Genetic analysis reveals candidate species in the Scinax catharinae clade (Amphibia: Anura) from Central Brazil

Abstract Scinax (Anura: Hylidae) is a species-rich genus of amphibians (113 spp.), divided into five species groups by morphological features. Cladistic analyses however revealed only two monophyletic clades in these groups: Scinax catharinae and Scinax ruber. Most species from the S. catharinae clade are found in Atlantic rainforest, except for Scinax canastrensis,S. centralis, S. luizotavioi, S. machadoi,S. pombali and S. skaios. In the present work, specimens of Scinax collected in Chapada dos Guimarães, central Brazil, were morphologically compatible with species from theS. catharinae group. On the other hand, genetic analysis based on mitochondrial (16S and 12S) and nuclear (rhodopsin) sequences revealed a nucleotide divergence of 6 to 20% between Scinax sp. and other congeners from the Brazilian savannah (Cerrado). Accordingly, Bayesian inference placed Scinax sp. in the S. catharinae clade with high support values. Hence, these findings strongly indicate the presence of a new species in the S. catharinae clade from the southwestern portion of the Brazilian savannah. To be properly validated as a novel species, detailed comparative morphological and bioacustic studies with other taxa from Brazil such asS. canastrensis, S. centralis, S. luizotavioi, S. machadoi, S. pombali and S. skaios are required.

The S. catharinae group (Frost, 2015) is characterized by the lack of an anterior process in the suprascapula, m. depressor mandibulae without an origin at the dorsal fascia of the m. dorsalis scapulae, distal division of the middle branch of the m. extensor digitorum comunis longus, and insertion of this muscle at the medial side on the tendon of the m. extensor brevis medius digiti IV (Faivovich, 2002). The vocalization of frogs from this group is usually composed of short notes and, sometimes, displays harmonic structure (Pombal et al., 1995a, b).
Most species in this group are distributed throughout the Atlantic rainforest (Faivovich et al., 2005). The only exceptions reported so far include S. canastrensis, S. centralis, S. luizotavioi, S. machadoi, S. pombali and S. skaios, which were observed in gallery forests within the Brazilian savannah (Cerrado) and in central and southeastern Brazil (Pombal and Bastos, 1996;Pombal et al., 2010;Lourenço et al., 2013).
Park in the southwestern Cerrado, some samples of Scinax morphologically compatible with species of S. catharinae group were collected, but these specimens were differentiated from all other species described so far. Therefore, the goal of the present study was to perform a molecular analysis of these samples as an additional tool to their taxonomic identification, besides verifying the presence of a putative new representative in the S. catharinae clade in areas distant from their center of origin.
Eight individuals of Scinax sp. were collected on April 04, 2006 in deep gallery forests alongside headwaters of the Coxipó River, in Chapada dos Guimarães, state of Mato Grosso, Brazil ( Figure 1, Table 1). The specimens were deposited in the Vertebrate Collection of the Universidade Federal de Mato Grosso (UFMT). Approximately 25 mg of muscle were removed from the inner thigh of each specimen and preserved in ethanol 95% at -20°C for molecular analyses.
The PCR conditions consisted of an initial denaturation step at 95°C for 5 min, followed by 35 cycles of denaturation at 94°C for 40 s, annealing at 55°C (12S and 16S) or 49°C (rhodopsin) for 40 s and extension at 72°C for 30 s, plus a final extension step at 72°C for 7 min. Subsequently, the reaction products were purified and sequenced in an ABI 3500XL Genetic Analyzer automatic sequencer (Applied Biosystems). Sequencing reactions were carried out by using terminal dideoxynucleotides (Sanger et al., 1977). The sequences were then aligned with Clustal W available in the software BioEdit v. 5.09 (Hall, 1999). The software GBlocks 0.91 (Castresana, 2000) was used to eliminate poorly aligned positions and divergent region portions of 16S, according to the following parameters: minimum number of sequences for a flank position to 10, maximum number of contiguous nonconserved positions to 08, minimum length of a block to 2, and allowed gap positions to within half.
To estimate the divergence matrix and phylogeny we added sequences of seven other anuran species from GenBank to our data set: S. catharinae, Scinax berthae, Scinax uruguayus, Scinax garbei, Scinax elaechroa, Scinax nasicus and Hypsiboas faber (outgroup). Two other species from the S. catharinae clade collected in Bahia, northeastern Brazil and Espirito Santo, southeastern Brazil, were also included in our analysis: Scinax strigilatus and Scinax agilis (Table 1).
A Bayesian phylogeny was inferred using the software MrBayes 3.1 (Ronquist and Huelsenbeck, 2003). The best mutation model was estimated according to Akaike Information Criteria -AIC in the software jModel Test 0.1 (Posada, 2008). Two runs (four chains each) with 20 million generations were performed with trees being sampled at every 1000 generations. Adequate burn-in was determined by examining likelihood scores of the heated chains for convergence on stationarity, as well as the effective sample size of values in Tracer 1.5 (Rambaut and Drum-Genetic analysis of Scinax catharinae 51 mond, 2007). We discarded 10% of the generations/trees. We considered relationships strongly supported when posterior probabilities were equal to or higher than 0.95. Eighteen sequences of 16S and 12S were obtained from each of the nine Scinax species, comprising 423 bp (164 variable sites) and 386 bp (131 variable sites) for each fragment respectively. For rhodopsin, 13 sequences of 316 bp with 51 variable sites were obtained from eight Scinax representatives.
The Bayesian consensus phylogeny (16S + 12S + rhodopsin) placed Scinax sp. as a distinct clade with strong support, being closely related to S. berthae, S. catharinae, S. strigilatus and S. agilis, all belonging to the S. catharinae clade (Figure 2 and Table 2). The four species from the S. ruber clade also formed a monophyletic group with strong support.
Even though the cytochrome C oxidase I (COI) gene has been elected as a universal DNA barcode in animals (Hebert et al., 2003), the 16S gene seems to be more effective to discriminate amphibian species (Vences et al., 2005), thus being used in the present study. Indeed, the genetic distances of 7 to 10% in 16S rDNA observed between Scinax sp. and the other known species in the S. catharinae clade (S. berthae, S. catharinae, S. strigilatus and S. agilis) ( Table 2) are higher than the minimum value of 3% in nucleotide divergence proposed by Fouquet et al. (2007) to discriminate anuran species. Moreover, sequences of 12S and rhodopsin (nuclear) were also included to provide additional support to our hypothesis of a new species in the S. catharinae clade occuring in the Chapada dos Guimarães.
Many researchers advocate the integration of multiple approaches (molecular, cytogenetic, morphological and ecological studies) for identifying species (Dayrat, 2005;Padial et al., 2010). According to the nomenclature rules established by Vieites et al. (2009), Scinax sp. could be classified as an "unconfirmed candidate species" (UCS), depending on additional morphological, ecological and vocalization studies to confirm its taxonomic status.
In conclusion, our molecular data provide evidence of a new species in the S. catharinae clade occurring in the Chapada dos Guimarães region, central Brazil. However, further morphological and bioacoustical analyses should be performed and focused on comparative data with other species from the S. catharine clade from Brazilian savannah, such as S. canastrensis, S. centralis, S. luizotavioi, S. machadoi, S. pombali and S. skaios.