Molecular assessment of Gymnotus spp. (Gymnotiformes: Gymnotidae) fishing used as live baitfish in the Tietê River, Brazil

ABSTRACT The capture of live bait for sport fishing is an important activity for fishing communities. The main species used for this purpose are members of the genus Gymnotus, which comprises numerous species of cryptic nature that are difficult to identify based on external morphology. The aims of this work were to identify through partial sequences of the COI gene Gymnotus species fished in the Jacaré-Guaçu River, SP, and to develop a molecular diagnostic approach using PCR-RFLP to identify these species. Partial COI sequences were compared to those of other species deposited in GenBank. The sequences were assessed in the NEBCutter program to determine restriction sites in the sequence and the enzymes to be tested. Phenetic analysis performed by Neighbor-Joining method showed that the specimens sampled belong to two species preliminary identified here as G. cf. sylvius and G. cf. cuia, with G. cf. sylvius accounting for 95.2% of the individuals sampled. The enzymes NlaIII and SacI generated fragments that allowed distinguishing the Gymnotus species using PCR-RFLP. This analysis can be used to accurately identify these species, which is fundamental for monitoring Gymnotus fishing and assessing the conservation of this genetic resource.


Introduction
Species of Gymnotus Linnaeus, 1758 are an important fishing resource for fishermen communities and the principal source of commercial live bait for sport fishing (Sousa et al., 2017). The genus Gymnotus is a member of the family Gymnotidae (order Gymnotiformes). This genus consists of many species, represented by 43 valid species in six clades (Tagliacollo et al., 2016), with wide distribution in the continental waters of South and Central America, except in Chile and Belize (Albert, 2001).
The identification of species in this genus based on morphological characters is sometimes challenging due to the morphological conservatism among species 2 e190075 [2] and high phenotypic plasticity within species. Many efforts have been made based on studies of morphology, cytogenetics, molecular, and electrical signals to elucidate the taxonomy of electric fishes, including Gymnotus (Alves-Gomes et al., 1995;Albert, Crampton, 2003;Fernandes et al., 2005;Margarido et al., 2007;Maxime, Albert, 2009;Scacchetti et al., 2011;Milhomem et al., 2012;Tagliacollo et al., 2016;Sousa et al., 2017;Craig et al., 2017Craig et al., , 2018. The existence of cryptic species in combination with poor-quality data from landing records for biomass estimation and catch quotas, may imperil the longterm survival of given fish genetic resources (Hilsdorf, Hallerman, 2017). In addition, worldwide concerns on the indiscriminate use of live baitfish, the ecological risks of introducing alien species, and the depletion of local species, are on the agenda of fish diversity conservation in many countries (Drake, Mandrak, 2014;Sousa et al., 2017). Therefore, the correct identification of species is one of the first steps towards successfully establishing conservation and management measures to protect local baitfish species. Hebert et al. (2003) proposed the use of a fragment of approximately 650bp of the mitochondrial cytochrome c oxidase subunit 1 (COI or COX1) as a global identification system for most animal species presently known as the DNA barcoding.
Supported by this background, we use partial sequences of COI to identify Gymnotus spp. captured as live baitfish and develop a rapid and low-cost molecular methodology to identify unequivocally the species being caught to help ensure the sustainability of Gymnotus fisheries and preserve the biodiversity of species.

