstrate the efficacy of DNA barcoding for discriminating known species and to flag new ones, alone or associated with other genes. Despite the concerns of Hickerson et al. (2006) to the contrary, DNA barcoding revealed the existence of separate taxa with low divergence rate or recent radiation. We also substantiate the use of DNA barcode sequences as part of the formal description of species. These data can be useful when morphological characters are insufficient or too weak to define a species and, importantly, because they apply to any sex or life stage, can help to disambiguate the application of names in future studies.

Acknowledgements

The authors are grateful to Renato Devidé and Ricardo Teixeira for their help with the fish collection. Financial support for the present study was provided by CNPq and FAPESP.

Declaration of interest: Financial support for this study was provided by Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP) and Conselho Nacional de Desenvolvimento Cientifico e Tecnológico (CNPq).