Using DNA barcodes to confirm new records of Amazon longfin herrings Pellona castelnaeana Valenciennes, 1847 and Pellona flavipinnis (Valenciennes, 1837) (Clupeiformes: Pristigasteridae) in the Branco River sub-basin

This note demonstrates the use of a DNA barcoding methodology in confirming new occurrence records of Pellona castelnaeana and Pellona flavipinnis in the Branco River sub-basin. The DNA barcode result was verified by identification based on morphological characters of both species. Thus, these records increase the species’ ranges by more than 600 km in the Amazon and show evidence of high genetic variability in P. flavipinnis.


Introduction
Fishes of the genus Pellona, commonly known as "apapás" or "sardas" in Brazil, are included in the family Pristigasteridae (longfin herrings). Two species have a wide distribution in the Amazon basin: Pellona castelnaeana Valenciennes, 1847 and Pellona flavipinnis (Valenciennes, 1837). Their general geographic distribu-tion was presented by Whitehead (1985), while more up-to-date occurrence records can be accessed at the Global Biodiversity Information Facility (http://www. gbif.org/). Nevertheless, these sources do not indicate records of these fishes from some important river systems of the Amazon basin, including the Branco River sub-basin in Roraima state. There are, however, personal observations reported by Ferreira et al. (2007). With this NOTES ON GEOGRAPHIC DISTRIBUTION note we establish occurrence records of P. castelnaeana and P. flavipinnis ( Fig. 1a and 1b, respectively) in the Branco River.
To confirm the identification of the specimens, we assessed meristic characters commonly used in the identification of these species: the number of the post-ventral scales and the number of the lower-arch gillrakers (Santos et al. 2007). For molecular identification, we used the cytochrome c oxidase subunit I (COI, DNA barcoding sensu Hebert et al. 2003). As reference sequences from Amazon-sourced Pellona were not available on GenBank, we used samples of P. castelnaeana and P. flavipinnis caught from known localities (Coari and Manaus, respectively; Fig. 2a) as reported by Whitehead (1985) ( Table  1). The samples of P. castelnaeana and P. flavipinnis from known localities are from unpublished data from Ximenes (2014) and were used here to compare with specimens from Branco River (Fig. 2a). The sequence data were deposited in GenBank under the following accession numbers: KX462719 to KX462723 for P. castelnaeana and KX462724 to KX462728 for P. flavipinnis.
DNA was extracted using a standard phenol/chloroform protocol (Sambrook et al. 1989), and its quality was visualized on a 1% agarose gel and quantified using a NanoDrop 2000 UV-Vis spectrophotometer. The primers used and the amplification of COI was conducted as described by Ivanova et al. (2007). The PCR products were purified using EXO-SAP (Exonuclease -Shrimp Alcaline Phosphatase) and subjected to fluorescent dye-terminator (ddNTP) sequencing following the manufacturer's recommended protocol for BigDye sequencing chemistry (Life Technologies). Sequences were aligned using the Clustal W algorithm (Thompson et al. 1994) as implemented in BioEdit (Hall 1999).   Table 1. Individuals used in this study (ID code used in the map, Figure 2). Voucher and tissue sample codes, sampling localities, site localities, GenBank accession numbers for all the sequences and references. In bold are the new records of this study. To visualize the relationships among specimens and clusters, we generated a distance-based tree using the neighbor-joining (NJ) method and the Kimura 2-Parameter (K2P) substitution model, using the software MEGA 6.0 (Tamura et al. 2013). Bootstrap resampling (Felsenstein 1985) was applied to assess the support for individual nodes using 1,000 pseudo-replicates. The intra and interspecific genetic distances were inferred based on the Ward (2009) standard parameters for DNA barcoding of fishes. The marine pellona Pellona ditchela (GenBank accession number AP011609) was used as outgroup.

Results
Four specimens of P. castelnaeana were collected from the Branco River, along with 1 reference specimen from Coari; one specimen of P. flavipinnis was collected from the Branco River, along with 4 reference specimens from Manaus. For both species the NJ tree shows individuals collected from the Branco River grouped with individuals from the reference populations (Fig. 2b). The maximum intraspecific genetic distances obtained were 0% for P. castelnaeana (mean 0%) and 1.2% for P. flavipinnis (mean 0.6%); the minimum interspecific distance between the two was 4.9% (Table 2). These results are in agreement with the barcoding parameter for fishes (Ward, 2009), which suggest a limit up to 2% (maximum intraspecific distance) to consider several individuals belonging to the same species.
Morphological characters, however, can be used to confirm that the specimens collected in the Branco River belong to the species P. castelnaeana and P. flavipinnis.

Discussion
One interpretation for the presence of the P. castelnaeana and P. flavipinnis in the Branco River is based on their migratory life history. The apapás undertake migration for spawning in headwater tributaries (Ikeziri et al. 2008, Le Guennec andLoubens 2004) and mainly use white waters for this purpose because they are rich in nutrients and primary productivity (Barthem and Goulding 2007). The Branco River is considered an important white-water Amazon tributary, and most likely, these fishes are using the river for spawning.
We can conclude that the individuals collected in the Branco River sub-basin correspond to the species P. castelnaeana and P. flavipinnis, thus expanding the occurrence area for both species by approximately 650 km north of the Amazon basin. Additionally, given the high levels of exploitation of apapás by commercial fisheries (Ikeziri et al. 2008), we consider that ecological and genetic studies on these species are needed to reveal their population dynamics, population structure, and wider taxonomic status. These studies are especially important for P. flavipinnis, whose diversity may be underestimated, since our Branco River individual represented a single haplotype with divergence up to 1.2% from the Manaus reference sequences.
The DNA barcoding methodology was shown to be efficient for the identification of Pellona species, and can be considered an additional tool for confirmation of new   records for other Amazonian taxa, but should be verified against morphological data.