Material and Methods
Collection and Sampling Site. Taxon sampling was carried out between October 2015 and March 2016 along the Jacaré-Guaçu River, a tributary of the Tietê River, southeast region of Brazil (Fig. 1). Eight individuals from four fishing sites were collected in two season periods, in a total of 64 individuals. All specimens were photographed, labeled and a small piece of muscle or fin clips were taken from each fresh fish and stored in 95% ethanol at -20°C. Voucher specimens were deposited in the Ichthyology Collection of the Museu de Zoologia da Universidade de São Paulo (MZUSP), SP, Brazil, receiving the register numbers MZUSP 124930 and MZUSP 124931.
The procedures carried out in this study were in accordance with the ethical standards of the Colégio Brasileiro de Experimentação Animal (COBEA) and were approved by the Comitê de Ética em Experimentação Animal do Instituto de Pesca (CEEAIP) protocol number 01/2016.  DNA extraction, PCR and Sequencing. Total genomic DNA was extracted following the protocol of saline extraction method (Aljanabi, Martinez, 1997). Partial sequences of the cytochrome c oxidase 1 (COI or Cox1) gene have been amplified by the polymerase chain reaction (PCR) using the universal primers FishF1 (5 ' -TCA ACC AAC CAC AAA GAC ATT GGC AC-3 ') and FishR1 (5 ' -TAG ACT TCT GGG TGG CCA AAG AAT CA-3') (Ward et al., 2005). PCR was conducted for final volume of 50 µL containing 1x buffer solution, 0.2 mM of each DNTP, 25 mM MgCl 2 , 0.2 µM of each primer (sense and anti-sense), 1.0 U of Platinum Taq DNA Polymerase High Fidelity (Invitrogen TM , Carlsbad, CA, USA) and 0.5 -2 ng of DNA. The amplification reaction was conducted in a PTC-200 thermal cycler (MJ Research, Waltham, MA, USA) with an initial cycle of denaturation at 94°C for 2 min, followed by 35 cycles (94°C for 40 S, 58°C for 1 min, 72°C for 1 min), and the final extension at 72°C for 7 min.
PCR products were purified with CleanSweep™ PCR Purification Reagent (Applied Biosystems™), according to the protocol provided by the manufacturer. For the sequencing reactions the BigDye™ Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems ™) was used. Sequencing was performed on the ABI 3730 DNA Analyzer (Applied Biosystems™).
Sequence Analysis. The COI partial sequences of all samples were edited in the Chromas V.2.6 program. The Codon Code Aligner V 1.5.2 program (Codon Code Corporation, Dedham, Massachusetts, United States) was used for quality evaluation and obtaining the consensus sequence (contig) from the forward and reverse sequences.
The sequences were aligned with the ClustalW command (Thompson et al., 1994) implemented in the MEGA 6 program (Tamura et al., 2014). Neighbor-Joining (NJ) based on Kimura two Parameter (K2P) genetic distance were used to analyze the phenetic relationship between the samples, as recommended by Barcode of Life. The bootstrap re-sampling analysis was performed using 1,000 replicates (Nei, Kumar, 2000) and the genetic divergence was calculated by the K2P distances (Kimura, 1980).

PCR-RFLP.
Gymnotus partial COI sequences amplified herein, and those retrieved from GenBank were assessed using the NEBCutter program (Vincze et al., 2003). The diagnostic enzymes were used to distinguish among the Gymnotu sampled individuals.
Digestion was carried out at 10 µL final volume, including 1.0 µL of endonuclease (10 U.mL-1) (BioLabs), 1 µL of 10x NEBuffer 4 (50mM Potassium Acetate, 20mM Tris-acetate, 10mM Magnesium Acetate, 1mM dithiothreitol (DTT), pH 7.9, 25ºC; Biolabs) and 5 µL of the PCR product. Solutions were incubated for 1-2 h at 37°C in the thermal block. The resulting sizes of restriction fragments were determined in 3.0% agarose gel electrophoresis by comparison with a standard DNA molecular weight marker. Images were digitalized for further analysis of the banding patterns.

Molecular Identification.
Out of 64 sequences examined only two showed poor quality and could not be used. After editing, the sequences were 585 bp long. Of the 62 successfully sequenced samples, 59 showed similarity of 99 to 100% with the species G. sylvius Albert & Fernandes-Matioli, 1999 and the other three sequences presented a best identity with G. inaequilabiatus (Valenciennes, 1839) when compared with other GenBank sequences in the MegaBLAST program. However, a species level match was not made, presumably due to misidentifications of the species names in GenBank.
The bootstrap values indicate branch statistical supports. Analysis showed that specimens from the fishing in the Jacaré-Guaçu River grouped in two different clades: one with sequences of G. sylvius (clade A) and another one with sequences of G. inaequilabiatus (clade C) (Fig. 2). The first one comprising 95.2% of the individuals sampled and the second one 4.8%. Specimens grouped in clade A yielded 5 haplotypes (Photos S2, S3, S4, S5 and S6) with 4 polymorphic sites (n = 59) and in clade C, 2 haplotypes (S7 and S8) with 3 polymorphic sites (n = 3).

Discussion
Successful conservation plans need precise species identification (FAO, 2013). Thus, genetic markers have been applied widely for traceability of fish and fish product as well as for the species identification and populations to enforce fisheries regulation for the conservation of fisheries resources (Ogden, 2008). Partial sequences of COI obtained in this study demonstrated that the Gymnotus species caught as live bait by artisanal fisher community in the Jacaré-Guaçu River belong to two distinct species -G. cf. sylvius and G. cf. cuia.
Species of Gymnotus known to occur in the Upper Paraná basin include: G. carapo australis Craig, Crampton & Albert, 2017, G. cuia, G. inaequilabiatus, G. pantanal, G. pantherinus, G. paraguensis Albert & Crampton, 2003 and G. sylvius (Craig et al., 2017;Craig et al., 2018). Of these species, G. carapo australis, G. cuia, and G. pantanal are ecologically abundant and G. inaequilabiatus, G. pantherinus, G. paraguensis and G. sylvius are not abundant. Gymnotus inaequilabiatus is often misidentified or mistaken for other species, despite the presence of multiple salient external diagnostic characters (Maxime, Albert, 2014) that place it in the G. tigre clade rather than the G. carapo clade. Further, G. inaequilabiatus is unique among congeners in its large adult body size, growing to one-meter total length. We found inconsistent taxonomic identifications when comparing the sequences of the specimens classified here as G. cf. sylvius and G. cf. cuia with the sequences available in GenBank. Three sequences presented better identity with G. inaequilabiatus (clade C), but also grouped with G. paraguensis, G. cf. paraguensis, G. carapo and Gymnotus sp. (Fig. 2). Our sequences of G. cf. cuia presented also 99% of identity with Gymnotus new sp. RS2 that, according to Tagliacollo et al. (2016), belongs to G. carapo clade and not to G. tigre clade, which includes the species G. inaequilabiatus and G. paraguensis (Maxime, Albert, 2014;Craig et al., 2018). It is possible that the specimens identified as G. inaequilabiatus in previous studies based on Graça, Pavanelli (2007) and Ota et al. (2018) may be actually G. cuia. These discordances were probably caused by misidentification of specimens used in previous studies, 6 e190075 [6] once new data from literature, as the recent revisions of Scacchetti et al. (2011), Craig et al. (2017 and Utsunomia et al. (2018) are now available to elucidate this complex group.
Despite of these considerations, we demonstrated that at least two different species are fished as live bait in Jacaré-Guaçu River. These two species, preliminarily identified here as G. cf. sylvius and G. cf. cuia, presented 4.7% to 5.6% of genetic divergence (Tab. 1) and they can easily be discriminated by PCR-RFLP method developed in this work. This technique has been proven to efficiently identify different taxa for distinct purposes (Wolf et al., 1999;Bieliková et al., 2010;Schmidt et al., 2015). The PCR-RFLP technique developed herein showed that the NlaIII and SacI endonucleases cut patterns clearly distinguished G. pantanal and G. pantherinus species where they occur in sympatry with G. cf. sylvius and G. cf. cuia, as shown in Tabs. 2 -3. The data demonstrated that the most frequent species was G. cf. sylvius corresponding to 95.2% of the fished specimens. Fernandes-Matioli et al. (2000) also observed Gymnotus populations of different species living in sympatry in some watersheds of Southeastern Brazil and always one of the species occur at a significantly lower frequency. The sequences alignment and analysis of the haplotypes retrieved from GenBank database allowed a carefully examination of polymorphic restriction sites to detect variation among species, excluding sites with intra-specific variation.
The identification of Gymnotus species fished for live bait is extremely important for the management of this fishery resource and as a result to determine proper fisheries regulation. The current fishing legislation in the Paraná River Basin refers to Gymnotus species as G. carapo and determines the minimum capture size of 20 mm (IBAMA, 2009). Thus, the present outcomes show the need for legislation change to include other Gymnotus being caught as live bait. The minimum capture size depends on the captured species, since each species has its reproduction cycle. Sousa et al. (2017) also suggested a revision of capture legislation in the Pantanal of Mato Grosso do Sul, because it recognizes two Gymnotus species. However, these authors found three species of Gymnotus commercialized as live bait in the region.
The use of live bait has been concerning worldwide due to risks of introduction of invasive alien species by sport anglers and the ecological impact on the native biodiversity once this live bait species become established (Sá et al., 2017). At the same time, even local fish species being indiscriminately used as live bait can impact the long-term survival of local species population. Species identification and management through legal monitoring is central to maintain sustainable fishing by local communities. The molecular surveillance tool developed herein allows the implementation of a rapid and low-cost diagnostic system by PCR-RFLP to identify Gymnotus species in the sampling fishing area of this study, which can be applied to yield seasonal Gymnotus fishing data and support a reliable fishing regulation to use this fisheries resources sustainably.